CLASSICS

A lot has been written about Ramanujan's unconventional story and how his talent found an expression
through G H Hardy who invited Ramanujan to England, and collaborated with him. The third leg of the
triangle at that time was Littlewood. This review by Littlewood of Ramanujan's collected papers published
after the latter’s death is unusual. Littlewood takes occasion to make enlightening remarks about what he
thought of specific talents of Ramanujan and how some of these were nurtured by Hardy. Littlewood
mentions some insightful comments in his unusual review after prefacing them by “To do it justice I must
infringe a little the rules about collaboration.” This is a very revealing review of historical value as well.

B Sury

COLLECTED PAPERS OF SRINIVASA RAMANUJAN

Edited By G H Hardy, P V Seshu Atgar and B M Wilson
Pp. xxxvi+355, 30s net. 1927. (Cam. Univ. Press.)

Ramanujan was born in India in December 1887, came to Trinity College,

Cambridge, in April 1914, was ill from May 1917 onwards, returned to India
in February 1919, and died in April 1920. He was a Fellow of Trinity and a
Fellow of the Royal Society.
Ramanujan had no university education, and worked unaided in India until
he was twenty-seven. When he was sixteen he came by chance on a copy of
Carr’s Synopsis of Mathematics ; and this book, now sure of an immortal-
ity its author can hardly have dreamt of, woke him quite suddenly to full
activity. A study of its contents is indispensable to any considered judg-
ment. It gives a very full account of the purely formal side of the integral
calculus, containing, for example, Parseval’s formula, Fourier’s repeated in-
tegral and other “inversion formulae”, and a number of formulae of the type
recognizable by the expert under the general description “f(α) = f(β) if
αβ = π 2 ”. There is also a section on the transformation of power series

*
Reproduced with permission from The Mathematical Gazette, Vol.14, No.200, pp.425–428, Apr., 1929.

RESONANCE ⎜ September 2013 859

 CLASSICS
into continued fractions. Ramanujan somehow acquired also an effectively
complete knowledge of the formal side of the theory of elliptic functions (not
in Carr). The matter is obscure, but this, together with what is to be found
in, say, Chrystal’s Algebra, seems to have been his complete equipment in
analysis and theory of numbers. It is at least certain that he knew nothing
of existing methods of working with divergent series, nothing of quadratic
residuacity, nothing of work on the distribution of primes (he may have
 
known Euler’s formula (1 − p−s )−1 = n−s , but not any account of the
ζ-function). Above all, he was totally ignorant of Cauchy’s theorem and
complex function theory. (This may seem difficult to reconcile with his
complete knowledge of elliptic functions. A sufficient, and I think a neces-
sary, explanation would be that Greenhill’s very odd and individual Elliptic
Functions was his text-book.)
The work he published during his India period did not represent his best
ideas, which he was probably unable to expound to the satisfaction of
editors. At the beginning of 1914, however, a letter from Ramanujan to
Mr. Hardy (then, at Trinity, Cambridge) gave unmistakable evidence of his
powers, and he was brought to Trinity, where he had three years of health
and activity. (Some characteristic work, however, belongs to his two years
of illness.)
I do not intend to discuss here in detail the work for which Ramanujan
was solely responsible (a very interesting estimate is given by Prof. Hardy,
p.xxxiv). If we leave out of account for the moment a famous paper writ-
ten in collaboration with Hardy, his definite contributions to mathematics,
substantial and original as they are, must, I think, take second place in
general interest to the romance of his life and mathematical career, his un-
usual psychology, and above all to the fascinating problem of how great a
mathematician he might have become in more fortunate circumstances. In
saying this, of course, I am adopting the highest possible standard, but no
other is appropriate.
Ramanujan’s great gift is a “formal” one; he dealt in “formulae”. To be
quite clear what is meant, I give two examples (the second is at random,

860 RESONANCE ⎜ September 2013

 CLASSICS
the first is one of supreme beauty):

{(1 − x5)(1 − x10)(1 − x15) · · ·}5

p(4) + p(9)x + p(14)x2 + · · · = 5 ,
{(1 − x)(1 − x2 )(1 − x3) · · ·}6

where p(n) is the number of partitions of n;

 ∞
cos πx 1
dx = (a > 12 ).
0 {Γ(a + x)Γ(a − x)} 2 4Γ(2a − 1){Γ(a)}2
But the great day of formulae seems to be over. No one, if we are again
to take the highest standpoint, seems able to discover a radically new type,
though Ramanujan comes near it in his work on partition series; it is fu-
tile to multiply examples in the spheres of Cauchy’s theorem and elliptic
function theory, and some general theory dominates, if in a less degree,
every other field. A hundred years or so ago his powers would have had
ample scope. Discoveries alter the general mathematical atmosphere and
have very remote effects, and we are not prone to attach great weight to
rediscoveries, however independent they seem. How much are we to allow
for this; how great a mathematician might Ramanujan have been 100 or 150
years ago; what would have happened if he had come into touch with Euler
at the right moment? How much does lack of education matter? Was it
formulae or nothing, or did he develop in the direction he did only because
of Carr’s book—after all, he learned later to do new things well, and at an
age mature for an Indian? Such are the questions Ramanujan raises; and
everyone has now the material to judge them. The letters and the lists of
results announced without proof are the most valuable evidence available
in the present volume; they suggest, indeed, that the note-books would give
an even more definite picture of the essential Ramanujan, and it is very
much to be hoped that the editors’ project of publishing them in extenso
will eventually be carried out.
Carr’s book quite plainly gave Ramanujan both a general direction and the
germs of many of his most elaborate developments. But even with these
partly derivative results one is impressed by his extraordinary profusion,
variety, and power. There is hardly a field of formulae, except that of
classical number-theory, that he has not enriched, and in which he has not
revealed unsuspected possibilities. The beauty and singularity of his results

RESONANCE ⎜ September 2013 861

 CLASSICS

is entirely uncanny. Are they odder than one would expect things selected
for oddity to be? The moral seems to be that we never expect enough; the
reader at any rate experiences perpetual shocks of delighted surprise. And
if he will sit down to an unproved result taken at random, he will find, if
he can prove it at all, that there is at lowest some “point”, some odd or
unexpected twist. Prof. Watson and Mr. Preece have begun the heroic
task of working through the unproved statements; some of their solutions
have appeared recently in the Journal of the London Mathematical Society,
and these strongly encourage the opinion that a complete analysis of the
note-books will prove very well worth while.
There can, however, be little doubt that the results showing the most strik-
ing originality and the deepest insight are those on the distribution of primes
(see pp. xxii-xxv, xxvii, 351, 352). The problems here are not in origin
formal at all; they concern approximative formulae for such things as the
number of primes, or of integers expressible as a sum of two squares, less
than a large number x; and the determination of the orders of the errors
is a major part of the theory. The subject has a subtle function-theory
side; it was inevitable that Ramanujan should fail here, and that his meth-
ods should lead him astray; he predicts the approximative formulae, but is
quite wrong about the orders of the errors. These problems tax the last re-
sources of analysis, took over a hundred years to solve, and were not solved
at all before 1890; Ramanujan could not possibly have achieved complete
success. What he did was to perceive that an attack on the problems could
at least be begun on the formal side, and to reach a point at which the
main results became plausible. The formulae do not in the least lie on the
surface, and his achievement, taken as a whole, is most extraordinary.
If Carr’s book gave him direction, it had at least nothing to do with his meth-
ods, the most important of which were completely original. His intuition
worked in analogies, sometimes remote, and to an astonishing extent by em-
pirical induction from particular numerical cases. Being without Cauchy’s
theorem, he naturally dealt much in transformations and inversions of order
of double integrals. But his most important weapon seems to have been a
highly elaborate technique of transformation by means of divergent series
and integrals. (Though methods of this kind are of course known, it seems
certain that his discovery was quite independent.) He had no strict logical

862 RESONANCE ⎜ September 2013

 CLASSICS

justification for his operations. He was not interested in rigour, which for
that matter is not of first-rate importance in analysis beyond the under-
graduate state, and can be supplied, given a real idea, by any competent
professional. The clear-cut idea of what is meant by a proof, nowadays so
familiar as to be taken for granted, he perhaps did not possess at all. If
a significant piece of reasoning occurred somewhere, and the total mixture
of evidence and intuition gave him certainty, he looked no further. It is
a minor indication of his quality that he can never have missed Cauchy’s
theorem. With it he could have arrived more rapidly and conveniently at
certain of his results, but his own methods enabled him to survey the field
with an equal comprehensiveness and as sure a grasp.
I must say something finally of the paper on partitions (pp. 276-309) written
jointly with Hardy. The number p(n) of the partitions of n increases rapidly
with n, thus:
p(200) = 3972999029388.
The authors show that p(n) is the integer nearest
2
1  √
(1) √ qAq (n)ψq (n),
2 2 q=1

where Aq (n) = ωp,q e−2npπi/q , the sum being over p’s prime to √ q and less
than it, ωp,q is a certain 24qth root of unity, ν is of the order of n, and

d √ 1 √2
ψq (n) = (exp{C (n − 24
)/q}), C=π 3
.
dn
We may take ν = 4 when n = 100. For n = 200 we may take ν = 5; five
terms of the series (1) predict the correct value of p(200). We may always
√
take ν = a n (or rather its integral part), where a is any positive constant
we please, provided n exceeds a value n0 (a) depending only on a.
The reader does not need to be told that this is a very astonishing theorem,
and he will readily believe that the methods by which it was established
involve a new and important principle, which has been found very fruitful
in other fields. The story of the theorem is a romantic one. (To do it justice
I must infringe a little the rules about collaboration. I therefore add that

RESONANCE ⎜ September 2013 863

 CLASSICS

Prof. Hardy confirms and permits my statements of bare fact.) One of Ra-
manujan’s Indian conjectures was that the first term of (1) was a very good
approximation to p(n); this was established without great difficulty. At this
1
stage the n − 24 was represented by a plain n—the distinction is irrelevant.
From this point the real attack begins. The next step in development, not
a very great one, was to treat (1) as an “asymptotic” series, of which a
fixed number of terms (e.g. ν = 4) were to be taken, the error being of the
order of the next term. But from now to the very end Ramanujan always
insisted that much more was true than had been established: “there must
be a formula with error O(1).” This was his most important contribution; it
was both absolutely essential and most extraordinary. A severe numerical
test was now made, which elicited the astonishing facts about p(100) and
p(200). Then ν was made a function of n; this was a very great step, and
involved new and deep function-theory methods that Ramanujan obviously
could not have discovered by himself. The complete theorem thus emerged.
But the solution of the final difficulty was probably impossible without one
more contribution from Ramanujan, this time a perfectly characteristic one.
As if its analytical difficulties were not enough, the theorem was entrenched
also behind almost impregnable defences of a purely formal kind. The form
of the function ψq (n) is a kind of indivisible unit; among many asymptoti-
cally equivalent forms it is essential to select exactly the right one. Unless
1 d
this is done at the outset, and the − 24 (to say nothing of the dn ) is an
extraordinary stroke of formal genius, the complete result can never come
into the picture at all. There is, indeed, a touch of real mystery. If only
we knew there was a formula with error O(1), we might be forced, by slow
stages, to the correct form of ψq . But why was Ramanujan so certain there
was one? Theoretical insight, to be the explanation, had to be of an order
hardly to be credited. Yet it is hard to see what numerical instances could
have been available to suggest so strong a result. And unless the form of
ψq was known already, no numerical evidence could suggest anything of the
kind—there seems no escape, at least, from the conclusion that the discov-
ery of the correct form was a single stroke of insight. We owe the theorem
to a singularly happy collaboration of two men, of quite unlike gifts, in
which each contributed the best, most characteristic, and most fortunate
work that was in him. Ramanujan’s genius did have this one opportunity
worthy of it.

864 RESONANCE ⎜ September 2013

 CLASSICS

The volume contains a biography by the second of the editors, and the
obituary notice by Prof. Hardy. These give quite a vivid picture of Ra-
manujan’s interesting and attractive personality. The mathematical editors
have done their work most admirably. It is very unobtrusive; the reader is
told what he wants to know at exactly the right moment, and more thought
and bibliographical research must have gone into it than he is likely to
suspect.

J. E. LITTLEWOOD

Nature, stretching Horace Davenport out, had forgotten to stretch him sideways, and one
could have pictured Euclid, had they met, nudging a friend and saying: “Don’t look now,
but this chap coming along illustrates exactly what I was telling you about a straight line
having length without breadth”
– P.G.Wodehouse.

“Every mathematician worthy of the name has experienced ... the state of lucid exaltation
in which one thought succeeds another as if miraculously... this feeling may last for hours
at a time, even for days. Once you have experienced it, you are eager to repeat it but
unable to do it at will, unless perhaps by dogged work”
–Andre Weil.

“I have often pondered over the roles of knowledge or experience, on the one hand, and
imagination or intuition, on the other, in the process of discovery. I believe that there is
a certain fundamental conflict between the two, and knowledge, by advocating caution,
tends to inhibit the flight of imagination. Therefore, a certain naivete, unburdened by
conventional wisdom, can sometimes be a positive asset”
–Harish-Chandra.

“A person who can, within a year, solve x^2 - 92 y^2 = 1 is a mathematician”

–Brahmagupta

RESONANCE ⎜ September 2013 865