Guilford Press

S&S Quarterly, Inc.

Marx and Mathematics

Author(s): Dirk J. Struik
Source: Science & Society, Vol. 12, No. 1, A Centenary of Marxism (Winter, 1948), pp. 181-
196
Published by: Guilford Press
Stable URL: http://www.jstor.org/stable/40399882
Accessed: 22-05-2016 19:13 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Guilford Press, S&S Quarterly, Inc. are collaborating with JSTOR to digitize, preserve and extend access
to Science & Society

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
MARX AND MATHEMATICS

DIRK J. STRUIK

received his early training in mathematics at the Gym-

nasium of Trier (Trves), the Rhineland city where he was
born. At his graduation, in 1835, his knowledge of mathe-
matics was considered adequate. This means that he started his career
with some knowledge of elementary arithmetic, algebra to the quadratic
equations, and plane and solid geometry. He also may have had trigo-
nometry, and a little higher algebra, analytical geometry and calculus.
There are no indications that he showed any interest in mathematics
during the turbulent years before and after 1848, in which he and
Engels developed their outlook on the world. The first token that Marx
had returned to his study of mathematics is from the period in which he
had settled in London and was working on his great scientific projects.
In a letter to Engels of Jan. 11, 1858,1 he wrote:

During the elaboration of the economic principles I have been so

damned delayed by computational errors that out of despair I undertook
again a quick scanning of the algebra. Arithmetic was always alien to me.
Via the algebraic detour, however, I catch up quickly.
From this period until his death in 1883 Marx showed continued
interest in the study of mathematics, often returning to it as a diversion
during his many days of illness.
His study of algebra was followed by that of analytical geometry
and the calculus. In a letter to Engels of July 6, 1863 he reported
progress:

In my spare time I do differential and integral calculus. Apropos!

I have plenty of books on it and I will send you one if you like to tackle
that field. I consider it almost necessary for your military studies. It is
also a much easier part of mathematics (as far as the purely technical
side is concerned) than for instance the higher parts of algebra. Aside
from knowledge of the common algebraic and trigonometric stuff no
preparatory study is needed except general acquaintance with the conic
sections.2

It seems therefore that Marx found algebra easier than arithmetic

and the calculus easier than algebra. But he was not so much interested
in the technique of the calculus. He was irresistibly drawn to the age-old

l Marx-Engels Gesamtausgabe (Berlin, 1930), Abt. Ill, Bd. 11, p. 273.

2 Ibtd., ni, p. 14g.

181

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 182 SCIENCE AND SOCIETY

question of the foundation of the calculus, the more so since in the

books which he consulted this subject was treated in a most unsatisfactory
and occasionally in a controversial way. Marx, like so many dialectical
thinkers before and after him, found unending fascination in the
different definitions of the derivative and the differential, as is shown
by a large amount of manuscript material which was found among his
papers.
In the years after 1870 Marx even tried to develop his own views.
Engels reports on this phase in the preface to the second volume of
Capital:

After 1870 an intermission set in again, mainly due to sickness. The

content of the many notebooks with abstracts of this period consists of
agronomy, American and especially Russian agrarian relations, money,
market and banking systems, and finally natural science, geology and
physiology, and especially independent mathematical papers.8

Marx, in the later days of his life, cast some of his reflections con-
cerning the differential calculus into a readable form and dispatched
the manuscript to Engels. A letter of August 18, 1881 shows that Engels
had studied them:

Yesterday I found at last the courage to study your mathematical

manuscripts even without reference to textbooks, and I was glad to see
that I did not need them. I compliment you on your work. The matter
is so perfectly clear (sonnenklar) that we cannot be amazed enough
how the mathematicians insist upon mystifying it.4
Engels continues to present Marx' viewpoint in his own words and
to compare it with Hegel's views, with which both he and Marx were
thoroughly familiar. He ends with the words:
The matter has taken such a hold of me that it not only turns around
in my head the whole day, but that also last week in a dream I gave
a fellow my shirt buttons to differentiate and this fellow ran away with
them [und dieser mir damit durchbrannte]*

Marx, who at that time was preoccupied with his wife's sickness-
she died in December of the same year- did not, it seems, return to the
subject in his subsequent correspondence. When, however, Engels re-
ported to Marx (November 21, 1882) on an exchange of letters between
him and their friend Sam Moore on the subject of Marx' mathematical

S Capital (Chicago, 1919), u, p. 10.

4 Marx-Engels Gesamtausgabe, Abt. Ill, Bd. iv, p. 513.
B Ibid., p. 514.

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 MARX AND MATHEMATICS 183

theories, Marx made a prompt reply the next day. We return to this
correspondence later in this article.
Marx died before he could add anything more to his ideas. Engels
later thought of publishing Marx' mathematical manuscripts together
with his own on the dialectics of nature. In the preface to the second
edition of the Anti-Dhring (1885) he mentions his own studies in
mathematics and the natural sciences, and adds that he had to dis-
continue them after the death of Marx. He concludes: "there will per-
haps later be an opportunity to collect and to publish the obtained
results, together with the posthumous, and very important, manuscripts
of Marx."6
Engels did not find the time to accomplish this work, and the papers
of Marx and Engels dealing with the exact sciences remained in the
archives. The German Social Democrats, who inherited the papers of
Marx and Engels, were unable to appreciate the dialectics of mathematics,
physics and chemistry. Understanding had to wait until the Russians
began to show the fundamental importance of Marx' and Engels* philo-
sophical work. Lenin's Materialism and Empirio-criticism (1908) was a
trail blazer, but it did not become known outside of strictly Russian
circles until it was published in German, long after the revolution of 1917.
Later the Russians published Engels' Dialectics of Nature; first in
Russian, then (1927) in the original German.
Both Lenin's and Engels' books are now available in English, Lenin's
in a translation of 1927, Engels' in a translation of 1940.
Still later some of the most characteristic of Marx' mathematical
manuscripts were published, but only in a Russian translation.7 Our
study is based on the papers published by the Russians. It is to be hoped
that all of his mathematical note books will eventually be published,
not only in Russian, but also in the original German.

6 Antt-Dhring (New York, 1939), p. 17.

iMarksiim i Estestvoznanie (Moscow: Partisdat, 1933). The Russian translation of
the manuscripts occupies p. 5-61; it is followed by articles by . Kolman, S. Ianovskaia,
D. J. Struik, H. J. Mller and others. The original German text of the manuscript
has not, as far as I know, been published, though there seem to have been plans;
see Unter dem Banner des Marxismus (1935), no. 9, p. 104, n. 1. I received in 1935
a typewritten copy of the original German text of the published mathematical manu-
scripts from the Marx-Engels Institute in Moscow, and the quotations in the present
article are translated from this text.
8 The information concerning the general character of the 900 pages of Marx mathe-
matical manuscript is taken from S. Ianovskaia, "O Matematicheskich Rukopisiakh
K. Marksa," p. 136-180. See also E. Colman, Science at the Cross Roads (London, 1931).
P. 133-135.

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 184 SCIENCE AND SOCIETY

The extent of Marx' interest in mathematics is shown by the fact

that the Marx-Engels-Lenin Institute in Moscow has obtained, since
1925, photographic copies of about 900 pages of Marx' mathematical
manuscripts, all of which have been deciphered and put in order.8
They consist essentially of abstracts of textbooks, studied by Marx, often
with notes; of comprehensive accounts of special subjects, and of inde-
pendent investigations, expressing different stages in Marx' studies, from
preliminary sketches to finished manuscripts probably prepared for the
benefit of Engels. Only a few pages, hardly twenty-four, are devoted
to computational work.
By far the most voluminous of these manuscripts deal with algebra,
which Marx studied from Lacroix', Maclaurin's and perhaps from other
texts. Most of this algebra deals with the solution of equations of
higher degree, but Marx also showed an interest in series, notably diver-
gent series. There are also abstracts dealing with analytic geometry,
notably from a book by Hymers.
Other manuscripts contain Marx' reflections on the differential
calculus. There are again plenty of abstracts and comprehensive ac-
counts based on the textbooks of Lacroix, Boucharlat and Hind, sup-
plemented by those of Hall and Hemming, all popular school texts
from the early decades of the nineteenth century. This work deals mainly
with the conception of function and of series, of limit and of derivative,
the series of Taylor and Maclaurin, and the determination of maxima
and minima. Marx showed particular interest in Lagrange's famous
use of the Taylor series for the "algebraic" foundation of the calculus,
and compared the different definitions of the derivative and the dif-
ferential in the various texts. Marx, in one of his own notes, reproduces
the derivation of the binomial theorem from Taylor's theorem, and
remarks that "Lagrange, on the contrary, derives Taylor's theorem from
the binomial theorem," a fact which he often repeats and to which he
devotes some thought. One of his manuscript papers is entitled "A some-
what modified development of Taylor's theorem on purely algebraic base
according to Lagrange,"9 others have such significant headings as:
"Taylor's theorem- is based on the translation from the algebraic language
of the binomial theorem into the differential way of expression," and
"Maclaurin's theorem is also only translation from the algebraic language
of the binomial theorem into the differential language." Two notebooks,

"Nach Lagrange somewhat modified Entwicklung des Taylorschen Theorems auf

bloss algebraischer Grundlage."

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 MARX AND MATHEMATICS 185

probably dating from a later period in Marx* life, contain examples of

the method of differentiation which Marx eventually preferred, as well
as a paper on the differential and a historical sketch of the methods of
differentiation used by Newton, Leibniz, D'Alembert and Lagrange.
These notebooks present the position which Marx seems to have placed
before Engels. They also contain a long paper on the integral calculus,
which contains a critical analysis of Newton's "Analysis per aequationes
numero terminorum infinitas." Their published contents form the sub-
ject of the present article.
Marx studied the calculus from textbooks which were all written
tinder the direct .influence of the great mathematicians of the late
seventeenth and the eighteenth centuries, notably of Newton, Leibniz,
Euler, D'Alembert and Lagrange. He was not so much interested in the
technique of differentiation and integration as in the basic principles
on which the calculus is built, that is, in the way the notions of derivative
and differential are introduced. He soon found out that a considerable
difference of opinion existed among the the leading authors concerning
these basic principles, a difference of opinion often accompanied by
confusion. This confusion only increased in the school textbooks written
by the minor authors.11 Different answers were given on such questions
as whether the derivative is based on the differential or vice versa,
whether the differential is small and constant, small and tending to zero,
or absolutely zero, etc. Marx felt the challenge offered by a problem
which had attracted some of the keenest minds of the past and which
dealt with the very heart of the dialectical process, namely the nature
of change. Not finding any satisfying answer in the books, he tried to
reach an answer for himself in his own typical way: by going to the
sources, comparing the results, and forging beyond them into new
regions. It may perhaps strike the reader that among the sources studied
by Marx there seems to be no reference to Augustin Cauchy- at any rate
as far as we can judge from the published material. Cauchy's work, which
underlies the exposition of the foundation of the calculus in our present
day textbooks, could have been available to Marx.12 The reason that
Marx took no notice of Cauchy may be that Cauchy's ideas only slowly
penetrated into textbooks, so that they might have escaped Marx, who

11 A good survey of the various theories is given by F. Cajori, "Grafting of the

Theory of Limits on the Calculus of Leibniz," Am. Math. Monthly, xxx (1923),
P. 223-34.
12 A. Cauchy, Rsum des leons donnes a l Ecole Royale Polytechnique sur le
calcul infinitsimal (Paris, 1823).

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
186 SCIENCE AND SOCIETY

did not move among professional mathematicians.18 A more likely reason

is that Cauchy's way of defining the derivative was essentially that of
D'Alembert, so that Marx did not consider his method a new one.
Whatever Marx* reasons were to ignore Cauchy's work, his feeling
of dissatisfaction with the way the calculus was introduced was shared
by some of the leading younger professional mathematicians of his day.
In the same year (1858) in which Marx resumed his study of mathematics,
Richard Dedekind at Zrich felt similar dissatisfaction, in his case while
teaching the calculus. Writing in 1872, he first stated that in his les-
sons he had recourse to geometrical evidence to explain the notion of a
limit; then he went on:
But that this form of introduction into the differential calculus can
make no claim to being scientific, no one will deny. For myself this feeling
of dissatisfaction was so overpowering that I made the fixed resolve to
keep meditating on the question till I should find a purely arithmetic
ana perfectly rigorous foundation for the principles of infinitesimal
calculus.14

This led Dedekind to a new axiomatic approach to the conception of

continuum and irrational number, which was one of the great pioneering
efforts in what we call the arithmetization of mathematics. Some years
later one of the other pioneers of the new methods of rigor in mathe-
matics, Paul Du Bois Reymond, exclaimed:
What mathematician would deny that-especially in its published
form- the conception of limit and its closest associates, the conception
of the limitless, the infinitely large and the infinitely small, the irrational,
etc., still lack rigor? The teacher in writ and word is used to hurry
quickly through this questionable entrance to analysis, in order to roam
Uie more comfortably on the well blazed roads of the calculus.15

It was not until the last decades of the nineteenth century, under the
influenc of Dedekind and Du Bois Reymond, as well of Weierstrass and
Cantor that the thorough overhauling of the principles of the calculus

13 The preface to the sixth edition of Boucharlat's book (1856), which Marx con-
sulted, though mentioning in detail the work of Newton, Leibniz, D'Alembert and
Lagrange, is silent about Cauchy. One of the first widely used textbooks which ex-
plicitly used Cauchy's methods was C. Jordan, Cours d'analyse, which appeared in
1882.
MR, Dedekind, Stetigkeit und Irrationalzahlen (1872). Translated in "Essays on the
Theory of Numbers" (Chicago, 1901), p. 1 f.
15 P. Du Bois Reymond, Die allgemeine Funktionentheorie, i. (1882), p. 2. The
author was the brother of the physiologist Emil, who framed the slogan of agnosti-
cism: "Ignorabimus."

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 MARX AND MATHEMATICS 187

took place which underlies modern methods, and has shown that Cauchy's
approach can lead to full rigor. This work appeared too late to influence
Marx and Engels.16 The result is that Marx' reflections on the founda-
tions of the calculus must be appreciated as a criticism of eighteenth
century methods. We feel however that his work, developed contem-
poraneously with but independently of the leading mathematicians of
the second half of the nineteenth century, even now contributes to the
understanding of the meaning of the calculus.
We should never forget, of course, that Marx never published his
material, and that there is not even an indication that he intended pub-
lication, even though Engels seems to have played with the idea. Marx
worked on mathematics in spare hours, for relaxation, often in hours
of sickness, guided by some books which he happened to have in his
library, such as Boucharlat's, which introduced the principles of dif-
ferentiation in an unsatisfactory way. He looked for elucidation in the
sources quoted in Boucharlat and similar books, which led him to New-
ton, Leibnitz, D'Alembert and Lagrange. His notes were in the first
place intended for his own clarification, after reading those classics in
attempts to understand the often obscure texts. Struck by the unsatisfac-
tory formulations in these books, he tried in characteristic way to
straighten out the difficulties for himself.
The difficulties which Marx tried to overcome are at present as real
as in his time, even if our formal apparatus is more carefully elaborated
and practically foolproof. These difficulties are as old as Zeno of Elea
and as young as the latest philosophical or physiological attempt to
understand how rest can pass into motion, and how motion can lead to
rest. This is the reason why Marx studied so carefully the conception
of the derivative of a function and the related conception of the di-
ferential. He found that there are three main methods by which these
conceptions have been developed. Marx classified them, called them the
mystical, the rational and the algebraic method (connected with the
names of Newton-Leibnitz, D'Alembert and Lagrange respectively), and
then opposed to them his own mode of understanding the derivative,
the differential, and the calculus in general. Let us explain the difficulty
by differentiating the function y=x8 in the different ways criticized by
Marx.

l It is even doubtful if any pertinent information on the work of the great German
mathematicians of the second half of the nineteenth century reached Marx and Engels.
The England of their days was an excellent place to study capitalism, as well as physics
and chemistry and biology, but it was backward in mathematics, except in some spe-
cialized branches of geometry and algebra.

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 188 SCIENCE AND SOCIETY

1) Newton-Leibnitz* ("The mystical differential calculus").17 - x changes

into x + xt in Newton, into x + dx in Leibnitz; we follow Leibnitz. Then
y changes intoni =j> + dy and
yi =y + dy - (x + dx)* - x* + 3x2 dx + 3x (dx)* + (dx)*.
Since (*)2 and (dx)* are infinitesimal as compared with 3x2dx, they may
be dropped, and we obtain the correct formula
dy 3x2dx.
This is highly mysterious, and the mystery does not disappear if we first
divide dy by dx
dy/dx = 3x2 + 3xdx + (dx)2
and then let h = dx be zero. It is true that we obtain the right formula

but as Marx remarks:

the nullification of A is not permitted before the first derived function, here
3x' has been liberated from the factor h by division, hence (y' - y)/h = 3*2
+ 3xh + h2. Only then can we annul (aufheben) the finite difference. The
differential coefficient dy/dx = 3x2 must therefore also originally be de-
veloped before we can obtain the differential dy = 3x2 dx.

In other words, we knew in advance what the answer must be, and build
up some reasoning to make it plausible. It was this loose way in which
Newton and Leibnitz usually founded the calculus which led Bishop Berkeley
to his famous criticism in The Analysts of 1734. Here he asked whether the dx
are zero or not zero, called them "ghosts of departed quantities" and con-
cluded that no mathematician who believed these absurdities could reason-
ably object to the miraculous tenets of religion. It has not been the only
case in which foundation difficulties in science have been exploited for
idealist and obscurantist reasons.

Mathematicians felt the difficulty and tried to cope with it by suggesting

more exact ways of founding the calculus.18 The most important contribu-
tions were those of D'Alembert and Lagrange.

2) D'Alembert ("The rational differential calculus").19 In Marx' words:

D'Alembert starts directly from the starting point of Newton and Leibnitz
xi = x + dx
but he makes immediately the fundamental correction xi = x + Ax, that

17 Leibnitz issued his first publication on the calculus in 1684, Newton his in 1693.
18 See e.g. F. Cajori, A History of the Conceptions of Limit and Fluxion in Great Britain from
Newton to Woodhouse (Chicago and London, 1919).
19 D'Alembert on "Diffrentiel" in Diderot's Encyclopdie (1754)

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 MARX AND MATHEMATICS 189

means, Ax becomes an undetermined, but prima facie finite increment, which

he calls A. The transformation of this A or Ax into dx (he used the Leibnitz
notation, like all Frenchmen) is only found as the last result of the develop-
ment or at least just before closing hour (knapp vor Tor schiusi), while it
appears as starting point with the mystics and the initiators of the calculus:

h xi - x A h

Now, by placing h = o, the expression |/(x + A) - f(x)j/h changes

into -J- :

:--*-/
The way in which D' Alembert differentiates is very much akin to Cauchy's
method. We write at present with Cauchy
dy lim /(* + A) -/(*).
dx h+o A

Marx' objection to this method is that though it is formally correct, the

derivative/ (x) is already present in 3 x2 + 3xh + A2, that is, before dif-
ferentiation. It is simply the first term of a sum, 3 x2 + 2 xh + A2, and
D'Alembert's method only consists in devising a way in which to get rid
of the member (or members) of the sum which follows 3 x2. Marx calls this
Loswicklung (separation); while the correct method should be Entwicklung
(development):
The derivation therefore is the same as in Leibnitz and Newton, but the
ready-made derivative is in strictly algebraic way separated from its further
context.20 There is no development but a separation of the /'(*), here 3x2 21
from its factor A and the members which appear next to it in the other mem-
bers marching on in rank and file. What has really been developed is the left
hand symbolic side, namely dx, dy and their ratio, the symbolic differential
coefficient dy/dx or o/o (rather in the other way o/o = dy/dx), which in its
turn again provoked some metaphysical shudders, though the symbol was
mathematically derived. D'Alembert had, by stripping the differential
calculus from its mystical garb, made an enormous step ahead.

Marx' evaluation of D'Alembert's work as "an enormous step ahead" still

stands. This is the more remarkable, since even modern historians of mathe-
matics have a way of glossing over it. Marx next proceeds to Lagrange.
3) Lagrange (" lhe purely algebraic differential calculus").

*-(* + h)% = * + 3*2* + 3*A + A'

10 "losgewickelt von ihrem sonstigen Zusammenhang."

11 "Es ist keine Entwicklung, sondern eine Loswickimg des/ (*)."

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 190 SCIENCE AND SOCIETY

Lagrange simply defines the coefficient of A as the derivative:

j = /'(*) = 3x2, or more generally by Taylor's theorem for a general /(x):

ax

yx = f{x + h) -, (or/*) + dh + g + . . .
Marx then paraphrases Lagrange's method in the words:
In the first method (1), as well as in the rational one (2), the required
real coefficient is fabricated ready made by the binomial theorem and can
be found already as second term of the series expansion, hence in the term
which necessarily contains A1. The whole further differential procedure, be
it as in (1) or be it as in (2), is therefore luxury. Let us therefore shed the
useless ballast. We know once and for all from the binomial expansion that
the first real coefficient is the factor of A, the second one that of A1, etc. These
real differential coefficients are nothing but the derived functions of the original
function in x, expanded binomially in succession . . . The whole real problem
reduced itself to the finding of methods (algebraic ones) of expanding all
kinds of functions of x + A into integral ascending powers of A, which in
many cases cannot be effected without great prolixity of operations. u Up
to now there appears nothing in Lagrange, but what can be found directly
from D'Alembert's method (since this also includes the whole development
of the mystics).

The objection which Marx raised against the classical writers was that all
four had the derivative already prepared before the process of differentiation
really begins. Marx wanted a method which actually folowed the process of
variation of the variable and in this process itself defined the derivative as
o/oy in which case it can be endowed with a new symbol dy/dx. The deriva-,
tive, he claimed, should be derived by a process of differentiation, not be pro-
duced from the beginning by the binomial theorem.

Whether we start falsely from x + dx or correctly from x + A, if we sub-

stitute this undetermined binomial into the given algebraic function of x, we
change it into a binomial of a definite degree, e.g. (x + A)1 instead of x1, and
this in a binomial in which in one case dxy in the other case A, figures as its
last member. Hence it also figures in the expansion only as a factor, with
which the functions, derived from the binomial, are externally affected.21

This lack of internal development can be avoided in the method which

Marx suggests, say for^ = xs:
j, -/()-,
/(*i) -/(*) - J>i - J> - * - * - (*i - *) M + **i + x')
fMjzM . ym . 4 + XXl + A
X' - X X' - X

M We now know that often it cannot be done at all, but this requires an extension of the
functional conception beyond Lagrange's horizon.

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 MARX AND MATHEMATICS 191

When x = x', or x - x' = o, we obtain:

-sss^ssx2
o dx dx + xx + x2ss3x2m
In this method, writes Marx, we obtain first a preliminary derivative, namely
x2 + xxi + x2, and this passes by x = x' into the definite derivative. This
passing from x' to x does away with any "infinitesimal" approximation, i*
shows that the derivative is actually o/o, obtained when xi - x is actually
zero:

Here we see in striking form:

Firstly; to obtain the derivative we must place x' = *, hence x' - x = o
in the strict mathematical sense, without a trace of only infinitesimal approxi-
mation.

Secondly: Through the fact that x' has been placed = *, hence x' - x = o,
nothing symbolic enters into the "derivative." The quantity x', originally
introduced by the variation of *, does not disappear, it is only reduced to its
minimal boundary x. It remains an element introduced as new into the
original function, which by its combination partly with itself, partly with the
x of the original function produces at the end the "derivative," that is the
preliminary "derivative" reduced to its minimum value.
. . . The transcendental or symbolic accident (o/o = dx/dy = 3x2) occurs
only on the left hand side, but it has already lost its terror, as it appears now
only as the expression of a process that already has shown its real content
on the right hand side of the equation.24
At the moment that x' = x the quotient Ay /Ax becomes 0/0. Since in this
expression 0/0 every trace of its origin and of its meaning has disappeared it
is replaced by the symbol dy/dx, in which the finite differences Ay and A*
appear in symbolical form as liquidated (aufgehobene) or vanished (verschwundene)
differences. At this moment algebra disappears and the differential calculus,
which operates with the symbols dy/dx, begins.
In order to understand Marx' intentions better, we translate here part of
the letter which Engels wrote him August 18, 1881, after he had read Marx'
manuscript:
When we say that in y = f(x) the x and y are variables, then this is, as
long as we do not move on, a contention without all further consequences,
and x and y still are, pro tempore, constants in fact. Only when they really
change, that is inside the function, they become variables in fact. Only in that
case is it possible for the relation - not of both quantities as such, but of their
variability - which still is hidden in the original equation, to reveal itself.

28 "nur als Faktor, womit die durch das Binom abgeleiteten Funktionen usserlich
behaftet sind."
*4 "Das transzendentale oder symbolische Unglck ereignet sich nur auf der linken
Seite, hat aber seine Schrecken bereits verloren, da es nun als Ausdruck eines Prozesses
erscheint, der, seinen wirklichen Gehalt bereits auf der rechten Seite der Gleichung
bewhrt hat."

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 192 SCIENCE AND SOCIETY

The first derivative Ay /Ax shows this relation as it occurs in the course of the
real change, that is in every given change; the final derivative dy/dx shows
it in its generality, pure. Hence we can come from dy/dx to every Ay/ Ax,
while this itself (Ay /Ax) only covers the special case. However, to pass from
the special case to the general relationship the special case has to be liquidated
as such (als solcher aufgehoben werden). Hence, after the function has passed
through the process from x to xf with all its consequences, xr can be quietly
allowed to become x again, it is no longer the old x, which was only variable
in name, it has passed through real change, and the result of the change re-
mains, even if we liquidate it again itself (auch wenn wir sie selbst wieder aufheben).
We see here at last clearly, what many mathematicians have claimed for
a long time, without being able to present rational reasons for it, that the
derivative is the original, the differentials dx and dy are derived.
The difference between Marx' method and D'Alembert's method (and
also that of Cauchy) should not be misunderstood and rejected as trivial or
insignificant (xf - x = A versus xf = x + A). Marx, as I see it, was per-
fectly satisfied that D'Alembert's method is formally correct. However, he
wanted to come to an understanding of the process of differentiation itself.
Is the derivative obtained by letting x (and y) pass through a sequence of
constant values, or is it necessary to let x (andy) really change? Thus under-
stood, we see the old "paradox" of Zeno emerging: can the motion of a point
be obtained by following a sequence of positions of this point at rest? Zeno
showed that a sequence of such positions will never produce motion; he also
showed by a similar reasoning that Achilles will never reach the tortoise.
D'Alembert's method, Marx claimed, represents a mode of thought which
does not do justice to the actual event which happens when a function is
differentiated. What happens is a real change, and this is better understood
when we first write Ay /Ax as a function of x and an entirely new x', and then
let x = xr. Moreover, A = x' - x does not only approach zero, A becomes
zero. Emphasis is placed on the fact that the derivative only appears when
both Ay and Ax are absolutely zero. This never became clear with the
"mystics" Leibnitz-Newton, and appeared as an accidental thing in D'Alem-
bert-Lagrange.25 It is so little understood that in some popular texts, such
as Hogben's Mathematics for the Million, the impression is given that the
process of differentiation is only approximately true. But even in our modern
textbooks, though they use a formal apparatus which is unimpeachable,
some of the thought behind the apparatus is not fully clarified.
Let us take, as an example, the textbook Pure Mathematics of G. H. Hardy,
who is one of our greatest living mathematicians. The derivative is explained
in the Cauchy-D'Alembert way:
,M _ lim 0(s + A) - 4>(x)

26 More information in F. Cajori, "The History of Zeno's Arguments on Motion," vi,

Am. Math. Monthly xxn (1915), p. 143-149.

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 MARX AND MATHEMATICS 193

which means that {<f> (* + h) - <f> (*)} /h tends to a limit when h tends to
zero. What does this mean? We are told that 0 (y) tends to the limit / as y
tends to zero, if, when any positive number 8, however small, is assigned, we
can choose^o(5) so that |4 (y) - /| < when o <y J>o($).26
This definition is exact, in the sense that we have a correct and subtle
criterium to test any limit. But 4> {y) always hovers near the limit, since we
are told that y "tends" to zero. Similarly, 4>f (*) is defined by means of an
h which "tends" to zero. The question is, is the event h = o ever reached?
Marx not only affirms it, he stresses it. The usual modern textbook definition
does not take this question seriously, because it is satisfied with a pragmatic
criterium which allows us to recognize a limit when it appears.27
The result is that much teaching of the elements of the calculus proceeds
as follows - and I confess to it myself in my own teaching. First, it is shown
that a limit can be approached as closely as we like, but never reached. Then
the derivative is defined with the aid of this conception of limit. And then
suddenly we begin to work with this derivative, which could never be
reached (as we have before demonstrated) as if it actually had been reached.
The case h = o, xr = #, though present in the formal apparatus, is somehow
obscured in the reasoning. An exception is found in the work of Moritz
Pasch, who in his very careful analysis of the derivative develops a formal
apparatus in which there is full room for the case h = o.28
Marx therefore belonged to that school of thinkers who insist on utmost
clarity of thought in interpreting a formal apparatus. His position contrasts
sharply to that of those mathematicians or mathematical physicists who
believe that the formal apparatus is the only thing that matters. Marx'
position was that of the materialist, who insists that significant mathematics
must reflect operations in the real world.
It is interesting to notice that the differences between Marx' and D'Alem-
bert's formal apparatus diminish when we consider more complicated
functions. For the case y = sin x the derivative, in the D' Alembert way of
differentiation, is still obtained by separation (Loswicklung), but by y = log x
the derivative can only be obtained from Ay/ Ax by letting h pass through
a real change.
28 G. H. Hardy, Pure Mathematics (Cambridge University Press, 6th ed., 1933) esp. p.
116, 198. This definition of limit is valid when 7 tends to zero by positive value. In a similar
way a definition of limit can be reached when^y tends to zero by negative values.
27 See e.g. F. Cajori, Am. Math. Monthly, xxn (1915), p. 149, concerning variables
reaching their limits: "In modern theory it is not particularly a question of argument, but
rather of assumption. The variable reaches its limit if we will that it shall; it does not reach
its limit, if we will that it shall not." Such a reasoning seems to lead to the conclusion that
it depends on our will whether Achilles will reach or will not reach the tortoise.
28 M. Pasch, "Der Begriff des Differentials," in Mathematik am Ursprung (Leipzig 1927)
p. 46-73, esp. p. 61, 68.

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 194 SCIENCE AND SOCIETY

As soon as dy/dx is established as the result of a real change, it becomes itself

the subject of a calculus, the differential calculus. Marx, in a manuscript on
the meaning of the differential, derived as one of the first formulas of this
calculus, that the derivative of y = uz, u and z functions of x, is given by
dy du dz

When uz =/(*), then dy/dx can be written /'(*), and "the/'(x) stands
opposed to dy/dx as its own symbolic expression, as its double or symbolic
equivalent."
The symbolic differential coefficient has become an independent starting pointy
whose real equivalent has first to be found. The initiative has been moved
from the right hand pole, the algebraic one (in dy/dx =/'(*)) to the left
hand one, the symbolic one. With this, however, the differential calculus
appears also as a specific kind of computation, operating already independ-
ently on its own territory. Its starting points du/dx, dz/dx are mathematical
quantities which belong exclusively to this calculus and characterize it. And
this reversal (Umschlag) of the method resulted here from the algebraic
differentiation of uz- The algebraic method changes automatically into its
opposite, the differential method.

Now, by removing in the equation (a), -r = Z x + u L *e common

denominator dx, we obtain (b), d (uz) = dy = udz + zdu, in which every
trace of its origin from (a) has been removed.
It (b) is therefore valid in the case that u and z depend on x as well as in the
case that they only depend on each other without any relation to x. It is
from the beginning a symbolic equation and can serve from the beginning
as a symbolic operational equation.
The differential is therefore a symbolic form - we would say an operational
form - dy =/'(*) dx appears as just another form of dy/dx = /'(x) and is
always convertible into the differential form. Modern mathematicians will
have no fault to find with this method, and V. Glivenko has specially shown 29
how Hadamard, the French mathematician, had stressed the operational
character of the differential. Marx does not mention, however, the now
common interpretation that dy should be/'(x)Ax, obtained by arbitrarily
placing dx = Ax. This way of representing dy, which dates back to Cauchy,
may have escaped Marx (he criticizes Boucharlat for his introduction of the
differential, but Boucharlat's method is an antiquated one). We believe
however that Marx would in any case have objected to this equation dx = Ax,
which established an identity between two conceptions with an entirely
different operational meaning. The interpretation of dy by Cauchy, which has

19 V. Glivenko, "Der Differentialbegriff bei Marx und Hadamard," Unter dem Banner
des Marxismus (1935) no. 9, p. 102-110; Russian text in Pod Znamenem Marksi&na 1934,
no. 5. See J. Hadamard, Cours d'analyse, i (Paris, 1927), p. 2 and 6.

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 MARX AND MATHEMATICS 195

found its way in all our texts, is mechanical and can only be justified by the
use to which the formula dy = /' (x) dx can be put as an approximation to
certain changes of a constant x into an equally constant x + A#.80 And the
fact that this difference between dx and Ax, dy and Ay can be neatly repre-
sented in a figure would not have impressed Marx and Engels, whose
interest was in the arithmetical-algebraic relationship of the symbols of the
calculus with the real process of change. This may be shown from the follow-
ing correspondence between Marx and Engels after Sam Moore had written
his opinion on the manuscript material of Marx:
Enclosed first a mathematical attempt by Moore. The result that "the
algebraic method is only the differential method disguised" refers of course
only to his own method of geometrical construction and is there also rela-
tively correct. I have written to him that you do not care about the way in
which the matter is represented in the geometrical construction, the applica-
tion to the equation of the curves is indeed sufficient (reiche ja hin). Moreover,
the fundamental difference between you and the old method is that you make
x change into *', hence make them really vary, while the other one departs
from x + , which is always only the sum of two quantities, but never the
variation of a quantity. Your x therefore, even when it has passed through
x1 and has again become x, is yet another than before; while x remains
constant during the whole period when h is first added to x and later again
subtracted. However, every graphical representation of the variation is
necessarily the representation of the past process, of the result, hence of a
quantity which became constant, the line x; its complement is represented
as x + A, two segments of a line. From this already follows that a graphical
representation of how x becomes x1 and x* again becomes x is impossible
(Engels to Marx, Nov. 21, 1882).31

Marx5 answer followed the next day:

Sam, as you have seen immediately, criticizes the analytical method which I
have used by simply pushing it aside, and instead keeps himself busy with
the geometrical application, to which I did not devote one word
I could in the same way get rid of (konnte damit abspeisen) the development
of the proper so-called differential method - beginning with the mystical
method of Newton and Leibnitz, then continuing with the rationalist method
of D'Alembert and Euler, and finishing with the strictly algebraic method
of Lagrange (which however always starts from the same original principle
as Newton-Leibnitz) - I could get rid of this whole historical development
of analysis by saying that practically nothing essential has changed in the

10 Compare C. De la Vallee Poussin, Cours d'analyse infinitesimale , i (Louvain, Paris, 1923),

p. 52: "For the substitution of dx for A* in the equation d/(x) =/ (x)Ax there is no necessity,
but it is hallowed by custom and this custom is justified."
11 The words between quotation marks are in English in the letter - see Marx-Engels
Gesamtausgabe, Abt. Ill, Bd. u, p. 571.

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms
 196 SCIENCE AND SOCIETY

geometrical application of the differential calculus, that is, in the geometrical

representation (Versinnltchung).32

This last remark of Marx shows affinity with that of Dedekind, who also
endeavored to build up the calculus independent of the geometrical repre-
sentation of the derivative. We can consider this as one of the characteristics
of Marx' analysis, in which it agreed with our modern approach. Another
important feature was his insistence on the operational character of the
differential and on his search for the exact moment where the calculus
springs from the underlying algebra as a new doctrine. "Infinitesimals" do
not appear in Marx' work at all. In his insistence on the origin of the deriva-
tive in a real change of the variable he takes a decisive step in overcoming
the ancient paradox of Zeno - by stressing the task of the scientist in not
denying the contradictions in the real world but to establish the best mode
in which they can exist side by side.32 Here his position is directly opposite
to that taken by Du Bois Reymond, who thought that the increments dx, dy
have to be taken as being at rest, invariable,33 or of the modern mathema-
tician Tarski, who denies the existence of variable quantities altogether.34
Marx' position in this respect will be appreciated by most mathematicians.
We believe that this survey of Marx' opinions on the origin of the calculus
demonstrates that publication of his other mathematical manuscripts is
also desirable.

Massachusetts Institute of Technology.

82 Marx-Engels Gesamtausgabe, Abt. Ill, Bd. iv, p. 572. Compare Marx, Capital, Part I,
ch. 3, Section 2: "The Metamorphosis of Commodities," (Engl. translation, ed. 1889,
p. 76).
83 Du Bois Reymond, op. cit., p. 141, states his dislike for the conception of dx as a
"quantit vanouissante," since he disapproves (geht mir entschieden wider den Mann) quan-
tities which begin to move only when we look at the formulas: ''As long as the book is
closed, profound rest prevails. As soon as I open it, the race to zero begins of all quantities
provided with the d" Marx, without coming to Du Bois Reymond's conclusion, might
have shared his criticism, since he wanted to express not only a change on paper, but a
change in reality.
34 A Tarski, Introduction to Logic (New York, 1941), p. 4.

This content downloaded from 193.140.194.57 on Sun, 22 May 2016 19:13:32 UTC
All use subject to http://about.jstor.org/terms