CATS’

PAWS
and
CATAPULTS
Mechanical Worlds of
Nature and People

STEVEN VOGEL

Illustrated by Kathryn K. Davis

with the author

W. W. NORTON & COMPANY

New York • London
For Jane
 Contents

Preface
1. Noncoincident Worlds
2. Two Schools of Design
3. The Matter of Magnitude
4. Surfaces, Angles, and Corners
5. The Stiff and the Soft
6. Two Routes to Rigidity
7. Pulling versus Pushing
8. Engines for the Mechanical Worlds
9. Putting Engines to Work
10. About Pumps, Jets, and Ships
11. Making Widgets
12. Copying, in Retrospect
13. Copying, Present and Prospective
14. Contrasts, Convergences, and Consequences
Notes
References
Index
 Preface

Life is what biology’s about. Technology is something else altogether. Or so I

believed before I got into a kind of biology that’s about technology as well as
life. More to the point, it—biomechanics—looks at the technology of life, at the
mechanical world of nature. Sometimes that world resembles the mechanical
world that we humans have created. But sometimes the two differ strikingly.
This book compares those technologies. It’s about the ordinary things and
creatures around us; it intends, immodestly, to change the way you look at your
surroundings—at least a little. It has some other missions as well.
I’ve come to realize that engineers are as curious about our world as we are
about theirs. Some suspect that a look at organisms might help them create
designs and fabricate devices. How could a biologist disagree? But shifting from
one world to the other isn’t a trivial matter, and the traveler needs a road map
and a guidebook. This book tries to provide them by introducing biomechanics
in a point-by-point comparison with the more familiar world of our own
technology.
At the same time I want to inject an element of sobriety into our romantic
view of living things. The elegance of natural design seduced a lot of us into
becoming biologists. Nature does what she does very well indeed. But—and
here’s the rub—why should she do so in the best possible way? And why should
she provide a model for what we want to do? I want to ruffle our tendency to
view nature as the gold standard for design and as a great source of technological
breakthroughs.
Beyond that, I want to argue that natural design provides no honest foil for
skewering human technology. In getting together the material for this book, I’ve
repeatedly bumped into an antitechnological literature—nature-worshiping,
engineer-bashing tracts. Its authors take a thoroughly unrealistic view of our
contemporary situation and prospects and lay blame inappropriately; hanging the
social consequences of technology on engineers amounts to hanging airport
congestion on the Wright brothers.
About this last mission I cheerfully admit personal bias. I have no formal
background in engineering and only a primitive knowledge of the underlying
physics and mathematics. I couldn’t have done what science and writing I’ve
done without the unfailing generosity and support of engineers. For nearly forty
years, and at several institutions, they’ve explained things, steered me out of cul-
de-sacs, pulled my foot out of my mouth before I published nonsense, and
suggested accessible source material; in short, they’ve done everything
imaginable to welcome me to their domain.
Doing this book has treated me to an intellectual feast. I indulged my
fondness for building things by making a version of an ancient Egyptian drill,
and I had a pretext to lay hands on (and read) the first book printed on paper
derived from wood pulp. I had an excuse to use nine libraries on the Duke
campus (and several elsewhere), not to mention interlibrary loans, CD-ROMs,
government documents, various on-line databases, old newspapers on
microfiche, and network news groups.
I remain daunted by the clear relevance and separate sophistication of
anthropology, archaeology, paleontology, economics, architecture, geometry,
geography, law, and the histories of science, technology, exploration,
domestication, and culture. All the complex interconnections bring to mind the
biomechanical problems of keeping your finger in the air while your ear’s to the
ground and of keeping your feet on the ground while your head’s in the clouds.
I’ve drawn quite shamelessly on professional colleagues and other friends. In
no other project have I received such a treasure of useful suggestions, ideas, and
examples. I’m especially indebted to Matthew Healy, Michael LaBarbera,
Catherine Loudon, Jane Vogel, and Stephen Wainwright, each of whom read the
entire manuscript in first draft and made copious but always kind and tactful
suggestions. In addition, useful ideas emerged from conversations with David
Alexander, Michael Blum, Richard Burian, Steven Churchill, Ruth Day, Martha
Dunham, Betsey Dyer, Shelley Etnier, Robert Full, Margaret Hivnor, Diane
Kelly, Peter Klopfer, Daniel Lieberman, Dan Livingstone, Anne Moore, M.
Patricia Morse, Bruce Nicklas, Francis Newton, Fred Nijhout, George Pearsall,
Charles Pell, Henry Petroski, Jeffrey Podos, Michael Reedy, Knut Schmidt-
Nielsen, Kalman Schulgasser, John Sharpe, Robert Teer, Edward Tenner, Lloyd
Trefethen, John Wourms, and many other people of whom I lack a proper list.
I’m grateful to a host of helpful librarians, especially Richard Hines and David
Talbert.
 Edwin Barber, my editor at W. W. Norton, has done more for my writing
than anyone since my student days. In particular, he has given detailed guidance
as well as general admonishment in the struggle for spontaneity and the battle
against academic pretentiousness and obfuscation.
Finally, I’d like to pronounce words of passionate appreciation and advocacy
for libraries where what you need—or don’t know you need until you see it—sits
on end user–accessible shelves. Or their electronic equivalents.
Several of the figures derive from previously published material. For
permitting redrawing, I gratefully acknowledge the following copyright holders.
Figure 2.4: Columbia University Press (Fig 8.5 of B. D. Dyer and R. A. Obar,
Tracing the History of Eucaryotic Cells); Figure 4.15: Academic Press, Inc. (Fig.
2 of J. T. Finch and A. Klug, J. Mol. Biol. 15: 344 and Fig. 5.4 of R. E. F.
Matthews, Plant Virology. 3rd ed.); Figure 5.12: Dr. Mimi Koehl (Figure 1-9 of
Duke University Ph.D. dissertation); Figure 8.4, American farm and Darrieus
rotor windmills: Van Nostrand- Reinhold, Inc. (Figs. 22, 42 of F. R. Eldridge,
Wind Machines, 2nd ed.); Figure 10.10: Dr. Vance A. Tucker (Cover, Science, 14
Nov. 1969); and Figure 13.3: John Wiley and Sons, Inc. (Fig. 5.19 of M. E.
Rosheim, Robot Evolution). For permitting direct copying of Figures 11.1 and
13.2, I acknowledge the copyright holders, Gordon and Breach Scientific
Publishers, Inc. (Fig. 27 of J. Kastelic et al., Connective Tissue Research 6: 11)
and Dr. William M. Kier (Fig. 11 of Duke University Ph.D. dissertation),
respectively. For allowing redrawing of the mollusk shell of Figure 6.8 (from S.
W. Wise and W. W. Hay, p. 427 in Trans. Amer. Micr. Soc. 87), I thank Dr. Vicki
Pearse, editor, Invertebrate Biology. For loaning the Arctium for Figure 12.12, I
thank Dr. Robert Wilbur, director, Duke University Herbarium. Sam (Soft and
Mellow) Cat suggested the title’s feline allusion; he never confuses the mouse of
one technology with that of the other and remains ever hopeful that the printer
will emit something better than paper.
STEVEN VOGEL
Durham, North Carolina
CATS’ PAWS
and
CATAPULTS
 Chapter 1

NONCOINCIDENT WORLDS

When some of us were much younger—for me the late 1940s—we read Flash
Gordon every Sunday in the comics. With the casual confidence of kids, we
assumed that space travel to extraterrestrial civilizations was just around the
corner. Mr. Gordon seemed as close to our world as George Washington and a lot
closer than Julius Caesar. While Star Trek, with Mr. Spock, is leagues ahead of
Flash Gordon in sophistication, reaching other beings has become a much
dimmer prospect. At fault is neither the loss of our youthful certitude nor any
inferiority of Spock to Gordon. What’s happened is that the arrival of the space
age, like any reality, has brought its potion of sobriety, its diminution of
innocence. Space travel has proved a lot trickier—and more expensive—and
extrater- restrial civilizations a lot more remote than we ever imagined.
Nonetheless, the success of Star Wars and Star Trek and the continued
popularity of science fiction testify that the allure remains. Much of the appeal
obviously turns on their views of alternative cultures, things in short supply here
on earth. Ours is a single world. The civilizations of the Orient and those of
Europe and Africa have interacted for more than a thousand years, and extensive
intermingling with those of the Americas has now gone on for five centuries.
Human technology may have become vastly more complex, but it has lost
diversity and frozen into a stereotype. The global convergence leaves no Atlantis
in the offing. Meanwhile we’re ever more doubtful that we’ll bump into any
other technology.
A shame, perhaps, but not all that bad. We do have an alternative technology
as a mirror in which to view our own—the technology of organisms, the result of
evolution by natural selection over the past few billion years. Life forms a
technology in every proper sense, with a diversity of designs, materials, engines,
and mechanical contrivances of every degree of complexity.
As systems to compare we could ask for nothing better than nature s designs
and human inventions. Nature’s technology occurs on the surface of the same
planet as that of human culture, so it endures the same physical and chemical
limitations and must use the same materials. But nature copes and invents in a
way fundamentally different from what we do. At the very least, the rate at
which she alters herself is glacial by our cultural standard.
The very shapes of the two technologies differ dramatically. Just look around
you. Right angles are everywhere: the edges of this page, desk corners, street
corners, floor corners, shelves, doors, boxes, bricks, and on and on. Then look at
field, park, or forest. Where are the right angles? Absent? No, but rare, which
raises questions. Why so few right angles in nature? Why do civilizations find
them so serviceable?
Natural and human technologies differ extensively and pervasively. We build
dry and stiff structures; nature mostly makes hers wet and flexible. We build of
metals; nature never does. Our hinges mainly slide; hers mostly bend. We do
wonders with wheels and rotary motion; nature makes fully competent boats,
aircraft, and terrestrial vehicles that lack them entirely. Our engines expand or
spin; hers contract or slide. We fabricate large devices directly; nature’s large
things are cunning proliferations of tiny components. One can easily continue;
indeed, much of this book is simply an exploration of such contrasts—of their
mechanical aspects in particular.
At some very basic level all of us recognize how different are the products of
humans and of nature. Artists take advantage of that subconscious perception to
jar us by depicting one culture using the forms of the other. The cubists draw
human faces with flat sides and harsh, straight edges, often at right angles to
each other. Salvador Dalí, in painting, and Claes Oldenburg, in sculpture, re-
create the hard objects of manufactured technology—watches, engine blocks,
and such—with the natural world’s lack of rigidity. The incongruity startles,
intentionally and dependably.
FIGURE 1.1. Corrugated or fanfold surfaces as cheap routes to stiffness: scallop shell, corrugated
paperboard, and ridge and valley roof.

But one can easily make too much of these differences. Both bicycle frames
and bamboo stems take advantage of the way a tube gives better resistance to
bending than a solid rod. A spider extends its legs by increasing the pressure of
the fluid inside in much the same way that a mechanical cherry picker extends to
prune trees or deice planes. Both technologies construct things using curved
shells (skulls, eggs, domed roofs), columns (tree trunks, long bones, posts), and
stones embedded in matrices (worm tubes, concrete). Both use corrugated
structures (as in Figure 1.1) to get stiffness without excessive mass—whether the
shell of the scallop, one of the rare swimmers among bivalve mollusks, or the
stiffening structures of doors, packing boxes, and aircraft floors, or fan-folded
paper and occasional roofs. Both catch swimming or flying prey with filters
through which fluid flows—whether spiders or whales, gill-netting fishers or
mistnetting birders.

I care very much about the ways of engineering, but by profession and
predilection I’m a biologist, not an engineer. The implied equivalence between
the two fields is a little misleading, though; the two are not merely opposite sides
of the same fence. The biologist studies something that exists: nature, in all its
splendor. The engineer, by contrast, creates. Further, the engineer’s successes
have more immediate impact than those of the biologist, and failure exacts
penalties far beyond the approbation of a few peers.1
 In practice, calling my calling biology isn’t specific enough since the
mechanical interests of most biologists don’t go beyond keeping their scientific
equipment cooperative. So we who look at nature’s mechanical aspects call our
field biomechanics2 (or, as on a reimbursement check I once received from the
American society of the same, biomechantics— whether in high spirits, high
dudgeon, or pure accident). Awkwardly, something called biotechnology looms
large on the contemporary scene. As it happens, biotechnology in its current
sense is quite a different endeavor and will play almost no role here.
Biotechnology is largely a synthetic rather than our analytic activity; in addition,
it mainly focuses on the molecular and microscopic while we care more about
the mechanical and macroscopic. By contrast, what goes by the name biophysics
is analytic enough, but it’s similarly rooted in a molecular domain.
The biomechanic ought to make a candid, if unflattering, admission at the
outset. Simple logic suggests examining nature’s mechanical technology as a
first step toward both creating and understanding human technology—at least to
establish the range of possibilities. Who can deny that nature got here first? In
fact, the shoe is almost always on the other foot. The biomechanic usually
recognizes nature’s use of some neat device only when the engineer has already
provided us with a model. Put another way, biomechanics mainly still studies
how, where, and why nature does what engineers do.
One might reasonably expect any proper biologist to find natural, human-
free systems impressive, even aesthetically standard-setting. And often we do;
without our affection for nature, we might have chosen other ways to spend our
time, so almost without exception biologists are biophiliacs. But loving nature is
not at all the same as finding her perfect, a gold standard for design. A lot of
famous and otherwise estimable people have viewed nature as some Edenic
perfection of process and product. Using short quotations out of context may be
a little unfair, but as a biologist I cringe at statements such as those that follow. I
can forgive the ancients more easily than my contemporary post-Darwinians.

If one way be better than another, that you may be sure is Nature’s way. (Aristotle, fourth century
B.C.E.)3

Human ingenuity may make various inventions, but it will never devise any inventions more
beautiful, nor more simple, nor more to the purpose than Nature does; because in her inventions
nothing is wanting and nothing is superfluous. (Leonardo da Vinci, fifteenth century)4

Sources of hydraulic contrivances and of mechanical movements are endless in nature; and if
machinists would but study in her school, she would lead them to the adoption of the best
principles, and the most suitable modifications of them in every possible contingency. (Thomas
Ewbank, mid-nineteenth century)5

One handbook that has not yet gone out of style, and predictably never will, is the handbook of
nature. Here, in the totality of biological and biochemical systems, the problems mankind faces
have already been met and solved, and through analogues, met and solved optimally. (Victor
Papanek, contemporary)6

This casual attitude toward nature’s automatic excellence can’t be casually

dismissed. For one thing, it implies that the engineer or entrepreneur who copies
nature will leap ahead of those plodders who rely on mere human ingenuity. For
another, it appeals all too persuasively to those who blame engineers for the ills
of the modern world.7 I find neither attitude attractive. (But engineer bashing
has receded somewhat, even as an antiscientific community has flourished;
maybe some of its acolytes of the sixties and seventies got hoisted in the eighties
and nineties by their undeniable affection for personal computers.)
So is this a book about copying nature? Emphatically not. As we’ll see, on
surprisingly few occasions has copying proved useful. Indeed, felicitous transfer
of bits and pieces should not be expected. We’re dealing with separate contexts
of mechanical design, each system uniquely integrated by its own elements of
internal harmony and consistency. Moreover, one of these systems, even though
it includes ourselves (we are, after all, creatures of nature), is a far stranger
mechanical technology than commonly realized. So nature’s version needs
special attention, and thus a biologist-biomechanic feels compelled to write a
book about technology—or, better, about technologies.
 Chapter 2

TWO SCHOOLS OF DESIGN

Almost anything starts from a plan, whether a blueprint, template, some

macromolecular chemical code, or just a scheme held in mind. But no plan is
without antecedent, whether in an individual’s mysterious alchemy of experience
or as the result of innumerable ancestral adjustments. And neither human nor
natural technology represents a single act of creation. But nowhere do they
diverge more than in how their plans originate, in the processes we might call
design.
Nature’s process is that mechanism Darwin uncovered, evolution by natural
selection. Human technology springs from what is variously called invention,
discovery, development, or planning. A little confusingly, the word “evolution”
has recently been associated with human technological progress.1 Sometimes
that implies a kind of selective process, but most often it just alludes to
incremental change, with things building one upon another.

THE NATURE OF NATURAL SELECTION

Oddly enough, the familiar act of human creation is harder to explain in
acceptably scientific terms than the way nature creates her devices. The very
everyday character of human creativity, though, to anyone who has drawn a
picture, written a poem, or baked a cake allows us to evade any precise
formulation. By contrast, the intuition recoils when we are faced with evolution
by natural selection. It can have direction and can perhaps even make progress,
yet do so with no semblance of planning. Its information is carried and dealt
with by molecules, and molecules lack proper perceptual reality. And its time
course ordinarily exceeds anything in our direct personal experience. Even if its
reality is now beyond serious doubt, the whole thing just seems overwhelmingly
unlikely.
What concerns us here is its mechanism, for in that specific mechanism
called natural selection lie both the power and the disabilities of the evolutionary
process. Putting that mechanism as a series of observations and interconnected
statements (see the following box) should emphasize the persuasive underlying
logic as well as offset a lot of misconceptions and mystical notions of nature’s
perfection.2

1. Observations:
a. Every organism can produce more than one offspring, so populations, if unrestrained, will
increase continuously.
b. Every organism needs some minimum amount of material from the environment to survive
and reproduce.
c. The material available to a population of organisms is finite in extent, restraining its increase.

2. Consequence of a, b, and c:
d. A population in a given area will rise to some maximum size.

3. Consequences of a and d:
e. For a population at this maximum size, more individuals will be produced than the
environment can support.
f. Some individuals will not be able to survive and reproduce.
4. Further observations:
g. Individuals within populations vary in ways that affect their success in reproduction.
h. At least some of this variability is inherited; individuals resemble their parents more than they
do more distantly related individuals.

5. Consequences of e through h:
i. Characteristics that increase the number of an individual’s surviving offspring will be more
prevalent in the population in the next generation.

The final statement, of course, encapsulates evolution by natural selection.

Notice that nowhere in the scheme does “design” appear. Using the word as a
noun raises no hackles, but making it an honest verb is nearly impossible. To
design ordinarily requires a designer. In evolution, though, change happens as a
blind result of selection for whatever improves reproductive success. In this
sense, nothing in nature is “designed” or has “purpose.” Nonetheless, most
conspicuous bits of biological structure do serve specific functions. How else
could they increase reproductive success? The ear of an animal that’s preyed
upon enables it to hear a predator’s approach, to take evasive action, and perhaps
to live and breed. So in another sense, the one relevant here, the ear’s “design”
certainly does have a “purpose.” A seminal work in biomechanics has provoked
neither controversy nor offense with the name Mechanical Design in
Organisms3.
In short, design in nature is a process of generating variability (item g above
—mutation and recombination, to get technical about it) and then selecting those
variants that are beneficial. Naturally, most variations are either neutral or
detrimental. The late George Beadle, who received a Nobel prize for his work in
genetics, used the analogy of a typist repeatedly copying a page of manuscript.
Each copy is proofread for errors; if an error appears, the copy is discarded—
except for the rare error that improves the prose. If one of these appears, then the
new version becomes the model to be copied. So nature’s process is inefficient
but inexorable, given a supply of errors and some kind of selection. Random
errors yield nonrandom change, and anticipation or planning needn’t occur.
This thoroughly stupid process of natural selection can generate results we’d
hardly characterize as progress. Consider the consequences of a mammalian
mating system in which males have some innate preference for large-breasted
females. With remorseless logic, selection will supply increasingly well-
endowed females, quite beyond any utility for nursing neonates and well into the
range of mechanical awkwardness. Perhaps the example might not be
hypothetical; even with our prothoracic cultural bias, breast reduction surgery is
far from uncommon.

THE LIMITATIONS ON EVOLUTIONARY

CHANGE
The dazzling diversity of the living world too easily disguises the fact that the
evolutionary process faces constraints far more severe than anything impeding
human designers. We biologists recognize these constraints, but we don’t often
rise above our natural chauvinism and make enough public noise about them.
Every organism must grow from an initially smaller to an ultimately larger
size. Nature, in effect, must transmute a motorcycle into an automobile while
providing continuous transportation. The need for growth without loss of
function can impose severe geometrical limitations. Consider the possible shapes
of mollusks, a widespread and diverse group that includes scallops, slugs, snails,
and squid. Mollusk shell doesn’t grow, so a shell cannot enlarge except by
adding incrementally to edges and inner surfaces. For most shapes, such
incremental additions would quickly lead to awkward changes in proportions, as
in Figure 2.1. For example, enlarging a cylindrical shell by lengthening would
make it relatively skinnier and more vulnerable to breakage; a long rod is more
easily broken than a short one, as you can show with a strand of dry spaghetti.
Compensatory internal thickening would reduce the relative volume available
for the critical guts and gonads. Hollow cones are much better than cylinders or
most other shapes if what matters is growth in size without simultaneous change
in shape. And mollusk shells are basically cones, whether single cones, as in the
snail- and nautiluslike forms, or doubly conical, as in the clamlike forms.
Hypothetical mollusk shells can be generated as conic derivatives with a
computer, something first done by the paleontologist David Raup thirty years
ago.4 The constraint on their shapes is so severe that almost all the forms of real
shells can be produced by a fairly simple program.

FIGURE 2.1. Increasing size by adding to an edge. Cone and pyramid retain their shapes, while cylinder
and rectangular solid are disproportionately elongated.

Life can’t easily pause periodically for renovation, although such things as
the pupal stage of insects come close to a growth hiatus. Arthropods in general—
insects, spiders, and crustaceans, mainly—grow fitfully, periodically shedding
their outside casings as well as some of the inside equipment. Molting thus
provides an alternative to the edgewise growth of molluscan skeletons. But it
restricts structural possibilities also, surely precluding many useful internal
pillars, trusses, and the like. In addition, while insects are consummate fliers, no
insect molts functioning wings, so the flying stage is always a final, nongrowing
one; little flies are never baby flies. Finally, molting costs material and imposes
periods of mechanical vulnerability.5
One major group of animals has reconciled support and growth. We
vertebrates have a skeletal system that can grow and remodel itself continuously.
By contrast with mollusk shell and arthropod cuticle, bone is a living tissue, a
complex and unusual accomplishment of fishes, frogs, birds, and people. A
growing skeleton may be the greatest vertebrate innovation, the central item in
our success as moderate-size to large creatures.
Organisms must also reproduce, so they (the females, in species with
separate sexes) need the full equipment to make more of their kind. That’s
adding a requirement that an automobile tow behind it a factory for making
automobiles or at least making motorcycles. Again, some partial evasions are
known among complexly colonial creatures. Among bees and ants, for instance,
workers are nonreproductive. Among colonial coe- lenterates (the group that
includes jellyfish and sea anemones) only a few specialized polyps of the colony
—what looks to us like an individual is really a colony—can have offspring.
Even so, the reproductive units are still recognizable organisms—queen bees and
reproductive polyps. No organism has invented a proper out-of-body progeny
production plant.
After growth and reproduction comes dispersal. Sometimes the three may be
done by a single form of an organism, as in creatures like us that have simple life
histories. In other cases elaborate metamorphoses may separate dramatically
different forms. We’re the unusual ones; metamorphoses and the resulting
complex life histories characterize most plants, many insects, and almost all of
the great diversity of marine invertebrates. We humans tend to take mobility for
granted, but it’s tough for a tree, an oyster, a parasitic worm, or a sponge to get
from here to there. Each of these discharges special samples of itself to spread its
kind. In many barnacles and clams, tiny, swimming, predatory larvae convert
into sessile, reproductive adults that feed by filtering microorganisms from
water. In butterflies, leaf-munching growing crawlers become nectar-sipping
reproductive fliers. Such conversions must seriously constrain the range of
possible designs.
Other limitations are imposed by what we might call informational
constraints. The plan for making any organism has a problem that deserves more
attention. In this era of bytes and computers, we recognize information as
something quantitative and measurable. The basic unit, the bit, resolves the
uncertainty in a choice between two equally probable alternatives, the
information you gain when you look at which way a flipped coin lands.6 In
essence, plans are stores of information. To construct a human or similar animal,
a fertilized egg has available around 1010 (1 followed by 10 zeros) bits of
information in its DNA. That may sound like a lot—until one realizes that each
of us has about 1014 cells, a number no less than 10,000 times greater. So
10,000,000,000 bits isn’t such a big deal. As we’ve learned from our computers,
two-dimensional representations—graphics—absorb far more memory than
mere text. Whether a picture is worth a thousand words, it surely uses as much
disk space. Organisms, though, are three-dimensional, and details as fine as a
millionth of a millimeter are important. Specifying that much detail should
require a truly vast store of information, many millions of times greater than the
1010 bits in egg or sperm. Thus the shape of an organism has to be set by,
relatively speaking, a very sketchy set of plans. That the chickens in a flock look
identical cannot result from identical specification of every detail of each.

FIGURE 2.2. A helix of identical wooden blocks and a model of a microtubule. The latter has thirteen
elements per turn. each consisting of a pair of protein molecules. In either case, each element is in a
position equivalent to that of every other.

This shortage of information clearly underlies a lot of biological design.

Back in 1950 a prescient physicist, Horace R. Crane,7 predicted that a lot of
subcellular structures (he didn’t know which) would turn out to be helical in
form, not because helices necessarily worked best but because they could be
assembled with especially simple instructions. A helix can be built from identical
subunits (as a wall is built from identical bricks); also, every subunit is inserted
in exactly the same way as every other, as in Figure 2.2. If you know how to
install one subunit, you know how to do the rest. Crane anticipated not only the
double helix of DNA but its supercoiling (a helix of helices), the so-called alpha
helix of parts of many proteins, and, on a larger scale, helical microtubules and
microfilaments important in maintaining the shape and motility of cells.
Microtubules and microfilaments have a remarkable capacity for self-assembly;
if all the components are put together (with perhaps a bit of the formed structure
as a starter), they ordinarily fall into place without any need for mold or
scaffolding or, more important, for any additional information.
Building large organisms out of lots of cells is probably made necessary by
that shortage of information. Cells may look diverse, but they all have a lot in
common; if you can build one kind, you need only a little more information
(relatively, of course) to build all the others. Furthermore, in the development of
each individual, one group of instructions can set more than one structure. In
humans, hand size is an excellent predictor of foot size; before stretch fabrics
were common, the salesperson wrapped a sock around your fist to tell whether it
would fit your foot. A single alteration of the genetic material—a mutation—
ordinarily affects both sides of the body of an animal. A mutant fruit fly doesn’t
have one white eye; it has two. Beyond these informational economies are
others. The hearts and lungs of all of us are in the same places, but at some level
of detail the locations of our parts are unpredictable. Anatomy students learn the
names of the large blood vessels, but the small ones stay blessedly anonymous—
simply because their arrangement varies from one person to the next.
Nor does evolution easily manage fundamental change. For one thing, the
variation that provides its raw material—genetic mutations and the juggling of
characters accomplished by sexual recombination—consists mainly of small
changes. Natural selection merely tests changes, such as a little thicker fur or a
little longer ear for possible reproductive advantage. For another, natural
selection operates on every individual. Thus any change must yield a fairly
immediate advantage if it’s to increase its representation in a population. Much
less useful is a structure whose reproductive advantage is realized only when
some other change appears.8 As the evolutionary biologist Richard Dawkins
eloquently argues (without controversy), the evolutionary process is a tinkerer
rather than a proper designer, a perpetual modifier rather than a creative
innovator.9
Quite obviously major innovations have occurred, and biologists have
exercised lots of ingenuity devising possible scenarios in which such innovations
don’t require long periods of elaboration of yet functionless structures. Birds,
bats, and insects fly with flapping wings. Since no small protowing makes a
proper aircraft, we worry about how wings got started in each of these
lineages.10 Did a running birdlike creature make longer hops with outstretched
feathered arms? Or did such appendages permit longer jumps from branch to
branch? Or did extended appendages— longer “arms”—help shed heat produced
by extended runs? Both flying squirrels and gliding lizards have membranous
skin between fore and hind limbs. Are these creatures reasonable models for
birds or bats at a stage before active, powered flight?
The evolutionary process has its hands tied in yet another way. Every
organism is a product of its particular evolutionary history. Such a history limits
design far more than ensuring that today’s disk will work in yesterday’s
computer. It’s tempting to assume that every organism is optimally attuned to its
personal circumstances as a result of its lengthy evolution, but it’s profoundly
wrong. Ancestry traps an organism. Consider a few fine features that appear only
in one lineage.
Earlier we spoke of the shells of mollusks, the cuticle of arthropods, and the
bones of vertebrates. Mollusks and arthropods never figured out how to make a
proper growing skeleton. Each adopted a scheme for coping with the problem of
support, one with the advantage (whether incidental or not, one can’t easily say)
of providing outer protection. Vertebrates solved the problem of making a
skeleton that could grow although in doing so, some of us—turtles and
armadillos, at present— had to make additional structures to gain that outer
protection.
Arthropods make an elastic protein, resilin, that has a higher resiliency (more
of the energy used to stretch it is recovered when it’s released) than anything
found in a mollusk or vertebrate. Mollusks like clams and scallops use an
alternative protein, abductin, to open their paired half-shells, as when a scallop
swims by clapping them together. Vertebrates use yet another protein, elastin, in
ligaments and blood vessel walls: See Figure 2.3. One of resilin’s main roles is
storing the energy that decelerates an insect wing at the end of one stroke to
accelerate the wing at the start of the next. Is making insect wings beat
efficiently more crucial than either making a good hinge ligament to connect the
half shells of a swimming scallop or making a low- loss elastic to support the
head of a grazing sheep? Probably not; most likely scallop and sheep would be
better off if they could trade proteins with a fly.

FIGURE 2.3. Three different elastic proteins, characteristic of three different organisms. The scallop uses
abductin in the hinge of its shell; the fly has a pad of resilin in the hinge of each wing; the ligament
connecting the head and thoracic vertebrae of a cow is largely elastin.

Mollusks aren’t inevitably losers in these comparisons. They alone have

mastered the trick of keeping a muscle contracted without expending energy.
Ideally, you don’t need energy to produce a force that just supports a load; you
need it only to move a load. Exerting force against an immovable load—Atlas
holding up the sky or a chain supporting a chandelier—ought to use no energy.
In practice, though, our own muscles take energy to do anything, even if nothing
moves. But the sedentary clam can stay clammed up at virtually no cost.
So nature must follow an inherited plan. The human designer, on the other
hand, can borrow devices from other designers. If the device is covered by
patents, then royalties or litigation form part of the process, but most useful
items are common knowledge in the public domain. For that matter, two
competing manufacturers can buy identical parts from the same supplier. Nature
has trouble doing anything analogous. Still, crosslineage transfer of technology
is not absolutely ruled out, and a pair of examples illustrates its nicely offbeat
flavor.
1. Termites harbor in their digestive systems protozoa called mixotrichs that
are critical to their ability to get energy by digesting cellulose. Colonial life
in termites may even have originated as a device to ensure adequate
opportunities for passing along “infections” of mixotrichs.11 These
protozoa were long assumed to be propelled by organelles called
undulipodia on their surfaces. Oddly enough, the undulipodia turn out to be
a bunch of bacteria, spirochetes in particular, arranged as shown in Figure
2.4. The protozoa have adopted bacteria as engines the way a human might
use a team of horses. The symbiotic association of course is much closer;
it’s obligatory for both kinds of organism.12 Even more curious, this kind
of symbiosis has occurred on at least one other occasion, with a different
bacterium in a different intestinal protozoan in another termite. In this latter
case the engines are the flagella of rod-shaped bacteria. Each of several
thousand bacteria on each protozoan has about a dozen flagella, oriented on
its surface so they all contribute to moving the protozoan around.13
2. Only coelenterates, such as jellyfish, know how to make certain special
stinging cells, their nematocysts. Contact with a big coelenterate (the
Portuguese man-of-war is especially vicious) is extremely unpleasant for a
person and often fatal for a fish. A few creatures somehow manage to
contact coelenterates without triggering the nematocysts. These, including a
few fish, then use the coelenterates for shelter, protection, and even food.
Certain nudibranch mollusks (marine snails without shells) take the trick a
giant step farther. Not only can they contact coelenterates without damage,
but they can incorporate the nematocysts into their own skins quite
undischarged and still defensively potent.14 Here no symbiosis occurs;
rather a jellyfish technology is appropriated and redeployed: They steal
loaded guns from the army.
FIGURE 2.4. A mixotrich with a pox of spirochetes. While the mixotrich has several flagella of its own (at
top), for some reason it doesn’t use them for locomotion.

In both these cases something mechanical crosses from one lineage to

another. More frequent are chemical transfers, in which, for instance, an animal
becomes poisonous to potential predators by ingesting a normally toxic plant (as
when monarch butterflies eat milkweeds). The general point is that recognizing
such cases shows that we know what to look for. So we can confidently declare
such transfers uncommon and assert their difficulty for natural selection.
As a designer, then, nature is not only glacial in speed but lacking in
versatility and erratic in performance. Fundamental innovation comes hard, and
once achieved, it disseminates almost entirely within a lineage. To a remarkable
extent the dazzling diversity in nature represents superficial features of systems
of an exceedingly conservative and stereotyped character. No patents exist to be
licensed; infringement by copying is impossible (although fortuitous
infringement—convergence—carries no penalty); and the bottom line is
immediate profit—surer reproduction. Trial, error, patience.
DESIGN FROM A COMPARATIVE PERSPECTIVE
Design in human technology is far less constrained. All the complications of life
histories are swept away. Fundamental change may be hard to bring about, but it
faces no fundamental barrier. Of course, for the most part we build on past
accomplishments, and we do so within limits set by human ingenuity and modes
of thinking, by materials at hand, and by the social support at any time and place
for innovation. The inventor, after all, must eat, and society must be willing to
adopt an invention. Without a doubt, human societies have run the gamut from
hostility to hospitality toward technological innovation. The Roman Empire,
given its size and duration, must be rated quite low; ships, building materials,
even weaponry changed little for hundreds of years. By contrast northern Europe
and North America during the nineteenth century were obviously high;
innovations included railroads, steamships, telecommunications, synthetic fibers,
and electric motors and lighting, just to name a few.
Design in human technology generates change and progress in a way that’s a
lot easier on the intuition than design in nature. On the other hand, if’s a lot
harder to encapsulate succinctly. An inventor, a concept, a model; testing,
approval, dissemination, improvement—the elements come easily to mind.
Alexander Bell thinks of a simple way to convert sounds to electrical signals and
back again. After a lot of tinkering he builds a successful model, he patents the
model, and then he and others commercialize the telephone. Thomas Edison’s
more efficient mouthpiece then supersedes this element of the Bell system. The
design process clearly involves the planning, anticipation, and deliberation of
which natural selection is incapable. But the really basic items of human design
are so old that we know little or nothing of their beginnings. Who, after all,
invented the right angle? Who first fabricated things from metals? Good
histories of technology have been written, but they can’t see back more than a
few millennia.
While the processes of design differ dramatically between the two
technologies, important elements are common to both—perhaps more than are
commonly realized.
Cultural dissemination. Nothing forbids invention and cultural transmission
among animals (plants are a different matter). Monkeys and apes invent
copiously, but my favorite case involves birds, not known as such a brainy
bunch. Some years ago individuals of four species of tits in Britain found that if
they pecked just right at the caps of milk bottles left on doorsteps, they’d be
rewarded with fine meals of cream. Cream might seem an unusual food for
birds, but if they can get it, it’s a grand source of fat, their fuel for flight. Those
cream-fattened fliers might have enjoyed greater reproductive success; their
offspring might have preferred pecking at objects in some class that included
bottle caps; natural selection might have fine-tuned the behavior in succeeding
generations. Dissemination among British tits in fact proved far more rapid.
Milk bottle pecking spread quickly as birds learned from one another—until a
change in the design of the caps foiled birds and some fascinated biologists. As
our biobard John Burns quipped, the case “smacks of the loosely preadaptive
inasmuch as tits were the birds to try it.”15

Natural selection. Nor does anything prevent accident and selection by

percipient humans. Lucky accidents must have been fairly important in early
human history. Domestication of plants and animals, the origin of cooking, the
use of a strangely shaped stone followed by deliberate shaping of others—all
need only minimal foresight or initial deliberation. Certainly folk medicine is
mostly selection from among nearly random ministrations. We learned that
chronic aspirin therapy lowers the risk of clot-associated cardiovascular diseases
from observations incidental to the use of aspirin for arthritis. Many a toothsome
recipe must derive from accidental alteration of a routine procedure.
Both natural and human technologies bow to economics. Improved
reproductive success—“fitness”—and corporate advantage are much alike.
Corporate advantage is especially close to what the biologist means by fitness
according to the economist John Kenneth Galbraith.16 He points out that
corporate managers typically favor expansion—equivalent to population growth
—over payout of profits to anonymous stockholders. Furthermore, one hears
complaints that the corporate culture takes an unduly short-term view of what’s
beneficial; immediate utility is thus the main measure of success in both
technologies.
The role of isolation. A lot of evolutionary change apparently occurs in small,
genetically isolated populations, in which competition with other members of a
species is limited by geographic or ecological barriers. In general, organisms that
don’t do a lot of moving around generate more different species than those that
do. When a barrier falls, a now well- tuned form may come into renewed
competition and displace another— part of why biogeography has become such
an important part of evolutionary biology.
With a globally uniform technology, we’ve less opportunity for such
temporary sheltering, and continuous stirring replaces the rare mixing of a
Marco Polo. But we still set up isolated habitats of reduced competitive
pressures. For better or worse, a lot of secret work done under military auspices
ultimately appears in open, nonmilitary circumstances. The space program of the
United States sheltered and subsidized the microelectronics industry. Surely
government-supported biomedical research (with results available in the public
domain) and the profitability of the pharmaceutical companies are joined at the
hip.

The conservative bias. No dominant technology displaces easily in a purely

competitive encounter. Cars long ago standardized on the Otto cycle internal-
combustion engine. Its nearly universal use doesn’t prove it the most
economically appropriate way to power passenger vehicles. A hundred years of
refinement and easy access to fuel and maintenance are advantages not easily
overcome. If that engine is ever displaced by another, it won’t be through pure
competition in an open, global marketplace. Present standards for the resolution
of broadcast television are clearly anachronistic, but with hundreds of millions of
sets in use, they’re proving hard to shake off. Similarly, insects constitute the
majority of the world’s animal species despite the awkwardness of periodic
molting and the resulting difficulty of making large terrestrial forms. In neither
technology does fundamental superiority of a newcomer give assurance of
success. Their respective histories hold each a captive.

The time course of change. In times past we believed in the steady advance of
human culture; everything went “onward and upward.” Nowadays we recognize
that our history is bumpier, more episodic. In biology we now generally accept
the idea that evolutionary change may occur as intermittent bursts, something
first brought to our attention by Niles Eldredge and Stephen Jay Gould.17 The
issue remaining is just the relative importance of their “punctuated equilibrium”
and steadier change through time (phyletic gradualism). Most of the examples on
which arguments turn are fairly arcane. But consider the explosive radiation of
an already old group of small creatures called mammals following the demise of
the dinosaurs sixty-five million years ago. Millions of years of stagnation and
then, wow! Another such episode happened more recently, a mere twenty million
years ago. Grasses evolved, and with them came extensive grasslands. Grass
may be easy to get to, but it’s poor fodder for the unprepared, abrasive stuff with
a low energy content for its bulk. Within a relatively short time, teeth in many
lineages of mammals evolved into a form (Figure 2.5) that could manage to
munch this miserable material.18 Less dramatically, punctuated equilibrium
implies that most of the time natural selection just maintains forms well arranged
for what they do; once an organism is well established, random alterations are
especially unlikely to improve it.
Looking for equivalent bumpiness in human technology requires that we
speed up our time scale. Doing so reveals the same variation in how fast change
occurs. The small electric fan you buy today has changed little in eighty years,19
but the laptop computer was almost unimaginable forty years ago. Fans, toasters,
and other small appliances came hard on the heels of light bulbs and the wiring
of households. Even for a stand-alone device, change is irregular. The body of a
single-lens reflex camera looks and works like one of twenty or fifty years ago,
but between skin and skeleton a layer of novel electronics has recently been
interposed. Probably the development of semiconductor amplifiers (transistors)
in the late 1940s and the related creation of digital integrated circuits (chips) in
the 1960s have been the main instigators of the most explosive changes in
contemporary human technology. By contrast, our homes, our vehicles, our
household appliances, and our clothing have changed far less than we might
have predicted forty or fifty years ago. Besides radically new electronics and a
few pieces of soft plastic, I see precious little novelty when I look around my
house.

FIGURE 2.5. A horse’s teeth are typical of those of large, grazing mammals. Vertical columns and layers of
materials (enamel, cementum, and dentin) differ in hardness. In use the harder material will always
protrude farthest, so the teeth do not get smooth as they wear down.

Incremental progress. Natural selection, as emphasized a few pages ago, is an

evolutionary rather than revolutionary process, one building on small changes.
Of course we humans can make major technological revolutions. But what do
we really do? Early education gives the impression that James Watt invented the
steam engine and Henry Ford the automobile. When examined in detail, though,
almost every recent human invention is part of an incremental sequence. Key
individuals certainly play important roles, but almost every recent history of
technology views the development of specific technologies as more gradual than
what most of us learned.20 That long-term progress is unsteady doesn’t mean
that short-term change isn’t incremental; just as in nature, geologically episodic
change isn’t revolutionary when viewed generation by generation.
New uses for old devices. Evolutionary biologists speak of preadaptations,
preexisting features that make organisms suitable for new situations. For
instance, amphibians and the rest of us legged vertebrates appear to have
evolved not from familiar ray-finned fishes but from a group of lobe-finned
fishes. (One representative of this group survives in deep water off the east coast
of Africa. Until its discovery in 1938 we thought that coelacanths had been out
of the picture for the past hundred million years.) Living in oxygen-poor
swamps, these lobe-finned fishes had evolved air-breathing lungs and muscular
fins that allowed them to keep their heads above water, perhaps as in Figure 2.6.
The lungs and lobe fins that permitted life in ancient swamps amounted to
preadaptations for our kind of ambulatory, air-breathing, terrestrial life.21

FIGURE 2.6. Not prehistoric lobe-finned fish but contemporary ray finned ones, a pair of mudskippers.
They prop themselves up in the manner suggested for creatures in the transition from fishes to tetrapods. No
proposal of descent is involved; the mudskippers just show that fish can do it.

The sudden expansion of human culture has brought to light all kinds of
preadaptations among organisms. For instance, a weed with seeds that are hard
to separate from those of a crop plant can suddenly find itself superbly well
suited to a vast new habitat. And some small flies (chironomids), whose larvae
normally live attached to the rocks in rapid streams, now thrive as a nuisance in
the aerators of sewage treatment plants.22
Preadaptation may be so common in human technology that no one pays it
much attention. Computers that displayed programs as text and were instructed
from a keyboard were obviously preadapted to become word processors. The
waterwheels long used as power sources provided a way to apply rather than
extract power in the first generation of steamboats. Indeed, using preexisting
things in novel ways enjoys a special countercultural mystique, giving particular
gratification if the items have outlived their original applications. For instance,
an old oil drum can be sliced lengthwise to make a rotor for one kind of
windmill. Some of us do it all the time. I recently pressed a plastic water pipe
into service as a curtain rod and used an ordinary metal lathe as a torsion-testing
machine.

Parallel developments. Nature often makes the same thing in several different
lineages. Such convergence includes some truly remarkable cases.23 Fleshy,
spiny, leafless plants evolved in the deserts of the Old World within one family,
the euphorbs (such as the crown of thorns plant); and similar plants evolved in
the American deserts within another family, the cacti. The most recent common
ancestor of cacti and euphorbs was not fleshy, spiny, and leafless; they’re truly
convergent. Marsupial and placental mammals are in many cases amazingly
similar. Marsupial mice, moles, flying squirrels, dogs, and others look like their
placental equivalents but (on overwhelming evidence) represent an independent
bush of mammalian evolution. The human (vertebrate) eye and the octopus
(cephalopod) eye look alike and work alike but form another case of independent
and convergent evolution. A good design is a good design, and that different
lineages are driven by natural selection in similar directions shouldn’t be
surprising. Convergence tells us a lot about functionally important characters,
since anything that converges must make a difference to reproductive success. It
also directs attention to what is relatively easy (in some sense) for the
evolutionary process.
 We pay less specific attention to convergence in human cultures and
technology, but it’s there. Some parallels, such as the invention of the calculus
by Leibniz and Newton in the seventeenth century or the suggestion of evolution
by natural selection by Darwin and Wallace in the nineteenth, are purely
intellectual. In each instance the intellectual climate must have been right.
Technological parallel development is probably commoner. Did Marc Brunel or
Eli Whitney invent interchangeable parts, Howe or Singer the sewing machine,
Swan or Edison the light bulb? Or, in each case, both? Bell–s first telephone
patent beat a competitor by a few hours. Others were so close to flight that the
Wright brothers had to strike a careful balance between immediate public
disclosure and proper patent protection for their aircraft. When technology is
changing, obvious next steps (need and ease, as in nature) will occur to more
than one person.
In early human history, when contacts between cultures were much more
limited, convergence must have occurred. Whether it’s common or uncommon is
debated between anthropologists of isolationist (independent origin) and
diffusionist (spread from a single origin) predilections. Did such things as
weaving, archery, and metallurgy arise more than once or only on single
occasions? Impressed with the extreme commonness of convergence in nature, I
ally more easily with the isolationists.

Extinction. We usually think of extinctions in Darwinian terms, but they result

as much from general change in habitat or circumstance as from directly
competitive inferiority. Characteristics beneficial during normal times are likely
to work against an organism faced with catastrophic change.24 After all, those
normally nice characteristics were selected when the future mirrored the past; a
plant that does well in the shade will have a hard time when the forest
disappears! At the least, when the world changes rapidly, extreme specialization
is likely to be counterproductive. One wonders whether the role of habitat
change in large-scale extinctions and the disadvantage of specialization hold
practical lessons for human technology. Certainly, the causes of extinction in
human technology must be comparably complex. Saddlemakers and farriers
declined not because of automation but because of automobilization. Perhaps the
internal- combustion engine will fade too, in the face of more expensive fuel or
unacceptable emissions (in effect, habitat change), not a competitor that’s
superior under current conditions.
Much has been written about evolutionary and revolutionary change. But these
terms acquire special connotative hazards from their biological associations and
political analogs. Maybe we ought instead to stick with less burdened words,
such as “gradual” and “jumpy” or (more pedantically) “incremental” and
“saltatory” for changes in small steps and great leaps. We’re biased toward a
jumpy view of history since we like to focus on heroes and to divide things up
into discrete events, time periods, and categories. An incomplete fossil record
makes us think that nature is similarly jumpy. Perhaps recognizing that we’re
prone to one bias will alert us to the pitfalls of the other.
The larger point is our recurrent theme that looking at both natural and
human technologies forces us to think about each in novel ways. Here we’ve
seen surprising similarities in practice despite the vast difference in underlying
mechanism. Similarities have had center stage; farther along the differences will
loom larger.
 Chapter 3

THE MATTER OF MAGNITUDE

Size matters, and like evolution, it will pervade all that follows. For one thing,
an effective design for large things often works poorly for small things, and vice
versa.1 For another, our two mechanical technologies span an enormous range,
from a virtual macromolecule to the largest of human structures. For yet another,
nature’s products are generally smaller than ours, although the ranges overlap
extensively. Since the two technologies share the same planet, they experience
the same pressures, temperatures, gravitational accelerations, winds, and water
currents. In many ways, though, the influence of such physical factors on the two
often proves profoundly different; practical reality depends very much on how
big something is.
How much size matters has long been recognized. Galileo gave it his full
attention, correctly calculating that (in the absence of air resistance) an animal of
any size ought to be able to jump as high as any other. This means that relative to
body length, the small ones win hands down. (Even in the real world of draggy
air, fleas are truly impressive, clearing the bar at several hundred times their own
length.)2 The great seventeenth-century French polymath Descartes put the
matter this way: “The only difference I can see between machines and natural
objects is that the workings of machines are mostly carried out by apparatus
large enough to be readily perceptible by the senses (as is required to make their
manufacture humanly possible), whereas natural processes almost always
depend on parts so small that they utterly elude our senses.”3
What confuses our intuitions—but didn’t mislead Descartes—is our own
atypical size. The smallest fully competent organism (thus excluding viruses), a
bacterium that gives you a mild pneumonia, is about 0.2 micrometers long, about
a fifth of a thousandth of a millimeter and just visible as a dot in a good light
microscope. The largest in volume is a large whale, a little more than 20 meters
or 60 feet long. That’s roughly a hundred million—fold range. On an appropriate
geometric scale, as in Figure 3.1, we humans hug the upper end. At about 2
meters long, we’re about ten times shorter than the biggest whale but ten million
times longer than the smallest bacterium. We use, for these comparisons, a
geometric rather than an arithmetic scale, counting each additional zero, or order
of magnitude, as an equivalent increment. As a pioneer of biomathematics,
D’Arcy Thompson, puts it, “It is a remarkable thing, worth pausing to reflect
upon, that we can pass so easily and in a dozen lines from molecular magnitudes
to the dimensions of a Sequoia or a whale. Addition and subtraction, the old
arithmetic of the Egyptians, are not powerful enough for such an operation.”4
(D’Arcy Thompson needs a few words. He’s almost exclusively known for On
Growth and Form, a large book written in 1917 and again in 1942,
unquestionably the best- known work on mechanical aspects of biology. Part of
the book’s continuing impact—it’s still in print—comes from its shear linguistic
splendor; more, perhaps, reflects its accessibility, startling breadth, and creative
insight. While certainly worth reading, as biology it’s strange and anachronistic,
a search for a kind of geometrical perfection in nature to which evolution by
natural selection is largely irrelevant. Nonbiologists such as architects often
assume that On Growth and Form is in the mainstream of biology or
biomechanics. So I hasten to explain that Thompson is a much-beloved
godfather rather than someone whose intellectual genes we proudly carry.)
FIGURE 3.1. The size ranges of organisms and our mechanical devices. Some arbitrary judgments beg
forgiveness. The longest (if very thin) organisms are probably some multinucleate fungi, and the Great Wall
of China is vastly longer than the longest bridge. Nor are subcellular items, such as microtubules and
bacterial flagella, included.

Not only are most organisms smaller than we, but in most groups smallness
is the ancestral condition and largeness the specialization.5 Big fossils are
impressive, but little ones are more likely to lead somewhere. Nature starts
small. Organisms are basically built up from cells rather than divided into cells;
the earliest fossils are microscopic. Human technology goes the other way. Our
ships, buildings, and bridges may be larger than ever, but the factor of increase
has been small and the times involved have been long. More impressive is the
way our systems (or their parts) have gotten smaller. The first steam engines
were enormous, operating slowly and at low pressures. Jet turbines are small,
fast, high-pressure devices. Most extreme of course are electronic devices;
compare today’s microscopic semiconductor junctions within large-scale
integrated circuits with the huge vacuum tubes of the 1930s.

LENGTH, SURFACE AREA, AND VOLUME

Length, surface, and volume aren’t at all the same kind of thing. Consider a pair
of cubic boxes, as in Figure 3.2. If an edge of one is twice as long as that of the
other, then the larger one will have not twice but four times the surface area of
the smaller. At the same time the larger will have fully eight times the volume of
the smaller. Similarly, if one of the cubes has edges ten times longer than the
other, it will have a hundred times as much surface area and no less than a
thousand times as much volume. Put as a general rule, area increases as the
square of length (22 = 4; 102 = 100), while volume increases as the cube of
length (23 = 8; 103 = 1,000). The rule works for any set of similarly shaped
objects, such as spheres or (at least roughly) salmon. When things grow big,
volume increases more drastically than does surface area. Therefore, being big
means having lots of inside relative to your outside; being small means having
lots of outside relative to your inside.

FIGURE 3.2. Two cubes, one with sides twice as long as those of the other. The bigger has four times the
area, eight times the volume, but only half the surface relative to volume of the smaller.

Biological objects, whether trees, people, or bacteria, don’t vary much in

density—all are about as dense as water—so mass and weight follow the rules
for volume. A fish twice as long as another of the same shape will weigh about
eight times as much; let cooks take notice.
Thus variables that follow volume, such as weight, will increase faster than
variables that follow surface or length. The consequences aren’t trivial. For
instance, heat is generated throughout an animal’s insides but is lost across its
surface. If two animals, a large one and a small one, produced heat at the same
rate (relative to their volumes), the larger one, volume rich and surface poor,
would be warmer. But body temperature varies little among mammals and birds
of all sizes. We larger creatures simply produce less heat (relative to our
volumes). We need proportionately less food. We can get by with a thinner shell
of insulating fur, fat, or feathers. We can also walk or swim about in a cooler
climate. Being warm-blooded would have been no enormous accomplishment
for a large dinosaur but a lot more remarkable for the small mammals
contemporary with the dinosaurs. Indeed, warm-bloodedness occurs only in
animals above a few grams in mass; as a fine convergence, the smallest birds
(hummingbirds) are about the same size as the smallest mammals (shrews). Both
hummingbirds and shrews are voracious eaters, and at night both let their
temperatures drop, essentially hibernating, lest they starve before morning. For
warm-blooded aquatic animals the minimum size is still greater. Warm-
bloodedness is no small trick for tiny animals; all that surface makes trouble.
Our technology makes elaborate use of heat—for instance, in fabricating
materials. But we use large ovens and fabricate in large batches, so the actual
energy requirements aren’t that bad. For an organism only a millimeter or
centimeter across, making a hot spot either internally or externally would be far
more costly, relative to its volume. Keeping a large building heated is cheaper,
relative to its volume, than is heating a small house. Colonial bees can heat their
nests communally; no solitary insect can do so. Nor is this business of heat
exchange the only consequence of how the relationships among length, surface
area, and volume depend on size. All processes that involve exchange of
material with the surroundings are ruled by those relationships—for both better
and worse. If you’re small, getting oxygen in and out is relatively easy, even
without resort to lungs or gills, but at the same time you’re more vulnerable to
chemical assault by predator or pollution since no part of your inside is very far
from your surface.
As noted, organisms have to grow in size without serious interruption of their
functioning. That raises peculiar complications for size-dependent variables. An
adult can’t be, and in fact isn’t, just an enlarged child, as you can see from
Figure 3.3. Consider two people of ordinary corpulence. One is tall, and the
other short. (It doesn’t matter if the short one is a small adult rather than a
youngster.) The weight of each ought to follow the cube of height, at least
roughly, and the soles of the feet of each will have to bear that weight. But soles
—now we’re talking about an area. If the tall person were just an enlarged
version of the short one, then the tall one would impose more pounds (weight)
on each square inch (area) of those soles. As it happens, though, we’re more
subtly made. As investigators at the Nike Research Center found out, tall people
have disproportionately large feet, just disproportionate enough to fix the weight
per area of sole.6 Of course weight here is a kind of anticipated weight based on
the cube of height; if you get fat, your feet don’t enlarge in compensation, even
if they flatten a bit in response.

FIGURE 3.3. An adult and an infant of about five months drawn to the same apparent height, each with
head, limbs, and torso in correct proportion. To emphasize the change in shape, the baby has been given
adult posture.

Another example: A falling body falls faster and faster, until its increasing
drag just equals its weight and it gets to a final, steady velocity. The surface of a
falling object determines its drag while the volume of the object sets the weight
that draws it earthward. Since a larger body has more volume for its surface and
since weight equals drag at its final velocity, the larger body will fall faster, as
you can see from Table 3.1. Falling is mechanically hazardous for a human; it’s a
serious danger for a nestling bird only because predators may lurk below. As a
great biologist, J. B. S. Haldane, put it in an essay entitled “On Being the Right
Size,” “a mouse is uninjured, a man is broken, a horse splashes.”7 (In a bleak
book about coal mining in England, The Road to Wigan Pier, George Orwell—
who later wrote Animal Farm and 1984—wondered about how mice get into the
mines: “. . . possibly by falling down the shaft—for they say that a mouse can
fall any distance uninjured, owing to its surface area being so large relative to its
weight.”8 Common knowledge to Orwell and other British Socialists at least—
Haldane’s essay first appeared in their newspaper.)

DIAMETER FALLING SPEED

1 meter 330 m/s 738 mph
or
10 centimeter 104 233
1 centimeter 15 33.6
1 millimeter 3.6 8.1
0.1 millimeter 0.27 0.6

TABLE 3.1. Final falling speeds for spheres of water’s density at sea level. (Over this huge range one can’t
simply assume that drag varies with the square of speed. I’ve ignored transonic phenomena in the
calculations.)

In short, we big, terrestrial animals live in a gravity-dominated world. For an

animal of more ordinary size, gravity matters a lot less. For aquatic animals, it’s
of no great gravity at all.

SIZE AND FLIGHT

No mechanical feat is more impressive than flight. Nature showed us that flight
was possible, something that could not have been self-evident. But while flying
animals pointed the way to airplanes, they misled us badly on the particulars—
mainly because the practical problems of flight are strongly size-dependent. Put
another way, a flying machine’s size controls both its design and performance.9
First, a flying machine has two separate size-dependent missions, staying
aloft and making headway. For a tiny flying insect, staying aloft against gravity’s
pull is easy. But making headway against the drag of the air is tougher than for a
bird or airplane. At issue, again, is how drag and weight scale with size. To stay
aloft takes an upward force that just counterbalances weight; going forward
requires a force equal to drag at that flying speed. Weight depends on volume,
while drag depends on surface area. Halving body length reduces weight fully
eightfold while reducing drag about fourfold. Thus the smaller creature finds
weight less troublesome but drag more so. Thus it flies more slowly and finds
itself more severely affected by any wind—for better (taking advantage of air
currents) or worse (dealing with headwinds and navigational complications).
Second, a wing’s lift varies with its area, just as its drag does. Therefore,
doubling length (while keeping shape unaltered) gives a craft four times the lift
but eight times the weight, which doesn’t sound auspicious. One solution is
having really large wings on the larger craft; another is to fly somewhat faster:
Like drag, lift goes up with speed through the air. So here again, larger ought to
mean faster, roughly what we see among flying animals, from tiny insects to
large birds and larger planes. It needn’t (and doesn’t) mean very much faster,
though, since a doubling of speed increases both lift and drag, not twice but
about four times. A fruit fly might hit three miles per hour, a bumblebee can do
twelve or so, only large birds can exceed forty or fifty miles per hour, while for
airplanes that’s about the slowest they can go.
A third size-dependent factor complicates design and performance. The best
wings produce a lot of lift while suffering little drag. This relationship between
lift and drag depends slightly but significantly on size and speed, especially for
small wings traveling slowly. Lift relative to drag gets worse as the craft gets
smaller. The culprit is a property of fluids, whether liquids or gases, known as
viscosity. Put simply, viscosity is a fluid’s resistance to flowing, its internal
stickiness. Its influence gets steadily more pernicious as systems get smaller and
slower. The wing of a tiny insect moves through air a bit like a rod pushed
through thick syrup; its shape, upon which its lift depends, is considerably
obscured by the air carried with it. Very small and slow wings have more drag
relative to their lift; here nature is the designer with the harder assignment.
Consider gliders, whether living or not. The angle at which a glider descends
in still air depends almost entirely on that ratio of lift to drag; maximization of
the ratio gives the flattest, most nearly horizontal glide. Birds can’t make glides
as flat as sailplanes, simply because they’re smaller, as in Table 3.2. In practical
terms, a bird can’t glide as far from a given height. Is human technology better
since our wings have less drag for their lift than those of birds? The comparison
is so clouded by size that no simple judgment is fair or useful. As gliders, insects
are still worse off than birds, and only large ones, such as locusts and butterflies,
do much gliding at all.10
Gliding through moving air makes matters still murkier. Both human gliders
and gliding animals ride air currents to prolong and direct their flights; it’s called
soaring. So time aloft may be quite as important as the distance that might be
covered in a simple still-air glide. Time aloft depends equally on descent angle
and descent speed. Smaller generally means both steeper and slower, so while
the gliding bird may descend more steeply than the sailplane, it descends more
slowly. Thus time aloft is about the same for an eagle and a sailplane. Gliding
insects descend still more steeply but more slowly yet, and the monarch butterfly
is little worse than bird or plane.

FLIER MINIMUM GLIDE

ANGLE
Sailplane 1.5°
Small airplane, engine 3.0°
off
Albatross 3.0°
Falcon 3.5°
Pigeon 9.5°
Monarch butterfly 12°
Flies, etc. (calculated) 30°

TABLE 3.2. Least (and thus best) angles of descent for a variety of gliders.

Size affects the design of flying machines in a fourth way, a subtle one that
misled most of our early attempts to fly. Flying animals use beating wings to
produce both lift and thrust. Efficient human aircraft (which helicopters and
harrier jets are not) divide lift and thrust production between fixed wings and
propellers. Should birds emulate our separation of the two functions? Or why
can we achieve decent efficiency only by disentangling the two while birds
needn’t bother?
An aircraft must push air rearward faster than its flying speed in order to
keep going forward. At the same time it has to push air downward faster than its
ascending to keep up its ascent. But ascent speeds are tiny compared with
forward speeds; indeed, for most of any flight of bird or airplane the ascent
speed is zero. For level flight, any downward push will make lift, and the
greatest efficiency (for a reason best avoided at this point) is realized when the
largest amount of air is given the least downward speed. If the craft is going
forward, though, it must push air that’s already moving fast. Long, fixed wings
deflect a lot of air downward; short propellers spin rapidly, so pushing less air
but giving it that necessarily greater speed.11 The separation of functions buys
efficiency—smaller engines and less fuel—for airplanes. But flying animals,
being smaller, proceed more slowly and don’t encounter such rapid, oncoming
wind. Thus separating propeller and wings buys little advantage, and flapping
wings make both lift and thrust quite nicely. We’ll return to this comparison in
Chapter 10.

SURFACE TENSION AND DIFFUSION

If we humans, so large and lumbering, care little about viscosity, we care even
less about surface tension and still less about diffusion. But for the tiny insect
that accidentally touches a wing to a puddle of water, surface tension can be a
matter of life or death. Nature gives it close attention, making surfaces that either
wet easily or vigorously repel water, depending on the roles they play.
Surface tension comes from mutual attraction among the molecules of a
liquid such as water. If molecules attract each other, they prefer to cluster—
which is to say that they prefer to form the least amount of outer surface. Surface
tension works much like a bunch of people trying to get close together, perhaps
to lessen heat loss. A droplet of water tends to take the shape that gives it the
least surface for its volume. If the droplet of water rests on a surface to which its
molecules are less attracted than they are to each other, it will round up into a
flattened sphere, as, for instance, a raindrop on a well-waxed car. In an orbiting
spacecraft the droplet will be almost perfectly spherical. If, by contrast, the water
molecules are more strongly attracted to the surface than to each other, the
droplet will spread as a thin film over the surface.
We see that water rises in a thin glass tube since water and clean glass attract
each other strongly; we note that mercury drops in the same tube because of its
low attraction to glass. We commonly add a little detergent to water to reduce its
surface tension and make it give a more cleansing wash. We notice (usually
without knowing why) that an absorbent, say, a cotton ball or a cloth of natural
fiber, compresses as it dries; “absorbent” means that water adheres to it, and the
surface-minimizing tendency of water pulls the wet fibers inward as the water
evaporates and loses volume. One can use the phenomenon to make a self-
starting, small-scale siphon; a glass of water will empty if you drape a piece of
cotton fabric from the bottom of the glass up and over its rim and down into an
adjacent, lower sink. Still, such things are far from critical to our daily lives.
But for a water strider, as in Figure 3.4, surface tension is life itself. Its legs
have a waxy coating and don’t attract water; its weight depresses the water’s
surface under each leg; the water pushes upward as it tries to flatten its surface
and thus minimize its area; the water’s surface ends up depressed just enough so
that its upward force offsets the water strider’s weight. If the insect lifts two legs
off the water, the other four depress their surface dimples slightly more. In its
world surface tension is a major player; some surface insects move forward by
squirting detergent behind. What about us? Why can’t we walk on water? A
downward force, weight, has to be balanced by an upward force, surface tension
times the length of the foot-water-air contact line. The downward force is a
matter of a volume, the upward one of a length. We’re so big that we weigh too
much for feet of any manageable edge length. My sixty-kilogram mass would
require feet with eight thousand meters (five miles) of edge, but a ten- milligram
mosquito-size insect needs a mere millimeter of total foot edge.

FIGURE 3.4. A water strider standing on the surface of a pond. Notice the dimples in the water’s surface
under each leg.

The downside of being small enough to walk on water is being unable to get
through the water’s surface. That surface has about the same behavior for an
insect about a millimeter long as the canvas wall of a tent has for us. We can
dive in, we can row a boat by repeatedly dipping the oars, and we can swim
using crawl, butterfly, or backstroke; the tiny insect, though, must remain above
or below the surface.
Nor is the surface of pond or puddle the only place surface tension matters.
Water rises in the conduits of trees and evaporates from leaves. If you stop
sucking on a straw, air enters the top, and the soda goes back down, so why
doesn’t air go back into the leaves? Surface tension turns out to be critical for
keeping the air out; the pores in a leaf’s cell walls will let water evaporate out,
but they’re just too small for air to get in. Around thirty atmospheres of pressure
would be needed to pull air through the air-water boundary in pores a ten-
thousandth of a millimeter across. Again, the relevant size (here pore diameter)
is small, so surface tension can be enlisted for a major role.12
Diffusion comes into play on an even smaller scale. It’s a consequence of the
endless random and independent wandering of every molecule of every gas or
liquid; left alone, substances such as oxygen and nitrogen or fresh and salt water
mix together. The usual demonstration of molecular diffusion involves opening a
bottle of perfume in a classroom; after a short while everyone smells the aroma.
The odorant is spread, it is claimed, by that random wandering of molecules—to
diffusion.13 Not so. Except for the tiniest bit next to one’s nasal epithelium, the
perfume has been carried around by the irregular and turbulent motion of air in
the room, something totally different from random molecular motion. So
ubiquitous is such convective motion of air and water that diffusion is almost
impossible to demonstrate on a perceptually relevant scale.
But as we go from small to smaller, to subcellular dimensions, diffusion
becomes a potent agency of transport and mixing—within ourselves and all
other organisms. Impulses most often go from one nerve cell to another by the
diffusive spread of a transmitter substance. With a gap between cells of only
about one fifty-thousandth of a millimeter, the diffusional delay is about one ten-
thousandth of a second. For subcellular distances, diffusion is certainly speedy.
Indeed, almost all transport of material within animal cells takes advantage of
diffusion. But the usefulness of diffusion depends drastically on size; a tenfold
increase in distance slows diffusive transport a full hundredfold. Animals made
up of more than one or a few cells can’t ordinarily rely just on diffusion for
moving material within themselves. They have to augment it with hearts, blood
vessels, pumped lungs, digestive tubes, and other devices that force fluids to
move.14
The machines of human technology, much larger than cells, only
occasionally make use of diffusion. One blood-cleansing machine used for
dialysis of people with kidney failure relies on diffusion in and out of very tiny
pipes with a huge aggregate surface area. One famous (or infamous) process
takes advantage of the different rates of diffusion of molecules of different sizes.
The rare but fissionable uranium 235 moves faster than the common, bigger, and
nonfissionable uranium 238 as they diffuse (as gases) through a porous barrier.
Being big and impatient, we ordinarily resort to stirring and pumping to get
around the slowness of diffusion over appreciable distances, doing just what
animals do when they make circulatory systems. A cell-size creature might well
wonder why we bother.

GRAVITY AND INERTIA

We’ve now noted three phenomena—viscosity, surface tension, and diffusion—
especially important in small systems. Others—in particular, gravity and inertia
—dominate large ones. Gravity has already raised problems here. It causes large
objects to fall faster than small ones. It doesn’t let large creatures support
themselves with surface tension on water’s surface. And it makes it necessary for
large aircraft to fly faster if they’re to stay up with decent economy.
Gravity’s size-related mischief takes more subtle forms as well. Consider the
moving waves made by wind blowing across a body of water. What keeps them
wavy is the water’s inertia. What makes the water flatten out are the water’s
surface tension and weight. For ripples two-thirds of an inch or less between
crests, surface tension is the more important thing flattening the water; its
molecules pull together and minimize its surface area. For larger waves, weight
—gravity—predominates, and water’s preference for flowing downward is what
flattens the water. The shift makes big waves and small waves behave
differently. In particular, the relationship between the size of waves and the
speed at which they roll along depends on whether they’re big or small. For big
waves, bigger is faster; increase their crest-to- crest distance, or wavelength,
fourfold, and waves travel twice as fast. An ordinary boat can’t easily exceed the
speed of waves as long as its hull, so a boat four times as long can go twice as
fast before the cost of propulsion starts to rise disproportionately: Large ships go
faster than small ones, and even small ones go faster than ducks and muskrats.
But for tiny ripples, ones less than two-thirds of an inch apart, the rule is just the
opposite: Smaller is faster. The world of a minute surface boat, such as a
whirligig beetle, must be something like a freeway with small, fast sports cars
and large, slow vans.
Inertia, a property of both solids and fluids, is the tendency for something
either to remain at rest or to keep moving unless persuaded otherwise by some
external force. Put another way, to get an object going takes force, and a moving
object exerts a force when it stops. More specifically the force equals the mass
of the object times the acceleration or deceleration that alters its motion. What
matters here is that the force associated with inertia follows an object’s mass. A
massive system can exert a lot of force by stopping suddenly. Conversely, bullets
must have enormous speeds to offset their limited masses, and the lower speeds
of short-muzzle handguns are commonly offset by using heavier projectiles.
Humans have long used stone-ended clubs and metal-headed hammers, sledges,
mauls, picks, and axes—heavy things that stop abruptly. Large pieces of metal
can be shaped by dropping even larger pieces on them, something of industrial
importance for well over a century. A large animal can inflict substantial damage
on another by kicking; even a human can injure another by punching. But
inertial aggression without weaponry has little value for creatures much smaller
than we are; the most pugnacious ants don’t kick their antagonists. Even for us,
the effectiveness of a punch depends on the inertia—the mass—of its mark.
Kicking your cat is nasty; kicking a mouse is ineffectual.
Put the other way around, small things, with less mass, are easier to start and
stop—to accelerate and decelerate. As noted earlier, all of nature’s jumpers
could, if air resistance didn’t matter, achieve about the same height. That implies
that their takeoff speeds must be the same. But the short-legged flea achieves
that speed in a vastly shorter distance than does the long-legged kangaroo; the
flea’s acceleration is much greater. Bigger may mean higher speed, but bigger
also means lower acceleration, a rule of thumb that works for both living and
nonliving systems. Try catching a resting housefly in your hand! The jackrabbit
starts faster than the best racehorse and the most violent drag racer. Once started,
though, the large mover can coast better than the small one. Stopping a large ship
takes miles; a ferry must reverse its engines or it will smash the dock;
automobiles must be equipped with brakes. But a swimming microorganism will
halt almost instantly—typically in less than its body length. Its surface-
dependent drag is relatively huge; its mass-dependent inertia is trivial.
Inertia also affects how fluids flow. At small sizes, viscosity predominates,
and flows are orderly, laminar affairs. Each bit of fluid does nearly the same
thing as its neighbors. At larger sizes, inertia increasingly offsets viscosity, so the
bits of fluid tend to keep doing whatever they have been, despite any different
motion of their now more temporary associates; we call such flows, with their
chaotic eddying, turbulent. (Figure 3.5 illustrates the difference.) Flow
immediately around an aircraft or ship is inevitably turbulent; flow around a
microorganism is as assuredly laminar. (In between some tuning of transition
points may be achieved by changes of shape or surface texture.) Turbulent flow
really stirs things up; laminar flow is surprisingly ineffectual at mixing. A
microperson’s spoon wouldn’t easily stir milk into coffee. Blood flow in all but
the largest vessels of large mammals is laminar; almost all flows in industrial
and household plumbing are turbulent.

FIGURE 3.5. As speed increases, the flow of a liquid within and from a pipe shifts from being laminar
(above) to turbulent (below). The bigger the pipe, the lower the speed at which the transition occurs.

The two regimes don’t just differ in self-stirring; almost every rule for fluid
flow comes in two versions. Viscosity is a kind of internal stickiness, and it
causes small objects moving slowly through fluids to carry along a lot of the
fluid. Seeds of such plants as dandelions and milkweeds can thus descend slowly
by using a bunch of fine hairs as an analog of a parachute, as in Figure 3.6; the
hairs carry along enough air so the bunch behaves like a balloon. But the device
scales up badly since for larger sizes and speeds, viscosity becomes less
important than weight and inertia. So neither technology can use the fluffy seed
solution for slowing the descent of larger items. For slowing these, rather
curiously, the two have gone separate ways. The parachutes used by humans or
our bundles of baggage have only fairly crude analogs among terrestrial or
arboreal organisms. Nature prefers another design, that of spinning, autogyrating
seeds (fruits, strictly) of maples and other trees. While these passive autogyros
scale up satisfactorily and have been considered for use by humans, parachutes
have consistently proved handier.15 Physical reality precludes using tufts of fluff
if one is large, but it imposes no rigid choice between autogyro and parachute.

COLUMN AND BEAM

If you double the length of a column that supports a roof, how much fatter must
you make it? Even a casual look at some of the rules used by mechanical
engineers reveals still another role of size. These rules must apply equally to
designs in nature. For a simple example, we’ll consider two circular cylinders
(Figure 3.7) made of an ordinary material and carrying loads that don’t vary over
time.

FIGURE 3.6. Three ways to descend more slowly in air: the drag-increasing fibers of a dandelion seed, the
lift-producing autogyrating samara of a maple, and a drag-increasing conventional parachute.

Look first at the upright column, a cylinder supporting its own weight and
the weight of some load on top. As you might guess, failure by crushing will
happen only in a short, fat column. We’ll worry about one long and thin enough
so it fails by sudden buckling to one side or another, as when the ends of a piece
of dry spaghetti are pushed together. What does it take to start such a collapse?
The critical force varies with the fourth power of the column’s diameter divided
by the square of the column’s height. That’s a combination sufficiently hard on
the unaided intuition to demand specific numbers. What, then, would happen if
we make the column twice as big, doubling both diameter and height? The force
that starts buckling then goes up by 24 divided by 22, or 16/4, or fourfold. Swell,
a twofold size increase gives a fourfold increase in resistance to buckling.
But it’s really not at all good. If we’re being completely consistent about
doubling size, we end up increasing fully eightfold the weight of both the
column and whatever loads it. So scaling the entire system up by a factor of two
gives a column that is four times stronger, to be sure, but one that must bear
eight times the load! At best, the safety factor is halved; at worst, the column
breaks. For the larger column to serve as well as the smaller one, it must be
fatter. Worse, being fatter, it suffers still greater self-loading, which requires it to
be fatter still. At the least, the larger structure will need different proportions,
and if the differences in size are very great, it may need a stiffer material or have
to be designed differently. Large mammals have stiffer (and thus more fracture-
prone) bones than small ones. The daddy longlegs (or harvestman) walks easily
on spindly, multiply flexed legs, while the legs of the elephant are straight,
substantial columns.
FIGURE 3.7. A cylindrical column with a load on top; a similar circular cylinder supported near its ends
and loaded in the middle.

The same rule applies to a cylinder serving as a horizontal beam between

two supports. And it works whether the load acts at a single point in the middle
or uniformly over its length. For the cylinder to bend downward in the same
proportion after a doubling of size, something must be altered; its material must
be stiffer or its thickness must be more than doubled. Put another way, if the
distance between the beam’s supports, the beam’s diameter, and the length,
width, and height of the load all are doubled, then the beam will sag downward
not twice as far but four times as far. Once again, larger is weaker, relatively,
whichever technology is in charge. Thus, if the same design is used, the larger
bridge will incur a greater penalty from self-loading; scaled up sufficiently, it
will collapse from self-loading alone. Elephants are bonier than cats but still
must tread more carefully.

These examples are just that—examples, picked from a rich diversity of size-
related phenomena. But they show how severely size affects design, both
imposing constraints and affording opportunities. Gravity is important if you’re
big, diffusion if you’re small. And so on. Of particular relevance here is the fact
that the technology of people must be different from that of nature simply
because the two span different size scales. A rule for design may apply to both,
but if the rule has in it a size-related factor, the particular way it applies will
differ between the two.
 Chapter 4

SURFACES, ANGLES, AND CORNERS

Turning from size to shape, we start with things of such everyday commonality
that they normally pass unnoticed. That’s of course one of this book’s main
purposes: drawing attention to what we might otherwise overlook.

FLAT VERSUS CURVED SURFACES

We humans have a great affection for flatness. We make floors flat. And walls
and roofs and steps and desktops and paper and books. We make our roads as
flat as possible; the fancier the highway, the flatter the paving. Beneficence in
the Book of Isaiah includes “making the rough places plane”—not domed or
trough-shaped, but flat. Of course we’re not fanatically flat. Our automobiles and
airplanes have few flat surfaces. Jars, cans, and pipes are almost always
cylindrical.
Nature, on the other hand, makes very few flat surfaces. The only ones both
common and fairly large are photosynthetic structures: the leaves of many plants
and the fronds of large algae. Flatter is better for these latter, since surface area
facing skyward matters most for gathering light. Looking for flat surfaces
beyond leaves and fronds, one scrapes around among such small fry as fish
scales, bat wings, and duck feet.
First, the advantages of being flat. A floor that’s easy to walk on at any point
and in any direction will be flat—uniformly horizontal. So flatness has definite
utility in a world dominated by gravitation, which is the world of large,
terrestrial creatures like us. But the gravitational imperative can’t be the only
virtue. A wall of minimal area that separates two compartments will ordinarily
be a flat one. If you want a set of surfaces that pile smoothly on top of one
another in any alignment, flat surfaces are the easiest to make and the most
versatile in use. How would one read a hemispherical page? How would one
shelve a library of conical or dome-shaped books?
Roofs, whether horizontal or sloping, don’t have to be flat; domes and arches
have long and distinguished histories. But flat roofs are especially easy to make
and handy to use, at least if they’re tipped so water runs off. Identical, straight
beams can be laid parallel to each other, flat boards to fit on top need cutting in
only two rather than three dimensions, roofing paper can be unrolled and
fastened down with minimal fitting, and shingle or slate can be applied as a
strictly two-dimensional operation. Intersections between roof elements form
straight lines, so junction devices are simple. By contrast, intersections of
elements of vaulted or domed roofs (as in Figure 4.1) are usually curved, often
complexly so.
In short, flat is easy and convenient, and our technology capitalizes on that.
We stockpile material in a limited number of sizes and shapes; with only simple
cuts we then make a wide range of structures. The practicality of both sawmill
and paper mill rests on the flatness—and consequent versatility—of their
products.

FlGURE 4.1. Each of these intersecting barrel vaults supports the outward thrust of the other’s walls. The
resulting groin vault, used first by the Romans, makes efficient use of stone construction, but at the cost of
awkwardly curved roof intersections.
FIGURE 4.2. Wires, unless absolutely weightless, must sag between supports. A beam that’s supported at its
ends may look uncurved, but only because its thickness conceals the effective curvature; it has a virtual
hammock inside.

At the same time flatness has its disabilities. In the real world extending a
straight line indefinitely is no trivial task, unless the line is exactly vertical.
Neither a person nor a spider has trouble stretching a thread between two points.
But neither can avoid gravitational sag in the middle. For a long rope the sag is
obvious. The longer the distance to be spanned, the harder the surveyor must
pull on the measuring tape to get an accurate reading. To achieve perfect
straightness, to thwart the malevolence of gravity and self-loading, takes an
infinitely forceful pull. That of course will break thread, rope, or tape. Hang
something in the middle, and the situation grows worse, the sag more obvious.
So, as in Figure 4.2, telephone wires always sag between poles, and hammocks
droop earthward in use even if minimally swaybacked on their own.
We now see that a flat floor is something special. How might we make one?
Stretched sheets simply won’t work, and in practice we use beams—structural
members with some thickness so they can resist bending. In a very crude sense
(as the figure shows) a beam’s thickness hides a curve, a concealed sag. The
greater the load (including, of course, self-loading), the thicker must be the floor
or the horizontal beams that support it.1 The flat surface exacts a considerable
price, paid as thickness in flat roofs, sagless bookshelves, and so forth.
Bookshelves—mention of them brings up a concrete example that could
easily repay the cost of this book. A bookshelf is a beam supported at its ends
like the one in Figure 3.7. How does the sag of the shelf depend on its load and
dimensions? First, the distance the center sags follows the load. (We’ll ignore
self-loading since books greatly outweigh shelves.) Doubling the load doubles
the sag. Second, the sag varies with the cube (third power) of the length of the
shelf. With the same load, a thirty-six-inch shelf sags 73 percent farther down
than a thirty-inch-long shelf. But ordinarily the load itself increases with the
length of the shelf since the longer shelf carries more books. So as a practical
rule the sag increases with the fourth power of length, and a thirty-six-inch shelf
actually sags just over twice as far as a thirty-inch one.2
So stick with short shelves! If, however, you insist on long ones, bear in
mind that even a little extra thickness helps a lot. Sag varies inversely with the
cube of beam thickness. So the thirty-six-inch shelf need be only a little thicker
—about 40 percent—to sag no more than the thirty-inch shelf. A thickness of
one inch instead of three-quarter inch will do nicely.3 That’s not an off-the-shelf
piece of lumber, but glue and clamps don’t deter the determined bibliophile. In
any case, assiduously avoid furniture suitable solely for bric-a-brac, and load a
sample shelf before booking delivery.
(The general point—longer means weaker when something is bent— came
up in the last chapter and will reappear later. In recent months it has become a
personal matter. To illustrate the principle for a television program on size, I first
put a board between two sawhorses three feet apart and loaded it with a 140-
pound weight: me. Midpoint sag was half an inch. I then separated the sawhorses
by six feet to get the four-inch sag expected in theory and by my test the day
before. But I had inadvertently switched to a board with an oblique grain, and I
crossed the line between treading the boards and walking the plank. Fortunately
the board was the only real casualty.)
How does nature manage flatness? Dealing with the awkward problem of sag
underlies the design of leaves quite as much as that of bookshelves. Leaves
commonly stick out from branches more or less horizontally. Those veins on the
undersurfaces of many leaves, as in Figure 4.3, may look trivial, but they’re a
way to increase the functional thickness of leaves with only a little extra
investment in material. They are in fact beams, and with them, leaves sag a lot
less than they would otherwise.
A second approach effectively thickens and thus stiffens a flat surface with a
little curvature. You can’t make a flat sheet of paper stay upright when you hold
it at the bottom—unless you curve it a little, as you do every time you hold up a
page to read it. Quite a few biological surfaces get stiffness with just a little bit
of curvature. A lot of leaves have a pair of downwardly concave surfaces, one on
each side of the midrib. The southern magnolia is a good example; besides their
midribs, its leaves lack conspicuous venation. But with that little curvature (and
some thickness of the blades themselves) they form decently stiff beams.

FIGURE 4.3. Thin leaf surfaces avoid bending in various ways. Veins may provide supportive trusses (a),
the whole leaf may be cambered lengthwise (b), or pleats can make a ridge and valley self trussing system
(c).

Feathers are similarly curved on either side of their lengthwise central axes.
The curvature helps offset the pull of gravity, as with leaves. For feathers it also
serves an aerodynamic function, one particularly important for the long feathers
that form the fingerlike tips of many wings. About a century ago several people
discovered that airfoils work best if they’re curved, with their concave faces
downward.4 Early aircraft really did use curved plates for wings, and modern
planes just hide the curvature a little by covering the lower concavity with a flat
surface. Similarly, we make the blades of electric fans of thin, curved plates of
plastic or metal, getting both stiffening and better aerodynamic performance
from the curvature. Incidentally, stiffening increases whichever way a blade is
curved, but the aerodynamics improve only if the concavity is on the surface that
faces more nearly downstream. Check your fans; you’ll occasionally find one
with its blades installed backward, and it’s a breeze to reverse the blades on the
shaft.
A third way to stiffen a flat surface: Give it a set of pleats running in the
direction in which bending is expected. A piece of paper extending between two
supports will sag under self-loading alone, but if you fanfold the paper a few
times so the creases go from one support to the other, it will hold many times its
own weight. By pleating, you’re increasing effective thickness without going to
the trouble of adding proper beams beneath the surface. We use the device in
corrugated cardboard, as well as in grooved and rippled ceilings and roof panels.
Nature sometimes uses the scheme in large leaves, particularly those with veins
that radiate rather than branch.
When it comes to stiffening flat plates with a minimum of material, nothing
touches insect wings. Insects commonly invest only about 1 percent of body
mass in their wings. Yet the wings move at several meters per second through
the air, and many reverse their movement several hundred times each second. To
get sufficient stiffness for this demanding application, they combine curvature,
veins, and lengthwise pleats. Curvature may be an aerodynamic necessity in all
but very small insects, but it presents a special problem for beating wings: The
direction of curvature that’s right for a downstroke is wrong for an upstroke.
Many insects have a fine solution: Their wings are built so they’re curved by the
force of the wind that strikes them. Since the wind on the wings reverses
between upstroke and downstroke, their curvature reverses too.5
Thin, flat surfaces deflect with even small loads. The rounded shapes of our
automobiles seem made for minimizing drag and wind noise and for maximizing
sales appeal—the latter perhaps by tacit allusion to well- curved human bodies.
In fact, the roundness of cars serves primarily to stiffen them, if more attractively
than folds or ripples. Pressing a piece of metal into a curved shape is much
simpler and uses less material than spot-welding a lot of stiffening strips to a
plate. Sometimes we use a cruder fix; heating ducts are most often rectangular in
cross section, so they form huge arrays of thin, flat, metal sheets. Minor changes
in pressure or even temperature all too easily bow these sheets sideways and
produce disconcerting noises. We minimize the problem (but don’t fully fix it, at
least in my house) by putting slight diagonal creases in the flat surfaces.
One can take a more general view of the problem of flatness. Consider a
hollow sphere made of a thin material with a greater pressure inside than outside.
What pressure can the sphere withstand before bursting? In effect, what you’re
asking (as in Figure 4.4) is how much stretch—or tension— the surface can take
before you’re left with two hemispheres. What’s the relationship between the
pressure difference across the wall of the sphere and the tension generated in that
wall? Oddly enough, the amount of tension produced by a given pressure
depends on the size of the sphere. The rule, often called Laplace’s law,6 is that
the tension is equal to the pressure times one-half the radius of the sphere. For a
given pressure, the bigger the sphere, the greater is the tension stretching each
bit of its surface. Or, for a bigger sphere, less pressure is needed to get a given
tension. (Starting to blow up a large balloon is easier than starting a small one;
you need to blow less hard to get enough tension to do the job.)
One more piece of the argument: A bigger sphere has flatter, less sharply
curved walls. Thus, in general, the flatter the wall, the more tension a given
pressure difference produces. A fully flat wall would occur only on an infinitely
large sphere; any pressure difference across such a flat wall would generate an
infinite tension. That’s just the same problem as drawing a string out straight
between two supports: Any load (equivalent to pressure) and the string can’t be
drawn straight unless pulled outward with infinite force (equivalent to tension).

FIGURE 4.4. The amount of tension in the walls of thin-walled spheres or cylinders produced by a given
pressure inside depends on their size. The larger the sphere or cylinder; the greater is the resulting tension.

Among its other implications, Laplace’s law rules out making flat walls on
balloons or any other internally pressurized structures. (One can of course fake
flatness by using thick enough walls to conceal curved lines of tension, the way
we use thick floors to internalize the curvature.) It instructs us, among other
things, to make our pipes round rather than rectangular in cross section lest they
split their seams with little provocation. It also explains why a rectangular carton
of milk always (if filled) has bulgy sides. The sides must bulge, creating some
curvature, or else they’ll split. Only when packed together can full cartons have
flat walls because only then do the pressures across their walls balance out.
Aircraft fuselages are almost always cylindrical or elliptical. Grace has nothing
to do with this, and drag minimization has only a little. Pressure is what counts;
most aircraft in flight are pressurized to carry people at high altitudes, and that
rules out thin, flat walls. One common commuter craft, a Shorts, has flat walls—
it looks like a bus with wings—but since it can’t be pressurized, it’s limited to
low altitudes.7 Canisters for tea leaves can afford flat sides, but cans for beans or
soup ought to be cylindrical. Their shape doesn’t, as I once heard claimed,
represent deliberate phallic symbolism, although can and phallus do have
Laplace’s law as common denominators.
Living things are usually made of flexible materials, and both they and their
parts often have different pressures inside from outside. So Laplace’s law tells us
much about why nature abhors flat surfaces. Worms, guts, blood vessels, the
alveoli of lungs, free-living cells—all must be cylindrical, elliptical, or spherical,
indeed anything but flat-walled.
A similar rule relates the size of a dome and the load it can support. A bigger
dome is less curved and therefore relatively weaker under load. Above some
particular size, its weight alone becomes an unbearable load. Analogous rules
apply to arches, to the main cables of suspension bridges, and so forth. In each
case, the flatter the curve, the worse the tension generated by a given load, which
limits how large we can make them. Nature may not use many rigid flat surfaces,
but her stiff domes are legion; examples include eggshells, nutshells, clamshells,
and our own heads.

RIGHT ANGLES
For humans, they’re just right. In our world they’re so ubiquitous that their roles
defy short summarization: pages, desks, windows, floor and ceiling tiles, walls,
shelves and drawers, boxes, shingles, any set of straight lines encompassing the
wheels of a car or the legs of most furniture. Almost all the stock sizes of wood
in the lumberyard are rectangular solids, as are bricks and cement blocks. Both
Egyptians and Mayans made pyramids of rectangular blocks on square
foundations. That twentieth-century style of art cubism (Matisse’s derisive
name) flaunts our obsessive rectilinearity. By contrast, from tropics to Arctic,
simple dwellings are more often round: cones, domes, upright cylinders with
conical or domed roofs, and so forth. I recently visited a museum that traced tens
of thousands of years of archaeological and historical change; I noticed that
millennium by millennium right angles become better established.8 They’re
almost unfailing signatures of cultures of high technological complexity.
Nature has neither clear affection for right angles nor clear antipathy toward
them. At least one bacterium is square,9 and right angles occur between the
edges of the skeletons of certain protozoa, the foraminiferans. Where a surface is
covered with a single layer of cells, the walls or membranes that separate
individual cells form right angles with that underlying surface. Where trees grow
upward from level land, each trunk makes a right angle with the land; where the
land slopes, there’s still a right angle between trees and ultimate horizon. The
pine trees in my front yard, for instance, have almost perfectly vertical trunks.
We have to ask, then, not why nature avoids right angles but why we prefer
them. Of particular interest in this regard are the semicircular canals of our inner
ears. A set (Figure 4.5) consists of three canals, each at a right angle with the
other two. They’re important parts of the system with which we keep track of
our orientation and acceleration. Do we like right angles just because our
sensory equipment regards them as a simplest case in a geometrically complex
world?10
In the last chapter size loomed large, and it plays a major role here as well.
We’re big creatures; between our size and terrestrial habitat, we’re ruled by
gravity. You don’t tip over as long as you keep your center of gravity—the
effective location of your weight—over your feet. So you stand vertically. You
prefer to walk on horizontal surfaces, which much of the earth’s surface is
anyway. Vertical plus horizontal generates right angles, for you and the ground,
for the walls and floor of a building, for a tree and the horizon. Stacks of things
must stay close to vertical, or they’ll tip. Walls of stacked blocks are far easier to
make if upright, and the walls of large structures have over most of human
history been made from stacked blocks.
Not so for smaller structures, for dwellings light enough to be packed up and
moved, for ones easy to keep warm. For these, round and perhaps inwardly
tilting walls are better; tepees and similar conical and hemispherical houses have
been invented by numerous cultures. Anthropologists and archaeologists11 tell
us that round houses typify nomadic or seminomadic societies; curvilinear
buildings are more economical of material and easier to erect. By contrast,
rectangular houses characterize sedentary societies. They permit more buildings
in a small area—say, within a walled compound or city. Their interiors partition
more easily, and since their outer walls can serve as common walls for adjacent
structures, they’re easier to add on to. On average, the round houses of primitive
societies have less than half the floor area of the rectangular ones. A family may
occupy several round houses, with still others serving as storage structures; a
round building as a whole works like a single room of a rectangular one.

FIGURE 4.5. We have three semicircular canals (beyond the arrow) associated with our inner ears. Each
forms a plane at right angles with the others, so they occupy mutually perpendicular planes, as in the inset.

One’s view of one’s world is shaped by personal experience. For most of us

that reflects our incorrigibly rectilinear culture. A peculiar test reveals an odd
acquired bias. Back in the 1950s some psychologists tested the way two groups
of Zulu reacted to a particular visual illusion. A rural group lived in traditional
circular huts, while the other was urban, housed in conventional rectangular
dwellings. The illusion, called the Ames window or rotating trapezoidal illusion
(Figure 4.6), consists of a model of a window that’s rotated about a vertical axis
in front of a subject. Now windows are normally rectangular and not, as in the
model, trapezoidal. When conditions for depth perception are marginal (one eye
closed or the model quite distant, for instance), the urban Zulu (as do non-Zulu)
commonly perceive the model as an oscillating rectangle slanted away from the
viewer. In effect, we imagine a perspective that forces the model to fit our notion
of a properly rectangular window. The rural Zulu are much less easily fooled.
For that matter, their language lacks words for “square” or “rectangle.”12 Other
cross-cultural comparisons yield similar results. For instance, tepee-dwelling
Cree Indians of northern Canada are more resistant to the illusion than are other
Canadians.

FIGURE 4.6. An Ames window. This is not a perspective drawing.

Back to stacks. If you want to build with stacked blocks, rectangular solids
are wonderfully versatile. As suggested in connection with flat surfaces, pieces
of some shapes constrain designs more than pieces of other shapes. Curved
blocks determine a specific size of room or house; foot- long blocks curving by
ten degrees each will make a circular house 36 feet in circumference, about 11.5
feet in diameter. Uncurved rectangular blocks, by contrast, make rectangular
houses of any size and internal partitioning. Rectangular houses can be roofed
over with a set of identical beams, whether the roof is flat or sloped. Once you
have a rectangular house, rectangular furniture is positively made to order.
Rectangular books then fit nicely on rectangular bookshelves, and rectangular
drawers into rectangular cabinets. So one right angle leads to another, starting
with the decision to stack blocks.
Stacking blocks or bricks is particularly attractive for a technology that lacks
good adhesives and cables. A stack is held together by compressive forces,
blocks above pushing down on blocks below. Tensile— that is, pulling—forces
matter little. Resisting tensile forces takes adhesives, cables, or tensile joinery,
all more sophisticated than blocks or bricks. We put bricks together with mortar,
but the brick-mortar joint isn’t particularly resistant to tension. The mortar
actually has two other roles: It keeps bricks from slipping sideways across one
another, and it fills the spaces between them so bricks press evenly on one
another without the separate contact points that might make these brittle
materials crack. When building wooden houses, we join boards with nails. Like
mortar, though, nails merely keep adjacent boards from slipping sideways under
so-called shearing loads. For resisting tension, nails are hopeless; indeed, one
uses just that weakness (and a claw hammer) to pull out misplaced or bent ones.
So nailed structures are essentially stacked ones also, as you can see by
examining the frame of a house (Figure 4.7). When wooden structures must be
pulled on, we use screws and glue instead of nails, but that’s too slow and
expensive for making wooden houses. Boats face far less uniform and less
predictably compressive loads than houses, so builders of wooden boats have
had no choice but to master tensile joinery.
Furthermore, rectangular solids—such as bricks, beams, and boards—
facilitate storage and delivery. Most piles of identical items— sugar crystals, dry
seeds, gravel, nails—are unstable if their walls are steeper than some critical
angle of repose, but rectangular solids evade the problem by fitting easily into
regular, nonrandom piles. Stockpiles are simply larger rectangular solids made
up of smaller rectangular solids, with vertical walls and no internal cavities. If
one matches the larger solids with hollow boxes, also rectangular solids, one has
a fine system for packing as well as stacking. The only awkwardness in the
scheme is cultural. Having five fingers per hand, we persist in counting with a
ten- based, or decimal, system. Rectangular solids persist in packing into larger
solids in rather different arrays. Consider convenient packs or stacks made up of
individual items. You might arrange eight items two across, two deep, and two
high. Or you might arrange items in other arrays:

FIGURE 4.7. The frame of the wall of a wooden house. With real care and straight up-and- down stacking
(and no wind) it could be self-supporting without nails.

2 x 3 x 2 or 12 items 2 x 4 x 4 or 32 items
2 x 4 x 2 or 16 3 x 2 x 5 or 30
2 x 3 x 3 or 18 3 x 3 x 4 or 36
2 x 4 x 3 or 24 3 x 4 x 4 or 48

Multiples of ten aren’t prominent. Hence the continued popularity of

wholesale purchasing by the dozen and by the gross and the persistence of those
words in our decimalized world.
Nature rarely builds by stacking but rather laces things together with cables
of one kind or another. These include ligaments, muscles, and tendons in
creatures like us; membranous outer stretched sheathing in worms, caterpillars,
sea anemones, and such; and internal tensile fibers in plants. We have good
access to nature’s tension-resisting materials—we’ve long made our ropes of
natural fibers—but nature more readily attaches tendons to bones than we attach
cables to struts. Conversely, stacking isn’t so good if your materials aren’t very
dense; weight is what gives a stack decent resistance to sideways forces, such as
those that winds produce. Stacking loses its appeal if your materials are soft.
Forget about stacking low-density material underwater, where buoyancy offsets
weight and where the forces of flow get quite extreme. She may not stack much,
but nature does sometimes pack things. Still, as we’ll see, she doesn’t pack them
into rectangular boxes.
Orthogonality—right angleness—even if less useful for nature, sounds ideal
for us. “Ideal,” though, sweeps further relevant matters under the rug. If four
rigid struts are joined by hinge pins to form a foursided, two-dimensional
structure, as in Figure 4.8, the result has no set shape and is unstable. Such an
arrangement is called a mechanism—a poorly chosen term but one that at least
implies mobility. As the figure shows, mechanisms make nice mechanical
linkages. We use them for all manner of machines. Nature even uses them for
complex movements, such as those of snake jaws, where animals swallow
remarkably large and unmasticated mouthfuls,13 and fish jaws (as in the figure),
where a mouth can open and extend forward in a single motion.14
However good for eating, such sets of struts won’t support much. By
contrast, three rigid struts joined by hinge pins have a fixed and reliable shape;
such arrangements (Figure 4.9 gives a more complex one) are called statically
determined structures.15 If you frame a wooden wall with vertical and horizontal
pieces (such as the vertical studs running between the top and bottom horizontals
of Figure 4.7), you’ve unfortunately made a mechanism. Some cross bracing
must stabilize it against lateral loading; either a diagonal or two must be worked
in, or else a sheet of plywood that will remain reliably rectangular must be nailed
on. We make great use of such diagonals to create stable, rigid triangles out of
the unstable, flexible rectangles that our overall designs keep generating. And
we’ve been doing it for a long time. I was amused to see just this in a Bronze
Age rock carving of some sort of sled or boat, near the west coast of Sweden,
shown in Figure 4.10. A less aesthetic example is the scoreboard of the football
stadium of my own campus, shown in the same figure.

FIGURE 4.8. Mechanisms. An especially simple one and two views of some (not all!) of the stiff elements of
a complicated, biological one—the scheme by which a fish (a wrasse), on approaching prey, suddenly
protrudes both upper and lower jaws.
FIGURE 4.9. A statically determined structure. This one is nonredundant. Remove any pin, and it becomes
a mechanism, as you can check with cardboard and straight pins.

Nature doesn’t use triangles much—partly because she rarely goes in for real
stiffness and partly because she gains what stiffness she uses in other ways. A
few trusses formed of triangles, though, are recognizable—inside the wing bones
of large birds, for instance. I know of no case in which rectangles are, as an
afterthought, transmogrified into triangles with cross bracing. Not that natural
afterthought is unthinkable. Nature is forever adding a little here and a little there
to make an existing structure serve some new function. That’s the point of Steve
Gould’s fine essay about the thumb of the panda; the leading popular expositor
of evolution describes how pandas make functional thumbs by dividing the first
digit on each hand to give an apparent sixth finger, one unrelated to our own
thumbs.16
FIGURE 4.10. A contemporary structure in which the critical cross bracing looks like an afterthought, and
a shallow carving (painted for contrast) of a cross-braced structure. The latter, from the Bronze Age, is on a
rocky outcrop near the west coast of Sweden.

Since four end-to-end struts are nicely flexible, one might imagine that a
construction system with no predilection toward stiffness would prefer
quadrilateral rather than trilateral arrangements of struts, but that’s not quite the
case. Most sponges have flexible skeletons made of stiff struts with their ends
interconnected by flexible pads of protein, as in Figure 4.11. I once looked at a
lot of sponge skeletons in search of triangles, geodesics, and other efficient
strutting and trussing arrangements. None was obvious, and I realized that I was
looking at sponges with my human bias toward stiff structures. I then looked for
quadrilateral strutting but came up almost as dry; I had just brought to bear
another human bias. While the tiny struts (spicules) never formed triangles,
quadrilaterals were no more common than five-, six-, and seven-sided arrays.
The greater the number of struts hitched end to end in each loop, the greater the
overall flexibility that’s possible.
FIGURE 4.11. A sponge (left) and the skeleton of a sponge (right). The skeleton is a meshwork of protein in
which stiff struts are embedded. (Natural bath sponges are the supportive systems of animals that have the
protein meshwork but, unusually lack the stiff and scratchy struts.)

Yet another problem of a world of right angles. For a chair or table that sits
on the ground, stability minimally requires, first, three points of ground contact
that form the corners of a triangle and, second, a center of gravity somewhere
above that triangle. Four legs would appear better; the location of the center of
gravity will be less constrained, so load shifting will be less likely to make the
structure tip over. A four-legged chair tips less readily than a three-legged one.
Furthermore, a little redundancy sounds attractive. The fourth leg, though, makes
subtle trouble. With three, leg length isn’t critical, and the structure will behave
itself on almost any floor. That’s why we mount cameras and telescopes on
tripods. Making simultaneous contact with all four means that floor and legs
must be carefully adjusted to fit each other or that one or the other must be a bit
flexible. Our ordinary quadrupedal tables often come with height-tuning screws
on their legs to minimize wobble on uneven floors. Tripedal tables need no such
adjustability.
Cats, camels, and crocodiles use four legs, but analogy with tables isn’t
appropriate. For one thing, jointed legs are naturally adjustable in length. More
important, a quadruped with one leg lifted forms a nicely stable tripod. Watch a
cat slowly stalking; it keeps a leg off the ground for long periods. (That’s a lot
harder for people, who wobble a bit when standing monopedally.) But four legs
may be too few for easiest walking since only one can be lifted at a time without
losing stability. Most insects walk on six legs, so they can use their legs as
alternating tripods—one on one side and two on the other—without loss of
stability. That looks like a better arrangement.17 Six looks like the optimal
number of legs for a legged, walking biomimetic vehicle, about which more in
Chapter 13.18
But we’ve strayed a bit from right angles, to which we now return. On
reasonably flat terrain, we divide land into rectangular (and often square) plots,
whether eighth-acre house sites or state and national boundaries. By far the
largest number of arbitrary surveyed borders between states in the United States
and provinces in Canada run east-west or north-south and thus intersect at right
angles. Again, no rule requires it. Some diagonals have been used, and one
interstate border (between Delaware and Pennsylvania) is a surveyed circular
arc.
Simplicity suggests rectangular surveying. But it’s not uniquely easy or
automatically virtuous. Equilateral triangles divide a surface without gaps, and
these can be laid out by drawing taut a set of strings of equal length. Since no
angles need to be measured or divided, it may be the simplest surveying method
of all. Its main drawback is the large amount of boundary of its plots relative to
their enclosed areas. Regular hexagons, in which all six sides are of the same
length, also divide a surface without gaps. As an amalgam of six equilateral
triangles, a hexagon is no more difficult to lay out. For boundary relative to area,
hexagons are as good as can be done. And they have another advantage if our
whole spherical planet has to be divided. With the addition of twenty pentagons
to make the system close on itself, the earth can be partitioned into hexagons
without any distortion from its curvature. (Wyoming looks rectangular, but its
northern border is shorter than its southern one.) But hexagonal division
precludes running unidirectional roads along sequential boundaries. Neither
triangles nor hexagons give the farmer fields that can be planted in rows of equal
length. And neither facilitates arranging rectangular structures. Of course
buildings don’t have to be rectangular. A behavioral laboratory near where I live
uses a hexagonal room surrounded by six others. An observer in the middle of
the central room can keep watch on every bit of five others (the sixth provides
access). Such reasonable schemes are merely unfamiliar and less than
convenient to make from standard components. Figure 4.12 shows a street plan
with the predominant elements hexagonal; to my eye it looks as functional as
what we usually do.
We’ve clearly preferred to use rectangular plots of land for a very long time.
While the ancient Egyptians used triangles of stretched ropes for surveying, the
sides of their triangles weren’t equally long but had a 3:4:5 ratio. A 3:4:5 triangle
has a right angle between the shorter sides, so they could keep their corners
square. Nor were the Egyptians alone in that preference. The great Pythagorean
theorem applies to a 3:4:5 triangle or any other triangle with a right angle: The
squares of the short sides add up to the square of the long side, as, for instance,
32 + 42 = 52. (Or 9 + 16 = 25.) Only people concerned with right angles should
care about that subtle theorem. It was known (and quite likely was independently
discovered) in ancient China, India, and the Middle East well before the formal
proof that we attribute to the school of Pythagoras.19 Some logical person once
suggested that we mark a bare place on the earth (such as the Sahara) with a
pattern that demonstrates the theorem, the one shown in Figure 4.13. Anyone
looking at the earth would then know that it harbored intelligent life.

FIGURE 4.12. A street plan that makes substantial use of hexagonal elements. Going “around the block” is
a more literal matter in this arrangement.

Back to hexagons. Consider how a group of animals might divide a habitat

into individual territories. Our use of squares and rectangles certainly excels for
plowing and road building. But other animals do little of either. If a border to be
defended must be minimized, if all area is part of one or another territory, if all
area is equally good to live in, and if individuals are equally effective in
establishing and defending territories, then territories should be hexagonal. Even
with such stringent requirements, natural selection occasionally finds the logic
sufficient. Partitioning that closely approaches hexagonal has been found among
sandpipers in the tundra, terns on the barrier islands off North Carolina, and
bottom-living African cichlid fish in a breeding tank.20 Of course the most
famous cases of hexagonal partitioning in nature are the honeycombs and larval
cells of bees and wasps. We make occasional use of the arrangement for internal
stiffening partitions in hollow doors and for stiffening beamwork under the
floors of airplanes. The latter is no minor matter. Flatness costs material, and
aircraft designers are properly fanatical about weight economy. And the stress—
force per unit area—of a stiletto heel on a floor is remarkably high. Thin floors
braced with aluminum honeycomb puncture a lot less easily than if they were
braced with well-spaced beams.

FIGURE 4.13. A geometric representation of the Pythagorean theorem. If the angle between sides a and b is
ninety degrees, then area C is exactly as big as A and B combined.

Just as squares don’t give the best edge-minimizing partitioning of area, so

cubes don’t give the best surface-minimizing partitioning of volume. The best
geometric solid for such close packing is distinctly arcane. It’s a fourteen-sided
figure, one made with eight triangular faces (a regular octahedron) that then has
its six apexes cut off so each of those faces beomes a hexagon. For most of us
that’s beyond imagining, so Figure 4.14 is a necessity. Anyhow, such solids have
eight hexagonal faces and six square faces (yes, right angles do appear), and
mirabile dictu, they actually pack with no voids. Close approximations of them
can be made by compressing a container filled with identical lead shot until the
shot deforms enough to squeeze out all the air.21 So where in nature might such
a shape occur? The thin-walled cells that fill the middles of the stems of many
herbaceous (nonwoody) plants approach that ideal shape, with an average of
about fourteen faces on each.22 I know of no specific use in human technology
of this strange truncated octahedron, although it probably occurs fortuitously
where large pieces of foam are made from small beads of the stuff.

FIGURE 4.14. An eight-sided regular solid called an octahedron (a). Cutting off each of its six apexes
produces a fourteen-sided figure (b), a cuboctahedron or (gulp!) an orthotetrakaidec- ahedron. If the cuts
are done just deep enough so all edges are again the same length, then (c) the cuboctahedrons will pack
together without any spaces between. They turn out to do so with less surface area than any other volume-
partitioning, close-packing shape.

Nor are right angles best for making shells of struts or flat panels; the
geodesic domes of R. Buckminster Fuller are much more efficient ways to invest
material. As in Figure 4.15, a series of struts (or intersections between panels) in
such structures follows what, on a globe, we call great circle routes. Nowhere do
they intersect at right angles. I once built several geodesic domes out of metal
tubing. They worked well as jungle gyms for climbing—they were light in
weight, were stable under load, and needed no anchorage—but when I tried to
design a geodesic cabin— something more than bare metal strutting—I
discovered their impracticality. In a culture of rectangular lumber and plywood
sheets, the labor needed to make a satisfactory geodesic structure is excessive,
and wasted material ordinarily offsets its efficiency.
FIGURE 4.15. Geodesic domes are derivatives of icosahedrons, figures with twenty triangular faces and
twelve points at which five edges converge. The dome in the middle divides each face into nine smaller
triangles (notice the five-strutted junctions retained from the original icosahedron). A viral shell of
equivalent form (such as the cowpea chlorotic mottle virus on the right) has 180 protein molecules; the old
icosahedral apexes are replaced by rings of five and the old faces by rings of six.

Organisms don’t make much use of geodesic domes either. Perhaps that’s
because they make unstrutted, smooth domes when they want such regular
things. Perhaps they rarely need hollow, regular, truncated spheres. One
collection of geodesics is usually cited: the protein coats of spherical viruses.
Identical protein molecules combine in such coats in just the same numbers—80,
180, and 320—as the number of faces in full-sphere geodesic structures. I think,
though, that neither mechanical strength nor material economy explains their
occurrence. More likely we’re glimpsing the shortage of information brought up
in Chapter 2. These structures permit a virus to cover itself with multiple copies
of a single protein, with each molecule of that protein fitting into an equivalent
position, something that can self-assemble with minimal complication.

CORNERS AND CRACKS

When two components come together, they usually form a corner. We’ve been
asking, in effect, whether such a corner should form a right angle. Now let’s look
at the corners themselves. Here once again nature and human technology have
different preferences. We build things with sharp corners—unless good reason
dictates otherwise. Nature builds things with rounded (faired) corners—again
unless some functional imperative supervenes. The underlying questions here are
simple. First, what’s wrong with sharp corners? Second, for human technology at
least, where will sharp corners actually prove intolerable?
Answers to both questions flow from a third. In a stiff structure under load,
where will the first crack appear? We all know the answer to that one: where
there’s the most force on the structure. Straighten out a bend or corner in a brittle
bar, and any crack will start on the inside of bend or corner. Moreover, starting
the crack takes a much lower force than you’d need to start one in a straight bar.
The effect doesn’t depend on some weakness caused when the bar was bent in
the first place; a bar cut with a bend from a larger piece of material or one cast
with a bend will behave in the same way. The inside of a bend is apparently a
naturally weak place. What determines how much the bar is weakened by the
bend is not so much the overall angle of the bend as, oddly enough, the
sharpness of the inside of that angle.23 Making a bend deliberately less sharp,
fairing it, can gain a lot of strength.
To get a quick look at the effect of fairing the inside of a corner, cut out a flat
strip of aluminum foil, about an inch wide, with an angle of about sixty degrees
at its midpoint—a roughly boomerang-shaped piece. Pull on the ends with it flat
on a table, and it tears in two, beginning at the inside of the angle. Make another
strip, using a paper punch or a carefully rounded cut to get a less sharp inside
angle, and pull again. The force required to tear the second is probably greater;
fracture is foiled. Doing the same thing with plastic wrap may (depending on the
kind of plastic) be even more dramatic; the rounded corner may prevent tearing
altogether. Force concentration—that’s what’s wrong with sharp corners. The
stiffer the material, the worse the problem becomes. We use the phenomenon to
cut glass. Scratching the glass makes a kind of corner or at least a place where
force is concentrated and from which a crack will predictably propagate. We can
also get cloth to tear where we want it by making an initial slit, at least if the
cloth isn’t too stretchy. Tearing then perpetuates the slit, so the process easily
continues.
Worse even than a sharp corner is a sharp corner at which two pieces of solid
material are joined together, as at the corners of a picture frame. To the intrinsic
fragility of the corner are added all the difficulties of applying fastenings that
resist tensile loads; straightening a corner stretches its inside edges. But that’s
just how we build window and door frames, wooden boxes, drawers, furniture
with legs, and so forth. You know from experience where such things fail—at
the joints and usually at their insides. We know that making joined corners is a
bad practice, but it’s ever so convenient for construction!
Rounding corners helps, as you perhaps saw with the foil. Our technology
does just that, but usually only where mandatory. I once broke an interior door
handle of a car; examination showed a sharp inside corner in the bit of cast metal
within its plastic trim. In the “exact factory replacement part” that corner was
faired; someone had wised up. Long ago the teeth of gears were found less prone
to break off if the bottoms of the gaps between the teeth were rounded rather
than sharp-cornered. The portholes of a ship are always round to avoid initiating
cracks when waves stress the ship’s hull. We deliberately use windows with
rounded corners on aircraft. An airplane’s windows are gaps in the skin of the
craft, and the skin is part of the mechanical structure, not just some covering that
just keeps people and air inside and wind and weather outside. The claw of a
decent hammer has a bit of fairing where it connects to the rest of the head. If we
make two pieces come together at a corner, we often round the welded joint or
attach a third piece that rounds or cross-braces the corner. And so on.
Nature’s structures, more limber, must be less prone to cracking at corners.
In addition, natural structures only rarely use separate pieces rigidly joined at
corners. Most often we find single pieces of material, grown from a single site,
with fairing incorporated into the basic element rather than added adventitiously.
The key here, I think, is the growth process. Whatever the complications it
entails, growth greatly simplifies the construction of faired corners on single
structural elements. Figure 4.16 gives a pair of examples. The long ridge on a
mammalian scapula (shoulder blade) is nicely faired with the rest of the bone.
The branch of a tree grows smoothly out from the trunk with wood fibers
running across the upper (functionally the inside) angle of the joint.24 One of
my less smart acts was to hit a wedge with a sledge into such a crotch in a piece
of oak. Smashed eyeglasses a fraction of a second later gave real impact to a fair
lesson about fairing, a lesson reinforced by a small scar on my brow. Where
nature makes a sharply angled junction between pieces, the pieces usually form
part of a mobile mechanism, as where the forearm attaches to the upper arm.
There, of course, cracking isn’t an issue.
We’re gradually making more things from single pieces of material with
nicely rounded internal and external corners. That we’re deliberately copying
nature is unlikely; rather we’re taking advantage of plastics that can be cast or
molded into more complex forms than we can get by stamping things out of
metal sheet. The curves not only make the products easier on shins and simpler
to clean but, by reducing force concentrations and lessening the problem of
crack initiation, also permit useful economies of weight and material. Garden
carts made mainly from single plastic pieces can have far more intricate shapes
than wheelbarrows with stamped pans. Fiberglass shower-baths share the same
advantages over metal tubs and enclosures.
FIGURE 4.16. Faired corners in nature: the scapula, or shoulder blade, of a cat and a branch forking from
another in a tree.

Three contrasts in mechanical design between nature and human technology are
evident even at this geometric level. We build flat, nature builds curved; we hold
right angles dear, nature is untouched by such affection; our corners are sharp,
hers are rounded. Even in these most ordinary matters, we see pervasive
differences between the two technologies. Of greater importance, we see
repeated examples of the insights available if one avoids premature judgment of
superiority and inferiority. Of still greater importance, we begin to see how each
technology forms a separate, well- integrated entity, operating in an internally
coherent context.
 Chapter 5

THE STIFF AND THE SOFT

Natural and human technologies diverge as much in the materials they use as
in their shapes. We prefer stiff and brittle materials while nature opts for pliancy.
An airplane’s shape at the gate differs little from its shape in flight, but a leaf in
still air looks nothing like one in a storm. The difference takes us into materials
science, a field both interesting and immediately relevant, one deserving more
attention than it gets from us nonengineers. Blame it, if you wish, on how we’re
taught about the physical world. Elementary physics classes assume that masses
act as if concentrated at points and that solid bodies are perfectly rigid. These
fine abstractions, useful fictions, prove inadequate for the present story. If,
horrors, an object is flexible, then forces acting on it may change its shape and
the center of its mass, complicating things considerably. Worse yet, the changed
shape may then incur different forces. Such analytical complications, though,
represent our problem, not nature’s. As we’ll see, flexibility provides all manner
of opportunities for clever design.
We need special tools with which to describe this more complex world.
“Solid” will mean only that something isn’t fluid, that it doesn’t flow. Forget
perfect rigidity; no solid is perfectly rigid. Even the formal test for telling a solid
from a fluid presumes that solids give at least a little. Forcibly distort (without
breaking) a solid object, and it snaps back when the force is removed; distort a
fluid similarly, and it obligingly adjusts its shape. The jelled salad, a soft solid,
retains the shape of its mold; by contrast, the coffee fits any cup.
Furthermore, what usually matters is not force but stress, force divided by
the area over which it acts. Even a small force will push a needle into a hard
material; the needle is sharp, so the effective area of action is minute, and the
small force creates the large stress that does the job. Likewise, a sharp knife
penetrates better than a dull one when both are pressed with the same force; the
sharp one exerts a greater stress on steak or string bean.

MATERIAL EVIDENCE
A solid object stretches when you pull on it. If stone or bone, it won’t stretch
visibly, but it will certainly stretch measurably.1 You might plot data for a
gradual stretch on a graph where distance stretched runs horizontally and force
runs vertically, as on the left in Figure 5.1. The graph, though, describes only
your particular object. More useful is a graph relating force and stretch, not for a
specific object but for a kind of material—for stone or bone, Vermont granite or
cow femur. To achieve that generality, instead of force we use stress—pulling
force divided by the cross-sectional area of what we’re pulling on, once again.
Instead of distance stretched, we divide stretch by the original length of the
object and call the result strain, or fractional elongation. Then, as on the right in
Figure 5.1, we plot stress upward and strain across. In materials science, “stress”
and “strain” are far from synonymous. Anyway, odious nomenclature aside, we
now have a graph that applies not just to our object but to a material. A tensile
stress of a million newtons per square meter (abbreviated MN/m2, about 150
pounds per square inch) might give a strain of 0.1, or 10 percent; that’s the value
for one kind of rubber.2

FIGURE 5.1. A graph of force against stretch and the same converted to stress against strain. Strain is
unitless; it’s extension divided by original length or (multiplying by 100) percent extension. Stress is usually
given in newtons per square meter or (as here) meganewtons per square meter. A newton is a force about
equal to the weight of an apple.

With nonrigid solids, the world gets still more complicated; a rubber band
and a lump of taffy lack rigidity in obviously different ways. The stress-strain
graph of Figure 5.2 may help the reader as we make our way through a ménage
of relevant properties of materials. One such property is how forcefully a
material can be stretched before it breaks—the maximum stress that it can
withstand. That we call its strength. A second is how far the material can be
stretched before it breaks—the maximum strain it can withstand. That’s usually
known as its extensibility. On the graph, these are the maximum height of the
plotted line and the maximum distance it extends to the right. For a lot of
materials (many metals, for instance) something else complicates extensibility. If
extended up to a certain point, these materials snap back properly, but if
extended farther, they stretch and stay stretched. Such materials have an elastic
limit, and that may be the extensibility that matters in use. The elastic limit isn’t
limited to stretching; a bar bent beyond the elastic limit of its material stays bent.

FIGURE 5.2. Four material properties that can be extracted from a stress-strain graph: strength,
extensibility, stiffness, and work of extension.
 Beyond strength and extensibility we encounter a third property: stiffness.
Pull on something, and ask how much pull (stress) it takes to get a given stretch
(strain). That’s stiffness—stress divided by strain. Rubber is easy to stretch,
which is to say that it has a low stiffness. For the rubber just mentioned, a stress
of 1 MN/m2 divided by a strain of 0.1 gives a stiffness of 10 MN/m2. Steel,
vastly stiffer, is harder to stretch by the same amount. One kind demands a stress
of 20,000 MN/m2 (3 million pounds per square inch) to stretch by the same 10
percent and so has a stiffness of 200,000 MN/m2. Engineers call stiffness
Young’s modulus of elasticity after Thomas Young (1773–1829), one of the
people who struggled to apply Newtonian mechanics to real materials and to
whom, incidentally, we owe our notion of energy.3 But “stiffness” will do for us
since it corresponds quite closely to both our perceptions and our common
speech. On the graph, stiffness appears as the slope (the steepness) of the plotted
line.
Another slight complication: For many materials, stiffness depends on how
hard or far they’re stretched; a material may be hard to start or resistant to the
last bit of stretch before it breaks. On the graph, its stress- strain line will be
curved rather than straight. As it happens, the materials we humans most
commonly use—metals, especially—give fairly straight lines, so we’re
accustomed to quoting specific values for the stiffnesses of specific materials.
Almost all biological materials, though, have dramatically curved stress-strain
lines, sometimes curved upward and sometimes curved downward. On the one
hand, values for their stiffnesses either are only rough averages or else apply
only to specific conditions. On the other hand, the complications permit great
flexibility (to pun slightly) in how different materials can be used. For instance,
as in Figure 5.3, the material of our tendons gives a curve that’s concave
downward with a lot of area underneath it. By contrast, the material of our
arterial walls gives a curve concave upward with a relatively small area
underneath.
FIGURE 5.3. Stress-strain curves for a tendon (data from Alexander, 1988) and for a large blood vessel
(data from Shadwick, 1994).

Those areas underneath the curves represent a fourth property. Stretching

something takes energy, and stretching something thicker or stretching
something farther takes more energy; everyday experience does not mislead. We
use “work of extension” as a measure of that energy (strictly, energy per unit
volume) needed to stretch a material. On the graph that’s the area under the
stress-strain line, from zero out to the material’s breaking point. Were all stress-
strain lines straight, this wouldn’t be a particularly interesting property. But as
just noted, nature’s materials give diverse curves whose shapes are fraught with
functionality. “Work of extension” is sometimes called toughness,4 an agreeably
intuitive term that we’ll find quite useful.
And another, property number five—resilience—what rubber has that taffy
lacks. Say you keep track of the work of extension while stretching an object
(such as a rubber band or steel spring) to something short of the breaking point.
Then you release the object and keep track of the work or energy it returns. In
our imperfect world the two won’t be equal. The work going in as you stretch it
will always be more than what comes out when it’s released.5 The work you get
back divided by the work you’ve done is the material’s resilience, a property
alluded to several chapters ago when we considered the hinge pads of insect
wings and the hinge ligamerits of scallops. On the stress-strain graph it comes
from two areas, the one representing the work obtained during release divided by
the one representing the work done in the stretch.
At this point let’s summarize in a list (with some nonbiological examples) for
quick reference farther along:

MATERIAL PROPERTIES THAT LOW HIGH

MATTER
Strength stress (force per area) jelly steel cable
needed to break a material
Extensibility, strain (relative stretch) brick rubber band
needed to break a material
Stiffness: stress divided by strain as a soft plastic brick
material is stretched
Work of extension, toughness: energy cast iron spring steel
to stretch a material to breaking
Resilience, energy output upon taffy rubber band
release divided by input during
stretch

WHICH PROPERTIES MATTER WHEN?

With five properties of materials rattling around,6 superiority or inferiority can’t
be judged on a single scale. A lot of talk goes on about the wondrousness of
biological materials. Much of it is simplistic blather; one has to consider the
suitability of materials for specific applications. Even here, though, a little
mental extensibility is in order since almost any task can be done in a variety of
ways. Some examples ought to flesh out these points.
We fashion tension-resisting things out of many materials: steel and other
metals; natural fibers, such as jute, sisal, manila, and hemp; synthetic polymers,
such as nylon, polyethylene, and polypropylene. We also make them in many
forms: chains, bands, and bars as well as twisted and braided ropes and cables.
Their task seems specific enough: to resist pulling (tensile) forces. Making them
best should be simply a matter of maximizing strength relative to either weight
or cost. But no, the business is more subtle and multifaceted.
Consider stiffness. If you anchor a boat with a line made of unstretchy stuff,
you’ll get into trouble. If the boat moves about its anchorage, sooner or later the
line will come taut and try to stop it. How much force will that take? According
to the best authority, Sir Isaac Newton, the force will equal the large mass of the
boat times the deceleration of the boat. If the line comes taut at a very specific
length—that is, if it has a high stiffness— the boat will stop abruptly, with a lot
of deceleration and thus a lot of force. That might break even a hefty anchor line,
or worse, it might rip off part of the boat or dock. Much better to use a stretchier,
less stiff rope, one that stops the boat less abruptly and thus with a lower force.
Not only is damage less likely, but a weaker rope will do; using a material of
lower stiffness allows using one of lower strength! (Similarly, a longer mooring
line may be better than a short one. Extensibility may be the same for long and
short, but the actual extension will increase with the length of the line.) In other
applications, stiffness is clearly a virtue. Both bicycle chains and the belts on
your car’s engine work by resisting tension, and the less they stretch in the
process, the better. Flex, yes; stretch, no.
The same argument applies to living systems. A large, grazing mammal has a
ligament that runs from a ridge on the back of its skull, beneath the skin on the
back of the neck, to the bony upward extensions of the vertebrae of its neck and
chest (Figure 5.4). That ligament has fairly low stiffness—it’s stretchy—which
helps keep a heavy head from being badly shaken or the ligament from getting
torn loose when the animal trots along. If you buy an unsliced lamb’s neck (a
good source of chunks of meat for skewering), you can dissect out that ligament
and feel its elasticity. By contrast, tendon, connecting muscle to bone, has to be
stiffer stuff. A muscle works by shortening, usually pulling two bones closer
together. Any stretch of its tendons means that the bones move that much less.
The shank end of a leg of lamb has a nice long tendon that you can remove and
pull on to get a feel for its resistance to stretch. Its response contrasts sharply
with that of a neck ligament.
Through the 1800s natural technology far outdid humans in making unstiff
tensile elements—that is, stretchy ropes. We humans did play some tricks with
our stiff ones, creating what might be called virtual elasticity. One can, for
instance, use a rule mentioned earlier about not being able to draw a string out to
a perfectly horizontal line. Try pulling on a heavy chain that sags in the middle;
you have to pull harder and harder as the slack is diminished. The chain doesn’t
stretch, but its weight makes the two ends move apart as if it did. If you must
moor your boat with a chain, use a heavy one, or else hang a weight from the
chain’s midpoint. These days, of course, we routinely make stretchy stuff, such
as rubber7 and other polymeric materials. We’ve converged with nature not only
in making such materials but in what we make them from; even our fully
synthetic ones are mostly the products of carbon-based organic chemistry.

FIGURE 5.4. The approximate location of the nuchal ligament that supports the head of a grazing animal,
such as this horse, and the hamstring on the rear of the hind leg roughly equivalent to our Achilles tendon,
which connects the big muscle on the rear of the calf to the heel.

For resilience as well as stiffness, what’s best depends on the application. If

one wants a good hinge ligament for a swimming scallop, then high resilience is
important. In clapping its half shells together—to expel water in a pair of jets—a
scallop contracts a large (and tasty) muscle. What reopens the shell is the elastic
recoil of the hinge ligament; in shortening, the muscle not only has closed the
half shells but has strained the ligament, giving it the energy it will then use for
reopening. The higher the resilience, the less of that energy is wasted.
Conversely, high resilience may be quite undesirable. For a spider to catch
insects that fly into its web, the prey must remain attached to the silk. A web of
high resilience would, like a trampoline, tend to fling the prey back out. Best
would be a web made of silk with high strength, extensibility, and toughness but
low stiffness and resilience. Spider silk8 fits this peculiar bill superbly, compared
with the other materials of both technologies. But the requirements are unusual,
and what’s noteworthy is the match of material and application.
Resilience management highlights the dichotomy between nature and human
technologies. We don’t do well at making materials that extend a long ways and
then return to their original lengths with low resilience— materials like spider
silk. To do so, a material must turn into heat most of the energy put in, but it
mustn’t flow (as a fluid) or get self-destructively hot in the process. Our springs,
whether of steel or rubber, are just that: nicely springy. Our normal way to get
lower resilience, as in Figure 5.5, is to use additional mechanical elements in
parallel with springs, so-called shock absorbers, as found alongside or within the
springs of our cars and as parts of door closers. These devices, properly called
dampers, have no preferred length and no resilience at all. Like springs, they
resist any stretch or crunch; unlike springs, they don’t spring back, converting
mechanical energy into heat instead of storing it elastically.
Keeping resilience and rebound low is equally important in nature; the orb
web of a spider is only an extreme example. For instance, for a tree to sway in
the wind with too much resilience—too little damping— wouldn’t be at all
handy. The magnitude of sway would increase, and the roots would gradually
work loose. Nonetheless, nature doesn’t limit resilience with anything quite like
our shock absorbers. Instead, she commonly provides damping right in her basic
materials, such as the wood of trees. Materials of low resilience predominate in
nature. High resilience is a sign of something special going on, such as
reversible energy storage (as in hinge pads and ligaments) or deliberate flow-
induced shaking (as in our vocal cords and the seed-flinging equipment of some
plants).

FIGURE 5.5. The shock absorber of an automobile, consisting of a cylinder with a leaky piston inside, and
a spider web of silk in which a single material provides both elasticity and (through low resilience)
damping.

Animals often offset unwanted motion with schemes much fancier than
simple damping: with sensors, feedback circuits, and contracting muscles.
You’re more stable when standing upright than any department store mannequin.
Why? Your base of support is equally narrow, and your tendons are highly
resilient. But if you shift your center of gravity slightly outward from your base
of support, sensors in your tendons and muscles detect the shift, and your brain
then tells a host of muscles to make slight adjustments in their lengths.
(“Postural reflexes” is the general subject in textbooks of physiology and
neurobiology.) Only occasionally does human technology achieve such
sophistication. Large airplanes, for instance, use the same kind of active controls
to reduce the bumpiness of flying through irregularly moving air. You may
notice flaps on the rear of the wings that seem to wander slightly but
disconcertingly up and down as the plane flies along. They’re computer-
controlled and not under the pilot’s direct command; no pilot could react fast
enough as the plane whizzes along. One hears talk of active automobile
suspensions, systems that would detect a bump just before a wheel hit it and then
extend or retract the wheel as needed.
No material better illustrates the complex diversity of mechanical properties
than wood. To most of us it’s ordinary stuff, with nothing more than superior and
inferior varieties. But specifying quality demands that we look at the match of its
wide range of mechanical properties with an equally wide range of applications.
To identify our local trees, I use a book that mentions the traditional uses of the
wood of each species.9 The specificity may represent tradition, but the tradition
comes from experience, not arbitrary ritual. White pine makes matchsticks, red
spruce makes canoe paddles and the sounding boards of pianos, black willow
makes artificial limbs, black walnut makes airplane propellers, white oak makes
whiskey barrels, American beech makes clothespins and spools, Osage orange
makes bows, yellow poplar makes barrel bungs, lignum vitae makes pulleys, and
basswood makes drawing boards. That’s in addition to structural timbers, for
which yellow pine is notable among the softwoods and red oak among the
hardwoods. These things were sorted out by generations of craftspeople,
ignorant of the arcana of stress-strain graphs but finely attuned to the practical
behavior of what they had available.

INTERRELATIONSHIPS AND CONSEQUENCES

Spider silk is extensible, strong, and tough but not particularly stiff or resilient.
Collagenous tendon is stiff, tough, and resilient but not very extensible or strong.
The mechanical properties we’re worrying about may be defined quite distinctly,
but they don’t necessarily go their separate ways in the real materials available to
the two technologies. In short, the full range of properties in all combinations
isn’t available to either technology. Put another way, seeking a material with a
desirable value of one property limits one’s choice of values of the other
properties. So a design that capitalizes on advantageous characteristics of a
particular material must also deal with that material’s less handy properties.
More generally, if two technologies use different basic materials—say, concrete
versus fresh wood—many other differences will follow.
Consider stiffness, strength, and toughness. Stiff materials are usually not
very tough, but strong materials (if not too stiff) are commonly quite tough.
Bricks, for instance, while stiff are notably nontough. The problem with such
stiff materials is that cracks all too readily extend through them. A sharp blow by
a hard object cracks a brick, especially if the brick is supported at its ends and
the blow is delivered in the middle. Mortar keeps bricks supported along most of
their surfaces lest the bricks crack under force concentrations that they can’t
redistribute (as does wood) by sagging a little. By contrast, mild steel, nylon,
spider silk, and fresh wood are strong materials; none is especially stiff; all are
nicely tough. Objects made of them can’t easily be cracked in two but must
usually be cut completely across. You don’t hear of people splitting blocks of
freshly cut wood with karate chops.
James Gordon, retired naval engineer, student of crack propagation, and
biology watcher, has pointed out that humans usually build to a criterion of
adequate stiffness while nature most often builds to a criterion of adequate
strength. Perhaps it implies too much judgment by both technologies to call their
contrasting preferences for stiffness and strength a fundamental difference in
philosophy of design. For both the preference is essentially accidental, since no
one’s really in charge, and incidental, a matter of the materials most readily
available. While far from absolute— we’re really viewing tendencies or
alternative defaults—the difference certainly represents a pervasive divergence
between human and natural technologies.
Bricks, cement blocks, cast concrete, stone, ceramics, and glass are the most
extreme of our stiff materials. Cast iron and high-tensile steel are notably stiff as
well, and wood in the form of cut and dried timber is much stiffer than the stuff
in trees even if less stiff than steel or ceramics.10 Making value judgments,
cracks about being stiff and uptight, are all too easy. They’re also unfair. The
materials we’ve found plentiful, versatile, and durable happen to have been stiff
ones. If your tool is a hammer, you’ll prefer nails over screws; if you discover a
good mortar, then stones suddenly look attractive; given a power saw, you might
build cabins of squared-off timbers rather than logs. With an ample supply of
rot-resistant cypress or red cedar, you might choose wooden pilings over stone
piers. Furthermore, tall structures on land—and thus fully vulnerable to gravity
—are simpler to build from stififer than from less stiff material.
What’s important are the consequences of our predilection for stiff materials,
the strictures it puts on our structures. First, it makes them peculiarly vulnerable
to any accidents or unusual loads that might start cracks. These include both
unanticipated loads imposed by our uses of the structures and loads from
extreme but rare environmental forces, such as hurricanes, heavy snow or ice,
and earthquakes. Second, fractures are more perilous than deformations; a
structure can’t so easily bounce back or grow back from a complete break. So
our stiff materials court disaster more than nature’s flexible materials. These
disasters have stimulated both soul-searching and useful analysis by engineers as
well as some very readable literature.11
Gordon also pointed out a third and more subtle consequence. Most amply
stiff structures will be sufficiently strong as well, but adequately strong
structures may not be especially stiff. Put another way, more material is usually
needed to build something that’s stiff enough than to build something that’s
strong enough. Should we conclude that our technology is wasteful because it
puts such a high value on stiffness? We certainly find unstiff floors, ones with
too much “give,” disconcerting, even when they’re strong enough to avoid
breakage quite safely. I rebuilt a deck on my house a few years ago using thicker
beams; the original deck had given long service even under heavy partying, but
it felt just a bit too lively. The low stiffness of modern skyscrapers, suspension
bridges, and airplane wings bothers quite a few of us. But any value judgment
between strength and stiffness as design criteria is unfair unless we examine the
practical and historical options and alternatives. Still, earthquakes are undeniably
harder on stiff houses than on limber trees.
The bones of large animals are fairly stiff; they support the animals against
gravity and serve as the levers and attachments that allow muscular engines to
drive body movements. Just as a stretchy tendon would defeat the action of a
muscle, so a leg bone that bends would lack standing. While nature does use stiff
materials, she ordinarily reserves them for applications where stiffness is crucial.
Much stiffer than bone are the real biological ceramics, the ones with less
protein and more inorganic material than bone. Hard coral is a particularly stiff
material, but coral forms new and more widely dispersed colonies after a reef
breaks up in a storm, so failure of stiffness doesn’t mean failure of fitness. The
shells of marine mollusks are stiff too. But the shells provide enviable
protection. Anyway, both mollusk shell and hard coral skeleton are made of
calcium compounds that may cost very little. The ocean isn’t at all short of
calcium. Quite the opposite, it’s supersaturated with the stuff, and very little
energy need be invested to extract calcium from seawater. The enamel of our
teeth is even stiffer than mollusk shell, but soft teeth would lack bite. For each
instance we can recognize a specific rationale for making stiff materials and
structures.
Like engineers and architects, nature must deal with the lack of toughness of
stiff materials. Teeth and bones and hard coral, after all, do crack. Significantly,
both technologies commonly use their stiffest and thus most crack-vulnerable
materials in small pieces. Tooth enamel forms only thin layers or outer surfaces;
teeth contain more dentin than enamel. Similarly, if you want to cut hard
materials or get very long service from the blade of your power saw, you choose
carbide-tipped blades. “Tipped”—we limit the stiff ceramic to tiny pieces at the
tips of the saw’s teeth.

THE VIRTUES OF FLEXIBILITY

Using unstiff materials can do far more than circumvent a few cracks on the
cheap, and here nature wins hands down. I’ve picked examples from a wide
range of possibilities to emphasize two general points. First, flexibility pays off
both under tough external conditions—winds, waves, impacts, and the like—and
in connection with internal operations—the flow of blood and the tugs and twists
of muscles and tendons. Second, flexibility can do more than improve toughness
and impact resistance. It provides a way to make structures that can change their
shapes when loaded in highly specific and useful ways.12
How might one design a long structure that needs a lot of surface area and is
exposed to rapid water currents? Something stiff, like the hull of a ship, requires
substantial bracing. It may break suddenly, and it’s in real trouble if it hits
another stiff structure. A kind of marine alga or seaweed—a kelp—that lives on
a wave-swept rocky coast (as in Figure 5.6) may be as much as a 150 feet long,
as long as a big ship. The attachment to a rock at one end seems remarkably
weak for a long structure subjected to the drag of storm-generated waves, and
the whole plant seems not just flexible but positively flimsy.
FIGURE 5.6. An especially long marine alga, Macrocystis, from the Pacific coast of North America.

Mimi Koehl, who works on all kinds of flexible marine organisms, identified
the kelp’s trick. She’s a flexible thinker and recognized what earlier observers
hadn’t considered.13 Waves produce flows that reverse direction every few
seconds. When a nicely flabby kelp grows longer than the distance the
surrounding water travels between reversals, its drag stops increasing. Any
additional length simply goes with the flow. Its speed is zero relative to the water
around it, so it has no drag! If water flows three feet per second and reverses
every four seconds, then only the first dozen feet of kelp pull on the attachment.
Koehl’s seaweeds grow to great lengths to avoid drag. But the trick works only if
you’re really flexible: One part has to extend in a different direction from
another. Long kelps flex like ropes.
Our flags and pennants have a distressing tendency to tatter in strong winds,
distressing unless, as do some plant ecologists, you use the tattering as a measure
of exposure to wind.14 Flags in flows are draggy things; for the same shape and
area, a flag has around ten times as much drag as a rigid weather vane. So
flexibility clearly confers no automatic advantage. The kelp’s evasion is
impractical on land because winds are too fast and don’t reverse often enough.
With a wind speed of forty miles per hour (fifty-nine feet per second) and a
reversal even as often as every ten seconds, a limp object would have to be
almost six hundred feet long to reach the dragless zone. Nature certainly flies
flexible flags on long poles; hers are called leaves, and they aren’t hauled down
during storms. To do its business, capturing energy from sunlight, a tree needs a
big area of leaf, which could make big trouble. Most of the drag of a tree comes
from its leaves; the trunk contributes little. Transfer of that force, drag, from
leaves to trunk to roots makes trees go down in storms.
So what’s the drag of a leaf? A few years ago I did some measurements in a
highly turbulent wind tunnel at speeds a leaf might meet in a storm.15 A leaf
may be flexible, but leaves experience more nearly the low drag of a rigid
weather vane than the high drag of a flag. Their strategy involves a clever use of
flexibility. For instance, as a wind pulls a single maple or tulip poplar leaf away
from the tree, the wind catches the lobes on either side of the stem end of the leaf
blade. Those lobes bend upward, and the blade curls into a cone, as in Figure
5.7. As the wind increases, the cone rolls ever tighter. Even in highly turbulent
and fluctuating winds, the cone is stable and experiences relatively low drag—
around a quarter of that of a square flag of the same area as the leaf. One can
imitate the behavior with simple models of paper or plastic, but they work only
crudely; mere flexibility isn’t enough. It takes just the right amount in just the
right places. The leaf’s solution, though, isn’t without its disadvantage. Curled
up, a leaf exposes less area to the sky, and exposure to light, after all, is central to
its business. So a maple tree curls its leaves the way a sailing ship furls its sails,
to lower its drag temporarily.

FIGURE 5.7. The leaf of a tulip poplar (also called a tulip tree or yellow poplar) in still air and winds of
11, 33, and 44 miles per hour (5, 15, and 20 ms).
FIGURE 5.8. The compound leaf of a black locust in still air and winds of 11, 33, and 44 miles per hour (5,
15, and 20 ms).

Curling into a cone is only one scheme leaves use to reduce drag in windy
weather. Leaves made up of pinnate—that is, feather-arranged— leaflets, such as
those of black locust and black walnut, reconfigure into cylinders that tighten as
the wind rises (Figure 5.8). Besides individual tricks, leaves play communal
ones. In rough weather the needles of pines splay out less and get more like the
hairs on the tail of a horse, once again achieving relatively lower drag. The stiff
leaves of the American holly swing together over the stem, packing like a
multilayer sandwich (Figure 5.9). Leaves that curl into individual cones can also
curl into communal cones when they grow close tdgether. Other strategies for
leaf reconfiguration undoubtedly exist, but no one has yet done a systematic
survey.
Not that we’ve never done what leaves do; in high winds the blades of many
old windmills would furl, if made of fabric, or pivot, if rigid. Perhaps we rarely
face the tree’s problem since we don’t often need to expose a lot of surface to the
sky while that surface is well off the ground. Would some slight saving in
material make us tolerate utility poles and antennas that bent over in storms?
FIGURE 5.9. A group of leaves on a branch of an American holly in still air and winds of 22, 33, and 44
miles per hour (10, 15, and 20 ms).

The mechanical bases of some of these reconfigurations interest us here. A

cluster of leaves usually feels even less drag than do individual leaves. To
cluster, the flexibility of the stems of the individual leaves is as critical as that of
their blades since the stems have to twist readily. That presents a queer problem
in design. To hold the blade outward so it can absorb sunlight, the stem of a leaf
must resist bending. Thus it must be resistant to bending loads while being
compliant to twisting loads; it has to twist more easily than it bends. “Twisting in
the wind” isn’t just a slogan from the last days of the Nixon presidency.
One way to aid twisting while preventing bending turns on a simple device
that’s easily demonstrated. Try first to bend and then to twist either the cardboard
core of a roll of foil, plastic wrap, or toilet paper or else a plastic soda straw.
Then make a lengthwise slit in your cylinder, and try again. The slit weakens it,
but not equally so in both respects; the cylinder’s resistance to twisting is
reduced far more than its resistance to bending. In fact, any lengthwise groove or
line of weakness or even any deviation from a round cross section will do the
same, if less dramatically. Many leaf stems have lengthwise grooves along their
tops (Figure 5.10). Together with some peculiarities of the cells and fibers inside,
the grooves ease twisting relative to bending, which reaches four or five times
that of cylinders of plastic or metal.
FIGURE 5.10. Lengthwise grooves on the upper side of the leaf stem (petiole) of a dogwood and on the
underside of the wing feather of a bird.

The feathers that form the tips of a bird’s wings face an equivalent problem.
Here again, ease of twisting is functionally important. Wings propel a bird the
way a propeller moves an airplane; they just beat up and down instead of
spinning around and around. A propeller blade has to be twisted lengthwise to
give a decent push to the air passing across it. If it were to spin in the other
direction, the twist would have to be reversed. But twice per stroke the feathers
of a beating wing reverse direction and so must reverse their twist. At the same
time the feathers must resist bending; after all, the wings lift the bird, so in flight
its body truly hangs from its wings. Again a structure must twist but not bend,
and as on many leaf stems, a lengthwise groove runs along each feather’s shaft.
Melina Hale, now an accomplished biologist but at the time a first-year
undergraduate, drew my attention to the possibility, which measurement then
confirmed.
One difference, though, between wing feather and leaf stem: The groove on
the feather is on the bottom, while that of the leaf stem is on the top. That too
makes some sense, as you can see with your cardboard core or plastic soda
straw. In bending, one side is stretched while the other is compressed, and the slit
makes least trouble for bending when it’s on the side that’s stretched. For the leaf
stem that’s the top since the weight of the blade bends it downward. For the wing
feather it’s the bottom that’s stretched; its own lift makes the feather bend
upward.
A lengthwise groove, then, indicates twistiness. Where else and for what else
might the device be used? Consider a starfish. Using the suction cups on the tiny
feet that protrude beneath its five arms, it can (very slowly) grab a clam. It then
pulls steadily until the clam tires and gives enough so the starfish can insert its
stomach between the half shells and digest its dinner. But a starfish bears arms
that must adjust to fit all clams. Patricia O’Neill looked at how starfish, sand
dollars, and other echinoderms work. She found that the groove that runs
outward along the bottom of each arm permits a starfish to twist each arm to
perfect its grip—as it must, since living clams clam up forcefully and
persistently.16
In this look at bending and twisting, we again encounter something possible
but rarely used by human technology. We mainly put twist-prone structures
where no forces will twist them. A road sign neither bends nor twists much if it’s
mounted on a round pole. But it twists readily on a cheaper pole, one merely
creased lengthwise (Figure 5.11). We just avoid any wind-caused twist of
creased poles by putting only well-centered signs on such poles. Similarly, an
ordinary I-beam girder is very prone to twisting, but we hide that flexibility. We
always use at least two such beams adjacent to each other, so each keeps the
other in line.
Another aspect of how flexible structures bend matters to nature more than to
us, or at least nature makes a virtue of what for us is mainly a nuisance. The way
a structure bends depends not only on the stiffness of its material but on the
amount of material and the way it’s arranged. Altering the amount of material
even a little can have a large effect; recall how slightly thicker bookshelves
sagged much less. For a cylinder, resistance to bending follows the fourth power
of its radius; doubling the radius gives a fully sixteenfold greater rigidity. Or,
looked at the other way, a simple kind of bending joint can be made by judicious
local thinning. Mimi Koehl showed how a tall sea anemone (see Figure 5.12)
lets a current bend its crown of tentacles just enough to position it best for
capturing the small edibles that the current carries. Normally a sideways force on
the top has its greatest effect lowest down; an anemone of uniform stiffness and
diameter would bend over at the bottom. But the trunk of the anemone is a little
skinnier at the top, and that’s enough to make an adequate bending joint about
which the crown rotates.17
FIGURE 5.11. The creased poles often used for road signs twist easily, but we mount the signs
symmetrically so winds cause little twisting force. For less well-centered loads, we use more twist-resistant
cylindrical poles. I-beams twist easily, but we circumvent that weakness by using them in pairs or groups.

FIGURE 5.12. Both a large sea anemone, Metridium, and the flower stem of a daffodil bend at specific,
predetermined places.

Something like this happens when a daffodil flower emerges. The main
difference is that it happens only once. Initially the bud points skyward, but it
bends at a specific point just before opening. What appears to be happening is
creation of a temporary joint, more likely by temporarily decreasing stiffness
than by thinning, and then letting gravity do the work. I’ve made strange-looking
daffodil flowers by inserting the stems upside down in siphons so the buds point
down instead of up. No bending then happens, and the resulting flower, turned
upright again, looks expectantly skyward rather than shyly downward.
Not that we never use localized flexibility. On my desk I have a small file
box made of a single piece of soft plastic. The hinge is just a horizontal thin zone
between trough and lid that makes bending happen where it’s wanted. The box
has now opened and closed when prompted for about a decade, and it’s not
obviously worse for the experience.
 SWELL PIPES
But enough twisting and bending of protruding parts. Let’s go inside.
Contraction of your left ventricle generates a hearty pressure that forces blood
out into your arteries. Then that ventricle relaxes and gets refilled. Blood
pressure at the heart varies in each stroke from about 0 to 120 millimeters of
mercury. Zero? Why then do we measure a range of about 80 to 120 with a cuff
on the arm? Simply because our arteries are stretchy enough to damp the
pressure fluctuations of the heart. When the heart contracts, the blood it pumps
stretches the arterial walls. When the heart relaxes, the arteries deflate and in the
process do a little passive pumping on their own. The synchrony is automatic,
and it’s why you can count heartbeats by feeling arterial diameter changes at
your wrist or neck. The damping is a good thing too. Blood entering the small
vessels moves a lot more smoothly, and lower peak pressures produce sufficient
flow of blood. Atherosclerosis—stiffening of the arterial walls—means trouble.
One signal of its presence is a wider spread between the maximum and
minimum pressures measured on your arm.
Many ordinary pumps (for instance, the piston pumps we use to inflate
bicycle tires) are as pulsatile as any heart. We might smooth the flow with
arterylike elastic pipes, but we ordinarily take a different tack. Our pumps often
have several chambers that work at different points in the cycle. Thus each
chamber produces its peak pressure at a slightly different time, and the overall
pressure never drops to zero. For the same problem we use a different solution,
one that doesn’t need materials that are especially flexible and, as I’ll now
explain, most peculiarly flexible.
Inflate a cylindrical balloon or a rubber condom. What happens isn’t like
stretching a rubber band. Starting to stretch a rubber band is easy, while it takes
more and more force to stretch it farther and farther. Starting to inflate any
balloon takes as much pressure as (and often more than) further expansion; after
the start a constant pressure does the rest of the job. Furthermore, one part of a
cylindrical balloon inevitably expands almost to the bursting point before the
remainder does much at all. This odd (if familiar) behavior reflects something
that came up in the last chapter, Laplace’s law. In expanding, the wall of a
balloon gets flatter. The rubber may be stretched farther, but pressure stretches a
flatter wall more effectively. So the rubber gets harder and harder to stretch
while the pressure gets better and better at stretching. The two just about
balance. (Only the fact that the stress-strain line for rubber has an extra steep bit
just before the breaking point permits cylindrical balloons to work at all. If the
line were fully straight, one part would actually break before the rest stretched at
all.)
In an artery, such ordinary elastic stretchiness would produce a local bulge,
an aneurysm, something even worse than atherosclerosis. Fortunately, normal
arteries expand uniformly, unlike balloons. How do they manage this crucial
trick? To achieve uniformity, a stretchy pipe has to be very flexible when
inflation starts but get stiffer and stiffer as the process continues; expansion must
start easily but then become more and more difficult, disproportionately so. In
their walls arteries have fibers of collagen, the same unstretchy material that
makes up most of a tendon. But the fibers kink up when the walls aren’t
expanded, so they’re mechanically irrelevant; a slack rope withstands no pull. As
in Figure 5.13, expansion of the wall extends more and more of the fibers to the
point where they provide tensile stiffening. This very unordinary stretchiness
produces the upwardly curved stress-strain plot for arterial wall of Figure 5.3.
Again we see how flexibility in nature is a subtle and multidimensional business.
We (our arteries, at least) are really special.
Still, however nicely designed, we’re not unique. Robert Shadwick, of the
Scripps Institution of Oceanography, and his associates looked at some creatures
whose circulatory systems work much like ours but that are about as distantly
related to us as animals can be.18 They found almost precisely the same variable
flexibility in the arteries of squid and octopus. The main difference was that
these creatures tune their arterial stretchiness to work at their lower blood
pressures, as do our low-pressure closer kin, toads and lizards. Most startling,
octopus and squid achieve their variable flexibility with a different elastic
protein from the one we vertebrates use. So their arterial flexibility must have a
substantially different genetic basis and must represent an independent
evolutionary innovation. Wherever a pulsating heart pushes blood through
flexible vessels, the vessels simply must be designed to avoid aneurysms.
FIGURE 5.13. Arterial wall unstretched and stretched, showing how initially kinked fibers straighten out,
making the material gradually get stiffer.

As if to emphasize the point, a third group of animals has evolved aneurysm-

resistant vessels. Shadwick found variably flexible arteries in crabs and lobsters,
with yet another material basis and once again tuned to work at the appropriate
blood pressures. He also recognized that this tuning of flexibility permits
portentous predictions. Given a sample of its arterial wall, one can pretty well
guess an animal’s blood pressure. By this test, giant squid (which we’ve never
kept alive) apparently have blood pressures as high as our own.
An upwardly curved stress-strain plot means that little energy is stored up
when the material is stretched; the line has no great area underneath it. That can
be a safety feature. A balloon, with a less curved plot, absorbs a lot of energy
when you inflate it (whew!), and most of the energy gets liberated violently if
the balloon bursts. Upwardly curved plots are common among biological
materials such as skin, including the stretched skin of bat wings and duck feet;
such as cartilage, as in our external ears; and such as ligaments. A cut or
puncture doesn’t release much energy, and (here we anticipate the next chapter)
release of energy is what keeps cracks propagating. So a slight injury to such a
material doesn’t have the catastrophic consequences of a pinprick in a balloon.
But upwardly curved plots aren’t universal. The plot for tendon (as in Figure
5.3) has what looks like a dangerously high area underneath. But on that area
hangs functional significance of another kind. Tendon is stiff and doesn’t stretch
very far—10 percent beyond original length is about the limit—nor could it and
still do its job. R. McNeill Alexander, probably the foremost investigator of
bioelasticity, found that its energy storage even at these low strains permits a
kangaroo to jump more cheaply. On landing, tendon is stretched; tendon and
animal then rebound, with about 40 percent of the absorbed energy reappearing.
We do likewise when we run. An Achilles (heel) tendon absorbs a lot of the
energy released when a leg decelerates at the end of one step. That energy then
helps reaccelerate it at the start of the next.19 Energy storage makes getting
about on legs much more efficient. Legs with energy-storing tendons don’t reach
the efficiency of wheels, but they’re better than legs without step-to-step storage.
These tendons, then, represent still another way to put nonrigidity to practical
purpose.
If you want specific imagery for the present comparisons, consider a cat’s ear
and a door hinge. In one technology, orientation is changed by making things
bend; in the other, by making them slide or roll. My plastic file box has bending
hinges; when my joints move, my bones slide along one another. Once again the
distinction between the technologies is one of degree and default. And once
again such things as historical and evolutionary continuity, the availability of
materials, and the modes of manufacture underlie the distinction, in short, the
factors that underlie the different ways objects get designed and built.
If you want a context for the present comparisons, consider whether without
the contrasting world of natural design, you would have wondered about the
consequences of living in structures and using devices that are built of stiff stuff.
Or would you have guessed how multifaceted is flexibility? What’s most
familiar biases our thinking, and what’s most familiar is mostly what we
ourselves make.
 Chapter 6

TWO ROUTES TO RIGIDITY

Nature teaches us the virtues of flexibility. Leaves, large algae, and feathers
show us how to economize on material, how to change shape as environmental
forces change, how to enlist the environmental forces themselves to produce
those changes. Still, nature doesn’t entirely avoid stiffer stuff. Shell, coral, tooth,
bone, and wood are relatively stiff. So too are the chitinous exoskeletons of
lobsters and big beetles. Stiff materials may not be necessary for building large
creatures, but they certainly help. But while both technologies use stiff materials,
they differ in what kinds they use. One difference is startlingly absolute: no
“relative to this” or “preponderance of that.”

THE NONMETALLIC WORLD OF ORGANISMS

No organism that we know uses any piece of metal for any mechanical purpose;
for that matter, none biosynthesizes a piece of metal at all. Here “piece of metal”
means something fairly specific. Alloys of different metals are included; we use
precious little in the way of unalloyed metals ourselves. But excluded from the
definition are both organic and inorganic materials in which chemical
compounds bind metal atoms with nonmetal ones. No legalistic subterfuge is
meant since such metal-containing compounds don’t behave mechanically like
metals. Steel and bronze are proper metals because the metal atoms bind directly
to each other, while iron oxide or copper sulfate are simply metal-containing
compounds.
Quite peculiar, this complete absence of metallic materials in nature. To start
with, most (perhaps all) organisms contain atoms of the kinds of metals that
might do mechanical tasks. These atoms aren’t there by accident; creatures
require them, and they have enzymes that can synthesize large metal-containing
molecules. The most familiar of course is hemoglobin, which contains iron.
Other iron-containing compounds help cells transfer energy from compounds
that store it to ones that fuel biosyntheses, that work muscles, and so forth.
Hemoglobin itself has evolved on numerous occasions in both animal and plant
kingdoms. To these iron- containing compounds within us we add ones
containing copper, zinc, chromium, tin, and possibly nickel. One group of
animals, the ascidians or tunicates (called sea squirts or sea pork; one is shown
in Figure 6.1), has blood cells mysteriously loaded with vanadium. A few other
metals are needed in small amounts by other organisms.1 Small quantities, to be
sure; our own most abundant metal is magnesium, still only a twentieth of 1
percent of our weight. As a pure metal magnesium reacts too readily to make
safe structures; burning magnesium made the flash of old-fashioned
photographic flash powder. Of iron, an adult has only about four grams (mostly
in hemoglobin), ten times less than our magnesium.2 Still, we contain metal, we
use metal, we require metal.

FIGURE 6.1. Several adult ascidians, one cut away to show how water passes through its filtration
apparatus. These animals are reasonably common on rocks and wharf pilings, usually below the low-tide
line. They're in the same phylum, the Chordata, as are vertebrates such as we.

Some organisms even build serious mechanical devices out of metal-

containing compounds. Many mollusks (snails in particular) feed with organs
called radulae (Figure 6.2). A radula works like a cross between a cat’s tongue
and a chain saw. The flexible, horny structure goes in and out, rasping food from
a surface and carrying it mouthward. Hard denticles on the radula do the rasping,
and some of these contain a lot of metal. But the metal turns out to be in the
form of metal salts—of iron or copper—suitably hard, but minerals rather than
metallic materials. Nor are radulae unique in this respect. Chewing poses a
peculiar problem for animals that feed on leaves and wood—cows, caterpillars,
and such. For their volume, these parts of plants don’t provide much
nourishment, so the animals have to eat large amounts. Worse yet, the plants
seem to have taken considerable trouble (evolutionarily) to make themselves
difficult to bite off and grind up. Grasses and tropical woods are full of sand
(silicon dioxide), most likely just to hobble herbivory. So tooth wear poses a
worse problem for herbivores than for carnivores; recall the horses’ teeth of
Figure 2.5. Metal salts can harden nonmetallic materials; herbivorous insects, for
instance, have zinc or manganese in their mandibles, but again as salts rather
than as metallic materials.3

FIGURE 6.2. A snail and its radula, a toothed strap that comes out of the snail’s mouth and moves in and
out, scraping like a rasp.

About twenty years ago a short paper announced the presence of “iron-rich
particles” in bacteria that oriented and moved directionally in magnetic fields.4
Superficially the magnets sounded metallic. But the original report made no such
claim and gave no indication of the form of the iron. It proved to be not metallic
iron but a compound of iron and oxygen called, for obvious reasons, magnetite.
Nothing obscure or radical—we use compounds such as magnetite to coat
magnetic tape and computer disks. Magnetite has now turned up just about
everywhere that magnetic sensitivity is known and in some places where it’s
only suspected: in bird brains, in honeybees (in the abdominal segments, as it
happens), in salmon, in some rodent brains. For that matter, it occurs in our own
brains.5 Our sensitivity to magnetic fields remains uncertain.

WHY DOESN’T NATURE USE METALS?

So, ironically, no chunks of metal in organisms. In a sense this absence of
metallic materials tests one of our main points: that natural design is severely
constrained. If you take the opposite viewpoint—that nature will do all that’s
best and brightest—then you have to make the case for metals being unavailable,
inappropriate, or at least no better than naturally produced nonmetallic materials.
How good is that case?

Supply limitations. Perhaps living things find metals scarce because their
distribution on the earth’s surface is too spotty. After all, humans long ago found
that ores or crudely refined metals were worth moving long distances despite
their weight and bulk. In classical antiquity, Spain, Brittany, and Cornwall
exported tin, the crucial additive that made soft copper into hard bronze.
Organisms build themselves of more widely distributed elements: carbon, from
the carbon dioxide in the atmosphere by way of photosynthesis in plants;
hydrogen, from water, also through photosynthesis; oxygen, from the
atmosphere, where it is a by-product of photosynthesis;6 nitrogen, from the
atmosphere through the action of nitrogen-fixing bacteria and a few other
agencies. Calcium and phosphorus complete the list of elements in us in amounts
over 1 percent. Calcium occurs in many rocks and in most natural waters.
Phosphorus makes up less than a tenth of 1 percent of the earth’s crust and a still
smaller fraction of seawater, but it’s well spread around. So far so good with the
argument for effective scarcity of metals.
Still, the list of major ingredients in organisms corresponds poorly to the
composition of the surface of the earth. While none of the elements just
mentioned is rare, some elements present in abundance aren’t used in any
quantity. Aluminum, a splendid structural material, is the most notable; it makes
up about 8 percent of the earth’s crust. Iron, long our favorite nonprecious metal,
makes up fully 5 percent. While rich ores are spotty, decent supplies of
aluminum and iron are widespread at least on land. Besides, organisms do well
at acquiring elements from dilute sources. The ascidians that concentrate
vanadium get it from seawater, where it constitutes only two parts per billion.
Upon closer scrutiny, therefore, ill distribution weakens as a basis for arguing the
effective scarcity of metals. Metals are better spread than one might think, and
organisms get many elements from sparse sources.
Nonetheless, for at least two reasons the scarcity argument shouldn’t be
casually dismissed. First, using a small amount of metal as a cofactor for some
enzyme isn’t the same as using enough to build a respectable amount of
mechanical equipment. Our bodies use iron, but we use far less of it than, for
instance, calcium. Calcium, a major component of bones, makes up two hundred
times as much of our weight as does iron. Maybe organisms can’t afford the cost
of extracting large amounts of iron. Second, abundance in the earth’s solid crust
may be a misleading criterion if the earliest and most biochemically innovative
episodes of evolution occurred in the sea. Seawater doesn’t just contain calcium;
it’s supersaturated with the stuff. So the sea gives up its calcium for free. Iron,
copper, and aluminum are vastly scarcer in the sea—one part in twenty-five
hundred for calcium against one part or less in one hundred million for these
latter.

Chemical trouble. Perhaps chemistry limits the use of metals. Organisms may
be versatile chemists, but not all reactions are equally easy for them.
Mechanically useful metals occur in nature almost entirely as compounds rather
than as pure materials; they’re combined with such other elements as oxygen,
chlorine, silicon, and sulfur. Getting metal out of ore demands a lot of energy;
worse, it requires a concentrated energy input, such as high heat or high voltage.
In chemical terms, metals have to be reduced from the oxidized compounds in
which we find them. A scale called an electrochemical series gives a good view
of the relative difficulty of purifying different metals; it looks at the barrier to
sticking enough electrons on to the oxidized form to get the pure stuff:

Easiest to reduce to pure metal

gold

mercury, silver
copper
lead, tin, nickel
 iron

zinc
vanadium
titanium, aluminum

magnesium, calcium, potassium,

sodium

Hardest to reduce to pure metal

The scale runs from very stable, inactive gold, so indifferent to oxygen that it
doesn’t even need periodic polishing, to highly reactive sodium, which combines
with just about anything it touches and spontaneously burns when exposed to air.
The reactivity of the chemical compounds of these elements runs in the other
direction. Compounds of mercury and gold are weakly bound and unstable. The
top three elements occur in free, uncombined form in nature, copper ores
sometimes contain metallic copper, and metallic lead is very rare. Their ores
easily yield copper and lead. Reducing iron, farther down, from compounds to
metallic form is tougher. Reducing aluminum, still farther down, is even harder.
Metals harder to isolate are also harder to maintain. The issue here is
spontaneous oxidation, rusting. Silver tarnishes a little; sodium reacts violently
even with water. We can use nearly pure aluminum only because it forms a white
rust, aluminum oxide, a nicely impervious coating that protects the underlying
metal. Rusting is rapid, but it’s also self-limiting. (We also use various coatings
and other processes to offset aluminum’s reactivity in the presence of oxygen.)
Iron oxidizes less eagerly, but the ordinary oxide has the inconvenient habit of
flaking and peeling to expose fresh metal. Most organisms are aerobic; they use
oxygen to oxidize carbon-containing compounds as energy sources. Under
oxidizing conditions, perhaps metallic structures are impractical or less desirable
than alternatives. In other words, preventing rust may be more trouble than it’s
worth.
An indication that these chemical limitations—the cost and difficulty of
reduction (blast furnaces and so forth) and the subsequent cost and difficulty of
preventing reoxidation (periodic painting of steel ships and bridges)—may be
important comes from a dog that didn’t bark. Aluminum is the third most
abundant element on the surface of the earth, after oxygen and silicon. Every
clay contains the element. Yet as far as I know, no organism does anything with
it, even as some trace nutrient or biochemical cofactor. Exposure to aluminum
leads to internal accumulation of the element but to no notorious pathology; it’s
much less toxic than the heavier metals.7 The fact that aluminum (and titanium,
for that matter), far down the electrochemical series, isn’t used at all suggests
that iron, a little higher, can be used only with difficulty.
Excessive density. Perhaps organisms don’t use metals because life evolved in
water. Of the metals that might make decent structures, only chemically
troublesome aluminum isn’t particularly dense. It’s only 2.7 times as dense as
water. The equivalent figure for copper is 8.9, for iron 7.9, for tin 5.8; even
titanium is fully 4.5 times water’s density. Just a little structural iron, for
instance, would make an organism a lot denser than water. Thus gravity would
impose a serious load if an organism grew up from the bottom of a body of
water, and it would just about rule out swimming. The minerals with which
nature builds bones, shells, and the like are less dense. Silicon dioxide has a
density 2.2 to 2.6 times that of water, calcium carbonate 2.7 to 2.9, and calcium
phosphate 2.2 to 3.1; the ranges reflect different crystalline forms. Organisms
that don’t have air inside (as do trees) are mostly denser than seawater, but only
slightly. Even a thick-shelled clam sinks more slowly than a chunk of iron. A
small deposit of fat or a tiny bladder of gas can offset slightly denser minerals
and keep an organism from sinking at all. With substantial use of iron or copper,
density compensation would take much greater investments of space, material,
and energy.
This difference in density between metals and these minerals may also be
important for animals that live in sediments beneath bodies of water. If agitated,
particles sort themselves out by density (and, of course, size). Particles of most
sands, silts, and clays have densities between two and three times that of water,
so animals made of lightweight organic compounds,8 even with a lot of mineral,
are less dense than the surrounding sediments. By contrast, an organism with
substantial amounts of metal might be self-burying. The habitat is far from
unusual; most organisms are small, and an extremely diverse fauna of tiny
creatures live in the spaces between the grains of lake bottoms, coastal beaches,
and continental shelves. Yet despite these consistent and suggestive differences
in density, the heaviness of metals doesn’t seem a weighty enough factor to
account for their complete absence in nature.
Inability to grow. I once encountered the argument that metallic structures
couldn’t grow the way the nonmetallic components of organisms grow. I’m not
persuaded that they couldn’t grow. Furthermore, neither mollusk shell nor
arthropod cuticle grows, but both grace eminently successful phyla of animals.
Similarly, trees make their wood once and for all, increasing their girth by
adding new wood peripherally as annual rings. In short, the argument is so much
vertebrocentricism.
When we ask why organisms don’t use metals, we need to consider another
possibility: Maybe metals lack mettle and aren’t all they’re cracked up to be. To
explore that possibility, we need to look both at what makes metals special and at
how human technology uses them.

THE UTILITY OF METALLIC CONSTRUCTION

Humans have long been willing to expend enormous effort to obtain metals.
Ancient Middle Eastern and Mediterranean metallurgy is well known. Iron
smelting was widespread in Africa. The Indians of North America strove to
obtain metallic copper where they had any access to it, as in northern Michigan
and adjacent areas of Canada. The Inuit of Cape York in northern Greenland
laboriously pounded bits of iron from a meteor deposit to use as edges for their
tools.9 Did human technology have easy access to some better alternative to
metals? The possibility seems remote. Why, then, are metals so useful? In
particular, they’re malleable and ductile; with minimal risk of breakage, they can
be pressed or hammered into shape, or they can be drawn out into thin wires and
sheets. Good malleability and ductility imply as well an agreeably high level of
toughness, defined in the last chapter as requiring a lot of energy to be broken.
What’s special about metals is easiest to see by stretching a sample of one
and plotting the result on a now-familiar stress-strain graph. Figure 6.3 gives a
plot for steel: for the lower edge of an I-beam that’s being bent downward or the
bowed-out side of a round column that’s pushed sideways. The stretch initially
meets a high resilience response; removal of the weight on the beam or the push
on the column restores the original length immediately and forcefully. In this
region the plot ascends along a fairly straight line. Double the stress, and you
double the strain. Used like this, a piece of metal makes a handy spring for a
scale to weigh things; the numbers on the scale will be equally spaced, and the
scale will rezero itself between uses.10 Most biological materials such as
tendons give curved rather than straight lines in equivalent plots of stress versus
strain. Organisms, as we saw, put the curvature to good use. That metals give
straight lines indicates intrinsic superiority only for making simple spring scales.
What matters is their high resilience.
Things get special beyond this initial upward line, beyond what’s called the
elastic limit. Typically, the plot levels (or nearly levels) abruptly as the metal is
stretched farther; it enters its plastic region. Without any (or much) additional
loading, the metal can be deformed a lot further, which is what ductility is all
about. This horizontal extension of the stress-strain line generates a lot of
functionally important behavior.
If you use a metal in its resilient, elastic region, you’ve got something that
can be loaded repeatedly without getting permanently bent out of shape. The
piece of metal tolerates abuse. Overload it, and it deforms plastically. That may
spoil it for further use, but metal hasn’t actually broken, and with good design,
catastrophe needn’t follow. The structure just bears its load in a slightly different
position. The bicycle frame may be bent, but you haven’t been dumped off, and
you may even be able to ride home. The bed and side walls of my old pickup
truck got pretty dented, but only the resulting rust demanded attention. Recall
that the area under the stress-strain curve represents the energy absorbed in
straining something. For a metal accidentally distorted beyond its elastic limit,
all that area under its curve enhances its safety; straining the material soaks up
energy that might otherwise make trouble. A well-strained resilient material,
such as rubber, will snap back dangerously when unloaded again. But metals in
the plastic region aren’t at all resilient; they’re plastic, not elastic, and the work
of deformation gets benignly converted into heat rather than stored as elastic
energy. Repeatedly stretch or bend a piece of rubber, and it warms almost
imperceptibly (tires are a little warmer after rapid driving). Repeatedly bend a
strip of metal beyond its elastic limit, and it warms noticeably.
FIGURE 6.3. Stress-strain curves for mild steel and for cow bone (data from Currey, 1984).

Another view of breakage comes from a look at how cracks propagate.

Cracks extend because of force concentrations at their tips. Cracking relieves the
force, releasing energy that powers further progress of the crack. Force
concentration means that the stress at a crack’s tip is much higher than just the
force applied to the object divided by its overall cross-sectional area. In their
plastic region, ductile metals respond to higher local stresses by stretching rather
than cracking. Stretching distributes the force over a greater area and thereby
reduces the stress—or at least ensures that an increase in force doesn’t cause a
corresponding increase in stress. By contrast, cracking merely moves the
location of the high stress from one place to another as the crack extends. That
can increase the stress further since less material then remains intact to bear the
load. Put one way, metals are fairly safe considering how stiff they are. Put
another way, ductile failure is better than brittle failure; a dented cup holds water
better than a shattered one.
We make good use of the ductility and particular stress-strain curves of
metals when we make things of them. Push or pull on a metal piece with enough
violence, and it will bend to your will as you force it into the plastic region.
Remove the load, and the metal retains the new shape, but it now has a new
elastic region; it now follows the dashed line of Figure 6.3. The new shape can
then give long, dependable service, snapping back elastically when modestly
deformed. Neither the stiffness nor the strength of the material has suffered
much by the treatment. That’s what its maker did to my pickup truck before I
bought it, and that’s how almost every curved panel on almost every automobile
is made. With the aid of huge presses and hammers, that’s how we make all
kinds of ordinary objects, such as nails with heads. For large objects we extend
the plastic region by working with hot metal; the process is called forging. To
make wire and hollow pipes, we take similar advantage of the ductility and
malleability of metals.
The stress-strain plot of bone, to take a well-studied biological material,
looks superficially like that of mild steel. It has an initial upward-sloping elastic
portion and then behaves plastically at greater extensions (Figure 6.3, again).
Loading is most often in the elastic region, and the plastic portion provides a
margin of safety. But as John Currey, who has loaded all kinds of bones in all
kinds of ways (picking a bone with Currey is a serious matter), points out,11
what happens in bone differs from what happens in metals. Bone isn’t ductile but
is instead viscoelastic. That is, in its plastic region it’s viscous; it flows. Worse, it
develops tiny cracks in all directions and softens from their interactions. A bone
so loaded then has to be rebuilt, even if it hasn’t actually fractured. Rebuilding
depends on its being a living, growing material. As we use it, bone works about
as well as does a decent metal, perhaps even a little better in terms of the mass
needed to support a given load. But it does so only by virtue of continuous
micromaintenance.12 Without such rebuilding, bone loses much of its value.
Humans have long had access to bone but have mainly used it—as do the Inuit
in particular—when little else, such as wood, stone, or metal, has been available.
Any proclamation of the magnificence of metals must be tempered by a few
cautions. Metals are even more mechanically diverse than are woods, and no one
metal proves perfect in all properties at once. Our cultural biases, stemming from
both personal experience and the way we teach history, may give metalworking
undeserved credit for permitting technological complexity. The use of metal for
more than ornamentation was a glory of Mediterranean and Middle Eastern
civilizations. We mustn’t forget that sophisticated but largely nonmetallic
cultures flourished elsewhere, particularly in the Americas. Furthermore, early
metals weren’t particularly good. Copper is soft; if hammered, it gets harder but
more brittle. Bronze is better, but it still takes an edge far less sharp than flaked
obsidian. I was much impressed watching an anthropologist eat a meal with an
obsidian knife he’d made. He said (and I have no doubt) that he could shave with
such a knife. Bronze, though, is less likely to crack when inexpertly used. That
last point, in fact, may hold the key to the initial attractiveness of metals. Stone
axes are hard, and flaked flints are sharp, but both are brittle, thus fragile, and
their use calls for both skill and care.13
 Finally, metals can be stamped, drawn, forged, cast, ground, sliced, and
sawed. But living systems need to do none of these; they make materials into
artifacts by internal growth and surface deposition. So our wonderfully diverse
array of fabricational techniques may hold little appeal for nature. Once again,
we’re looking at two distinct but individually well- integrated technologies; an
impressive aspect of one may have little relevance to the other.

HOW NONMETALLIC MATERIALS CAN AVOID

CRACKS
Organisms may eschew metals, but human technology uses both metallic and
nonmetallic materials. Through most of our history and prehistory we’ve been
almost as nonmetallic as organisms; truly large scale use of metals is less than
two centuries old. James Gordon14 points out that even the main cylinder of
Fulton’s steamboat of 1807 (although not the boiler) was made of wood, and he
claims that the tenfold drop in the cost of iron was the most important thing that
happened during the long reign of Queen Victoria. The main materials of human
technology have been stone, selected, shaped, flaked, or polished; ceramics,
including brick, pottery, and glass; wood, from tied bamboo poles to seasoned
timber and glued plywood; and a diverse assortment of natural fibers, mainly
keratin from animal hair, and cellulose from plants. During the past century or so
we’ve added a few things: rubber, starting with Goodyear’s vulcanizing process;
plastics, from Bakelite, the first, to ever more diverse polymers; artificial fibers,
such as nylon and polyester; and more complex materials, such as chipboard and
fiberglass. Our use of metals, relative to other materials, has probably passed its
peak. During the fifty-odd years that I’ve observed such things, one metallic
object after another has been re-created in plastic: file boxes, garden carts, even
the side panels of our new car.
In this respect our technology is moving toward nature’s. But I don’t mean
back to nature. We rarely use nature as explicit model, much legend to the
contrary; as we’ll explore in Chapter 12, we didn’t learn spinning from spiders.
We also subject naturally synthesized materials to increasingly complex
processing before reusing them; for instance, we make both paper and rayon of
the cellulose from wood, but neither much resembles nature’s original product.
Of greater interest is how we’re adopting nature’s approach to dealing with
the trade-off between stiffness and toughness in nonmetals. As noted, metals
have decent values of both properties. (Nonetheless, improvement in one usually
entails deterioration of the other. High-tensile steel stretches less than does mild
steel, but it also cracks more easily.) Nonmetallic systems are worse, at least if
made of ordinary, single-component materials. Glass is wonderfully stiff but
cracks so easily that it probably wouldn’t be approved as a new material today.
Plexiglas15 is less stiff, but it still cracks without much provocation. The softer
plastics, such as the polyvinyl chloride of pipes, are nicely tough, but “softer” is
just a gentle way of saying “less stiff.”
Something odd, though, is at work. Our stiff, nonmetallic, single-component
materials, such as glass or Plexiglas, bricks or ceramic tiles, are much more
brittle than nature’s wood, horn, or bone. Even seemingly brittle natural
materials are often tougher than they look. Mollusk shell may look like ceramic
tile, but it resists shattering much better when hit or drilled. I once made hanging
ornaments from a bunch of scallop shells left over from lab work. Drilling a
small hole near the thin edge of each took no special tools or care, and no shell
shattered in the process. A number of primitive cultures fashioned fully
functional fishhooks out of shell.16 Somehow organisms deal with the awkward
tendency of stiff nonmetals to crack.
From measurements of the strength of a material such as steel, one can
calculate the strength of beams made of the material. But by the early 1900s sad
experience showed that such calculations provided poor guidance for how beams
behaved in use. Ships broke in half at stresses that their steel should have easily
withstood. The culprit turned out to be force concentrations—small places where
the stress was unusually high— and cracks that spread from such places. At least
steel suffered less from crack propagation than many other materials, particularly
stiff, nonmetallic ones. Sharp corners and preexisting cracks were bad things, but
the exact nature of the problem remained mysterious.
Around 1920 a British engineer named A. A. Griffith, working at something
other than what he was supposed to be doing, found that glass fibers could be
pulled harder than could glass rods—that is, fibers could withstand great stress.
The thinner the fibers, the more stress they could take, down to the finest that
could then be made, with diameters around one ten-thousandths of an inch.17
Put another way, a bundle of fine fibers would support a lot more load than a
single rod of the same thickness, even where no bending was involved. Still, the
strength of the finest fibers didn’t quite reach the strength of their chemical
bonds. So the real issue wasn’t the strength of fibers but the weakness of rods.
What made the difference were microscopic cracks.
Griffith went on to provide the basis for our present understanding of what
makes cracks extend. The tip of a crack is a place where force is concentrated, as
mentioned earlier and crudely illustrated in Figure 6.4. If you cut halfway
through a sample of a stiff material and then pull on it, it usually breaks at far
less than half the force needed to break an uncut sample. The concentration of
force makes the crack extend, the sample is fractured further, more force is
concentrated at the crack tip, and in an instant the unbroken sample is history.
Whether a crack will propagate depends on the initial depth of the crack and the
sharpness of its tip; the deeper and sharper the initial crack, the more likely it is
to extend farther.
Making additional surface on either a solid object or a body of liquid absorbs
energy—whether slicing bread, cracking eggs, or even as a water strider’s leg
makes a downward dimple on a pond. Since cracking makes surface, it absorbs
energy. But cracking also releases energy as it relieves the force concentration at
its tip. As a crack gets deeper, more force is concentrated at its tip. Thus, as a
crack progresses, it relieves ever more force and releases more and more energy.
Eventually the energy released by relieving the force concentration at the crack’s
tip surpasses the energy needed to make new surface as the crack propagates.
Then stand back, for the system is out of control, and the crack will
spontaneously extend at a fair fraction of the speed of sound. You deliberately
crack a brittle material such as glass or plasterboard by making (scoring) an
initial groove—a potential force concentration. You then apply a slight force,
and the material breaks in two where you’ve started the crack. Breaking a piece
of glass incompletely isn’t at all easy.
FIGURE 6.4. Force trajectories bunch just beneath a crack, so force is more concentrated (stress is higher)
there.

All real objects have cracks. What matter are their depths, the sharpness of
their tips, and the loads on the objects. Griffith’s critical crack length is reached
when the energy released becomes enough to keep a crack advancing. Thinner
objects normally have shallower cracks, which make them stronger, able to
withstand greater stresses. That’s why a bundle of skinny glass fibers is stronger
than a thick glass rod. Using a bundle of skinny fibers instead of a single rod
gains something else as well. If the stress is high enough so a fiber does crack in
two, that’s usually the end of the matter. What chance that the next fiber will
have a preexisting crack just where necessary for breakage to continue? When
you break a bundle of spaghetti, the individual strands don’t all break at the same
place.
Stress a piece of metal, accommodatingly ductile, and its plastic stretch
blunts the tips of its cracks. As a result, critical crack lengths in metals are
between ten thousand and a million times longer than those in stiff nonmetals
like glass. Cracking still matters—ships do break up, so portholes and hatchways
still need to be rounded—but a steel rod is far less fragile than one of glass.
How might this crack-prone fragility of stiff materials, especially
nonmetallic ones, be avoided? The obvious fix is to use very skinny fibers and
run lots of them in parallel, as a kind of ropework. It works as long as the
individual fibers can slide fairly freely against each other. A rope is weakened if
the individual strands are glued together—perhaps a problem when ropes get
iced up. But ropes take pulls and not bends. A second and better (or more
general) fix uses an interface as a crack stopper. An interface is nothing special;
in fact, it’s no thing at all, just the place where one material stops and another
begins—where the rubber meets the road, as a recent ad put it. But an interface
can act in a curious way.
Whether a crack extends or not depends, again, on the sharpness of its tip.
You can often stop a crack by making a round hole at its tip; an elderly machinist
introduced me to the trick when I was a graduate student, and I suggested how
you might demonstrate it with aluminum foil in Chapter 4. How, then, to blunt
the tips of cracks? Think about that ubiquitous material Styrofoam. For its
weight, it’s strong stuff. What keeps strength up and weight down are the
microscopic holes that make it a foam. Plastic and air come together at the edges
of those holes; they’re interfaces. A crack moving across the stuff can’t avoid a
hole, which blunts its tip. So the crack can’t build up a sufficiently concentrated
force on the other side of the hole to keep itself going, as put diagrammatically
in Figure 6.5. Thus a simple way to build in crack stoppers is to make a foam of
the material. The small, hard elements that stiffen and support such echinoderms
as starfish are formed of such a foam. They’re made of calcium carbonate (as
crystals of calcite), which is pretty brittle stuff. But the interconnected holes,
between one two-thousandths and one fiftieth of an inch across, have rounded,
smooth surfaces, as in Figure 6.6.18
One crack may even stop another; a crack is just an interface between a solid
and air, and interfaces can have the same blunting effect as round holes. Imagine
a material with a lot of lengthwise cracks, perhaps ones so small and subtle that
you’d not normally notice them. Cracks running lengthwise are innocent
bystanders as far as load bearing is concerned, with about as much effect as the
medians of divided highways have on traffic capacity. But when a crosswise
crack, the kind that breaks an object, runs into a lengthwise crack, it’s stopped—
like a car running into one of those medians.

FIGURE 6.5. Voids, if oriented lengthwise or if rounded, can reduce the ease with which a crack propagates
across a loaded piece of material.
FIGURE 6.6. The hard elements (ossicles) of echinoderms are really foams of calcite and round-surfaced
voids.

How might lengthwise cracks be built into a material? Bundled fibers might
make good ropes, but they won’t make good beams and columns. For the latter,
less dramatic interfaces work better, interfaces between two solids of very
different mechanical properties rather than between a solid and air. Fiberglass
uses just this trick. Glass still bears the stress, but the strength of a fiberglass
pole dramatically exceeds that of a glass rod. In fiberglass, as shown in Figure
6.7, hard, stress-bearing glass fibers are joined with a relatively soft glue. After
passing through the hard stuff, a crack hits the soft stuff, which stretches (gives)
rather than transfers the crack across to the next hard element.
Materials that use this scheme are called composites, the name alluding to
having more than one component. If you want to experience a composite, mix
wheat bran, for fiber, and egg white, for glue, and bake the product until hard;
shaping can be done either before or after solidification. Adjusting the mixture,
baking time, and temperature changes the character of the product, while adding
some sweetening and flavoring gives you a satisfyingly tough, munchable
material. (Don’t do as several students did when 1 challenged them to produce
an edible composite. Mixing marshmallow and Rice Krispies makes a material
midway between solid and fluid, so shapes slowly sag into amorphousness.)
FlGURE 6.7. Tough composite materials are made of fibers or layers of stiff material separated by material
of much lower stiffness.

Fiberglass is neither the most common nor the best composite material that
human technology has produced. Particleboard combines wood chips and binder;
it’s now the stuff beneath the veneer of our furniture. Soft steel in the reinforced
concrete of our highways and public buildings offsets the brittleness of pure
concrete. Composites compose a versatile technology. Individual fibers need not
run the entire length of a structure; short ones (whiskers) are effective as long as
the glue binds well. Nor must all fibers run in the same direction, as in a
fiberglass fishing pole; the direction is random (in two dimensions) in fiberglass
sheet. Nor are fibers the only useful geometry for the hard material; thin, flat
plates or sheets work well.
For human technology, the main drawback of composites is cost, typically
greater than that of ordinary metals or single-component plastics. In addition,
material and structure usually have to be made together, something a little
complicated and unfamiliar. Sheets or rolls of metal can be cut, bent, and riveted
together to make an aluminum canoe. By contrast, a felt of glass fiber is
combined with epoxy glue over a mold to make fiberglass and canoe
simultaneously.
Nature uses composites for all her hard materials, usually quite complexly
organized composites; Figure 6.8 illustrates just two. Wood is a composite of
cellulose, a hard, fibrous material, and lignin, a glue. The cuticle of arthropods is
a composite of chitin fibers in a proteinaceous matrix, with some calcium
carbonate salts used to stiffen the larger crustaceans. The shells of mollusks are
made of layers of hard mineral separated by a critical few percent of protein.
Bone is a composite of the protein collagen, some other protein, and calcium
phosphate salt. Even your teeth are composites of mineral and protein; when
teeth are drilled, burning protein makes the sulfurous smell. In each case the
structure is highly tuned for its specific applications. Not only does wood vary
from one kind of tree to another, but the wood of a tree’s roots has very different
properties from the wood of its trunk. Bone isn’t just bone, even in an individual;
its composition and properties depend very much on what it’s called on to do.

FIGURE 6.8. Natural composites: the shell of a mollusk (left) and the egg of a bird (right). Both are made
largely of brittle calcium salts with small but critical amounts of softer material to toughen them and
prevent crack propagation.

A NONMECHANICAL INTERLUDE
Let’s briefly leave the world of materials and structures. Important
nonmechanical differences between our two technologies arise from the absence
of metallic materials in organisms. In particular, metals and nonmetals differ
greatly in thermal and electrical conductivities. Both kinds of conductivity are
hundreds to thousands of times greater for metals than for nonmetals, and both
properties determine how we do many things.
To give a few examples, the thermal conductivity of copper is more than
3,000 times the conductivity of wood, 500 times that of glass, 660 times that of
water, and 1,000 times that of fresh leaves. The consequences aren’t minor. The
higher the conductivity, the faster heat moves from the hotter parts to the colder
parts of an object, so the object will more rapidly approach a uniform
temperature. If you cook on an electric stovetop, pots and pans of high
conductance19 give a more even heat distribution and take much less stirring
and scraping of their contents. Relative to its volume, aluminum has about two
and a half times the conductance of iron or steel and thus works better as
cookware. Pans of pure stainless steel are a nuisance, and cast iron is satisfactory
only when thick and heavy; this slows its response to your adjustments of the
stove and trades one disadvantage for another. Ceramic vessels, even if
heatproof, heat up still less uniformly and mainly find use (their manufacturers’
claims notwithstanding) as ovenware.
When the food reaches the table, though, high thermal conductivity becomes
a nuisance. The coldness of metal and the warmth of wood— celebrated in
aphorisms and figures of speech (“cold steel,” etc.)—is the perceptual signal of
the difference in thermal conductivity. A piece of aluminum feels colder than a
piece of pottery because the aluminum more rapidly conducts heat away from
your hand. Similarly, a silver or aluminum dish conducts the heat away from the
food, increasing the area over which the heat is shifted to the surrounding air and
moving heat to the place where you’re holding the dish. We ate cold food from
metal mess kits on scout trips but blamed it entirely on the weather. Tea or coffee
cools more rapidly in a metal pot, especially in aluminum, copper, or silver,
which have the highest conductivities. According to at least one guide, the
governor of colonial Williamsburg, in Virginia, ate off silver dishes—the worst
of all possible materials. Since cooking was done in an outbuilding, one suspects
that no governor ever enjoyed a nice hot meal at home. Conversely, metal
handles often get uncomfortably warm. Stirring a cup of boiling liquid with a
silver spoon shows the downside of high thermal conductivity. The upside is
good value as radiators for cooling internal-combustion engines and for steam
heating in homes.
The high thermal conductivity of metals would be useful to living things in a
number of situations. When large and moderate-size animals do heavy physical
activity, they generate waste heat that they have to transfer to their surroundings.
High conductivity would help get the heat from muscles to skin, but human
conductivity is low, about that of water. So we resort to convection and
evaporation instead of conduction. In convection, heat is moved from one place
to another by physically moving some heated material rather than just shifting
the heat from one bit of material to the next. The usual materials moved are
fluids—hot blood and hot breath—and moving them requires pumps and energy.
In evaporation, heat is moved by making water evaporate and then moving or
discarding the water vapor. Evaporation takes energy, which the vapor contains;
when the vapor drifts away, the body is left cooler. We sweat; dogs pant; we both
lose water thereby. But water must be obtained and carried around, and both
activities can prove troublesome.
Leaves do likewise. When the wind drops and they’re still exposed to the
sun, they get considerably hotter than the air around them; they can warm as
much as 20° C. Their centers, less exposed to what air movement remains, face
the problem at its worst. Hot centers would be less hot if leaves were made of
nicely conductive metal,20 but no such luck. Instead plants do things that must
limit other activities or interfere with other desirable features of their design.
They make elaborately lobed leaves to get a lot of edge and perhaps absorb less
light. They use a lot of water for evaporative cooling, which requires a decent
water source. They make thickened leaves, which takes material, so they heat
more slowly and can endure longer lulls in the wind. And when the sun is strong,
the wind is low, and water is scarce, many leaves droop down to a less sun-
struck and more airflow-exposed vertical position, which compromises
photosynthetic activity.21
We also take advantage of the high electrical conductivity of metals. Every
electric wire is metallic; the only nonmetallic components of most electronic
devices are active ones, such as transistors, or components, such as resistors,
whose function depends on low conductivity. What about nerves? We may think
of them as wires since nervous systems and electronic devices do similar tasks.
But the analogy between nerve and wire misleads. A nerve impulse moves along
the axon of a nerve cell by a different physical scheme from that of a pulse of
electricity traveling along a copper wire. Which is better? I’d go with the wire.
For instance, nerve conduction is glacially slow compared with that along wires.
One hundred twenty meters per second, a mere 270 miles per hour, is about as
fast as any nerve conducts impulses; by contrast, wires transmit electrical pulses
some five million times faster.22 A thousand impulses each second is about the
limit for a nerve; wires carry millions of pulses per second. All the fancy stuff
we do with our brains requires the most massive parallel processing—using
many nerves running alongside each other and great numbers of circuits working
simultaneously.
When rigidity is the aim, we make great use of metals and a little of composites;
nature makes elaborate use of composites but none at all of metals. I’ve
suggested several reasons why nature doesn’t use metals, but none was fully
persuasive. Two further explanations, alluded to earlier, share the same
uncertainty—admittedly an unsatisfying state of affairs.
Perhaps nature doesn’t use metals simply because she has something every
bit as good. If we ask whether natural composites are in the same class with
metals, we have to say yes. If, for instance, stiffness relative to density is the
criterion, wood and steel are about equal. If work of extension is what matters,
and weight rather than volume is the reference, then yew wood (as used in
English longbows), collagen (as used in Roman catapults or ballistae), bone
(much used by Inuit), and horn (used in Chinese composite bows) all surpass
spring steel. One can pick alternative comparisons in which our manufactured
materials come out ahead. Overall, the rigid materials of the two technologies,
though dramatically different, do about equally well. Our present aggressive
development of composites is driven not primarily by the excellence of nature’s
but by the good performance of even the crude ones we already use. At least in
small quantities, we’ve now made composites that, in the ways that matter to us,
do better than anything in nature.
How curious a process is “design” in nature! For better or worse, composites
are what a mindless, blundering, information-starved, and minimally coordinated
system might be expected to make. Specifically, their properties are highly
sensitive to tinkering with the amounts and arrangements of their constituents on
a microscopic level. In short, they’re what one ought to expect from microscopic
improvisation, nature’s way, as opposed to macroscopic deliberation, our human
mode. And that brings us to the last possibility.
Maybe in noticing that nature uses composites rather than metals, we’re
simply looking at a result of the conservative, noninnovative, stick- with-the-
tried-and-true process of evolution by natural selection. Once organisms of
nonmetallic construction became established, what chance had initially crude
metallic forms? Future benefit is something evolution knows nothing about; she
has precious little venture capital. But even if evolutionary inertia is a reasonable
explanation, defending it is difficult. How could one ever come within shouting
distance of either proof or disproof? Perhaps our best bet is to regard
evolutionary inertia as a hypothesis of last resort—distasteful but unavoidable,
its mention deferred as long as possible.
 Chapter 7

PULLING VERSUS PUSHING

Ropes resist having their ends pulled apart; bricks resist having their ends
pushed together. To formalize the distinction, let’s call structural elements that
resist pulls ties and those that resist pushes struts. For ties we use cables, ropes,
belts, and some glues as well as some metal, wooden, and plastic rods and bars.
For struts we use walls, columns, rods, and bars of brick, stone, concrete, wood,
and plastic. Nature’s ties include muscles, tendons, ligaments, strands of silk,
and the stems of fruits. Most bones, hard coral, and tree trunks and a lot of insect
cuticle serve as struts. In this simple—perhaps over-simplified—distinction
between ties and struts lie some interrelated contrasts between human and
natural technologies. The contrasts aren’t as tidy as those between metals and
composites, but they’re equally pervasive and no less significant. What human
technology does about pulls and pushes is ancient in origin and cross-culturally
consistent—so ordinary and familiar that only viewing an alternative reveals its
peculiarity.
Two chapters back, the reader’s toughness and resiliency were stressed and
strained by a great batch of information about mechanical properties. In talking
about all the properties, though, we considered only what happened when we
pulled on things—tensile tests on samples of different materials. But when push
comes to shove, more than tension deserves our attention. One can also press on
a material, loading it in compression, and one can shear a material, distorting,
say, a rectangular block into a so- called parallelepiped, as in Figure 7.1. So three
stresses matter: tensile, compressive, and shearing.
In reality one doesn’t test a material per se but only a particular piece with a
particular shape. In a tensile test, shape matters very little. After you pull on a
round rod or an I-beam or a tendon, you divide the force of the pull by the cross-
sectional area of the piece to get stress. For tensile stresses, the shape of that
cross section doesn’t matter; a rope, a rod, and an I-beam give the same results.
By contrast, in a compressive test, shape matters a lot. Push on something short
and fat, and it squashes. Push on something long and thin, and it bows out to one
side before it breaks. Push on a tube with a very thin wall—such as a drink can
—and the walls crumple suddenly from a single point. Round, square, and
flattened rods of the same cross-sectional areas collapse under different loads.
Pulling is simple; pushing is anything but. When we worried only about
differences between materials, we could (usually) get away with tensile testing
alone. But materials serve as components of structures, and to understand
structures, we have to worry about shape.

FIGURE 7.1. Three ways to stress a sample of material.

FIGURE 7.2. The stresses on a protruding beam bent downward by a weight at its end and what happens if
the beam lacks resistance to shear.

Besides tensile (pulling), compressive (pushing), and shearing loads, two

more complicated loads matter for structures: bending and twisting. When a
structure as a whole is either bent or twisted, parts of it experience each of the
three simple loads. Think about what must happen when a weight on its end
bends a long, protruding beam, as in Figure 7.2. The weight is eager to move
toward the center of the earth, while the right- minded beam opposes any such
tendency. In the process the top of the beam stretches, at least slightly. Stretching
—that’s what ties resist, so we can view the top as a tie. A little less obviously
the bottom of the beam compresses, again at least slightly. Compression—that
means the bottom works as a strut. But if we actually substitute tie and strut,
disaster ensues: Instead of the tie taking tension and the strut taking
compression, the whole thing collapses downward and inward. The beam,
initially rectangular in side view, has become a parallelogram. Subtly but
crucially the middle of the beam has been resisting shear. Thus, when a beam
bends, all three stresses—tensile, compressive, and shearing—load its material,
and the amount of each stress varies from place to place in the beam.
FIGURE 7.3. The stresses—tensile, compressive, and shearing—on a cylinder as you twist one end one way
and the other end the other.

Something analogous happens when a structure, such as the cylinder in

Figure 7.3, gets twisted. Tension loads the outside, and compression loads the
inside. You can see both by twisting a wet cloth: Tension on the outside is
obvious, and compression of the middle is what squeezes out the water. In
addition, shear loads the whole thing (except the exact center line). That’s clear
enough if (as in the figure) you draw a square on a long balloon and then twist it;
the square becomes a parallelogram. Yet again, a single load on a structure
induces three stresses in the material of which it’s made.
We humans take great advantage of this combination of stresses in twisted
cylinders. We lay alongside one another short natural fibers, such as linen,
cotton, or wool, and twist (spin) them into long threads or ropes. Even though
their fibers aren’t joined end to end, these ropes are remarkably strong. The
fibers are kept from pulling apart by their resistance to shearing (sliding) across
one another when they’re compressed. Pulling on a spun thread or a twisted rope
has the same effect as shearing a structure; it presses together the fibers in the
middle. If the fibers aren’t too slick, they’ll break before they slide apart. This
tension-induced compression is especially obvious with fluffy yarn. Spinning,
either for ropemaking or prior to weaving, dominated many ancient cultures,
consuming a large fraction of the working hours of women, who in almost every
case performed the task.1
Nature never does this wonderful trick. She makes her long threads and
ropes as continuous strands—spider silk, for instance. Or, as in vines, she uses a
system of joinery based on something other than helical twisting and frictional
resistance to shear between fibers. Not that multiply stranded helices are at all
rare in nature’s load-bearing structures—the double helix of microtubules
(Figure 2.2) and the triple helix of collagen are far from obscure—but these
twists don’t gain tensile strength by resisting shear. Spinning is a human activity,
despite our misuse of the word for thread extrusion by silkworms and spiders.
Our protruding beam bent simply and obviously when weighted at the end.
Less obviously almost the same thing can happen when a long structure gets its
ends pushed together. The classic case is an erect column, what the old
Egyptians and Greeks liked to put around temples to keep their roofs aloft. The
force of gravity pulls down on the column; since the column doesn’t accelerate
downward, forces must be balanced, and the ground must be pushing upward.
Thus, overall, the column faces a compressive stress. But as soon as it starts to
bend, a compressively loaded column develops tension in the side that’s
outermost, as in Figure 7.4. The side inside the bend experiences additional
compression, and shear develops within the column just as it does in a beam. So
except for very short, fat columns, where the significant hazard is simple
crushing, columns, just like beams, get loaded in tension, compression, and
shear. The main difference is that the column can buckle in any direction,
whereas most beams know which side is up and which is down and therefore
where tension and compression will occur. A proper I-beam should be higher
than wide, while columns can’t deviate far from circularity.

TENSION VERSUS COMPRESSION

Say you’re designing a structure and you’re free to use struts and ties in any
combination. What factors bear on your choices? Primary among them are the
properties of the materials at your disposal. Of these properties, what matters
most is relative performance under tensile and compressive loading.
Stone, bricks, masonry, and other ceramic materials are terrific at resisting
compression. If you put something on top of them, they stay put and stay intact.
An ordinary brick can support several hundred thousand pounds. Conversely,
these ceramics resist tension poorly, with at least ten times less strength.2 In part
that goes along with their stiffness; again, stiff materials tend to be brittle—
which is to say that they don’t resist crack propagation. Cracks present far more
of a problem in tensile loading, which tends to spread them apart, than in
compressive loading, which has no such reprehensible habit.3 In addition, the
traditional cements and mortars used to join bricks and stones are only weakly
adhesive as well as being brittle themselves. Since in our uncertain world, purely
compressive loading can’t be guaranteed, a little tension insurance pays,
presumably why (as one learns in Exodus 5:7-5) sun-dried bricks do best with a
little straw. It’s also why plaster for cornices was made with horsehair.

FIGURE 7.4. Counterintuitively, local tension can crack a column that overall is loaded in compression.

However, if you can be assured that all loads will be compressive, and if
weight economy isn’t uppermost (as it is in structures that move), then ceramics
are excellent materials. Their high stiffness makes it safe to ignore their weight
in all but very large structures; bricks at the bottom of a wall aren’t squashed
much by the ones above. Gothic cathedrals, truly spectacular structures, must be
the greatest things humans have ever built that tolerate only compressive
loading.4 Vaulted roofs press outward on the walls, and external buttresses press
inward, with the two in very close balance. The less massive the walls, the better
must be that balance, or the walls—just piles of stones—will topple inward or
outward. Building a cathedral entailed a lot of mid-course adjustments of
vaulting and buttressing to keep everything in proper compression.5 Failures
were common during construction, in part why building one sometimes took a
century. St. Peter’s in Rome took no less than 181 years to construct, but St.
Pierre’s of Beauvais, France, retains the record. Two spectacular collapses
punctuated 350 years of intermittent building, and the cathedral remains
unfinished to this day. But once correctly done, a cathedral proves durable and
maintenance-free. The same can be said for stone arch bridges, aqueducts, and,
of course, pyramids.
Ropes, cables, chains, and vines resist tension, but compression causes
immediate flabby collapse. Hammocks are loaded in tension, and suspension
bridges are tensile structures except for their towers and roadways. Tensile
structures form subsystems that support elevators and the masts of cranes.
Tension in the rubber walls of tires—as much as air pressure—keeps automobile
bodies off the road, and tension in their outer fabric maintains the shape of
blimps. But compared with compression- resisting structures, the items are fewer
and smaller, and more of them are contemporary.
One of the best things about metals is that they take pulls and pushes about
equally well. That suits them for use in many of our best-loved devices, which
must withstand both stresses, sometimes in rapid alternation. Consider the
framework that resists a car’s weight and motion as the car speeds along a
bumpy road. Or think of the hull of a ship in a wavy ocean. If the ship spans two
wave peaks, its ends get pushed upward and its middle sags. If the ship straddles
a single wave, its middle gets pushed upward and its ends sag. The fine ability of
metals to deal with both compression and tension means that the casual observer
often can’t guess from structural details what kind of load a particular member
was designed to resist.
Biological materials differ only slightly. Like cables, our muscles, tendons,
and ligaments resist only tension. Muscles pull and are pulled on, and you suffer
pulled muscles, not pushed ones. Like steel beams, our bones and cartilage resist
both compression and tension. The differences, though, are significant. For one
thing, purely compression-resisting elements, such as bricks, almost never occur.
Even teeth must resist tension; the way we load our teeth, imagine how much
more prone to breakage they’d be if weaker in tension. For another thing,
biological materials don’t behave as similarly in tension and compression as do
metals. Wood, thoroughly fibrous stuff, is about twice as good at taking tension
as compression: Bend a live twig or a thin dowel, and it buckles on the inside of
the curve before cracking on the outside.6 Bone, by contrast, is slightly stronger
in compression than tension. Not surprisingly, mollusk shells are stronger in
compression. What’s interesting about shells and other biological ceramics is
that most are only about three times weaker in tension, far less than the tenfold
of conventional ceramics of human manufacture. That’s of course the advantage
of being composites or foams.

TIE VERSUS STRUT

Not surprisingly, both human and natural technologies use both struts and ties.
But functional requirements don’t fix the exact mix of the two. The designer of
something complex, like a large bridge, has considerable flexibility in deciding
how much to rely on struts and how much on ties. I think each technology has a
perceptible bias in its designs. Historically, humans have done better at handling
compressive loading; in struts we trust, not in the ties that bind. Nature prefers to
pull and is fonder of ties.
Such a bias will have a dramatic impact on the structures you’ll make. If you
stand with struts, you might build an arch bridge (Figure 7.5), in which almost
everything is compressively loaded, and the ends push (unsuccessfully) the
banks of the river apart. That’s why arches figure so prominently in stone-based
classical architecture. If you’re hung up on ties, you’ll be more inclined toward a
suspension bridge, where the ends try to narrow the river. Besides cost,
properties, and availability of materials, several other factors might bear on the
choice between struts and ties.7

• As we shift attention from materials to structures, size takes on renewed

importance. But it does so in counterintuitive ways—the general point of
Chapter 3. Fishing line labeled “Ten pound test” can take a ten-pound pull no
matter how long a piece you unreel. To support a given load, a long tie need
be no thicker than a short one, at least until the weight of the tie itself—self-
loading— becomes significant. (Digression: In theory an earth satellite can
dangle a rope down to the earth’s surface yet stay in orbit. If the satellite’s
center of mass is high enough so the orbit is geosynchronous, the rope won’t
even shift around. Terrific. For something to get into orbit thereafter, it
 simply climbs up the rope, pushing upward just enough so the center of mass
of the whole system doesn’t drop. So what’s wrong? Physics isn’t offended,
but material science fails. No known material has sufficient tensile strength
for a rope that long not to break of its own weight).8

FIGURE 7.5. The arch of an arch bridge is loaded in compression, while the main cables and suspenders of
a suspension bridge are loaded in tension.

For compressive loading, though, making something longer makes it

weaker. A short stick is harder than a long one to break by pushing the ends
together. Try it with any long, thin, stiff thing such as a dowel, a soda straw,
or a strand of spaghetti. The guilty party is sideways buckling, to which the
long one is more vulnerable. Proportionately increasing the thickness of a
strut as its length increases doesn’t solve the problem. As noted in Chapter 3,
doubling the size of a strut-supported system without changing its shape
doubles the stress on the struts. To keep the stress constant, their thickness
must increase disproportionately.
Nature appreciates the problem of scaling struts. Among trees an increase
in height comes with a disproportionately great increase in trunk diameter.9
The bones of large terrestrial mammals make up a larger fraction of their
body masses than do the bones of small mammals. The skeleton of a 10-
 pound cat makes up about 7 percent of its weight; a 130-pound person is 8.5
percent skeleton; a 1,300-pound horse is 10 percent; the skeleton of a
15,000-pound elephant is all of 13 percent of its weight. In fact, even these
increases in skeleton don’t keep all mammals equivalently sturdy, and the
larger ones are more fragile than they appear.10 As D’Arcy Thompson
poetically puts it, “Elephant and hippopotamus have grown clumsy as well as
big, and the elk is of necessity less graceful than the gazelle.”
Thus when scaled up in size, the strut gets into trouble that the tie knows
nothing about. Gibbons provide an instructive case, since they both walk on
two feet and swing branch to branch (brachiate) from two arms. Their arms
are loaded almost entirely in tension as they swing.11 The weight to be
managed may be the same, but the bones of their arms are longer and thinner
than the bones of their legs. Ties can be long, but struts should be kept as
short as possible.

• The stability of struts and ties differs in a curious but entirely reasonable
way. Once a strut starts buckling, it does so more and more, and it quickly
collapses. The failures of the cathedral at Beauvais must have been scary.
Problem: As soon as a strut begins to bend, it loses strength, and the more it
bends, the weaker it gets. In part that instability comes from the increasing
leverage of the bent strut as it moves outward. In part it comes from the way
crosswise cracks extend ever faster as buckling proceeds. As an engineer,
Michael French, pointed out, once Samson pushed enough to bow the
columns of the temple by some critical amount, he could ease up and let the
roof do the rest.
In this respect a tie is intrinsically stable. Stretch a tie between two posts,
and hang a weight in the middle; the tie will deflect downward. More weight
means more deflection, as either the posts bend or the tie stretches. But at no
point does the process become self-sustaining; clotheslines pose few hazards.
For stability as well as for scaling up to larger size, ties do better than struts.
Once again, nature responds rationally. Larger mammals may be bonier, but
they’re no more muscular. Forty percent muscle does for us all.

• But building with ties is trickier than building with struts. Struts go together
easily; they stack, with gravity providing the glue. Ties are hard to connect to
each other or to struts, as we’ve noted several times already. The
technologies of fasteners and adhesives have challenged every culture that
has built things by any scheme other than piling rocks or blocks one upon
another. In a dome, for instance, tension loads the lower periphery. Great
tensile chains of iron, hidden within the masonry, encircle Brunelleschi’s
great dome of the Cathedral of Santa Maria del Fiore in Florence, Italy.12 By
making rings in which the chain connects only to itself, Brunelleschi
circumvented the problem of tensile fastening. Another problem for ties is
gravitational loading. Most of our structures extend upward from the earth’s
surface, something that can be done with struts alone but not solely with ties.
Wooden ships provide an especial challenge to effective joinery, and by long
tradition shipwrights enjoy more prestige than carpenters; according to an
old saying, even a poor shipwright makes a good carpenter.
Not that ties aren’t used in ordinary construction. Conventional frame
houses with peaked roofs have horizontal members (stringers) that run at
right angles to the ridge of the roof; they often form the floor of an attic. Two
roof supports (rafters) and a stringer form a triangle (Figure 7.6).
Gravitationally loaded, the roof will tend to force the walls apart; the stringer
is the tie that binds the walls. Alternatives are available, so a peaked roof
needn’t be tied by stringers. One, again, is external buttressing, providing a
counterforce pushing inward at the tops of the walls, as in Gothic cathedrals
and some A-frame houses. Another, which gives my house the pleasant
openness of so-called cathedral ceilings, is a large, stiff beam just under the
ridge that’s held up by inconspicuous vertical columns. That lengthwise,
horizontal ridge piece then keeps the rafters from spreading.
For some reason (or, more likely, reasons), tensile joinery doesn’t faze
nature. Once in a while a tendon pulls loose from a bone, but the commonest
examples of tensile disconnection—leaves, seeds, and fruits from their
parent plants—can’t be dismissed as cases of failure.
FIGURE 7.6. A conventionally framed house in which stringers act as tension-resisting elements and a post
and beam house that manages without stringers.

• Getting the stiffness that we humans like presents difficulties for a structure
that makes elaborate use of ties. A chain or steel cable may be stiff in our
original sense of material stiffness—that is, in resisting stretch when pulled
lengthwise. But it isn’t stiff in a structural sense; it bends with little
provocation. That’s the downside of the length and slenderness tolerable in
ties. Since nature tolerates low structural stiffness, she’s better poised to take
advantages of ties.
Nowhere have we used tensile ties on a larger scale than in suspension
bridges. Aside from primitive suspension bridges built of vines, the scheme
goes back only as far as the availability of reasonably cheap metal—first
wrought iron and then steel—that could resist tension. (Most suspension
bridges use steel wire for both the main cables, in wrapped bundles, and the
road hangers. But a lovely wrought-iron one spans the gorge of the Avon at
Bristol, England. It has no wires. Bone-shaped links (eyebars) make up the
main cables, with transverse pins connecting them, and the hangers are very
long rods that descend from those connecting pins. The great engineer
Isambard Brunel designed it in 1831; when it was built in 1864, steel had
become cheaper, and the bridge was anachronistic.) Suspension bridges are
flexible things. More than a century ago railroads found them unsuitable. A
steam locomotive is a heavy, moving load—a bad combination for the
continued integrity of both bridge and rail- bound train. On at least one
occasion the flexibility of a suspension bridge proved its undoing. The best-
known American bridge disaster was the plunge of the Tacoma Narrows
suspension bridge into Puget Sound in 1940. The immediate cause wasn’t
 just wind but the violent oscillations that resulted from the interaction of
wind from a particular direction with the deck of this narrow, graceful, but
flexible bridge.13 Collapse took an hour or so as spectators were awed, and a
movie—shown to almost every engineering student since— recorded the
event.

TENSION VERSUS COMPRESSION, AGAIN

Recall that a load stretches the top of a protruding beam, compresses the bottom
of the beam, and shears the middle. How might we design an efficient beam?
One popular solution is the I-beam, so called because its cross section looks like
the letter I. Its excellence depends in part on the similarity in performance of
metals under tension and compression; the identical upper and lower flanges (the
serifs of the 1) represent lots of material well above and below the center line so
they can better oppose the beam’s bending. Vertical “webbing,” which takes the
shear, separates the flanges.

FIGURE 7. 7. A protruding truss. The compressively loaded elements are the darker and fatter ones.

The word “webbing” hints at how one can do a bit better. We can and do
replace it with a latticework of diagonal elements—an actual web. To withstand
shear, the elements of the lattice have to run diagonally, since a set of verticals
would fold very nicely as the protruding beam sagged into a parallelogram;
recall Figure 7.2. When the beam is equipped with this latticework, as in Figure
7.7, we call it a truss, but little has really changed. Consider that lattice more
closely: Half its diagonals are loaded in tension and the other half in
compression. As with the flanges, if we’re using metal, the diagonals can be
identical. But weight might be saved if we use cables for the tensile diagonals.
Of course we’ll lose a little overall stiffness, and we’re limited to downward
loads lest the cables come slack. In fact, we can push the logic a step farther. If
the load is consistently downward, then the top flange always feels tension and
can also be replaced by a cable—again if we accept the loss of stiffness. Struts
and ties thus emerge as identifiable items, solid bars and cables.
Nature uses such specialized beams to support the heads of many mammals,
as D’Arcy Thompson pointed out long ago. As in Figure 7.8, the lower,
compression-resisting flange is provided by the bony centers of the vertebrae of
the neck and thorax and the cartilaginous intervertebral disks between them. The
tension-resisting upper flange is provided by muscle and by a ligament, the one
mentioned in Chapter 2 as having a lot of stretchy elastin. The compression-
loaded diagonals are bony extensions. of the vertebrae, and the tension-loaded
diagonals, running oppositely, are made up of more muscle, tendon, and
ligament.14 The whole thing may be flexible, but being stiff-necked isn’t the
object. The trusses used in human technology, by contrast, rarely make any
distinction between their struts and their ties. In a sense, we construct only struts
and use half of them as ties, accepting a slight loss in weight economy for
simpler construction and greater stiffness.

FIGURE 7.8. Another protruding truss, this one (thoroughly idealized) the vertebrae, intervertebral disks,
muscles, tendons, and ligaments supporting the head of a large mammal. Tension-resisting elements are
shown as lines.
 A horse carries most of its weight on its forelegs. These don’t attach to the
backbone and rib cage any more directly than the attachment between the wheels
and body of a car. Even the best of roads have bumps, and so teeth don’t get
rattled or torsos tossed, proper vehicles need suspensions—springs, at least, and
perhaps dampers (as mentioned in connection with spider silk) to keep the
springs from repeated rebounding. When road and body move closer, the springs
are compressed; when they move apart, the springs are stretched. But both horse
and car have weight, so the springs bear loads even when the road behaves itself
—what’s called preloading. A suspension can be designed so either preloading
and active loading can be either tensile or compressive. A weight (such as a
rider) or an antiweight (such as a pothole) can be made to work either way,
depending on how the suspension’s components are arranged.

FIGURE 7.9. Automobile bodies are commonly supported on compressed coil springs near each wheel,
while horses and other quadrupedal mammals have tensile suspensions; their torsos effectively hang from
their pectoral girdles and forelegs.

Horse and car happen to use opposite loading and preloading. As Figure 7.9
shows, a horse’s body is hung from its pectoral girdle. Sturdy muscles, tendons,
and ligaments descend from the scapulae (shoulder blades, in humans) to the
backbone and rib cage. An increase in weight on the horse raises the tension on
these components above their tensile preloading; compression of the whole
system stretches its springs! Most cars have springs coiled underneath, arranged
so an increase in the car’s weight compresses them. Metal springs work well in
either tension or compression, and cars have been built that (like horses) are
really suspended from their suspensions. Such cars have a mildly disconcerting
way of leaning inward rather than outward on turns, since their masses are
centered below the level of attachment of their suspensions.15 Putting
compressed springs underneath is simply more convenient, given the other
imperatives and constraints on the designer. We make wide use not only of
tensile and compressive springs but of bending, torsional, and even occasional
shearing ones as well—springs that resist every load that’s been mentioned.
Nature’s springs are much less diverse. Only a few, such as the hinge pads of
scallops and other bivalve mollusks, absorb energy by compressing rather than
stretching. Even less often, as in the wishbones of birds during flight, do bones
or exoskeletal components act as bending springs like the leaf springs on some
cars.16 The vast preponderance of natural springs works by stretching.
Deliberate preloading, though, is more than a minor convenience for vehicle
suspensions. Both technologies make wide use of the device to ensure that
materials will be loaded in the ways they withstand best. We put preloading to
particularly good use in what we appropriately call prestressed concrete. We cast
a large piece with steel rods running through it and then tighten the rods, or else
we stretch and hold rods in tension while concrete hardens around them. Either
way, the steel is preloaded in tension, and the concrete in compression.
Stretching the concrete then just relieves preexisting compression and doesn’t
tickle its dangerous weakness in tension. We sometimes get even fancier. Using
curved rods, we can prestress a long beam with a bend that will then be offset by
its weight and live load. Prestressing doesn’t simply provide reinforcing and
crack stopping, as when concrete is cast around rods of steel; it’s a finer thing
altogether.
Just as a material that behaves badly in tension can be precompressed, so a
material that does worse in compression can be pretensioned. Wood is such a
material, and trees do exactly that. The inner portion of a tree trunk, which
bending affects little, is normally compressed by more than just the weight of
wood above it. Additional compression comes from tensile squeezing by the
wood surrounding it. When the trunk bends, the outside of the bend is stretched,
a load wood withstands well. The inside of the bend is compressed, a load wood
takes poorly. But that compression mainly relieves preexisting tension, thus
making better use of wood as a structural material.
 THE FORMS OF TIES AND STRUTS
Cables and bars are unmistakably different, even when both are made of the
same material. The best shape for a tie will rarely coincide with the best shape
for a strut—surely no surprise at this point. At the least, compression elements
should be relatively short, whereas the length of tensile elements doesn’t much
matter. Both technologies recognize the principle, but they apply and
compromise it in different ways.
Suspension bridges, again, give a good view of our best efforts: They’re big,
their builders care about minimizing cost and self-loading, they use both tensile
and compressive components, and most of their components are visible at a
glance. The tensile elements—the main cables and the road- suspending
descenders—are long and thin. The compressive structures, the towers, have a
sturdier look. If made of masonry, as in nineteenth-century designs, the towers
are imposing piles, tall but far from thin. Even if made of steel, they’re much
fatter than even the main cables. Steel towers may be made of latticework grids
or their equivalent in plated tubes. But they work much like solid columns in that
individual struts (or their equivalent in the walls of the plates) are fairly short.
Buckling must be prevented, something that just isn’t a problem with tensile
loading.
Other structures also follow the general rule of keeping individual
compressive elements short. The two-by-fours that form the vertical studs in the
walls of a house may be eight or ten feet long, and individually they’re prone to
side-to-side buckling. But we nail siding, Sheetrock, and various kinds of
paneling to the studs and thus make them into an interactively self-bracing array.
Even flimsy siding can keep the studs from buckling; preventing the initial
lateral movement is easy even if the process gets more forceful further on.
In nature the picture is less clear, even with a great diversity of tensile
elements to look at. Purely tensile structures may be quite thin relative to their
lengths; look at the threads of a spider web or at a hundred-foot- long seaweed.
But compressive structures aren’t always fat—consider the trunks of tropical
trees and the leg bones of gazelles and storks—although compressive structures
are thicker relative to their lengths than are tensile ones. Tall, woody terrestrial
plants have evolved on several occasions. No lineage has made extensive use of
external tensile braces—guy ropes— despite the fine strength of wood loaded in
tension.17 Still more peculiar, to me at least, is how arthropods so often use very
skinny compression elements. At least some members of each major group of
arthropods have what seem to be counterproductively thin appendages: Long,
skinny legs characterize crane flies (insects), harvestmen (arachnids), spider
crabs (crustaceans), some centipedes (myriapods), and sea spiders
(pycnogonids). All are jointed and have muscles inside, so they must be
compressively loaded. Figure 7.10 shows a few of these mechanically bizarre
creatures. Our book of biomechanics still lacks whole chapters!

FIGURE 7.10. Some arthropods that have very long, thin legs: scutigerid centipede (left), harvestman or
daddy longlegs (center top), crane fly (center bottom), and sea spider or pycnogonid (right, considerably
enlarged).

One group of animals deserves special mention. Most sponges, as described

in Chapter 4, are supported by tiny hard spicules of calcium salts or silica
(essentially glass) laced together by flexible pads of a protein much like that of
our tendons. The entire group of animals lacks substantial muscles, so they don’t
need long, stiff struts for their ties to pull against. With that major constraint
relaxed, the general rule about short compressive elements becomes decisive.
Their hard parts, the compression-resisting spicules, are less than a millimeter
long.18 Sponges aren’t all small or weak—individuals three feet high survive
hurricanes and typhoons—but sponges built this way are fairly flexible.
A small jump from sponges leads to something we haven’t yet encountered:
an attractive-looking scheme that’s little used by either technology. Where struts
and ties are distinct elements, we expect the struts to form a continuous,
interconnected network with ties inserted here and there. How else to resist
anything but purely tensile loads? In fact, the continuously interconnected
network of struts isn’t crucial. Structures that resist bending, twisting, and
compression can be made in which the continuous array—one hesitates to call it
a framework—consists only of ties. The struts need make no contact with one
another. Whether such a structural scheme was invented by a specific person
isn’t clear, but R. Buckminster Fuller took out the key patents and named the
scheme tensegrity.19 In rich language for a patent, he spoke of “islands of
compression in a sea of tension.”
A picture such as Figure 7.11 serves better than words for explaining how a
mast or tower, for instance, can stand up without interconnected struts. The key
is putting ties where pushing, bending, or twisting the structure as a whole
increases their tensile loading. Ties, after all, resist only tensile loads and
collapse helplessly under any other kind. Masts and towers aren’t the only
possible tensegrities; domes, for instance, aren’t hard to design. One can, I’m
sure, make a fabric tent with noninterconnecting pockets for stiff battens;
perhaps it has already been done. The main disadvantages of structures based on
tensegrity are those of tensile construction in general: the lack of stiffness and
the dependence on fasteners. So tensegrities remain mainly an art form.
Sponges may make some use of the concept; at least their spicules most
often don’t interconnect directly. But the arrays of spicules bear only limited
resemblance to Fuller’s designs, probably because even tensegrities are stiffer
than optimal for sponges—animals that accept rather than minimize flexibility.
Spicular skeletons—that is, tension-resisting tissue with small pieces of hard
material embedded in them—aren’t limited to sponges. They occur in some soft
corals, sea cucumbers, stalked barnacles, and elsewhere. These look even less
Fulleresque and more like steps in a continuum between deviant tensegrity
structures and composite materials.20
FIGURE 7.11. A tensegrity tower or mast. The strutted tetrahedrons don’t touch except through sets of ties,
yet the thing will stand erect. Imagine pulling on a rope and raising the flagpole as well as the flag; a
tensegrity tower permits such intuitive preposterousness.

My best guess about why tensegrity hasn’t been elaborated by nature is as

much of an evasion as those guesses I gave for the absence of metals. Where
stiffness is important, tensegrities are too flexible. Where flexibility is tolerable,
something better is available. We turn, then, to the ultimate expression of tensile
support, hydrostatic systems.

FLUID STRUTS AND HELICAL TIES

Something close to home for half of us. Consider the penis of yourself or your
nearest and dearest; in an immediately evolutionary sense fitness depends on
stiffness. Discontinuous struts were odd enough, but here’s a stiff structure with
no solid strut at all! Does a strut really have to be solid to resist compressive
stress? In fact, it doesn’t. Consider how the three ordinary states of matter
behave. Solids resist tension, compression, and shear; liquids resist tension and
compression; gases resist only compression.21 Thus all three states of matter
fight back—they resist compression—when pushed into a corner. So a fluid,
whether liquid or gas, might serve as a strut. That sounds peculiar, but maybe we
just haven’t taken a sufficiently broad view of possible struts. Fluid struts would
certainly be economic; water and air are boundless resources. How, then, can we
make them?
Making a fluid strut is simple. A tension-resisting sheath need only be
wrapped around a body of compression-resisting fluid to get a structure that has
a discrete shape as well as appreciable stiffness, strength, and so forth. We’re
talking about nothing more than a balloon of air or a canvas fire hose of water.22
Not to prolong any suspense, we’re also talking about blimps and buildings
supported by air pressure as well as about lots of worms and a host of cells—
plus plenty of penises. Of course the character of the structure depends a lot on
whether it’s filled with gas or liquid. Water is eight hundred times denser than air
and much more resistant to compression. A water-filled balloon has an almost
constant volume whatever the load, but its shape reflects gravity’s downward
pull. The shape of an air-filled balloon cares little about gravity, but its volume
varies more. We lack a single word for such inflated, pressurized, balloonlike
structures: If water-filled, they’re hydrostatic; if air-filled, they’re aerostatic.
As practical pressurized bodies, rubber balloons have serious limitations. For
one thing, rubber isn’t too stiff, so the whole structure ends up a bit flabby. For
another, rubber’s stress-strain plot is nearly straight rather than concave upward.
As a result (and as brought up in Chapter 5), balloons are prone to aneurysms; a
cylindrical one doesn’t inflate uniformly. Furthermore, inflation stores up a
potentially dangerous amount of elastic energy; a balloon fails violently. Using a
less stretchy wall material improves the situation. But practical and versatile
aerostats and hydrostats depend on complex, multicomponent walls containing
tension-resisting fibers. These most often run not lengthwise or in circular rings
around the cylinder, but in helices, as in Figure 7.12
The two technologies use this general scheme in different ways and to
different extents. Humans make relatively limited use of it, and a gas usually
serves as the fluid of choice, as in air-supported buildings. Sometimes, though,
we fill the inside with liquid, as in collapsible fire hoses and fiber-reinforced
storage tanks. Nature uses the device extensively, and her fluid of choice is either
water or some muscle or mucus. Here and there she does use gas, as in various
floats, such as that of the notorious Portuguese man-of-war.
While the classic zeppelins had rigid frames and internal gas bags, modern
blimps are proper aerostats, with the helium inside doing structural service as the
compressive counterpart to the tensile skin. Walking into an inflatable building
through the antechamber—a real breezeway— may feel a little strange, but you
can’t feel any increased internal pressure once you are inside. All the inflatable
rubber tires of our vehicles do the same thing on a smaller scale but with higher
pressures. But none of these applications is both common and large. Hydrostats
are scarcer still. Besides the hoses and tanks already mentioned, a few barges for
carrying liquid cargo have been built as tensile bags, as have the skins of large
rockets with internal fluid propellants.23

FIGURE 7.12. If a cylinder that's wrapped with helical fibers is compressed, the fiber angle increases; if it
is stretched, the fiber angle decreases.

Nature may rarely inflate with air, but her water-filled hydrostats are
wonderfully diverse in appearance and function. In addition to cases already
mentioned, they provide the main stiffening for the tiny tube feet of starfish and
sea urchins. They add stiffening to that provided by the skeleton in swimming
sharks. And they contain the pressurized water that a squid squirts in its jet
propulsion system. To understand how these schemes work, we need to look at
how the volume of a cylindrical hydrostat changes with alterations in its length,
as in Figure 7.13. At one extreme the cylinder is squashed into a volumeless disk
with the fibers running circumferentially, while at the other end it’s a volumeless
line with the fibers lengthwise. The cylinder has its greatest volume almost
midway between, where reinforcing fibers run helically at a fifty-five- degree
angle—a little more circumferentially than lengthwise. At that angle an increase
in internal pressure has the least chance of increasing the volume of the cylinder
and an equal tendency to make the cylinder get longer or fatter. That specific
angle represents a divide between two ways that nature uses hydrostats.

FIGURE 7.13. The relationship between volume and length as fiber angle (noted on the curve) is changed.
The helical fibers of the hydrostat are assumed to be inextensible. Beneath the curve the hydrostat is limp;
above the curve it has exploded.

Consider how a squid squirts a jet of water. It has an outer mantle with
muscles running almost circumferentially around its girth and helical fibers
running more lengthwise than circumferentially. Contracting these muscles
might make the mantle get longer (as when you squeeze a cylinder of clay or
dough), but that would make the fibers run even more nearly lengthwise, moving
the system to the right on the graph and reducing its volume. Water doesn’t
compress; instead it squirts out as a jet, and the squid is suddenly elsewhere.
Alternatively consider a limp worm that wants to burrow. If it’s a fairly
simple, unsegmented worm, fibers in its outer cuticle run almost
circumferentially. Contracting lengthwise muscles might make the worm get
fatter, but that would make the fibers run even more hoop-wise, moving it to the
left on the graph and reducing the worm’s volume. Since the worm is closed,
what happens is that the muscle contraction greatly increases its internal
pressure, stiffens it, and facilitates burrowing. Contracting muscle on one side of
the body curves as well as pressurizes it, so the worm can move by manipulating
its hydrostat.24 That’s good for the worm, although perhaps not so good for us;
we’re explaining (among other things) how some parasitic worms can make their
way through our flesh.
Another application of this mechanism doesn’t involve a passive core of
compression-resisting water, blood, and guts but instead uses active muscle
throughout. Muscle of course can only shorten, so all muscular devices need
some way to extend—either initially or when getting reset for their next pull. If a
cylinder has a constant volume, then decreasing its diameter must make it
lengthen. That’s how our tongues, the arms and tentacles of squid and octopus,
and even the trunks of elephants manage to extend.25 These “muscular
hydrostats” are versatile. Muscles can be arranged so the overall cylinder
extends relatively slowly and forcefully for a short distance (as does an
elephant’s trunk) or relatively rapidly and less forcefully for a longer distance (as
does a squid’s tentacle). We’ll defer more specific explanation until we consider
the general topic of leverage. The general theme is what matters here; resisting
compression, what struts do, doesn’t require solid materials.
Two final questions about aerostats and hydrostats. First, why does nature
make elaborate and creative use of hydrostatic support in aquatic systems but
rarely use them in terrestrial ones? While hydrostats can’t be made without
water, even terrestrial creatures are full of the stuff. Our tongues and penises,
some plant cells and small stems, and a few other cases don’t come close to the
several phyla of worms, to snail and starfish feet, to all cephalopod arms, and to
sharks. I suspect, and the suspicion may be relevant to human use of the
mechanism, that gravity poses a heavy problem. Hydrostats can’t achieve great
stiffness, penises notwithstanding, a serious disability where gravity dominates
and organisms must extend above the ground. Tree trunks and long bones, for
instance, must not be prone to bending if trees and mammals are to remain erect.
The second question concerns one of those odd omissions in nature. She
makes no blimps—no lighter-than-air craft. Flight by heavier-than-air craft
doesn’t come cheaply, dispersal matters in nature’s scheme of things, and a
lighter-than-air flier should be splendidly well suited for completely passive
dispersal. Is making hydrogen beyond nature’s ability? Unlikely—
photosynthesis in every green plant starts by splitting water into oxygen and
hydrogen. Every one of our cells continuously removes hydrogen from fat or
carbohydrate as it rechannels stored chemical energy. My best guess, but one
lacking great conviction, invokes scaling. Nature starts small and perhaps finds a
tiny blimpish seed or fruit hard to make or not too useful. To keep the thing
buoyant, wall thickness must not exceed some fixed fraction of diameter, so a
very small blimp must have a very thin wall, raising difficulties of gas
penetration and mechanical integrity.

This book uses natural technology to provide perspective on our own. Going
beyond a scientist’s analysis to an engineer’s synthesis can aid in gaining that
perspective. As an exercise in creative synthesis, you might try devising (on
paper) an alternative technology based on ropes, hydrostats, and other tensile
schemes. Imagine a culture whose solid materials resist only tension and not
compression, a culture in a world in which compression can be resisted only by
fluids, either gaseous or liquid. How might buildings, vehicles, furniture, and
other everyday structures and machines look and work?
 Chapter 8

ENGINES FOR THE MECHANICAL

WORLDS

Up to now we’ve mainly looked at static systems—at the geometric, material,

and structural aspects of buildings, bridges, and their living counterparts. We
turn now to dynamic systems, ones that move themselves or their parts. Whether
they are living or not, the same basic principles must apply, but once again the
practicalities will prove remarkably different.
Making moving machinery involves two or three kinds of component. First,
energy must be fed into the system. Neither natural nor human technology really
makes energy in any normal sense, but instead each manages to get it from
external sources. Bending the ordinary definition a little, we’ll label as “engines”
all devices that feed nonmechanical energy into mechanical systems. These
engines use forms of energy, such as heat and electricity, to push, pull, swell,
shrink, bend, turn, or slide. Second, the mechanical energy has to be applied to a
specific task, such as galloping, grinding grist, or gathering gold. Stretching
another definition, we’ll call all the requisite coupling devices, from simple
cables and shafts to complex pulleys and gears, transmissions. Third—at least
sometimes— mechanical energy is stored for a while between being generated
by an engine and being used to move something around. The energy may be
supplied in pulses (as with the separate firings of an engine’s cylinders), or its
task may be intermittent (as with any back-and-forth saw) or it may be needed in
more concentrated packets (as when the slowly developed momentum of the
hammer suddenly assails the nail). We’ll call all such energy accumulators
batteries. The focus of this chapter will be engines; of the next, transmissions
and batteries.
Both technologies use energy for purposes other than moving matter around:
for such things as biological growth and the synthesis of useful chemicals, for
heating bodies and buildings, and for communications with nerves or with wires.
The big bang, where it all began, left aside, the ultimate source of almost all this
energy is the sun. Living systems capture it through the photosynthetic activity
of plants and bacteria. Human technology gets most of its energy from the same
process, preferring, though, the well-aged product, fossil fuels. Both
technologies make a little use of aero- and hydropower, ultimately just different
pathways for obtaining solar energy. We get some additional energy from
nuclear reactions and a very tiny bit from geothermal sources, neither of which is
a significant source for nature. Finally, we co-opt nature’s engines themselves—
our own muscles and those of our domestic animals. Figure 8.1 puts all this
energy transfer in diagrammatic form. It should point up two things: the central
position of photosynthesis for both technologies and the greater diversity of
processes used by humans.
Another digression: Talking sensibly about motility requires a few additional
terms in their scientific senses. So then . . .

Work. It gets done when something that resists being moved is nonetheless
moved some distance. Quantitatively it’s the force that’s used times the distance
the thing moves. If you hoist a ten-pound something or other a foot upward,
you’ve done ten foot-pounds of work. The resistance you’re working against is
the gravitational attraction between earth and weight. Lifting five pounds two
feet does the same amount of work, irrespective of how the lifting is done. The
same work is done on the object if, with a lever, pulley, or windlass, you exert
only one pound while moving your end of the device ten feet. Only force times
distance matters, and both must be in the same direction. Lever, pulley, and
windlass are just transmissions.
FIGURE 8.1. How energy gets into the mechanical domains of the two technologies. The thicknesses of the
connecting lines give a crude indication of the relative magnitude of the pathways.

Energy. This is mysterious stuff. But the standard evasion, “the capacity for
doing work,” will do for our mundane purposes. What matter here are two
points. First, energy and work are nearly the same thing, and we measure both on
a yardstick marked with the same units: foot-pounds, or calories, or (in official
scientific usage) joules. Second, the notion of energy permits statement of some
general rules about our physical universe. The first is what’s called a
conservation law: In any real process energy is neither created nor destroyed but
is at most shifted from one form to another. When we say that energy is “used,”
we really mean that it’s transformed from a potentially useful form (the
electricity in a battery or the high water behind a dam) into some less useful
form (heat, in particular). Conservation laws serve us so well that a quantity such
as energy is worth inventing just so we can state one.
Power. It’s simply how fast a process uses energy or does work—energy or
work divided by time. We measure it in horsepower, or foot-pounds per second,
or (officially) watts. Quite straightforward, an engine’s output power is how fast
it does mechanical work, and a more powerful engine takes less time to do the
same work. The elevator with the bigger motor gets you up in shorter order.
Power can be viewed another way too: Force times distance divided by time is
the same as force times speed. So, just as a certain amount of work can be done
with any combination of force and distance, so a certain amount of power can be
expended in any combination of force and speed. Things that swim and fly care
about this last trade-off. One can invest an airplane’s fuel in big, slowly turning
propellers or small, rapidly spinning jets.

Efficiency. Even though the notion isn’t complicated, we need to be careful to

use it precisely. Efficiency is what you get out of some device divided by what
you put in. It’s the weight of the elevator times the height it ascends divided by
the energy consumed by its motor. If work or energy is the currency (energetic
efficiency), then that conservation law just mentioned limits the quotient to one.
You can’t get an efficiency over one, or 100 percent. In short, you can’t get
something for nothing, or shorter, there’s no free lunch. In our real world all
mechanical devices have energetic efficiencies below 100 percent, which means
that some energy ends up where it does you no good. If you run, you get hot, and
the engine of your car needs a radiator to dispose of heat.

The main measures of an engine are power output (or power relative to
weight) and efficiency. Relevant also, if a little less so, is how the power output
of an engine changes as its operating speed changes. For instance, many of our
electric motors deliver proper output with decent efficiency only within a very
narrow range of speeds. The operating speeds of automobile engines, while less
critical, are still so limited that the engines need complicated and costly
transmissions. Old-fashioned steam engines were much more tolerant; steam-
driven cars and locomotives connected their engines directly to their wheels.

THE VARIETY OF ENGINES

Nature and humans have engines of great importance and engines with minor
roles. For contemporary humans the main players—the prime movers, so to
speak—are heat engines. These include all the external- and internal-combustion
engines that burn fuel to make steam or that move pistons or turbines directly—
almost all the engines we use to move our cars, boats, and airplanes. Nuclear
power must be included here since it gets fed into our systems by transferring
heat. A nuclear power plant uses heat from its reactor to generate steam, which
then does the same job as the steam in a coal-fired plant. Our second team
consists of electric motors, although we don’t take electricity as such from the
environment, so these aren’t quite analogous to heat engines. Instead they’re
paired with generators as systems to transfer power from generating plant to
kitchen. Electric motors serve the same role as the array of overhead shafts and
descending belts that, in a nineteenth-century stream-side factory, drove all the
machines from a common waterwheel. Beyond heat engines and electric motors,
the list of engines grows long, but the overall contributions are slight:
hydroelectric plants, windmills, tidal and wave-energy plants, and so forth.
For most of human history, the principal engine has been muscle— first
human and then, increasingly, animal. To be useful, animals must be
domesticated, trained, and their power harnessed toward plowing, hauling, or
whatever else we want them to do. Harnesses—transmissions, essentially—were
rudimentary before the Middle Ages. Waterpower—as waterwheels—played
some role in the classical world, but wide use of wind power, except in the sails
of boats, is more recent. By the Middle Ages both windmills and waterwheels
were common.1 Hero (or Heron) of Alexandria invented a steam engine in the
first century C.E. but the design wasn’t auspicious for practical applications, as
will be clear when we look at jet engines specifically. Thus our major engines
are recent. Practical heat engines appeared in the eighteenth century, and electric
motors came into wide use barely more than a century ago.
Nature’s engines are of no greater fundamental diversity. The most familiar
is muscle, as noted earlier, about 40 percent of the weight of a typical mammal.2
We also use motile cilia, microscopic cylinders that can actively bend, for such
things as propelling sperm and moving mucus in our respiratory passages. Such
cilia are widely used by animals both big and small for locomotion, pumping,
and other purposes. (Curiously, motile cilia are absent in the entire phylum
Arthropoda—insects, crustaceans, arachnids, centipedes, and such. Recall the
comments in Chapter 2 on the limitations imposed by one’s lineage.) Beyond
these relatively rapid major engines loom some slow but not insignificant ones,
at least if we relax our zoocentric sense of speed enough to give plants proper
attention; over its lifetime a corn plant or a tree lifts many times its own weight
of water from soil to leaves. In addition, a few engines extract power directly
from moving air and water, doing what windmills and waterwheels do, if in
different ways.

ENGINE OUTPUT,WATTS/KG OUTPUT,HP/LB

18th-century steam pump 10 0.005
 Cilia 30 0.02
Skeletal muscle 200 0.12
Electric motor 200 0.12
Automobile engine 400 0.24
Motorcycle engine 1,000 0.6
Aircraft engine, piston 1,500 0.9
Aircraft engine, turbine 6,000 3.6

TABLE 8.1. The power output relative to mass of a variety of engines.3

If our criterion of excellence is power relative to mass, nature’s best engines

match only low-level examples of human technology. By this measure, modern
heat engines are truly superb. Table 8.1 compares the power output (watts or
horsepower) per unit mass (kilograms or pounds) for some living and nonliving
engines—devices we rarely consider at the same time.

COMBUSTION ENGINES
At some risk of oversimplification, heat—that is, combustion—engines can be
divided two ways. Combustion may be external, with some working fluid
carrying energy from the heater into the engine proper, as in all steam engines,
whether piston-driven locomotives or modern steam turbines. Or it may be
internal, as in our automobiles, where the burning of fuel in the cylinders
generates the pressure that moves the pistons. The motion may be intermittent
and reciprocating, as when pistons go to and fro. Or it may be continuous and
rotational, as in all turbines, whether the steam-injection ones of power plants or
internal-combustion jet engines.
Heat engines are unique to human technology, and the obstacles to their
development were formidable. No natural analogs provided hints for their
design, understanding of the underlying science came slowly, and metallurgical
and fabricational limits long precluded steady operation at high temperatures.
The earliest practical combustion engines were about what one might guess.
Thomas Newcomen’s mine pump (Figure 8.2), popular during much of the
eighteenth century, was an external-combustion piston engine using steam at
atmospheric pressure as the working fluid. Motion of the piston was generated
not by expansion of steam in a cylinder but instead by the press of the
atmosphere on the other side of the piston when a spray of water lowered the
temperature and made the steam condense. Thus especially high temperatures,
good pressure-resistant fittings, and carefully machined crankshafts were not
needed; the engine made modest technological demands. Conversely, the low
pressure difference (one atmosphere at most) demanded huge pistons to produce
much power, and wasteful heating and cooling of the cylinder’s walls guaranteed
very low efficiency. Still, these huge, slow machines gave fine service,
especially for pumping water out of coal mines, where fuel supply was not an
issue.4
Nineteenth-century steam engines were much better machines, efficient
enough in weight and fuel consumption to move themselves and some payload
around—at least on hard rails and with gentle grades. James Watt, late in the
eighteenth century, took advantage of improved metals and better machining to
make steam push pistons instead of condensing the steam in the cylinder and
getting only the push of the atmosphere outside. His engines were better in other
ways as well. Paired steam inputs pushed the pistons in both directions,
incoming water was preheated with waste heat, the cylinders were kept hot by
condensing steam outside, and piston rods turned wheels rather than just rocked
arms. Watt and his contemporaries conceived of the steam turbine—after all, the
principle is simply that of windmill or waterwheel—but both the precision
needed to construct such an engine and the availability of transmissions adequate
for its high speeds delayed its practical realization until the present century.
FIGURE 8.2. The main components of Newcomen’s steam engines of 1712 and thereafter. A huge rocker
arm connected the piston’s chain and the pump’s chain. The power stroke of the piston was the downward
one as steam was condensed by water sprayed into the cylinder. The counterweight then pulled the piston
back up while more steam was let into the cylinder. In the early versions a person working the controls
alternately sprayed in water and admitted steam.

The story of internal-combustion engines is a similar tale, needing only a

time shift of fifty or a hundred years. Piston engines came into wide use toward
the end of the nineteenth century, and turbines around the middle of the
twentieth. What must be emphasized, again, is the novelty of this whole
technology next to anything in nature. Only the fuels are similar: hydrocarbons
resulting from the biosynthetic activities of organisms. Metals, high
temperatures, gases at high pressures, the large size of early engines, the high
speeds of modern ones, pistons, crankshafts, flywheels— none has a close
natural analog. Even the basic mode of operation of heat engines is unnatural;
these engines work by either pushing (stretching the point just a little for
Newcomen’s atmospheric engine) or rotating.
Why doesn’t nature use anything like our heat engines? Because the
fundamental rules that govern heat engines are alien to the living systems on our
planet. And why is that? A heat engine needs not only a high temperature but
also a difference in temperature—both a hot source and a cooler sink. Down this
temperature gradient energy will flow, as when hot coffee cools in air or warm
water melts ice. The temperature difference sets an absolute limit on how good a
heat engine can be. The basic rule is simple, but it does require that we use a
temperature scale with a real zero point—the coldest possible cold. To convert a
Celsius (centigrade) scale to one with a real zero, just add 273°; to do the same
with a Fahrenheit scale, add 459°. Using such a scale, the maximum efficiency
equals the sink temperature divided by the source temperature with the result
subtracted from one. (Multiply by 100 to get percent efficiency.) Thus one could
get perfect efficiency (1.0 or 100 percent) only if the sink were at an unrealizable
-273° C or -459° F or the source at an unthinkable infinite temperature. For a
steam engine, the hotter the steam put in and the colder the steam finally
released, the better this underlying thermodynamic efficiency.5
Compare what our technology does with what nature might do. We can
easily use a source at 1,000° C (1,800° F) and a sink at the boiling point of
water, 100° C (212° F); thermodynamic efficiency is then 71 percent6—not at all
bad. Nature might use a source at 40° C (104° F) and a sink of 0° C (273° K),
about the maximum range of temperature tolerated by active animals. If so, the
efficiency of her engine will be less than 13 percent. For a more easily realized
ten-degree difference, ideal efficiency drops to a disastrous 3 percent. And these
maximal thermodynamic efficiencies always exceed those of real devices.
Incidentally, nature’s nonuse of heat engines carries a message that’s familiar
to almost every engineer but to few of the rest of us. Might we use the
temperature difference between surface water and deep water in lakes or oceans
to tap an unlimited and benign energy source?7 Yes, the energy is there and
available; after all, a temperature difference indisputably exits. From the rule
about energy conservation one can easily calculate that monumental quantities of
energy are available. But the result is monumentally misleading. What must be
factored in is the thermodynamic tax imposed by the small temperature
difference; the energy used to run such things as pumps would most likely be
greater than the energy obtained from the underlying resource.

ELECTRIC ENGINES AND GENERATORS

The other major engine of contemporary human technology is the electric motor.
Practical versions range from microdevices a fraction of an inch in length to
ones weighing many tons that drive the propellers of large ships. They perform a
wide range of tasks with remarkable effectiveness and reliability. My not
especially high-tech household has more than sixty small electric motors. Try
such a count yourself, including such easily forgotten ones as the fans,
compressor, and defrost timer of the refrigerator; you’ll be surprised at how
many you happily harbor. Electric motors are more efficient than combustion
engines and don’t get too hot when running. Still, their high efficiency can
mislead us since (as noted earlier) most electric motors just provide output
devices for large heat engines, much like multiple dumb terminals for a
mainframe computer. Their heat engines are of course the fossil fuel and nuclear
power converters in generating plants. (Hydro- and wind-powered generators
make only a small contribution to our production of electricity.) Thus a full
accounting should consider overall efficiency, from fuel to mechanical power
output, including substantial losses in the electrical generating plant and
transmission lines. We’ll come back to that shortly.
Once again we’re looking at a distinctly human class of engines. Electricity
is common among organisms; every cell has a charge across its outer membrane,
for instance, with potentials approaching a tenth of a volt. Not too many cells in
a series would give quite a good voltage, nor would all that many, working in
parallel, deliver an eventful current. Several kinds of electric fish modify
muscles into arrays with series and parallel connections that deliver stunning
performances, reportedly up to 650 volts.8 While that’s an eclectic habit, it
shows that producing electric power in quantity takes only minimal alteration of
normal tissue and ordinary metabolic chemistry. Our muscles, even those of our
hearts, may be electrically controlled, but they’re not electrically powered.
Why doesn’t nature build electric motors? Must these motors use unnatural
wheels and axles and engage in unnatural rotational motion? Probably not.
While all familiar electric motors rotate, rotation simply serves a technology that
finds the wheel and axle easy to make and likes the versatility of belts and gears.
Every kind of rotary electric motor has its linear or reciprocating counterpart.
Linear engines have been developed for driving trains; the track forms half the
motor. Short-stroke repetitive pullers (solenoids) do chores such as opening and
shutting the water lines and drains of our washing machines; turn on the
electricity, and a metal core moves forcefully through a coil. The earliest electric
motors of Joseph Henry, in the United States, and Charles Wheatstone, in
England, in the first half of the nineteenth century were reciprocating devices.
One early reciprocating electric motor drove a pump the same way a Newcomen
steam engine did.9 Closer to home, motors in which a piece of metal vibrates
rapidly back and forth sometimes power electric razors, massagers, and
reciprocating sanders.
A more likely obstacle to living electric motors is that dullest of components’
wire. Good conductivity comes hard without metals, as noted two chapters ago.
The salt solutions of cells don’t come close. For instance, a strong solution of
potassium chloride (seventy-one grams per liter, a so-called one molar solution),
an especially conductive brew, is still nine million times less conductive than
copper.10 Equal performance would require that a copper wire a mere tenth of a
millimeter across be replaced by a pipe of potassium chloride at least a foot in
diameter. Nothing remotely appropriate to get power from generator to rotor
exists in nature’s armamentarium; electric motors are mainly practical in a
metallic technology. Neuromuscular systems use electricity in a way peculiarly
adapted to or limited by (take your choice) their low conductivity. In a nerve,
electrical currents flow only for short distances. Long-distance transfer of
information takes place by an odd scheme in which a local electrical event
incites a similar event adjacent to it—the way a wave travels across the ocean
without moving any bit of water very far.
Electric power’s popularity comes from its handiness as a transportable
intermediate form of energy. We generate electricity in one place and then
transmit it for long distances at (to reduce power losses) several hundred
thousand volts. For safety and convenience, we then reconvert the high-voltage
electric power to lower voltage (240 and 120 volts, mainly) near the sites of
use.11 We vertebrates do much the same thing to power our muscles, although
we do it chemically rather than electrically. We move the sugar that results from
carbohydrate digestion through a special set of veins to our livers, which store it
(as a polymer, glycogen) and meter it out to the rest of the circulation.12 In
muscle cells the sugar charges up a more immediately usable energy transfer
system: Breakdown of sugar to either lactic acid (short term) or carbon dioxide
and water (in sustained activity) converts adenosine diphosphate (ADP) to
adenosine triphosphate (ATP). This last then feeds power into the actual
muscular motor. The coupling of sugar breakdown to ATP synthesis in muscle
has about the same function as the conversion of high- to low-voltage electricity
in an electrical transformer.

WIND AND WATERPOWER

First, we should remind ourselves that getting power takes more than just air or
water in motion. A windmill on a freely floating balloon won’t turn; the balloon
moves with the wind, and no apparent motion of the earth beneath helps at all.
Having its feet on the ground matters as much for a windmill as having its rotor
in the air. Extracting power requires a device that experiences a difference in
speed: between air and ground for a windmill, between water and ground for a
waterwheel, or between air and water for a balloon towing a submerged turbine.
The requirement has the same basis as that for two poles on a functional battery,
for different water levels on the two sides of a power-producing dam, and for a
hot source and cold sink in a heat engine. It’s incumbent on both our law-abiding
technologies.

FIGURE 8.3. Three kinds of horizontal-shaft waterwheels.

The classic water-powered engine is the waterwheel, of which Figure 8.3

shows several useful versions. Water runs over the top in overshot wheels, where
the weight of the water (gravitational energy) pulls one side down. The water’s
motion (kinetic energy) pushes paddles on the bottom in undershot wheels.
Water enters in the middle, tapping both resources, in breastshot wheels. Less
commonly, a stream of water may fall against inclined paddles around a wheel
with a vertical shaft.13 None of these arrangements achieves great efficiency,
and for at least a century they’ve been superseded by various kinds of turbines—
rapidly turning rotors concealed within the ductwork of large hydroelectric
plants. Small-scale hydropower as used by old grain and sawmills has been
largely abandoned, as a result both of the widespread availability of electricity
and good electric motors and of the change to the large-scale production of flour
and lumber.
The classic wind-powered engine is the windmill, of which Figure 8.4 gives
representative designs. Windmills are more recent devices than waterwheels for
the same reason that internal-combustion engines followed external-combustion
ones: The technological problems are more challenging. To develop much useful
power, the rotors of windmills must be large, but if large, they become
vulnerable to storms. The speed of atmospheric winds varies a lot more than
does the pressure of water stored behind a dam. The basic setup is also less
forgiving for primitive devices; a waterwheel needn’t be completely immersed in
water, but a windmill must work entirely in air.
Almost all windmills use one or the other of two physical mechanisms.14
The oldest ones had vertical shafts and turned in a horizontal plane. Like
whirling cup anemometers, they turned in winds coming from any direction. But
turning depended on blades that had higher drag when facing into the wind than
when facing away from it, a sure recipe for inefficiency. A recent version is the
Savonius rotor, which enjoyed a brief countercultural fashion since it could be
made from a metal drum sliced lengthwise.15 By contrast, most windmills of the
present (second, not third) millennium have horizontal shafts, with a set of
propellerlike blades that rotate in a vertical plane. While the blades are useful
during their entire circuit, the whole structure must turn to face the wind.
Aerodynamically they work by producing lift, a force at a right angle to the wind
direction, rather than depending on the difference in the drag of bodies facing
into and away from the wind. Generating lift is a subtle business whose physical
basis wasn’t understood until the early part of this century. (The airplane’s
demand for really good wings and propellers of course provided the impetus.)
Not surprisingly, old windmills were aerodynamically poor, and the drag on the
blades and their supports was unnecessarily high relative to the power that could
be extracted.16
FIGURE 8.4. Windmills of various designs. Above are three with horizontal shafts: classic Dutch, American
farm, and a modern high-efficiency design. Below are two less conventional ones with vertical shafts:
Darrieus turbine and Savonius rotor.

All these waterwheels, turbines, and windmills use wheels and axles. So
none of this sounds like a technology that has natural analogs, and in a strictly
mechanical sense it doesn’t. Nevertheless, it does have energetic analogs.
Sometimes nature draws power from speed differences between the ground and
flows of air or water. We know of a number of cases and several different
arrangements.
Consider, as in Figure 8.5, a wind across the two openings of some U-shaped
pipe beneath the ground, a pipe whose openings differ in height. The higher
opening will ordinarily be exposed to faster flow, and, by Bernoulli’s principle,
the faster the flow, the lower the pressure. Since any fluid—gas or liquid—left to
itself flows from high pressure to low pressure, then flow within the U-shaped
pipe will go from low opening to high opening—irrespective of the direction of
the external flow driving it.17 The arrangement is used by the prairie dogs of the
North American Great Plains to ventilate their long, deep, multiapertured
burrows. I was involved in investigating the phenomenon in the early 1970s. At
the time I thought that the ventilation supplied the animals with oxygen, but now
I’m less sure; it’s far better than necessary for that purpose. More likely, airflow
through the burrow gives the animals the scents of what’s above; the scheme is
an olfactory sensation. With water as the moving fluid, various worms and
crustaceans may use an analogous arrangement to irrigate their burrows in sandy
substrata beneath shallow bays.
If air or water moves across an elevation, normally the air or water will flow
fastest at the peak; ridges are more windy than valleys. If pipes or a porous
medium (such as sand) connect a valley with a peak, as in Figure 8.6, then air or
water inside will move from one to the other. Again, the direction of the external
flow makes no difference, and the physical basis remains Bernoulli’s principle
(plus some secondary effects). This second arrangement makes wind draw air
and thus oxygen through giant termite mounds in the open country of East
Africa. Similarly, sponges, which live by filtering microscopic organisms from
seawater, use the flow of water around themselves to reduce their cost of
filtration. That improves the (net) profit that determines if they’re viable in a
particular place. And slotted sand dollars form slight elevations on sandy bay
bottoms, so water flowing across them draws water and edible particles up
though their slots from interstices in the sand.

FIGURE 8. 5. A scheme for using ambient flows of air or water to induce a secondary flow in a burrow or
other passage through the substratum, and a prairie dog viewing the world from the crater-shaped opening
on one end of a burrow that uses the scheme for ventilation.
FIGURE 8.6. Another scheme for using ambient flow to induce a secondary flow, this one through some
elevated structure; and a keyhole limpet, which uses the scheme to draw water in beneath the edge of its
shell, over its gills, and out the apical hole.

In a still simpler arrangement, shown in Figure 8.7, one opening of a pair is

directed upstream and the other faces crosswise to the flow. This third
arrangement depends, though, on proper orientation with respect to the external
flow. At least one kind of insect larva does it; some caddisflies that live in
streams make appropriate tubes in the substratum and equip the tubes with catch
nets. In addition, many fish that use nostrils in interconnected pairs put enough
of a hood on the front one so it faces upstream. Since streams flow downhill and
fish swim forward, orientation to flow presents no problem in either case.
We know a few other schemes by which nature takes advantage of flows
across solid surfaces to obtain a little power. Some seabirds, for instance,
repeatedly make long vertical loops without beating their wings. Gliding
alternately high and low is a way to use altitudinal velocity differences to stay
aloft without expending energy.18 For that matter, we sometimes ventilate
simple buildings, such as Indian tepees and partially buried mound houses, with
the same tricks as sponges and prairie dogs. Mine ventilators often use reversible
fans to take advantage of any wind-induced flows. Such flows will occur in all
tunnels that have multiple openings except where the external wind is the same
over all openings, the openings are geometrically identical, and the land is
perfectly flat.19
FIGURE 8.7. A more potent scheme for using ambient flow to induce a secondary flow, but one no longer
independent of the direction of the ambient flow. This one drives water in the mouth, across the gills, and
out the operculum of many fish when they’re swimming rapidly.

All these cases use wind or water currents to drive internal flows of air or
water. One flow induces another, which takes a minimum of transducing
machinery. However interesting as bits of natural history, they’re insignificant
next to photosynthesis as a way by which nature gets energy from the
environment. Moreover, none provides energy in a storable form, the most
important feature of photosynthesis.

MUSCLES AND CILIA

By this point the reader may wonder whether nature always comes off second
best. For one type of engine, though, nature has the only team on the field. I refer
here to what are coming to be called molecular motors. Human technology
works large, dealing with materials in bulk and using such processes as casting,
pressing, slicing, dicing, and crushing. We struggle mightily to make ever tinier
components for our complex electronic devices so they can be powerful without
getting dysfunctionally large. Nature works the other way; hers is a microscopic,
even submicroscopic technology aggregated with some minimal integration into
larger systems. Molecular motors, if a value judgment is permissible, show
nature both at her finest and at her most distinctive. Two of these—muscles and
cilia—drive her fastest motions.
FIGURE 8.8. A cilium cut across (imagine a sliced strand of cooked spaghetti), showing the typical set of
nine doublet microtubules surrounding a pair of singlet microtubules.

Cilia (and flagella, any distinction being unimportant here) of a fairly

stereotyped kind occur in a wide variety of animals and plants.20 They’re tiny
hairlike machines that push stuff around by waving, beating, or wiggling, either
singly or in coordinated arrays. Some stick out of single cells—such as our
sperm—and other tiny creatures, propelling them along. Others line surfaces
(clam gills, for instance) and pipes, acting as pumps and filters. In fallopian
tubes, they propel eggs toward the uterus. All use chemical energy in the form of
adenosine triphosphate (ATP), as does muscle. Figure 8.8 gives a simplified look
at the structure of a cilium. Of particular relevance to us are two proteins of their
internal structure. A polymer of one, tubulin, makes up the nine lengthwise
doublet tubules; a polymer of the other, dynein, forms a pair of arms on each of
those doublets. Tubulin and dynein slide along each other to generate motion;
the dynein arms cyclically attach to and detach from sequential sites along the
adjacent tubules. Bending of the overall cilium results when sliding occurs only
on one side. In short, the basic motor moves by ratcheting along somewhat the
way a person does when poling a boat up a shallow stream.
Muscle uses the same chemically powered ratcheting motion, however that
differs from what seems to happen when you lift something by shortening your
biceps. Shortening occurs because the basic filaments increasingly interdigitate
as they ratchet along each other, as when you slide the fingers of one hand
between those of the other, as in Figure 8.9. The fundamental motion may
resemble that of cilia, but the main proteins of muscle, myosin and actin, have
evolved separately from dynein and tubulin. Muscle can also tighten without
actually getting shorter; it can just develop an end-to-end pull. So either
microscopically or macroscopically, were misled a little by the term
“contraction” for what muscle does. A muscle develops force, which may or may
not (depending on the load) bring its ends closer together and make its belly
fatter.21

FIGURE 8.9. The basic contractile unit of ordinary muscle. Cross bridges on the thick myosin filaments
alternately attach and detach from sequential sites on the thin actin filaments. That makes the two sets of
filaments interdigitate farther so the whole apparatus gets shorter.

Muscle’s need to use energy to develop force, even when it doesn’t shorten
and thus do work, generates much of our common confusion between the terms
“force” and “work.” The peculiarity is peculiar to muscle. The chandelier’s chain
exerts enough force to hold up the chandelier, and it does so year after year with
no fuel supply whatsoever. (Oddly, one kind of muscle can maintain a force
without additional expenditure of energy. It’s the muscle that holds together the
half shells of clams, mentioned in Chapter 2. But offsetting that cheap, steady
pull is a slowness of action perhaps tolerable only by an unhurried clam.) Muscle
has another slight disability compared with the engines of our technology. Active
contraction is completely irreversible. Not only can’t muscle actively reexpand,
but stretching it doesn’t make it produce chemical energy. If you turn the shaft of
an electric motor, you can get electricity back out; you’ve made a generator.
Forcing pistons back and forth or spinning the rotors of turbines makes them
work as pumps. A loudspeaker can be used as a microphone. But you can’t
regain energy by running downhill or by putting a motor on your exercise bike to
drive your legs.
In at least one way, though, muscles behave like most of our internal-
combustion engines. The maximum power that either kind of engine can produce
depends on how long it has to sustain that power. We get higher power by briefly
supercharging (pressurizing the intake) when an aircraft takes off or by briefly
tolerating high heat production in the starter motors of our cars. Nature goes us
one better, tailoring muscle for specific durations of use. For good sustained
output, muscle sacrifices some of its contractile fibers (and thus power) to
provide space for more oxygen-processing metabolic machinery. Dark meat on a
bird or fish is the sustained output muscle; light meat is the more fiber-rich,
higher output, intermittently active version. Lots of so-called white muscle (a
little lighter in color) makes for success in a hundred-yard dash; red muscle
produces less power but wins in the long events. How much of each kind you
develop depends on your training regimen. The strategy for brief bouts of high
output may be the same for heat engines and muscle, but the tactics differ. Our
heat engines increase their peak output by laying on extra fuel and oxidant.
Muscle, by contrast, changes the way it uses its fuel, temporarily doing without
oxidant and accumulating (to our discomfort) lactic acid instead of carbon
dioxide.
Musclelike systems occur elsewhere among organisms. All nonbacterial cells
contain the basic motor proteins, actin and myosin. Such things as the
movements of the contents of plant cells and the locomotion of amoebas seem to
be driven by ratcheting interactions of actin and myosin filaments. Similarly,
when a cell divides, tubulin appears involved in moving the chromosomes
around, although probably without the same kind of ratcheting as in cilia. (We
find it much easier to discover what proteins are present in a system than to
understand how they operate.)
While organisms are a diverse lot, these proteinaceous motors aren’t. Nature
invented a few versions early on and stuck with them. Perhaps that’s not too
surprising. Enzymes, virtually all of them proteins, are nature’s chemical
machines. But they do their chemistry in a mechanical way: They work by
grabbing, manipulating, and releasing other molecules. Both myosin and dynein
are enzymes; they just move other proteins, actin and tubulin, instead of
performing more conventional chemical transformations.
Muscle does something very different from what any human machine can
manage. This presents a terrible problem for biomedical engineers as they try to
contrive prostheses for muscular organs. A heart is quite a simple thing,
compared, say, with a kidney or a liver, but if’s mainly muscle. We can make
fine valves (although as replacements pig valves still have certain points in their
favor), but we can’t yet make a full prosthesis thats anything like one’s original
equipment.

OTHER NATURAL ENGINES

Besides cilia, muscles, and a few other schemes in which proteins ratchet along
one another, nature, as noted earlier, has several other kinds of engines up her
sleeve. While they’re slower and thus not so obvious to us impetuous animals,
they’re undeniably powerful. They also have few parallels in our technology, and
they’re of particular interest because they work without solid parts that move.
Three are especially common.

• An ordinary corn plant lifts about four quarts of water each day from the
soil. Lifting is work, so a corn plant must have an engine, as must almost all
terrestrial plants. The main engine is simple but strange: a direct-acting solar-
powered evaporative engine. If (1) pipes filled with water run continuously
from the roots up into the leaves, if (2) water can evaporate into the
atmosphere from the leaves without air leaking in, and if (3) the pipes are
stiff enough so they don’t collapse, then water lost to evaporation will be
replaced by water ascending from the soil by way of roots, trunk, and
branches. The necessary conditions appear to be met, and most of the ascent
of water in a corn plant or a tree is due to this pull from the top.22
Evaporation of course takes a lot of solar energy; evaporating a gram of
water at room temperature takes a bit more energy even than boiling off a
gram in a hot pot. We’ll return in Chapter 10 to this remarkable machine that
may pull against pressures of more than one hundred atmospheres (nearly a
ton per square inch) without any moving parts.

• Most carbohydrates (starch, mainly) and proteins are hydrophilic; they

avidly attract water. Cornstarch and gelatin do just that in the kitchen, which
is why we use them as thickening agents. Put a lump of either starch or
protein (many kinds of protein, at least) in water, and the lump will swell
almost irresistibly. If dry seeds beneath concrete pavement get wet, their
 expansion can crack the pavement. Germinating seeds commonly use this
engine to split their seed coats and to penetrate soil. One case of this so-
called imbibition is known in animals: The male mosquito hydrates pads of
protein to raise the hairs on its antennae in order to sniff out the location of
an odoriferous, receptive female.23

• Like all other molecules in gases and liquids, water molecules diffuse around
—that is, they continuously move around at random, mixing themselves.
Work can be extracted from this random motion in the following way. If
water is more concentrated in one place than in an adjacent place, more
water will move from the concentrated place to the dilute place than the
other way around simply because there is more water in the concentrated
place. How to do it? Say we dilute some water by adding a solute such as a
salt or sugar. We then place the diluted water in a compartment that’s
separated from another compartment filled with pure water. As a barrier
between the compartments we use something through which only water can
pass, as in Figure 8.10. The compartment with the added solute will swell up
as more water enters than leaves it. While it may swell slowly, it will do so
very forcefully, working as yet another expansion engine. Pressure generated
by such a scheme ejects the stingers (nematocysts) of jellyfish from their
parent cells. Roots absorb water from soil and pump it into stems with a
version of this engine. Movements of leaves and other parts of plants depend
on pressure changes inside nonrigid cells; such pressure changes are
generated by adjusting the concentration of solutes in the cells and letting
diffusion do the rest.
FIGURE 8.10. The basic pressure-generating osmotic engine. Water enters a compartment in which it is
less concentrated, passing through a barrier that’s impermeable to the solute that dilutes the water.

COMPARING EFFICIENCY
Both power relative to weight and energetic efficiency measure the quality of
engines. But even together they don’t provide a full yardstick, and for some
engines, such as the evaporative, imbibition, and osmotic ones just described,
neither measure is particularly revealing. Still, energetic efficiency merits at least
the attention that weight economy got in Table 8.1. For electric engines one
ought to include the generator’s efficiency as a factor. For none of the engines is
it easy to peg the costs of obtaining and processing fuel, whether by mining or
by digestion. And the costs of transporting fuel vary a lot. Nor can we easily
account the use of “waste” heat for thermoregulation in warm-blooded animals
(and some large, active fish and flying insects) and for space heating of our
building. Bearing these limitations in mind, let’s look at a few figures for
energetic efficiency.24
Piston engines have a lot of virtues, but efficiency is not high among them.
At best automobile engines reach about 25 percent, but we usually run them far
below their optimal output level. Diesel engines do somewhat better, but their
efficiency also is often compromised by the ways we use them. In steam piston
engines where the steam expands in successive cylinders, efficiency got up to
about 17 percent by 1900 and can reach about 19 percent currently. (Watt’s
single-stage engines did about 2 percent.)
Turbines do better and have superseded piston engines for power plants,
large ships, and most aircraft. A coal-burning steam power plant might have an
efficiency of 40 percent, a nuclear steam plant (using, for safety, slightly cooler
steam) an efficiency of 32 percent, or a gas (internal- combustion) turbine a 26
percent efficiency.
Consider an electrical system composed of a turbine, a generator, and a
motor. The turbine has an efficiency of 40 percent or less, so it represents a huge
power loss. Large generators are better than 95 percent efficient, so very little is
lost at this step. Electric motors are widely variable, ranging from 20 percent for
the motor of a small electric fan to a most impressive 90 percent for a hundred-
horsepower polyphase induction motor optimally loaded. So the overall
efficiency of only the best systems will exceed 30 percent. One does better with
hydroelectricity—water turbines pass around 90 percent of the energy of the
stream to the generator—but since one has to have a decent head of water on tap,
the benefits are not universally available.
What about muscle? It turns out to be flabby, if functional. Even ignoring
losses in, for instance, locomotory machinery, it’s still less than 25 percent
efficient, and sometimes it’s a lot less. The shortening machinery isn’t too bad,
but the chemical operations that transfer energy from sugar to adenosine
triphosphate and thence to myosin suffer substantial losses. Also, in muscles (as
in other engines), efficiency depends a lot on speed of operation, something that
varies a great deal from muscle to muscle, animal to animal, and moment to
moment. Muscle may win no prize, but on the other hand, it’s no disaster either.
Both nature and human technologies give energetic efficiency a fairly high
priority in their respective designs. The evidence of this for natural technology
may be indirect, but it’s persuasive: Most structural and behavioral aspects of
organisms make sense only if energy is considered precious. For human
technology we have good historical records, whether we look at improvements
in the harnesses of draft animals or at the historical development of steam
engines. The similarity of engine efficiencies is curious, even if accidental—in
the 20 to 30 percent range for gas turbines, steam or gasoline piston engines,
electrical systems, and muscle. If either technology has any slight edge, it’s our
human one. Also, we’re still improving, as nature probably isn’t.

Beyond all the detail, what might we say about the good and bad of all these
engines? Heat engines and engines that expand or rotate dominate one
technology; constant-temperature engines, most of which shorten or shear,
dominate the other. One technology transports energy over long distances, either
as electricity or as chemical fuel; the other transports energy for only short
distances and uses electricity only for signaling. While human technology may
be marginally better in energetic efficiency, we definitely win in power output
relative to weight; nothing comes close to modern aircraft engines.
Nonetheless, we can still look with envy on muscle nature’s preeminent large
and fast engine. An individual muscle of a tiny insect might weigh a microgram;
a large muscle of a big whale may approach a hundred kilograms—several
hundred pounds. Those masses are a hundred billion times different, 1011-fold,
and performance doesn’t deteriorate noticeably at either extreme. Very little in
either technology does so well over such a gigantic size range.
Besides, muscle is good to eat. This is no mere Parthian shot. Expeditions
that used beasts of burden sometimes exercised the option of consuming them.
The Lewis and Clark expedition to the Pacific Northwest of the United States in
extremis ate horses, and a century later Amundsen’s expedition to the South Pole
ate its dogs on a predetermined schedule. Try that with your internal-combustion
engine.
 Chapter 9

PUTTING ENGINES TO WORK

Making an engine run efficiently doesn’t make it run usefully; it must be

persuaded to do its assigned task. Once in a while sagacity or sheer luck lets us
couple an engine directly to a machine, the way the shafts of their motors bear
the blades of lawn mowers, fans, and blenders. More often machines need things
between motor and output device to couple the two—transmissions. While the
machines of our devising usually need transmissions, muscle-powered machines
always need them. That’s because our rotary engines, at least, make full cycles
under their own power. Muscles can’t do this; a muscle does its work by
shortening, but it always needs something to restore it to its original length. In
human technology transmissions are ubiquitous, but in nature they’re universal.
For engines the important factors are power produced relative to power
consumed, power produced relative to weight, and how power varies as speed
changes. For transmissions, the only relevant power is the maximum that can be
handled without failure. Weights run well below and efficiencies well above
those of engines. Two things, though, matter a lot for transmissions. First is the
particular trade-off of force and distance (for a given amount of work) or force
and speed (for a given power level). You start a car moving with the
transmission in a low range, meaning a lot of force (to accelerate) while the car’s
speed is low. You then shift (or the car shifts) to a higher range, with less force
but a higher speed. Meanwhile the engine only has to operate over a narrow
range of speeds (RPMs). Second is the match of motion between engines and
final components, what’s called the kinematics of machines—the way the
hipbone is connected to the thighbone, and so forth. Proper kinematic design lets
a piston going up and down make wheels turn and lets the rotor of an electric
motor make a saw blade go back and forth.
In both technologies the diversity of transmissions exceeds that of engines.
But each technology has its bag of favorites, with less overlap between the two
than one might guess. Transmissions in fact contrast almost as sharply as do
engines. That’s partly driven by differences in the components from which
they’re crafted, partly by differences in the initial engines, and partly by
differences in what the machines finally do.

LEVERAGE
Probably no mechanical device is older than the lever; simple and versatile, it’s
no doubt older than we humans. Prying with a lever is something that must be
done at least occasionally by the finch of the Galapagos Islands, which uses a
thorn to extract insects from bark, or by a chimp dealing with a moundful of
delectable termites. Levers are powerful devices—Archimedes exaggerated only
a little in claiming he could move the earth if given a fulcrum for his lever—and
ubiquitous devices; without levers how could you pry the cap off a bottle,
puncture a can, tighten a nut, switch on a light. . . ?
A lever at its most basic consists of a stiff rod along which three locations
can be recognized. One is the fulcrum, the spot around which the lever turns.
Another is the load, where what’s being worked on makes contact. Finally
there’s the effort, where the driving force is applied. Levers come in two
precisely opposite versions, both shown in Figure 9.1. In one version, a force
amplifier, the distance between fulcrum and effort is greater than the distance
from fulcrum to load. In the other, a distance or speed amplifier, the distance
between fulcrum and effort is less than that from fulcrum to load.1 We’ll defer
talking about speed for a few pages.
FIGURE 9.1. Two basic kinds of levers. The crucial feature that distinguishes them is the relative distance
of load and effort from the fulcrum.

A lever is about the simplest possible transmission. It does no work, which

we have defined as force times distance. It just changes the particular mix of
force and distance of an engine (muscle or motor) into a more useful mix of
force and distance. Levers allow one to apply an effort force less than the
loading force (as with a prying bar). They also allow moving the effort a shorter
distance than it moves the load (as when one swings a golf club or baseball bat).
Levers are just the simplest of a diversity of devices that trade force against
distance and find use as transmissions. All give the same choice between force
amplification and distance amplification. Figure 9.2 shows a few that find use in
human technology: a windlass in which ropes run around coupled wheels of
different sizes, a so-called block and tackle in which a rope goes around several
pulley wheels, and a belt drive that uses pulleys of different sizes. The windlass
resembles a simple lever in both structure and operation; for the others one
shouldn’t let structural differences obscure the functional equivalence. In natural
technology, levers are common enough, but lever analogs like the block and
tackle aren’t quite so obvious.
In this distinction between force amplifiers and distance amplifiers lies an
interesting difference between natural technology and the gadgets we humans
use. We have a strong preference for force-amplifying devices, a few of which
are shown in Figure 9.3. Your appendages move far, but they don’t move very
forcefully, so you use force amplifiers to get, as we say, “leverage.” Look around
your kitchen. Almost all the hand-operated tools are force amplifiers. In our
kitchen the only distance-amplifying devices are a salad server, hinged between
short handles and long tines, and a pair of wooden tongs, hinged at one end, that
hoists English muffins from the depths of the toaster. Among the hand tools on
our workbench are screwdrivers, pry bar, wire snips, metal shears, socket
wrenches, and pliers—force amplifiers all. Among our garden tools the only
distance-amplifying device is the pair of snips used to trim grass that’s too close
to barriers for the power mower to reach; all others are force amplifiers.

FIGURE 9.2. Devices that get “leverage,” shifting the relative magnitudes of force and distance or speed.

We use hand-operated force amplifiers whenever we crank something.2 If

you haul a boat onto a trailer, you use the crank handle of the winch or windlass
so a lot of arm motion by the crank exerts a lot of force on the wire cable. Door
handles and faucet levers are just small-scale cranks. Our bias toward force
amplifiers shows in the way we use the term “mechanical advantage” for “force
advantage”; evidently distance advantage isn’t as advantageous. Thus we say
that a lever in which the effort-to-fulcrum distance is twice the load-to-fulcrum
distance has a mechanical advantage of two, not one-half.
A biologist would have made the opposite choice. In nature, distance
amplifiers rule the roost. They do so because muscles are short-stroke engines;
they make lots of force but shorten a relatively short distance. A muscle does its
greatest work (again, load times distance) when it shortens by only about 10
percent of its length, even though most muscles can shorten as much as 30
percent with lighter loads. To make arms and legs swing through substantial
arcs, to make their far ends move substantial distances, to permit muscles to run
close in alongside the bones that they move—such tasks require distance
amplifiers. Consider the muscles and bones that raise and lower your forearm,
shown in Figure 9.4. Both the biceps muscle of the front of your upper arm and
the triceps behind your humerus are substantial distance amplifiers.3

FIGURE 9.3. Household devices using levers that increase force at the expense of distance or speed: jar lid
loosener, nutcracker, garlic press, and can opener.

So our muscles are force specialists, and our bodies compensate with
distance-amplifying internal levers of tendons and bones. That makes our
appendages distance specialists, for which our technology compensates with
force-amplifying hand tools—our can piercers and pliers. Paradoxical, even
irrational perhaps, but we do prefer to use our tools without first being surgically
adjusted.
When substantial force is still needed, less drastic distance amplifiers are
used, as with the attachments of its muscles to the forearms of a mole, which
burrows through soil. Its bones are shorter and the muscles attach farther out
from the joints than do our equivalent ones. The same choice sets the relative
positions of jaw muscles and teeth in mammals and reptiles.4 The front teeth of
a protruding jaw snap at prey, with lots of distance but little force, while the rear
teeth, closer to the jaw-closing muscles, chew and grind with lots of force. Still,
both the mole’s forearms and the lion’s molars remain distance amplifiers, just
less drastic ones than our forearms and the lion’s incisors. Conversely, where
great distance (or speed) is needed, distance amplifiers get really extreme. The
flight muscles of many insects shorten less than 5 percent of their lengths—such
a short stroke may be necessary for insects to achieve their high wingbeat
frequencies—and since the muscles are tucked inside the insects’ thoraxes, they
have to be short. Insect wing tips may move as much as a hundred times farther
than their muscles shorten.

FIGURE 9.4. The muscles that work the forearm flank the bone (the humerus) of the upper arm. The
arrangement allows short, forceful contraction of these muscles to give long but less forceful movements of
the forearm.

To move body parts either farther or faster, nature often raises distance
advantages by hooking up muscles in a peculiar way. Instead of running straight
from the tendon on one end to the tendon on the other, the fibers are short and
oblique, usually running from a pair of outer tendons to a central one, as in
Figure 9.5. The outer tendons attach to one skeletal element, the central one to
another. The arrangement greatly limits the muscle fibers’ shortening distance,
but it much enlarges the total crosssectional area of the fibers, on which the force
depends. The arrangement is especially common in insects and crustaceans,
which need it because their muscles run inside their tubular skeletons. With
muscles inside, getting a substantial sideways distance between the hinge of a
joint and the attachment of a muscle is nearly impossible. Muscles simply can’t
shorten very much, so they have to produce a lot of force and then turn to
distance amplification. A lobster claw is a fine case, closing its pincer with one
of these pennate (penlike) muscles and opening it with another. A look at the
relevant anatomy might provide adequate excuse to eat one;5 otherwise Figure
9.5 will have to suffice. Another example is the fattest part of the hind leg of a
grasshopper; fiber directions correspond to ridges visible on the outside.

FIGURE 9.5. An ordinary muscle, such as one of those of the previous figure, and the really drastic high-
force, low-distance arrangement of a pennate muscle.

Several other arrangements let nature move things farther, faster, and less
forcibly than does a direct muscular hookup. Near the end of Chapter 7 a class of
hydrostatic devices called muscular hydrostats was mentioned; examples
included the tentacles and arms of squid, various tongues, and the trunks of
elephants. If a squid’s tentacle is made entirely of muscle, as it very nearly is,
and if it’s long and thin, as it certainly is, then a slight decrease in diameter ought
to give a great increase in length. If muscle contraction reduces the diameter of
an incompressible cylinder by 10 percent, then the cylinder will lengthen by
almost 24 percent. But a tentacle is already long and skinny. If its length is
twenty-five times its diameter, then in absolute, not percentage, terms a one-unit
decrease in diameter will give an almost sixty-unit increase in length. Force, of
course, will work the other way, with a sixtyfold decrease in force from input to
output. Squid snag moving prey by extending tentacles far and fast but not very
forcibly.6
A more widespread and personally relevant way to make the most of limited
shortening is to wrap a thick coat of incompressible muscle around a spherical
chamber, as in a heart. A little calculation reveals that if the volume of the wall is
twice that of the chamber inside, shortening the muscle fibers by 6 percent will
drive out half the contents of the chamber, and shortening by 13 percent will
fully empty the chamber. Six percent and 13 percent span muscle’s efficient
range of shortening.7 Similar calculations can be done (the reader might be
inclined to try) for muscle-wrapped cylinders, as in our intestine. The wall is
thinner, a little greater shortening is needed, and we use a different kind of
muscle, but the principle is the same.
Nor is the problem of converting high force to high distance unique to
muscle-driven machines. The proteins that ratchet past one another in a cilium
are near enough to the center of the cilium so a little ratcheting gives the whole
thing a substantial bend. And the wilting of the leaf of a plant from horizontally
outward to vertically downward takes only a small loss of volume of a few cells
in its stem, as shown in Figure 9.6.
The engines of our technology are more diverse in their force-distance
behavior than are muscles, cilia, and swelling cells. But in most instances their
moving parts travel far but less forcibly—quite the opposite of living engines.
What particularly characterizes engines of recent vintage—whether electrical or
combustion—is that they move fast. High speed has consequences similar to
long distance: Power output is force times speed just as work output is force
times distance. Higher speeds let us use smaller and lighter engines. For
instance, we use particularly highspeed motors to power hand-held power tools
and aircraft, cases in which size and weight must be minimized. In effect, higher
speed means that a given engine is used more often. Higher speed also means
that we often have to slow things down between engine and application, and a
speed- reducing transmission is a force-amplifying one.
In short, muscles most often need distance-amplifying things like tendons
and bones to couple them to their tasks, while rotary motors commonly need
force-amplifying gearboxes to drive useful machines.
Still, distance and speed aren’t quite the same, and our arms and legs move
relatively far while the engines of human technology move relatively fast. As a
result, motorizing a hand-operated machine rarely works well. Years ago almost
every household had a hand grinder designed to be turned about once each
second by a willing arm. Most of its tasks now get done by a food processor, a
radically different design with a directly connected electric motor. Less common
is a grinder of the old design but with a motor. We have one, and its performance
is unimpressive.

FIGURE 9.6. Osmosis can generate spectacularly high pressures. Slight but forceful swelling and shrinking
of the large cells in the lower part of this leaf stem (petiole) raise and lower the leaf’s blade as sunlight and
water supply change.

Where simple motorization works, though, the comparison is instructive.

Consider the hand-operated and electric forms of meat grinders and ice-cream
freezers. For hand operation a long crank is used, the usual force amplifier
undoing the distance amplification between arm muscles and hand. The
motorized versions replace that long lever with motors that turn between fifty
and a hundred times each second, so a problem of excessive speed replaces one
of excessive distance. Lest the motor stall or the contents be immediately
liquefied, a set of gears must be used to reduce the speed and increase the force.
We make textile thread with machines not at all like old spinning wheels,
spinning blade power lawn mowers have displaced the motorized reel type of
mower, and the most effective bits for use in electric drills differ from those that
work best in a hand-cranked drill.

WHEELS
Human technology turns on wheels. Except for the odd sled or sledge, all our
land vehicles ride on them. Our ships are driven by rotating paddle wheels or
propellers. Our aircraft use rotary turbines with or without rotary propellers.
Snow blowers, trench diggers, conveyor belts, and chain saws go in circles
around axes. Internally we use crankshafts, rotary electric motors, rotating
pulleys, gears, capstans, hinges, cams, windlasses, ratchet wheels, roller
bearings, and spindles—just to mention the more obvious ones. Nature, with
only one exceptional case, doesn’t go around using wheels. Only metal use
provides as dramatic a contrast.
When I was a student, the issue was simply put. “Nature has never invented
the wheel,” went the textbooks. But science progresses, and we now know that a
perfectly fine and true wheel and axle does occur in nature. It’s new only to us
since the organisms equipped with wheels are of enormous antiquity. The
discovery of this wheel—by Howard Berg and his associates during the 1970s8
—demands that we ask not why nature hasn’t invented the wheel but why she
uses it in only one instance. First, of course, we have to examine that instance.
The talk about cilia and flagella in the last chapter quietly excluded bacterial
flagella. They’re much smaller than the standard appurtenance of higher
organisms, and they lack the “normal” internal gear with which cilia actively
wave around. In the high magnification of an electron micrograph, a bacterial
flagellum usually looks like a carefully drawn set of very regular waves—
suspiciously like a rigid structure. It turns out to be a rigid helix much like a
corkscrew. Instead of passing one wave of bending after another along its length,
it spins around, ten to one hundred times each second, as in Figure 9.7. The base
of the flagellum forms a driveshaft that passes through the cell membrane,
connecting it to a rotary engine. And the membrane works like a proper set of
bearings. The engine bears a curious similarity in both appearance and operation
to our electric motors. It’s even reversible. The whole thing—engine and
corkscrew—either singly or in groups, pushes or pulls a bacterium around much
the same way a propeller pushes a ship or pulls an airplane.
FIGURE 9.7. A bacterium with its flagellum, and the base of the flagellum in more detail as we presently
understand it. The magnification here is extremely high, about three hundred thousand times, so specific
details represent interpretations of smudges. How this rotary engine works is far from clear.

How well does the bacterial flagellum perform? In terms of power output per
unit weight it’s more than fifty times better than muscle, better even than a gas
turbine. Still, even protozoa, some only a little larger than a typical bacterium,
swim with conventional flagella. These latter rely on the much less potent
tubulin-dynein engine described earlier. This situation is powerfully puzzling.
Could it be that key bits of information simply never got passed on to the
nonbacterial world? Odder yet, higher organisms have adopted other bits of
bacterial machinery by the drastic step of symbiotically expropriating the
bacteria themselves, as noted back in Chapter 2.9 Perhaps for some reason the
bacterial engine can’t be enlarged: impracticality of making larger bearings,
difficulty in electrical transmission over any longer distance in a nonmetallic
world, or something else entirely.
By wheels we mean proper wheel and axle devices that can rotate without
limit with respect to the rest of a machine. If you roll down a hill, your whole
body may be a wheel, but you’re no wheel and axle. So we’re not talking about
tumbleweeds, or about the tiny turds that dung beetles roll homeward for their
grubs, or about a few crustaceans that get around by rolling as a whole. Nor are
we worrying about how far we can rotate our fists around our arms or our heads
on our shoulders. By “rotation” we also mean something fairly specific. When
you draw a circle on a piece of paper, do you rotate your hand? You may move it
in a circle, but you don’t truly rotate it; after all, your hand at all times points in
the same compass direction. Human dances make elaborate use of such circular
but nonrotational motion, most likely because it doesn’t make us dizzy. Not all
dances, of course; waltzes are rotational and, one suspects, intentionally
vertiginous. The wheels of a bicycle rotate; your feet and the pedals just move
around in circular paths. The Ferris wheel rotates as a whole, but the seats and
people just go in circles. In this precise sense— excluding both rolling as a
whole and merely going in circles—the only known instance of a wheel and axle
in nature is the bacterial flagellum.
The classic view is that wheels are terrific, but that nature (sweeping the
bacteria under the rug) just can’t figure out how to make them do decent service.
Stephen Jay Gould made just this point,10 noting the difficulty of moving
nutrients into structures with sliding connections to the rest of an organism. That
suggestion fits nicely with one of the main points here; the different contexts of
human and natural technologies. He mentioned, additionally, the problem of
evolutionary continuity. How could an incompletely evolved wheel have
conferred benefit on a creature? The argument, like the equivalent one made here
for metals, may be attractive but isn’t fully persuasive. Gould’s whole
discussion, though, hinges on the absolute superiority of wheeled transport.
As with so many issues, close scrutiny uncovers complications. At least two
people, Michael LaBarbera and George Basalla, a biologist and a historian,11
have taken exception to Gould’s freewheeling assumption of superiority. They
admit that wheels may give cheaper transport than legs. After all, bicycling is
more efficient than walking or running; adding twenty-five pounds of passive
machinery reduces the cost of transport severalfold. But they point out that
wheels realize their advantage only on smooth surfaces—on roads or floors. A
wheel on a cart can go over bumps no higher than a quarter of its diameter, and
even those bump up the cost of transport. Prairie schooners had enormous
wheels, much larger than the ones on ordinary on-the-road wagons. To use
wheeled transport routinely and productively, a culture must be sufficiently
settled and organized to do a lot of civil engineering. For most organisms, much
smaller than we, the natural world is an even bumpier place.
Wheels, however wide our use of them, don’t define human technology.
They seem to have appeared first in the Middle East, about five thousand years
ago, in two applications: wheeled vehicles and potters’ wheels. Whether the two
were linked or the wheel was invented on more than one occasion remains
unclear. But they were absent from the pre-Columbian New World; pack animals
and sledges were used for transport, and even axisymmetrical pottery was built
up by coiling long, thin cylinders of wet clay rather than by turning on potters’
wheels. A visit to a museum of Incan or Mesoamerican artifacts should persuade
anyone that these weren’t technologically primitive cultures. Moreover, they
knew about wheels; Mayan and Aztec toys have been found in which animals
had wheels on the ends of their legs.12 Nor was the Western Hemisphere the
only site of non-wheel-ridden cultures. The inhabitants of sub-Saharan Africa,
Southeast Asia, and Australia got along without them as well.
For wheeled transport to make sense, streets must be wide enough for a cart
and draft animal to turn around since the combination can’t easily back up. Draft
animals of sufficient size must be domesticated, and that seems not to have
happened in pre-Columbian America. Furthermore, roads must be hard. Shag or
well-padded carpet almost completely immobilizes a rider-propelled wheelchair,
and government standards prohibit thick carpeting in public buildings. Wheeled
vehicles were given up in North Africa and the Middle East between the third
and seventh centuries (C.E.) and not used there again for a thousand years.
Domesticated camels that carried rather than pulled apparently provided a better
alternative under the circumstances.13
So one may be skeptical of claims of wheeled transport’s absolute
superiority. But the arguments don’t shed much light on other uses of wheels in
technology. Here I think the case for the broad utility of wheels and axles is
much clearer. Wheel-based transmissions are highly versatile and efficient. Pairs
of simple gears (Figure 9.8) routinely pass power from one shaft to another with
better than 99 percent efficiency. The efficiencies of such things as bevel gears
(which connect shafts at right angles), belting, and roller chains range between
95 and 99 percent. In worm gearing, a long, turning driver (the worm) slides
through the teeth of the gear it turns (the wheel), but even this turns out to be
about 80 percent efficient.14
Roads may limit the use of wheels for vehicular transport, but bearings limit
their general use as mechanical elements. Doing without bearings is possible
only by using such unattractive schemes as placing free rollers (such as tree
trunks) under an object and moving the rollers as they come free from the rear to
the front of the object. Large stone blocks were certainly moved this way by the
builders of ancient monuments, and I’ve found it a useful way to move long,
unsplit logs for short distances. Alternatively, of course, the whole vehicle can
simply roll. But in properly upstanding vehicles some parts must slide against
one another, the sliding surfaces must bear loads, and friction between sliding
surfaces costs power. So a cart needs smoothly turning bearings, either between
the wheels and their axles or between the axles and the frame.

FIGURE 9.8. Several kinds of gears and such, a far from exhaustive selection (clockwise from the upper
left): spur gears, a cam, a worm and wheel speed reducer, and bevel gears.

So everything turns on bearings and on the twin problems of friction and

wear. These must have made big trouble until metal lathes of decent precision
were developed a few hundred years ago. The underlying problem is a nasty one.
A thicker axle gives good support with low stresses on its material, but its
bearing surfaces move around one another at higher speeds and thus with more
heat and wear. The opposing surfaces of a thinner axle move more slowly,
generating less heat and wear, but stresses are higher, and the axle is more likely
to break. Good bearings, like electric wires, are mundane but critical items
whose present excellence we take for granted—except, of course, when they
seize up or screech thirstily for lubricants. Or during wartime, when ball-bearing
factories become targets of the highest priority.
Even though she uses no wheels, nature makes some impressive bearings.
On the ends of our bones are layers of porous cartilage. A lubricant (synovial
fluid) oozes through them, so adjoining bones never actually touch each other.
The resulting friction is about as low as anything in engineering practice—unless
you suffer from arthritis or bursitis. One can get a feel for the impressive
slipperiness of our joints by dissecting the muscle and connective tissue away
from the knee or hip of a lamb and then moving the bones on either side of the
joint while pressing them together as hard as one can. At least perceptually, the
joint is frictionless.15 Nature’s bearings won’t bear blame for the absence of the
wheel and axle.
In using wheel-based devices, humans can’t copy nature. Our widespread use
of them proves their technological value better than any verbal argument or
calculation. Their scarcity in nature, though, isn’t so simply explained. At least
we don’t understand what mix of factors might matter. Are they less useful to an
unstiff technology in a bumpy world? Are they hard to evolve and maintain?
Had they worn out their welcome when organisms got much over a thousandth
of a millimeter long?

MATING MOTORS AND ROTORS

The lack of wheels and rotational motion among organisms, on one hand, and
their commonness among human devices, on the other, raise a peculiar problem:
how to turn a rotary machine. If, as in every human- or animal-powered
machine, the engine doesn’t rotate, then linear or reciprocating motion must be
converted to rotary motion. We commit mechanical miscegenation when our
pushes and pulls work rotating machines. But the conversion problem doesn’t
face only muscular drivers.
How can a push-pull engine, whether it has muscles or moving pistons, make
rotation? The engines of our cars use piston rods to turn crankshafts, and some
of us recall the slider cranks attached to the driving wheels of steam
locomotives. But these devices weren’t common until late in the eighteenth
century, when James Watt introduced a version of the crankshaft so his
stationary steam engines could turn wheels.16 Nor was Watt the first to get
rotation from a reciprocating engine. Rotary motion had sometimes been
produced with Newcomen beam engines, using a heroic but inefficient scheme:
The engine drove a pump that raised water that then dropped onto an overshot
waterwheel.
This problem of getting repeated rotation out of muscle can be circumvented
if the engine itself goes in circles, as happens when an animal tethered to a long
cranking arm walks in monotonous circles. Otherwise one needs cranks like
those of our piston engines. Modern cranks may have been invented by
Archimedes, about 200 B.C.E.; they’re first mentioned in Chinese literature in
31 C.E., surely as an independent development. But the idea goes back to the
ancient Egyptians, although their version required practice and skill to use
effectively. The difference between theirs and ours lies in how much freedom of
motion the main shaft is allowed. Most of our cranks use fully guided rotors in
which a set of bearings keeps the main shaft from doing anything other than
turning about its axis. The operator’s hand or foot is guided around a
predetermined circle, as in an old-fashioned meat grinder or a bicycle. Using the
opposite extreme, an unguided rotor such as a spinning lasso, takes real
dedication.
Intermediate is the partially guided rotor, which is what some Egyptian relief
sculptures seem to show. As in Figure 9.9, one end is fixed and the other swings
in a circular path. It’s all too prone to wobble; with only one end fixed the
operator has to keep the shaft from swinging to one side or the other. The
scheme does make a perfectly serviceable drill, although one that takes a little
practice to operate and works best for shallow holes, where a little wobble is
acceptable.17 We still use a few partially guided rotors. A conventional brace
and bit require that the operator maintain the alignment of the drill shaft. But
alignment only requires that the aligning hand act as a stationary upper bearing
rather than move in a specific circle. On the other hand, the brace and bit
occupies both of the operator’s hands, while the old Egyptian device left a hand
to hold the object that was being worked on. A hand-operated eggbeater is
another partially guided rotor, and like a brace and bit, it demands a little skill.
A device that’s now fully obsolete, the bow drill, may be used to make a
muscle-powered arm drive a rotor. As in Figure 9.10, a flexible cord connects
the ends of a bow; the cord is looped around the shaft to be spun. The lower end
of the shaft holds a drill or is otherwise pointed, while the upper end turns
against a hand-held thrust bearing. Pulling the bow back and forth makes the
shaft revolve first one way and then the other. True conversion to rotary motion
isn’t strictly involved, since periodically reversing the shaft balances the
clockwise and counterclockwise turns. Still, it realizes most of the advantages of
rotation for tasks such as drilling. Bow drills are ancient and multicultural. North
American Indians, who used no other obviously rotary devices, used them to
generate sufficient heat (through friction with the lower bearing block) to start
fires, a fine, nonobvious (as a patent office would put it) invention.

FIGURE 9.9. A partially guided crank, in this case an interpretation of the mechanics of an ancient
Egyptian drill. The details come from my test version rather than actual antiquity. The operator grasps the
handle and moves it in a circle, watching to make sure the drill stays vertical and doesn’t wobble. The
weight swings out and around on its own; it smooths the motion and provides downward force on the drill.

HYDRAULIC CONNECTIONS
In a hydraulic device something solid gets pushed by a pressurized liquid.
(Where a gas gives the push, if’s a pneumatic device, but the principles are the
same.) We’ve already run across such devices in various guises. In Chapter 7 we
looked at hydrostatic and aerostatic supportive systems, such as worms and
blimps; in Chapter 8 we considered water transport, water absorption, and
evaporative engines (mainly in plants); and earlier in this chapter we talked
about muscular hydrostats like a squid’s tentacle. Except for a few aerostatic
blimps and buildings, all the examples involved organisms, and they ranged
from unicells to trees and sharks. Nature apparently finds this an “easy”
technology. That shouldn’t surprise us since nature pumps a lot of fluid around
for other purposes and since a muscle encircling a spherical chamber or a
cylindrical tube suffices for pressurization.

FIGURE 9.10. A bow drill, shown with a modern bit. The upper, hand-held block has a small concavity on
its lower side to guide the drill. Making and using one of these are easy, although you’ll need to roughen
the bit’s surface so the string or leather thong doesn’t slip.

The brakes of cars are our most familiar hydraulic machinery. You push on a
pedal that pushes a piston farther into a cylinder. That in turn forces brake fluid
out of the cylinder, through some pipes, and into another cylinder, where the
fluid pushes another piston outward. The second piston then pushes the brake
lining against the brake drum (or disk), creating lots of car-slowing friction. The
system works because a pressure applied at any point in a closed fluid-filled
system appears everywhere else in that system; pressure is transmitted almost
undiminished and instantly.
The handiness of hydraulics comes from its basic principle. The pressure in
the hydraulic fluid is the force on the piston times the area of its face, and that
pressure remains the same throughout the system. Thus a wide range of forces
can be produced by doing nothing more devious than changing the face area of
the pistons. If, as in Figure 9.11, you push inward on a piston with a force of ten
pounds in a cylinder one square inch in cross section (a pressure of ten pounds
per square inch), a piston in a cylinder of four square inches will be pushed
outward with a force of no less than forty pounds. Although this sounds like
getting something for nothing, work and energy in fact are properly conserved.
The piston in the big cylinder may move more forcibly, but it won’t move as far.
If you push the small piston in an inch, the large one will move outward only a
quarter of an inch. In short, the device behaves just like a lever. What a nice way
to move force, work, or power from one place to another, trading force and
distance (or force and speed) whenever convenient!

FIGURE 9.11. A hydraulic connection that achieves a fourfold force amplification.

Where does nature use hydraulic transmissions? Put a freshly cut flower in a
vase of water, and the water is hydraulically sucked up the stem from the vase. A
starfish moves around on a thousand or so tiny tube feet, hydrostatic devices
interconnected in a low-pressure hydraulic system. A worm adjusts the pressures
in its sequential compartments so that with the aid of some rearward-pointing
bristles, it can penetrate soil. A spider flexes its legs with an ordinary set of
muscles, but instead of using extensor muscles, it moves its legs outward
hydraulically; it squeezes together the top and bottom of its body (the
cephalothorax), increasing its blood pressure and extending its legs. On
emerging from its pupal skin, a butterfly briefly pulls its abdominal segments
inward to increase its blood pressure, inflating the veins of the wings, and
expanding their membranes. We run the filtration equipment (glomeruli) in our
kidneys hydraulically, using our high arterial blood pressure to force plasma (but
not blood cells) through tiny pores as the first step in making urine. That’s partly
why keeping proper fluid balance depends on having a good heart. We (male
humans and some but not all other male mammals) use the same blood pressure
to erect our penises. At least we do so in part, with valves and local muscle
activity then pushing the pressure up to much higher levels. Hydraulic devices
are nothing if not widespread among organisms.
But for human technology, hydraulics and pneumatics have come harder. We
have no handy squeezer like muscle, and most of our manufactured pipes are
rigid. We make pressure with pistons; indeed we use pistons for little else. But
pistons must fit snugly and smoothly in their cylinders if they’re to slide freely
yet not leak. (Detecting such leakage is the purpose of a compression test on an
automobile engine.) The ancients could do none of this, and hydraulic devices
were limited to siphons and other pistonless schemes that worked at close to
atmospheric pressure. Leakage between pistons and cylinders limited the
pressures that could be used in early steam engines. Now that leakage is a
manageable problem, hydraulic and pneumatic devices find an ever wider
variety of uses. The technology may have presented great difficulties early on,
but it turns out to be so marvelously handy. But while our applications resemble
nature’s, our machinery looks thoroughly different.18
Consider, again, the brakes of a car. These days the brake pedal works not
one but two cylinders, a safety feature to make the system leak-tolerant. If your
car has power-assisted brakes, the reduced pressure associated with the engine’s
intake works as a pneumatic element to generate the hydraulic pressure. If the
system leaks a little, you add extra brake fluid to the reservoirs or remove air or
froth from the system (“bleeding” the brakes). Some cars use hydraulic
connections to run their clutches as well; the device simplifies construction of
models available with either right- or left-hand drive. Farm tractors usually have
a hydraulic system that raises and lowers various attachments, with the hydraulic
pressure produced by an engine-driven pump. Control surfaces—rudder,
elevator, ailerons, and flaps—and retractable landing gear on aircraft are often
hydraulically operated. Heavy construction equipment steadily increases its use
of hydraulic power transmission. Closer to home, door closers, shock absorbers,
and hand-operated squirt or spray bottles are all hydraulic or pneumatic devices.
A particularly subtle and clever use of hydraulic power transfer is basic to
almost every automatic automobile transmission. These fluid couplings aren’t
especially efficient—early automatic transmissions had to be water-cooled to
dispose of their waste heat—but they have the nice property of slipping badly (or
usefully) when run slowly. That means that the driver can let the engine idle
impotently without mechanically disconnecting it from the wheels (as a manual
clutch does). The basic scheme of a fluid coupling is shown in Figure 9.12. It
works in the following way. First, any mass prefers to go straight rather than to
turn in a circle; if you turn around while holding a string with a weight at the
other end, the weight flies outward and pulls the string taut. Liquids have mass,
so when spun, the liquid in a chamber tries to move outward, experiencing
what’s ordinarily called centrifugal force. Second, moving mass in toward the
axis of rotation makes a system rotate faster, a notion familiar from watching
skaters increase their spinning rates by pressing their arms closer to their bodies.
(To look up the details, try “conservation of angular momentum.”) If fluid moves
outward as a result of the spin of an input shaft and moves inward again in an
adjacent chamber attached to an output shaft, the output shaft will speed up.
Rotary motion and force (together, torque) have thus been transferred from input
to output. Within limits, the faster the input shaft is spun, the more efficient the
coupling.19 So when you speed up the engine, the wheels recall that they have
an engine attached.

FIGURE 9.12. A fluid coupling. Input and output shafts each have hollow half toroids running around their
circumference. The half toroids, filled with oil and baffling, are sealed together but are free to slide against
each other. Imagine the well-buttered sliced surfaces of a halved bagel pressed together.

BRIEF BATTERIES
Speeding up or slowing down anything that has mass takes force. To exert a
force over a distance takes work, and that implies expenditure of energy. Yet a
pendulum repeatedly accelerates downward and decelerates upward, needing
only slight periodic pushes to keep it going. Where does the pendulum’s energy
come from? What it does is shift its energy investment four times for each full
swing. As it rises and slows, it gets out of energy of motion (kinetic energy) and
into gravitational energy. Only a small amount (offset by the occasional push) is
diverted to heat. As the pendulum swings earthward again, the gravitational
investment is cashed in and reappears as increased speed. It then swings up the
other way and does two more energy transfers. In effect, the pendulum
repeatedly stores kinetic energy in a gravitational battery.
Besides using gravity, how else can energy be stored? The next most obvious
agency is elastic resilience, the reversible deformation of material—what the
springs of our cars do. In addition, rechargeable electric batteries are common in
cars, portable appliances, the clocks of computers, and elsewhere. Finally energy
can be stored inertially, by making a flywheel spin; a machine can then be run by
slowing that spin. Some old potters’ wheels took only an occasional push,
maintaining their motion by doling out the energy of that push.
Human technology uses all four storage modes: gravitational, elastic,
electrical, and kinetic. Pendulums have long served as speed regulators for
mechanical clocks, and counterweights within the frames of windows and in the
shafts of elevators allow us to raise heavy things with the energy from previous
lowerings. In a medieval catapult, energy was slowly put in by raising a
counterweight; letting it fall released the energy, which was used to throw a
missile. We occasionally use gravitational storage on a large scale when we
generate electricity; the scheme is called pumped storage. Our use of electricity
varies a lot—but predictably so—with the time of day, so utilities must be able to
adjust production. Nuclear plants aren’t much help because they cost so much to
build and so little to run that they’re best operated near full capacity. Coal-
burning plants are better for meeting intermittent demand, but they leave a utility
with a lot of unused capacity for much of each day. And electrical batteries to
smooth out the demand would be uneconomically large. A pumped storage
system makes no new electrical energy. It just uses excess capacity when
demand is low to pump water into an uphill reservoir; it then operates as a
hydroelectric plant when demand is high. But it requires a handy mountain and
local people willing to tolerate a reservoir whose level fluctuates daily.
Local acceptability can’t be brushed aside. Pumped storage works best in
undeveloped hilly country near centers of population—a formula for maximizing
offense to people who value access to scenic nature. During the 1960s New York
City’s main electrical utility proposed building such a plant about fifty miles up
the Hudson River, near West Point, in the beautiful and historic Highlands. A
mighty outcry eventually put an end to the matter.20 As a biologist who grew up
only a few miles from the site I am perhaps predictably biased: Keeping the
Highlands unsullied merits a little energy conservation in the lower Hudson. One
must remember that, their image makers notwithstanding, utility companies are
in business to sell, not save, power.
More common is elastic storage, although most of the springs in our cars and
appliances are there more for our comfort or because they permit simpler designs
than for storing great amounts of energy. Mechanical typewriters used large
numbers of springs, and every dishwasher, tape player, and camera has a few.
They provide more serious storage in clocks, watches, and wind-up devices,
such as toys and an occasional lawn mower starter. As mentioned a few chapters
back, metals have high resilience and accept a wide range of deformations:
tensile, compressive, flexural, and torsional. In addition, springs may be
nonmetallic (rubber bands, for instance) or even gaseous. Squeeze a gas such as
air, and it can reexpand with almost no loss of energy. Elastic storage was of
great importance in weaponry before explosive charges came into use: An arrow
is propelled by energy stored in the bent bow, and the great rock-throwing
ballistae of classical antiquity used twisted cow tendon. Still, the use of elastic
storage has declined relative to the next scheme.
Until recently most rechargeable batteries stored electricity to restart
internal-combustion engines. Such batteries now grace all kinds of small
appliances, even though they’re heavy for the energy they store; electric vehicles
run into just that drawback. In addition, batteries deteriorate far more rapidly
than springs and pendulums. Fashion may play a role; fine wind-up mechanical
shavers, devices that could be left in car or briefcase and didn’t care at all about
chargers or local voltage, were available a few years ago.
Finally, we use an occasional flywheel, nothing more than a large mass
rotating in a wide circle. These simplest of short-term batteries may also be the
most ancient; intermittently driven potters’ wheels have been used for more than
five thousand years. We use flywheels to smooth out the action of our engines,
both piston engines and (in turntables and tape recorders) electric motors, and
we use them in toys, such as tops, Yo-Yos, and tiny model cars. Occasionally
one hears suggestions that automobiles might employ flywheels along with
smaller internal-combustion engines that run at more efficiently high and steady
speeds; the flywheel would intermittently supply energy for acceleration and hill
climbing.
Two of these four schemes, gravitational and elastic energy storage, find
wide application in nature, probably more than in human technology. Just about
every way that animals move around—walking, running, jumping, flying,
swimming—uses one or the other. Three of nature’s peculiar disabilities cry out
for short-term energy storage. One is her lack of wheels and consequent reliance
on reciprocating or pulsating devices like legs and wings. A second is muscle’s
inability to reexpand itself and the resulting need for some easy way to undo its
contraction; an opposing muscle is bulky and expensive. The third is that muscle
contraction isn’t instantaneous, and chemical explosives aren’t part of nature’s
ordinary armamentarium, yet motion with high initial acceleration has obvious
advantages: Predators from alligators to cats get their food by springing forward.
Through slow accumulation and sudden release, energy storage permits pulses of
extreme power output.
Elastic storage must be more common than gravitational storage simply
because the latter works only for terrestrial organisms of significant weight and
the bulk of the earth’s creatures are aquatic or tiny or both. No absolute choice,
though, need be made; a single animal may use both. Consider how we get
around on our hind legs. Legged locomotion isn’t especially efficient—recall
that the cost of getting from one place to another can be substantially reduced by
attaching a bicycle to oneself— but getting around on legs would be worse
without energy storage. Gravitational or elastic? As shown in Figure 9.13, we
use both, switching back and forth effortlessly and almost mindlessly. We
cheapen our walking gait with a little gravitational, pendulumlike storage
between strides. As we walk faster, we tend to swing our legs farther rather than
more frequently. Just as with a pendulum, one frequency is “natural” or most
efficient for a given leg length. Eventually, of course, your anatomy resists any
greater amplitude, and you switch to a gentle jog, another gait altogether. For a
typical human, that happens at about a twelve-minute mile.
FIGURE 9.13. The essential difference between walking and running. In the former what matters is the
weight of the appendage since storage between strides is gravitational. In the latter the elasticity of tendons
provides the equivalent reservoir for absorbing and releasing energy.

R. McNeill Alexander, of Leeds University, who has done the best work ever
on the mechanics of walking and running, found the rule that sets the switch
point. Except for the value of a constant, the formula for the gait transition point
is the same one that gives the period of a pendulum. Whether you’re a crow or
kangaroo shifting from walking to hopping, or a human or dog (or even an
insect) switching from walking to trotting, you follow Alexander’s rule.
Switching happens when the square of your speed is around half of gravitational
acceleration times the distance between your hip and the ground. For a middle-
size human, again, that’s about a twelve-minute mile or five miles per hour. For
a smaller animal, the transition speed is lower; you walk while a small child,
dog, or cat must trot to keep pace.21
While gravitational storage isn’t of much value above the transition speed,
energy storage is still very much in the picture. When jogging or running, you
use elastic rather than gravitational energy storage, stretching tendons rather than
swinging legs up and down. The storage mode differs, and the two gaits are
unmistakably distinct. In a way, gravitational storage is the special case, used
only as legs rise and fall in walking. All the other common gaits—trotting,
galloping, hopping, and so forth— depend on elastic storage. A hopping
kangaroo, for instance, regains about 40 percent of the energy absorbed in
landing when it bounces up again. In storing the energy by stretching tendon, it’s
using the protein collagen as its battery—the same stuff the ancients took from
cows for their ballistae and the main material of our own tendons. Collagen has a
resilience of about 93 percent—that is, 7 percent of the energy put in when it’s
stretched fails to reappear in mechanical form when it springs back. That’s not
bad, better than the ordinary rubber we make from the sap of rubber trees. But
where collagen brings home the bacon is in the amount of energy it can store
relative to its weight—nearly twenty times that of spring steel.
The problem of appendages that repeatedly change directions isn’t limited to
terrestrial locomotion with legs. Insects beat their wings up and down at
frequencies up to almost (in the smallest ones) a thousand times a second. We’ve
known about (and wondered at) these remarkable rates for a long time. During
the 1940s an unusually gifted Finnish investigator, Olavi Sotavalta, compiled a
compendium of wingbeat frequencies simply by listening to them. Or perhaps
not so simply: Sotavalta not only had perfect pitch but had trained himself to
distinguish fundamental notes from overtones, avoiding, as he put it, the
“soprano-tenor error.” Tone-deaf people like me resort to microphones,
recorders, and other bits of electronic assistance. (In my youth I tethered fruit
flies by fine wires to phonograph needles.) Sotavalta’s data have always proved
reliable. Incidentally, one can force the frequency even higher by unloading the
system—snipping the ends off the wings. The record is 2,218 beats per second,
held (and checked by others from a recording) by the same Olavi Sotavalta;22
that’s by far the most rapid back-and-forth movement made by any organism.
As with stride frequencies in walking, wingbeat frequencies don’t vary much
for a given insect, and changes in wingbeat amplitude and other variables handle
adjustments of flying speed. A sharp optimal frequency characterizes almost any
gravitational and elastic energy storage system— why we use either pendulums
or tiny springs to regulate our mechanical timepieces.
Flying insects use elastic storage, and in their wing hinges are pads of the
best elastic polymer known to either technology, resilin, which got brief mention
in Chapter 2. Resilin was discovered about 1960 by a great Danish scientist,
Torkel Weis-Fogh, who measured its resilience as an astonishing 97 percent.
Resilin thus loses only 3 percent of the energy put in rather than the 7 percent of
our collagen. The difference in power economy, though, is probably trivial, for
97 percent is only a little better than 93 percent. What must matter more is the
mischief caused by any energy that’s lost: It turns into heat, and 3 percent is less
than half of 7 percent, which means that the muscular flight motor can be run a
lot harder without cooking itself.
Reextension of muscle was mentioned as another place where shortterm
energy storage is common. We use some gravitational storage every time we
raise an arm. But we mostly use muscles in pairs or groups so one muscle can be
used to reextend another. For instance (as in Figure 9.4), the biceps on the front
of your upper arm both raises your forearm and extends the triceps on the other
side of your humerus. That triceps both lowers your forearm and extends the
biceps. A more important role for storage occurs in bivalve mollusks, such as
clams and scallops. The two half shells are brought together and held together by
muscles; elastic hinge ligaments reopen the shells and reextend the muscles. The
process happens rapidly in scallops, which repeatedly clap their half shells
together and squirt out water in brief bouts of swimming. Their hinge ligaments,
of the protein abductin, have a respectable 91 percent resilience. The 9 percent
loss to heat shouldn’t matter much for a water-cooled machine that runs for only
a few seconds at a time.
An important case of elastic energy storage came up in Chapter 5. Our hearts
beat; they’re pulsating pumps. Part of a heart’s work pushes blood directly
around our circulatory circuits. But part of that work, using the blood as a
hydraulic fluid, pushes out—stretches—the walls of our arteries. In between
beats the arteries then rebound, constricting elastically and pushing blood
onward. In this way the elasticity of your arterial walls reduces the extreme
blood pressure fluctuations generated by your beating heart and makes blood
flow more smoothly through the capillaries and other small vessels downstream.
Finally, energy storage can be used to get high acceleration, something of
especial importance to small creatures. To go any decent distance, a projectile
needs a high initial speed—no matter whether it’s jumping, being kicked, or
being shot. A projectile’s speed is at its highest when it leaves its pusher. The
smaller the creature, the shorter the distance in which that speed (“muzzle
velocity” for a firearm) must be reached. So a shorter distance demands a greater
acceleration to reach the same final speed. A flea, a grasshopper, and a kangaroo
have about the same takeoff speeds, but the flea must have about a hundred
times and the grasshopper about ten times the acceleration of the kangaroo. The
kangaroo can jump mostly with direct muscle power, but fleas and grasshoppers
use their muscle to store up energy by deforming resilient material—resilin pads
for fleas and chitinous cuticle for grasshoppers. They then use trigger
mechanisms to release the accumulated energy in short order. Many plants do the
same thing in order to throw their seeds. Their devices form a wonderfully
diverse lot. One way or another, all crank energy into elastic material; triggering
is variously done by drops of water, by drying to a critical point, by brushing
against animals, and so forth.

We commonly use four different ways of storing mechanical energy for brief
periods: gravitational, elastic, electrical, and inertial. Nature uses only two:
gravitational and elastic. Here again, ours looks like the more versatile
technology. Nonetheless, short-term energy storage matters more for nature.
Why? Simply because most of nature’s brief batteries solve problems that using
rotational devices, explosives, and other such tricks, we don’t ordinarily
encounter.
Two general points about the engines and transmissions of the two
technologies. First, the very intricacy of nature’s alternatives perhaps best points
up the usefulness of rotational machinery. Second, the behavior of engines
interacts with the character of transmissions. Each technology makes a great
diversity of transmissions perhaps because each, in its own way, calls on engines
that are few in number and fairly similar in performance to do a wide range of
tasks.
 Chapter 10

ABOUT PUMPS, JETS, AND SHIPS

How can human technology help us understand the nonhuman world? Let’s
ask the question of more complex devices than we’ve worried about so far,
turning to whole systems rather than individual components. We’ll look, in
particular, at three cases where we can make sense of what nature does by
combining physical rules with the practical experience of human designers. By
“make sense” I mean seeing order amid diversity and discerning rules that
transcend mere accidents of ancestry. The three cases are pumps for moving
fluids around, propulsion by means of jet engines, and swimming on the surface
of bodies of water. Notice that all are physical rather than biological categories.
Getting the most mileage out of these comparisons requires us to begin with
physical phenomena and consider pump, jet, and surface swimming rather than
heart, squid, and duck.

PUMPS
Our blood circulates; sap rises; a clam filters food; a squid jets. These share a
crucial common feature: In each case a pump moves a fluid. Still, hints of
commonality get hidden in the dazzle of diversity. Perhaps that diversity should
come as no surprise; beyond the diversity of pumping creatures, the pumps
manage a ten million-fold range of pressure.1
What does a pump—any pump—do? It uses power to raise the pressure of
fluid flowing through it. So three things matter: power output, increase in
pressure, and rate of flow. Power is the increase in pressure times how fast the
fluid flows (volume flowing per time, not the speed of a bit of fluid). Put another
way, a pump makes fluid go through some system (load) that it wouldn’t go
through by itself. If the load resistance is high, as when fluid has to flow through
a long and skinny pipe, then a lot of pressure produces only a modest trickle.
Conversely, if the load resistance is low—as with a short, fat pipe—a little
pressure gives a great gush of fluid. Obviously a pump needs sufficient power to
do its intended job. A bit less obviously a pump ought to be appropriately
matched to the resistance of its intended load.
The designer faces a choice. One pump might produce a great pressure but
deal with a relatively low flow rate. Another pump might invest its power in a
high flow rate but give only a small increase in pressure. If you use a pump
that’s a good pressure producer for an application that mainly needs a high flow,
things will go awry. I say this with the conviction that comes from learning the
hard way. Long ago, so long that the embarrassment has faded, I built a
recirculating flow tank (or flume) that used a two-horsepower centrifugal pump
to pump its water around. That pump, the largest that could be plugged into the
room’s electrical outlet, moved only two gallons per second. A few years later I
built a flow tank that instead used a marine propeller to push the water around; a
half-horsepower motor now moved over thirty gallons per second. So I got sixty
times as much flow relative to power expended, and flow is what matters in a
flow tank.
My first pump would have been a good choice for lifting water ten or twenty
feet or making it squirt through a nozzle; it produced unnecessarily high pressure
and too little flow. The propeller, by contrast, did good service as a low-pressure,
high-volume pumper. At least I wasn’t the first to pick a pump that gave the
wrong combination of pressure and volume. Before the Normandy invasion in
1944 harbor components were prefabricated and then, to prevent storm damage,
flooded with water and sunk off southern England. The pumps expected to
empty them for floating across the English Channel, while big enough, were the
wrong sort and couldn’t generate enough pressure. An alert naval officer forced
his superiors to face the problem by staging a persuasively unsuccessful
demonstration. To save the day, the pumps of the London fire department were
borrowed, no trivial accommodation at a time of air raids.2
FIGURE 10. 1. Several positive displacement pumps in common use. The piston pump is of course a tire
inflator. The diaphragm pump is often used to push gasoline into a car's engine, and the gear pump moves
its lubricating oil.

Two general classes of pumps can be recognized by their particular trade-offs

between pressure and flow. The first are so-called positive displacement or fluid
static devices; see the examples in Figure 10.1. They include the piston pumps
we commonly use as hand-operated tire inflators, the diaphragm pumps that
serve as fuel pumps in most automobiles, and the vane or gear pumps that move
a car’s lubricating oil around. In all these, either the volume of a chamber is
decreased, pushing fluid out some intended opening, or else the position of a
chamber is moved, and the fluid it contains moves with it.
Those of the second class are called fluid dynamic, or rotodynamic, or
kinetic pumps; several are shown in Figure 10.2. Some, just enclosed versions of
aircraft propellers, push fluid lengthwise down pipes. Others (like my centrifugal
pump) spin the fluid and thus fling it outward. Still others use the motion of one
stream of fluid to move another stream of fluid, without having any moving parts
themselves. One version of the latter is a jet pump. It squirts fluid from a nozzle,
drawing other fluid along with it. Another is an aspirator. Here, as in automobile
carburetors, the pressure reduction caused by the motion of one fluid can draw in
another fluid—what in Chapter 8 we saw in sponges, prairie dog burrows, and
similar systems.
FIGURE 10.2. Several fluid dynamic pumps. Centrifugal blowers are used for tiny cooling fans and for the
large circulators of forced air furnaces. In an aspirator (once in every automobile) fuel is drawn in by
motion of air past an orifice. A jet pump works in the opposite manner, with a high-speed flow out of a small
pipe drawing with it a flow in the large pipe.

While “fluid dynamic” sounds superior to “fluid static,” in fact neither one
wins in any general sense; each is suitable for a particular range of applications.
Fluid static or positive displacement pumps produce high pressures and low
flows, doing best with loads of high resistance. You need one, for instance, if
you’re to raise water up from a deep well. Fluid dynamic pumps produce higher
flows and lower pressures. Within each class, pumps cover a wide range of
pressures; thus the pump that was too strongly biased toward pressure for my old
flow tank was still a fluid dynamic one. Also, functionally competent exceptions
exist. For instance, air entering a jet engine first goes through an axial
compressor (Figure 10.5), a fluid dynamic pump that manages to reach high
pressures. It does so step by step, compressing air as it passes through a long
series of alternately rotating and stationary blades.
What of living pumps? Can the distinction that works for the pumps we build
help us recognize common features among nature’s? Nature’s pumps certainly
look nothing like ours, but they do the same job of pushing liquids and gases and
do it over an equivalently wide range of operating conditions. Table 10.1
summarizes the character and performance of several biological pumps.
Without much doubt,3 the biological pump operating against the highest
resistance is the evaporative sap lifter of trees and vines. It’s about as different as
can be from any common machine, but it’s a true positive displacement pump.
Recall how it works. A continuous column of liquid water extends from roots to
leaves through conduits (xylems) less than a millimeter in diameter. Evaporation
through the fibrous meshwork of the walls of cells in the leaves reduces the
volume of sap in them, and that in turn pulls water up from below. A column of
water a little more than thirty feet, or ten meters, high exerts a pressure of one
atmosphere on its base. So for every ten meters of height, a tree has to pump
water against a gravitational pressure of one atmosphere. In fact, it has to work
against two other sources of resistance. The skinny conduits offer a
hydrodynamic resistance to flow about as high as their gravitational resistance.
In addition, the soil around the roots (especially if nearly dry) may hold its water
with great tenacity. Separating water from soil is much like wringing out a wet
towel; it’s harder and harder to extract water as the supply decreases. Faced with
the problem, we inevitably switch schemes and resort to a state change—
evaporation—to extract the final bit and dry the towel. But the tree can only
raise liquid water so its roots can’t make that switch. Desert plants, in the driest
soil, have to pull the hardest; to raise water, they may work against pressures of
as much as a hundred atmospheres.4 That’s the pressure that would be exerted
on the base of a column of water thirty-three hundred feet high or the pressure
on the hull of a submarine thirty-three hundred feet down.

TYPE CATEGORY SYSTEM EXAMPLES

RESISTANCE
Evaporative positive displ. highest leaf sap sucker
Osmotic positive displ. very high root sap pusher
Valve and positive displ. high heart, bird lungs,
chamber squid jet
Peristaltic positive displ. high intestine, some
hearts
Piston positive displ. medium some tubicolous
worms
Vane or gear positive displ. medium other tubicolous
worms
Valveless chamber positive displ. medium jellyfish jet,
mammalian lung
Paddles fluid dynamic medium crustaceans in
burrows
Propellers fluid dynamic low hive ventilating
honeybees
Ciliary layer fluid dynamic low bivalve mollusk
gills
Flagellar fluid dynamic low sponge pumping
cells
Venturi aspirator fluid dynamic very low prairie dog
 burrow
TABLE 10.1. Various kinds of pumps used in biological systems, arranged in descending order of the
system resistance they encounter.

Another positive displacement device is an osmotic pump, also one of the

engines described in Chapter 8. Like the evaporative pump, it involves no
moving parts and strikes us as a strange machine. Offsetting relatively ordinary
concentration differences takes pressures of tens of atmospheres,5 so an osmotic
pump can work against a high load resistance. Osmotic pumps find their greatest
use in small-scale systems of one or a few cells, where very high pressures are
associated with only modest tensile stresses (recall Laplace’s law from Chapter
4). But they’re important in water secretion in our pancreases, in water
absorption in roots, and in pushing sap upward in stems—the latter a lower-
pressure process complementary to the evaporative sap lifter.
The best-known positive displacement pump is the valve and chamber heart,
shown as a generic device in Figure 10.3. A muscular chamber between two
conduits forms the basic element, with a one-way valve where each conduit
connects to it. By allowing only one-way flow, the valves ensure that one
conduit leads fluid in and the other leads it out. These pumps often produce
higher overall pressures by using several pumping chambers in sequence—atria
and ventricles, for instance— but none comes anywhere near the pressures
generated by the sap lifter. Our hearts, for instance, manage no more than a
quarter of an atmosphere, not tens or a hundred. The highest pressures, reached
by giraffe hearts and squid jets, are still less than half an atmosphere. The gills of
many fishes and the lung inflators of frogs also use valve and chamber pumps.
All the jet engines of animals appear to be positive displacement devices. Some
have valves (squid, for instance); others (such as jellyfish and the anal jets of
dragonfly larvae) have single, valveless conduits through which both squirting
and refilling happen, as in a kitchen baster. Nature’s valve and chamber pumps
resemble our piston and diaphragm pumps despite the absence of sliding parts
like piston rings in nature and of any actively contractile element like muscle in
our own devices.
FIGURE 10.3. Positive displacement pumps in nature: a human heart, with its valves and chambers, and a
marine worm, Arenicola, that lives in subtidal sand and, by passing waves of contraction down its body,
pumps water down through the sand and up through its burrow.

Yet another kind of positive displacement pump propels digesting food down
our intestines. In one of these peristaltic pumps, waves of constriction pass down
a muscular tube and push along blobs of fluids or slurries. Many worms have
hearts—or large blood vessels, since the distinction gets blurry—that pump
peristaltically. The ugly lugworm, Arenicola (shown in the figure), moves water
through its partially sand-filled (and thus high-resistance) burrow by passing
peristaltic waves down its body. Peristaltic pumps find only limited
technological use (they’ll come up again in Chapter 13, and one is shown in
Figure 13.1), mainly where we want to keep the fluid in the tube that conveys it
and where the inefficiency of the devices (lacking a decent analog of muscle)
isn’t prohibitive.
In our technology fluid dynamic pumps mostly use wheels and axles, so
nature’s versions (a couple are shown in Figure 10.4) don’t look much like ours.
Furthermore, our fluid dynamic pumps are big, fast things, so pushers, such as
fan blades, work well. By contrast, most of nature’s fluid dynamic pumps are
smaller and slower, so pushing with propellerlike blades is hydrodynamically
less effective.6 Instead the basic push works better if done in the fashion of a
submerged oar: by alternately moving something downstream in a high-drag
orientation and then upstream again in a low- drag position. Human technology
hasn’t used that scheme much since the demise of paddle wheels; propellers
blow away oars if they’re large. But despite such divergent appearance and
operation, the basic principles and the applications of natural and human-made
fluid dynamic pumps are every bit as close as those for positive displacement
pumps.

FIGURE 10.4. Fluid dynamic pumps in nature: a ciliated epithelium (as lines much of our respiratory
tracts) pushing a layer of mucus across itself and a sliced sponge, in which chambers of flagellated cells
draw water in through the general surface and out into the central cavity and apical opening. This sponge
is highly diagrammatic since in reality the chambers and passages are far too small to be visible in an
overall view.

Living or not, fluid dynamic pumps deal with low-resistance systems. Some
of the lowest resistances are handled by devices that we don’t think of as pumps
at all. These move large volumes of an external rather than an internal fluid.
We’re talking here about swimming and flying by moving some appendages,
whether by beating wings or paddles or by oscillating cilia or flagella. All such
locomotion requires processing huge volumes of fluid but imparting only slight
pressure increases. Slightly higher, but still relatively low, resistances face the
fluid dynamic pumps (mainly ciliated surfaces) that push mucuses around. Most
of the rest of nature’s fluid dynamic pumps are involved in suspension feeding.
Far more animals than most people imagine make livings in this quiet but
demanding way—from the simplest sponges to the largest whales. Most natural
waters contain edible particles, but they don’t often reach high concentrations.
So a lot of water has to be processed to get a little food. For the nutritional
benefit to exceed the processing cost, a separation system can’t afford great
pressure differences. Thus the task calls for fluid dynamic pumps. For
suspension feeders, life becomes a competitive battle of extraction efficiencies,
with the winner the one that can put the most energy into growth and
reproduction and the least into food collecting.
We have lots of data on suspension feeding pumps, mainly because they’re
used by bivalve mollusks, of both ecological importance and culinary appeal.
Soft-shell clams and mussels, pushing hard, can press about one two-thousandths
of an atmosphere, but in normal operation they (and sponges too) work at about
one hundred-thousandths of an atmosphere. But while these pressures sound
trivial, they push impressively high volumes; water equal to as much as half of
an animal’s body volume gets processed per second. (Our circulatory systems
don’t come close to these volume flows; even during vigorous exercise the
human heart pumps less than 1 percent of body volume each second, fifty times
less.) Similarly low pressures are involved when sponges use local water
currents to augment their pumping and prairie dogs ventilate their burrows; only
very low resistance systems can take advantage of ambient flows. Yet another
fluid dynamic pump working in the same pressure range is the honeybee hive
ventilator, a pump consisting of stationary bees beating their wings at the hive’s
entrance. This last can be a multistage pump, like the compressor of a jet engine,
since the bees often stand one behind another. Lacking decent ductwork, though,
it’s suitable for only low-resistance work.
Thus the distinction we make between our two general classes of mechanical
pumps helps us see order in some diverse natural devices; the same resistance-
based difference in performance drives the choices made by both technologies.
And we need help if we’re to understand nature. Just assuming that natural
selection leads to good design doesn’t take us very far. Not only do organisms
differ dramatically in size and anatomy, but their different lineages form separate
technological microcosms. The sap lifter isn’t available for use by a vertebrate
heart, nor can a plant order up any muscular device. But a few thousand years of
human technology let us see the forest through the trees. At the least it
counteracts the unnatural division of natural systems among people who study
different kinds of organisms, who consider different biological functions, who
publish in different journals, and who write different chapters in different
textbooks.
The analysis also helps us understand why certain arrangements don’t occur.
Our circulatory system uses tiny pipes—capillaries—to exchange material
between blood and cells, and it uses large pipes—arteries and veins—to
interconnect different capillary beds. Pumping of course is done with muscular
hearts joining the largest vessels. Why not build a circulatory system that uses
ciliated capillaries instead of muscular hearts? After all, blood flows through
capillaries at speeds consistent with ciliary pumping, and clams pump vast
amounts of water with cilia. Some years ago Michael LaBarbera and I guessed
that cilia were simply so much less efficient than muscles that the scheme would
be impractically costly— whatever the advantages of decentralization of
function.7 We should have argued that cilia, as fluid dynamic pumps, don’t
match the high resistances of circulatory systems. Keeping resistance low
enough for ciliary pumps would demand huge interconnecting vessels containing
a great volume of blood, along with extremely short capillaries—if the system
could fit into the body at all.
One final note on nature’s pumps for fluids, a point that might not occur to us
except through a comparison with human technology. The biologist studies what
occurs, not what doesn’t occur, since what doesn’t occur isn’t available for
examination.8 But the comparisons in this book have repeatedly drawn our
attention to oddly glaring omissions in the natural world. Each of these has either
told a tale or raised some provocative question. What follows is another of
nature’s peculiar omissions.
Potential mismatches such as using cilia to drive blood through a mammal
can be fixed with a device that trades off pressure change against volume flow.
We’re invoking nothing more radical than what levers do with force and distance
or electrical transformers do with voltage and current. Our technology uses many
such converters and has done so since antiquity. A device of the ancient
Mediterranean world called a noria is shown in Figure 10.5. A noria used the
flow of a stream either to drive paddles that ran a revolving chain of buckets or
to drive an undershot waterwheel equipped with buckets. Either way, a lot of
slightly descending flow in the stream moved a lesser volume to a much greater
height— high pressure and low flow from low pressure and high flow. A so-
called hydraulic ram, still in occasional use, accomplishes the same conversion
in a different way. A stream flowing downhill makes a small amount of water
flow considerably farther uphill—here by capturing the energy released when
some of the moving water stops. In effect, the sudden end of a surge of flow
drives a smaller volume to a higher level—repeatedly. A best-selling book of the
late 1940s, The Egg and I, talks about how much life on the farm improved
when the family got its ram; the modern urban reader might wonder about the
odd ovine allusion. All of nature’s valve and chamber pumps produce pulsating
flows, but no hydraulic ram has yet been described in nature.
We reduce pressures and increase volumes as well as the opposite. A ducted
fan-jet engine, as in Figure 10.5, uses a big fan in front to move additional air
through ducts that go around the engine itself. It thus gets greater volume flow
with less overall pressure change. (We’ll face the “why bother” shortly).
Animals commonly make one flow induce another, but no squid or jellyfish uses
such a converter. The combination of long fixed wings and small propellers on
an airplane does the same conversion. The propeller produces its thrust by giving
the small stream passing through it a large pressure increase. The long wing uses
some of that horizontal thrust to make a lot of air flow just a little bit downward.
So it converts a high-pressure, low-volume flow to a low-pressure, high-volume
flow in a different direction. But among flying animals only beetles make any
use of this separation of propeller and wing.

FIGURE 10.5. Old and new ways to trade pressure against volume flow. A noria run by an undershot
waterwheel increases pressure and decreases flow, so the motion of a stream can lift water into an
irrigation system. A fan-jet engine decreases pressure and increases flow to improve efficiency at subsonic
speeds. Air enters the front; fuel is pumped directly into the combustion chamber.

Nature uses many devices that make flows go faster or slower by changing
the sizes of pipes. The nozzle on the squid’s jet speeds the water, working just
like the nozzle on a garden hose. But all these trade speed for area, not pressure
for volume flow. The scarcity of pressure flow converters in nature is a puzzle.
Perhaps we’re missing something or looking at these systems in the wrong way.
Perhaps the extreme range of pressures spanned by nature’s pumps reduces her
need for converters. But that makes the improbable assumption that natural
selection has a wide choice of pumps in each specific instance. Or perhaps the
small size of organisms is an insurmountable burden. For small, slow flows,
viscosity places a severe tax on the efficiency of fluid mechanical devices, so
maybe nature simply can’t build converters efficient enough to be worthwhile.

JET PROPULSION
What could be simpler? Eject a fluid in one direction, and get propelled in the
other. If ambient fluid (usually air) is used, we call the engine a jet; if the fluid
comes entirely from within, we call it a rocket. The distinction doesn’t matter
much here, so we’ll refer to what all such reaction engines do as jet propulsion.
The process can drive an airplane, a surface ship, or a submarine; it can even
work in a vacuum. What could be more natural for an animal? One need only
wrap a muscle around a bag of fluid and give the bag a squeeze; out will come
fluid through any hole, deliberate or fortuitous. Someplace or other, it’s done by
almost every animal big enough for us to see. Squeezing our hearts sends blood
outbound; contracting our leg muscles helps send blood inbound again; a
traveling squeeze of the esophagus pushes food stomachward, while a similar
squeeze of the intestine propels the processed slurry farther aft. Expel rather than
just propel the fluid, and you have a jet engine.9 Furthermore, the most ordinary
fluid, liquid water, is nicely dense, flows easily, and thus serves admirably.
Plastic water rockets are splendid toys; just half fill the rocket with water and
pump air into the remaining space. When the rocket is launched, reexpansion of
the air expels the water downward. Recoil sends the rocket up a hundred feet,
and the experienced operator gets only minimally dampened by the effluent.
Nature has many such engines. Perhaps no other locomotory scheme has
independently evolved in so many different lineages. Jellyfish do it. And the
cephalopods—squid, octopus, and cuttlefish—do it. A scallop swims in short
bursts using a pair of jets on either side of the hinge of its shell. A young
dragonfly swims in a pond by squirting water out its anus. Frogfish maneuver by
squirting the water that has passed over their gills out through nozzles placed
more or less amidships. And on and on.
All these jets work like toy water rockets. A squid, certainly nature’s
champion jetter, has a water-filled cavity between an outer muscular sheath
(which got notice earlier as a hydrostatic skeleton) and its various viscera. When
a squid tightens that sheath, the water inside is forced out through a nozzle, and
the squid can get elsewhere in a hurry; fifteen to twenty miles per hour is
impressive for an aquatic animal less than a foot long. Avoiding an approaching
mouth, a squid can briefly outdistance all but cetaceans or the fastest fish. It can
shoot sixteen feet up from the ocean’s surface or take an arcing aerial trajectory
as much as fifty feet long.
So jets sound terrific. But while jets may be simple and common, no jet-
driven animal ever goes both fast and far. Squid can’t maintain top speed for
more than a few pulses. Jellyfish may swim steadily, but they reach only about a
quarter of a mile per hour. Frogfish do only about twice that, terribly slow by
piscine standards. Dragonfly larvae swim at a little more than one mile per hour,
and scallops up to one and a half miles per hour; both, like squid, perform only
intermittently. Where’s the rub? Is the limitation biological or fluid mechanical?
How we ourselves use jets is revealing. Jet and rocket engines may be the
oldest devices that burn fuel to make motion; in principle they’re the simplest of
all heat engines. The earliest-known steam engine was a direct jetting device; it
had only one moving part and needed little precision in its construction. This
was the famous engine of Hero of Alexandria, in the first century C.E., shown
(with necessary artistic license) in Figure 10.6. A fire heated water in a spherical
chamber mounted on an axle. The chamber had two tangentially pointing
nozzles through which steam emerged; the jets of steam thus spun the chamber.
But Hero’s engine never made anything better than a heroic steam whistle for
reasons that, as we’ll see, cast no aspersions on Roman culture.10 The steam
engines that drove the Industrial Revolution were based on a completely
different scheme.
Even after fifty years of development we mostly use these reaction engines
for very fast aircraft and almost never for cars or trains or boats. Why do both
natural and human technologies take this peculiar arm’s- length attitude toward
what seems an attractively simple and powerful scheme? Because the scheme
has a major and fundamental flaw.
The crux of the problem is the low efficiency of jets under all too many
circumstances. Here’s how it arises. By pushing fluid rearward ejecting
momentum—a jet engine produces a forward force. One might take a lot of fluid
and increase its rearward speed just a little. Or one might take a small amount of
fluid and greatly increase its rearward motion. All that matters is mass flow rate
multiplied by velocity. Air or water passing at one kilogram per second and
given an extra speed of two meters per second yields the same thrust as two
kilograms per second given one extra meter per second.
But the efficiency of the device cares a lot about the particular mix of
amount and speed of expelled fluid. For a given flow rate, the rearward thrust
you get depends directly on the speed you give the passing fluid. But the cost in
energy needed to produce that push increases not directly with the speed but with
the square of the speed. Double the speed, and you double the thrust, but
doubling the speed costs you four times as much energy. So you want to keep the
speed of the expelled fluid as low as possible; minimizing cost means using lots
of fluid and giving it only a little rearward push. There’s the rub. That’s just the
opposite of what jets and rockets do: They squirt small but rapid streams.

FIGURE 10.6. The steam engine attributed to Hero (or Heron) of Alexandria, first century C.E. A fire
beneath a metal sphere spins the sphere by causing a mix of water and steam to squirt out the jets.

How can an engine be made to use lots of fluid? Definitely not by making all
the fluid pass through its middle and out a nozzle; too little fluid can get through,
so what goes through has to go too fast. You need to take the opposite approach.
Instead of forcing a little fluid through the engine, you ought to force a lot of
fluid around the engine. Better to make the engine go through the fluid than to
make the fluid go through the engine. How to do this? Attach long, moving
appendages to the engine—fins, wings, paddles, or propeller blades. Any of
these will beat the jet in propulsion efficiency. All, though, are more complicated
than Hero’s engine or the simple squirting machine of a jellyfish. Nonetheless,
the complication pays off. A trout, waving its body and tail, deals with about ten
times as much water per unit time as a jetting squid of the same size. As a result,
the trout takes half the power to go twice as fast.
Should jets be dismissed as primitive because of their simplicity and
inefficiency? Are they just the bad hands dealt out by an animal’s ancestry?
Wrong as well as unfair. To swim or fly by pushing the local fluid backward, a
machine (living or not) must push it faster than the rate at which the craft goes
forward.11 Consequently, the speed of fluid emerging from its propulsion device
sets the craft’s maximum forward speed. If going fast is what matters, then jet
engines look a lot more attractive. That’s how we use jet aircraft: We make both
small and large ones, but we don’t make slow ones. A squid does much the
same. With small fins on its rear, it has a choice of systems. For slow swimming,
as when feeding, it mainly uses its fins. But when pursued by a fish or cetacean,
it turns on its jet and zips away. In a brief life-or-death maneuver, biological
fitness doesn’t turn on energetic efficiency!
For that matter, the jet engines of commercial aircraft have changed over the
past few decades, processing ever larger amounts of air and reducing their
average output speeds. They’ve added turbine-driven ducted fans, evolving from
pure jets to ducted fan jets (Figure 10.5), as noted when we considered pressure
flow converters. Early jets had small intake openings, and all the air coming in
entered a fanlike compressor prior to receiving its charge of fuel. Fan jets have
big intakes with large, conspicuous entrance fans. Engine designers have worked
hard to increase the amount of air going around relative to that going through the
combustion chamber—the bypass ratio—to improve efficiency and thus get
lower fuel consumption and greater range and payload.
Poor propulsion efficiency makes jet aircraft impractical for flight at the
sizes and speeds of animals. We know of no living jet aircraft—almost certainly
nature never made one—some engaging bits of science fiction (and flatulent
humor) notwithstanding. But oddly enough, jets aren’t all that bad for slow
swimming. In steady swimming, the thrust produced equals the drag incurred.
Low speeds generate disproportionately low drag, so balancing it takes very little
thrust. Some squid make lengthy migrations using their jets, but they do so at
speeds of around a body length per second, less than a mile per hour. Evading
drag simply by being slow probably underlies the success of slow jetters such as
jellyfish.
While the same fluid mechanical rules govern motion through air and water,
flying is a lot harder than swimming. The lower density of air may help a craft
go forward, but it demands an additional force to keep the craft aloft. This
additional component consumes (except for buoyant blimps and balloons) a lot
of power, whether or not the craft goes forward at all. For flight, never mind
moving slowly to take advantage of low drag. To the cost of going anywhere
must be added the cost of staying aloft, and slower flight means longer periods
aloft. In short, the drag-evading slow swimming of jellyfish or frogfish provides
no model for any heavier- than-air flying machine.12
To stay aloft, an aircraft must push air downward. Some air must be given an
increase in speed, but now the speed that matters is downward. Again the choice
is between giving a large amount a small speed increase and giving a small
amount a large speed increase. But for staying aloft by making air flow
downward, the relevant speed of the craft is how fast it’s ascending. If (as is
usual) it’s simply holding altitude, then ascent speed is zero. So the lower the
speed of the downward airflow, the better. Concomitantly, the greater the volume
of air moved downward, the better. Thus directing a small, high-speed stream of
air—a jet—downward is a particularly inefficient way to stay up. A small
downward-facing propeller works only a little better.
That’s why helicopters have long rotors and why tilting the engines of an
ordinary propeller plane from horizontal to vertical produces a very inefficient
hoverer. One military aircraft, the Harrier, is a jet that can hover, but it gulps fuel
when it does. The opposite extreme would be a helicopter with blades infinitely
long; it could hover at no cost at all!
This issue of propulsion efficiency explains a basic difference between flying
animals and ordinary airplanes. Birds, bats, and insects get both thrust, to go
forward, and lift, to stay aloft by beating their wings. Their wings ordinarily beat
not just up and down but to some extent fore and aft. Slower flight usually
comes with less up-and-down and more fore- and-aft motion; a creature just tilts
the plane of beating backward when it slows down, as in Figure 10.7. When
hovering at flower or feeder, a hummingbird has its head up and tail down, and
its wings mainly move back and forth. What a helicopter does when hovering is
almost identical. Like a hummingbird’s wings, a helicopter rotor moves in a
horizontal plane when it hovers, tilting the plane front down a bit for forward
flight. But ordinary airplanes use propellers to go forward and fixed wings to
stay aloft, successful airplanes preceded successful helicopters by about thirty-
five years, and helicopters are still profligate fuel consumers. We managed to
build airplanes when we gave up the bird or helicopter arrangement and began
using airfoils in two different ways on the same craft. (A crosswise slice of a
propeller reveals a shape just like that of a wing; they’re the same kind of
device.)
FIGURE 10.7. A hummingbird switches from hovering to rapid forward flight mainly by changing the plane
in which its wings beat from horizontal (back and forth) to vertical (up and down).

The fixed wing airplane produces its thrust by moving air backward at a
speed greater than that of the plane’s forward motion, which is usually quite
rapid. It produces its lift with long, fixed wings that give a large amount of air a
little downward push. What’s subtle is the power source for operating fixed
wings. It can only be the propeller since that’s the only thing that has an engine
attached. What happens is that the power to stay aloft is felt by the propeller and
engine as an additional drag that requires more thrust; when producing lift, a
wing (unless it’s infinitely long!) has more drag than when simply sticking out
into the moving air. In effect, the airplane produces forward thrust by using small
airfoils or jets operating at high horizontal speeds. It converts some of that thrust
to lift by using large wings operating at near-zero vertical speeds, as we noted
when talking earlier about pressure flow converters. The arrangement is
efficient, which is why, long after fine helicopters first became available, we
persistently build planes that take long runways to reach high takeoff speeds.
How then do birds, bats, and insects manage so well with only one set of
airfoils? The argument, alluded to in Chapter 3, turns on the relative sizes of the
flying machines of the two technologies. A wing produces lift in proportion to its
area. But an aircraft requires lift in proportion to its weight. After all, for steady,
level flight, lift and weight must balance just as do thrust and drag. Consider
what happens if the size of a craft doubles, with no change of shape or density.
All lengths—length, wingspan, etc.—will double. All areas—total external
surface, wing area, etc.—will go up fourfold. Volume and weight, though, will
increase no less than eightfold. So weight relative to area will double; the bigger
craft will have twice the weight relative to its wing area. Thus it will be
relatively lift-deprived, something not at all propitious for flight.
 Two evasions suggest themselves. The first is to use disproportionately large
wings for larger craft. The original Wright Flyer of 1903 had huge wings as does
the ultrasophisticated Gossamer series of human-powered aircraft.13 But both
are intentionally slow, disdaining the second evasion, which is to go faster.
Double the flying speed, and the lift of a wing goes up fourfold. (So a doubling
of weight can be balanced by a 1.4-fold increase in speed.) Ordinary aircraft may
weigh more relative to their wing areas than do birds, but they fly faster. Only a
few large birds can exceed fifty miles per hour in level flight, while only a few
specialized airplanes can fly so slowly. Larger size not only permits higher speed
(less total surface to incur drag relative to volume) but practically requires it
(less lifting surface relative to volume). Higher speed, in turn, requires a smaller
and faster air pusher. Airplanes gain efficiency by separating wings and
propellers. Birds, smaller and thus slower, find larger and slower air pushers
practical, and they have little cause to use fixed wings for lift and flapping ones
for thrust.14
But even flying animals aren’t completely out of the woods. Big animals
weigh more, relative to their wing areas, than do small ones. Big birds may fly
faster than small ones, but they don’t have enough wing area to hover
effectively; their propulsive systems move too little air too fast for decent
efficiency in hovering. Large aquatic birds often need to taxi furiously to reach
takeoff speed. A pigeon can hover for a few seconds, but it can’t maintain
sufficient power output to keep it up for long. Only hummingbirds are proper
avian hovercraft. But for small flying insects hovering is routine. These small,
slow creatures have a high ratio of lift-producing wing area to lift-requiring body
weight. With all that wing area they move a lot of air downward, so they don’t
have to move it downward very fast.
A pair of lessons: First, something absent in both, jet engines propelling craft
of moderate size at moderate speeds, draws attention to the unromantic but
inescapable issue of propulsion efficiency. That in turn provides proper
skepticism of proposals for jet backpacks, jet boats, and the like. Second comes a
point made repeatedly. Comparing situations where fixed wings work best with
those where flapping wings prove practical turns on the subtle, pervasive,
perhaps even pernicious influence of size. Size differences all too often
confound judgmental comparisons between what we make and what nature
makes.
 SWIMMING AT THE SURFACE
Two kinds of machines, surface ships and submarines, swim, and both
technologies make both. That every submarine we’ve built can also swim at the
surface shouldn’t be allowed to obscure basic differences between the two kinds
of swimming. One might guess that surface swimming is easier; only at the
surface can a thrust-producing appendage enjoy a recovery phase in low-density
air. Rowboats, canoes, and paddle-wheel steamers use aerial recovery, but it’s
ignored by all our better boats as well as by nature’s. So aerial recovery can’t be
all that valuable. A more important difference, one with the opposite effect,
emerges from the various ways water resists a swimmer’s motion. Beneath the
surface lurks ordinary drag, most of which can be avoided by careful
streamlining. The surface ship must also contend with drag from surface waves.
With less of its hull submerged, the surface ship avoids some ordinary drag, but
the saving is less than the ship pays in wave-induced drag. A submarine can go
faster with less fuel than a surface ship of the same size.
Both technologies build both kinds of swimmers, just as both build positive
displacement and fluid dynamic pumps and just as both build jet engines. Here,
though, we see a major bias. Nature’s swimmers are almost all submarines, with
only an occasional duck, muskrat, or water strider moving at the surface. Our
swimmers are nearly always surface ships, with self-propelled submarine
technology of any kind barely two centuries old. Even now, submarines are
limited to damn-the-cost military use. So we have two questions: What limits
submarines for us, and what’s wrong with surface swimming for living boats?
Answering the first question is easier but not as interesting as the second. For
one thing, we like to breathe air at sea-level atmospheric pressure, so we build
our submarines with rigid, pressure-resisting hulls. Such hulls have to be rigid
because with air inside, a submarine faces a peculiar instability. As it dives
deeper, the rising pressure tries to compress hull and air. That makes the overall
density of the submarine rise, so it gets less buoyant and all too eager to go
deeper yet. (Whale and skin diver contain air only in the lungs, although
compression of air in the lungs and the resulting transfer of nitrogen gas from
lungs into blood do present a serious problem.) In addition, most of our engines
breathe air, and they demand more oxygen than the humans inside the craft. So
fully submerged boats were first hand-cranked, then driven by electric motors
and rechargeable batteries and most recently by nuclear reactors driving steam
turbines.
 If nuclear engines were commercially available, we might use submarines at
least for transporting bulky and incompressible cargoes such as oil for long
distances. Fuel efficiency would be better, vulnerability to storms might be
reduced, and the whale-shaped ships themselves would be more compact than
present supertankers. A small pressure hull would of course be needed for the
crew.
The trickier issue is why nature’s swimmers so rarely keep their heads above
water. Avoiding the surface is remarkably widespread. Ducks swim on the
surface, but they do so slowly; they’re far faster either submerged or in flight.
Air breathers, such as penguins, seals, and cetaceans, do most of their swimming
underwater even though they have to come up to breathe. Nature’s submarines
vary in size from the smallest microorganisms to the largest whales, while her
surface ships encompass a size range from just below a centimeter to less than a
meter in length. By contrast, think how ancient and successful are our surface
ships, how varied in materials and designs, how diverse the cultures that have
built them, how many places on earth were settled long ago by people who came
by boat. In short, why do surface ships present a problem for nature, or does she
simply build such good submarines that surface ships are superfluous?
The culprits, waves, pose a worse problem for swimming animals than for
boats. The waves that give the most trouble aren’t the ones that winds generate
but the ones that surface swimmers themselves make. Neither duck nor ocean
liner can avoid making waves as it moves along. As in Figure 10.8, a floating
craft moving along the surface creates a pair of waves separated by roughly its
own waterline length—so-called bow wave and stern wave. That waterline
length (hull length) thus sets the distance between waves, the wavelength. Waves
always move, and their wavelengths set their speeds. As has been known for at
least a century, wave speed increases with the square root of wavelength; double
the wavelength, and you increase the wave speed about 1.4 times. By making
waves of longer length, the bigger boat makes waves that travel faster. With a
string, a stopwatch, and some toy boats (add weight if they ride too high) you
can test the assertion in any swimming pool.
FIGURE 10.8. An ordinary boat generates a pair of wave crests as it moves—a bow wave and a stern wave
—separated by about the length of its hull (here marked “L”).

Trouble comes because an ordinary floating ship or a swimmer finds it hard

to go faster than its waves. If it tries, a hill of water faces it, and it has either to
cut its way through or go perpetually uphill—up the down escalator as it were,
as in Figure 10.9. So a surface ship faces a practical speed limit, what’s called its
hull speed. Above its hull speed its drag increases suddenly and severely. Since
hull speed depends on hull length, the larger ship can go faster before hitting that
limit. Tinkering with the shape of the hull can have some effect on the speed and
suddenness of that drag increase, but the underlying problem can’t easily be
evaded. We build fairly large ships, and a hundred-foot-long ship reaches hull
speed at about a respectable fifteen or so miles per hour. Hull speed for a ten-
foot- long boat, about as small as we ever use, is about five miles per hour. For
the foot-long duck hull, it’s only one and a half miles per hour—for an animal
that can fly at thirty. For a muskrat the speed limit is even lower. Even
swimming humans, far from streamlined, can go faster underwater than on the
surface, despite giving up any aerial recovery phase for their arms. So much
better is underwater swimming for us that excessive use of it has been banned
for races lest fanatic competitors injure themselves; it’s healthy to breathe when
working maximally.
FIGURE 10.9. A small boat hits hull speed at a low speed; for this rubber duck being towed in a tank it's
uphill all the way above about one mile per hour. Notice the slight tilt of the duck at the highest speed
shown.

One can evade the limit by hydroplaning, scooting across the surface, which
we do in small to medium-size motorboats in fairly calm water. A few birds,
such as razorbills and loons, hydroplane for short distances. But it’s just not a
practical way to carry large masses for respectable distances. The problem then
is one of size. Surface ship technology is more attractive if you’re big than if
you’re small. So the profound divergence turns on nothing more obscure than the
difference in scale between the two technologies.
As if in the interest of symmetry or fair play, quite a different way of getting
around on a liquid’s surface works better for nature than for human technology.
Whirligig beetles, small black ovals (Figure 10.10), use it to dash around on
streams and ponds. An ordinary surface wave involves two competing
phenomena. The inertia of the moving water keeps the surface wavy15 while
gravity tries to flatten the surface, and the interaction of these phenomena sets
the speed of a wave. Gravity, though, isn’t the only agency trying to flatten the
surface; it’s just the important one for large waves. The phenomenon that makes
droplets round up, that causes water to bead on a waxy surface, that permits us to
rest a clean needle on the surface of a pan of water, that we minimize by using
soaps and detergents—surface tension—also flattens the water’s surface. Gravity
works at large scales, while surface tension becomes significant when things are
small. So the interaction of inertia and surface tension rather than inertia and
gravity rules waves shorter than about two-thirds of an inch.
If only gravity were relevant, then a centimeter-long hull could make only
about five inches per second. But when surface tension is taken into account, the
speed limit reaches almost ten inches per second. For a halfcentimeter hull,
surface tension makes even more of a difference: twelve inches per second
instead of three and a half. So the speed limit is by no means as bad as we might
have guessed for creatures in the millimeter to centimeter range of hull length.
What’s more, in this peculiar domain smaller is faster, not slower.

FIGURE 10.10. A whirligig beetle on a pond is a surface ship so small that it encounters waves whose
speeds are determined more by surface tension than by gravity.

Nonetheless, whirligig beetles are among the few inhabitants of this world of
small surface ships. (Water striders, springtails, and a few other creatures walk
on the surface rather than swim in the present sense.) Using surface tension
requires fairly calm water.16 Also, it’s not a friendly world for animals less than
a few millimeters long. For these still-smaller animals the speed limit isn’t the
problem. Instead surface tension itself becomes a trap, as you may have noticed
when a tiny fly falls onto the surface of a pond or puddle. Sprinkle a little talcum
powder in a bowl of water, and you’ll see the problem. Talc is several times
denser than water, but the particles get caught on the surface and stay there
persistently.

Pumps, jet propulsion, and surface swimming are each used by both
technologies. What do we learn from the comparisons? Both technologies make
elaborate use of pumps of diverse designs; for both the same rules link basic
design with practical applications. Both make substantial use of jet propulsion,
but each gives the scheme something less than a clear and sweeping
endorsement—for the same reason. Jets get used when something transcends
their inefficiency, something such as the desire for high speed, the need for a
brief burst of speed, the acceptability of very low speed, or the simplicity with
which preexisting structures can be modified. Both technologies use surface
swimming, but human technology finds it far more attractive than does nature.
The problem for nature is one of size, the difficulty that the mechanics of wave
propagation poses for small ships trying to move rapidly. Looking at pumps, jets,
and ships, we’re once again reminded that however divergent their aims and
appearances, the underlying constraints and imperatives are very much the same
for human builders and nature.
 Chapter 11

MAKING WIDGETS

Before they work, things must be made; to keep them working, they must be
repaired. We’ve looked extensively—exhaustively, the reader might declare—at
what natural and human technology make, but we’ve yet to ask about how each
produces and services its things. In no aspect do the two diverge farther. The
primary story is the biological one; being competent organisms gives us no
intrinsic sense of how we operate! But human technology needs a little attention
as well; participating in a modern industrial society doesn’t instill much
understanding of that either.

THE LIVING FACTORY

We large creatures operate in two distinct domains. On one hand, our pieces
make up a single product, an organism tested against other organisms in the
marketplace of natural selection, with reproductive success as criterion. On the
other hand, the factory that makes our pieces is the cell. A tiny organism may be
just a cell, but the animals and plants we see about us are decidedly
supercellular. A human contains about 100,000,000,000,000 cells (a hundred
trillion, or 1014). Assuming a typical diameter of a hundredth of a millimeter, in
single file such cells would stretch a million kilometers, a thin line encircling the
earth about twenty- five times. These microscopic production units vary little to
either side of their typical length of four ten-thousandths of an inch. It’s no
accident that the ancient icon of the biologist, the light microscope, views
structures at this scale. Biology textbooks discuss levels of organization such as
organs and communities, but only the species approaches in importance the cell
and the organism.
If this dualism of cell versus organism strikes you as theological or
metaphysical, expunge such heretical thoughts; this is reporting, not New Age
analysis. Never mind, at this point, why a cellular scheme of organization was
retained almost every time macroscopic organisms evolved. What matters here is
the strangeness of a manufacturing system whose products are larger than its
factories. This is cottage industry with a vengeance, and it has important
consequences for how nature makes and maintains her things and for what
particular things she makes.

Building Up, Not Down. The basic production units, cells, make things bigger
than themselves, organisms, from parts smaller than themselves, molecules. The
nearest equivalent of a single cut, cast, or pressed part is a single protein
molecule. If a cell were simply filled with such molecules and nothing more, it
would hold roughly 10,000,000,000 of them (ten billion, or 1010). Despite their
small size, these parts are complex. A protein may be a polymer of a few
hundred monomeric chemical units strung together, but it differs in two
important ways from any polymer produced industrially. The monomeric units
(amino acids, if you need a name) are nonidentical, and the sequence in which
different ones are strung together is all-important. Not just the proportions but
the specific sequence matter. Complexity starts within the molecules themselves.
Making proteins is the most important thing that cells do. These proteins
either get used within cells or get exported, mainly to form intercellular
structures or to flow in intercellular fluids. As a manufacturing system a cell
should find it easiest to make things smaller than itself. Of course assembling
subcellular things made of several kinds of proteins isn’t necessarily automatic.
But it should be a lot less tricky for a cell than making things larger than itself.
The evidence of microfossils in the oldest rocks accords with this view; for most
of the time that life has existed on earth, supercellular structure was limited to
whatever gluey stuff stuck a few cells together.1
FIGURE 11.1. The hierarchical structure of a tendon.

The larger things made by organisms reflect this process of building up from
the molecular level. A steel beam is uniformly composed of steel, while the
wooden beam taken from a defunct tree has level beneath level of structure. In
making metals and plastics, we try to achieve material homogeneity, while living
systems seem strongly biased against homogeneous materials. Homogeneity
isn’t reached until one gets well below cellular dimensions. A tendon (Figure
11.1), for instance, isn’t just a stiff elastic band; it’s divided into fascicles, which
are made up of fibrils, which are formed of subfibrils, themselves composed of
microfibrils, the latter mostly bundled triple helices of amino acid chains. Even
hair, of a single protein, keratin, has a hierarchical (or as Julian Vincent
shamelessly quips, hairarchical) structure.2 Nature routinely and easily
introduces organizational complexity at a microscopic level, at least compared
with the barriers that face our own fabricational techniques.
Nature uses composites for all her rigid materials while we crow a bit when
we make a composite that’s competitive in a nonmilitary marketplace. Perhaps
nature just does what comes naturally, taking advantage, though, of some
coincidence in how cell size coincides with the best size for the components of
good composites. Only a little coordination among cells yields highly
anisotropic composites—composites with regularly rather than randomly
arranged components, composites whose properties depend on the direction of
loading. We do that with fiberglass, making sheets in which the glass fibers run
in the same plane and rods in which all the fibers are parallel. But even the
fanciest fiberglass is monotonous next to wood or bone, as in Figure 11.2.
FIGURE 11.2. Recognition of the complex composite character of bone and wood is no new thing. Here are
micrographs of lengthwisey glancing (tangential) sections of the two, from the 1878 edition of Gray’s
human anatomy and Sachs’s plant physiology of 1882.

Making a biological composite amalgamates the outputs of a multitude of

microfactories to good advantage. For other biological products, any superiority
of cellular over macroscopic synthesis is less clear. What can’t be doubted,
though, is that cellular synthesis has consequences far beyond the internal
arrangements of structural materials. Consider muscle: It differs from an electric
motor in a way not yet mentioned. The motor is a single machine; omit almost
any part, and it will fail to work, and no portion of the motor can run by itself.
The muscle, by contrast, is an amalgam of small, identical units, the sarcomeres
(Figure 8.9), each about two micrometers, about one ten-thousandths of an inch,
long. Whether one operates doesn’t depend at all on whether the others can do
so. We put that independence to use, adjusting how hard a muscle works by
changing the number of motor units—somewhat larger operational elements than
sarcomeres—that are active.
Nor is muscle the only living machine in which larger simply means more of
the basic units. A kidney and an intestine and a liver look substantial enough,
with their functions, as organs, described in textbooks. But the functions of these
organs are just the functions of identical elements made up of individual cells or
small groups of cells. Each of us carries a strong sense of organismic
individuality, and none of us holds special affection for our cells as individuals,
yet we’re minimal confederations of these tiny elements. In even the most
complex organism, far more information moves within cells than between cells.
Even our brains are bit players compared with the rate at which information
from the genetic material directs the synthesis of protein, an entirely intracellular
activity.
Put another way, nature achieved something glorious when, something more
than half a billion years ago, she invented well-integrated, multicellular,
macroscopic organisms.3 Nature builds up more often than she miniaturizes.
With rare exceptions, large size is the specialization, whether in how creatures
work, how they mature as individuals, or how they evolve. Evolution has
engendered big things many times, but major evolutionary change happens
mostly in small organisms.4 In short, big creatures usually descend from small
ancestors, not the other way around.

Recipe or Blueprint? “Bake until golden brown.” “Boil, stirring constantly,

until the temperature reaches 230 degrees.” “Reduce heat,until just barely
simmering.” Each instruction specifies an end point rather than a specific course.
Each demands that you observe something and then alter a process on the basis
of that observation. Each employs feedback, called that because you feed
information about results back to control a process.5 In the first two instances
you terminate the process, while in the third you adjust something. By analogy
with cookbooks, we’ll call these kinds of instructions recipes, even if recipes for
cooking don’t inevitably involve feedback. By contrast, instructions may be
outcome-independent and lack such an informational loop; we’ll call these latter
blueprints. The level of detail doesn’t matter; what’s relevant is whether or not
the instructions depend on the results. Figure 11.3 puts a result-dependent
scheme in more general form.
Feedback control may be automatic, with no human link in the feedback
loop. The earliest clear cases of automatic controls, according to Mayr,6 were a
few float valves of classical and early Islamic civilizations. These worked like
the ones we use to regulate the water level in the tanks of household toilets: High
water raises a float and turns off the water supply, as in Figure 11.4. Early in the
seventeenth century Cornelis Drebbel, a remarkably prolific Dutch inventor
about whom we rarely hear, invented thermal controls for ovens and incubators.
In them, high temperature reduced the rate of fuel combustion. One of James
Watt’s finest accomplishments was a governor for a steam engine; an increase in
the speed of the output shaft cut back the steam entering the engine. Feedback
controls now occur everywhere: temperature controls on ovens, furnaces, and
refrigerators; error-correcting protocols in modems; load-dependent speed
controls on motors; and on and on. Nor have feedback controls with human links
become obsolete or superfluous. You close a feedback loop when you steer a car.
If it drifts toward the left, you turn the front wheels a little rightward; when it
drifts too far rightward, you turn toward the left. No road runs so straight that
you can take aim and then close your eyes.

FIGURE 11.3. The crux of a feedback system is the ability to adjust what it does depending on conditions
outside itself where those conditions include the results of its own actions. In a strictly mechanical and
formal sense, it has self-awareness.

In organisms, feedback controls in physiological systems get most attention.

Such things as the tension produced by a leg muscle, the output of the heart, the
rate of breathing, and the diameter of the pupil of each eye are result-regulated.
Lift one leg off the ground, and the muscles that stiffen the other leg detect the
increase in load and pull harder in compensation; you don’t sag down toward the
ground, as would a system that wasn’t keeping track of what it was
accomplishing. Less widely appreciated is how extensively feedback loops lace
together the process of making an organism in the first place. Its genetic
material, DNA, is often referred to as the blueprint for making an organism. If
“blueprint” implies a detailed and complete set of instructions that need only be
read out by the synthetic machinery, then the word misleads.7
FIGURE 11.4. A specific feedback control system, as machinery and as a formal scheme—the water-level
control of a household toilet. Not only will it restore the level after a flush, but it will compensate for
evaporation from the tank and is unaffected by changes in water pressure.

DNA in fact functions as a recipe in the present sense. No instructions could

be precise enough to make something as complex as an organism without
adjusting course on the basis of how things were going. You might program a car
to drive a hundred miles without running off the road, but only if you equipped it
with lane-detecting sensors whose outputs feed back to the steering apparatus.
Otherwise, if there’s even the tiniest alteration in tire pressure, if a passing cloud
causes the road temperature to drop ever so slightly, if a crosswind comes up,
you’re done for. Chemical processes, such as those central to the development of
an organism, depend both on the concentration of reactants and on temperature,
and different reactions differ in their exact dependence. A minor change in
environmental temperature, in the concentration of some minor ion, or in any of
a host of other variables would fatally derail development. Without elaborate use
of feedback, the possibility of success would be remote.
The realization of how self-regulating were the processes that generated
organisms caused shocks that rumbled through biology and philosophy. In 1891
Hans Driesch, a German embryologist, let a fertilized sea urchin egg divide into
two cells. He then separated the cells and watched the further development of
each. To his surprise (frog eggs had given the opposite result for Wilhelm Roux
a few years earlier) each cell went on to make a perfect, albeit smaller, larva.
That any cell could so reorganize its developmental program converted Driesch
from his mechanistic view of life into a staunch supporter of vitalism—the idea
that life involves something not explicable in terms of physics and chemistry.
Retrospectively, Driesch’s problem was the unimaginability of feedback control
at a cellular level to the biology of his time.
Up to this point the development of an organism doesn’t sound all that
different from modern industrial practice, where machines and operators provide
lots of nice error-correcting feedback control. But nature takes things a step
farther, and the word “recipe” gets even more apt. To a large extent,8 two
products of a modern manufacturing process perform alike because they’re
formed alike. We most often achieve acceptably precise products by making
them of equally precise components. By contrast, two organisms, even
genetically identical ones, resemble each other only superficially. The closer one
looks, the greater the differences one sees. We put names on the large blood
vessels but not the small ones; the latter vary from individual to individual.
Nature strives for fitness rather than for precision or consistency, and at every
stage development is goal- regulated rather than programmed precisely. The
investigator can often remove large pieces of early embryos with little or no
postpartum effect. In their DNA itself, different individuals vary greatly; nature
tolerates and (in an evolutionary sense) even values diversity. That of course is
why we can distinguish among individuals by testing the DNA of blood samples
and other bits of body. Some of the variation among individuals may be put to
use in social species to facilitate recognition of one member by another; that’s a
fancy way to say that we’d have problems if all of us looked alike. But most of
the variation in detail serves no immediate purpose; it’s tolerated because it’s
without effect on our fitness.
Without the gene you couldn’t make the structure—that’s usually how we
identify genes—so the gene is indisputably necessary. But while necessary, the
gene isn’t sufficient, even in a purely informational sense. An earlier generation
of developmental biologists coined a name, equifinality, for the phenomenon
they uncovered when they (starting with Driesch) saw how much abuse embryos
could stand without affecting the viability of the complex organism that resulted.
The word “equifmality” isn’t much used anymore. Words may be poor
substitutes for understanding, but this one did call attention to how nature, using
elaborate feedback control, makes machines only crudely similar in structural
detail yet exquisitely similar in performance.

Scheduled Maintenance. Almost no machine serves its full term without

attention to its well-being. Beyond incidents and accidents, some part will wear
out before the whole contraption deserves scrapping. Perhaps a temporary part,
such as a filter or a flexible gasket, is handier or more economical than a
permanent one. Or perhaps a part works by wearing out, as with the brake pads
on a car. Or perhaps routine operation of the machine depends on cleaning filters
and periodically adding lubricants. Do living machines need such scheduled
maintenance? Being out of service for repair risks the continued existence of an
organism; the weak, the sick, and the injured are neither effective predators nor
evasive prey. Perhaps sleeping and hibernating play some role in maintenance,
but most organisms do neither. Or are organisms designed so maintenance
doesn’t put them out of action?
What happens is truly remarkable. Organisms continuously rebuild nearly
everything within themselves! So unexpected and inconspicuous is the
phenomenon that it long escaped either suggestion or notice. Rudolph
Schoenheimer came across it during the late 1930s, when he began feeding mice
isotopically labeled compounds, which had just become available.
Schoenheimer’s account, entitled The Dynamic State of Body Constituents, is
still worth reading. He missed on a few details, but most of the work has stood
up well. Schoenheimer died in 1941, shortly after giving the lectures from which
the book derived. He would probably have been awarded a Nobel prize had he
lived; unfortunately the prizes aren’t given posthumously.
Isotopic labeling allows marking atoms so they can be distinguished from
other, chemically identical atoms. By feeding animals food that had labeled
nitrogen atoms, Schoenheimer could look at where those new nitrogens went
without being swamped by the huge number of nitrogens already in their bodies.
He wanted to see how the extra nitrogen came out—which excreted compounds
ended up with the labeled atoms. Specifically, he fed nongrowing adult mice
protein with labeled nitrogen. Extra protein input led to extra nitrogen excretion,
in the form of urea, the ordinary and expected result. But, unexpectedly, much of
the labeled nitrogen failed to appear in the murine urine at all; unlabeled
nitrogen took its place. That unlabeled nitrogen could have come only from the
existing protein of the bodies of the mice. Since the mice weren’t shrinking or
otherwise deteriorating, they must have been adding a lot of new protein to their
bodies, protein that exactly matched in amount and composition the old stuff that
was converted to urea and excreted. The expectation was confirmed by finding,
postmortem, labeled nitrogen in their body proteins.
The conclusion is inescapable. An organism isn't an engine in which food
simply supplies energy and any material needed when the engine needs
servicing. Instead much of the food becomes part of the living machine even
when it’s not growing; simultaneously an equivalent amount of machine breaks
down and gets excreted. This happens almost all the time in all organisms and
for almost all the material within the cells of organisms. Remarkable:
Intracellular material gets continually replaced without changing the overall
composition of either cell or organism. The organism retains the same structure
and remains made of the same kinds of molecules, but the particular molecules
residing and participating are ephemeral. Hence Schoenheimer’s phrase
“dynamic state of body constituents.”
This dynamic state represents a profound difference between living and
nonliving systems. The same molecules the ancient Egyptians hauled into place
more than four thousand years ago still make up the pyramids, but you're not the
same person you were a year ago. Organizationally, yes, but materially, no. The
organization of the individual organism persists far beyond the material of which
it’s made. And the replacement process doesn’t just copy what’s scheduled for
demolition; it’s an entirely fresh synthesis. The instructions for making our most
critical and complex material, protein, get continuously reread from the genetic
material. Cells don’t need to divide to do it, and organisms give no
morphological or microscopic hint of their dynamic state.
Why bother? And why bring up protein turnover, a biochemical story, and
the dynamic state of body constituents, a physiological story, in a biomechanical
tale? To begin with, proteins are not the stablest of compounds, and their
stability decreases (they rot faster) as the temperature goes up. Warm-blooded
animals—mainly mammals and birds—keep their bodies at temperatures high
enough for protein to go bad at a significant rate,9 and even non-warm-blooded
organisms face a slower version of the same kind of spontaneous deterioration.
That makes maintenance mandatory. The schedule for replacement tells us a lot
about an organism’s problem. The more stable structural proteins are the most
persistent; the less stable soluble proteins the least. It’s handy to speak of “half-
life” the way the physicists do when they compare how fast different radioactive
isotopes break down; 'half-life here is how long it takes for half of a given kind
of material to be replaced. The average half-life of the proteins of a rat’s carcass
—muscle, tendon, and bone—is twenty-one days, while the half-life of the
proteins of its liver and blood plasma is only six. The average half-life for the
proteins of an adult human is about eighty days. So perfect is the replacement
process that you can recall the events of many years ago.
All intracellular proteins participate in this, dynamic state. (Spontaneous
changes in the genetic material, DNA, by contrast, are repaired—usually—
without wholesale replacement.) Extracellular stuff, such as hair, we simply
make and discard continuously. Mammalian red blood cells are discarded in their
entirety, with an average life-span of 120 days. But they’re not true cells; with no
nuclei, they can’t direct the resynthesis of their protein. So we have to remake
them in their entirety.
In short, nature doesn’t make her widget but keeps making her widget in an
unending Sisyphean process. Once again, why bother? Probably the instability of
proteins makes continuous maintenance necessary if they’re to be usable within
organisms. Probably because they’re made up of specified sequences of amino
acids, they’re complex and informationally rich enough that no error detection
system could do the whole maintenance job. So probably their combination of
complexity and instability leaves organisms with no alternative but a wholesale
replacement schedule.

Unscheduled Maintenance and Dynamic Alteration. In addition to random

molecular disarrangement, organisms sustain large-scale breakage and injury.
Bones fracture, skin tears, and appendages get bitten off. Sometimes controlled
proliferation of adjacent cells just covers wounds with scar tissue, while
sometimes injuries get repaired so precisely that no mark remains. We’re battle-
scarred by adulthood, despite modern medicine’s intervention. How much repair
and regeneration go on varies greatly among organisms and their parts, but all
multicellular animals and plants manage some degree of restoration.
Repair and regeneration may be familiar and expected, but they’re no trivial
tasks. Even making a bit of scar tissue where a branch has broken from a tree or
where a person has been cut or burned demands that the system have
extraordinary self-knowledge. Cell proliferation has to be stimulated, directed
appropriately, and then turned off. For full regeneration, at which we humans
happen not to be especially good, the right kinds of cells have to be made in the
right places at the right times, blood and nerve supplies have to be reconnected,
external covering layers have to be extended, and so forth, all as things were
before. Somehow nearly autonomous cells, acting in response to their individual
instructional codes, must carry out the complex task in precise coordination.
Injured vertebrates are especially vulnerable to death from infection or blood
loss, but they still patch themselves up after severe tissue loss. Salamanders and
lizards can completely regenerate lost tails; a salamander can re-create an entire
leg. Invertebrates, though, can do much more spectacular feats. For instance, a
starfish can regenerate well over half of itself. This came as an unpleasant
surprise to some oyster gatherers, who had been inadvertently increasing the
numbers of their adversaries by cutting oyster-eating starfish in half and tossing
the halves back in the water. Some simpler flat- worms and sea anemones that
live on sufficiently hard surfaces deliberately confound regeneration and
reproduction; one becomes two when parts simultaneously have a mind (or half
a mind) to crawl off in opposite directions.
Nor are replacement and regeneration the full extent of nature’s provision for
perpetual care. Another is so familiar we forget how special it is. Exercise a
muscle, and it gets larger; put an appendage in a cast for a month or so, and the
muscles atrophy as the body recycles their constituents. Bone responds similarly,
losing mineral and thus softening if out of use and increasing in density in
response to repeated loading. Increase the blood flow to an organ—say, by
exercise-induced proliferation of its capillaries—and the blood vessels that
supply that organ widen.10 A lot of the degeneration associated with aging may
reflect disuse rather than old age itself. Lengthy spaceflights cause all kinds of
serious physiological deterioration; fortunately most of the changes are
reversible. The components of animals like us continuously readjust themselves
in response to changes in how we use them.
Usage-tuned structural change isn’t only an animal thing. Add weight to the
branch of a tree, and it grows thicker. Bend a branch down to the ground, attach
a weight to it, and it will (if the weight isn’t excessive or the branch damaged)
slowly lift the weight off the ground. Cut down one tree, and the adjacent ones
grow longer branches where their illumination has been increased. Load one side
of a tree, and the trunk will grow additional wood neatly positioned to offset the
additional stress.11 The evolutionary theory that preceded Darwin’s, the
Lamarckian idea that increased use of a feature led directly to its hereditary
amplification, simply extended these familiar phenomena from individuals to
lineages. The extrapolation, an entirely reasonable one, just doesn’t happen to be
correct for life on earth.
We now know a lot about how organisms manage to be so responsive, but
the details matter less than two larger points. First, the routine character of such
responses in living machines stands in sharp contrast with the absence of almost
any equivalent in nonliving machines. An archaeologist decides whether a rock
might have been a tool, or how a tool was used, or whether some item was
decorative or utilitarian by checking for wear and examining the pattern of wear.
By contrast, the same archaeologist learns how people did repetitive tasks, such
as grinding grain, by studying how bones have hypertrophied—worn larger, as it
were. Only extracellular parts, ones that weren’t strictly living in their functional
heyday, wear down from use—tooth surfaces and mollusk shells, for instance.
The second point is that this repair, regeneration, and demand- responsive
alteration turns on feedback control. The system must compare what ought to be
with what actually is, and that takes information about what it has or hasn’t
done. The whole process depends on taking action to minimize any difference
between goal and present state, defining, again, feedback control. The key
elements of the system are the sensors that inform it about its present state. More
than anything else, they distinguish natural technology from our own. We
understand quite well the sensors used in controlling the positions of our
appendages or the rates at which our hearts beat—parts of high-speed
neuromuscular and neuroendocrine feedback systems. We also know a lot about
the information-carrying links, things such as chemicals that diffuse away from a
sensor or else move in blood or other internal fluids. One rarely thinks of
sensory equipment in plants. But growth away from the earth depends on gravity
detectors, and growth toward sunlight can’t happen without photoreceptors—just
to point to a few.
Organisms control themselves at an even more basic level. With few
exceptions, every cell contains the full set of genetic instructions the complete
recipe book. Most of the time, though, most of the information in most cells
stays on the disk, even under the dramatic demands of regeneration. Except in a
few instances—cloning being the most notorious—we can’t call the information
back into functioning software. Organisms are highly inhibited, elaborately
repressed. To keep everything from being attempted all at once, a lid must be
kept on almost all the possibilities. The repression may fail, as when cells that
should know better produce a cancerous growth in a reproductive orgy. In a
certain sense this restrained totipotency isn’t uniquely biological—only
predominantly so. Long ago the neurobiologist Sir Charles Sherrington pointed
out (by analogy with the brain, as it happens) that when you use a telephone, the
important thing isn’t so much getting the right connection as avoiding the entire
world of wrong numbers.
In nature, then, the processes of production and maintenance intermingle so
thoroughly that we can’t make any tidy distinction. Some individual organisms
(like ourselves) may grow to fixed sizes while others (such as many trees and
fish) may keep growing however long they live, but for neither does
development ever finish.
How Good Is Good Enough? Living machines are a competitive bunch, doing
overt or subtle battle for nutrients, for energy, for space, and for mates.
Competition produces winners and losers, those more fit and those less fit. The
conflicts resemble those of unrestrained capitalistic economies enough to have
generated the nasty doctrine of social Darwinism. The latter is basically the
claim that well-to-do individuals, races, or nations are well off because they are
biologically fitter and that while the not so well off might deserve sympathy,
attempts to alleviate their plight will inevitably founder on their lower fitness.12
The inapplicability to human society of fitness in anything like its evolutionary
sense is fortunately a fatal flaw for the notion.
But recognition of competition in nature brings up a different parallel with
human activities. How good is biological design? The assumption of good
design in nature, at least within her intrinsic constraints, pervades studies of how
animals work. Whether one justifies the assumption by natural selection or by
divine omniscience makes little practical difference.
In recent years the assumption of good design has been variously reassessed,
criticized, qualified, and subjected to quantitative analysis. Strict optimality
(what we might call perfectionism) has taken a well- deserved beating from the
evolutionary biologists.13 To a large extent, though, they’ve criticized a straw
man; good design isn’t a rigorous principle but a working hypothesis of
physiologists. We know that organisms aren’t paragons of perfection—after all,
no biological element has evolved for an unlimited time in an unconstrained
context—but the assumption that we’re rationally assembled without a host of
useless features provides a reasonable starting point for investigating how our
features function. We do seem well tuned, however hazardous the assumption of
good design. The practical problem for studying how we work is that most
structures are multifunctional, and it’s rarely self-evident which function was
preeminent in determining the design of a structure.14
But well tuned or well designed doesn’t imply perfectly precise production;
we’re just not built that way. Uncertainty and scatter in scientific data often
come from inaccuracy in measurements, but in biology they as often reflect real
variation among the things being measured. Constancy of the speed of light or
the mass of a carbon atom may be limited only by our ability to measure them.
But the diameters of human liver cells vary intrinsically. Such natural slop and
scatter don’t get much attention. At least from anecdotal evidence, nature’s
tolerance for variation itself varies from structure to structure. She builds some
things to very tight specifications and other things less consistently; natural
selection must target variability per se.
Many things in nature must vary less than what would be tolerable without
loss of fitness. Where no extra cost comes from excessive standardization it
ought to persist. For instance, many substitutions can be made in the amino acid
sequences of proteins without changing how well they work.15 But most such
substitutions rarely occur; the synthetic machinery doesn’t have the option of
inserting one amino acid instead of another when one is more abundant and the
consequences of change are insignificant. (On the other hand, individual-to-
individual variation may enhance fitness. Where offspring are released into an
uncertain environment—climatically or otherwise—and only a few will survive,
cloning may not maximize the number of survivors. That’s one of the arguments
commonly advanced for the ubiquitousness of sexual reproduction.)
We can ask in another way about the quality of natural design. The sagacious
engineer, mistrusting all estimates and assumptions, designs everything to be
better than minimally adequate. We prefer to overbuild. Does nature? The ratio
of the load that would produce failure to the greatest load expected in use is
usually called the safety factor. Any safety factor above one indicates
overbuilding.16 But determining nature’s safety factors turns out to be far from
easy. For one thing, natural selection doesn’t anticipate, and safety factor in its
usual sense means anticipating possible failure rather than reacting after the
event—whether by redesign by the engineer or by subsequent selection in
nature. For another, we can’t easily estimate the greatest expected load for a
natural structure. Some things are loaded steadily and predictably while others
face highly variable loads. For still another, the service life of natural products
varies greatly. Worse, organisms vary in their life histories as well as in their
lifespans. An oak tree that blows over in year ten loses all its fitness if it
normally grows skyward for twenty years before making acorns. The vine on the
tree loses a lot less if it has been putting out seeds since year five. Most human
products give full service from the start, while for organisms true profit awaits
the reproductive payoff.
Despite such formidable complications, some biologists have had the
temerity to tackle safety factors and the cost-benefit analyses underlying them.
The loss of a branch for a tree or a tail for a lizard costs little, and both losses are
ordinary events. So branches and tails aren’t attached with high safety factors.
The buoyancy chambers of two deep-sea cephalopod mollusks, cuttlefish and
nautilus, face collapse from the pressure of the water around them. Their safety
factors are only about 1.4 relative to the pressures at which they live. But those
pressures are especially constant and predictable. Bones and tendons have higher
safety factors, running between 2 and 6, with tendons usually lower than bones.
Flying animals have lighter bones than nonfliers; probably their bones have
lower safety factors because for creatures that fly, excess weight carries a great
penalty. Tree trunks have safety factors around 4, and the stems of annual plants
around 2, although data are limited and uncertain.17 We care a lot how plants
behave since we deliberately alter them for agricultural purposes, since storms
are notoriously irregular, and since whole fields of crops and whole plantations
of farmed trees do blow over from time to time.

AND THE NONLIVING FACTORY

Since few of us ever set foot on a factory floor, how we make things is only a
little more familiar than how nature makes them. That’s unfortunate because we
live by the products of division of labor, mass production, and assembly lines.
Handcrafts are viable only for occasional and optional use, unless we accept
drastically reduced economic conditions. Farming methods increasingly reflect
the same economics of specialization, scale, and labor minimization; the
purveyors of fast food bring the same imperatives to yet another domain.
Even though ancient Egypt, Rome, and China manufactured a diversity of
items, mass production is recent. Many historians, especially in America, have
described the evolution of the modern factory; the New World may have led the
Old in the process. North America experienced a slightly different industrial
revolution from northern Europe, one marked by waterpower rather than by
steam engines, by small cities located where fast creeks dropped into navigable
rivers rather than by large ones at rail hubs, and by the incentive to mechanize of
high-cost labor. The story comes with characters whom we elevate to heroic
status.18 Thus we meet Eli Whitney and his revolutionary idea that parts could
be produced precisely enough by minimally skilled labor to be interchangeable,
Frederick W. Taylor and the Gilbreths with their time and motion studies that
systematized the organization of the manufacturing system itself, and Henry
Ford and the integration of multiple assembly lines for large products. As
chauvinistic Americans we ignore major figures such as Marc Brunel, who
mass-produced wooden tackle blocks for the Royal Navy during the Napoleonic
Wars.
Size Trends. We build downward rather than upward. The factory—even the
factory that assembles our largest airplanes, Boeing 747s—dwarfs the product.
For exceptions, such as large construction projects and the global
telecommunications network, one might argue that factory and product simply
need redefinition. In any case, we’re then not talking about mass production in
the usual sense.19 We build down in another sense as well, one that reverses
Copes rule about how size most often increases in the evolution of living
lineages. Early steam engines were huge, slowly moving things, slow enough at
their inception that inlet and outlet valves were manually operated; recall
Newcomen’s engine, in Figure 8.2. Increasingly precise manufacturing permitted
higher pressure differences and faster operation, so ever smaller engines
produced as much power. Early waterwheels were huge and slow; modern
turbines are small and fast. The dramatic miniaturization of electronic devices
started with the decrease in size and increase in number of elements within
electron tubes in the 1930s, 1940s, and early 1950s; the recent elaboration of
digital integrated circuits just takes to a previously unimaginable level a trend
begun much earlier.

Control. Relative to the complexity of the tasks, industry probably uses more
specific and detailed instructions than does nature. Conversely, we make less use
of feedback, at least in the number of loops involved in carrying out a particular
task. Not that feedback isn’t important in manufacturing; modern machinery is
unimaginable without it.20 This use of feedback has a curious history. In a way
the most extreme use of feedback antedates factories since piece-by-piece
handwork depends critically on the artisan’s sensitivity, visual and tactile, even
auditory and olfactory, to what’s being made. The machine lacks such
sophisticated sensory input, so it must make do with fewer and simpler, if faster,
loops. Ultimately the precision and versatility of the machine become dependent
on the level of feedback—on its awareness of what it’s accomplishing at each
stage. Testing to determine whether the end products meet some standard
becomes really an afterthought. Shifting from artisan to machine-based
manufacture provides an incentive to push machines toward the level of sensory
equipment and judgment of skilled operators.21 Much of robotics is about just
that: sensory equipment, fast and sophisticated calculations, and feedback loops.
Maintenance. Only in a loose sense does human technology do anything
analogous to organisms’ continuous turnover of material. Here and there we
replace things on schedules determined by their expected safe service lives. The
more technically demanding the device, the more hazardous any failure, the
greater the cost of redundancy in design or of unscheduled downtime, the greater
our willingness to replace bits and pieces before they’ve failed in the normal
sense. Of course what I’ve described is a commercial airplane. We ask airplanes
to give long service now that the basic technology has stabilized and they don’t
quickly become obsolete. But after ten or twenty years little except the frame
may remain of the craft that first flew from the factory. Replacement without
failure happens elsewhere as well. I’m told that in some large buildings light
bulbs are changed on schedule whether or not they’ve burned out, and the
discards are sold on the cheap to users with dimmer cost accountants.

The Criteria of Quality. While reproductive fitness is a uniquely biological

attribute, suitability for the task at hand provides a similar general criterion. For
nature, uniformity matters only when it correlates with fitness, and we encounter
in the phenomenon of equifinality a world largely beyond our technological
experience. For human technology, uniformity in detail takes on specific
importance. Uniformity in performance usually depends on consistency in
construction, and interchangeability of parts demands a similarly high level of
consistency. Nature, by contrast, cares almost nothing about interchangeability
and may even be actively hostile to it. To make us inhospitable to pathogens, our
immune system has become so potent that we must almost destroy it in order to
interchange tissues and organs among individuals. Transplanting a heart presents
fewer difficulties than persuading the body to accept it. (But the rejection
phenomenon isn’t biologically general; insects, for instance, are tolerant of
dramatic transplants of glands and other organs. And French grapes grow on
American roots, roots of a species resistant to a particular pathogen.)
Quality in human technology has an aspect roughly analogous to biological
fitness. “As good as possible” isn’t a useful way to specify either a part or an
entire machine. For the machine what’s important is being good enough to do its
intended task satisfactorily. Does making it a little better repay the cost, or does
making it a little worse generate a true saving? For an element on an assembly
line, what matters is how bad it can be, how far from some ideal it can deviate
and still fit in satisfactorily— “fitness” in a different sense. An important
criterion of design quality for complex devices is the tolerable range of variation
of their parts. The better design is the one that can be made with sloppier parts.

Safety Factors. These of course have a long and honorable history in human
technology; the biomechanic borrows the concept but must strain it almost out of
recognizable shape. Whether they represent uncertainties about loads that might
be encountered or uncertainties in our analysis of a design,22 modern
engineering without safety factors is unimaginable. Real truth, full certainty, and
fully assured safety exist only in the domains of lawyers and theologians. The
rest of us have to make do with a fallible world of purely statistical anticipations,
simplified assumptions, and inattentive inspectors. We learn well from failure,
but we prefer other schools of education.

One other aspect of human manufacturing ought to be noted. Historically,

human technology has both reduced its direct use of natural materials and
increased the degree to which natural materials are modified before use. Partly
that comes from our increasing sophistication in metallurgy and polymer
chemistry. Less obviously it recognizes that natural materials were naturally
selected for their suitability for natural structures and not for our applications. In
native form they’re usually less suitable for modern manufacturing methods,
which demand homogeneous composition and consistent properties. Natural
materials call for the sympathetic treatment of the craftsperson rather than the
rapid and uniform processing of the assembly line. Wood grows by itself, but
canoes of aluminum or fiberglass are cheaper than those of wood. Stone need
only be quarried while bricks must be formed and baked, but a wall of brick has
a lower final cost than one of stone.

The point made at the start bears reiteration. The production methods of natural
and human technologies appear different at first glance, and a more penetrating
analysis and more extensive search for underlying factors find them not closer
but even more divergent. They’re so divergent that even a common terminology
often proves elusive. Everyday terms such as “assembly,” “polymer,”
“blueprint,” “safety factor,” “design,” and “intended application” were meant for
our production systems, and we run considerable risk of self-deception when we
use them for natural systems. Similarly “selection,” “fitness,” “regeneration,”
“dynamic state,” and “derepression” describe biological phenomena, and using
them for human production risks real danger of inadvertent linguistic
misguidance.
 Chapter 12

COPYING, IN RETROSPECT

Making better widgets by copying nature—bioemulation—is no new notion.

In classical mythology Daedalus and Icarus fly from captivity in Crete on wings
cleverly copied from birds: “Then he fastened the feathers together with twine
and wax at the middle and bottom; and, thus arranged, he bent them with a
gentle curve, so they looked like real birds’ wings.” Then Ovid (43 B.C.E.–17
C.E.)—in the Metamorphoses—makes a second reference to profitable copying.
After the death of Icarus (his waxy wings were definitely not FAA-approved)
Daedalus takes on a twelve-year-old apprentice of real creativity: “This boy,
moreover, observed the backbone of a fish and, taking it as a model, cut a row of
teeth in a thin strip of iron and thus invented the saw.”1
Legendary benefits aside, copying nature holds at least three attractions for
us. Foremost is the impression of nature’s superiority conveyed by the
sophistication and diversity of her technology. A tall tree in a storm, a running
horse, a spider’s web, a flying bird, a jumping flea— their commonness doesn’t
obscure awesome mechanical performances. Nor does close examination dispel
one’s initial sense of excellent design. Each device of nature does something
beyond easy reach of our technology, and a lot more besides.
Second comes a more curious motivation, one entwined with contemporary
attitudes. For most of human history, the natural and human worlds stood
opposed. Nature was something to be tamed and utilized; we had the ordinary
attitude of organisms toward other species. Nowadays the natural world intrudes
far less but gets venerated far more. And why not? When one’s meat is bought in
a store, when locusts don’t threaten one’s corn crop, when central heating and
plumbing are the norm, the aesthetics of nature hold greater appeal. We embrace
a kind of pantheism or, to use E. O. Wilson’s less pejorative term, “biophilia.”2
That affinity for nature drives our eleventh-hour efforts at conservation. It also
drives the feeling of a natural rectitude, a moral superiority in nature’s ways of
doing things.
The third attraction reflects a combination of culture and economics. Support
for science and technology rests on a steady supply of explicit promises at least
as much as on the record of past success. Whatever the real motives of the
participants, such promises work best if couched in terms of practical payoff, not
intellectual or spiritual enlightenment. Several kinds of promises are especially
effective: industrial profit, alleviation of ill health, and military superiority. Each
of these fits well with suggestions that we might make dramatic leaps forward by
copying nature.
“Copying,” though, isn’t the ideal motto to march behind, so better words
have been coined. First came “bionics,” defined about 1960 by J. E. Steele as the
“science of systems whose function is based on living systems, or which have
the characteristics of living systems, or which resemble these.”3 The word
“systems” came naturally to those, mostly engineers, initially involved; neural
systems and physiological controls formed biological parallels to human
technology’s cybernetics and systems theory. Pattern recognition and feedback
devices got particular attention. Use of “bionics” has receded lately. “Robotics”
and “artificial intelligence” now hold center stage. A more recent designation is
“biomimetics,” whose imperatives are more explicitly mechanical—composite
materials and walking vehicles, for instance.4
But does it work? Not as well as every book, article, and symposium on
bionics and biomimetics would have us believe. Most of the lovely allusions to
past successes do no more than recognize elements of mechanical commonality.
That the jet emulated the squid, that the suetion cup copied the octopus sucker
should not be our default explanation. A common physical context is far more
likely to drive technological commonality. For that matter, we’re taking a pretty
dim view of human creativity when we assert that copying has been extensive
and successful in the past.
I claim in fact that successful copying has been rare. But defending the claim
against the more appealing affirmative claim puts me in an awkward position.
My best recourse is to search hard for good cases of copying. I’ve therefore
examined the track record with some care, playing historian and aided by the
professional reference librarian to whom I’m wedded.
We set some ground rules to circumscribe our search. The present book is
about mechanics, so we restricted our purview accordingly. Copying had to be
both credible in concept and documented in practice. The result had to be a
practical thing that has achieved fairly wide use, not a prototype or a proposal.
At the end we were left with fewer than a dozen acceptable cases of
bioemulation.5 These, though, turn out to be far more interesting than mere
items for a list or count.

BUCOLIC ROMANTICISM?
Let’s consider, for a start, three repeatedly cited cases that wilt under scrutiny.
All happen to be British and of roughly the same antiquity.

The Oak Tree and the Eddystone Light. Atop a shoal about fourteen miles
from Plymouth, England, stands the Eddystone light, which has guided ships in
the English Channel for three hundred years. The first Eddystone light fell in a
storm, and the second (of timber) burned. Between 1756 and 1759 the first great
British civil engineer, John Smeaton, built the third lighthouse (Figure 12.1)
from interlocking stones prepared at Plymouth. Instead of using the rectangular
cross section of its predecessors or the uniformly tapered cones now in favor,
Smeaton chose the graceful taper of, as he said, “a large, spreading Oak.” More
of his own words, written in 1791: “Let us now consider its peculiar figure.
Connected with its roots, which lie hid below ground, it rises from the surface
thereof with a large, swelling base, which at the height of one diameter is
generally reduced by an elegant curve, concave to the eye, to a diameter less by
at least one-third, and sometimes to half its original base. From thence its taper
diminishes more slowly, its sides by degrees come into a perpendicular, and for
some height form a cylinder.”6
Two problems becloud a claim of copying. First, by any engineering
standards, these specifications are much too vague. “Generally reduced,” “at
least one-third,” and “sometimes to half” are far from ample instructions—
analogy or inspiration, perhaps, but not a quantitative model. The other problem,
pointed out by Alan Stevenson in 1850, is that no enlightened engineer would
emulate an oak tree.7 Its main load comes from the drag of its leaves, so it’s an
end-loaded rather than area-loaded beam. Moreover, it’s made of light, tension-
resisting wood rather than heavy, compression-resisting stone. Smeaton is
simply calling to his reader’s mind something that’s sufficiently similar to the
lighthouse to substitute for an illustration, which the account lacks. Incidentally,
the lighthouse still exists, although in a different location. The rock on which it
originally stood began breaking up, so in 1882 the lighthouse was removed and
replaced with a bigger one a short distance away. Reassembled, the upper part
now stands as a monument to Smeaton on the headland above the Plymouth
waterfront.
FIGURE 12.1. The third Eddystone lighthouse, built between 1756 and 1759 by John Smeaton.

The Shipworm and the Tunneling Shield. Early in the nineteenth century Marc
Isambard Brunel bored a vehicular tunnel—still in use by the London
Underground—under the Thames. Little experience was available to guide this
first tunnel under a river. Preliminary borings suggested a river bottom much
drier, more stable, and altogether more suitable for tunneling than actually
proved the case. In fact, during the seventeen years between starting and
completion, just about everything that could go wrong did: money, labor,
Brunel’s health, and so on. Everything, that is, except Brunel’s tunneling
technology. It centered on his new tunneling shield, which needed almost no
modification as work proceeded.8 The shield (Figure 12.2) allowed thirty-six
workers to dig at once at the advancing face of the tunnel with a minimum of
unbraced excavation.
The burrowing equipment of a shipworm supposedly provided the model for
that critical shield. The shipworm isn’t strictly a worm at all, but an infamous
bivalve mollusk whose paired shells, much smaller than the rest of the animal,
serve as the hard parts for tunneling in wood. According to a biography of Marc
Brunel, written by a younger engineer who had worked with him on the tunnel:
one day, as he himself related to me, when passing through the dockyard, his attention was attracted
to an old piece of ship timber which had been perforated by that well known destroyer of timber, the
Teredo navalis. He examined the perforations, and subsequently the animal. He found it equipped
with a pair of shelly valves which enveloped its anterior integuments, and that with its foot as a
fulcrum, a rotatory motion was given by powerful muscles to the valves, which acting on the wood
like an augur, penetrated gradually but surely. . . . To imitate the action of this animal became
Brunel’s study.9
FIGURE 12.2. The front end of the shipworm Teredo with its rasping shell halves, and the drawing of Marc
Brunel’s tunneling shield in Beamish’s 1862 biography.

This sounds splendidly specific, but it simply can’t be true. Boring by Teredo
resists observation and was first described (after much trouble) by a zoologist a
century later.10 It doesn’t happen by a rotating augurlike action at all, but instead
the shells rock back and forth as rasps, scratching off bits of wood that are then
ingested. Furthermore, neither what Brunel supposedly saw nor how Teredo
really works resembles the way the tunneling shield operated. Within the shield a
workman removed a single board, excavated perhaps a foot in front of it,
replaced the board, and then did the same with the next one. The shield as a
whole was pushed forward bit by bit with jackscrews. The shipworm must
penetrate hard wood rather than soft sediment, but as a fully aquatic creature it
faces no air-water interface or pressure difference. So Brunel’s problem was
exactly the opposite: providing access to an all too soft substratum without
letting in the river. Teredo may have provided inspiration for tunneling rather
than bridging, but the rest is mythology. Brunel deserves all the credit.
FIGURE 12.3. The underside of the floating leaf of Victoria amazonica and Paxton’s patented ridge and
valley roofing system being installed during construction of the Crystal Palace—from the Illustrated
London News, October 19, 1850.

The Giant Water Lily and the Crystal Palace. In 1850 Joseph Paxton designed
and built an enormous exhibition hall, the Crystal Palace, in London. In every
aspect the structure—see Figure 12.3—was extraordinary.11 Opened less than a
year after the design was accepted, the building made unprecedented use of glass
and modular components, its appearance was dramatic and unlike any
contemporary style, and it was later successfully disassembled and reconstructed
on another site. Paxton is often referred to as a gardener. This, although strictly
true, gives a false impression of his prior accomplishments and reputation.12 He
was the preeminent innovator of his time in greenhouse construction and, among
other things, patented the ridge and valley system that permitted extensive areas
to be covered by a horizontal self-draining glass roof, a key feature of the
Crystal Palace.13
This roofing system is repeatedly cited as a successful case of copying
nature, in particular a giant water lily native to South America, Victoria
amazonica (formerly V. regia).14 Now this is no ordinary lily. Its leaves span as
much as six feet and form boats buoyant enough to support a child. An elegant
system of interconnected trusses on its undersurface stiffens each flat leaf. But
it’s still a floating structure, with the trusses offsetting small waves and the
lateral forces of currents rather than downward gravity. Paxton was the first to
raise these lilies in England, in a special structure he built for his patron the duke
of Devonshire; the crucial innovation was provision of a continuous current of
slowly recirculating water in its pond.
 The claim of copying from nature comes right from the horse’s mouth, from
a speech given by Paxton to the Royal Society of Arts while the Crystal Palace
was under construction. From the Times (London), November 14, 1850:
It was determined in 1836 to erect a new curvilinear greenhouse, 60 feet in length and 26 feet in
width. . . . This house was subsequently fitted up for the Victoria Regia, and it was here I invented a
water-wheel to give motion to the water in which the plant grew; and here this singularly beautiful
aquatic plant flowered for the first time in this country on November 9, 1849. [He then shows a
leaf.] You will observe that nature was the engineer in this case. If you will examine this, and
compare it with the drawings and models, you will perceive that nature has provided it with
longitudinal and transverse girders and supporters, on the same principle that I, borrowing from it,
have adopted in this building.

But Paxton’s words, taken literally, don’t support the usual story. Victoria uses a
trussing system in which all the trusses, in whatever direction each runs, stay in
contact with the supported surface. That’s quite different from Paxton’s ridge
and valley system. Horizontal iron girders running in two perpendicular
directions supported the ridge and valley roof of the Crystal Palace, but these
girders were arranged with one set beneath the other in the ordinary way, not in
the same plane, as are the lily’s. And of his innovative ridges and valleys the leaf
of Victoria shows no trace, although the large aerial leaves of some other plants
are pleated into just this arrangement, as shown in Figure 4.3. But something
else is wrong. “This building,” the final words quoted above, refers not to the
Crystal Palace of 1850 but to his greenhouse of 1836. Giving the greenhouse a
roof in the pattern of its intended occupant indulged aesthetics more than it
solved a problem of engineering. Finally, there’s not a word about Victoria in the
supplement to the Illustrated London News of a month earlier entitled “Mr
Paxton’s History of the Building for the Great Exhibition of 1851. ”15
In all three cases nature may have played some role, but inflating her
contribution demeans splendid engineering achievements. Perhaps the legends
persisted until canonized through an antitechnocratic and bucolic romanticism
that came with the Industrial Revolution, something obvious enough in
eighteenth- and nineteenth-century English novels, poetry, and paintings. The
poet William Blake intended no praise when he called attention to “those dark
Satanic mills.”

WHERE COPYING HAS WORKED WELL

Having aroused the reader’s skepticism, we can turn to a series of more
persuasive claims. Yes, successes exist, and impressive ones at that.
Trout, Dolphins, and Streamlined Bodies. A body that travels through air or
water experiences least resistance (drag) if it’s rounded in the front and tapers to
a rear point in the familiar, streamlined shape of tuna or whale.16 Watch a proper
marine animal—fish, seal, porpoise, or penguin—glide about underwater. No
illusion: The animal moves almost effortlessly because it meets little drag,
around ten times less than would a sphere or a person of the same size.
How can a rounded front and elongated, pointed rear so dramatically reduce
drag? This subtle matter defied explanation until the present century. But long
before, around 1809, Sir George Cayley had devised the first deliberately
streamlined, low-drag shape, using the best thing at hand: animals that moved
rapidly through fluids. He was explicit about what he did: “It has been found by
experiment that the shape of the hinder part of the spindle is of as much
importance as that of the front in diminishing resistance. . . . I fear, however, that
the whole of this subject is of so dark a nature as to be more usefully
investigated by experiment than by reasoning and in absence of any conclusive
evidence from either, the only way that presents itself is to copy nature;
accordingly I shall instance the spindles of the trout and woodcock.”17
Cayley measured the girth of a trout at a series of points along its length. He
then divided each datum by three, and he used the results as diameters to make
an elongate wooden body. As Theodore von Karman, a great twentieth-century
aerodynamicist, pointed out, Cayley’s streamlined body closely matches the
form of the best modern low-drag airfoils and hydrofoils. So he did very well
indeed. Cayley got analogous data from a dolphin (Figure 12.4), but the work on
trout has received more attention.
What happened next was less serendipitous. Cayley split his wooden model
lengthwise and used the shape of the resulting half body for a boat hull, saying,
“We should then be deriving our boat from a better architect than man, and
should probably have the real solid of least resistance.”18 But that approach
doesn’t give an auspicious hull for a shipshape boat, either for low drag or for
decent rolling stability. Recall that most of the resistance met by surface ships
comes from gravity waves at the surface, not from the kinds of drag faced by
trout or submarine. More rationally shaped hulls came along a few decades later.
FIGURE 12.4. A page from George Cayley’s notebooks, with a dolphin and the streamlined body that he
derived from measurements of its girth.

Bird Wings and Cambered Airfoils. Airplane wings have curved tops and
flatter bottoms as in Figure 12.5; they’re spoken of as “cambered.” With that
asymmetrical combination they get much more lift relative to their drag than
either inclined flat plates or inclined wings symmetrical top to bottom. High lift-
to-drag ratios mattered a lot in the early years of aviation, when planes flew at
lower speeds with less weight-efficient engines and less refined designs. As with
low-drag bodies, practical experience preceded theory; several decades elapsed
between the discovery of cambering and an adequate theory.
During the 1880s two people showed the superiority of cambered airfoils
over tilted flat plates. For both, bird wings provided the key—or least very
important—models. In England, Horatio Phillips tested a variety of shapes,
including the wing of a rook (Figure 12.6).19 Otto Lilienthal, in Germany, made
a much more extensive series of measurements. He found that plates cambered
just slightly gave the best results— an upward bowing of about a twelfth (8
percent) of the distance from the front to the back of a wing. This, he noted, is
the camber of the wings of the best birds. The dramatic effect of such slight
curvature surprised him and probably reinforced his conviction that birds were
worth close emulation. In the years before his accidental death in 1896,
Lilienthal simultaneously studied the flight of birds and constructed aircraft that
we’d now call hang gliders. The ultimate aim was powered flight using a
flapping wing craft. Lilienthal’s main legacy is an impressive book entitled Bird
Flight as the Basis for Aviation20
FIGURE 12. 5. Airfoils in cross section, showing the convex upper surface and (sometimes) concave lower
surface.

FIGURE 12.6. The airfoil cross sections tested by Horatio Phillips, along with the results he obtained.

(Neither Phillips nor Lilienthal but a third person who looked at curved
airfoils in the 1890s, Frederick Lanchester, took the first major steps toward
understanding why they work as they do.21 The explanation of how an airfoil
generates lift is based on the fact that air flows faster across the top of a wing
than across the bottom. By Bernoulli’s principle, that gives a lower pressure
above than below, so the plane is pulled upward. But the origin of that faster
flow is much more peculiar than one might guess from the polite fictions we’re
taught in schools and museums.)
 As with Cayley’s boat hull, the sequel makes the protagonists look less
prescient and heroic. The Wright brothers initially trusted Lilienthal’s numbers;
he was, after all, a proper engineer. So they used both his airfoil shapes and his
data for their first full-size glider. The glider gave too little lift, so they then built
a wind tunnel and got their own data. Lilienthal’s errors may have arisen because
he swung wings on the end of a whirling arm rather than put them in a wind
tunnel; after the first revolution on the whirling arm, the test object meets the
disturbed air of the previous circuit. I’m sympathetic since I once ran afoul of
the same problem.22
Birds and Turning Aircraft. In the two-dimensional world of automobiles and
boats, steering involves nothing more complicated than changing one’s heading
by reaiming the front wheels or turning a rear rudder. Airplanes, though, fly in
three dimensions: They can roll from side to side or pitch up and down as well as
turn left and right. Early attempts to fly paid little attention to three-dimensional
control. Some designs relied on familiar rudders; in others the flier was supposed
to shift position like a bicycle rider so the craft would make banked turns
without any specific aerodynamic adjustment.
More than any others of their pioneering generation, Wilbur and Orville
Wright took control seriously. Their initial and most important patent described a
system of control, and the basic scheme they worked out is still almost
universally used. Bird watching helped, although much later Orville tended to
minimize its contribution. But a letter written in 1900 from Wilbur to Octave
Chanute contains the following: “My observation of the flight of buzzards leads
me to believe that they regain their lateral balance when partly overturned by a
gust of wind by a torsion of the tips of their wings. If the rear edge of the right
wing is twisted upward and the left downward the bird becomes an animated
windmill and instantly begins to turn, a line from its head to its tail being the
axis. . . . In the apparatus I intend to employ I make use of the torsion
principle.”23
In short, the Wrights found that a bird adjusted the angles of the tips of its
wing for roll control. One tip was tilted so its front was slightly upward, and that
would increase its lift; the front of the other was tilted downward to decrease the
lift. That asymmetry in lift would cause banking, and in a banked position the
overall lift would be directed slightly sideways instead of straight upward. That
sideways force (with perhaps a little compensation from the rudder) would pull
the plane around in a curve. The key, then, was to twist or warp the wings, as the
Wrights did with the ingenious arrangement of cables shown in Figure 12.7. You
can demonstrate the change in shape, as they did, by twisting an elongate
rectangular box (such as a milk carton) without ends. Of course, as Orville said
about learning flight from birds, “After you once know the trick and know what
to look for you see things that you did not notice when you did not know exactly
what to look for.” The only major change in the system since the Wrights has
been replacement of warping by a pair of flaps or ailerons, one on the outer rear
end of each wing. That’s better for the more rigid wings of later airplanes.
Ailerons, though, work the same way, producing banked turns by raising one
wing and lowering the other.24

Wasps and Paper from Wood. While papermaking is an old art, only recently
have we made use of wood fiber as our normal starting material.25 Up through
the eighteenth century most paper was made from cotton and linen rag, and the
limited supply of rag became troublesome as increasing literacy and more
complex commerce raised demand. The dead were buried in wool (in England
by law) to save cotton and linen for papermaking. Around 1719 the great French
entomologist and polymath René-Antoine Réaumur suggested making paper
from wood, as were the nests of paper wasps (Polistes and related genera):

FlGURE 12.7. The system used by the Wrights for wing warping—from How We Invented the Airplane, by
Orville Wright.

The American wasps form very fine paper, like ours; they extract the fibers of common wood of the
countries where they live. They teach us that paper can be made from the fibres of plants without the
use of rags and linen, and seem to invite us to try whether we cannot make fine and good paper from
 the use of certain woods. . . . The rags from which we make our paper are not an economical
material and every papermaker knows that this substance is becoming rare. While the consumption
of paper increases every day, the production of linen remains about the same.26

Réaumur himself made no paper, but during the century that followed a number
of people attempted to produce paper from wood, and decent evidence links
Réaumur and the wasps with their efforts. The German Jacob Christian Schäffer
made paper in the 1750s from a wide variety of plant material (and from wasps’
nests themselves) with only a small fraction of rag. He clearly followed
Réaumur’s path, with conspicuous drawings of the adult wasp, its larvae, and its
nest in his treatise on papermaking. In London, in 1800, Matthias Koops (an
otherwise obscure figure) managed to make paper from both straw and wood
with no rag at all, and his paper was suitable for printing presses. He
demonstrated his achievement with a small book whose final pages were printed
on his paper; the subject was—what else?—the history of papermaking. (No
other area of technology seems to leave as extensive a paper record as
papermaking.) In the book he cites both Réaumur and Schäffer as important
predecessors with “ideas on substitutes for paper-materials.”27 Wasps are not
explicitly mentioned, but the connection is plausible.
The wasps thus set the stage by showing what was possible: that cellulose
fiber from wood, an almost unlimited source, could be separated from its binder,
lignin, and re-created as a two-dimensional mat. They were much less
forthcoming with practical guidance, and a long and arduous struggle followed
the original suggestion. Koops himself went bankrupt after building a large mill;
despite the fine product and the high cost of rag paper, wood paper could not yet
compete. But progress thereafter was steady, and within a few decades paper
mills started eating forests in quantity.
Silkworms and Extruded Textile Fibers. The same Réaumur had another
suggestion that ultimately proved practical. Just before they pupate, moth larvae
produce the protein we know as silk. Immediately after extrusion through a fine
orifice as a liquid, the protein solidifies into a continuous fiber. The silk designed
by nature for the cocoons of silkworms (the family Saturniidae) has long been
appropriated by humans for beautiful and commercially valuable textile fiber.
We either chop the cocoons and dissolve a gluing protein (wild silk) or, in the
domesticated species of silkworm, Bombyx mori, unwind the fiber. Robert
Hooke, in the seventeenth century and Réaumur, in the eighteenth, suggested
that a textile fiber might be manufactured by an analogous extrusion process.
Hooke viewed silk as “a dried thread of glue” and casually speculated “that
probably there might be a way found out, to make an artificial glutinous
composition, much resembling, if not full as good, nay better, than that
excrement, or whatever other substance it be out of which, the silkworm
wiredraws his clew. If such a composition were found, it were certainly an easy
matter to find very quick ways of drawing it out into small wires for use. I need
not mention the use of such an invention. . . .”28
Hooke’s “easy matter” proved, to say the least, wishful thinking. During the
nineteenth century this possibility—extruding or “drawing out” fiber from an
orifice—was explored by a number of people. Louis Schwabe, in England,
extruded glass fiber as early as 1842.29 Georges Audemars, in Switzerland,
drew threads of “artificial silk” of cellulose nitrate in 1855. One M. Ozanam, in
1862, suggested that silk scrap might be reconstituted by dissolving and
reextruding.30 In the 1880s Hilaire de Chardonnet, with great labor and expense,
finally developed a commercially viable process for making an extruded
artificial silk, initially the same dangerously flammable cellulose nitrate. The
silkworm’s footprints are unmistakable. Schwabe, Audemars, and Chardonnet
each were involved in the natural silk industry Schwabe made silks in his mill
for Queen Victoria, and Chardonnet worked with Louis Pasteur on diseases of
silk moths—an insufficiently appreciated branch of veterinary medicine.
Chardonnet later said he intended “imitating, as closely as possible, the work of
the silkworm.”31
By the start of the twentieth century both rayon and cellulose acetate fiber
were being made by versions of the silkworm’s (and Chardonnet’s) extrusion
process, and we now make many other fibers the same way. Was the silkworm a
useful model? Although indirect, the evidence is persuasive. To start with, the
name of the industrial extruder, spinneret, is the same as that given the
silkworm’s organ. Better yet, the word is thoroughly inappropriate for either, a
term borrowed earlier by entomologists32 from the spinning process long used
to make long threads from short fibers. In the spinneret of neither silkworm nor
fiber mill does anything spin, rotate, or go around in any sense at all. Silkworms
(and spiders) don’t actually “spin” their cocoons and webs, and in a fiber mill
thread is spun from extruded fiber in a subsequent operation. So industry
probably adopted the existing lingo of entomology.
Early mechanical extruders in fact looked very much like the analogous parts
of insects—small pipes tapering down to fine apertures, as in Figure 12.8. These
pipes were made of glass, and the demanding work of pulling out hot glass just
enough (to a little under one tenth of a millimeter) produced a delicate product
all too prone to breakage and clogging.33 Around the turn of the century
multiple-orifice precious-metal spinnerets came into use. These thimblelike
devices (as in the figure) work much better, although, operating with high
pressure across a flat surface, they’re not the kind of arrangement likely to occur
in nature. Still, as described in 1930:

FIGURE 12.8. The silk gland and extruder of a silkworm, an early tapered spinneret and a modern
multiple-orifice spinneret.

The method of producing artificial silk [rayon, etc.] closely simulates the process by which real silk
is formed. The silkworm extrudes fibroin through two apertures below its mouth, and cements the
two threads together with sericin, which is ejected from the glands at the same time. In the
manufacture of artificial silk, the silk glands are represented by large tanks and the apertures by fine
jets. In the spinning process, i.e. the transformation of the viscous mass into threads, it is customary
to make a distinction between dry spinning and wet spinning. . . . Dry spinning utilizes single
spinnerets consisting of thick-walled glass tubes contracted to a fine capillary, with an internal
diameter of 0.08 mm., and the separate filaments are assembled into a thread.34

Eardrums and the Telephone’s Transducers. By the 1870s telegraphic

communication was routine. But the advantage of an instant connection was
offset by the need to use Morse’s dots and dashes, with slow and cumbersome
coding and decoding. A telegraphic circuit had only two states, connected or not
connected. (Such binary coding is what our present computers use, but they can
switch from one state to the other millions of times more often.) Among others,
Alexander Graham Bell worked on the problem of transmitting voices instead of
mere telegraph signals. One possible scheme broke a complex sound into
separate frequencies that were transmitted along parallel lines; the receiver then
recombined them. Bell’s important insight was that such complexity was
unnecessary; a single device could convert all frequencies of sound into a single
electrical signal. Bell, no electrical guru, got the idea from an analogous
biological device.
At the time Bell was a professor of vocal physiology at Boston University.
He and his father pioneered in teaching deaf people to speak intelligibly, making
sounds that they themselves had never heard, so he was immersed in the
physiological and behavioral aspects of making and perceiving sounds. Bell
recognized that the eardrum was a single device that handled all frequencies at
once. Its movements in and out set the bones of the middle ear into motion, and
these in turn communicated with the liquid-filled inner ear, where the neural
equipment was located. In his own words, “It occurred to me that if a membrane
as thin as tissue paper could control the vibrations of bones that were, compared
to it, of immense size and weight, why should not a larger and thicker membrane
be able to vibrate a piece of iron in front of an electromagnet. . . and a simple
piece of iron attached to a membrane be placed at the other end of the
telegraphic circuit?”35

FIGURE 12.9. Bell’s diagram of the telephone transmitter (or microphone) and receiver.

Not that all then went smoothly, but the invention of this microphone broke
the back of the problem. The same device in reverse served as a receiver to get
the sound back out, as in Figure 12.9, taken from what may be the most
profitable patent in history.36 Thus the device was automatically bidirectional.
Although Bell’s transmitter was soon replaced by Thomas Edison’s more
sensitive carbon microphone, his receiver survives in earphones and (at least its
basic principle) in loudspeakers.
Modern life would be impossible without aircraft, cheap paper, artificial
fabrics, and telecommunication of speech—major matters, all of them. Three
further cases meet our criteria for emulation but involve task-specific devices of
narrower application and lesser everyday impact.

Barbed Wire. Keeping livestock pinned within hedgerows of thorny plants is an

old practice, one especially useful where wood or stone for fencing is in short
supply.37 Settlers of the North American prairies faced an ever-worsening wood
shortage as they moved westward. The plant of choice for the Midwest was a
shrubby tree native to East Texas and nearby areas—the Osage orange (Maclura
pomifera)—and a small industry during the 1860s and 1870s supplied its seeds
and seedlings for use farther north.38 This thorny bush, though, had substantial
disadvantages. Growing an effective hedge took about three years, the
grapefruit-size but inedible fruits were a nuisance, and the hedge was both
immovable and a nuisance to maintain. Michael Kelly’s patent of 186839 for an
early form of barbed wire was explicit: “My invention [imparts] to fences of
wire a character approximating to that of a thorny hedge. I prefer to designate
the fence so produced as a thorny fence.” Indeed, the wire was produced by an
enterprise called the Thorn Wire Hedge Company, perhaps advertising its utility
by drawing attention to a familiar antecedent. Figure 12.10 shows the similarity
of plant thorns such as those of the Osage orange to this early form of barbed
wire.

FIGURE 12.10. Branch and thorns of an Osage orange, Kelly’s “thorny fence” of 1868, and the barbs of a
modern version.

Kelly barbed wire was eclipsed by two competing brands of cheaper wire
after 1874; as with wings, spinnerets, and telephone transmitters, fidelity to
nature guarantees no economic magic. Patents for the new types were held by
Joseph Glidden and Jacob Haish. With the usual personification of invention,
Joseph Glidden is often listed as the inventor of barbed wire. Haish, almost
certainly not coincidentally, had a lumberyard that sold Osage orange seed. As
the historian George Basalla puts it, “barbed wire was not created by men who
happened to twist and cut wire in a peculiar fashion. It originated in a deliberate
attempt to copy an organic form that functioned effectively as a deterrent to
livestock.” Barbed wire has been an enduring success. Current consumption in
the United States runs to well over a hundred thousand tons a year.

Chain Saw Cutters. Small gasoline engines were developed early in this
century. A saw blade in the form of an endless chain with teeth on the links was
patented in 1858. Chain saws both with attached gasoline engines and with
electrically powered ones designed to work with adjacent generators appeared in
the 1920s and 1930s. Yet commercial logging still made much use of hand-
operated crosscut saws. Motorized crosscuts and circular saws were huge,
unwieldy things quite unsuitable for felling trees in a forest. Chain saws didn’t
work smoothly and required frequent resharpening; even with large, heavy
motors, they cut slowly. The arrangement of teeth (shown in Figure 12.11) so
effective in the crosscut40 just doesn’t cut the mustard for a saw that cuts a
wider kerf (groove) with teeth that move unidirectionally as small, guided links.

FIGURE 12.11. The teeth of a crosscut saw, one mandible of a timber beetle, and a section of a saw chain
showing cutter and depth feelers.

During the 1940s a machinist working as a logger took a careful look at the
tunneling equipment and technique of the larva of a large woodboring beetle,
Ergates spiculatus.41 These beetle larvae are among the few insects (termites are
others) that digest wood—that is, they break the cellulose polymer down into
metabolically usable simple sugar. But they start by biting off wood with pairs of
mandibles that (in insect fashion) move sideways rather than (as in vertebrate
jaws) up and down. Unsurprisingly, these wood borers (the subfamily Prioninae
of the family Cerambycidae) have robust mandibles. They splay out from the
head so their sharp front edges cut at the walls of the extending tunnel—the
equivalent of the bottom and sides of a saw’s kerf.
This ex-machinist-logger, Joseph Cox by name, then devised a chain with
cutters of the shape and in the equivalent position of those of the beetle larva.
They didn’t move side to side, and they pointed alternately left and right
(adjacent pairs would have jammed immediately), but they were unquestionably
beetlelike. These wood-cutting teeth worked so well that millions of portable
chain saws now use the design, so well that the company Cox founded in 1947
now dominates the business. If you buy a chain saw, it will probably have an
Oregon chain.

Velcro. This flexible and alignment-independent hook-and-loop fastener has

found a secure niche in modern life, gradually replacing shoelaces, buttons,
zippers, snaps, picture wire, curtain rings, and many other modest but ancient
adhesive devices. It does yeoman joinery, socially rehabilitating those
disadvantaged by limited manual dexterity. For the more traditional among us,
the sound and feel of separating halves still carry a sense of something self-
destructing, but we’ll eventually adjust. Velcro (from “velvet” and “crochet”)
gained wide acceptance more slowly than barbed wire or the modern chain saw
because it neither satisfied a critical need nor replaced predecessors found
wanting. But like them, it capitalizes on a model from the living world.
About 1948 a Swiss engineer and avid walker named Georges de Mestral
contemplated the burs that clung to his socks and dog after a hike in the local
hills.42 The burs, variously reported as cocklebur (Xanthium) or burdock
(Arctium), had tiny hooks at their tips that engaged anything fuzzy, as in Figure
12.12. Nylon, then less than ten years old, turned out to be an ideal material to
make the hooks—when heated and then cooled, it would retain a curve in the
face of all kinds of abuse—but the hooks had to be made from cut loops, which
required the invention of some novel machinery. In addition, the hooks were
more fastidious in choosing what to grab than were the natural burs, so Velcro
ended up as a pair of mating surfaces—hooks and loops. Improvements and
derivative products have come along, such as stainless steel Velcro (stronger)
and silent Velcro (for military use), but the basic material has changed little.
A person could hang from a patch of standard Velcro less than five inches in
diameter, but such a patch can be peeled loose with one hand. To hold strongly
but release with ease, Velcro (like adhesive tape) takes clever advantage of
what’s called a dimensional reduction. It holds against a force extending over a
broad area, while it peels in response to a force concentrated along a line; it
attaches in two dimensions but releases in one. A pair of scissors or a zipper does
the same thing, except that instead of reducing an area to a line, it reduces a one-
dimensional line to a nondimensional point. To bring matters full circle,
dimensional reduction is important in nature as well. As they get food, lots of
animals peel and tear—that is, they reduce an area to a line or a line to a point;
that’s how a cat uses its claws and a caterpillar cuts through a leaf. Conversely,
some limpets have broad skirts that make them hard to peel off rocks. Also,
grasses have lengthwise veins that help them resist crosswise tearing.

FIGURE 12.12. The hooked burs of a plant, Arctium minus (courtesy of the Duke University Herbarium),
and those of Velcro.

WHEN COPYING WORKS, WHY DOES IT?

These cases don’t add up to a statistically significant set of data. But we can
dimly discern some common threads that, if not definitive, are at least
suggestive.
First, mere imitation isn’t likely to be productive. George Cayley made a leap
of abstraction by carving not a trout but an axisymmetrical body derivative of
trout girths. Horatio Phillips’s best airfoils gave greater lift-to-drag ratios than
his rook’s wing. Otto Lilienthal’s airfoils were arcs of circles in section, not
cross sections of actual bird wings. And while wing warping may be what birds
do, Wilbur and Orville Wright did it in a thoroughly unavian way. Besides, wing
warping was quickly eclipsed by ailerons, a still less avian device. Barbed wire
is a tensile structure, a wire, closer to vines than to branches. The barbs were
originally attached rather than integral, with their radial orientation fixed in
nonbiological fashion. The way the cutters of a chain saw blade move scarcely
resembles the motion of larval mandibles; what’s common are the shape of the
cutters, their flexible connection to the saw or head, and where in the tunnel or
kerf they cut. Velcro is a paired adhesive system like zippers or snaps, while burs
are a single element more like the reversible adhesive of Post-It note papers.
Only the crucial hooks are a common element. The idea, the inspiration, or the
strategy—whatever one chooses to call it—not the details or tactics that humans
use, is what nature has provided. Practicality seems to lie somewhere between
general inspiration and exact emulation.
Second, success depends inversely on how well we understand the
underlying science. Where our science is strong, copying produces at best
narrowly targeted items, such as barbed wire, saw chains, and Velcro. But where
our science is weak, copying can generate devices of broad utility. Fluid
mechanics was a murky business before the present century, so streamlined
bodies, cambered airfoils, and ailerons could not easily be deduced from first
principles. As Cayley took pains to emphasize, copying was the best one could
do under the circumstances. Electrical signals with complex wave forms were
almost unknown, so appreciation of the eardrum allowed technology to leap
ahead of theory. Papermaking and fiber extrusion involve complex combinations
of solid mechanics, fluid mechanics, and chemistry, but such complexity doesn’t
bother natural selection, so nature provided useful hints both of what’s possible
and of how to proceed.
A third point, the most important, turns on the difference between the two
technologies. Nature’s is typically tiny, wet, nonmetallic, nonwheeled, and
flexible; human technology is mainly the opposite: large, dry, metallic, wheeled,
and stiff. Where one technology operates in what is normally the domain of the
other, emulation holds promise. As natural structures go, thorns and beetle
mandibles are especially stiff, so they’re closer to what we do with our materials.
Among devices made by humans, Velcro is relatively flexible, so with it we
brush against a world whose possibilities have been more thoroughly explored
by nature.

HUMAN FLIGHT AS A CASE HISTORY

To look at successes is to look with perfect hindsight. To counteract such post
hoc bias, we might focus on a specific human endeavor. The history of flight,
which has given us several examples of successful emulation, contains still more
of failure. Even ignoring the completely bizarre, such as men jumping off barns
with improvised wings,43 a nicely multidimensional story emerges. One side of
the story was put well by Hiram Maxim, in 1909. He had just spent a lot of
money—money made from his machine guns—on a spectacularly large and
unsuccessful flying machine: “Man is essentially a land animal, and it is quite
possible that if Nature had not placed before him numerous examples of birds
and insects that are able to fly, he would never have thought of attempting it
himself.”44 The other side of the story is that in most respects the flight of
organisms proved to be a notably bad source of guidance.45 As Maxim also put
it, “The successful locomotive was not based upon an imitation of an
elephant.”46 Quite beyond their nonflapping wings, the Wrights’ aircraft were
thoroughly unbirdlike: propeller-pushed biplanes with horizontal stabilizers in
front and large vertical control surfaces behind. To sharpen the argument that
nature speaks with forked tongue, consider several specific aspects of aircraft
design.
Using a single set of structures to produce both upward and forward force is
a reasonable scheme only for a flier of low wing loading, which is to say a small
one, an argument made in Chapter 10 and not unknown at the end of the
nineteenth century. Furthermore, nature’s reciprocating motion—up and down,
up and down—is a fairly cumbersome way for our motors to apply a large
amount of power. So a single pair of flapping wings is an inauspicious route to
human flight. No successful human- powered aircraft, where power is most
severely limited, works that way. The glider that killed Lilienthal was about to
get an engine that would flap the wings with a pair of hydraulic or pneumatic
cylinders. It’s extremely doubtful that Lilienthal would have flown the first fully
powered aircraft but for the accident that killed him.
Most of the early attempts at flying machines had very bird- or batlike
wings; if anything, they were even shorter and broader than those typical of
flying animals. Figure 12.13 gives an example. But other things being equal, the
best ratio of lift to drag is obtained with the longest wing, although that might
not have been completely obvious a century ago. For a given wing area, longer
and skinnier are better. Bird and bat wings are compromises; flapping is
probably easier for shorter wings, and shorter wings give greater
maneuverability and better ability to deal with erratic air currents. A little less
attention to animals and a little more experimentation with isolated wings would
have been helpful.
Probably the animals misled us worst by giving the impression that control
of flight was no big problem. In locomotory machines in general, and with
especial severity in aircraft, maneuverability and stability tend to be antithetical
characteristics. Each of the three extant lineages of flying animals is of
considerable antiquity, so, as pointed out by the evolutionary biologist John
Maynard Smith,47 they’ve had time to become unstable, time to evolve neural
systems that could manage the instability that goes with good maneuverability.
After all, maneuverability will almost always be advantageous for animals since
they’re either aerial predators or targets of aerial predators, or they fly among
obstacles, or they soar on atmospheric irregularities.
Lilienthal was especially single-minded about imitating birds. His book on
bird flight spells out his operational philosophy: “. . . natural birdflight utilizes
the properties of air in such perfect manner, and contains such valuable
mechanical features, that any departure from these advantages is equivalent to
giving up every practical method of flight.” Thus he was fatally trapped in an
excessively birdlike and unstable glider. The instability of Lilienthal’s craft
worried Octave Chanute, an older man who had experimented with gliders
earlier. Birds, for instance, have horizontal tail surfaces but no vertical rudders,
since they get adequate control of sideways turning with their wings alone, so
vertical tail surfaces were small or absent in many hopeful aircraft built between
about 1880 and 1905.
FIGURE 12.13. Clément Ader’s batlike Éole of 1890, which made a partially powered hop; the wing (top
view) of lgo Etrich’s plane of 1906, which copied the gliding fruit of the Javanese cucumber, and the gliding
fruit itself.

Some aircraft, just a little later, went overboard in the other direction, with
such great stability that they were almost unmaneuverable. At least one copied
another natural flier. Animals can do wonderful things with delicate sense organs
and fast-acting feedback loops. Plants obviously can’t, so gliding plant parts
must be highly stable. The samara (seed in fruit) of a maple, mentioned as an
alternative to a parachute, glides helically (it “spirals”) downward. The glide
slows its earthward movement so it can be carried farther by the wind that blew
it off the tree. A few fruits glide along a straight path, gaining distance from the
parent tree without the need for wind. Among these is the fruit of the Javanese
cucumber, a flying wing that glides earthward in the still air beneath the forest
canopy in Southeast Asia (Figure 12.13). After the death of Lilienthal, Ignaz and
Igo Etrich bought his remaining gliders, whose instability converted them
(scared them?) into advocates of extremely stable craft. Igo obtained flying fruits
from the Hamburg Botanical Museum and copied them in a series of craft—
unmanned, manned, and then powered. But as gliders they were almost
unmanageably stable and lacking in maneuverability. Worse yet, they combined
the worst of both worlds since too much stability was lost when motors were
installed.48
Eventually it became clear that an aircraft, unlike a bird, ought to be so
inherently stable that it didn’t require continuous activity by the pilot. At the
same time it ought not be too stable to be controllable, thus unlike a gliding fruit.
Modern aircraft manage the compromise quite well. As a former colleague,
Molly Bernheim, describes learning to fly, “So a pilot, before he flies safely,
must learn a very difficult lesson, one which is contrary to all his natural
instincts. A headlong descent toward the ground? Pulling back won’t help you!
You must let go of the stick, so that it moves forward. A wing that won’t come
up? Let go! The airplane can look after itself better, now, than you can do! Turn
it loose! Then, and only then, you may guide it gently where you want it to
go.”49
One domain, though, differs. Small, high-performance military aircraft are
made deliberately unstable to achieve great maneuverability. But they feel stable
to their pilots because they’ve adopted a crucial feature of animal flight. Modern
control technology, which means sensors and actuators in rapidly acting
feedback loops, gives them the best of both worlds.50 The pilot makes the
strategic decisions while the tactical details are handled by servomechanisms,
just as a mammal or bird divides control between the cerebral executive and the
middle management of the rest of the central nervous system.
Sticking close to nature’s methods has proved no boon in at least one other
form of locomotion. We’re most often told that Robert Fulton invented the
steamboat in 1807. In fact, steamboats existed, and Fulton knew about them. He
had a better engine (from Boulton and Watt, in England) and better financial
backing, but the main advantage of his boat lay in that very unbiological device
the rotating paddle wheel. James Rumsey’s boat of 1787 used a steam-driven
piston to make a squidlike pulse jet, taking water in beneath the bow and
expelling it forcefully at the stern. A piston engine naturally reciprocates, so a
few valves suffice to make it power a pulsating squirter. But Rumsey’s boat
suffered from the squid’s problem: the low-propulsion efficiency that comes with
a highspeed, low-mass output. John Fitch’s boat of 1790 used a set of
reciprocating rear paddles that alternately dipped into the water and pushed
rearward like a person doing a crawl stroke with duck feet as hands. Paddle
wheels, first in the rear and then at the sides of boats, were simpler, more
effective, and more efficient.51

This has been a skeptical and polemical chapter. But its message of disbelief and
irritation at excessive romanticism and self-deception is hard to paint in brighter
colors. Historians are fated to look backward, whereas scientists usually look
forward. In all the debunking, however, a positive message should not be lost:
The record of copying may be sparse, but good cases do exist. Blind emulation
may have nothing to recommend it, but we’ve sometimes been much smarter
than that. Indeed, a good argument can be made that as our technology takes on
more of the characteristic’s of natures—more flexible materials and structures,
increased miniaturization, greater use of nonmetallic materials, and so forth—
nature will be an increasingly useful teacher.
 Chapter 13

COPYING, PRESENT AND PROSPECTIVE

History isn’t destiny. If it were, then we’d have known in 1875 that cities were
doomed, that by 1925 the ever-increasing volume of horse manure would
produce total urban inundation. We’ve looked at the past; its messages about
what’s to come in the business of copying nature are both mixed and unreliable.
The best we can claim is that an informed guess is better than none.
Bioemulation makes the news—or at least a lot of corporate advertising copy—
but balance and perspective aren’t the hallmarks of day-to-day journalism.

A MIXED BAG OF MAYBES

Problems: Copying may work but hold insufficient advantage over an
alternative. Or copying may be assumed—mistakenly—to have worked. Or else
the “nature-copied” device works but turns out not to have been copied. Here are
examples of each, intentional fuel for the fires of skepticism. The concentration
on fluid mechanical cases should be blamed on the author’s experience and not
taken as defining the current mainstream of work in bioemulation.

Guts and Peristaltic Pumps. Our intestines act as their own pumps, pushing
along slurries of digesting food. Waves of muscle contraction move lengthwise
down our twenty feet of small intestine and eight feet of large intestine in the
process known as peristalsis—much the way we squeeze toothpaste from the
bottom of a nearly empty tube. The digestive systems of animals most often use
this kind of pump; even the circulatory systems of a lot of worms pump
peristaltically. Peristaltic pumps—the name borrowed from physiology—find
occasional small-scale use in human technology, especially in biomedical
applications; an example is shown in Figure 13.1. The various versions have in
common an arrangement for flattening successive elements of a tube with elastic
walls, pushing the contents along. Peristaltic pumps have two selling points. The
viscosity of the material being pumped can vary from that of water to that of a
sloppy slurry without affecting the pumping rate, as you know from everyday
personal experience. And the material doesn’t have to leave the elastic tube
while it passes through the pump, which avoids contamination from contact with
the rest of the pump and simplifies sterilization. Otherwise, peristaltic pumps are
a nuisance. Elastic tubing must be replaced often, and as pumps go, they’re
inefficient; deforming the elastic tube takes a lot of energy, little of which gets
recovered.
Was the peristaltic pump copied from nature’s peristalsis? The name suggests
emulation, but the evidence argues otherwise. In 1894 a British inventor, John G.
A. Kitchen, patented a pump much like one of our contemporary ones but
without using the name. He intended the pump as a hand-operated, portable
device for reinflating bicycle tires, a low- speed, occasional use in which rubber
tubes wouldn’t quickly fail from the repeated squashing.1 The present name
most likely comes (perhaps as advertising) from its use by a biomedical
community.

FIGURE 13.1. The major components of a modern peristaltic pump; it’s belt-driven through a pulley
immediately behind the plane of the drawing. John Kitchen’s version differed mainly in having one rather
than two peripheral rollers and a hand crank on the central shaft.

Dolphins and Low-Drag Submarines. A lovely story that tells how a dolphin’s
almost dragless swimming was discovered and then mimicked. In the mid-1930s
Sir James Gray, one of the founders of modern biomechanics, got a curious
result when he calculated how hard a dolphin had to work at its top swimming
speed. Flow around a body can be either laminar or turbulent, depending on such
factors as the body’s size, speed, smoothness, and shape. Gray calculated that
only if flow around a dolphin was laminar could a dolphin’s muscles produce the
power needed for the speeds it attains; turbulent flow would cause too much
drag. But a body as big and fast as a dolphin’s normally experiences turbulent
flow. The result became known as Gray’s paradox.
About twenty years later Max Kramer claimed that dolphins maintain
laminarity with a soft, compliant skin that damps incipient turbulence and that
his Lamiflo coating system for submarines imparted the same impressive effect.
Unfortunately the story doesn’t hold water. First, we now know that Gray
underestimated how much aerobic power can come from mammalian muscle and
overestimated how fast a dolphin could swim continuously. A dolphin should
manage well enough even if flow is turbulent. Second, no one has demonstrated
that dolphins work as Kramer suggested, although they have soft skin and do
appear to disturb the water remarkably little as they swim. Finally, the Lamiflo
coating never gave the promised reduction in drag.2

Fish Slime and Polymers for Drag Reduction. If you add to water substances
made of big molecules, the water, now more viscous, flows less readily. Under
some circumstances, though, the opposite happens. Adding small amounts of a
long, linear, soluble polymer to turbulent flows can reduce friction and speed up
flow through a pipe or across a body. Some fish seem to use such polymers on
some occasions, shedding fish slime into the water as they swim. Fish slimes and
other biological polymers can lower resistance to flow on fish, on flow through
pipes, or on solid test bodies. But practical use of the trick requires continuous
shedding, since the flow carries the polymer away from the surface. Even fish
probably secrete polymers only during brief predator-prey chases. Furthermore,
while a lot of effort has focused on fish, since they apparently use the stuff, the
original discovery and investigation of friction reduction by adding polymers
didn’t involve biological systems. It turned up in investigations of so-called non-
Newtonian liquids, complex liquid systems in which viscosity isn’t constant
from time to time and place to place.3

Shark Scales and Drag Reduction. Not all fish are slimy. Sharkskin, for
instance, feels like medium-grit sandpaper. Examined closely, the scales of
sharks turn out to have ridges running atop them, and the ridges look as if they
were aligned with the local direction of flow. These ridges are small, less than
one-tenth of a millimeter apart and still less in height, but they appear to have
evolved independently in several lineages of especially speedy sharks.
Experiments with artificial versions have worked well enough to spawn a
commercial coating (riblets) for use on high-performance racing yachts. But this
coating reduces skin friction at most by 10 percent and usually much less, and
skin friction, in any case, isn’t as important as wave drag for surface ships.
Moreover, whether sharks use their own ridging in the same way remains a bit
uncertain.4

One final note: We learn things by word of mouth more than we ordinarily
admit to students. Recently word of mouth has gained a whole new dimension as
we’ve plugged into another informal avenue of inquiry, the Internet. I explained
on several Internet news groups5 that I was looking for successful emulations of
nature. I gave the cases I already knew about and the guidelines I was following.
A number of the stories here originated in the replies, and I got a fine
introduction to work in progress. I also became more confident that I’d left no
major stone unturned.

THE PROMISE OF CURRENT WORK

Never have more people been chasing biomimetics. Moreover, we’re now seeing
careful and systematic investigation of living systems in concert with efforts at
emulation. An increasing number of people are now familiar with both sides of
the street: biomechanics and engineering. The past history of copying nature
may be unportentous, but its scattershot character is no longer representative. In
short, it’s a cautionary tale, not a trajectory. Here, then, are a few of the
endeavors to which biomimetic thinking has recently taken on relevance.6

Nanotechnology. Nature builds upward from molecules and cells, while we

usually don’t. Some things, though, are more easily made that way. For instance,
you hook together in specific sequences the different amino acids that form your
proteins as directed by your genetic code, not by brewing up a batch of identical
little molecules and making them stick to each other in random order. You also
make composites like bone by synthesizing and positioning the various
components just where they’re supposed to be, not by mixing the components
and letting the product solidify. One way for human technology to build upward
is to use microorganisms as host factories with DNA inserts of our choosing. We
already do this when we make small amounts of human insulin and other highly
valuable biological molecules. Nothing stands in the way of our making either
normal or modified structural proteins—nothing but horrible economics. Even
with the most devoted microorganisms, the cost of making such products by the
ton instead of by the gram would require that they have enormous advantages
over ordinary polymers. Where might they be worthwhile? Perhaps in medical
applications, where we’re willing to spend extravagantly. For instance, we might
want to make something that a human body, with its immune system, accepted as
a normal part of itself. Bacterially synthesized human insulin has that advantage.
Or imagine solid, prosthetic materials that not only would be accepted but that
would join in the body’s growth and replacement processes. Another way to
build upward is to design large molecules that self-assemble into microscopic
structures like microtubules and other cellular components.7
Muscle Analogs. Engines that work by contraction are rare in human
technology. So are room-temperature engines made of soft material that can be
made arbitrarily small. But nothing prohibits us from making musclelike devices
that convert chemical to mechanical energy; such devices just haven’t yet
reached the stage of practical applications. Anything really similar to muscle,
though, would need a well-developed nanotechnology for production. Just how
similar to muscle we’d want our engine to be is uncertain. While muscle is fine
stuff, it’s an engine that can operate only when cool and wet. Except, again, for
direct replacement for body parts, we don’t have to be limited to low-
temperature operation and shouldn’t have to do everything underwater.8

Composite Materials. Bone, enamel, coral, wood—nature is so good at making

hierarchical and composite materials that we surely should be able to learn from
her products. In fact, this area is especially active at present. Seashells seem to
hold lessons for making cermets—ceramic and metal composites. Beetle body
wall—cuticle—seems to have a superior system of bonding fibers and matrix.
The hard parts of sea urchins and other echinoderms are apparently single calcite
crystals, but they incorporate a bit of protein that makes them interestingly
fracture-resistant. And so on, from materials through large-scale structures
derived from bamboo stems and molluscan teeth. Even if close emulation awaits
effective nanotechnology, products that combine presynthesized components in
nature’s ways are clearly feasible in the short term. One wood-derived product,
for instance, has been patented and can be produced with the equipment now
used to make corrugated paperboard.9

Smart Materials. Muscle, bone, wood, skin, and some other natural materials
adjust their composition or quantity in response to changes in load. This capacity
for load-dependent reconstruction depends on sensing loads—on the sensory
physiology of bones, tree trunks, blood vessels, and other structures that we
don’t usually think of as capable of sensation. Sensory systems have classically
been the province of neurobiology, but in these structural materials well-
understood neural feedback loops play little or no role. Loading some living
materials such as bone and wood sets up electrical changes within them. The
phenomenon, called piezoelectricity, has long been known in crystals and
synthetic ceramics; press on such a material, and electrical charges appear on its
surface. We use piezoelectricity to convert mechanical changes to electrical
signals in devices such as phonograph pickups and microphones. Thus
something unusual for the biologists is standard fare for the engineers, and
piezoelectric sensation should be more readily emulated than neural sensation.
Still, sensation is only one element of a feedback system that must also be
capable of processing and acting on information. The self-awareness of hunger is
of no value without mechanisms to obtain food. The other elements—how
electrical charges trigger growth, for instance—remain mysterious. So making a
responsive material is no trivial matter. Still, the ability to put material where
and only where it’s needed and to compensate for wear and for changes in
loading would be no small advantage.10
Robotic Manipulators. Robots, industrial robots in particular, are of special
interest these days. No one outside the film industry gives serious thought to
visually anthropomorphic robots, but humans do have enviable dexterity and
tactile ability. One tricky task that a good robot should master is picking up and
positioning delicate, nonrigid objects. Humans do this well, using jointed
appendages and a combination of visual, tactile, and proprioceptive (telling
where your various parts are at any instant) senses. Various multijointed handlike
manipulators have been built, with nature’s versions (as with aircraft,
composites, and smart materials) a steady reminder of what’s possible. One
interesting approach uses flexible, pneumatically inflated structures rather than
jointed appendages, as in Figure 13.2, copying the muscular hydrostats of
tentacles, tongues, and elephant trunks in order to handle fragile objects. We
mammals mainly use jointed appendages, but they have no intrinsic superiority
over muscular hydrostats for biological uses, much less as models for robotic
technology. Indeed, close emulation of a tentacle may be a better bet than
copying a hand since the hand requires many more discrete and independently
controlled elements. The tentacle, though, pulls us into the unfamiliar world of
structures that lack hard parts.11

FIGURE 13.2. William Kier’s diagram of how a squid tentacle twists and (from U.S. Patent 4,792,173 of
1988) a diagram of one of James F. Wilson’s pneumatic manipulators.

Walking Vehicles. Wheels are better than legs in hard, flat contexts. The messier
the terrain, the better legs become. On a soft substratum—say, sand—that a foot
can still push against effectively, a wheel must climb out of its own rut as it
moves, doing something analogous to a ship sailing above hull speed. The
military worry about getting people and things around without first making
roads, so they’re attracted to legged vehicles and have paid for building some
rather fancy ones. Both the motor arrangements and the sensors and feedback
systems of these are enormously complex. One six-thousand-pound walker goes
only five miles per hour and carries only five hundred pounds; on the other hand,
it can climb a 60 percent gradient. A strong consensus favors six-legged
machines, so the walkers (such as the one in Figure 13.3) look a bit like giant
insects. Balance comes harder to four-legged machines; if one leg is lifted, the
thing may tip. But a hexapod can lift every other leg while retaining a stable
tripedal stance, or it can lift almost any two without loss of stability, as you can
demonstrate with clay (or a brownie) and toothpicks. Current work on walking
vehicles pays serious attention to insect biomechanics.12 Nonetheless, the logic
of hexapedalism doesn’t mean that good walkers will look like insects; such
things as the difference between muscles and motors and the difference in scale
will inevitably drive divergence.

Swimming by Bending. The fish most familiar to us swim with obvious ease
and agility in a manner unlike either surface ships or submarines.

FIGURE 13.3. A hexapedal insect and a hexapedal vehicle, not at the same scale!
FIGURE 13.4. A pair of twiddlefish invented by Stephen Wainwright and Charles Pell (courtesy of Twidco,
P.O. Box 542, Durham, NC27702).

These fish—trout, bass, and the like—accelerate impressively when they start
and turn sharply at any speed. In their mode of swimming, waves of bending
pass from head to tail. The waves increase in amplitude as they move aft, so the
whole body is an integrated propulsive unit powered by a large mass of edible
muscle. For our machines, that’s an even more awkward arrangement than
flapping wings. Even so, the waving motion has been simulated using a
complexly interconnected set of solid segments called the robotuna.13 It can also
be simulated—and much more cheaply—by a simple fish-shaped casting of soft
plastic. A twiddlefish, as in Figure 13.4 is persuaded to oscillate side to side by
twisting, first one way and then the other, by a vertical shaft that emerges
upward just behind its head.14 While the twiddlefish is now sold as a toy, it’s a
serious proposition. A version a foot or two long can propel a small person-
powered boat; a person sits on the boat and twists a shaft that goes down through
the hull to the twiddlefish. Larger ones, various geometries, combinations of
propulsive units, and different degrees of stiffness are currently being
constructed and tested.
Will such devices supplant marine propellers? Doubtful: Rotating propellers
are probably more efficient than any possible twiddlefish. Really fast marine
swimmers—tunas and whales—swim by swinging wide tails back and forth
behind stiff bodies, not by passing back waves of bending. That strongly
suggests (as does other evidence) that a trout, for instance, trades off propulsive
efficiency for maneuverability and acceleration.15 But just as legged vehicles
have specific uses that their intrinsic complexity and inefficiency don’t rule out,
so perhaps do soft, oscillating propulsive units. Imagine, for instance, a trolling
motor that moves a fishing boat through a tangle of submerged vegetation. An
oscillating propulsor ought to get fouled a lot less easily than a rotating propeller.

QUO VADIMUS?
If you want to make very small things or to fabricate materials that are
heterogeneous at a microscopic level, nature may hold both lessons and
assistance. The smaller the scale, the better the prospects for emulation. We’re
relatively better on large scales, where our metals and wheels come into their
own. This suggests that materials science is a fertile place for biomimesis, while
structures and mechanical systems are less so. Since the early cases of copying
were all large, we glimpse the possibility of a new path.
If you develop new materials that are more like nature’s, her designs should
show how to use those materials to best advantage. Using new materials is a
nontrivial business. The earliest iron bridges looked like those made of wood or
stone, while the earliest steel bridges looked like the bridges that had been made
of iron. Many of us remember that early plastic buckets, yard carts, and such
looked more like their metallic antecedents than do current designs. As
composites, including ones particularly like those of nature, become more
competitive in price with metals and homogeneous plastics, natural designs may
take on new attractiveness for us.
If you intend to make things of pliant rather than stiff materials, recognize
that nature has been there first. We need just point again to tentaclelike robotic
handlers and to twiddlefish, both cases in which flexible devices similar to their
natural analogs work well. Designing with pliancy in mind puts strength, not
stiffness, uppermost, and that holds promise of material economy: doing more
with less.
If you want to make prostheses, some advantages may come from using
materials and structures as close as possible in properties and operation to
nature’s originals. Here one has to fit an element of one technology into the
other. And to be crassly commercial, a prosthesis can command a spectacular
price relative to the amount of material in it.
Even though put as admonitions, each of these items points to a future
considerably brighter than implied by a simple historical account. We’re moving
toward ever smaller components in our various contrivances, in effect getting
closer to nature’s miniature world; remember that an average animal is only a
millimeter or so in length. We’re developing a great array of flexible materials to
supplement or supplant stiff metals and brittle ceramics. We’re exploring the use
of composite materials, composites far better than our old standby fiberglass, and
thus entering a world in which nature (perhaps for lack of metals) is an
experienced and versatile player. With improvements in small actuators and
complex controls, devices made of muscles, tendons, bones, and nerves are
increasingly attractive as models.
But once again the caveat. Nature may show what’s possible, but she’s a
poor guide to what’s worth doing. Flashy corporate advertising repeatedly pops
up heralding the imminent industrial synthesis of spider silk. Spider silks (there
are many variants) do have unusual properties. They’re strong, extensible, and
tough (although they have low stiffness and resilience). Even so, they’re not, as
the ads imply, leagues better than existing polymers such as Kevlar, which
happens to be stronger than any known silk. Making a spider silk would be an
impressive biotechnological accomplishment. The task extends well beyond
getting the amino acids linked together in the right order; it has to be extruded
with the right organization of crystalline and noncrystalline regions.
How badly do we want to make a spider silk? Getting manufacturing costs
down far enough is hard to imagine, since almost any application would need a
large amount of the stuff. For that matter, thinking of applications isn’t all that
easy. The properties of spider silk and indeed of structural proteins in general are
extremely dependent on their water content— level of hydration, strictly—and
on temperature. To complicate any uses even farther, changes that happen when
hydration and temperature are altered often aren’t reversed when hydration and
temperature come back to normal. What applications do we have that are
sufficiently permissive to put up with such trouble? Yet another trouble spot is
the low resilience of silk—perhaps its most unusual and thus potentially
desirable feature. The combination of high toughness and low resilience means
that lots of energy can be absorbed in a stretch and that the energy doesn’t come
out again as elastic rebound. In other words, spider silk is stretchy, but it doesn’t
behave at all like a rubber band; it absorbs and then keeps the energy. But
energy, according to the first law of thermodynamics, can’t be destroyed, and the
energy appears as heat.16 Now that may not pose a problem for a skinny thread.
None of its inside is at all far from its surface, so heat loss to the surrounding air
or water keeps it from getting hot. But imagine using a rope of spider silk to stop
a falling body or a moving aircraft. If the rope is thick enough to do such a big
job, the silk will be immediately ruined by the resulting increase in its
temperature.
So having easy access to large ropes of spider silk might hold little
advantage. They might be like some other biological items that are widely
available but not especially useful. The big and prosperous stems (properly
“culms”) of bamboo in my yard are a good example; they produce a substantial
annual harvest for which I’ve never found an application. What might be more
useful would be to understand how spider silk comes by its curious combination
of properties. That could help us devise materials that have its unusual and
desirable properties without its disabilities, materials that can be made cheaply
by our kind of manufacturing.

Poor prospects get weeded out quickly enough either because the people doing
the work get discouraged or because the sources of funding get discouraged.
Even skeptic that I am, I view the present avenues being explored as highly
promising. A kind of natural selection will operate among technological projects,
one driven by experience, wisdom, and financial support.
History may teach another lesson: that ignoring nature can be as much a peril
as mindless copying. Recasting history with hindsight may not be the fairest of
games, but consider the following. Screw propulsion for ships was invented in
the first half of the nineteenth century, mainly by Francis Pettit Smith and John
Ericsson. Their model was a pump long used in human technology, the
Archimedean screw (Figure 13.5). Contained in a close-fitting pipe, the device
will do service as a simple, if not especially efficient, pump. For propulsion of a
vessel such a screw holds some advantage over a paddle wheel: It’s more
compact and works entirely underwater. Smith’s first commercial vessel
(launched in 1838) even bore the name Archimedes. But as a thrust producer the
Archimedean screw isn’t much better than it is as a pump. An accident is what
established screw propulsion as the way to go. The outer half of a wooden screw
broke off, and the shortened screw worked better. A set of flat blades, none
extending for more than a fraction of a turn around the shaft turned out to work
best of all.17 (Recognition of the advantage of curving—cambering—the blades
awaited the twentieth century and the development of aircraft propellers.)18
Smith and Ericsson ignored a better model for an underwater propulsor. A
ship’s propeller employs the same fluid mechanical scheme as do the flukes of a
whale or the wide tails of fish such as tuna. The only real difference is what
we’ve repeatedly noted: rotational instead of reciprocating motion. If these
developers of screw propulsion had understood that they were making rotating
analogs of whale flukes or tuna tail fins, they might have done a lot better a lot
sooner.

FIGURE 13.5. An Archimedean screw used as a pump and as the propeller of a ship.

In viewing on one hand biomechanics and on the other bionics, biomimetics,

and robotics, we’re looking at the difference between pure science and applied
science. The old Dewey decimal system for book classification makes precisely
this distinction. Books on science, the 500s, are found in one part of the library;
books on applied science, the 600s, are found in another. And the people who
care about each are all too often in different places as well. Bringing them
together has to be mutually beneficial.
 Chapter 14

CONTRASTS, CONVERGENCES, AND

CONSEQUENCES

The more closely we look at the technologies of natural selection and of human
contrivance, the less similar they appear. We might well have guessed otherwise
in the light of their common situation. Life has proliferated on our planet for
several billion years, and we’ve been making things for a million or so—ample
time for underlying imperatives to make themselves felt. Yet those basic
differences persist:

• Nature uses fewer flat and more curved surfaces than we do.
• Ours is a far more rectilinear world while nature shows little bias in favor of
right angles.
• Corners in our technology are abrupt; nature’s are more often rounded.
• Numerous mechanically separate but individually homogeneous components
make up our devices; nature uses fewer components whose properties vary
internally.
• Nature’s designs take advantage of diffusion, surface tension, and laminar
flow; gravity, thermal conductivity, and turbulence matter more for ours.
• We most often design to a criterion of adequate stiffness, while nature seems
more commonly concerned with ample strength.
• Partly as a consequence, our artifacts tend to be more brittle while nature’s
are tougher.
• As another consequence, our things move on sliding contacts between stiff
objects whereas nature’s objects bend, twist, or stretch at predetermined
 places.
• As an additional result, we minimize drag with streamlined bodies of fixed
shape, but nature often does so with nonrigid bodies that reconfigure in
flows.
• Human technology makes enormous use of metals, while metallic materials
(as opposed to materials containing metal atoms) are totally absent in nature.
• As a result, we use the ductility of metals to prevent crack propagation;
nature does as well, but with foams and composites instead.
• We more commonly load materials in compression while nature more often
loads in tension.
• Concomitantly, we make greater use of shear preventatives such as nails and
mortar to keep stacked objects aligned.
• Structures with tensile sheaths outside and pressurized fluid inside are both
more common and more diverse in natural designs than in ours.
• For such hydrostatic and aerostatic systems, nature’s predominant fluid is
water while our structures mostly contain air or some other gas.
• We make profuse and diverse use of rolling devices based on the wheel and
axle; but things rarely roll in nature, and only one true wheel and axle is
known.
• Our prime movers—engines—are based on rotation or expansion; most of
nature’s are based on sliding or contracting.
• Many of our engines extract mechanical energy from temperature
differences, whereas all natural engines are isothermal.
• Levers in human technology most often amplify force at the expense of
distance, while nature’s commonest levers amplify distance at the expense of
force.
• Our devices store mechanical work as electrical, kinetic, gravitational, or
elastic energy; nature mainly uses the last two and most often the last one.
• Our fluid transport devices often interchange pressure drop and volume flow,
but equivalent transformers are rare in nature.
• Surface ships have long played an important role in human technology, but
nature overwhelmingly prefers submarines.
• Our factories dwarf the items they produce; nature’s factories make products
far larger than themselves.
• We judge our devices best when they need only minimal maintenance, but
nature’s devices get continuously rebuilt.
• Our technology is as dry as nature’s is wet.
Listing differences hints at interrelationships among them. If gravity dominates,
then stiff materials to resist it find special utility. Stacking becomes a reasonable
way to build things, with shear preventatives like nails to prevent sliding.
Permitting such sliding here and there then defines joints. And so on. Using
metals makes their special properties available (for instance, high thermal and
electrical conductivities) and allows using otherwise impractical devices (such as
wires) and construction methods (like pressing and forging). Composites drop
from crucial to simply useful. And so forth.
Each domain, nature’s and ours, thus develops its distinctive coherence,
consistency, and rationality, each a well-integrated entity in its particular context.
Might we mix and match among the features of the two technologies, generating
a vast number of further ones?1 All but a few would surely lack that degree of
coherence, consistency, and rationality; the combinations of features that mark
each of our two transcend historical accident. What determines the kinds of
devices that a technology finds effective? For one thing, its physical situation:
the size of a technology’s artifacts, whether its basic medium is air or water,
whether it works at a surface or suspended in a gas or liquid, and so forth. For
another, how it goes about doing things: production methods, degrees of
resistance to revolutionary change, relative ease of technological diffusion, and,
once again, so forth.
Even social interactions matter to how a technology goes about its business.
Nature faces severe limits when organizing and coordinating the efforts of
individuals; one might say “institutional limits” to sharpen the comparison with
human efforts. Casual verbiage to the contrary, organisms don’t naturally work
“for the good of the species.” Except (perhaps) for humans, neither the
individual organism nor its genes know anything about its species or feels any
obligation to it. Only to a limited extent do organisms work for the good of some
community, ecosystem, or biosphere.2 In nature, coordination of effort happens
with any regularity only under two specific conditions. First, an organism may
do things that increase a close relative’s chance of reproduction while decreasing
its own. Relatives have many genes in common, so a self-sacrificing action can
still enhance an individual’s genetic contribution to the next generation.3
Second, if by helping another, an organism helps itself, then public-spirited
action is acceptably selfish in evolutionary terms.4 These stringent conditions
must preclude all kinds of cooperative activities.
Human technology, for instance (since we’re now including hypothetical
ones), operates in a much different social context. Our large enterprises depend
on cooperation among people with no close kinship, although a paycheck does
represent respectably selfish reciprocity. Governmental and religious institutions
—the military being the most ancient and extreme—coordinate the activities of
people of the most remote familial relationships located in far-distant places. Tin
from mines in British Cornwall converted Mediterranean copper into bronze
well before a single political entity bound the areas. Our remarkable facility for
inducing people to pool their efforts permits great task specialization; think how
many individuals had some hand in bringing you a car or computer. Our facility
for generating transportation systems also means that we can transcend nature’s
need to make things out of locally available materials.

SIMILARITIES
A list of similarities turns out to be such a scattershot mix of major matters and
minor details that we gain little from the itemization. Most similarities between
the technologies emerge from inescapable physical rules and environmental
circumstances, both matters we’ve already dwelled upon. The best gravity-
resisting vertical column will be circular in cross section, whether it’s evolved or
manufactured, grown or cast, of wood or of concrete, of bone or of steel. Jet
engines find use by either technology only when some other special advantage
(such as high speed or mechanical simplicity) outweighs their inefficiency. Low-
pressure pumps capitalize on fluid-dynamic phenomena while high-pressure
pumps employ fluid statics. And so forth.
At this point more subtle and abstract similarities hold more interest:
similarities of process and historical trajectory rather than of product, similarities
for which common physical context provides no adequate explanation.
Addressing them returns us to issues quiescent since the second chapter.

• Major innovation is no easy thing for either technology, but for different
reasons. We persistently believe that the progress of human technology
depends on an adequate scientific base. Probably that’s only occasionally
true. George Basalla argues (and I don’t disagree) that we too easily and too
often exaggerate the contribution of science to technology.5 When the two
interact, technology more often drives science than the other way around;
steam engines stimulated thermodynamics, and aviation provided impetus
 for aerodynamics. More often, I suspect, the main difficulty of innovation in
human technology comes from the intrinsic complexity of introducing
something truly novel. A good idea isn’t enough. Steam turbines were
mentioned around 1800 by James Watt, but despite obvious advantages over
piston engines, they didn’t power ships for another century. Turbines made
unmanageable demands on metallurgy, techniques for precision fabrication,
and lubrication. A sail is a great device to spare human labor, but sailing in
all but the calmest of waters requires drastic redesign of oar-driven ships. A
computer as good as Babbage designed awaited electronics, a century later;
doing the job mechanically cost too much to hold sufficient appeal to launch
the technology. A developed technology has a great deal of momentum, more
often than not barring from the market competing schemes, even intrinsically
superior ones. About this last argument, more later.
Evolutionary innovation faces even greater difficulty. Natural selection
requires advantage all too immediately. That mandates continuity of design
and must rule out lots of life-forms that might ultimately have proved
superior. A system of inheritance that includes sexua! recombination and
recessive genes does allow changes in several features at once. But it doesn’t
encourage large- scale change in any particular way. Small changes
encounter different troubles. Competing with established designs must be
especially hard when a new design differs only marginally. In both
technologies refuges from the full burden of competition have obviously
been important—for instance, military programs for our products and
geographical isolation for nature’s.

• The time course of change or progress, depending on one’s outlook, contains

several curious parallels. When we look at relative rather than absolute
durations, both technologies seem to change slowly and steadily. Nature
achieved terrestrial life long after accomplishing large-scale organization and
mastered flight a long time later. Step by step, humans elaborated devices to
apply human muscle in ever more diverse ways; we then added, with
increasing effectiveness, the muscles of other animals; we further added,
again step-wise, nonliving energy sources. The same incremental increase in
potency marked food acquisition, construction of tools and dwellings, and
destructive social interactions.
Viewed more closely, slow and steady characterizes neither. In each
culture one can recognize some takeoff point that follows a very slow
 prepaleontology (for visibly large fossils) or prehistory (for record-keeping
cultures). Cells of sorts persisted for several billion years, both singly and in
small chains and aggregates, before they managed the specialization and
coordination that led to large creatures. But the latter appeared suddenly—
over only ten or so million years—and have been around for a paltry six
hundred million years. Humans occupied most of the earth long before the
modern era—less than ten thousand years—of specialized occupations and
elaborate coordination of individuals. Most of human history preceded all the
complex trade, transportation, and political organization that we think of as
definitively human. Whether change in nature or human societies continues
to accelerate beyond the period of takeoff is much less clear. The fossils from
the Cambrian, from more than five hundred million years ago, are
impressively complex, and a look at the artifacts, both large and small, of
ancient Egypt gives the same impression for things that humans make. Has
our technological progress in the twentieth century been greater than that
during the nineteenth? The nineteenth saw the spread of self-powered
locomotory systems, instantaneous communication with electrical devices,
cheap metals, and mass production. To these the twentieth has added mainly
electronics, aviation, modern medicine, and the development of polymeric
materials. The culture shock of moving back a hundred years would probably
be a lot less for one of us than for someone living in the 1890s.

• What about changes on a finer scale, changes in specific devices and

characters? In both technologies they’re at least as episodic as broad
organizational changes. A trigger gets the ball rolling, to mix metaphors,
initiating a period of rapid change. Some critical environmental circumstance
may alter: A seed lands on an island; a ship reaches a new continent. Some
novel material or general component becomes available: a growing skeleton
or edible grass in nature, a stronger metal or ship’s propeller for us. Or some
key constraint gets removed: Teeth appear that can tolerate the abrasives in
grass; dynamos supplant batteries as primary electrical sources. For either,
the range of possible triggers is wide.
Whether evolution is predominantly steady or episodic has engendered
considerable controversy, but we needn’t worry here about the debate
between “gradualists” and “punctuationalists.” At the molecular level,
evolution proceeds gradually, with the genetic material changing at a
relatively constant rate. At the level of the functioning organism, things
change more erratically; a given amount of genetic change simply doesn’t
cause a fixed amount of organismic alteration. No paradox attaches to this
difference; many complex and interactive processes intervene between a
sequence of bases in DNA and a functioning, multicellular structure. At the
molecular level gradualism reigns. At the organismic level a considerable
degree of unsteadiness has long been at least tacitly recognized; the present
debate just concerns the degree to which a punctuational model should be
considered the default. Evolutionary biologists care about evolutionary
change, so they’re jarred a little by the idea that most species most of the
time are well adjusted and do very little overt evolving.6
The episodic character of change in human technology—or in human
institutions in general—isn’t often questioned. Or at least not recently,
however persuasive was the steady “onward and upward” view of human
progress a generation or two ago. “Revolution” may be a buzzword, with an
agricultural revolution, an industrial revolution, and even a postindustrial
revolution, but it’s not inappropriate. Any history of technology provides lots
of examples of rapid change on the heels of some key innovation, whether
early bronze or cheap steel, steam engines or internal-combustion engines, or
the fast electronic switches of tubes and transistors. Not that “rapid” means
instantaneous. Applications lead to improvements in cost and efficacy of the
basic innovation, which then lead to other applications, but even such a
positive feedback process takes time to operate. A century elapsed between
Newcomen’s bulky steam-driven pump and engines sufficiently light and
efficient for self-propelled land vehicles—for Trevithick’s steam carriage and
Stephenson’s locomotive. But that much delay may be unusual, the special
result of engines worked by pressure differences in a technology still unable
to handle large pressures. Telegraph, telephone, and electric motors followed
closely on the easy availability of batteries and wire. Horse-drawn urban
transport completely disappeared within a few decades of the development of
practical internal-combustion engines.
The technologies share another factor that reduces the steadiness of
change. Improvements in new designs come more easily than improvements
in well-established designs. For nature, the chance (always low, of course)
that a mutation will be favorable and ultimately fixed in the population will
therefore be greater if it occurs in an organism that has recently undergone
other change or recently invaded a new habitat. A larger fraction of
mutations will be neutral or detrimental when they happen in creatures that
 have been doing the same thing in the same place for eons. In our domain,
improvement moves from the province of the casual amateur to the
experienced professional—whether one considers farm implements,
automobile engines, or computers.

• If revolutions are easy to recognize, cases of stasis are no more obscure. In

nature we needn’t invoke famous living fossils, such as the lobe-finned
coelacanth fish, the ginkgo, or the horseshoe crab. On the evidence of either
very similar fossils in beds of different ages or members of stay-at-home
species on different and distant islands (where accidental introductions must
be rare), change must be neither continuous nor inevitable.
Even amid the ostensible rush of contemporary change, many things
endure. Once stripped of a lot of adventitious machinery, the basic four-
cylinder engine and manual transmission of my car wouldn’t much puzzle a
mechanic who’d been asleep since 1930. A clothes dryer in our house
replaced one that served for twenty-five years; most of its parts are
interchangeable with those of its predecessor. Typewriters developed rapidly
after their introduction but then changed little for the next half century. Jet
aircraft have changed only slightly in the last thirty or so years. Stasis is
evident in the basic types of electric motor, in almost all our pumps and
industrial power transmission devices, and in the techniques of
manufacturing most of our basic materials. Information handling—the
nervous system’s analog— may have changed dramatically, but the final
effectors—the industrial equivalents of muscles and bones—have been much
more stable. Sensors, computers, and robots may replace human operators,
but they mostly run the same tools for cutting, shaping, and assembly.

• In both technologies a reduction in diversity can follow an episode of

innovation and initial diversification. Temporarily reduced competition
permits more experimentation than the norm, creating a climate amenable to
creativity and proliferation. Restoration of fully competitive selection by
nature or marketplace then prunes the progeny.
The first phase of the natural phenomenon is called ecological release: A
species expands and diversifies to take advantage of multiple habitats.
Particularly famous examples include the proliferation of cichlid fishes in the
great lakes of East Africa and of finches in the Galápagos Islands off Peru.
Cases of the second phase—ecological displacement—emerge when one
 compares the wide variability of particular kinds of organisms on islands
where they face no competition with their more limited variability where
they do.7 On a vastly larger scale, Stephen Jay Gould views the earliest
macroscopic fauna, the Ediacara of the late Precambrian, something more
than half a billion years ago, as a diversification of which only a small
element remained as ancestors for later life.8 But his view of the Ediacara
remains controversial. Other cases in which a few better designs displaced
many less effective ones are probable but problematic. Still, while evidence
of overall diversity reduction may be lacking, a decent consensus of
paleontologists agrees that the past half billion years haven’t seen a steady
increase in the diversity of life-forms.
The consolidation more clearly characterizes human technology, with a
concatenation of forces driving the shakedown. Industrywide, national, or
universal standards get established, from systems of screw threads and drill
gauges to computer codes. Superior supply and repair networks make some
designs more attractive to consumers. The familiar design is the comfortable
one—our fascination with novelty can easily be exaggerated—increasing the
dominance of whatever version gains initial preeminence in the marketplace.
And true technological superiority does certainly matter, despite all the
factors that dilute its impact. So the “mature” technology often ends up less
diverse than its “revolutionary” antecedent.9 (Sacred literature, curiously,
may undergo an analogous proliferation and then standardization; the early
Gospels, for instance, got pruned as they got canonized.)

• Another factor homogenizes both technologies. We do the same things in the

same ways over the entire earth, the result both of fad and fashion and of our
long-distance transportation and communication. The same global
transportation has spread many kinds of organisms beyond their original
geographic ranges. Some have been spread accidentally,10 and some
deliberately, most of the latter through movement of our domesticated plants
and animals. Whatever the impetus, homogenization decreases diversity in
both technologies; indigenous crafts (along with many other aspects of
indigenous culture) are supplanted, and local species are driven to extinction.

DRAWING ANALOGIES
Similarity of product need not imply similarity of process. I’m persuaded that
comparing the products of the two technologies lends breadth to our thinking
and gives insights not otherwise evident. About processes I’m more equivocal.
Natural selection is a most peculiar process, and its limitations are inadequately
appreciated. One often encounters analogies between the processes by which
human technology changes and evolution by natural selection. I think these
badly need the scrutiny of a biologist.11 To start with, a clear distinction ought to
be made between mechanism and history—the distinction we biologists make
between natural selection and natural history. The more basic problem is that an
analogy doesn’t explain. One judges an analogy not by whether it’s true, trivial,
or untrue but by whether it’s useful, nonuseful, or misleading.
The evolutionary analogy counteracts the personified, heroic view of
technological change that afflicts our early education. With that I have no quarrel
and will say little more. The analogy helps little, if at all, in understanding the
role of personal creativity in technological change. Too bad—the most crucial
and most mysterious element of our technology is the origin and nature of
human creativity, something without parallel in nonhuman affairs.12 Where
change hinges on inheritance and the latter cares nothing about personal effort,
no great inventor or political leader can play a role. Nature can only show that a
system need not be creative in intent in order to be innovative in result. While
the point isn’t intuitively obvious, it’s also not all that important. Who can doubt
the central role of creativity in the advance of human technology? For our
technology, increasing the role of marketplace selection relative to that of
creative intention would just add irrationality, inefficiency, and danger.
An analogy or point of similarity may either provide a good goad or a poor
substitute for further analysis, but in itself it’s nonanalytic. For instance, Robert
Heilbroner makes the statement that “advances, particularly in retrospect, appear
essentially incremental, evolutionary. If nature makes no sudden leaps, neither, it
would appear, does technology.”13 Nice words, but where does one goes next?
Incidentally, note his use of “advances,” with its implication of progress. That
presumes some equivalent type of progress in nature, something not as obvious
as we once thought. Perhaps evolution was once progressive but long ago
reached a state in which complexity and diversity change only randomly.
Perhaps, by contrast, human technology has yet to reach such a plateau and
remains progressive. If so, then we mislead ourselves by comparing their recent
alterations.
 Technological determinism is another notion for which natural selection
supplies an analogy of uncertain value. Does (or, more realistically, to what
extent does) technology drive history? Figures who debate the issue run the
gamut from Karl Marx to recent social critics of technology, such as Lewis
Mumford. Inventions suggested as historically critical include the stirrups, breast
straps, horse collars, and heavy plows that allowed efficient application of the
power of large animals in the Middle Ages.14 Equivalent determinism in natural
history is so obvious that the analogy helps very little. For organisms, little
matters beyond natural selection— that is to say, reproductive success. No social
consciousness, political expediency, informed judgment, or mass hysteria can
confuse the story.
A Darwinian or selectionist analogy attracts us because natural selection is
more straightforward and better understood than less dependably irrational
systems; the latter resist reduction to syllogisms, such as that of Chapter 2. But
no matter how attractive a selectionist model might be, it at least oversimplifies
and may do even worse. For instance, humans intentionally minimize selection
among competing designs by analyzing before fabricating and by field testing
before marketing; we do our best to anticipate the effectiveness of our devices.15
As Joel Mokyr puts the crux of the difference, “new ideas and mutations are
inherently different in that mutations are copying errors, while ideas are
deliberate attempts to make a change.” Furthermore, we learn from failure, a
factor well argued and illustrated in several books by my colleague Henry
Petroski. In addition, we can ask, “Wouldn’t it be nice if we had something that
would . . . ?” Human technology emerges from some complex combination of
the rational and the irrational, and we shouldn’t let familiarity disguise its
resistance to tidy analysis. At the same time we shouldn’t let its analytical
complexity disguise its practical effectiveness.
The selectionist model is dangerous in another way, one more of application
than applicability. What’s done by nature is by definition natural. But the word
“natural” carries a strong connotation of rightness, even of sanctity. We must
reject any general notion that nature’s ways indicate proper procedures for
people. Sometimes nature provides a model; in other instances veneration of the
natural is just snake oil for our contemporary troubles. Not that we don’t face
unprecedented and possibly insuperable problems, and not that some of these
aren’t unintended consequences of human technology, but we can’t indulge in
the simplistic confidence that these problems will be solved by going back
(whatever that may mean) to nature, to what’s natural, or to natural selection.
What nature does show is that a complex technological system can appear,
function, and persist without moment-to-moment micromanagement by some
prescient guiding hand. That, though, the continued existence of complex,
technological, capitalistic economies demonstrates as well. Furthermore, the
latter should be judged on their own merits, in particular on how they contribute
to our social goals, and not on any analogy with nature. Nature may be nicer
than the “red in tooth and claw” monster (Tennyson’s phrase) of a century ago,
but nothing should impel us to hold it up as model for emulation. Darwin
followed Adam Smith, not the other way around.
I don’t know a biologist who’s a nature worshiper in this quasi-theological
sense. We’re not more rational, merely battle-scarred by repellent social
doctrines that have used natural processes for justification, doctrines perpetually
reemerging with new names and new sponsorships. One is social Darwinism
(mentioned in Chapter 11), the turn-of-the- century idea that since nature
indulged in unfettered competition, it provided a proper operating principle for
human society. Another is biological determinism, in its extreme form the denial
of any possible amelioration of the deficiencies of one’s heritage, whether
personal or racial. But we’re wandering from the present argument.
Not only is natural selection attractively logical, but similarities between the
histories of life and of human technologies tempt wondering about common
mechanisms. One parallel is what economic historians call lock-in and some
paleontologists have termed the privilege of incumbency.16 Each observes that
an established device or life-form isn’t readily displaced, even by something
economically or selectively superior. In economics this strikes yet another blow
to the notion of the ideally competitive market. In biology it denies full
applicability of what’s called the principle of competitive exclusion—the idea
that if two groups (at the species level or above, so they can’t mix genetically)
are in complete competition, one will displace the other.
The privilege of incumbency for a well-established group of organisms is the
resistance to extinction conferred on it by a large geographic range, a large
number of species, or a large number of individuals. Its reality rests on less than
fully satisfying indirect evidence; because of the time spans necessarily
involved, we see only outcomes and not competitive battles. But the privilege of
incumbency rationalizes a lot of otherwise paradoxical features of the fossil
record. In Chapter 6, the possibility was raised that the nonmetallic character of
natural technology might be a case of incumbent privilege, but as noted then,
how can we ever know for sure?
For human technology, the evidence for lock-in is better but still imperfect,
even if the general idea is too logical to deny. The persistence of the QWERTY
typing keyboard provides the usual example of lock-in. The QWERTY
arrangement of keys was initially chosen to minimize jamming rather than to
maximize typing speed. After technical improvements fixed the jamming
problem, one might expect that an ergonomically superior keyboard would have
replaced QWERTY. But retraining costs time and money, and facility with two
keyboards may be counter-productive to performance with either. A faster
keyboard, the Dvorak, has never gained acceptance.17 But the case may not be
as good as it looks. Comparative tests of the two keyboards were badly flawed
and almost certainly biased. The QWERTY arrangement turns out to be better
than an alphabetical or a random one; forcing the hands to alternate to avoid
jamming enhances typing speed as well.18
Another flawed example is the persistence of the VHS videotape system in
competition with the purportedly superior Beta system. Any slight advantage of
Beta was offset by the disadvantage of a shorter recording time—initially an
hour, less than the normal length of a movie.19 The argument that the internal-
combustion engine got locked in, that we’d have good electric cars if we’d
standardized on them a century ago suffers from similar flaws. The argument
presumes that any technology can be improved without limit given sufficient
incentive or investment. But only battery weight limits electric cars, and military
interest in battery weight for submarine use (before the advent of nuclear power
in the 1950s) as well as similar imperatives of the space program must have put
maximum muscle (and a nice competitive shelter) into the search for weight-
efficient storage of electricity.
Nonetheless, lock-in—market failure because of persistence of initial choices
—must be common. Is the present computer screen the best shape or merely a
persistent version of a television set? Is the thirty-five-millimeter film format for
still cameras, with all that space given to perforations, an accident traceable to
the initial adoption of thirty-five-millimeter movie film? And lock-in may
represent no strict market failure. After all, a more complex item may be cheaper
to produce if it can be done in larger quantities since the cost of designing and
tooling up gets spread more widely.20 We have to ask, though, whether an
equivalent phenomenon in nature says anything about the existence and role of
lock-in. Market dominance, irrational behavior of consumers, existing
infrastructure to support initial choices—none has a precise analog in the factors
underlying nature’s privilege of incumbency.

CONES AND SPIRALS—ONE FINAL TALE

We’ve seen the deep pitfalls of arguments based on similarities between the
processes of natural and human technologies. Arguments based on similarities
between their products incur less risk of self-deception. But the risk is reduced,
not eliminated. A second look at a subject mentioned in Chapter 2 will do better
than any admonition.

FIGURE 14.1. Increasing the size of a cone without changing its shape, along with a limpet. Limpets are
essentially low, uncoiled snails.

FIGURE 14.2. Common cones of human technology: (top row) conical wheel bearings and their race from
an automobile, ground glass stopcock, rubber faucet washer, ground glass stopper; (middle row) threaded
pipe, cork, tubing connector, the needle valve of the water supply for an ice-making refrigerator, the
centering device of a metal lathe; (bottom) hammer stock with conical top.

Consider a simple cone. A cone can be characterized by two measures, its

diameter and its edge to apex distance, as in Figure 14.1. Enlarging the cone by
extending its lower edge changes its size but not its shape; doubling the diameter
doubles the edge to apex distance. By contrast, extending the end of a cylinder
makes it skinnier and skinnier.
Cones have another interesting property. If their inside and outside tapers are
the same, cones can be nested together as tightly as one wishes and in whatever
number one wishes—the way we ship and store cones for ice cream and conical
paper cups. Pressing them together snugs the connection. Cones of identical
taper are easy to manufacture, whether by cutting on a lathe, by casting in a
mold, or by any of several other techniques. Again, the contrast with cylinders is
sharp. Nesting cylinders find use for such things as telescoping radio antennas.
But their precise sizes fix once and for all the tightness with which they fit
together. Nor will identical cylinders nest together.
While both technologies make great use of cones, they do so for different
reasons. For nature, the essential advantage must be the first one noted here, the
ability to grow by adding at the edge without change of shape. That makes cones
good for things like mollusk shells, where growth takes place only by such
addition. Since we don’t make artifacts that grow, that advantage means little to
us. For us, the advantages that matter are their ability to nest and the way they
can be press- fitted with controllable snugness; Figure 14.2 shows several such
applications.
(We’re looking at severe biases, not absolute dichotomies. Some identical
cones do nest in nature. Many jellyfish in their sessile stage produce a stack of
what are called strobili, each of which will break off to form a recognizable
swimming jellyfish. The individual muscular units of the bodies of fish form
nested cones, as you can see by picking apart cooked specimens. And you make
use of the way cones permit ungrowth, if not growth, without change of shape,
every time you sharpen a pencil.)
The axial cutting tools on a lathe (drills, reamers, and so forth) fit into the
hole in the tailstock (the opposite end from the motorized head-stock) as the
male halves of a pair of nesting cones. Pressing inward, as happens when the
tool is used, tightens the fit, but if the tool suddenly jams in the work, it comes
loose from the tailstock and spins nondestructively. Not only are such nesting
cones easy to manufacture, but the angle of taper isn’t sensitive to the changes in
size that accompany heating and cooling. Old-fashioned ground-glass stoppers
were always conical for the same reasons: ease of achieving the necessary fit and
dependable behavior when tightened or loosened. By contrast, the pistons of
glass hypodermic syringes have to be individually fitted to their respective
cylinders. They’re often numbered, so they can be remated after communal
washing or sterilization. Corks for single use, as in wine bottles, are cylindrical;
corks for repetitive use (as in laboratories or in an earlier generation of Thermos
bottles) are conical so they will fit a range of bores and fit as well after long use.
Few of us notice how modern technology prefers cones over cylinders. The
threads on the end of a metal pipe differ from the threads of ordinary screws;
they’re not uniformly deep grooves on a cylinder but get deeper toward the
pipe’s end, forming part of a long cone. As a result, pipes can be screwed to the
same tightness even if their threading depth varies a bit. Automobiles have
conical rather than cylindrical roller bearings within their wheels, so tightening
an outer bolt to a standard torque can counter-act wear or variation in
manufacture.21 But—again—the advantages of these cones differ from anything
gained by the shell of a mollusk.

FIGURE 14.3. Two logarithmic spirals. On the left, each radial line is sixty degrees from the previous one
and is twice as long.
 Nature likes not just ordinary cones but a derived kind of cone as well. Horns
and tusks, clamshells and snail shells, even the tiny protozoa that formed the
white cliffs of Dover in southern England—all are conic derivatives. Recall how
height and diameter are proportional in cones. That same relationship holds for a
more general class of bodies, logarithmic or equiangular spirals; using curved
instead of straight lines produces these particular spirals instead of ordinary
cones. A line spirals outward from a starting point in such a way that the channel
between turns gets wider in proportion to how far outward the spiral has
grown.22 For most of us a drawing, as in Figure 14.3, works better than words.
Like cones, logarithmic spirals can grow without change of shape by the simple
addition of material. Moreover, they’re consistent with biologically typical
logarithmic growth—growth in which how fast material is added is proportional
to how much material is already present. In fact, what were loosely described a
few paragraphs back as cones, things like limpet shells, are really logarithmic
spirals of very low curvature.

FIGURE 14.4. Logarithmic spirals in nature: a pinecone, horns of a sheep, a section of a chambered
nautilus, and a microscopic foraminiferan (a shelled protozoan).

Early in this century three biologists became fascinated with the way these
spirals appeared again and again in nature; Figure 14.4 gives a few examples.
For James Bell Pettigrew, they revealed a curious bias of the Creator, the
centerpiece of a huge antievolutionary tome. For D’Arcy Thompson, they
provided the best example of the geometric idealism or at least the mathematical
tidiness of living forms. Thompson’s house at 44 South Street in St. Andrews,
Scotland, now wears a commemorative plaque well adorned with logarithmic
spirals. For Theodore Andrea Cook, the issue turned on aesthetics, without such
cosmic overtones.23 We’re still attracted to them, but by their informational
simplicity and by their handiness for the way organisms most often grow. The
same mathematics that determines the growth rate of a fresh bacterial culture
(and the yield of an investment with steady and reinvested interest) generates a
logarithmic spiral.
Should our designs pay more attention to these logarithmic spirals because
natural designs demonstrate their desirability? No. Their prevalence in nature
reflects their compatibility with the way she does things. They share that
compatibility with cones, in essence unturned spirals. Our interest in cones
doesn’t carry over to spirals; spirals don’t nest. So for the way we build things,
the logarithmic spiral is irrelevant.

CONVERGENCES
Biologists since Darwin have known that similar organisms don’t always come
from close common ancestors; similar structural arrangements often evolve
again and again. The image-forming eye of fish or mammal bears a remarkable
resemblance to that of squid or octopus. That they differ in certain fundamental
but functionally irrelevant aspects confirms our conviction that no common
ancestor had such an image-forming eye. We know of innumerable cases of this
phenomenon (mentioned in the second chapter) of convergence—that between
marsupial and placental mammals, for instance. For biologists (from Darwin,
incidentally), it makes trouble—not for the theory of evolution, which happily
accepts convergence, but for the practical business of classifying organisms.
Since similarity of structure doesn’t indicate relatedness, convergence makes it
harder to decipher lineages. For the biologist interested in function, by contrast,
convergence is a handy tool. Functional significance (plus a few other things,
admittedly) ought to drive convergence as a direct consequence of natural
selection: What’s useful will be selected. Therefore, cases of convergence ought
to tell us what’s functionally significant—that is, what aspects of the design of
an organism matters to its reproductive success. What’s selected must have been
useful.
 In addition, convergence ought to point to what’s easy for the evolutionary
process to generate. If something turns up repeatedly, it must be achievable
without heroic alteration of hereditary information. Biologists who reconstruct
evolutionary lineages look for continuous scenarios and descent schemes that
have the fewest innovative steps. So “easy” isn’t a trivial consideration.
Perhaps the same approach might prove productive when we look at how
human technology has developed. All our communicative channels deprive our
contemporary cultures of the independence of nature’s separate lineages. But
that hasn’t always been so. Humans may have originated from a small number of
ancestral hominids, but we spread over the planet before we acquired much in
the way of technology. Perhaps convergence signifies the same things for both
technologies: that cases of convergence are fingers pointing to what matters and
what’s easy. The idea got proper notice from Joseph Needham in his
monumental work Science and Civilisation in China, but Needham, probably not
coincidentally, was an excellent biologist as well as a spectacular sinologist.24
Consider the early histories of our use of metals. Without a persuasive
alternative, one should prefer the most incremental scenario, the “easy” scenario
that requires the least genius. From evidence from the Middle East, a likely
sequence runs as follows. Copper, almost alone among the mechanically useful
metals, occurs naturally in metallic form. Copper ores commonly contain pieces
of metallic copper. Melting copper with some associated ore (perhaps to separate
the two) will yield more product than will a pure melt, so the transition from
melting to smelting presents no great difficulty. Ores with certain impurities will
yield better metal; antimony and arsenic bronzes find more uses than pure
copper. Tin bronze, less likely to poison the artisan, is still better, but it’s also
less likely to be fortuitously made. In most ways iron is a still-better metal, but
producing it demands higher temperatures.
How reasonable or likely is that sequence from metallic copper through
smelted ore, arsenical bronze, and tin bronze to iron? We gain confidence in that
reconstruction when we learn that the same sequence from copper to tin bronze
occurred in South America as well as in the Middle East, if a few thousand years
later in the New World.25 One can alternatively invoke what the anthropologists
and archaeologists call diffusion as an alternative explanation, implying that the
technology spread through occasional contacts between separate civilizations.
But I view diffusion as (to pun slightly) a farfetched scenario for how the
Americans went metallic—to the limited extent that they did.26 (Assuming that
they didn’t get there on their own also happens to be insulting, patronizing, or
worse.)
The development of woven textiles in the Old World and New World
provides a similar story of convergence. While some elements may have crossed
the Bering land bridge with early invaders of North America, much of the
technology must have been developed independently. Both Old World and New
World used looms, but features consistent within either but differing between
them point to independent innovation. The sequence in the two hemispheres
looks similar, running from untwisted to twisted cordage, from there to basketry
and knotted netting, and on to weaving, with improvements in spinning
technique and the labor efficiency of looms. In both instances, long fibers, such
as flax and sisal, seem to have been used before short ones, such as cotton and
wool; the short ones can be produced in greater quantity but are harder to make
into adequately strong thread. Both cotton and wool were developed in both
hemispheres, the cotton from different species of the genus Gossypium and the
wool from different animals—sheep and goats in the Old World, llamas and
alpacas in the New.27 A careful comparative analysis might reveal some general
principles of early textile evolution.
The courses of development of both metals and textiles point to the
importance of a sequence with an easy continuity, one that requires a minimum
of large-scale innovation and one the evolutionary biologist would find familiar.
Nor are metals and textiles the extent of possible subjects for comparisons that
focus on convergence. Fired pottery was widely used, but with different
techniques for shaping clay in the two hemispheres. Other candidates include
techniques for trapping, fishing, plowing, seeding, tanning, and boat making, as
well as weaponry for both hunting and warfare.28
Convergences might also serve in a way that’s almost the opposite, showing
not what’s easy but what’s hard for one or the other and forcing us to ask why
something is hard. To do this, one should look for elements that are ordinary in
one technology but rare or unknown in the other. Consider the use of tools.
While we know many examples of minor tool use by mammals and birds, one is
struck not by their existence but by their scarcity and casual character. The
limited use of tools contrasts sharply with the way animals use materials from
the environment to construct domiciles: caddisfly larval tubes; worm tubes;
hermit crab shells; nests of fish, reptiles, birds, and mammals; beaver dams; and
so forth. The limited use of tools contrasts quite as sharply with the way tools
proliferate in human societies. Why the difference? My only suggestion, a casual
and uninformed one, is that elaborate tool use may happen only after some
critical takeoff point. That takeoff was reached only once by any animal—by
humans at some stage in our prehistory. Perhaps it came with the acquisition of
learned, symbolic language and the fabulous improvement in cultural
information transmission that such language permitted. But would one ask the
question in this way except in the context of a comparison between the
technologies?

MESSAGES
This view of nature as a technology has provided an unusual perspective on the
world around us. Recognizing that nature deals with the same variables as do
human designers—in the present case mostly mechanical engineers—led us to
compare both products and processes. But the main test of the comparison is
utility. In short, what do we learn?
Identifying specific devices that we might profitably emulate constitutes the
least of what we gain. What’s the chance that in searching through the pieces of
an electric motor you’ll recognize a part that will improve the performance of an
internal-combustion engine? How much more remote the chance when searching
through the proteins of a muscle! By contrast, one may well expect insight on
design where natural technology is preeminent, but enough noise about that has
already been made. That each technology is a coherent entity, remarkably
distinct from the other, can be either advantage or disadvantage. Perhaps the best
encapsulation, if a trifle trite, is that nature shows what’s possible.
Disparaging things were said about analogies, but real utility balances the
risk analogies pose of short-circuiting proper analysis and explanation. A useful
tool emerges from recognition that the technologies involve analogous time
courses of development and ways of operation. One can test the logic and
credibility of hypotheses about how one system operates by examining what
happens in the other.
We’ve seen places where the products of the two technologies coincided.
And we’ve seen places where the products proved surprisingly different. Each
may carry a prescient message. Coincidence between these vastly different
technological contexts directs attention to constraints that neither can escape,
constraints that we must try to identify. Different solutions to the same problems
or different devices for the same task imply something equally interesting: the
possibility of a third or fourth solution or device. Thus the comparison can
suggest where major innovation might be possible and where it’s unlikely. Since
major innovation isn’t easy, a little help might go a long way.
The histories of both science and technology are replete with cases in which
someone with an unusual background or outlook—an outsider— solved a long-
standing problem.29 Can one deliberately become an outsider? Or, better, how
might one get the fresh perspective of an outsider without sacrificing the
insider’s expertise? For the human designer, a perceptive look at nature’s
technology can do just that: It can provide the wide-angle view that reveals
possibilities that would otherwise escape consideration.
I began with statements extolling the superiority of nature. In a sense this
entire book is my skeptical response. Sure, nature is wonderful. But bear in mind
what we do that she doesn’t: However they came about, the unique achievements
of our species deserve our full appreciation. Metallic materials, ropes of short
fibers, woven fabrics of crossed threads, the wheel and axle, thermal expansion
engines, fast surface ships, electro-chemical energy storage, lighter-than-air
aircraft—humans, our engineers in particular, have made extraordinary things.
How we use them is another matter altogether.30

In these pages the reader has been introduced to my immediate scientific

interest, biomechanics. As thoroughly comparative as any area of science, it
looks at the mechanical problems of existence, in particular at how solutions to
these problems vary among organisms of different sizes, different lineages, and
different ways of managing growth, reproduction, and dispersal. It has always
borrowed its basic concepts from engineering, and it has done so to its great
advantage. We’ve made unfettered and hugely productive use of the existing
technology of humans in our efforts to understand what nature does. Our books
and articles make no attempt to disguise the process; indeed, we’re continually
asked by biology students encountering biomechanics, “What is this,
engineering?” The importance to us of human technology as a critical reference
is best put by reiterating the admission made at the start. We’ve only rarely
recognized any mechanical device in an organism with which we weren’t
already familiar from engineering. Perhaps we’re insufficiently creative and
imaginative, or perhaps the engineers enjoy a long head start and numerical
superiority. Or perhaps the situation indicates the value of an external reference
for any attempt at understanding.
 NOTES

CHAPTER l: NONCOINCIDENT WORLDS

1. The real counterpart of biology isn’t engineering but a tiny field concerned with the nature and history
of human technology. A few recent books (to which further reference will be made) are those of
Basalla (1988), Vincenti (1990), Adams (1991), Petroski (1992), and Cardwell (1995). A key journal is
Technology and Culture; a central organization is the Society for the History of Technology (SHOT).
2. General books on the subject are those of Wainwright et al. (1976), Alexander (1983, 1992), Vogel
(1988), and Niklas (1992).
3. Nicomachean Ethics 1099B, 23. Quoted in Mackay (1991).
4. Quoted in Schneider (1994).
5. T. Ewbank (1842), p. 514. The author has a fine (and enthusiastic) appreciation of physiology and
natural history.
6. Papanek (1971).
7. The antitechnological literature is large and diffuse. Examples are books by Ellul (1964) and Mumford
(1967) and pieces by Commoner, Mumford, and Ellul in the general collection of Burke and Eakin
(1979). Florman (1975) gives a good look at it as well as providing a sober rejoinder.

CHAPTER 2: TWO SCHOOLS OF DESIGN

1. Thus The Evolution of Useful Things, The Evolution of Technology, and The Evolution of Design-
books by Petroski (1992), Basalla (1988), and Steadman (1979). See also Mokyr (1991), Dasgupta
(1996), and the final chapter here.
2. This logical scheme isn’t original here; its origin, though, is obscure. See, for instance, Endler (1986)
or Futuyma (1986) for exposure to the full richness of the subject.
3. Wainwright et al. (1976).
4. Raup (1966). A good recent book on shells from an evolutionary perspective is that of Vermeij (1993).
More recent computer-generated “virtual shells” are given by Meinhardt (1995).
5. Our special attachment to soft-shell crabs is a case, perhaps trivial, of a predator’s preference for a
newly molted arthropod.
6. Pierce (1961).
7. Crane (1950).
8. But evolution hasn’t scorned the possibility. Having two nonidentical sets of genetic information in the
cells of most organisms (diploidy) and engaging in recombination (sex) permit sheltering variation and
testing it in different combinations.
 9. Dawkins (1986).
10. These are clearly separate evolutionary developments. As a biologist would put it, the nearest common
ancestor of birds and bats, for instance, lacked wings and didn’t fly-according to paleontological,
anatomical, and molecular evidence.
11. Termites, close relatives of cockroaches, constitute the only case of hypersociality in insects outside of
the Hymenoptera, the group consisting of ants, bees, and wasps. Repeated evolution of hypersociality
among Hymenoptera is a consequence of a predilection peculiar to their particular (and peculiar)
genetic system; some other explanation (such as infectious symbiosis) is needed for the termites. See,
for instance, Wilson (1980).
12. Dyer and Obar (1994) give a little further information and a useful context; the basic source is
Cleveland and Grimstone (1964). One hopes not to be too familiar with spirochetes, one of which is
the pathogen in syphilis.
13. Tamm (1982).
14. See, for instance, Thompson and Bennett (1969), for a case in which nematocysts of sufficient potency
to harm human swimmers were expropriated. The basic phenomenon has been known for almost a
century.
15. The case is described by Fisher and Hinde (1949) and Hinde and Fisher (1951). Griffin (1984)
summarizes it in prose, and Burns (1975) does the same in verse. My colleague Peter Klopfer tells me
that the birds could do even better when challenged. The types of milk were distinguished only by the
color of the bottle caps (presumably so returned bottles needed no sorting). Not only did the birds
choose the bottles with cream beneath the caps, but they rapidly altered their color preference when
caps were deliberately switched.
16. Galbraith (1967).
17. Eldredge and Gould (1972). What must be emphasized is that sudden bursts of change are claimed to
be sudden only in a paleontological sense-that is, when one is looking over very long periods of time.
No serious challenge to the underlying Darwinian model is implied, merely the recognition that the
relative gain in fitness by change in form is enormously circumstance-dependent. For that matter,
Darwin himself recognized (in The Origin of Species) that most species remain largely unchanged most
of the time.
18. Grasses and grassland are among the earth’s more recent acquisitions. Grasses are especially tolerant
of fire and herbivory, and many grasslands revert to forest if these factors are removed. But the
nutritive value of grass is low, and it’s especially mean stuff to masticate in quantity.
19. The portable electric fan was in fact invented a lot earlier-by Nikola Tesla around 1890.
20. See, for instance, Basalla (1988) and Cardwell (1995). The former stresses continuity while the latter
has a more saltatory or heroic bent but, even so, is more incrementalist than are low-level textbooks. I
take no stand on the controversy, which isn’t central here anyway. But I do believe that any analogy
with biological evolution is almost entirely irrelevant. Vincenti (1990) and Adams (1991) flesh out the
issue with specific cases; in essence the former is about what engineers know (as in its title) while the
latter is about what they do. One might also look at basic textbooks of mechanical design as used in
engineering courses.
21. The discovery and biology of the coelacanth are a fine story, well told by Thomson (1991).
22. Houston et al. (1989). Some other flies have colonized the water- filled traps under our sinks.
23. Some biologists make a distinction between “parallelism” and “convergence.” But it’s a hairsplitting
one (see Roth, 1996) that we’ll ignore.
24. See, for instance, Stanley (1987) and Raup (1992).

CHAPTER 3: THE MATTER OF MAGNITUDE

 1. Good general accounts of the biology of size are given in several books, especially McMahon and
Bonner (1983) and Schmidt- Nielsen (1984). For brief and engaging introductions, see Haldane
(1928); Chapter 2 of Thompson (1942); or Went (1968).
2. Bennet-Clark (1977).
3. Descartes’s Principia quoted by Kearney (1971), p. 156. Size mattered to Horace, much earlier:
“mountains will be in labor, the birth will be a laughable mouse.” I’ve thus, as James Thurber once
wrote, put Descartes before Horace.
4. Thompson (1942), p. 68.
5. The matter is called Cope’s law by evolutionary biologists, after the paleontologist E. D. Cope (1840–
1897).
6. Robinson and Frederick (1989). We’ve occasionally done this as a class exercise, with people tracing
the feet and recording the heights of each other and of their particularly large and small acquaintances.
Of course that may not be the sole reason why big people have bigger feet.
7. Haldane (1928). Galileo’s rule about all bodies falling at the same rate works only in the absence of air
or as an approximation for large, dense bodies.
8. Orwell (1937), p. 26.
9. I’ve taken most of the material on fluid mechanics-lift, drag, gliding, types of flow, diffusion, surface
tension-from my earlier book (Vogel 1994a), which contains copious references to the basic literature.
Tennekes (1996) gives an admirably accessible account of size and flight.
10. Cope’s law is mentioned in note 5. The three lineages of extant flying animals—birds, bats, and insects
—appear to be significant exceptions to the rule. That’s probably because flapping flight evolved from
gliding, and gliders work better if large. The tiniest insects, the microchiropteran bats, the
hummingbirds—all are members of relatively new groups within their general kinds.
11. The skeptical reader will ask whence comes the power source for staying aloft with fixed wings; after
all, propellers, but not wings, are attached to engines. Nothing ethereal is happening; a fixed wing
incurs more drag when it’s producing lift, and the propeller and engine have to work harder to offset
that drag.
12. These pores are in the feltwork of fibers that make up the walls of cells within leaves. They’re not the
stomata on the surface of leaves through which gases pass in and out, which are around a hundred
times larger.
13. The efficacy of diffusion depends substantially on the size of the diffusing molecules, and good
perfumes are complex mixtures. Were diffusion the main agency of spread, the character of a perfume
would depend on one’s distance from the source! Writers of science fiction might take note. My
curmudgeonish diatribe on the matter is in Vogel (1994d).
14. We touch here on a general argument for why cell size is fairly constant and why organisms use a
cellular scheme of organization, one elaborated in Vogel (1988, 1992).
15. We do use the scheme of the maple seed occasionally. Some obscure aircraft (called autogiros,
gyroplanes, or gyrocopters) use the principle, and a helicopter whose engine has quit slows its descent
with the maple seed’s trick. I’ve provided a more extensive explanation of these various ways to fall
more slowly elsewhere (Vogel, 1994a). Current interest centers on using such single winged
autogyrating aircraft to deliver payloads from orbit to the surface of Mars; see Stephen Morris’s work
at http://aero.stanford.edu/MapleSeed.html.

CHAPTER 4: SURFACES, ANGLES, AND

CORNERS
1. We commonly refer to the component of the load coming from the weight of the structure as dead load
 and that from useful activity as live load.
2. 363/303 = 1.73, a 73 percent increase. 364/304 = 2.07, a 107 percent increase. I did a quick
introduction to the relevant part of beam theory in a book called Life’s Devices. A better source (this
isn’t false modesty) is Gordon (1978).
3. One more consequence: If sag increased with length and decreased with thickness to the same power,
then bookshelves could be built with the two in simple proportion—twice as long being twice as thick.
But since sag increases with length to the fourth and decreases only with thickness cubed, the longer
shelf has to be disproportionately thick; twice as long needs to be four times as thick. That’s a major
reason why larger structures must have a disproportionate amount of supportive material, an argument
made with a slightly different example in the last chapter.
4. The reason why camber is aerodynamically important is fairly complex, more so than implied by the
usual prevarication. I give a fairly equation-free account in Life in Moving Fluids; other accessible
sources are Sutton (1949) and von Kármán (1954). The phenomenon will come up again in Chapter
12.
5. A. Roland Ennos (1988) shows that camber and lengthwise twist were largely reversed passively—that
is, by aerodynamic forces—rather than, as had previously been assumed, by highly coordinated
muscular action.
6. The rule often goes nameless; in any case it’s not clear that the French mathematician Pierre-Simon,
marquis de Laplace (1749–1827) is its originator. (“Laplace’s equation” is something else again.) For a
cylinder the rule is similar, omitting only the one half; cylinders are curved in one dimension, spheres
in two. Since the pressure is resisted by tension running in only one direction, twice as much tension is
generated. A cylinder with hemispherical ends (such as a balloon or boiled sausage), pressurized until
it explodes, will usually split its sides before blowing off its ends.
7. Shortses are marvelously roomy by comparison with the flying cigars into which we’re more often
inserted. They also seem to rattle a lot. Is that real and related or just my imaginative anticipation of
logical consequences?
8. This was the Israel Museum in Jerusalem; it has the special virtue of dealing with an unusually long
and continuous history of human occupation of a place exposed to input from diverse cultures.
9. Walsby (1980).
10. Put another way, is drawing graphs in Cartesian coordinates—that is, with mutually orthogonal x, y,
and sometimes z coordinates-in some sense natural for us?
11. The key reference is Flannery (1972); see also Hodges (1972) and Saidel (1993).
12. The original paper is by Allport and Pettigrew (1957). The illusion was reported by Ames (1951) and
is described well by Hochberg (1978). The work with the Zulu is discussed by Owen (1978) along with
work of Annis and Frost (1973) on Cree Indians.
13. Frazzetta (1966); Gans (1974).
14. Westneat and Wainwright (1989).
15. Parkes (1965), for instance. Neither three- nor four-strutted devices exhaust the world of mechanisms
and statically determined structures; we’ve the tip of an iceberg here.
16. Gould (1980).
17. Well-evolved walkers in fact aren’t consumed by some search for static stability. After all, when
running, you have no legs at all on the ground a fair fraction of the time.
18. Which the U.S. military has tried, very expensively, to do for many years. One might ask about using
more than six legs. Aside from the obvious increase in mechanical complexity and potential
interactions and stride-length limitations, another factor enters. Recall from the last chapter that skinny
columns buckle under lower loads than fat columns of the same length. The material investment in
using a large number of thin columns for support turns out to be greater than using a small number of
fatter columns, as nicely explained by Gordon (1988).
19. Between different languages and systems of mathematical notation, it seems unlikely that knowledge
 of the theorem spread after some single genius discovered it. See Bose (1971), Anderson (1972), Swetz
and Kao (1977), and Prakash (1987).
20. Grant (1968), Buckley and Buckley (1977), and Barlow (1974), respectively.
21. D’Arcy Thompson (1942), mentioned earlier in connection with size, has a nice discussion of the
relevant geometry and biology. His book, still available in its entirety or in an abridgment by J. T.
Bonner (Thompson, 1961), is at once unique in content, imaginative in approach, and magnificent in
the elegance of its prose.
22. Hulbary (1944) faces the complications of our less than ideal world.
23. A splendid treatment of the phenomena associated with cracking—fracture mechanics—is given in any
of three books written by James E. Gordon (1976, 1978, and 1988). All three are simply wonderful
works in general. The 1978 one, Structures, gets my vote for the best book ever written on a technical
subject for a nontechnical readership.
24. An engaging and accessible treatment of such aspects of the construction of trees is given by Mattheck
(1991).

CHAPTER 5: THE STIFF AND THE SOFT

1. In part, of course, because we can measure very small changes in length. Parts per million are no trick
at all for modern technology applied to very stiff samples.
2. From Gordon (1978).
3. Note that “elasticity” is being used in a most peculiar sense. First, the more easily stretched a material,
the lower is its Young’s modulus, and second, whether and how forcefully the material returns to its
original length are immaterial.
4. “Toughness” sometimes is used for the energy involved in the actual process of breakage as opposed to
that en route to the breaking point. In breaking, new surface is created, and that takes an additional
input of energy, an amount related to the resistance of the material to making surface and to the amount
of surface that’s made in a break. Real materials, after all, don’t all break evenly across their diameters.
While work of extension is an energy per unit volume because a whole specimen gets stretched,
“toughness” in this latter sense is an energy per unit area because the energy is invested in making
surface. The overall work of breakage of an object that’s initially unstressed involves of course both.
5. That’s one of many consequences of one of the most important of all rules in science, the second law
of thermodynamics. Stated anecdotally, it declares that in any real process you get out something less
than you put in, with the difference appearing as heat. The comments of C. P. Snow (1959) on it are
more relevant than ever as we become increasingly afflicted with people who view science strictly as
social construct.
6. Such as hardness, based on who scratches whom; properties such as ductility, important in
manufacturing; a host of time-dependent properties relating to differences in response depending on
how fast loads are applied and recovery after loading for various times; and various thermal properties.
7. One might argue that rubber is (or originally was) entirely a natural product. But in the rubber tree it’s
a liquid, used by humans mainly as a coating for waterproofing. So rubber (like rayon) is something
merely derived from a natural product; its mechanical utility began with Charles Goodyear’s discovery
of the vulcanization process in 1839.
8. Spider silk isn’t actually a single material. Different kinds of spiders produce silks of somewhat
different composition and properties, and those that spin nice orbs use several kinds in a single web.
Silk moth silk is still another group of related materials. On the mechanics of spider silk, see Gosline et
al. (1986) and Vincent (1990).
9. Harrar and Harrar (1962).
10. The pejorative “crackpot” must be a tacit cultural recognition of the lack of toughness of ceramics.
11. In particular, I like “A Chapter of Accidents” in Gordon (1978) and books by Levy and Salvadori
(1992) and my colleague Henry Petroski (1985, 1994).
12. A version of this section has been previously published as Vogel (1995).
13. See Koehl et al. (1991).
14. See Grace (1977) or Miller et al. (1987).
15. See Vogel (1984a, 1989, 1993).
16. O’Neill (1990).
17. Koehl (1977). We sometimes demonstrate the relationship between the diameter of a cylinder and the
ease with which it bends by using cylindrical pasta—spaghetti—of various diameters but of the same
brand. A strand is supported between two bricks, fishing sinkers are added, and sag is measured with a
ruler. Both the fourth power rule for weight versus sag and the third power rule for distance between
supports versus sag emerge reasonably well, at least for small sags—Vogel (1988), p. 337.
18. Shadwick et al. (1990); Shadwick (1994). Squid and octopus are cephalopod mollusks. Vertebrates
split from the ancestors of mollusks and arthropods (insects, spiders, crustaceans, etc.) at a simple,
wormy stage of animal evolution roughly half a billion years ago.
19. Alexander (1984a, 1988).

CHAPTER 6: TWO ROUTES TO RIGIDITY

1. Tungsten, for instance, has recently been shown (see Chan et al., 1995) to be a crucial component of
certain very unusual bacteria that grow best at temperatures up to the boiling point of water. A general
source on metals in organisms is Kendrick et al. (1992).
2. All these data are from Schmidt-Nielsen (1990), a textbook that really explains things rather than just
naming and mentioning them.
3. The main source here is Vincent (1990). Lowenstam (1967) points out that the teeth of chitons
contained enough magnetite so that substantial amounts of the magnetite in the sediments beneath
shallow seas might be of biological origin.
4. Blakemore (1975).
5. Magnetite is, for the chemically minded, Fe3O4 or Fe2O3FeO, distinguished from ferrous oxide, FeO,
and ferric oxide, Fe2O3. (Ferrum is the Latin word for “iron.”) Accessible references include J. L.
Gould et al. (1978) and Walcott (1979).
6. Photosynthesis, in case you’ve forgotten, splits water as a source of hydrogen and combines the latter
with carbon dioxide, liberating oxygen. Before life invented the process, the atmosphere had little or
no oxygen in it; thus oxygen is the first and greatest pollutant.
7. Association of exposure to aluminum with subsequent development of Alzheimer’s disease looks less
likely than it did a few years ago. The Merck Manual (Berkow and Fletcher, eds., 1987) mentions only
“dialysis dementia,” a side effect of administration of aluminum to counteract hyperphosphatemia in
chronic renal failure. Aluminum dissolved in natural waters is distinctly bad for fish, though.
8. Mainly materials such as muscle and collagen in animals, the last a major component of tendon, skin,
bone, and cartilage; also, cellulose in plants, the main constituent of wood.
9. See Wayman (1989a, b),Tylecote (1992), and Schmidt (1996).
10. In the trade, materials that give straight plots of stress versus strain are spoken of as Hookean, after
Robert Hooke (1635–1703) who stated what’s now known as Hooke’s law, that the amount of
deformation follows the deforming force, that strain is proportional to stress. Calling it a law smacks of
hyperbole, since it does little more than describe how some, but not all, materials behave.
11. Currey (1984). This is a lovely integration of biology and mechanics, sophisticated without being
 formidably technical.
12. It’s worth reminding ourselves that what I’m calling micromaintenance is unusual even in living
systems for large, stiff, structural elements. Bone is intracellular, by contrast with hair, arthropod
exoskeletal material, mollusk shell, and wood. The latter are built once and for all, with only very
limited reabsorption and replacement.
13. Hodges (1970) has a short but insightful account of how the use of metals may have begun. A more
elaborate analysis that focuses on the archaeological evidence is that of Singer et al. (1954).
14. Gordon (1976).
15. Note for the purists: Plexiglas is a trade name for a certain plastic mainly of polymethylmethacrylate
that has no common generic name. Perspex and Lucite are the same plastic from other manufacturers.
By contrast, “fiberglass” is a generic, with a double s and no capitalization.
16. Singer et al. (1954).
17. I’m following the account given by Gordon (1976), who was a major player in the crack game.
18. Wainwright et al. (1976). Vincent (1990) is a newer source, which, while less general, is especially
good on biological foams and composites.
19. “Conductivity” refers to a property of a material, while “conductance” refers to the property or
behavior of a particular object.
20. The advantage of metallic leaves has been inadvertently demonstrated in several studies of the thermal
behavior of leaves that used models inappropriately cut from sheets of metal.
21. I did some work a few years ago on thermal behavior and thermal conductivity of leaves. The upshot
of a paper on convective cooling at very low wind speeds (Vogel, 1970) and another on conductivity
(Vogel, 1984b) is that broad leaves either wouldn’t get so hot during lulls in the wind or could be less
constrained in shape and size—if leaves had the thermal properties of, say, copper.
22. But even that isn’t instantaneous. The largest and fastest computers are arranged to minimize the
lengths of wires connecting their components since conduction times constitute significant delays in
their operation.

CHAPTER 7: PULLING VERSUS PUSHING

1. A recent book (Barber, 1994) gives a fine introduction to both the technology and the social context of
spinning as well as to the etymology of the words, such as “life-span,” that we derive from “spin.”
2. Salvadori (1980) makes much of this point in a most enlightening book.
3. But an analog of a crack does occur in compression. Wooden beams such as diving boards accumulate
compression creases on their undersides. As Vincent (1990) cautions, subjecting such creases to tensile
loads is an exceedingly bad idea. Don’t turn an old wooden diving board upside down to expose an
unworn surface.
4. Several good sources on the mechanics of cathedrals are available—for instance, Mark (1978 and
1982).
5. Bronowski (1973) makes quite a point of this empiricism in cathedral construction.
6. The data for these comparisons are from Wainwright et al. (1976), Currey (1984), Vincent (1990), and
Niklas (1992). See note 3, above.
7. French (1994) has a good discussion of the matter. For considerations of size and scale, going back to
the second chapter of D’Arcy Thompson’s classic work (1942) is certainly worthwhile. For bridges in
particular, I enthusiastically recommend Billington (1983).
8. Isaacs et al. (1966) suggest the possibility.
9. Niklas (1994).
10. I give more data and a more detailed account in Life’s Devices (1988); the best source on scaling in
 mammals is Schmidt-Nielsen (1984). For aquatic mammals such as whales (and for fish), skeletal mass
is very nearly proportional to body mass, as it ought to be where the most severe loading is provided by
muscle contraction rather than gravity. If size is doubled, then both resistance to buckling and muscle
cross-sectional area (which muscle force follows) increase fourfold.
11. See, on brachiation, Hallgrimsson and Swartz (1995) and the references in that paper.
12. Salvadori (1980) gives a good account of Brunelleschi’s achievement.
13. More complete accounts and further references are given by Petroski (1985, 1991, and 1995).
14. Tendons usually connect muscles to bones; ligaments usually connect bones to bones.
15. Laithwaite (1984) makes quite a point of nature’s invention of the underslung suspension.
16. The wishbone of a bird bends when the bird flaps its wings up and down, according to Jenkins et al.
(1988). Springs can work by twisting as well, as in torsion bar automobile suspensions. I know of no
examples in nature unless one includes such structures as the trunks and branches of trees and the wing
feathers of birds that are loaded torsionally and have reasonable resilience. No relatively massive
components, though, are attached to their outboard ends, so it’s arguable whether they’re springs in the
present sense.
17. Vines of course are normally loaded in tension. But trees aren’t ordinarily held upright by the vines
they harbor. The large but thin buttresses of many tropical trees do seem to be tensile guys; they’re thus
the exact opposite of the buttresses (flying and otherwise) of Gothic cathedrals. On the buttresses of
trees, see Mattheck (1991) or Ennos (1993). Of course plenty of material is loaded in tension wherever
a trunk forks or a branch diverges; on this matter, see Mattheck (1991).
18. Bath sponges are unusual in that they lack spicules and have only the tension-resisting stuff spongin.
They’re not very large, and a mesh- work of spongin does have some compressive stiffness.
19. See, in particular, U.S. number 3,063,521, November 13, 1962 (Fuller, R. B.; Tensile-Integrity
Structures).
20. The best paper I know on spicular skeletons in general is Koehl (1982). At this point good mechanical
analyses of the supportive system of sponges don’t seem to exist.
21. Liquids care about shear, but what matters isn’t how far they’re sheared but how fast; they resist only
the rate of shear, and the coffee adjusts its shape to fit any cup fairly quickly. The resistance of liquids
to tension will come up again; it’s important in getting the sap up every tall tree, but our technology
finds the property impossible to employ. Another way to put the distinction, incidentally, is to
recognize that solids have shape and size, liquids have only size, and gases have neither shape nor size.
22. Notice how balloons keep arising here. My colleague Stephen Wainwright (with hydrostatic tongue
only slightly in membranous cheek) says that if you can’t demonstrate it with a balloon, it’s probably
not important anyway.
23. French (1994) gives these examples and a short treatment of the technology involved.
24. General descriptions of hydrostats in nature are given by Wainwright et al. (1976), Alexander (1983),
and Vogel (1988). Bending in hydrostats is analyzed by Alexander (1988).
25. See Smith and Kier (1989). Note the photograph of our late cat Fred with his tongue extended. To
make Fred perform it took two Vogels, one Kier, and an entire can of tuna.

CHAPTER 8: ENGINES FOR THE MECHANICAL

WORLDS
1. Various sources (perhaps copying one another) suggest the origin of windmills in seventh-century
Persia, although Tokaty (1971) mentions earlier cases. Derry and Williams (1960) note that more than
six thousand water mills for grinding grain were reported in the Domesday Book of England in 1086.
As Cardwell (1995) points out, technological innovation in Europe during the Middle Ages was much
 more impressive than during the Roman Empire.
2. The proportion is largely independent of the size of the mammal, which means that the edible fraction
doesn’t depend on how big an animal is. See Schmidt-Nielsen (1984).
3. Data for cilia, muscle, electric motor, and automobile engine from Nicklas (1984); for motorcycle,
from McMahon and Bonner (1983); for aircraft, from French (1994). The figure for the Newcomen
engine is my guess based on its 5.5-horsepower output (Derry and Williams, 1960) and an assumed
weight of half a ton. The output, at least, is reasonably reliable since the engine was employed lifting
water—in energetic terms a well-defined task.
4. A very fine exhibit of working models of early steam engines may be seen at the Science Museum in
South Kensington, London.
5. A good source on heat engines is Atkins (1984). This limitation on efficiency emerges from what’s
called the second law of thermodynamics.
6. 100 + 273 = 373; 1000 + 273 = 1273; 100(1 - 373/1273) = 71.
7. This isn’t a straw man or rhetorical device. See, for instance, Paturi (1976).
8. These are an electric catfish, the so-called electric eel, and electric rays; the trick has clearly evolved
several times. Some other freshwater fish generate low-strength electric fields that serve sensory
functions; see Denny (1993).
9. See Basalla (1988) for a description, a figure, and a good argument for the real rarity of technological
discontinuity. Laithwaite (1989) traces the development of linear electric motors.
10. Datum calculated from Edsall and Wyman (1958).
11. The ability to use transformers for easy and efficient conversion was the original reason for switching
to alternating current (60 Hz in North America and 50 Hz in most other places) from the direct current
used by Edison and others. Transformers don’t work with direct current.
12. Intermediate storage is one place where nature really eclipses anything in common use in our electrical
technology. We use storage of a form of carbohydrate, glycogen, in the liver for short-term use
(minutes to days) and storage as fat elsewhere for longer-term use. Analogous arrangements (such as
roots and tubers in plants) are nearly universal. Demand for electricity and for muscle power vary
similarly during the day; the power companies mainly accommodate the variation by adjusting the
output level of their fossil-fueled generators. (Nuclear generators, by contrast, are more economically
run at high and constant output since the plants are relatively more costly and the fuel is cheaper.)
Intermediate-term storage takes such awkward devices as pumped storage plants, in which water is
pumped to an uphill reservoir when power demand is low and released through generators when
demand increases.
13. Usher (1954) gives a good introduction to the subject. An experimental analysis from the days when
waterwheels were major engines (and a paper still worth reading) is that of John Smeaton (1759).
14. An exception, one using rotating cylinders and the same Magnus effect that makes spinning balls take
curved paths, was built by Anton Flettner in the 1920s and is described by Tokaty (1971).
15. See Tokaty (1971) for an account of various versions of Sigurd Savonius’s design, along with some
performance data.
16. I’ve given proper attention to this distinction between drag-based and lift-based thrust production in
connection with different modes of swimming in animals (Vogel, 1994a, pp. 154–55, and 283–87).
17. I’m avoiding proper explanation of why flow over the higher opening is more rapid as well as mention
of a secondary physical mechanism for driving the flow, one based on the viscous stickiness of real
fluids.
18. Dynamic soaring, it’s called. See, for instance, Vogel (1994a).
19. Quite a lot more on both mechanisms and cases can be found in Vogel (1978, 1994a). The original
suggestion was made by Vogel and Bretz (1972).
20. The flagella of bacteria, different things altogether, will get proper attention in the next chapter.
21. A good introduction to the mechanics of motility in animals is given by Schmidt-Nielsen (1990);
McMahon (1984) provides more information on muscle in particular.
22. The physical scientist will quickly protest that “lift” by suction to greater than ten meters isn’t possible
since what’s really doing the work is the atmosphere pushing from the bottom, and the maximum
pressure difference of one atmosphere can raise water only that high. The rejoinder to this objection is
that in tall plants the suction is real and that when properly contained in thin hydrophilic pipes, water
can withstand enormous tension. Zimmermann (1983) has a good account of our present understanding
of the process.
23. Nijhout and Sheffield (1979).
24. The figures come from a variety of sources. For human-built engines, I used Salisbury (1950), Croft
(1981), and Atkins (1984); for muscle, Heglund and Cavagna (1985).

CHAPTER 9: PUTTING ENGINES TO WORK

1. I’m avoiding the conventional classification into first-, second-, and third-class levers to which the
reader may have been subjected. It’s a structural rather than a functional scheme and proved confusing
and irrelevant to readers of the first version of this chapter. First-class levers, in case you care, have
their fulcra in the middle; second and third classes have fulcra at the ends. First-class levers may be
either force or distance amplifiers; second-class ones are always force amplifiers; third-class ones are
always distance or speed amplifiers. A first-class lever in which effort and load are equidistant from
the fulcrum is the exception; it changes only the direction of the force.
2. On rare occasions a crank is geared up to spin something rapidly—hand-operated centrifuges and
grinding wheels, for instance. Less drastically but more commonly, bicycle tires move faster than the
pedals except in the lowest gears.
3. The lower end of the humerus, no joke, is the “funny bone,” with no etymological pun whatsoever.
Distance advantages are 5.4 and 22, respectively (Currey 1984).
4. Alexander (1983) does a short and effective job on jaw mechanics and the topics that immediately
follow here.
5. The fibers are quite as easy to see after a few minutes’ immersion in boiling water to kill the creature
humanely. Leave the final, clawed segment intact, and work on the penultimate one. After looking at
and then consuming the muscle, you may be able to expose a pair of flat straps (properly called
apodemes rather than tendons since these are arthropods). The larger (pull on it) closes the claw while
the smaller opens it.
6. Smith and Kier (1989).
7. Vogel (1992) puts the calculation in context.
8. Berg and Anderson (1973) and Berg (1974). The figures in the next paragraph are from the latter
paper, to which my attention was drawn by R. Bruce Nicklas.
9. Mitochondria, the main sites of aerobic energy production in most cells, and chloroplasts, the sites of
the photosynthetic machinery of plant cells, appear to have originated as symbiotic bacteria. Another
case, the use of bacteria as locomotory organelles, is described in Chapter 2. For both, see Margulis
(1993) or Dyer and Obar (1994).
10. Gould (1981). “Wheels are not flawed as modes of transport; I’m sure that many animals would be far
better with them.”
11. LaBarbera (1983) and Basalla (1988). I’m not going to sort out their separate arguments.
12. Ekholm (1946). I don’t take seriously suggestions that these toys must have been copies of real
devices, now lost, since the wheels are on the ends of animal legs rather than on wagons.
13. On this last point, see Bulliet (1975).
14. These figures are all from a source I’d not be without, Machinery’s Handbook (Olberg et al., eds.,
1984).
15. On bearings in human technology, see French (1994); on those between bones, see Currey (1984).
16. Watt’s device incorporated an orbital gear largely as a way to evade infringing on a recent patent, so it
seems to us excessively complicated.
17. Sleeswyk (1981) talks about the general problem of coupling humans to machines in the context of
Egyptian antiquity and is my source for guided and misguided cranks.
18. A good source for this topic as well as for a lot of other matters relevant to this book is a two-volume
set by Amerongen (1967).
19. These transmissions use a version of a fluid coupling called a torque converter, which adds a little
complexity to this basic fluid coupling to get a trade-off of speed and force. And to improve fuel
economy, the input and output shafts may actually clamp together at high speeds to prevent any slip at
all. In addition, gearing is provided (hydraulically shifted!) to increase the car’s usable range of speeds.
20. Dunwell (1991) recounts the controversy, with the appropriate maps and pictures.
21. See Alexander (1984b or 1988) or (secondhand) Vogel (1988) on this topic and those that follow. If
you try the calculation, remember to keep your units consistent-for instance, height in feet, acceleration
in feet per second squared, and speed in feet per second.
22. Sotavalta (1953).

CHAPTER 10: ABOUT PUMPS, JETS, AND SHIPS

1. A more elaborate version of the present section appears in Vogel (1994b).
2. Miller (1945).
3. But some doubt. Martin Canny (1995) has cast an informed and skeptical eye on the evidence for the
extreme negative pressures in trees and suggests an alternative explanation for the rise of sap.
4. Zimmermann (1983) gives a good review of the system. The record for pull, 120 atmospheres, was
measured by a colleague, William Schlesinger (1982).
5. For example, no less than 22.4 atmospheres are needed to offset the propensity of a one-molar solution
of a nonelectrolyte to absorb pure water.
6. The underlying distinction between propellers and paddles or flukes and cilia, elaborated in Vogel
(1994a), is between lift-based thrust producers and drag-based thrust producers.
7. LaBarbera and Vogel (1982).
8. We do have an exception. A field called exobiology considers extrater-restrial life; on occasion one or
another of its devotees has proudly proclaimed this distinction and the uniqueness of a field that
doesn’t know if it has a subject matter.
9. Some of the material in this section appeared as Vogel (1994c).
10. A crude steam rocket made a fine toy for bathtub or pond back when small metal containers were
ubiquitous as 35 mm film canisters, tooth powder cans, and the like. One mounted a candle in the hull
of an open boat, with a partly filled can of water above. A pinhole in the rearward-facing top of the can
let steam escape to propel the boat forward. One shudders when thinking of its efficiency, much less of
its safety. A device distantly related to Hero’s engine did see service in the late nineteenth century: the
De Laval steam turbine, used to run high-speed electrical generators.
11. We’re assuming that the working fluid of the jet is of about the same density as the surrounding fluid.
A rocket in space isn’t concerned about the speed at which it passes through nothing at all.
12. Except, perhaps, for very tiny insects for which covering distance is more a matter of ambient wind
than their own navigation.
13. The life and times of both of these are the subjects of engaging popular accounts, Crouch (1989) and
Grosser (1981) respectively.
14. But oversimplifying is all too easy. The inboard part of a flapping wing goes up and down very little,
 so it does work very nearly as a fixed wing.
15. The motion that matters isn’t really the lateral progress of the wave—very little actual wave-wise
transport of water happens-but the orbital motion of the water immediately beneath the wave. See
Denny (1993), Vogel (1994a), or any introduction to oceanography.
16. Full support by surface tension is quite another subject from the speed limit set by waves. Whirligig
beetles are supported in part by surface tension and in part by their buoyancy, while water striders are
almost entirely supported by surface tension. Whirligigs are what matter here; water striders move in a
specialized way, one in which surface waves are less directly involved.

CHAPTER 11: MAKING WIDGETS

1. See, for instance, Schopf’s own contribution in Schopf (1992).
2. Vincent (1990). Of course I compound the sin by repetition.
3. Such multicellularity was in fact achieved in several separate lineages, and traces of these separate
origins exist in the different chemical and physical devices that present multicellular lineages use to
keep themselves operative.
4. See Chapter 3, note 10.
5. These are all cases of negative feedback: Action is taken to decrease the difference between the actual
state and some desired state. Positive feedback has another role altogether and isn’t immediately
relevant to the present discussion.
6. Mayr (1970). He raises such fascinating questions as why float valves were not used between the
twelfth and eighteenth centuries.
7. My colleague Fred Nijhout (1990) argues, in an elegant and persuasive essay, that we—ve been poorly
served by the metaphor of gene as unique and complete causative agent. Part of his argument is the
basis of the next paragraph.
8. Electronic equipment that operates with very high precision is ordinarily made from far less precise
components by using a variety of self-correcting schemes, most of which fit comfortably within the
present invocation of feedback.
9. Morowitz (1968) gives figures for the percentage of an average protein that will spontaneously spoil
(denature) per day as a function of temperature. In Celsius degrees, they are 1.1 percent at thirty-seven
degrees; 4.4 percent at forty degrees; 13.8 percent at forty-three degrees; 46.2 percent at forty-six
degrees; and 161 percent at forty- nine degrees. It does look as if mammals and birds, with body
temperatures around thirty-seven degrees to forty degrees, hug an upper practical limit for use of
ordinary protein.
10. The relative diameters of all the blood vessels in a vertebrate are constructed and, when necessary,
readjusted to minimize the operating cost of its circulatory system. Astonishingly, no overall
coordination is required or involved; each of the cells lining the vessels needs only to respond to
changes in the shear stress of the blood flowing past it. For details, see LaBarbera (1990) or Vogel
(1994a).
11. A reasonable introduction to a fairly diffuse literature on how trees respond to environmental forces is
a recent volume edited by Coutts and Grace (1995).
12. The doctrine was at its zenith in Britain around the turn of the century; it used the slogan “survival of
the fittest” to justify all manner of social conservatism, unrestrained capitalism, class stratification,
racial discrimination, and imperialism. Considerable opposition to evolution by natural selection has
been generated by this combination of an inappropriate analogy and the notion that a social system
based on natural selection has some natural and intrinsic merit. So in parallel with fundamentalist
opposition to evolution, we have left-wing opposition based on rejection of the predestination and
 social Darwinism often thought to be part and parcel of the biological world view.
13. The famous paper here, into which more is often read than a literal look suggests was intended, is
Gould and Lewontin (1979).
14. See the literature on symmorphosis, a term coined by Taylor and Weibel (1981); typical criticisms will
be found in Garland and Huey (1987) and Dudley and Gans (1991).
15. This isn’t to deny that sometimes single substitutions make dramatic differences. The sensitivity of the
product to such alterations isn’t at all uniform.
16. Two good references on safety factors in organisms are Alexander (1981) and Currey (1984). I sum up
their arguments in Vogel (1988).
17. Niklas (1992) sums up what information is available.
18. As with most historical accounts, complexities lurk beneath the surface. Thus Whitney may have been
one of the first with the basic idea of interchangeable parts, but full achievement was decades away
(see Woodbury, 1960). Interesting general (if short) sources are Derry and Williams (1960), Boorstin
(1965), and Reynolds (1991). Fridenson (1978) describes the transfer of this American system to
Europe.
19. During World War II Liberty ships, cheap and expendable freighters, were truly mass-produced, not
just made in quantity. But the case is exceptional.
20. About twenty-five years ago I took the standard tour of the local cigarette factory. The tour guide
halved a cigarette and tossed it back into the machinery; a few seconds later a pack was ejected some
distance away that contained 19.5 rather than 20.0 items.
21. Norbert Wiener’s now rather ancient book (1950) on the interrelationship between humans and
feedback devices, the field he named cybernetics, is still worth reading.
22. J. E. Gordon (1978) regards them as essentially factors of ignorance forming part of the theology of
design.

CHAPTER 12: COPYING, IN RETROSPECT

1. I’m indebted to Francis Newton, a neighbor and colleague in classical studies, for bringing in Ovid.
The quotations are from Book VIII, lines 183–259. Note, in the first, that the wings were curved; as
we’ll see, the crucial role of wing camber was not otherwise appreciated until the 1880s. The ancients,
especially the Greeks, took a dim view of human creativity; everything was given by the gods, stolen
from the gods, or copied from nature. Thus, according to Democritus of Abdera (ca. 420 B.C.E.), “We
are pupils of the animals in the most important things: the spider for spinning and mending, the
swallow for building, and the songsters, swan and nightingale, for singing, by way of imitation”
(Freeman, 1948). Poseidonius (135–51 B.C.E.) viewed millstones as copied from teeth (Cole, 1967).
2. E. O. Wilson (1984). He defines it as “the innate tendency to focus on life and lifelike processes.”
3. Quoted in Gérardin (1968) and elsewhere. Other books specifically about bionics include Bernard and
Kare (1962), Halacy (1965), and Marteka (1965). Winfield et al. (1991) have done an annotated
bibliography covering a broad sweep of bionics and biomimetics; more recent material is available by
searching under these headings in the relevant engineering and biological databases (NTIS, BIOSIS,
etc.)
4. “Biomimesis” was defined by Warren McCulloch in 1962 as all areas in which one organism copies
another; thus mimicry of distasteful insects by innocuous insects would be included. But that usage
seems not to have caught on.
5. Several disclaimers: Potentially profitable bits of technology are rarely reported in peer-reviewed
journals; more commonly they form the bases for patents, with laws and lawyers deliberately
compromising any semblance of historical verisimilitude. A patent tries to impress upon the world just
 how original is the invention, giving as little benefit as possible to any notion of continuity; it fully
intends to be heroic literature. That nature might have done the trick first is not usually something to be
forthrightly admitted, and the idea that the invention just copied nature is certainly best kept private.
As Basalla (1988) puts it (p. 60), “All of patent law is based on the assumption that an invention is a
discrete, novel entity that can be assigned to the individual who is determined by the courts to be its
legitimate creator.” For ancient and prehistoric possibilities, the chance of documentation is even less.
6. From Smeaton’s A Narrative of the Building and a Description of the Construction of the Edystone
Lighthouse with Stone, abridged by T. Williams (1882).
7. In Lighthouses, A Rudimentary Treatise, quoted by Majdalany (1960).
8. A wealth of history of the tunnel is provided by two biographies of the more famous Isambard
Kingdom Brunel, who as a young man supervised the work for his father. See Rolt (1959) and
Vaughan (1991).
9. Beamish (1862), p. 207. But the story must be older, since Ewbank (1842), p. 258, alludes to it, as he
does to Smeaton’s tree trunk.
10. Miller (1924).
11. Paxton and the Crystal Palace have not suffered obscurity; see Chadwick (1961), Beaver (1970), or
Kihlstadt (1984).
12. “[A] gardener as well known as his patron, the Duke of Devonshire,” according to Hix (1974), p. 50.
13. No. 13,186, dated July 22, 1850, enrolled January 22, 1851; in “Alphabetical Index of Patentees of
Inventions; Building Materials 1850–51, Vol. 184.”
14. Hertel (1963), Paturi (1976), Tributsch (1982), and Laithwaite (1994) all comment on the copying.
15. Illustrated London News, Supplement 17: 317–24 (October 19, 1850).
16. Forget about low-drag cars, whose shape is severely compromised by the requirement that they work
near the ground and not produce lift. And as noted just ahead, forget surface ships, whose drag comes
from surface waves.
17. Quoted by von Kármán (1954), in part, and more fully by Pritchard (1961).
18. Quoted by Gibbs-Smith (1962).
19. Phillips (1885); also reported by Chanute (1893).
20. The book was first published in 1889 as Der Vogelflug als Grundlage der Fliegekunst; a second edition
was prepared by his brother Gustav and published in 1910. That second edition was translated a year
later into English, under the title in the text. A short account, translated, is given as an appendix in
Chanute (1893).
21. A brief account of the difficult birth of airfoil theory is given by Giacomelli and Pistolesi (1934). They
also provide a glossary that’s needed to read Lanchester’s work; the latter coined and used a host of
wonderful Greek-and Latin-based terms that never came into common usage. Von Kármán (1954),
Sutton (1949), and Vogel (1994a) describe in nonmathematical terms how lift originates without
relying on the usual polite fiction. In the latter, bits of fluid diverge in front of the wing and then rejoin
at the rear; since the path across the top is longer, those half bits have to go faster; by Bernoulli (that at
least is okay), faster means lower pressure, so the wing is forced upward. The trouble with the
explanation is that no rule requires that the bits of fluid reassemble with their same partners at the rear,
something crucial to its logic.
22. At least that’s my best guess as to the trouble. My first attempt to calibrate an electronic anemometer
I’d built foundered on this particular rock. Von Kármán (1954) notes that the problem corrupted a lot of
nineteenth-century data.
23. Wilbur Wright to Octave Chanute, May 13, 1900, in McFarland (1953). The famous first powered
flight came in December 1903.
24. Recent biographies of the Wrights are those of Howard (1987) and Crouch (1989). See also Wright
(1953).
25. I’m ignoring old Chinese and Japanese paper craft as out of the historical sequence that led to present
mass-production technology, whatever its intrinsic interest or claims of priority.
26. Quoted in translation by Hunter (1947). Hunter is the best overall source I’ve seen, but Ainsworth
(1959) and Schlosser (1980) are also useful. Flatow (1992) gives a quick account without
documentation. About the wasps, see Hansell (1989).
27. Koops (1800). I was able to examine an original of Koops’s book; two centuries have treated it kindly,
with no deterioration obvious to my admittedly untutored eye. The slightly yellow-beige paper has a
rough surface, but the print is clear and dark, with no capillary “wicking” of the ink.
28. Hooke (1665), with modern spelling.
29. He demonstrated the process for the annual meeting of the British Association for the Advancement of
Science at its Manchester meeting in 1842; the event is noted in the report of the meeting.
30. Ozanam (1862). The occasional attribution of the invention of the spinneret to Ozanam on the basis of
these few vague and hopeful words (without the name) must be taken as an exaggeration. Schwabe’s
claim is both earlier and better.
31. Sources here are Avram (1927) and Leeming (1949). The quotation is from Chardonnet’s entry in the
Dictionary of Scientific Biography.
32. The Oxford English Dictionary traces it back to Kirby and Spence’s Entomology of 1825.
33. From Worden (1911).
34. Schober (1930).
35. From Bell’s lecture of October 31, 1877, to the Society of Telegraph Engineers, in London; quoted in
Prescott (1884). A good account is that of Bruce (1973); M. Gorman and W. B. Carlson, of the
University of Virginia, are making available Bell’s notebooks, currently as
http://jefferson.village.virginia.edu/-meg3c/id/AGB/index.html. John Wourms, of Clemson University,
called this case to my attention.
36. No. 174,465; March 7, 1876.
37. For instance, the prickly-pear cactus, a New World plant (as are almost all cacti), was imported into the
Middle East for hedgerow use about three hundred years ago. It’s now a nuisance.
38. A very nice encapsulation of this story is given by Basalla (1988); a more elaborate recounting is that
of the McCallums (1965). According to the former, in 1860 alone enough seed was sent north to
produce sixty thousand miles of hedge.
39. The Kelly patent is no. 74,379 (1868); the Glidden patent (commercially the most important one) is no.
157,124 (1874).
40. A crosscut timber saw with well-set and sharp teeth cuts impressively. “Cutters” splay outward very
slightly and make slits that define the outside of the kerf, while “rakers” work like chisels and shear off
the wood between those slits. With strong and determined sawyers, each pull of the saw may deepen
the kerf by more than half an inch.
41. The account is mainly that of Lucia (1975 and 1981 contain essentially the same material) together
with entomological information from Craighead (1915, 1923). I’m indebted to Professor George
Pearsall for calling my attention to the case.
42. A short history of Velcro is given by Flatow (1992) in an engaging, if poorly documented, book on
inventions. See also Budde (1995).
43. A good history of misguided people and misbegotten craft is that of Hart (1985).
44. Maxim (1909). His huge steam-engined biplane self-destructed after an expenditure of twenty
thousand pounds.
45. See, on this point, J. S. Harris (1989).
46. Quoted by von Kármán (1954).
47. Maynard Smith (1952)-who did aerodynamics before being lured into biology.
48. The basic account here is from Hertel (1963) and Paturi (1976), with the aerodynamics largely
confirmed by Bishop (1961). The plant was known to the Etrichs as Zanonia macrocarpa but is now in
the genus Alsomitra and hence A. macrocarpa. Good aerodynamic data on the fruit are now available,
in Azuma and Okuno (1987). In fact, it’s not an especially good wing, descending with a lift-to-drag
ratio of 3.5. For comparison, the hind wing of a locust, which operates under comparable conditions,
 manages a ratio of 8.2.
49. Bernheim (1959).
50. A particularly good introduction to the history of stability in aircraft is that of Vincenti (1988). Von
Kármán (1954) gives one a good feel for the various controls themselves.
51. See Boyd (1935) and C. M. Harris (1989). One wonders why neither Rumsey nor Fitch used a paddle
wheel; the device is nothing more than the then familiar undershot waterwheel used in reverse. Perhaps
Watt’s crankshaft for converting linear to rotary motion was insufficiently appreciated.

CHAPTER 13: COPYING, PRESENT AND

PROSPECTIVE
1. British patent 3587/1894. From Wilson (1975).
2. The relevant references are Gray (1936), Kramer (1965), and Riley et al. (1988).
3. The earliest report seems to be Toms (1948). See also Davies and Porter (1966), Wells (1969), and
Bushnell and Moore (1991).
4. See Reif (1985), Choi (1990), and Bushnell and Moore (1991).
5. sci.aeronautics, sci.engr, sci.mech.engr, sci.engr.marine.hydrodynamics, sci.mech.fluids, and
soc.history.science. I deliberately biased the selection toward news groups frequented by engineers,
and I received a number of very thoughtful and interesting responses. I thank all the people who
responded and hope that I managed to reply to each.
6. Two general sources are Winfield et al. (1991) and Sarikaya and Aksay (1995). Quite a lot of
information emerges if one searches databases such as that of the National Technical Information
Service (NTIS) or Biological Abstracts (BIOSIS) using “bionics” or “biomimetics” as keywords.
7. See Douglas and Clark (1990), Lombardi et al. (1990), Mann (1990, 1995), and Fournier et al. (1995).
8. See De Rossi et al. (1985), Pool (1989), Caldwell and Taylor (1990), Urry (1993), and Salehpoor
(1996).
9. See Gunderson and Schiavone (1989), Berman (1990), Mann (1990), and Sarikaya et al. (1994). The
patent (United States, 1983) is to C. R. Chaplin, J. E. Gordon, and G. Jeronimidis, no. 4,409,274,
entitled “Composite Material.”
10. This is somewhat further from practical realization, but it has been the subject of much speculation and
some experimentation. See Winfield et al. (1991).
11. See Mason and Salisbury (1985), Hayward (1993), Wilson et al. (1993), and Wainwright (1995).
12. See Raibert and Sutherland (1983), Song and Waldron (1989), and Manko (1992).
13. Stix (1994). The principals are Robert Barrett and Michael Triantafyllou. The principles are prefigured
in Triantafyllou et al. (1993).
14. The twiddlefish is the invention of Charles Pell and Stephen Wainwright, working at the Biodesign
Studio, at Duke. See McHenry et al. (1995).
15. Blake et al. (1995) have shown that tuna can’t turn especially sharply and suggest that their efficiency
in steady swimming involves some loss of maneuverability.
16. My main sources on spider silks are Gosline et al. (1986 and 1995).
17. Chatterton (1910) and Rouse and Ince (1957).
18. Maybe naval design is naturally conservative. Needham (1965) notes the cambered surfaces of ancient
Chinese kites. The matter is confused by the way sails automatically camber in a wind. See also
Smeaton (1759).
 CHAPTER 14: CONTRASTS, CONVERGENCES,
AND CONSEQUENCES
1. It’s now late enough in the game to admit something that has been kept sub rosa. Other mechanical
technologies do exist-in particular those of a variety of social organisms, such as bees, ants, termites,
and beaver. Each seems (adjusting for such things as scale) largely a compromise between the two
main ones under scrutiny here, so they have only minimal relevance to the present discussion.
2. Stable systems are more persistent than unstable systems. A system in which destructive interactions
can arise is more likely to disappear than one with self-improving interactions. This argument for
large- scale integration beyond what natural selection per se can produce was made by Pantin (1964).
It’s crucial to the Gaia hypothesis popularized by Lovelock and Margulis (1974).
3. The phenomenon is most often known as kin selection. It rationalizes things ranging from parent-child
altruism to the repeated evolution of complex societies in the genetically peculiar ants, bees, and wasps
(Hymenoptera). See, for instance, Dawkins (1976) or Hölldobler and Wilson (1994).
4. We use the term “mutualism” to make a needed distinction from symbioses (living together) in which
mutual and reciprocal benefit is absent—parasitism, for instance.
5. Basalla (1988). See especially pp. 91–92. Derek deSolla Price has noted the same thing on various
occasions.
6. The classic statement of the notion of “punctuated equilibria” in evolution—the extreme of episodic
change—is that of Eldredge and Gould (1972). The idea pervades many of the essays of Gould (1980,
and other collections from Natural History magazine). Less protagonistic statements can be found in
contemporary evolution textbooks such as Futuyma (1986).
7. For a more extensive but admirably accessible treatment, see Wilson (1992).
8. Gould (1989). A somewhat less opinionated view of the Ediacara is given by Runnegar (1992).
9. Petroski (1992) gives lots of examples involving some of the simplest of our industrial artifacts, such
as paper clips, forks, and hand tools.
10. See, for instance, Crosby (1986) or Vitousek et al. (1996).
11. See Chapter 2, note 1.
12. Dasgupta (1996) gives a lot of attention to this issue. Unless of course one includes the few odd cases
of acquired tool use and other minor elements of acquired behavior in animals.
13. Heilbroner (1967), an important and influential essay, reprinted in Smith and Marx (1995).
14. The main reference here is White (1962).
15. Dasgupta (1996) emphasizes this point.
16. Gilinsky and Rambach (1987); the idea was followed up by Rosenzweig and McCord (1991).
17. The usual citation for the argument is David (1985). It’s put in a larger context in Rogers (1983).
18. Liebowitz and Margolis (1990).
19. Liebowitz and Margolis (1995).
20. See, for a fuller argument, Arthur (1990). I’m not sure that his analogy of positive feedback in
economics with punctuated equilibrium in evolution is close enough to be of use.
21. One can get a good look at the importance of tapers—that is, conical surfaces—in contemporary
technology by glancing at handbooks such as Oberg et al. (1984).
22. A lovely recent book (Maor, 1994) puts the logarithmic spiral into both mathematical and historical
context.
23. See Pettigrew (1908), Cook (1914), and Thompson (1942, originally 1917). A general and realistic
view of alternative geometries is given by Pearce (1978).
24. Needham (1954), vol. 1, p. 229. At every possible occasion he compares Western and Chinese
technological achievements.
25. Tylecote (1992) is my main reference here; further information on Andean metallurgy is given by
 Lechtman (1988) and on the Old World by Hodges (1970). Native copper, incidentally, was mined for
about five thousand years in the Lake Superior/Isle Royale area of North America (Wayman, 1989a),
and meteoritic iron was hammered out into teeth for blades by prehistoric people in the Cape York area
of northern Greenland (Wayman, 1989b). Metallic technologies have arisen on quite a few occasions!
26. An extreme example of diffusionist argument, one relevant to the present case, is that of Needham and
Lu (1985).
27. My main sources here are Bird (1979), King (1979), Anton (1984), and Barber (1994).
28. The approach I’m suggesting isn’t really all that different from the analysis of weapons technology by
Churchill (1993) and probably other anthropological work.
29. The example of Bell, Elisha Gray, and the invention of the telephone is analyzed by Hounshell (1975)
and Gorman and Carlson (1990). Many other cases can be cited. Molecular biology, for instance, was
in large measure a creation of recruits from physics-part of the reason for using a name other than the
already established and familiar biochemistry.
30. See, for some recent musings, Tenner (1996).
 REFERENCES

Adams, J. L. (1991) Flying Buttresses, Entropy, and O-Rings: The World of an Engineer. Cambridge, MA:
Harvard University Press.
Ainsworth, J. H. (1959) Paper: The Fifth Wonder. Kaukauna, WI: Thomas Printing and Publishing Co.
Alexander, R. M. (1981) Factors of safety in the structure of animals. Sci. Prog., Oxford 67: 109–30.
____. (1983) Animal Mechanics, 2d ed. Oxford, UK: Blackwell Scientific Publications.
____. (1984a) Elastic energy stores in running vertebrates. Amer. Zool. 24: 85–94.
____. (1984b) Walking and running. Amer. Sci. 72: 348–54.
____. (1988) Elastic Mechanisms in Animal Movement. Cambridge, UK: Cambridge University Press.
____. (1992) Exploring Biomechanics: Animals in Motion. New York: Scientific American Library.
Allport, G. W., and T. F. Pettigrew (1957) Cultural influence on the perception of movement: the trapezoidal
illusion among Zulus. J. Abnormal Soc. Psychol. 55: 104–13.
Amerongen, C. van (1967) The Way Things Work, vols. 1 and 2. New York: Simon and Schuster. (Originally
Wie Funktioniert Das, 1963.)
Ames, A. (1951) Visual perception and the rotating trapezoidal window. Psychol. Monogr. 65: 324.
Anderson, C. N. (1972) My Fertile Crescent; Travels in the Footsteps of Ancient Science. Fort Lauderdale,
FL: Sylvester Press.
Annis, R. C., and B. Frost (1973) Human visual ecology and orientation anisotropies in acuity. Science 182:
729–31.
Anton, F. (1984) Ancient Peruvian Textiles. New York: Thames and Hudson.
Arthur, W. B. (1990) Positive feedbacks in the economy. Sci. Amer. 262 (2): 92–99.
Atkins, P. W. (1984) The Second Law. New York: Scientific American Library.
Avram, M. H. (1927) The Rayon Industry. New York: D. Van Nostrand Co.
Azuma, A., and Y. Okuno (1987) Flight of a samara, Alsomitramacrocarpa. J. Theor. Biol. 129: 263–74.
Barber, E. W. (1994) Women’s Work: The First 20,000 Years. New York: W. W. Norton.
Barlow, G. W. (1974) Hexagonal territories. Anim. Behav. 22: 876–78.
Basalla, G. (1988) The Evolution of Technology. Cambridge, UK: Cambridge University Press.
Beamish, R. (1862) Memoir of the Life of Sir Marc Isambard Brunel. London: Longman, Green, Longman,
and Roberts.
Beaver, P. (1970) The Crystal Palace. London: Hugh Evelyn, Ltd.
Bennet-Clark, H. C. (1977) Scale effects in jumping animals. Pp. 185–201 in T. J. Pedley, ed., Scale Effects
in Animal Locomotion. London: Academic Press.
Berg, H. C. (1974) Dynamic properties of bacterial flagellar motors. Nature 249: 77–79.
____, and R. A. Anderson (1973) Bacteria swim by rotating their flagellar filaments. Nature 245: 380–82.
Berkow, R., and A. J. Fletcher (1987) The Merck Manual of Diagnosis and Therapy. Rahway, NJ: Merck
and Co.
Berman, A., L. Addadi, Å. Kvick, L. Leiserowitz, M. Nelson, and S. Weiner (1990) Intercalation of sea
urchin proteins in calcite: study of a crystalline composite material. Science 250: 664–67.
Bernard, E. E., and M. R. Kare (1962) Biological Prototypes and Synthetic Systems. New York: Plenum
Press.
Bernheim, M. (1959) A Sky of My Own. New York: Rinehart and Co.
Bijker, W. E. (1995) Of Bicycles, Bakelites, and Bulbs: Toward a Theory of Sociotechnological Change.
Cambridge, MA: MIT Press.
Billington, D. P. (1983) The Tower and the Bridge: The New Art of Structural Engineering. Princeton, NJ:
Princeton University Press.
Bird, J. B. (1979) Fibers and spinning procedures in the Andean area. Pp. 13–17 in A. P. Rowe, E. P.
Brown, and A. L. Schaffer, eds., Junius B. Bird Pre-Columbian Textile Conference. Washington, DC:
Textile Museum.
Bishop, W. (1961) The development of tailless aircraft and flying wings. J. Roy. Aero. Soc. 65: 799–806.
Blake, R. W., L. M. Chatters, and P. Domenici (1995) Turning radius of yellowfin tuna (Thunnus albacares)
in unsteady swimming maneuvers. J. Fish Biol. 46: 536–38.
Blakemore, R. (1975) Magnetotactic bacteria. Science 190: 377–79.
Boorstin, D. J. (1965) The Americans: The National Experience. New York: Random House.
Bose, D.M. (1971) A Concise History of Science in India. New Delhi: Indian National Science Academy.
Boyd, T. (1935) Poor John Fitch, Inventor of the Steamboat. New York: Putnam.
British Association for the Advancement of Science (1842) Notes and Abstracts of Communications to the
B.A.A.S. at the Manchester Meeting, June 1842. P. 114 in Rept. of Annu. Meet.
Bronowski, J. (1973) The Ascent of Man. Boston: Little, Brown and Co.
Bruce, R. V. (1973) Bell: Alexander Graham Bell and the Conquest of Solitude. Boston: Little, Brown and
Co.
Buckley, P. H., and F. G. Buckley (1977) Hexagonal packing of royal tern nests. Auk 94: 36–43.
Budde, R. (1995) The story of Velcro. Physics World 8(1): 22.
Bulliet, R. W. (1975) The Camel and the Wheel. Cambridge, MA: Harvard University Press.
Burke, J. G., and M. C. Eakin (1979) Technology and Change. San Francisco: Boyd and Fraser.
Burns, J. M. (1975) BioGraffiti: A Natural Selection. New York: Quadrangle/New York Times Book Co.
Bushnell, D. M., and K. J. Moore (1991) Drag reduction in nature. Annu. Rev. Fluid Mech. 23: 65–79.
Caldwell, D. G., and P. M. Taylor (1990) Chemically stimulated pseudomuscular actuation. Int. J.
Engineering Sci. 28: 797–808.
Canny, M. J. (1995) A new theory for the ascent of sap: cohesion supported by tissue pressure. Ann. Bot. 75:
343–57.
Cardwell, D. (1995) The Norton History of Technology. New York: W. W. Norton.
Chadwick, G. F. (1961) The Works of Joseph Paxton 1803–1865. London: Architectural Press.
Chan, M. K., S. Mukund, A. Kletzin, M. W. W. Adams, and D. C. Rees (1995) Structure of a
hyperthermophilic tungstopterin enzyme, aldehyde ferredoxin oxidoreductase. Science 267: 1463–69.
Channell, D. F. (1991) The Vital Machine. New York: Oxford University Press.
Chanute, O. (1893) Progress in flying machines. Amer. Engineer and Railroad J. 67 (3): 135.
Chatterton, E. K. (1910) Steamships and Their Story. London: Cassell and Co.
Choi, K. S. (1990) Drag-reduction test of riblets using ARE’s high speed buoyancy propelled vehicle -
MOBY-D. Aeronaut. J. 94: 79–85.
Churchill, S. E. (1993) Weapon technology, prey size selection, and hunting methods in modern hunter-
gatherers: implications for hunting in the Palaeolithic and Mesolithic. In G. L. Peterkin, H. M. Bricker,
and P. Mellars, eds., Hunting and Animal Exploitation in the Later Palaeolithic and Mesolithic
ofEurasia. Archeological Papers of the American Anthropological Association 4: 11–24.
Cleveland, L. R., and A. V. Grimstone (1964) The fine structure of the flagellate Mixotricha and its
associated microorganisms. Proc. Roy. Soc. Lond. 159B: 668–86.
Cole, A. T. (1967) Democritus and the Sources of Greek Anthropology. Cleveland, OH: American
Philological Association/Western Reserve University Press.
Cook, T. A. (1914) The Curves of Life. London: Constable and Co.
Coutts, M. P., and J. Grace, eds. (1995) Wind and Trees. Cambridge, UK: Cambridge University Press.
Craighead, F. C. (1915) Larvae of the Prioninae. U.S. Dept. Agr. Rept. 107.
____. (1923) North American Cerambycid larvae. Bull. Dept. Agr. Can. 27 (NS): 1–237.
Crane, H. R. (1950) Principles and problems of biological growth. Sci. Monthly 70: 376–89.
Croft, T. (1981) American Electrician’s Handbook, 10th ed., W. I. Summers, ed. New York: McGraw-Hill.
Crosby, A. W. (1986) Ecological Imperialism: The Biological Expansion of Europe, 900–1900. Cambridge,
UK: Cambridge University Press.
Crouch, T. D. (1989) The Bishop’s Boys: A Life of Wilbur and Orville Wright. New York: W.W. Norton.
Currey, J. (1984) The Mechanical Adaptations of Bones. Princeton, NJ: Princeton University Press.
Dasgupta, S. (1996) Technology and Creativity. New York: Oxford University Press.
David, P. A. (1985) Clio and the economics of QWERTY. Amer. Econ. Rev. 75: 332–37.
Davies, G. A., and A. B. Porter (1966) Turbulent flow properties of dilute polymer solutions. Nature 212:
66.
Dawkins, R. (1976) The Selfish Gene. Oxford, UK: Oxford University Press.
____. (1986) The Blind Watchmaker. Harlow, UK: Longmans Scientific and Technical Press.
Denny, M. W. (1993) Air and Water: The Biology and Physics of Life’s Media. Princeton, NJ: Princeton
University Press.
De Rossi, D., P. Parrini, P. Chiarelli, and G. Buzzigoli (1985) Electrically induced contractile phenomena in
charged polymer networks: preliminary study on the feasibility of muscle-like structures. Trans. Amer.
Soc. Artificial Internal Organs 31: 60–65.
Derry, T. K., and T. I. Williams (I960) A Short History of Technology: From the Earliest Times to A.D.
1900. New York: Dover Publications.
Douglas, K., and N. A. Clark (1990) Biomolecular/solid-state nanoheterostructures. Appl. Phys. Lett. (USA)
56: 692–94.
Dudley, R., and C. Gans (1991) A critique of symmorphosis and optimality models in physiology. Physiol.
Zool. 64: 627–37.
Dunwell, F. F. (1991) The Hudson River Highlands. New York: Columbia University Press.
Dyer, B. D., and R. A. Obar (1994) Tracing the History of Eukaryotic Cells. New York: Columbia
University Press.
Edsall, J. T., and J. Wyman (1958) Biophysical Chemistry. New York: Academic Press.
Ekholm, G. F. (1946) Wheeled toys in Mexico. Amer. Antiquity 11: 222–28.
Eldredge, N. and S. J. Gould (1972) Punctuated equilibrium: an alternative to phyletic gradualism. Pp. 82–
115 in T. J. M. Schopf, ed., Models in Paleobiology. San Francisco: Freeman, Cooper, and Co.
Ellul, J. (1964) The Technological Society. New York: Alfred A. Knopf.
Endler, J. A. (1986) Natural Selection in the Wild. Princeton, NJ: Princeton University Press.
Ennos, A. R. (1988) The importance of torsion in the design of insect wings. J. Exp. Biol. 140: 137–60.
____. (1993) The function and formation of buttresses. TREE 10: 350–51.
Ewbank, T. (1842) Hydraulic and Other Machines for Raising Water Including the Progressive
Development of the Steam Engine. London: Tilt and Bogue.
Fisher, J., and R. A. Hinde (1949) The opening of milk-bottles by birds. Brit. Birds 42: 347–57.
Flannery, K. V. (1972) The origins of the village as a settlement type in Mesoamerica and the Near East: a
comparative study. Pp. 23–53 in P. J. Ucko, R. Tringham, and G. W. Dimbleby, eds., Man, Settlement,
and Urbanism. Cambridge, MA: Schenkman Publishing Co.
Flatow, I. (1992) They All Laughed. New York: HarperCollins.
Florman, S. C. (1975) The Existential Pleasures of Engineering. New York: St. Martin’s Press.
Fournier, M. J., T. L. Mason, and D. A. Tirrell (1995) Role of molecular genetics in polymer materials
science. Pp. 263–75 in M. Sarikaya and I. A. Aksay, eds.: Biomimetics: Design and Processing of
Materials. Woodbury, NY: American Institute of Physics.
Frazzetta, T. H. (1966) Studies on the morphology and function of the skull in the Boidae (Serpentes). Part
II. Morphology and function of the jaw apparatus in Python sebae and Python molurus. J. Morph. 118:
217–96.
Freeman, K. (1948) Ancilla to the Pre-Socratic Philosophers. Cambridge, MA: Harvard University Press.
French, M. (1994) Invention and Evolution: Design in Nature and Engineering. Cambridge, UK:
Cambridge University Press.
Fridenson, P. (1978) The coming of the assembly line to Europe. Pp. 159–75 in W. Krohn, E. T. Layton, Jr.,
and P. Weingart, eds., The Dynamics of Science and Technology. Boston: Dordrecht.
Futuyma, D. J. (1986) Evolutionary Biology, 2d. ed. Sunderland, MA: Sinauer Associates.
Galbraith, J. K. (1967) The New Industrial State. Boston: Houghton Mifflin Co.
Gans, C. (1974) Biomechanics: An Approach to Vertebrate Biology. Philadelphia: Lippincott.
Garland, T., and R. B. Huey (1987) Testing symmorphosis: does structure match functional requirements?
Evolution 41: 1404–09.
Gérardin, L. (1968) Bionics. New York: McGraw-Hill.
Giacomelli, R., and E. Pistolesi (1934) Historical Sketch. Pp. 305–94 in W. F. Durand, ed., Aerodynamic
Theory. New York: Dover Publications (1963 reprint).
Gibbs-Smith, C. H. (1962) Sir George Cayley’s Aeronautics, 1796–1855. London: H. M. Stationery Office.
Gilinsky, N. L., and R. K. Bambach (1987) Asymmetrical patterns of origination and extinction in higher
taxa. Paleobiol. 13: 427–45.
Gordon, J. E. (1976) The New Science of Strong Materials. Harmondsworth, UK: Penguin Books. (Reprint,
Princeton, NJ: Princeton University Press, 1984).
____. (1978) Structures; or, Why Things Don’t Fall Down. New York: Plenum Press.
____. (1988) The Science of Structures and Materials. New York: Scientific American Books.
Gorman, M. E., and W. B. Carlson (1990) Interpreting invention as a cognitive process: Alexander Graham
Bell, Thomas Edison, and the telephone, 1876–1878. Science, Technology, and Human Values 15: 131–
64.
Gosline, J. M., M. E. DeMont, and M. W. Denny (1986) The structures and properties of spider silk.
Endeavour 10 (1): 37–43.
____, C. Nichols, P. Guerette, A. Cheng, and S. Katz (1995) The macromolecular design of spiders’ silks.
Pp. 237–61 in M. Sarikaya and I. A. Aksay, eds., Biomimetics: Design and Processing of Materials.
Woodbury, NY: American Institute of Physics.
Gould, J. L., J. L. Kirschvink, and K. S. Deffeyes (1978) Bees have magnetic remanence. Science 201:
1026.
Gould, S. J. (1980) The Panda’s Thumb. New York: W. W. Norton.
____. (1981) Kingdoms without wheels. Nat. Hist. 90 (4): 42–48.
____. (1989) Wonderful Life: The Burgess Shale and the Nature of History. New York: W.W. Norton.
____, and Lewontin, R. C. (1979) The spandrels of San Marco and the panglossian paradigm: a critique of
the adaptationist programme. Proc. Roy. Soc. London B205: 581–98.
Grace, J. (1977) Plant Response to Wind. London: Academic Press.
Grant, P. R. (1968) Polyhedral territories of animals. Amer. Nat. 102: 75–80.
Gray, J. (1936) Studies on animal locomotion. VI. The propulsive powers of the dolphin. J. Exp. Biol. 13:
192–99.
Griffin, D. R. (1984) Animal Thinking. Cambridge, MA: Harvard University Press.
Grosser, M. (1981) Gossamer Odyssey: The Triumph of Human-Powered Flight. Boston: Houghton Mifflin.
Gunderson, S., and R. Schiavone (1989) The insect exoskeleton: a natural structural composite. J. Mineral
Metals, and Materials Soc. 41: 60–62.
Halacy, D. S. (1965) Bionics: The Science of Living Machines. New York: Holiday House.
Haldane, J. W. S. (1928) On being the right size. Pp. 20–28 in Possible Worlds. New York: Harper.
Hallgrimsson, B., and S. Swartz (1995) Biomechanical adaptation of ulnar cross-sectional morphology in
brachiating primates. J. Morph. 224: 111–23.
Hansell, M. H. (1989) Wasp papier-mâché. Nat. Hist. 98 (8): 52–61.
Harrar, E. S., and J. G. Harrar (1962) Guide to Southern Trees. New York: Dover Publications.
Harris, C. M. (1989) The improbable success of John Fitch. Amer. Heritage of Invention and Technology
4(3): 24–31.
Harris, J. S. (1989) An airplane is not a bird. Amer. Heritage of Invention and Technology 5(2): 19–22.
Hart, C. (1985) The Prehistory of Flight. Berkeley: University of California Press.
Hayward, V. (1993) Borrowing some ideas from biological manipulators to design an artificial one. Pp.
139–52 in P. Dario, G. Sandini, P. Aebischer, eds., Robots and Biological Systems: Toward a New
Bionics. NATO ASI Series F vol. 102.
Heglund, N. C., and G. A. Cavagna (1985) Efficiency of vertebrate locomotory muscles. J. Exp. Biol. 115:
283–92.
Heilbroner, R. L. (1967) Do machines make history? Technology and Culture. 8: 335–45.
Hertel, H. (1963) Structure, Form, and Movement. New York: Reinhold Publishing Co. (translation).
Hinde, R. A. and J. Fisher (1951) Further observations on the opening of milk bottles by birds. Brit. Birds
44: 393–96.
Hix, J. (1974) The Glass House. Cambridge, MA: MIT Press.
Hochberg, J. E. (1978) Perception. Englewood Cliffs, NJ: Prentice-Hall.
Hodges, H. (1970) Technology in the Ancient World. New York: Alfred A. Knopf.
Hodges, H. W. M. (1972) Domestic building materials and ancient settlements. Pp. 523–30 in P. J. Ucko, R.
Tringham, and G. W. Dimbleby, eds., Man, Settlement, and Urbanism. Cambridge, MA: Schenkman
Publishing Co.
Hölldobler, B., and E. O. Wilson (1994) Journey to the Ants. Cambridge, MA: Harvard University Press.
Hooke, R. (1665) Micrographia, facsimile ed., 1961. New York: Dover Publications.
Hounshell, D. A. (1975) Elisha Gray and the telephone: on the disadvantages of being an expert.
Technology and Culture 16: 133–61.
Houston, J., B. N. Dancer, and M. A. Learner (1989) Control of sewer flies using Bacillus thuringiensis var.
israelensis. I. Acute toxicity tests and pilot scale trial. Water Res. 23: 369–78.
Howard, F. (1987) Wilbur and Orville: A Biography of the Wright Brothers. New York: Alfred A. Knopf.
Hulbary, R. L. (1944) The influence of air spaces on the three-dimensional shapes of cells in Elodea stems,
and a comparison with pith cells of Ailanthus. Amer. J. Bot. 31: 561–80.
Hunter, D. (1947) Papermaking: The History and Technique of an Ancient Craft. New York: Alfred A.
Knopf.
Isaacs, J. D., A. C. Vine, H. Bradner, and G. E. Bachus (1966) Satellite elongation into a true “sky-hook.”
Science 151: 682–83.
Jenkins, F. A., Jr., K. P. Dial, and G. E. Goslow, Jr. (1988) A cineradiographic analysis of bird flight: the
wishbone in starlings is a spring. Science 241: 1495–98.
Kearney, H. (1971) Science and Change 1500–1700. New York: McGraw- Hill.
Kendrick, M. J., M. T. May, M. J. Plishka, and K. D. Robinson (1992) Metals in Biological Systems. New
York: Horwood.
Kihlstedt, F. T. (1984) The crystal palace. Sci. Amer. 251 (4): 132–43.
King, M. E. (1979) The prehistoric textile industry of Mesoamerica. Pp. 265–78 in A. P. Rowe, E. P. Brown,
and A. L. Schaffer, eds., Junius B. Bird Pre-Columbian Textile Conference. Washington, DC: Textile
Museum.
Koehl, M. A. R. (1977) Mechanical organization of cantilever-like sessile organisms: sea anemones. J. Exp.
Biol. 69: 127–42.
____. (1982) Mechanical design of spicule-reinforced connnective tissue: stiffness. J. Exp. Biol. 98: 239–
67.
____, T. Hunter, and J. Jed (1991) How do body flexibility and length affect hydrodynamic forces on sessile
organisms in waves versus in currents? Amer. Zool. 31: 60A.
Koops, M. (1800) Historical Account of the Substances Which Have Been Used to Describe Events and to
Convey Ideas from the Earliest Date to the Invention of Paper. London: T. Burton.
Kramer, M. O. (1965) Hydrodynamics of the dolphin. Adv. Hydrosci. 2: 111–30.
LaBarbera, M. (1983) Why the wheels won’t go. Amer. Nat. 121: 395–408.
____. (1990) Principles of design of fluid transport systems in zoology. Science 249: 992–1000.
____, and S. Vogel (1982) The design of fluid transport systems in organisms. Amer. Sci. 70: 54–60.
Laithwaite, E. (1984) Invitation to Engineering. Oxford, UK: Basil Blackwell.
____. (1989) A History of Linear Electric Motors. San Francisco: San Francisco Press.
____. (1994) An Inventor in the Garden of Eden. Cambridge, UK: Cambridge University Press.
Lechtman, H. (1988) Traditions and styles in central Andean metalworking. Pp. 344–78 in R. Madden, ed.,
The Beginning of the Use of Metals and Alloys. Cambridge, MA: MIT Press.
Leeming, J. (1949) Rayon, the First Man-Made Fiber. Brooklyn, NY: Chemical Publishing Co.
Levy, M., and M. Salvadori (1992) Why Buildings Fall Down: How Structures Fail. New York: W. W.
Norton.
Liebowitz, S. J., and S. E. Margolis (1990) The fable of the keys. J. Law and Econ. 33: 1–25.
____, and ____. (1995) Path dependence, lock-in, and history. J. Law, Economics, and Organization 11:
205–26.
Lilienthal, O. (1910) Birdflight as the Basis for Aviation. London: Longmans, Green.
Lombardi, S. J., S. Fossey, and D. L. Kaplan (1990) Recombinant spider silk proteins for composite fibers.
Proc. Amer. Soc. for Composites, 5th Technical Conference. 184–87.
Lovelock, J. E., and L. Margulis (1974) Atmospheric homeostasis by and for the biosphere: the Gaia
hypothesis. Tellus 26: 1–9.
Lowenstam, H. A. (1967) Lepidocrocite, an apatite mineral, and magnetite in teeth of chitons
 (Polyplacophora). Science 156: 1373–75.
Lucia, E. (1975) The Big Woods: Logging and Lumbering, from Bull Teams to Helicopters, in the Pacific
Northwest. New York: Doubleday.
____. (1981) Joe Cox and his revolutionary chain saw. J. Forest Hist. 25: 159–65.
McCallum, H. D., and F. T. McCallum (1965) The Wire That Fenced the West. Norman: University of
Oklahoma Press.
McCulloch, W. S. (1962) The imitation of one life form by another-biomimesis. Pp. 393–97 in E. E.
Bernard and M. R. Kare, eds., Biological Prototypes and Synthetic Systems. New York: Plenum Press.
McFarland, M. W., ed. (1953) The Papers of Wilbur and Orville Wright, Vol. 1: 1899–1905. New York:
McGraw-Hill.
McHenry, M. J., C. A. Pell, and J. H. Long, Jr. (1995) Mechanical control of swimming speed: stiffness and
axial wave form in an undulatory fish model. J. Exp. Biol. 198: 2293–2305.
McMahon, T. A. (1984) Muscles, Reflexes, and Locomotion. Princeton, NJ: Princeton University Press.
____, and J. T. Bonner (1983) On Size and Life. New York: Scientific American Books.
Mackay, A. L. (1991) A Dictionary of Scientific Quotations. New York: Adam Hilger.
Majdalany, F. (1960) The Eddystone Light. Boston: Houghton Mifflin.
Manko, D. J. (1992) A General Model of Legged Locomotion on Natural Terrain. Boston: Kluwer
Academic Publishers.
Mann, S. (1990) Crystal engineering: the natural way. New Scientist 125 (1707): 42–47.
____. (1995) Biomineralization, the inorganic-organic interface, and crystal engineering. Pp. 91–116 in M.
Sarikaya and I. A. Aksay, eds.: Biomimetics: Design and Processing of Materials. Woodbury, NY:
American Institute of Physics.
Maor, E. (1994) e: The Story of a Number. Princeton, NJ: Princeton University Press.
Margulis, L. (1993) Symbiosis in Cell Evolution. New York: W. H. Freeman.
Mark, R. (1978) Structural experimentation in Gothic architecture. Amer. Sci. 66: 542–50.
____. (1982) Experiments in Gothic Structure. Cambridge, MA: MIT Press.
Marteka, V. (1965) Bionics. Philadelphia: J. B. Lippincott.
Mason, M. T., and J. K. Salisbury, Jr. (1985) Robot Hands and the Mechanics of Manipulation. Cambridge,
MA: MIT Press.
Mattheck, C. (1991) Trees: The Mechanical Design. Berlin: Springer-Verlag.
Maxim, H. S. (1909) Artificial and Natural Flight. New York: Whittaker and Co.
Maynard Smith, J. (1952) The importance of the nervous system in the evolution of animal flight. Evolution
6: 127–29.
Mayr, O. (1970) The Origins of Feedback Control. Cambridge, MA: MIT Press.
Meinhardt, H. (1995) The Algorithmic Beauty of Sea Shells. New York: Springer-Verlag.
Miller, K. F., C. P. Quine, and J. Hunt (1987) The assessment of wind exposure for forestry in upland
Britain. Forestry 60: 179–92.
Miller, M. (1945) The Far Shore. New York: McGraw-Hill.
Miller, R. C. (1924) The boring mechanism of Teredo. Univ. Calif. Publ. Zool. 26: 41–80.
Mokyr, J. (1991) Evolutionary biology, technological change, and economic history. Bull. Economic Res.
43: 127–47.
Morowitz, H. (1968) Energy Flow in Biology. New York: Academic Press.
Mumford, L. (1967) The Myth and the Machine: Technics and Human Development. New York: Harcourt
Brace Jovanovich.
Needham, J. (1954) Science and Civilisation in China, vol. 1. Cambridge, UK: Cambridge University Press.
____. (1965) Science and Civilisation in China, vol. 4, part 2. Cambridge, UK: Cambridge University
Press.
____, and G.-D. Lu (1985) Transpacific Echoes and Resonances: Listening Once Again. Philadelphia:
World Scientific.
Nicklas, R. B. (1984) A quantitative comparison of cellular motile systems. Cell Motility 4: 1–5.
Nijhout, H. F. (1990). Metaphors and the role of genes in development. BioEssays 12: 441–46.
____, and H. G. Sheffield (1979) Antennal hair erection in male mosquitoes: a new mechanical effector in
insects. Science 206: 595–96.
Niklas, K. J. (1992) Plant Biomechanics. Chicago: University of Chicago Press.
____. (1994) Plant Allometry. Chicago: University of Chicago Press.
Oberg, E., F. D. Jones, and H. L. Horton (1984) Machinery's Handbook, 22d ed. New York: Industrial
Press.
O’Neill, P. L. (1990) Torsion in the asteroid ray. J. Morph. 203: 141–50.
Orwell, G. (1937) The Road to Wigan Pier. London: Victor Gollancz, Ltd.
Ovid (P. Ovidius Naso) Metamorphoses, tr. F. J. Miller (1966). Cambridge, MA: Loeb Classical Library.
Owen, D. H. (1978) The psychophysics of prior experience. Pp. 467–524 in P. K. Machamer and R. G.
Turnbull, eds., Studies in Perception. Columbus: Ohio State University Press.
Ozanam, M. (1862) Dissolution de la soie par l’ammoniure de cuivre. C. R. des Séances de l’Academie des
Sciences 55: 833.
Pantin, C. F. A. (1964) Homeostasis and the environment. Pp. 1–6 in G. M. Hughes, ed., Homeostasis and
Feedback Mechanisms: Symp. Soc. Exp. Biol. 18.
Papanek, V. (1971) Design for the Real World. New York: Random House.
Parkes, E. W. (1965) Braced Frameworks: An Introduction to the Theory of Structures. Oxford, UK: Oxford
University Press.
Paturi, F. R. (1976) Nature, Mother of Invention: The Engineering of Plant Life. New York: Harper and
Row.
Pearce, P. (1978) Structure in Nature Is a Strategy for Design. Cambridge, MA: MIT Press.
Petroski, H. (1985) To Engineer Is Human: The Role of Failure in Successful Design. New York: St.
Martin’s Press.
____. (1991) Still twisting. Amer. Sci. 79: 398–401.
____. (1992) The Evolution of Useful Things. New York: Random House.
____. (1994) Design Paradigms: Case Histories of Error and Judgment in Engineering. New York:
Cambridge University Press.
____. (1995) Engineers of Dreams: Great Bridge Builders and the Spanning of America. New York: Alfred
A. Knopf.
Pettigrew, J. B. (1908) Design in Nature. London: Longmans Ltd.
Phillips, H. F. (1885) Experiments with currents of air (anonymous report). Engineering 40: 160–61.
Pierce, J. R. (1961) Symbols, Signals and Noise: The Nature and Process of Communication. New York:
Harper and Row.
Pool, R. (1989) Making new materials with nature’s help. Science 250:1389.
Prakash, S. (1987) Geometry in Ancient India. Columbia, MO: South Asian Books.
Prescott, C. B. (1884) Bell's Electric Speaking Telephone: Its Invention, Construction, Application,
Modification, and History. New York: D. Appleton and Co.
Pritchard, J. L. (1961) Sir George Cayley, the Inventor of the Airplane. London: Max Parrish.
Raibert, M. H., and I. E. Sutherland (1983) Machines that walk. Sci. Amer. 248 (1): 44–53.
Raup, D. M. (1966) Geometric analysis of shell coiling: general problems. J. Paleontol. 40: 1178–90.
____.(1992) Extinction: Bad Genes or Bad Luck. New York: W. W. Norton.
Reif, W.-E. (1985) Morphology and hydrodynamic effects of the scales of fast swimming sharks. Fortschr.
Zool. 30: 483–85.
Reynolds, T. S., ed. (1991) The Engineer in America. Chicago: University of Chicago Press.
Riley, J. J., M. Gad-el-Hak, and R. W. Metcalfe (1988) Compliant Coatings. Annu. Rev. Fluid Mech. 20:
393–420.
Robinson, J. R., and E. C. Frederick (1989) Scaling of foot dimensions. XII Congr. Int. Soc. Biomech.
Abstr., p. 1074.
Rogers, E. M. (1983) Diffusion of Innovations, 3d ed. New York: Free Press (Macmillan).
Rolt, L. T. C. (1959) Isambard Kingdom Brunel, a Biography. New York: St. Martin’s Press.
Rosenzweig, M. L., and R. D. McCord (1991) Incumbent replacement: evidence for long-term evolutionary
progress. Paleobiol. 17: 202–13.
Roth, V. R. (1996) Cranial integration in the Sciuridae. Amer. Zool. 36: 14–23.
Rouse, H., and S. Ince (1957) History of Hydraulics. New York: Dover Publications (1963 reprint).
Runnegar, B. (1992) Evolution of the earliest animals. In J. W. Schopf, ed., Major Events in the History of
Life. Boston: Jones and Bartlett Publishers.
Saidel, B. A. (1993) Round house or square? Architectural form and socioeconomic organization in the
PPNB. Mediterranean Archaeology 6: 65–108.
Salehpoor, K., M. Shahinpoor, and M. Mojarrad (1996) Electrically controllable artificial PAN muscles. Pp.
116–24 in A. Crowson, ed., Smart Materials Technologies and Biomimetics. Proc. Int. Soc. for Optical
Engineering, vol. 2716.
Salisbury, J. K. (1950) Kent's Mechanical Engineering Handbook, 12th ed., vol. 2. New York: John Wiley.
Salvadori, M. (1980) Why Buildings Stand Up: The Strength of Architecture. New York: W. W. Norton.
Sarikaya, M., and I. A. Aksay (1995) Biomimetics: Design and Processing of Materials. Woodbury, NY:
American Institute of Physics.
____, C. E. Furlong, and J. T. Staley (1994) Nanodesigning and properties of biological composites.
Advances in Bioengineering. ASME BED Publication 28: 47–48.
Schlesinger, W. H., J. T. Gray, D. S. Gill, and B. E. Mahall (1982) Ceanothus megacarpus chaparral: a
synthesis of ecosystem processes during development and annual growth. Bot. Rev. 48: 71–117.
Schlosser, L. B. (1980) Papermaking and the industrial revolution: the search for new fiber. Amer. Book
Collector 1 (6): 3–12.
Schmidt, P. R., ed. (1996) Culture and Technology of African Iron Production. Gainesville: University Press
of Florida.
Schmidt-Nielsen, K. (1984) Scaling. Cambridge, UK: Cambridge University Press.
____. (1997) Animal Physiology: Adaptation and Environment, 5th ed. Cambridge, UK: Cambridge
University Press.
Schneider, M. S. (1994) A Beginner’s Guide to Constructing the Universe. New York: HarperCollins.
Schober, J. (1930) Silk and the Silk Industry. London: Constable and Co.
Schoenheimer, R. (1942) The Dynamic State of Body Constituents. Cambridge, MA: Harvard University
Press.
Schopf, J. W. (1992) Major Events in the History of Life. Boston: Jones and Bartlett Publishers.
Shadwick, R. E. (1994) Mechanical organization of the mantle and circulatory system of cephalopods. Mar.
Behav. Physiol. 25: 69–85.
____, C. M. Pollock, and S. A. Strieker (1990) Structure and biomechanical properties of crustacean blood
vessels. Physiol. Zool. 63: 90–101.
Singer, C., E. J. Holmyard, and A. R. Hall, eds. (1954) A History of Technology, vol. 1. Oxford, UK:
Clarendon Press.
Sleeswyk, A. W. (1981) Hand-cranking in Egyptian antiquity. Pp. 23–37 in A. R. Hall and N. Smith, eds.,
History of Technology, vol. 6. London: Mansell Publishing.
Smeaton, J. (1759) An experimental enquiry concerning the natural powers of water and wind to turn mills,
and other machines, depending on a circular motion. Phil. Trans. Roy. Soc. Lond. 51: 100–174.
Smith, K. K., and W. M. Kier (1989) Trunks, tongues, and tentacles: moving with skeletons of muscle.
Amer. Sci. 77: 28–35.
Smith, M. R., and L. Marx, eds. (1995) Does Technology Drive History? The Dilemma of Technological
Determinism. Cambridge, MA: MIT Press.
Snow, C. P. (1959) The Two Cultures. Cambridge, UK: Cambridge University Press.
Song, S.-M. and Waldron, K. J. (1989) Machines That Walk: The Adaptive Suspension Vehicle. Cambridge,
MA: MIT Press.
Sotavalta, O. (1953) Recordings of high wing-stroke and thoracic vibration frequency in some midges. Biol.
Bull. 104: 439–44.
Stanley, S. M. (1987) Extinction. New York: Scientific American Library.
Steadman, P. (1979) The Evolution of Design: Biological Analogy in Architecture and the Applied Arts.
Cambridge, UK: Cambridge University Press.
Stix, G. (1994) Robotuna. Sci. Amer. 270 (1): 42.
Sutton, O. G. (1949) The Science of Flight. Harmondsworth, UK: Penguin Books.
Swetz, F., and T. I. Kao (1977) Was Pythagoras Chinese? An Examination of Right Triangle Theory in
Ancient China. University Park: Pennsylvania State University Press.
Tamm, S. L. (1982) Flagellated ectosymbiotic bacteria propel a eukaryotic cell. J. Cell Biol. 94: 697–709.
Taylor, C. R., and E. R. Weibel (1981) Design of the mammalian respiratory system. I. Problem and
strategy. Respir. Physiol. 44: 1–10.
Tennekes, H. (1996) The Simple Science of Flight. Cambridge, MA: MIT Press.
Tenner, E. (1996) Why Things Bite Back: Technology and the Revenge of Unintended Consequences. New
York: Alfred A. Knopf.
Thompson, D’Arcy W. (1942) On Growth and Form, 2d ed. Cambridge, UK: Cambridge University Press.
____. (1961) On Growth and Form. Abridged, J. T. Bonner, ed., Cambridge, UK: Cambridge University
Press.
Thompson, T. E., and I. Bennett (1969) Physalia nematocysts: utilized by mollusks for defense. Science
166: 1532–33.
Thomson, K. S. (1991) Living Fossil: The Story of the Coelacanth. New York: W. W. Norton.
Tokaty, G. A. (1971) A History and Philosophy of Fluid Mechanics. New York: Dover Publications.
Toms, B. A. (1948) Some observations on the flow of linear polymer solutions through straight tubes at
large Reynolds numbers. Proc. Int. Rheological Congr., Scheveningen, Netherlands 2: 135.
Triantafyllou, G. S., M. S. Triantafyllou, and M. A. Grosenbaugh (1993) Optimal thrust development in
oscillating foils with application to fish propulsion. J. Fluids and Structures 7: 205–24.
Tributsch, H. (1982) How Life Learned to Live. Cambridge, MA: MIT Press.
Tylecote, R. F. (1992) History of Metalworking. London: Institute of Metals.
Urry, D. W. (1993) Molecular machines: how motion and other functions of living organisms can result
from reversable chemical changes. Angew. Chem. Int. Ed. Engl. 32: 819–41.
Usher, A. P. (1954) A History of Mechanical Inventions. New York: Dover Publications (reprint, 1988).
Vaughan, A. (1991) Isambard Kingdom Brunel: Engineering Knight Errant. London: John Murray.
Vermeij, G. J. (1993) A Natural History of Shells. Princeton, NJ: Princeton University Press.
Vincent, J. F. V. (1990) Structural Biomaterials, rev. ed. Princeton, NJ: Princeton University Press.
Vincenti, W. G. (1988) How did it become “obvious” that an airplane should be inherently stable? Amer.
Heritage of Invention and Technology 4(1): 50–56.
____. (1990) What Engineers Know and How They Know It: Analytical Studies from Aeronautical History.
Baltimore: Johns Hopkins University Press.
Vitousek, P. M., C. M. D’Antonio, L. L. Loope, and R. Westbrooks (1996) Biological invasions as global
environmental change. Amer. Sci. 84: 468–78.
Vogel, S. (1970) Convective cooling at low airspeeds and the shapes of broad leaves. J. Exp. Bot. 21: 91–
101.
____. (1978) Organisms That Capture Currents. Sci. Amer. 239 (2): 128–39.
____. (1984a) Drag and flexibility in sessile organisms. Amer. Zool. 24: 37–44.
____. (1984b) The thermal conductivity of leaves. Can. J. Bot. 62: 741–44.
____. (1988) Life’s Devices: The Physical World of Animals and Plants. Princeton, NJ: Princeton University
Press.
____. (1989) Drag and reconfiguration of broad leaves in high winds. J. Exp. Bot. 40: 941–48.
____. (1992) Vital Circuits: On Pumps, Pipes, and the Workings of Circulatory Systems. New York: Oxford
University Press.
____. (1993) When leaves save the tree. Nat. Hist. 102 (9): 58–63.
____. (1994a) Life in Moving Fluids: The Physical Biology of Flow, 2d ed. Princeton, NJ: Princeton
University Press.
____. (1994b) Nature’s pumps. Amer. Sci. 82: 464–71.
____. (1994c) Second-rate squirts. Discover 15 (8): 70–76.
____. (1994d) Dealing honestly with diffusion. Amer. Biol. Teacher 56: 405–7.
____. (1995) Better bent than broken. Discover 16 (5): 62–67.
____, and W. L. Bretz (1972) Interfacial organisms: passive ventilation in the velocity gradients near
surfaces. Science 175:210–11.
von Kármán, T. (1954) Aerodynamics. New York: McGraw-Hill.
Wainwright, S. A. (1995) What we can learn from soft biomaterials and structures. Pp. 1–12 in M. Sarikaya
and I. A. Aksay, eds.: Biomimetics: Design and Processing of Materials. Woodbury, NY: American
Institute of Physics.
____, W. D. Biggs, J. D. Currey, and J. M. Gosline (1976) Mechanical Design in Organisms. London:
Edward Arnold (reprint, Princeton, NJ: Princeton University Press).
Walcott, C., J. L. Gould, and J. L. Kirschvink (1979) Pigeons have magnets. Science 205: 1028.
Walsby, A. E. (1980) A square bacterium. Nature 283: 69–71.
Wayman, M. L. (1989a) Native copper: humanity’s introduction to metallurgy? Pp. 3–6 in M. L. Wayman,
ed., All That Glitters: Readings in Historical Metallurgy. Montreal: Canadian Institute of Mining and
Metallurgy.
____. (1989b) On the early use of iron in the Arctic. Pp. 99–100 in M. L. Wayman, ed.: All That Glitters:
Readings in Historical Metallurgy. Montreal: Canadian Institute of Mining and Metallurgy.
Wells, C. S., ed. (1969) Viscous Drag Reduction. New York: Plenum Press.
Went, F. W. (1968) The size of man. Amer. Sci. 56: 400–13.
Westneat, M. W., and P. C. Wainwright (1989) Feeding mechanism of Epihuus insidiator (Labridae;
 Teleostei): evolution of a novel feeding mechanism. J. Morphol. 202: 129–50.
White, L., Jr. (1962) Medieval Technology and Social Change. New York: Oxford University Press.
Wiener, N. (1950) The Human Use of Human Beings: Cybernetics and Society. Boston: Houghton Mifflin.
Williams, T. (1882) The Eddystone Lighthouses (New and Old). London: Simpkin, Marshall and Co.
Wilson, E. O. (1980) Sociobiology, Abridged ed. Cambridge, MA: Harvard University Press.
____. (1984) Biophilia. Cambridge, MA: Harvard University Press.
____. (1992) The Diversity of Life. New York: W. W. Norton.
Wilson, G. (1855) What Is Technology? Edinburgh, UK: Sutherland and Knox.
Wilson, J. F., D. Li., Z. Chen, and R. T. George, Jr. (1993) Flexible robot manipulators and grippers:
relatives of elephant trunks and squid tentacles. Pp. 475–94 in P. Dario, G. Sandini, P. Aebischer, eds.:
Robots and Biological Systems: Toward a New Bionics. NATO ASI, Series F, vol. 102.
Wilson, P. N. (1975) J. G. A. Kitchen, 1869–1940, and his inventions. Trans. Newcomen Soc. 45: 15–43 (for
1972–73).
Winfield, D. L., D. H. Hering, and D. Cole (1991) Engineering Derivatives from Biological Systems for
Advanced Aerospace Engineering. NASA CR- 177594; Research Triangle Institute, NC.
Woodbury, R. S. (1960) The legend of Eli Whitney and interchangeable parts. Technology and Culture 1:
235–53.
Worden, E. C. (1911) Nitrocellulose Industry. New York: D. Van Nostrand Co.
Wright, O. (1953) How We Invented the Airplane. New York: David McKay Co.
Zimmermann, M. H. (1983) Xylem Structure and the Ascent of Sap. Berlin: Springer-Verlag.
 INDEX

Page numbers listed correspond to the print edition of this book. You can use your device’s search function
to locate particular terms in the text.

Page numbers in italics refer to illustrations

abductin, 28, 87, 89, 143, 203

acceleration:
and predation, 200
in jumping, 32
projectiles, 204
Achilles tendon, 89, 104
adhesion, 269
aerodynamics. See camber, drag, lift, etc.
aerostats, 148, 311
ailerons, 261, 270, 271
aircraft, 16, 17, 37, 43–48, 46, 47, 51, 57
autogyrating, 53, 54
control, 91, 197, 260, 261, 272
fixed wings, 221
flow around, 52
helicopters, 220–21
history, 271–74, 273
jet-powered, 219, 296
lighter-than-air, 151
maintenance, 246
performance, 258
production, 245
shapes, 64
windows, 80
Wright Flyer, 222
airfoils, 259, 270
camber, 61, 258, 271
flapping, 272
power source, 215, 221
of propellers, 221
propulsion efficiency, 218
windmills, 97
 Wrights’, 260
albatross, gliding, 47
Alexander, R. McNeill, 104, 201
algae, 94, 95
aluminum, 247
cookware, 124
density, 112
in earth, 110, 112
oxidation, 111
alveoli, 64
Ames window, 66, 67
aneurysms, 103, 104, 148
angles. See right angles
animal territories, 74–75
anthropology, 37, 307–9
ants:
aggression, 52
reproduction, 24
archery, 37, 127, 194, 200
Archimedean screw, 287, 288
Archimedes, 192
Aristotle, 19
armadillos, skeleton, 28
arteries. See blood vessels
arthropods. See insects, spiders, etc.
ascidians, 107
vanadium in, 107, 110
atherosclerosis, 102, 103
Audemars, Georges, 263
autogyrating, 53, 54
automobiles, 35, 57
bearings, 303, 304
brakes, 195, 196, 237
carburetors, 207
control, 234, 235
engines, 33, 158
loads on, 134
pumps, 207
shape, 62
shock absorbers, 90, 90
suspensions, 90, 91, 134, 142, 142, 198
transmissions, 178, 197

bacteria:
flagella, 186, 187
magnetite in, 109
nitrogen fixation, 109
smallest, 40
spirochetes 29, 30, 30
 square, 65
ballistae, 127, 200
balloons, 63, 64, 102, 104, 148, 163, 220
bamboo, 17, 281, 287
barbed wire, 266, 267, 270
barnacles, 25, 147
Basalla, George, 188, 267, 293
bats:
flight, 27
wings, 58, 104
batteries, 154, 163, 198–204, 224, 290
elastic, 198, 199, 200, 202, 203
electrical, 198, 200, 295, 296, 302
gravitational, 198, 200
inertial, 198
in nature, 200, 204, 290
Beadle, George, 22
beams, 130, 140, 141, 231
between supports, 55
bookshelves as, 60
feathers as, 61
flexural stiffness, 130
head supports, 141
I-beams, 100, 132
leaves as, 60, 61
prestressed, 143
roofs, 58
sagging, 59
scaling, 55
shapes, 68, 140
strength, 118
tension, compression in, 140
torsional stiffness, 100
trees as, 253
under floors, 76
bearings, 186, 187, 189–91, 303, 304
bees:
flying speed, 46
heating hives, 43
honeycombs, 76
magnetite in, 109
reproduction, 24
beetles:
dung, 188
fixed forewings, 215
whirligig, 51, 226, 227, 227
woodboring, 268, 271
Bell, Alexander, 31, 37, 265
Berg, Howard, 186
Bernheim, Molly, 274
Bernoulli’s principle, 166, 167, 260
bicycles, 17, 192
chains, 88
cost of transport, 188
tire pump, 278
wheels and pedals, 188
biomechanics, 18, 22, 41, 145, 311
biomimetics, bionics, 250
see also copying nature
biophilia, 18, 250
biophysics, 18
biotechnology, 18
birds:
airfoils, 259
body temperature, 43
creativity, 32
falling, 44
flight, 27, 32, 220, 221, 222, 258, 260, 271, 272
gliding, 46, 167
hydroplaning, 226
magnetite in, 109
smallest, 43
territories, 76
wing bones, 71
wishbone springs, 143
see also eggs, feathers
Blake, William, 257
blimps, 148, 151, 220
block and tackle, 179
blood:
flow, 52, 125
pressure, 102
blood vessels:
shapes, 64
stress-strain graph, 86, 103
variability, 27
walls, 28, 86,102–5, 103, 203
boats, 16, 31, 41
acceleration, 52
anchoring, 87
building, 68, 309
flow around, 52
hulls, 94, 134, 138, 257
paddle wheels, 211, 223, 275, 287
portholes, 80
propulsion, 186, 287, 288
speed, and waves, 51, 225, 226
steamboats, 117
surface swimming, 223–28, 225, 226, 291
see also submarines
bone, bones, 17, 24, 70, 106, 128
bird wings, 71
calcium in, 110
as composites, 123, 231, 232, 280
cracks, 116
density, 112
fracture, 55
growth, 24, 116, 295
half-life, 239
in human technology, 127
joints between, 191
loads on, 135
responsive growth, 240, 241
safety factor, 244
scaling, 56, 137
shapes, 80
stiffness, 93
stress-strain graph, 114, 116
viscoelasticity, 116
bookshelves, 60, 68
bow drills, 193, 194
bows, 194, 200
brachiating, 137
breast-reduction surgery, 23
bricks, 68, 92, 128, 133, 248
bridges, 41
arch, 134, 135, 136
materials, 285
scaling, 56
suspension, 64, 134, 135, 136, 139, 144
brittle materials, 118
brittleness, 79, 290
bronze, 109, 117, 292, 295, 308
Brunel, Isambard, 139
Brunel, Marc, 37, 245, 253
buckling, 132, 135, 144
buildings, 41
aerostatic, 148
A-frame, 138
cathedrals, 134, 137, 138
framing, 69, 139, 144
geodesic, 77
heating vs. size, 43
shapes, 65
support schemes, 138
ventilation, 168
buoyancy, 70
Burns, John, 32
burrows, burrowing, 151, 166, 211
burs, 269
butterflies:
gliding, 46, 47
metamorphosis, 25, 196
buttressing, 134, 138

cables, 70
cacti, 37
caddisflies, 167
calcium, 94
in bones, 110
in nature, 109
reduction of, 111
in seawater, 110
camber, 61, 258, 271
capillarity, 48
capillary waves, 51, 227, 227
cartilage, 104
casting, 80
catapults, 127, 199
caterpillars:
hydroskeleton, 70
leaf cutting, 270
cathedrals, 134, 137, 138
cats:
boniness, 56, 137
ears, 105
stalking gait, 73
tongues, 108
Cayley, George, 257, 270
cells:
blood, 239
close packing, 77
coordination, 240
diffusion in, 50
as factories, 230
membranes, 65
number in organism, 25, 229
as precursors, 41
pressures within, 210
shapes, 64
variability, 243
cellulose:
digestion, 29
in human technology, 118
in wood, 123
centipedes, 145
cephalopods. See octopus, squid
ceramics, 117
composites, 281
 cookware, 125
stiffness, 94
cermets, 281
chains, 88
chain saws, 268, 270
from beetle mandibles, 267
radulas as, 108
Chanute, Octave, 273
Chardonnet, Hilaire de, 263
chitin, 106, 123, 204
cilia, 169
bending, 184
as motors, 157, 169
pumping with, 212, 212, 213, 214
circulatory systems, 50, 213
clams:
abductin, 28
metamorphosis, 25
muscle, 29, 170
suspension feeding, 213
close-packing, 76
cocklebur, 269
coelacanths, 35
coelenterates. See jellyfish, sea anemones, etc.
collagen, 91
in ballistae, 127
in blood vessels, 103
resilience, 202
triple helix, 132
see also tendons
colonial animals, 24
columns, 17, 53–55, 132, 133, 292
composite materials, 122–24, 122, 124, 126, 231, 232, 250, 280
compression:
in stacks, 68
testing, 129
compression elements:
examples, 128, 132, 135
fluid, 153–76
human preference, 135, 290
size and shape, 136, 144–46
vs. tensile elements, 135–44
concrete, 17
conductivity, electrical, 126, 162
conductivity, thermal, 124–26, 290
cones, 302–7
vs. cylinders, 303, 304
growth, 23, 24, 302, 305
in human technology, 303, 304
in nature, 304
control, 91, 233, 234, 235, 281
aircraft, 274
equifinality, 236
genetic information, 241
industrial, 245
responsive growth, 240–41
see also feedback
convection, 50, 125
convergence, 31, 36, 37, 43, 89, 307–9
see also evolution
Cook, Theodore Andrea, 306
cooking, 32
controlling, 233
edible composites, 122
size vs. time, 42
thermal conductivity, 124
Cope’s rule, 245
copper, 107, 109, 308
density, 112
human use, 116, 292
reduction of, 111
in seawater, 110
sources, 113
thermal conductivity, 124
copying nature, 19, 80, 117, 249–88
ailerons, 260
barbed wire, 266
cambered airfoils, 258, 259
chain saws, 267, 268, 270
composite materials, 280
Crystal Palace, 255
difficulties, 276, 310
Eddystone light, 251
extruded fibers, 263–65
flight, 271–74
low-drag submarines, 278
manipulators, 281
materials, 285
muscle analogs, 280
nanotechnology, 280
oscillatory swimming, 283, 284
papermaking, 261, 271
peristaltic pumps, 277, 277
polymer drag reduction, 278
rarity, 251
shark scales, 279
smart materials, 281
streamlined bodies, 257
telephones, 265, 266
tunneling shield, 253, 254
 Velcro, 268, 269
walking vehicles, 283, 283
when likely, 270–71
corals, 94, 106, 128, 147, 280
corners, 78–81, 118, 289
corrugated structures, 17, 62, 281
Cox, Joseph, 268
crabs, blood vessels, 104
cracks, cracking, 78–80, 93, 116, 119, 133
energetics, 119
in metals, 115
propagation, 115, 118–22, 290
stopping, 120–24, 121
Crane, Horace, 25
crane flies, 145
cranks, 180, 185, 186, 191, 192, 193
crankshafts, 191
Cree, 68
crushing, 54, 132
crustaceans, molting, 24
Crystal Palace, 255, 255
cubes, 76. See packing, right angles, stacking
cubists, 16, 65
Currey, John, 116
curved surfaces. See surfaces
cuticle, as composite, 123, 281
cylinders, 57, 131
as beams, 55
cans as, 64
as columns, 54
vs. cones, 303, 304
fuselages, 64
growth, 23, 24
hydroskeletal, 149, 150
inflating, 102
Laplace’s law, 62–64, 63
torsional stiffness, 98, 131
daddy longlegs, 55, 145, 145
Daedalus, 249
daffodils, stem stiffness, 101
Dalí, Salvador, 16
damping, 90, 142
dancing, 188
dandelions, 53, 54
da Vinci, Leonardo, 19
Dawkins, Richard, 27
de Mestral, Georges, 269
Descartes, 40
development of organism, 236
diffusion:
 cultural, 37, 308
in human technology, 50, 289
molecular, 50, 56, 173, 289
dinosaurs, 34, 43
dispersal, 25
DNA, 234, 236, 241, 280, 295
dogs, 37
dolphin, drag of, 278
see also whales
domes:
by Brunelleschi, 138
curvature, 64
geodesic, 77, 78
in nature, 64, 78
domestication, 32
drag:
falling, 44
flags, 95
flight, 45
kelp, 95
leaves, 96–99, 253
vs. lift, 46, 258
microorganisms, 52
pumping by, 211
vs. speed, 219, 225
streamlining, 257, 258, 278, 290
vs. surface area, 44, 45
from waves, 223, 224
dragonfly anal jets, 210, 216, 217
Drebbel, Cornelis, 233
Driesch, Hans, 235
drills, 192–93, 194
ducks, 51
feet, 58, 104, 275
surface swimming, 223, 224, 225, 226
ductility, 113, 115, 120, 290
dynamic state of body constituents, 237–39, 248

ear, 66
eardrum, 265, 271
echinoderms. See starfish, etc.
Eddystone light, 251, 252
Edison, Thomas, 31, 37, 266
efficiency, 156
flight, 47
jet propulsion, 217–20
motors, 161, 164, 174–75
muscle, 175
thermal, 160–61
 various linkages, 189
egg, eggs:
as domes, 64
information content, 23
shells, 17, 64, 124
elasticity. See resilience, stiffness
elastic limit, 85
elastic proteins. See abductin, elastin, resilin
elastin, 28, 28
Eldredge, Niles, 34
electrochemical series, 110
elephants:
boniness, 56, 137
legs, 55
trunks, 151, 183, 282
energy, 155
to contract muscle, 29
conversions, 152, 153
in cracking, 115
and resilience, 86, 89
sources, 154, 155, 294
storage, 28, 104, 115, 198, 201 204, 286 see also batteries
stress-strain graph, 104
of surfaces, 119
work of extension, 86
engineering, 18, 19, 309, 310, 311
engines. See motors
equifinality, 236
Ericsson, John, 287
Etrich, Ignaz and Igo, 274
euphorbs, 37
evaporation, 125, 209
evolution, 20–31, 213
analogy, 22
apparent altruism, 292
conservative character, 33, 127
convergence, 31, 36, 89, 307–9
design by, 22, 127
extinction, 37
eyes, 37
in human technology, 20, 307–9
incremental character, 35, 299
Lamarckian, 240
limitations, 23–31, 291, 298
mechanism, 21–22
of multicellularity, 230, 233
novelties, 27, 293, 294
preadaptations, 35
punctuated equilibrium, 34
role of isolation, 33
 size trends, 245
technological analogies, 298–302
time-course, 34, 294, 295, 296, 310
of wheels, 188
see also groups of organisms
Ewbank, Thomas, 19
extensibility, 84, 85, 87
of ropes, 88
extinction, 37
eyes, 37, 234, 307

factories. See manufacturing

failure:
crushing vs. buckling, 54
learning from, 300
fairing corners, 79, 80, 81
falling speeds, 44–45
fan-folding, 62
fans, 61
fat, 43
feathers:
aerodynamics, 61
as beams, 61
insulation, 43
torsional stiffness, 99
feedback, 233–37, 234, 235, 241, 245, 274, 281, 295
see also control systems
feet, sole area, 43
fiberglass, 122, 123, 231, 247
fibers:
in blood vessels, 103
glass, strength, 118
in hydroskeletons, 148, 149, 150
natural, 87, 131
synthetic, 31, 89, 92, 263–65, 269, 271
filtration, 17, 25
see also suspension feeding
fish:
conical myotomes, 304
cooking, 42
electric, 162
gills, 168, 210
jaws, 70, 70
mudskippers, 36
nostrils, 167
propulsion efficiency, 219
scales, 58
slimes, 278
streamlining, 257
 swimming, 283
territories, 76
Fitch, John, 275
fitness, 33, 236, 246, 247, 248
flagella:
bacterial, 29, 186, 187
see also cilia
flags, drag of, 95, 96
flat surfaces. See surfaces
fleas, jumping, 39, 52, 204
flexibility, 82, 106
blood vessels, 102
modern plastics, 101
in nature, 80, 93
of tensegrities, 147
and toughness, 94
uses in nature, 94–102
flexural stiffness, 100–102
beam, 130
bending, 17, 59, 79, 115
cylinders, 98
flat surfaces, 60–62
insect wings, 62
flies, gliding of, 47
flight:
by beating wings, 47
birds, 32, 220, 221, 222, 271, 272
control, 260, 261
by copying nature, 258
efficiency, 47
gliding, soaring, 46–47
history, 271–74
insects, 28, 45, 202, 220, 221
lighter-than-air, 151
origin, animal, 27
propellers, 47
safety factor, 244
size and scale, 45–48, 221–23
speeds, 46
vs. swimming, 220
see also aircraft
floors:
flatness, 59
stiffening, 76, 93
flow:
forces of, 70
heat convection, 125
laminar vs. turbulent, 52–53, 278, 290
non-Newtonian, 279
viscoelasticity, 116
 viscosity, 46
see also blood: flow
fluids:
as compression elements, 148
vs. solids, 83
flying, mechanisms, 212
flying squirrels, convergence, 37
flywheels, 200
foams, 77, 121, 122
foraminifera, 65, 305, 306
Ford, Henry, 35, 245
forging, 116, 291
fossils, size trends, 41
French, Michael, 137
frogfish, 216, 217, 220
frogs, lungs of, 210
fruit flies, speed of, 46
Fuller, R. Buckminster, 77, 146
Fulton, Robert, 275
fur, 43

gaits, switching, 201, 201, 202

Galbraith, John Kenneth, 33
Galileo, 39
gears, 80, 189, 190
geodesics, 72, 77, 78
gibbons, 137
giraffes, hearts of, 210
glass:
crack propagation, 79, 120, 118
in fiberglass, 122
strength, 118
thermal conductivity, 124
Glidden, Joseph, 267
gliding, 46–47, 272
animals, 27, 46–47, 167
plants, 273, 274
by Wrights, 260
Gordon, James E., 92, 93, 117
Gould, Stephen Jay, 34, 72, 188, 297
governor, 234
grasses, 34, 270
gravity, 46–47, 56, 58, 59, 65, 93, 132, 138, 148, 151, 154, 291
center of, 73, 91
energy storage, 198
pumping against, 209
waves, 225, 226, 226
Gray, James, paradox of, 278
greenhouses, 255
Griffiths, A. A., 118
growth:
bones, 24, 116, 295
cones, shells, 23, 24, 302, 304, 305
cylinders, 24
molting, 24
vertebrates, 24, 28
guns, 51

Haish, Jacob, 267

Haldane, J. B. S., 45
Hale, Melina, 99
hammers, 52, 80, 93, 116, 303
harvestmen, 55, 145, 145
heart, hearts:
control of, 234
as jet engines, 216
muscle action, 184
output, 213
as pumps, 196, 203, 209, 210, 211
transplanting, 247
heat:
from deformation, 115
from elastic loss, 203
in human technology, 43, 156, 158–60
in natural technology, 161
vs. size, 42
thermal efficiency, 160–61
Heilbroner, Robert, 299
helices, 26, 132
of fibers, 148, 150
self-assembly, 26
subcellular, 26, 231
helicopters, 220, 221
hemoglobin, 107
Henry, Joseph, 162
herbivory, 108
Hero of Alexandria, 157, 217
hexagons, 74, 75
hinges, 16, 105
Hooke, Robert, 263
horses, 89
boniness, 137
falling, 45
suspension, 142, 142
hull speed, 225–26, 226
humans:
aggression, 52
boniness, 137
 creativity, 293, 298
falling, 44
growth and size, 40, 43, 44
magnetite in, 109
number of cells, 230
protein half-life, 239
social organization, 292, 294
swimming, 225
thermal conductivity, 125
time course of technology, 294, 295–97
variation, 236
walking on water, 49
hummingbirds, 43, 220, 221
hydraulic ram, 214
hydraulics:
in human technology, 196, 282
in nature, 17, 194, 196, 203
principle, 195
transmissions, 193–98, 195, 197
hydroplaning, 226
hydropower, 164
hydroskeletons, 70, 151, 153–76, 149, 150, 194, 290

I-beams. See beams

icosahedra, 78
inertia, 51–53
information:
bit, defined, 25
in control systems, 233, 241
of egg, sperm, 25
of genes, 236
intra- vs. extracellular, 233
limitation, 25, 78
of organisms, 25–27
for regeneration, 239
insects:
flight, 24, 27, 45, 46, 202, 220, 221, 222
flight muscles, 182
growth, 24, 25
hexapedalism, 73, 283
resilin, 28
and surface tension, 48
wingbeat frequencies, 28, 62, 182, 202
wings, 62, 86
insulation, 43
interstitial animals, 113
intestines, 277
iron:
cookware, 124
 cost, 117
density, 112
in earth, 110
human history, 308
in humans, 108
oxidation, reduction, 111
in seawater, 110
use in nature, 107, 110

jaws:
leverage, 182
mechanisms, 70, 70
jellyfish:
locomotion, 209, 210, 216, 217, 219
reproduction, 24
stinging cells, 29, 174
strobili, 304
jet propulsion, 216, 228
efficiency, 217–20
engines, 158, 208, 215, 292
high speed, 219
in nature, 216, 219
pumps for, 210
scallops, 89
squid, 150
see also motors: turbines
joints, 191
jumping:
height, 39
take-off speeds, 52, 204

kangaroos:
gaits, 201, 202
jumping, 104, 204
Kelly, Michael, 266
kelp, 94, 95
keratin, 231
Kevlar, 286
kidneys, 196, 232
Kitchen, John G. A., 277
Koehl, Mimi, 95, 101
Koops, Matthias, 262
Kramer, Max, 278

LaBarbera, Michael, 188, 214

lambs, leg tendons, 88
Lanchester, Frederick, 260
land division, 74–75
Laplace’s law, 62–64, 63, 102, 210
lathe, 36, 303, 304
leaves:
as beams, 60, 61 drag, 96–99, 253
flatness, 60
flatness of, 58
herbivory, 108
reconfiguration, 96, 97, 98, 99
stems of (petioles) 99, 99, 185
thermal conductivity, 124
thermal relations, 125
of water lily, 255, 256
wilting, 184, 185
legs:
gaits, 201, 201, 202
number used, 73
vs. wheels, 104, 200
length. See size and scale
levers, 154, 178–82, 179, 181, 214
hydraulic equivalent, 195
natural vs. human, 179, 184, 290
as transmissions, 179, 180
life histories, 25
lift:
vs. drag, 46, 258, 272
vs. flying speed, 46, 222
origin, 260
power for, 221
vs. surface area, 46, 221
windmill blades, 165
ligaments, 28, 70, 88, 89, 104
lighthouses, 251, 252
Lilienthal, Otto, 258, 270, 272
linkages, 70
lizards, gliding, 27
lobsters:
blood vessels, 104
claw muscles, 183
exoskeletons, 106
lock-in, 301–2
locusts, gliding, 46

magnetite, 109
maintenance, 246, 281, 291
see also manufacturing
malleability, 113, 116
mammals:
body temperature, 43
bones, 55, 137
evolution, 34, 37
 grazing, 34
jaws, 182
neck beams, 141
neck ligament, 88
size, 43, 137
teeth, 34
manipulators, 281, 282, 285
manufacturing:
control systems, 245
in human technology, 244–48, 291, 299
mass production, 244
in nature, 229–39, 291
precision, 242–44, 246
product size, 243
repair, 239
routine maintenance, 237, 246, 291
mass production, 244
mass vs. volume, 42
materials. See specific ones, composite materials, etc.
bioemulatory, 280, 281, 285
materials science, 82
Maxim, Hiram, 271
mechanical advantage. See levers
mechanisms, 70, 71, 80
medicine, 32
metals, 16, 32, 106–17
conductivity, electrical, 126, 162, 291
conductivity, thermal, 124, 291
cracking, 115, 118
diversity, 116
fabrication, 81, 117
in human technology, 117, 290
metal-containing compounds, 108–9
not in nature, 106–13, 126, 290, 310
origin and history, 37, 308
plastic behavior, 114
pressing, 62, 291
properties, chemical, 110
properties, mechanical, 113–17
reduction of ores, 110
stiffness, 85
stress-strain graph, 114, 115
versatility, 134, 141
metamorphosis, 25
mice, 37, 45
microfilaments, 26
microphone, 266
microtubules, 26, 132, 169, 280
military:
design role, 33
 see also weaponry
milkweeds, 53
mist-netting, 17
mixotrichs, 29, 30
Mokyr, Joel, 299
molding, 80
moles:
convergence, 37
forearms, 181
mollusks, 23
abductin, 28, 87, 89, 143, 203
shells, 94, 167, 302, 306
suspension feeding, 213
see also scallops, squid, etc.
mortar, 68, 92, 93, 133, 290
motility:
cellular, 26
see also cilia, muscle, etc.
motors, 153–76, 290
combustion, 33, 38, 158, 160, 164, 174, 311
control, 234
efficiency, 161, 164, 174–75
electrical, 31, 161–63, 174, 186, 296, 297
energy sources, 153, 290
evaporative, 172
imbibitional, 172
linear vs. rotary, 162
muscle analogs, 280
muscles, cilia, etc., 168–72, 169, 170, 175, 186, 187, 232
nuclear, 224
osmotic, 173, 185
power output, 156, 158
rotational, 184, 186, 187
speed of operation, 184
steam, 35, 41, 139, 159, 217, 218, 245, 295
transmissions for, 178, 191–93
turbines, 41, 213, 215, 215, 219, 245, 292, 293
types, 156–58
wind and water powered, 36, 157, 163–68, 164, 165, 245
multicellularity:
development, 235
evolution, 230, 233
regeneration, 239
Mumford, Lewis, 299
muscle, muscles, 70, 89, 128, 134, 150, 151, 154
analogs, 280
antagonists, 203
control of, 234
efficiency, 175
fiber types, 171
 force vs. work, 170
half-life, 239
in hydroskeletons, 148
length change, 180, 184
locking, 29, 170
mechanism, 170, 232
as motors, 157, 169, 232
muscular hydrostats, 151, 183, 282
in neck truss, 141
pennate, 182, 183
power source for, 163
power output vs. time, 171
responsive growth, 240
size range, 176
squeezing heart, 184
transmissions for, 177, 181–84, 182, 183, 191–93
within exoskeletons, 145
muskrats, 51

nails, 68, 93, 116, 290, 291

nanotechnology, 280
natural selection. See evolution
nature, perfection of, 18–19
nature, copying. See copying nature
necks, 88, 89, 141, 141
Needham, Joseph, 307
nematocysts, 29–30
nerves:
conduction, 126, 163
diffusion between, 50
Newcomen, Thomas, 159
noria, 214, 215
nuchal ligament, 89
nudibranchs, 30
nylon. See fibers, Velcro

oak tree, 251

octopus:
arms, 151
blood vessels, 103
eyes, 37
jetting, 216
Oldenburg, Claes, 17
O’Neill, Patricia, 100
orthotetrakaidecahedron, 77
Orwell, George, 45
Osage orange, 266, 267
osmosis, 173–74, 173, 209, 210
Ovid, 249
packing, 68
paddle wheels, 211, 223, 275, 287
pandas, 72
Papanek, Victor, 19
papermaking, 261, 271
parachutes, 53, 54
Paxton, Joseph, 255
pendulums, 198, 201, 203
penises, 64, 147, 151, 196
pennate muscles, 182, 183
peristalsis, 277
Perspex. See Plexiglas
petioles, 99, 99, 185
Petroski, Henry, 300
Pettigrew, James Bell, 306
Phillips, Horatio, 258, 270
photosynthesis, 57, 109, 126, 152, 154, 168
piezoelectricity, 281
pigeons, gliding, 47
pipes:
resistance, 278
shapes, 64
see also blood vessels, pumps, tubes
plants:
evolution, 37
herbivory, 108
see also leaves, trees, etc.
plastics, 35
fabrication, 80
flexibility, 101
in human technology, 117, 285
Plexiglas, 118
pneumatics. See hydraulics
Portuguese man-of-war, 30, 148
postural reflexes, 91
potters’ wheels, 200, 309
pottery, 309
power, 155
bacterial flagellum, 187
for flight, 221
motors, 156, 158
muscle, 163, 171
in pumps, 206
prairie dogs, 166, 166, 208, 209, 213
preadaptations, 36
preloading, 143
pressure:
aircraft fuselages, 64
blood, 102
within cells, 210
 vs. depth, 210
evaporative pump, 172
across flat surfaces, 62
Laplace’s law, 62
osmotic, 173
in pumps, 206–14
see also hydraulics
prestressing, 143
privilege of incumbency, 301–2
propellers, 47, 186, 272, 284, 288, 288
airfoils, 221
making thrust, 215
vs. oars, 211
propulsion efficiency, 218
as pumps, 207
twisted blades, 99
propulsion efficiency, 218
prosthetic materials, 280, 285
proteins:
half-life, 238
instability of, 238
size, 230
synthesis, 233
tolerable variation, 243
turnover, 237
see also collagen, resilin, etc.
protozoa. See foraminifera, mixotrichs, etc.
pumped storage, 199
pumps, 205–16, 228
Archimedean screw, 288
evaporative, 208, 209
fluid dynamic, 207, 208, 209, 211–13, 212, 292
in nature, 208–16, 209, 211, 212
osmotic, 173–74, 173, 209, 210
peristaltic, 211, 216, 277
positive displacement, 207, 208, 211–13, 209, 211, 292
pressure vs. flow, 207, 208, 215
pulsatility, 102
for suspension feeding, 212
valve-and-chamber, 209, 210
punctuated equilibrium, 34
pyramids, 23, 65
Pythagorean theorem, 75, 76

QWERTY, 301

radula, 108, 108

rasps, 108, 254
Réaumur, René-Antoine, 262, 263
rectangles, 71
see also right angles
regeneration, 239
resilience, 86, 87
abductin, 28, 89, 203
collagen, 202
elastin, 28
metals, 114
relative utility, 89
resilin, 28, 203
rubber, 86
spider silk, 89
trees, 90
resilin, 28, 28, 203, 204
responsive materials, 281
Riblets, 279
right angles, 16, 32, 62, 64–78, 81, 289
in human technology, 64
land division, 74
in nature, 65
semicircular canals, 65, 66
rigidity. See stiffness
robotics, 246, 250
robots, 281, 297
Robotuna, 284
rockets. See jet propulsion
rods, 17
roofs, 57
corrugated, 17
Crystal Palace, 255
domed, 17
flat, 58
peaked, 138
ridge-and-valley, 17, 61, 255
vaulted, 58, 134
rope, 87–88, 120, 152, 310
construction, 131
kelp as, 95
natural, 70
sagging, 59
strength vs. stiffness, 87
as tensile element, 128, 134
rotary motion, 16, 162, 188, 192, 193
see also motors, transmissions, wheels
rotating trapezoidal illusion, 66, 67
Roux, Wilhelm, 236
rubber:
in human technology, 117
resilience, 86
stiffness, 84, 88
 stress-strain graph, 103, 104, 148
as tensile element, 134
Rumsey, James, 275
running:
vs. cycling, 188
energy storage, 104, 200, 201, 202
vs. walking, 201–2, 201
rusting, 111

safety factors, 55, 243–44, 247, 248

sagging, 59, 92
sailplanes, 46
sails, sailing, 157, 293
samaras, 54, 274
sand dollars, 167
sap lifting, 172, 208, 210
satellites, 136
saws, 94, 108, 249, 267, 268
see also chain saws
scale. See size and scale
scallops:
abductin, 28, 87, 89, 143, 203
corrugated shells, 17, 17
jet propulsion, 89, 216, 217
swimming, 28, 203
Schäffer, Jacob Christian, 262
Schoenheimer, Rudolph, 237
Schwabe, Louis, 263
science in technology, 293
sea anemones:
flexural stiffness, 101
hydroskeleton, 70
reproduction, 24
sea cucumbers, 147
sea spiders, 145
sea squirts. See ascidians
sea urchins, 149, 281
seaweeds. See kelp
seeds:
autogyrating, 53
flinging, 90
hydration, 173
parachutes, 53
self-loading, 60
semicircular canals, 65, 66
service life, 246
Shadwick, Robert, 103
sharks:
hydroskeleton, 149
 skin scales, 279
shearing, 290, 291
in column, 132
in ropes, 131
shearing load:
and flexural stiffness, 130
and torsional stiffness, 131
shell, shells:
as composites, 123, 281
density, 112
egg, 124
growth, 23, 24, 113, 302, 304, 305
loads on, 135
mollusk, 17, 17, 23, 94, 106, 124, 167, 241, 305, 306
as shipworm’s rasp, 253, 254
toughness, 118
see also domes, skulls, etc.
Sherrington, Charles, 242
ships. See boats
shipworm, 253, 254
shock absorbers, 90, 90, 197
shrews, night torpor, 43
silk, spider, 90, 128, 132, 286–87
properties, 91, 286
resilience, 89
strength, 92
silkworms, 132, 263–65, 264
size and scale, 23, 39–56, 291
bioemulation, 285
blimps in nature, 152
bookshelves, 60
cells, 229
and diffusion, 50–51
falling, 44
flight, 45–48, 221–23
gravity, 51
heat gain and loss, 42
human growth, 43, 44
human technology, 41, 56
inertia, 51–53
length, surface area, volume, 41–48, 42
nanotechnology, 280
size ranges, 39, 40
structures, 53–56
and surface tension, 48–50
ties vs. struts, 135–36, 144–46
viscous effects, 215
waves, 51, 226
wheels in nature, 191
skeletons:
 arthropods, 145, 145
growth, 18, 24, 295
horse, 89
sponge, 72, 73
turtles, 28
see also bone
skin, stress-strain graph, 104
skulls, 17
see also shells
smart materials, 281
Smeaton, John, 251
Smith, Francis Pettit, 287
Smith, John Maynard, 272
snails, radulas of, 108, 108
snakes, jaws of, 70
soaring, 47, 167
Social Darwinism, 242, 300
solids, 83, 84
Sotavalta, Olavi, 202
spheres:
falling speeds, 45
Laplace’s law, 62–64, 63
spicules, 72, 146
spider crabs, 145
spiders:
feeding, 17
hydraulic legs, 17, 196
molting, 24
spinning, 117
webs, 17, 59, 89, 90, 286
see also silk
spinneret, 264
spinning, 131, 265, 308
spirals, 302–7
logarithmic, 305–7, 305, 306
in nature, 306
spirochetes, 29, 30
sponges:
filtration by, 167, 208, 212, 213
pumping, 209, 212
skeletons, 72, 73, 146
springs, 86, 199–200
automobiles, 142–43, 142
in nature, 142–43, 142
resilience, 90, 114
square. See right angles
squid:
blood vessels, 103
giant, 104
hydroskeleton, 149
 jetting, 150, 210, 215, 216, 217, 219
tentacles, 151, 183, 282
squirrels, flying, 27
stacking, 65, 68, 138
stamping, 52, 81
starfish:
regeneration, 240
skeleton, 121
tube feet, 149, 196
twisty arms, 100
statically determined structures, 71
steam engines. See motors: steam
steel, 295
properties, 86, 92, 127
stress-strain graph, 114
stems, close packed cells, 77
Stevenson, Alan, 253
stiffness, 17, 84, 85, 87
of blood vessels, 102
of compressive elements, 132
crack propagation, 79, 93
vs. density, 126
honeycomb reinforcement, 76
in human technology, 82, 92–94, 92, 290
hydroskeletons, 151
and material economy, 93
of metals, 115
in nature, 82, 92–94, 290
non-constancy, 85
vs. other properties, 92
of ropes, etc., 87–89
of tensegrities, 147
of tensile elements, 139
vs. toughness, 94, 118
of various materials, 92
see also flexural stiffness, torsional stiffness
stone, stones, 133, 134, 253
brittleness, 117
in human technology, 117, 247
strain, 84, 115
streamlining, 257, 258, 270, 271, 290
strength, 84, 85, 87
in human technology, 92–94
of metals, 116
in nature, 92–94, 92
vs. other properties, 92
of ropes, 88
stress, stresses, 83, 84, 129, 129, 130, 131
stress-strain graphs, 83, 84, 84, 85, 87
biological materials, 104
 blood vessels, 86, 103
bone, 114, 116
metals, 114, 115
rubber, 103, 104, 148
skin, 104
tendon, 86, 86, 104, 114
stretch. See strain, stiffness
struts, 70, 72
see also compression elements
submarines, 210, 223–28, 258, 278, 283, 291, 302
surface area. See size and scale
surface ships. See boats
surface swimming, 223–28
surface tension, 48, 289
and insects, 48
vs. size, 48–50
water striders, 48, 119
waves, 51, 226
surfaces:
corrugated, 17
energy of, 119
flat vs. curved, 57–64,289
surveying, 74–75
suspension feeding, 209
ascidians, 107, 107
mollusks, 213
pumps for, 212
sponges, 167, 208, 212, 213
swimming:
at surface, 223–28
hydroplaning, 226
jet propulsion, 17, 219
mechanisms, 212

Taylor, Frederick W., 245

technology transfer, 29–31
teepee, 66
teeth, 106
as composites, 123
enamel, 94, 280
gears, 80
of herbivores, 34, 108, 295
mammals, 34
molluscan, 281
tensile loading, 135
wear, 241, 295
telephone, 31, 37, 265, 266
temperature:
body, 43, 238
 and chemical reactions, 235
and heat engines, 160–61
tendon, tendons, 70, 128, 134
Achilles, 89
attachments, 138
energy storage, 104, 200, 202
half-life, 239
microstructure, 231
properties, 91
resilience, 91
safety factor, 244
stiffness, 88, 93
stress-strain graph, 86, 104, 114
tensegrity, 146, 147
tensile elements, 128, 134, 152
vs. compressive elements, 135
nature’s preference, 135, 290
sheaths as, 148
size and shape, 135, 145
stiffness, 139
tension:
in column, 133
and corners, 79
in curved surface, 62
and joinery, 68
in string, 63
testing, 129
Teredo, 253, 254
termites:
intestinal fauna, 29
mound ventilation, 167
textiles. See fibers, weaving
thermal efficiency, 160–61
thermodynamics:
first law, 155, 286
second law, 156, 160
Thompson, D’Arcy, 41, 137, 141, 306
thorns, 266, 267, 271
ties. See tensile elements
tin, 107, 109, 111
tits, 32
toilets, 233, 235
tongues, 151, 282
toolmaking, 32, 241, 294, 309
torsion, 36, 100, 261
torsional stiffness, 98, 99–100, 131
toughness, 84, 86, 87
vs. brittleness, 290
metals, 113
vs. other properties, 92
 shell, 118
vs. stiffness, 118
towers, 146, 147
transformers:
electrical, 163, 214
fluid mechanical, 214–16, 219, 291
transmissions, 153, 154, 177–86, 180, 190
hydraulic, 193–98, 195, 197
levers as, 179, 181
for muscles, 177, 182, 183, 191–93, 194
wheels in, 189
transpiration, 172, 208
trapezoids, 66, 67
trees:
as beams, 253
damping, 90
lifting sap, 172, 208, 210
resilience, 90
responsive growth, 240
safety factor, 244
trunks, 17, 65, 128, 137, 144, 145
see also leaves, wood, etc.
triangles:
in human technology, 71
in nature, 71
surveying, 74
trusses, 61, 71, 140, 141, 142
see also beams
tubes, 17
see also cylinders, pipes
tumbleweeds, 188
tunicates. See ascidians
tunneling, 253, 254, 268
turbines. See motors
turtles, skeleton of, 28
Twiddle-fish, 284, 284, 285
twisting, 98, 131
see also torsion, torsional stiffness
typewriters, 199, 296, 301

undulipodia, 29

vanadium, 107, 110, 111

vehicles, 16, 35
legged, 74, 250, 283, 283
see also aircraft, automobiles, etc.
Velcro, 268, 269, 270, 271
vertebrates:
elastin, 28
 evolution, 35
eyes, 37
growing skeleton, 24, 28
VHS videotape system, 301
Victoria amazonica, 255, 255
Vincent, Julian, 231
virus coats, 78
viscoelasticity, 116
viscosity, 46, 53
and pumping, 277
vs. scale, 52, 215
vocal cords, 90
volume. See size and scale

walking, 73
vs. cycling, 188
energy storage, 200, 201
vs. running 201–2, 201
vehicles, 250, 283, 283
on water, 48–49
wasps, 76, 261
water:
composition, 110
for cooling, 125
density, 42, 112
water lily, 255, 255
water striders, 48, 49, 119, 223
water wheels, 157, 163, 164, 245
see also motors
Watt, James, 35, 160, 191, 234, 293
waves, surface:
capillary, 51, 227, 227
gravity, 51, 95, 224–27, 225, 226, 258
surface tension, 51, 227, 227
weaponry, 31, 52
archery, 37, 127, 194, 200
catapults, ballistae, 127, 199, 200
firearms, 51
longbows, 127
weaving, 37, 308, 310
weight, 42, 51
see also gravity
Weis-Fogh, Torkel, 203
whales:
feeding, 17
largest, 40
streamlining, 257, 258
swimming, 224, 284, 288
Wheatstone, Charles, 162
wheels, 16, 162, 166, 186–91, 290, 311
advantages, 189, 204
bacterial flagella, 186, 187
cultural factors, 189
disadvantages, 188, 189
history, 188
vs. legs, 105
in machinery, 189
on Mesoamerican toys, 189
potters’, 200
see also bearings
Whitney, Eli, 37, 245
wind:
over burrows and mounds, 166
and flight, 46
on leaves, 96
on wings, 62
windlass, 179, 180
windmills, 36, 157, 163, 164, 165
blades, 97
see also motors
windows, 67, 67, 80
windthrow, 244
wings. See airfoils
wires, 126, 291, 296
wood, 106, 253
applications, 91
as composite, 123, 231, 232, 280
digestion, 268
fresh vs. dry, 92
growth, 113
in human technology, 117, 247
joinery, 138
loads on, 135
making paper from, 261
preloading, 144
properties, 127
stiffness, 92
strength, 92, 145
thermal conductivity, 124
work, 154
work of extension, 84, 86, 87
see also toughness
worms:
hydroskeletons, 70, 151, 196
peristaltic pumping, 211, 211
shapes, 64
tubes, 17
Wright, Wilbur and Orville, 37, 260, 270, 272
Young’s modulus. See stiffness
Young, Thomas, 85

Zulu, 66
More praise for Cats’ Paws and Catapults

“[Steven Vogel] writes with unusual recognition of the needs of the inexpert
reader.”
—Philip Morrison, Scientific American

“Unceasingly, [Vogel] advocates the fun of science.”

—Peter Gorner, Chicago Tribune

“If [D’Arcy] Thompson’s work defined the classical period in the science of
form, the field has just entered its renaissance. One of the leaders of the
resurgence is the zoologist Steven Vogel.”
—Tyler Volk, The Sciences

“[F]ew scientific books show; most scientific books tell, recite strange names,
and wallow in complexity. But over the years I have come to expect that Steven
Vogel will always show me science. . . . Any of [Vogel’s] books could entertain
an expert or an amateur.”
—Mike May, American Scientist

“Who is the better technologist, Mother Nature-source of seashells, spider webs,

and birds’ wings—or the human engineer—creator of skyscrapers, nylon, and
airplanes? This engrossing question lies at the heart of a fine new book by
Steven Vogel, an expert in biomechanics with a flair for genial philosophizing.”
—Samuel Florman, author of The Civilized Engineer and The Existential
Pleasures of Engineering

“Full of ideas and well-explained principles that will bring new understanding of
everyday things to both scientists and non-scientists alike.”
—R. MCNEILL ALEXANDER, NATURE
“This elegant comparison of human and biological technology will forever
change the way you look at each.”
—MICHAEL LABARBERA, AMERICAN SCIENTIST
“Perfect for the lay reader.... It is Vogel's particular genius to illustrate the
principles of structural forces.”
—M. R. MONTGOMERY, NEW YORE TIMES BOOK REVIEW
Cats’ Paws and Catapults is about the ways living things work—and walk, run, jump, fly, and grow.
Inviting the reader into the surprising world of biomechanics, Steven Vogel explains how physical
law, size, and historical accident work together to determine both nature’s designs and the things
that we make.
Nature and human designers share the same physical environment, yet their designs have turned
out to be wildly dissimilar. Our technology, for example, goes around on wheels—and on rotating
pulleys, gears, shafts, and cams—yet in nature only the tiny propellers of bacteria spin as true
wheels. Our hinges turn because hard parts slide around each other, whereas nature’s hinges (a
rabbit’s ear, for example) more often swing by bending flexible materials. Why did these
“technologies” take such separate paths, and what can we learn from their differences?
 Copyright © 1998 by Steven Vogel

All rights reserved

First published as a Norton paperback 2000

For information about permission to reproduce selections from this book, write to
Permissions, W. W. Norton & Company, Inc., 500 Fifth Avenue, New York, NY 10110.

The text and display of this book were composed in Adobe Garamond
Desktop composition by Tom Ernst
Book design by BTD/Mary A. Wirth

The Library of Congress has cataloged the printed edition as follows:

Vogel, Steven, 1940-
Cats’ paws and catapults : mechanical worlds of nature and people
/ by Steven Vogel ; illustrated by Kathryn K. Davis,
p. cm.
Includes bibliographical references and index.
ISBN 0-393-04641-9
1. Biomechanics. 2. Mechanics. I. Title.
QH513.V64 1998

571.4—3-dc21 97-44807
CIP

ISBN 0-393-31990-3 pbk.

ISBN 978-0-393-35295-5 (e-book)

W. W. Norton & Company, Inc., 500 Fifth Avenue, New York, N.Y. 10110
www.wwnorton.com

W. W. Norton & Company Ltd., Castle House, 75/76 Wells Street, London W1T 3QT