1. What Is Knot Theory?

Why Is It In Mathematics?

In this chapter, we briefly explain some elementary foundations of knot theory. In 1.1,
we explain about knots, links and spatial graphs together with several scientific
examples. In 1.2, we discuss diagrams of knots, links and spatial graphs and
equivalences on knots, links and spatial graphs. Basic problems on knot theory are
also explained there. In 1.3, a brief history on knot theory is stated. In 1.4, we explain
how the first non-trivial knot is confirmed. In 1.5, the linking number useful to confirm
a non-trivial link and the linking degree which is the absolute value of the linking
number are explained. In particular, we show that the linking degree is defined directly
from an unoriented link. In 1.6, some concluding remarks on this chapter are given. In
1.7, some books on knot theory are listed as general references.

1.1 Knots, links, and spatial graphs

A knot is a tangled string in Euclidean 3-space R3 which is usually considered as a
closed tangled string in R3, and a link is the union of some mutually disjoint knots (see
Figure 1). The AYATORI game (= Cat's cradle play) let us know that a given knot can
be deformed into various forms and we feel that it is a difficult problem to judge
whether any given two knots are actually the same knot or not.

Figure 1: Knots and links

A spatial graph is the union of some strings with endpoints in R3 which are mutually
disjoint except the endpoints. A vertex of a spatial graph is a point in gathering more
than 3 strings and an edge of is a connected component of the strings obtained from
by cutting along all the vertices of . For example, Kinoshitas -curve in Figure 2 is a
spatial graph with two vertices and three edges.

Figure 2: Kinoshitas -curve

To find out a knot or a link phenomenon in natural science, at first a setting of an object
which one can consider as a string becomes important. Here, we show some examples
about such objects.
Example 1: It is possible to think a chain (see Figure 3) as one string if roughly seeing,
but also as a link which are twined round one after another like a string if a little more
minutely seeing.

Figure 3: A chain
Example 2A 3-braid is a knitting of three strings (see, for example, Figure 4). Since
this knitting pattern is known to be used in a pottery of the JOMON period (an ancient
time) in Japan, we see that the people of the JOMON period might understand the fact
that the 3-braid is a technology of making a long, strong string like a rope from some
short strings like straws. Then, by joining a and a, and b and b in Figure 4, and then by
deforming it (without changing near the ends), we can make AWABI MUSUBI (= the
abalone knot) of "MIZUHIKI" in Figure 5 used for the custom of the present in Japan
from ancient times. Also, although it is a little more difficult, we can also make the
same knot by joining b and b, and c and c in Figure 4, and then by deforming it
(without changing near the ends). In this way, the knot is also an interested study object
for cultural anthropology.
2

Figure 4A 3-braid

Figure 5

AWABI MUSUBI(=abalone knot) of "MIZUHIKI"

Example 3Assume that there are n particles moving without colliding in the plane
with a time parameter. Then the track of the motion forms an n-braid in the
three-dimensional space which is the product space of the plane and the time axis. In the
the case of 3 particles, we have a 3-braid like the example in Figure 4. Every knot and link
can cause from an n-braid by taking the closure.
Example 4When we consider DNA as one string long rope, there is one which
becomes a closed curve called a DNA knot (see Figure 6).

Figure 6A DNA knot (Acknowledgement :Professor N. R. Cozzarelli)

Example 5A molecular graph on a molecule in chemistry is a spatial graph whose
vertices correspond to the atoms in the molecule and whose edges express the
combination data between the atoms by bonds. Topology of molecular graphs has begun
to attract attention in researches of the synthetic-chemistry.

Example 6A protein molecule is considered to be a string consisting of amino acid

bases. Some protein molecules such as a prion protein of the Creutzfeldt-Jacob disease
and an amyloid protein of the Alzheimer's disease appear to be more or less related to
knot theory.
Example 7 A large scale structure on the cosmos is recently known to the
astrophysicists.
Example 8: The seismometer is a machine which draws the track of an
earthquake-motion of an observation point as a spatial curve, called an earthquake
curve with the time parameter. The analysis of this earthquake curve can be considered
as knot theory in the wide sense.

1.2

Diagrams and equivalence on knots, links, and spatial graphs

Figure 7

A crossing

A knot is in the three-dimensional space and we think that it is made of a very thin
string. We present it by a plane curve with only double crossings as they are shown in
Figure 7, which we call a knot diagram or simply a diagram. For a link, it is similarly
presented and called a link diagram or simply a diagram. By knots and links, we
mean their diagrams unless making confusion. A spatial graph is also presented to the
plane with only double points on the edges which we call a spatial graph diagram (see
Figure 3). For two knots, we say that they are the same knot or equivalent knots if we
can deform them into the same shape in the manner of AYATORI game (= Cat's cradle
play), i.e., by a finite number of Reidemeister moves I-III in Figure 8. For two links, we
say similarly that they are the same link or equivalent links if we can deform them into
the same shape in the manner of AYATORI game, namely by a finite number of
Reidemeister moves I-III. For two spatial graphs, we say that they are the same graph or
equivalent graphs if we can deform them into the same shape in the manner of
AYATORI game or in other word by a finite number of Reidemeister moves I-V in
Figure 8. A knot is called a trivial knot if it is equivalent to a circle in the plane like a
rubber band. Also, a link is called a trivial link if it is equivalent to an union of
separated trivial knots.
4

Figure 8: Reidemeister moves

Figure 9

The same arc knot

In daily life, we normally think of an arc knot in Figure 9. At that point, we consider
that the end points extend virtually long and endless. If we can deform them into the
same shape in the manner of AYATORI game without moving the parts that are
extended, then they are considered to be the same knot. A link with two end points can
be considered similarly to the case of an arc knot (See Figure 10). If a link has more
than two end points, then such a link no longer has any meaning unless the data
extending the end points are definitely given.

Figure 10A link with two end points

The main purpose of knot theory to solve the following two problems (which are related
to each other):
5

Equivalence Problem Given two knots (or links,

whether or not they are equivalent.

spatial graphs),

determine

Classification ProblemEnumerate all the knots (or links, spatial graphs) up to

equivalences.
To solve these problems, it is very important to develop topological invariants,
(namely, quantities which are invariant under the Reidemeister moves) of knots, links
and spatial graphs.

1.3 A brief history of knot theory

Knot theory is now believed that a scientific study to be associated with the atomic
theory of vortex atoms in ether around the end of the nineteenth century. However, it is
can be traced back to a note by J. B. Listing, a disciple of Gauss in 1849. In the note, it
has been that the mirror image of the figure-eight knot is the same knot (Figure 11). Up
until the 1930s, important researches were made by K. Reidemeister and H. Seifert in
Germany and J. W. Alexander in U.S.A., etc. From 1940 until the 1970's, one may say
that basic mathematical theory on knot theory was established with R. H. Fox in U.S.A.
as a center. In Japan, from around 1960, H. Terasaka, T. Homma, S. Kinoshita (later,
moving to U.S.A.) and K. Murasugi (later, moving to Canada), F. Hosokawa, etc. have
begun to make contributions to knot theory. From around 1980, knot theory came to
attention not only in almost all areas of mathematics, but also in the fields of science
that will be cutting-edge researches, such as gene synthesis, quantum statistical
mechanics, soft matter physics, biochemistry, polymer network, applied chemistry.
The international conference Knot theory and related topics received the world's first
was held at Osaka as a satellite conference of ICM Kyoto in 1990, from whose
proceedings Knots 90 (A. Kawauchi, e.d., Walter de Gruyter, 1992) one may feel a
fever of an expansion of knot theory.

Figure 11 : Equivalence of the figure-eight knot and the mirror image

1.4 The first non-trivial knots

Of knots and related concerns that are normally used casually in everyday life, let me
say here the simplest mathematical proof that there is a non-trivial knot. (This proof
uses an argument of the 3-colorability of a knot which is well-known to the experts of
knot theory as, algebraically, the representation theory of the knot group to the
symmetric group of degree three or as, topologically, the theory of three-fold branched
coverings of a knot.) Given a knot diagram, we color all the edges connecting the
crossings by using three colors (e.g., red, blue and yellow) by imposing in the vicinity
of every crossing the condition that we color the upper arc by the same color and color
the lower two edges by the same color or different colors so that one color or three
colors are used in the vicinity of every crossing. By this method, we can always color
every knot diagram by one color. For some knot diagrams, we can also color them by 3
colors (see Figure 12). Such a knot is called a 3-colorable knot.

Figure 12The trefoil knot is 3-colorable

Theorem

Every 3-colorable knot is a non-trivial knot.

The reason why this is true is because we can easily check that any knot diagram D
transformed from a 3-colorable knot diagram D by Reidemeister moves I, II, III (see
Figure 8) is also 3-colorable. We recommend to confirm this fact for various knots.

1.5 Understanding the linking number

The linking number of a link of two oriented knot components is the most fundamental
topological invariant in knot theory. However, because it takes a value in the integers,
this invariant cannot de defined without the notion of a negative integer. In this section,
we first define the linking number. Next, we introduce the absolute value of the linking
number, called the linking degree and computable directly from a link diagram without
using the notion of a negative integer.

Figure 13

A link diagram

1.5.1Linking number
We shall explain how to define the linking number of an oriented link. We consider
a link L of two oriented knot components as in Figure 13. Every crossing between the
distinct knot components coincides with one of the four crossings shown in Figure 14,
where the signs of two crossings in the left hand side are defined to be +1 and the signs
of two crossings in the right hand side are defined to be -1. Let m be the sum of signs of
all the crossings between the distinct knot components. We see that this integer m is
always even and does not change under Reidemeister moves IIIIII (see Figure 8),
namely m is a topological invariant. Then we define the linking number v of the link L
to be the integer m/2. When we reverse the orientation of one component of L, the sign
of every crossing between the distinct components is changed so that the linking
number v of L is changed into v. For example, the link diagram in Figure 13 has the
two (+1)-crossings and the four (-1)-crossings and the linking number v is given by
v = (2 - 4) / 2 = -1.

Figure 14 The sign of a crossing

1.5.2Linking degree
We consider a diagram of an unoriented link L with knot components J and K as in
Figure 15. To distinguish between J and K, we denote K by a bold line. We attach an
orientation to one of the components J and K, say J as shown in Figure 16. Every
crossing between the oriented component J and the unoriented component K coincides
with one of the two crossings in Figure 17. We construct a meridian loop, an oriented
loop surrounding K one time in every crossing between J and K as in Figure 18. For
8

example, we obtain the left-sided diagram in Figure 19 from Figure 16, from which
we obtain the right-sided diagram in Figure 19 by sliding these meridian loops along K.

Figure 15 An unoriented diagram

Figure 16 A link diagram with the component J oriented

Figure 17 The situations of a crossing

We denote by (P,Q) a pair of the subsets of loops with the same orientations in the set of
meridian loops around K obtained from the oriented knot J. Let p and q be the numbers
of the elements in P and Q, respectively, where we take pq. Then let n=q-p, and the
linking degree d of a link L consisting of the unoriented knot components J and K is
defined by
d = n/2 = (q-p)/2.
For example, the linking degree d of the link L in Figure 15 is computed from Figure 19
to be
d = (42)/2 = 1.
If we take the reversed orientation on J, then the orientations of all the meridian loops
are reversed and the difference n between the numbers of the meridian loops with the
same orientations is unchanged, so that the linking degree d is independent of a choice
of any orientation on J.

Figure 18: Constructing a meridian loop of K

Figure 19
We show that the same number n is obtained as the difference of the numbers of the
meridian loops with the same orientations even if we consider the meridian loops
around J obtained from the component K oriented in any direction (instead of J). In fact,
if we consider J and K as oriented knots, then we see that the orientation of the meridian
loop is locally determined as it is shown in Figure 20, so that a pair of the numbers of
the meridian loops around J with the same orientations obtained from the oriented knot
K coincides with the pair (up to ordering) of the numbers of the meridian loops around
K with the same orientations obtained from the oriented knot J. This implies that the
linking degree d of a link L is a non-negative rational number with denominator 2 which
is independent of choices of the components and the orientations of J, K.

Figure 20: The orientation of a meridian loop is locally determined

We show the following (1) and (2).
10

(1) The linking degree d is unchanged under Reidemeister moves IIIIII and
hence it is a topological invariant.
(2) The linking degree d takes a value in the natural numbers or zero. When we orient
a component J, the linking degree d is computed to be the difference of the numbers of
the meridian loops with the same orientations around the other component K on the
crossings of J which are upper than K, or the difference of the numbers of the meridian
loops around K with the same orientations on the crossings of J which are lower than K.
First, we show (1). The proof is made by considering a pair of the numbers of the
meridian loops with the same orientations around K obtained from an oriented J. The
numbers n and d are unchanged under Reidemeister moves except the moves relating to
both J and K. In particular, they are unchanged by the Reidemeister move I. For the
Reidemeister move II relating both J and K, we can see from Figure 21 that the numbers
n, d are unchanged under the Reidemeister move II.

Figure

21:
Reidemeister move II relating to J and K

For the Reidemeister move III relating both J and K, it is sufficient to consider the cases
in the left-hand side of Figure 22, if necessary, by changing the roles of J and K. By
examining the pictures in the right-hand side of Figure 22, we see also that the numbers
n, d are unchanged under the Reidemeister move III, showing (1).

Figure 22: Reidemeister move III relating to J and K

11

Next, we show (2). We can see from Figure 20 and the definitions of the linking number
and the linking degree that the linking degree d is the absolute value of the linking
number v. Therefore, if we assume that the linking number v takes a value in the
integers, we can see that the linking degree d takes a value in the natural numbers or
zero. Here, we give a direct proof about it. Let (P,Q) be a pair of the subsets of loops
with the same orientations in the set of meridian loops around K obtained from an
oriented knot J. Let E and F be the subsets of P obtained from the crossings of J upper
and lower than K, respectively. Let e and f be the numbers of the elements of E and F,
respectively. Also, let G and H be the subsets of Q obtained from the crossings of J
upper and lower than K, respectively, and g and f the numbers of the elements of G
and H, respectively. Then for the numbers p, q of the elements of P and Q, where we
assume pq, we have
p=ef,
q=gh.
We change all the lower crossings of J into the upper crossings of J, so that all the
crossings of J are upper than K by a crossing change operation in Figure 23. By this
change, the numbers p=e f and q=gh of the meridian loops with the same
orientations are changed into e+h and g+f, respectively (cf. Figure 18). Since now the
knot diagram of J is upper than the diagram of K, we can move the knot diagrams of J
and K by Reidemeister moves IIIII so that they do not meet. By the topological
invariance of the linking degree, we have
(eh) (gf) = 0, namely g-e = h-f.
Hence, we have
d = n/2 = (q-p)/2 = g-e = h-f
and 2is proved.

Figure 23: A crossing-change operation

1.5.3Computing examples of the linking degree using the meridian loops on the
upper crossings
Here are some examples of computations on the linking degree.
Example 0We introduce an orientation on J as in Figure 16 to compute the linking
degree of a link in Figure 15. Then the meridian loops on the upper crossings belonging
12

to J are the loops in Figure 24. Thus, we have d = 2-1 = 1.

Figure 24: The meridian loops on the upper crossings of the diagram of Figure 16
ExampleThe linking degree d of the Hopf link in Figure 25 is computed to be
d = 1.

Figure 25: A computing process for the Hopf link

Example 2The linking degree d of the Whitehead link in Figure 26 is computed to be
d = 1-1 = 0.

Figure 26: A computing process for the Whitehead link

Example 3The linking degree d of a 2-braid link in Figure 27 is computed to be
d = 2.

Figure 27: A computing process for a 2-braid link

Example 4The linking degree d of the parallel link of a trefoil knot in Figure 28 is
computed to be d = 3.
13

Figure 28: A computing process for the parallel link of a trefoil knot
Remark. The linking degree d of a twisted parallel link of a trefoil knot in Figure 29 is
computed to be d = 3-3 = 0.

Figure 29:

A twisted parallel link of a trefoil knot

Example 5The linking degree d of a parallel link of the figure-eight knot in Figure 30
is computed to be d = 2-2 = 0.

Figure 30: A computing process for a parallel link of the figure-eight knot

1.6 Conclusion
It is an ultimate purpose of knot theory to clarify a topological difference of knot
phenomena in mathematics and in science. In this study, a building power and a
computational ability in mathematics are needed in addition to the intuition power
having to do with a figure. We can watch a knot with eyes and our ability of space
perception will be grown up by playing with it. Knot theory is a subject suitable for
understanding nature deeply and desirable for learning in an early age.

14

1.7 References
There are many books on knot theory some of which are listed here. Among them, the
first appearing book is Knotentheorie written in German in 1933 by K. Reidemeister.
[1] C. C. Adams, The Knot Book, American Mathematical Society, 2004.
[2] G. Burde, H. Zieschang, Knots, de Gruyter, 1986.
[3] P. R. Cromwell, Knots and Links, Dover Publications, 2008.
[4] R. H. Crowell & R. H. Fox, Introduction to Knot Theory, Springer-Verlag,
1977 (This book was first appeared from Ginn and Co. in 1963).
[5] L. H. Kauffman, Knots and Physics, World Scientific, 1991.
[6] L. H. Kauffman, Formal Knot Theory, Dover Publications, 2006 (This book was
first appeared from American Mathematical Society in 1983).
[7] A. Kawauchi, A Survey of Knot Theory, Birkhauser Verlag, 1996
(This book is an expanded version of the book Knot Theory published in Japanese
from Springer Verlag, Tokyo in 1990).
[8] W. B. R. Lickorish, An Introduction to Knot Theory, Springer Verlag, 1997.
[9] C. Livingston, Knot Theory, Mathematical Association of America Textbooks, 1996.
[10] V. Manturov, Knot Theory, Chapman & Hall/CRC, 2004.
[11] K. Murasugi, Knot Theory and Its Applications, Birkhauser Verlag, 2004 (This
book was first appeared as a Japanese book with the same title from Nippon Hyoronsha
Co. Ltd. Publishers in 1993).
[12] K. Reidemeister, Knotentheorie, Ergebn. Math. Grnnzgeb. 1, Springer Verlag,
1933.
[13] D. Rolfsen, Knots and Links, American Mathematical Society, 2003 (This book
was first appeared from Publish or Perish, Inc. in 1976).

15

Index
(p.1) knot
link
AYATORI game (= Cat's cradle play)
(p.2) spatial graph
vertex
edge
Kinoshitas -curve
braid
AWABI MUSUBI (= the abalone knot)
(p.3) DNA knot
molecular graph
(p.4) earthquake curve
knot diagram
link diagram
spatial graph diagram
same knot
Reidemeister moves
equivalent knots
same link
equivalent links
same graph
equivalent graphs
trivial knot
trivial link
(p.6) Equivalence Problem
Classification Problem
topological invariants
figure-eight knot
(p.7) 3-colorable knot
trefoil knot
(p.8) linking number
meridian loop
(p.9) linking degree
(p.13) Hopf link
Whitehead link
16