Twenty-Five Years with

Nicolas Bourbaki,
1949–1973
Armand Borel

T
he choice of dates is dictated by per- To set the stage, I shall briefly touch upon the
sonal circumstances: they roughly bound first fifteen years of Bourbaki. They are fairly well
the period in which I had inside knowl- documented2, and I can be brief.
edge of the work of Bourbaki, first In the early thirties the situation of mathemat-
through informal contacts with several ics in France at the university and research levels,
members, then as a member for twenty years, until the only ones of concern here, was highly unsat-
the mandatory retirement at fifty. isfactory. World War I had essentially wiped out one
Being based largely on personal recollections, generation. The upcoming young mathematicians
my account is frankly subjective. Of course, I had to rely for guidance on the previous one, in-
checked my memories against the available docu- cluding the main and illustrious protagonists of the
mentation, but the latter is limited in some ways: so-called 1900 school, with strong emphasis on
not much of the discussions about orientation and analysis. Little information was available about
current developments abroad, in particular about
general goals has been recorded.1 Another mem-
the flourishing German school (Göttingen, Ham-
ber might present a different picture.
burg, Berlin), as some young French mathemati-
cians (J. Herbrand, C. Chevalley, A. Weil, J. Leray)
Armand Borel is professor emeritus at the School of Math-
ematics, Institute for Advanced Study, Princeton, NJ.
were discovering during visits to those centers.3
In 1934 A. Weil and H. Cartan were Maîtres de
This article is an outgrowth of a lecture given at the Uni-
Conférences (the equivalent of assistant professors)
versity of Bochum, Germany, October 1995, in a Collo-
quium in honor of R. Remmert, and at the International at the University of Strasbourg. One main duty
Center for Theoretical Physics, Trieste, Italy, September was, of course, the teaching of differential and in-
1996. With permission of the author, it is being published tegral calculus. The standard text was the Traité
concurrently in the Notices and in the Mitteilungen der d’Analyse of E. Goursat, which they found want-
Deutsche Mathematiker-Vereinigung. ing in many ways. Cartan was frequently bugging
1 The Archives of Bourbaki at the École Normale Weil with questions on how to present this mate-
Supérieure, Paris, contain reports, surveys, successive rial, so that at some point, to get it over with once
drafts or counterdrafts of the chapters, remarks on those and for all, Weil suggested they write themselves
resulting from discussions, and proceedings of the Con-
gresses, called “Tribus”. Those provide mainly a record 2See [2, 3, 6, 7, 8, 14].
of plans, decisions, commitments for future drafts, as well
as jokes, sometimes poems. 3For this, see pp. 134–136 of [8].

MARCH 1998 NOTICES OF THE AMS 373

 a new Traité d’Analyse. This suggestion was spread ous text in the same book or to an earlier book (in
around, and soon a group of about ten math- the given ordering). The title “Éléments de Math-
ematicians began to meet regularly to plan this trea- ématique” was chosen in 1938. It is worth noting
tise. It was soon decided that the work would be that they chose “Mathématique” rather than the
collective, without any acknowledgment of indi- much more usual “Mathématiques”. The absence
vidual contributions. In summer 1935 the pen of the “s” was of course quite intentional, one way
name Nicolas Bourbaki was chosen.4 for Bourbaki to signal its belief in the unity of
The membership varied over the years; some mathematics.
people in the first group dropped out quickly, oth- The first volumes to appear were the Fascicle of
ers were added, and later there was a regular Results on Set Theory (1939) and then, in the for-
process of additions and retirements. I do not in- ties, Topology and three volumes of Algebra.
tend to give a detailed account. At this point let At that time, as a student and later assistant at
me simply mention that the true “founding fa- the E.T.H. (Swiss Federal Institute of Technology)
thers”, those who shaped Bourbaki and gave it in Zurich, I read them and learned from them, es-
much of their time and thoughts until they re- pecially from Multilinear Algebra, for which there
tired, are: was no equivalent anywhere, but with some reser-
Henri Cartan vations. I was rather put off by the very dry style,
Claude Chevalley without any concession to the reader, the appar-
Jean Delsarte ent striving for the utmost generality, the inflexi-
Jean Dieudonné ble system of internal references and the total ab-
André Weil sence of outside ones (except in Historical Notes).
For many, this style of exposition represented an
born respectively in 1904, 1909, 1903, 1906, 1906— alarming tendency in mathematics, towards gen-
all former students at the École Normale Supérieure erality for its own sake, away from specific prob-
in Paris.5 lems. Among those critics was H. Weyl, whose
A first question to settle was how to handle opinion I knew indirectly through his old friend and
references to background material. Most existing former colleague M. Plancherel, who concurred,
books were found unsatisfactory. Even B. v.d. Waer- at a time I was the latter’s assistant.
den’s Moderne Algebra, which had made a deep im- In fall 1949 I went to Paris, having received a fel-
pression, did not seem well suited to their needs lowship at the C.N.R.S. (Centre National de la
(besides being in German). Moreover, they wanted Recherche Scientifique), benefiting from an ex-
to adopt a more precise, rigorous style of exposi- change convention just concluded between the
tion than had been traditionally used in France, so C.N.R.S. and the E.T.H. I quickly got acquainted with
they decided to start from scratch and, after many some of the senior members (H. Cartan,
discussions, divided this basic material into six J. Dieudonné, L. Schwartz) and, more usefully for
“books”, each consisting possibly of several vol- informal contacts, with some of the younger ones,
umes, namely: notably Roger Godement, Pierre Samuel, Jacques
I Set Theory Dixmier, and, most importantly, Jean-Pierre Serre,
II Algebra the beginning of intense mathematical discussions
III Topology and a close friendship. Of course, I also attended
IV Functions of One Real Variable the Bourbaki Seminar, which met three times a
V Topological Vector Spaces year, offering each time six lectures on recent de-
VI Integration velopments.
These books were to be linearly ordered: refer- Those first encounters quickly changed my vi-
ences at a given spot could only be to the previ- sion of Bourbaki. All these people—the elder
ones, of course, but also the younger ones—were
4See [3] for the origin of the name. very broad in their outlook. They knew so much
5They all contributed in an essential way. For Cartan, and knew it so well. They shared an efficient way
Chevalley, Dieudonné, and Weil I could witness it at first- to digest mathematics, to go to the essential
hand, but not for Delsarte, who was not really active any- points, and reformulate the math in a more com-
more when I came on board. But his importance has been prehensive and conceptual way. Even when dis-
repeatedly stressed to me by Weil in conversations. See also cussing a topic more familiar to me than to them,
[14] and comments by Cartan, Dieudonné, Schwartz in [3, their sharp questions often gave me the impres-
pp. 81–83]. In particular, he played an essential role in sion I had not really thought it through. That
transforming into a coherent group, and maintaining it methodology was also apparent in some of the lec-
so, a collection of strong, some quite temperamental, in-
tures at the Bourbaki seminar, such as Weil’s on
dividuals. Besides, obviously, Book IV, Functions of one real
theta functions (Exp. 16, 1949) or Schwartz’s on
variable, owes much to him. Some other early members,
notably Szolem Mandelbrojt and René de Possel, also con- Kodaira’s big Annals paper on harmonic integrals
tributed substantially to the work of the group in its ini- (Exp. 26, 1950). Of course, special problems were
tial stages. not forgotten—in fact, were the bread and but-

374 NOTICES OF THE AMS VOLUME 45, NUMBER 3

ter of most discussions. The writing of the books The underlying thought was apparently that re-
was obviously a different matter. ally new, groundbreaking ideas were more likely
Later I was invited to attend (part of) a Bourbaki to arise from confrontation than from an orderly
Congress and was totally bewildered. Those meet- discussion. When they did emerge, Bourbaki mem-
ings (as a rule three per year: two of one week, one bers would say, “the spirit has blown” (“l’esprit a
of about two) were private affairs, devoted to the soufflé”), and it is indeed a fact that it blew much
books. A usual session would discuss a draft of more often after a “spirited” (or should I say
some chapter or maybe a preliminary report on a stormy) discussion than after a quiet one.
topic under consideration for inclusion, then or Other rules of operation also seemed to mini-
later. It was read aloud line by line by a member, mize the possibility of a publication in a finite
and anyone could at any time interrupt, comment, time:
ask questions, or criticize. More often than not, this Only one draft was read at a given time, and
“discussion” turned into a chaotic shouting match. everyone was expected to take part in everything.
I had often noticed that Dieudonné, with his sten- A chapter might go through six or even more
torian voice, his propensity for definitive state- drafts. The first one was written by a specialist, but
ments, and extreme opinions, would automatically anyone might be asked to write a later one. Often
raise the decibel level of any conversation he would this was hardly rewarding. Bourbaki could always
take part in. Still, I was not prepared for what I saw change his mind. A draft might be torn to pieces
and heard: “Two or three monologues shouted at and a new plan proposed. The next version, fol-
top voice, seemingly independently of one an- lowing those instructions, might not fare much bet-
other” is how I briefly summarized for myself my ter, and Bourbaki might opt for another approach
impressions that first evening, a description not or even decide that the former one was preferable
unrelated to Dieudonné’s comments in [8]: after all, and so on, resulting sometimes in some-
thing like a periodicity of two in the successive
Certain foreigners, invited as specta- drafts.
tors to Bourbaki meetings, always come To slow down matters further, or so it seemed,
out with the impression that it is a gath- there were no majority votes on publications: all
ering of madmen. They could not imag- decisions had to be unanimous, and everyone had
ine how these people, shouting—some a veto right.
times three or four at the same time— However, in spite of all those hurdles, the vol-
could ever come up with something in- umes kept coming out. Why such a cumbersome
telligent.… process did converge was somewhat of a mystery
It was only about ten years ago, reading the even to the founding members (see [6, 8]), so I do
text of a 1961 lecture by Weil on organization and not pretend to be able to fully explain it. Still, I will
disorganization in mathematics [13], that I realized venture to give two reasons.
this anarchic character, if not the shouting, was re- The first one was the unflinching commitment
ally by design. Speaking of Bourbaki, Weil said, in of the members, a strong belief in the worthiness
part (freely translated): of the enterprise, however distant the goals might
seem to be, and the willingness to devote much
…keeping in our discussions a care- time and energy to it. A typical Congress day would
fully disorganized character. In a meet- include three meetings, totaling about seven hours
ing of the group, there has never been of often hard, at times tense, discussions—a rather
a president. Anyone speaks who wants grueling schedule. Added to this was the writing
to and everyone has the right to inter- of drafts, sometimes quite long, which might take
rupt him.… a substantial part of several weeks or even months,
with the prospect of seeing the outcome heavily
The anarchic character of these dis- criticized, if not dismissed, or even summarily re-
cussions has been maintained through- jected after reading of at most a few pages, or left
out the existence of the group.… in abeyance (“put into the refrigerator”). Many,
even if read with interest, did not lead to any pub-
A good organization would have no lication. As an example, the pièce de résistance of
doubt required that everyone be as- the second Congress I attended was a manuscript
signed a topic or a chapter, but the idea by Weil of over 260 pages on manifolds and Lie
to do this never occurred to us.… groups, titled “Brouillon de calcul infinitésimal”,
based on the idea of “nearby points” (“points
What is to be learned concretely from proches”), a generalization of Ehresmann’s jets.
that experience is that any effort at or- This was followed later by about 150 pages of
ganization would have ended up with elaboration by Godement, but Bourbaki never pub-
a treatise like any other.… lished anything on nearby points.

MARCH 1998 NOTICES OF THE AMS 375

 On the other hand, whatever was accepted would lot, and I owe to these friends a great
be incorporated without any credit to the author. part of my mathematical culture.7
Altogether, a truly unselfish, anonymous, de-
and by Dieudonné [8, pp.143-44]:
manding work by people striving to give the best
possible exposition of basic mathematics, moved In my personal experience, I believe
by their belief in its unity and ultimate simplicity. that if I had not been submitted to this
My second reason is the superhuman efficiency obligation to draft questions I did not
of Dieudonné. Although I did not try to count know a thing about, and to manage to
pages, I would expect that he wrote more than pull through, I should never have done
any two or three other members combined. For a quarter or even a tenth of the math-
about twenty-five years he would routinely start ematics I have done.
his day (maybe after an hour of piano playing) by
writing a few pages for Bourbaki. In particular, But the education of members was not a goal
but by far not exclusively, he took care of the final per se. Rather, it was forced by one of the mottoes
drafts, exercises, and preparation for the printer of Bourbaki: “The control of the specialists by the
of all the volumes (about thirty) which appeared nonspecialists”. Contrary to my early impressions
while he was a member and even slightly beyond. in Zurich, related earlier, the aim of the treatise was
This no doubt accounts to a large extent for the not the utmost generality in itself, but rather the
uniformity of style of the volumes, frustrating any most efficient one, the one most likely to fill the
effort to try to individualize one contribution or needs of potential users in various areas. Refine-
the other. But this was not really Dieudonné’s ments of theorems which seemed mainly to titil-
style, rather the one he had adopted for Bourbaki. late specialists, without appearing to increase sub-
Nor was it the personal one of other Bourbaki stantially the range of applications, were often
members, except for Chevalley. Even to Bourbaki discarded. Of course, later developments might
he seemed sometimes too austere, and a draft of show that Bourbaki had not made the optimal
his might be rejected as being “too abstract”. The choice.8 Nevertheless, this was a guiding principle.
description “severely dehumanized book…”, given Besides, many discussions took place outside the
by Weil in his review of a book by Chevalley [12, sessions about individual research or current de-
p. 397] is one many people would have applied to velopments. Altogether, Bourbaki represented an
Bourbaki itself. Another factor contributing to this awesome amount of knowledge at the cutting edge
impersonal, not user-friendly presentation6 was the which was freely exchanged.
very process by which the final texts were arrived This made it obvious that for Bourbaki current
at. Sometimes a heuristic remark, to help the research and the writing of the “Éléments” were
reader, would find its way into a draft. While read- very different, almost disjoint, activities. Of course,
ing it, in this or some later version, its wording the latter was meant to supply foundations for the
would be scrutinized, found to be too vague, am- former, and the dogmatic style, going from the gen-
biguous, impossible to make precise in a few words, eral to the special, was best suited for that purpose
and then, almost invariably, thrown out. (see [5]). However, the “Éléments” were not meant
As a by-product, so to say, the activity within to stimulate, suggest, or be a blueprint for, re-
Bourbaki was a tremendous education, a unique search (as stressed in [8, p. 144]). Sometimes I
training ground, obviously a main source of the have wondered whether a warning should not have
breadth and sharpness of understanding I had been included in the “Mode d’emploi”.
been struck by in my first discussions with Bour- All this bore fruit, and the fifties was a period
baki members. of spreading influence of Bourbaki, both by the
The requirement to be interested in all topics treatise and the research of members. Remember
clearly led to a broadening of horizon, maybe not in particular the so-called French explosion in al-
so much for Weil, who, it was generally agreed, had gebraic topology, the coherent sheaves in analytic
the whole plan in his mind almost from the start, geometry, then in algebraic geometry over C , later
or for Chevalley, but for most other members, as in the abstract case, and homological algebra. Al-
was acknowledged in particular by Cartan [7, p. xix]: though very much algebraic, these developments
This work in common with men of very 7“Ce travail en commun avec des hommes de caractères
different characters, with a strong per- très divers, à la forte personnalité, mus par une com-
sonality, moved by a common require- mune exigence de perfection, m’a beaucoup appris, et je
ment of perfection, has taught me a dois à ces amis une grande partie de ma culture ma-
thématique.”
6Called “abstract, mercilessly abstract” by E. Artin in his 8For instance, the emphasis on locally compact spaces in
review of Algebra [1], adding however “…the reader who Integration, on which P. Halmos had expressed strong
can overcome the initial difficulties will be richly rewarded reservations in his review [11], indeed did not address the
for his efforts by deeper insights and fuller understand- needs of probability theory, and this led to the addition
ing” (p. 479). of a chapter (IX) to Integration.

376 NOTICES OF THE AMS VOLUME 45, NUMBER 3

also reached analysis, via Schwartz’s theory of dis- gebraic geometry and soon came up with a version
tributions and the work of his students B. Mal- of the Riemann-Roch theorem that took everyone
grange and J.-L. Lions on PDE. Early in 1955 A. Wein- by surprise, already by its formulation, steeped in
stein, a “hard analyst”, had told me he felt safe from functorial thinking, way ahead of anyone else. As
Bourbaki in his area. But less than two years later major as it was, it turned out to be just the be-
he was inviting Malgrange and Lions to his insti- ginning of his fundamental work in algebraic geom-
tute at the University of Maryland. etry.
I am not claiming at all that all these develop- The fifties was thus outwardly a time of great
ments were solely due to Bourbaki. After all, the success for Bourbaki. However, in contrast, it was
tremendous advances in topology had their origin inwardly one of considerable difficulties, verging
in Leray’s work, and R. Thom was a main contrib- on a crisis.
utor. Also, K. Kodaira, D. Spencer, and F. Hirzebruch Of course there were some grumblings against
had had a decisive role in the applications of sheaf Bourbaki’s influence. We had witnessed progress
theory to complex algebraic geometry, but unde- in, and a unification of, a big chunk of mathemat-
niably the Bourbaki outlook and methodology were ics, chiefly through rather sophisticated (at the
playing a major role. This was recognized early on time), essentially algebraic methods. The most suc-
by H. Weyl in spite of the critical comments men- cessful lecturers in Paris were Cartan and Serre,
tioned earlier. Once R. Bott told me he had heard who had a considerable following. The math-
negative remarks on Bourbaki by H. Weyl in 1949 ematical climate was not favorable to mathemati-
(similar to those I knew about), but by 1952 the lat- cians with a different temperament, a different
ter said to him, “I take it all back.” Others, how- approach. This was indeed unfortunate, but could
ever (like W. Hurewicz, in a conversation in 1952), hardly be held against Bourbaki members, who
would assert that all that had nothing to do with did not force anybody to carry on research in their
Bourbaki, only that they were strong mathemati- way.9
cians. Of course, the latter was true, but the in- The difficulties I want to discuss were of a dif-
fluence of Bourbaki on one’s work and vision of ferent, internal nature, partly engineered by the
mathematics was obvious to many in my genera- very success of Bourbaki, tied up with the “second
tion. For us H. Cartan was the most striking illus- part”, i.e., the treatise beyond the first six books.
tration, almost an incarnation, of Bourbaki. He In the fifties these were essentially finished, and
was amazingly productive, in spite of having many it was understood the main energies of Bourbaki
administrative and teaching duties at the École would henceforth concentrate on the sequel; it
Normale Supérieure. All his work (in topology, sev- had been in the mind of Bourbaki very early on
eral complex variables, Eilenberg–MacLane spaces, (after all, there was still no Traité d’Analyse). Al-
earlier in potential theory (with J. Deny), or har- ready in September 1940 (Tribu No. 3), Dieudonné
monic analysis on locally compact abelian groups had outlined a grandiose plan in twenty-seven
(with R. Godement)) did not seem to involve brand books, encompassing most of mathematics. More
new, groundbreaking ideas. Rather, in a true Bour- modest ones, still reaching beyond the “Éléments”,
baki approach, it consisted of a succession of nat- also usually by Dieudonné, would regularly con-
ural lemmas, and all of a sudden the big theorems clude the Congresses. Also, many reports on and
followed. Once, with Serre, I was commenting on drafts of future chapters had already been written.
Cartan’s output, to which he replied, “Oh, well, However, mathematics had grown enormously,
twenty years of messing around with Bourbaki,
that’s all.” Of course, he knew there was much
9In this connection, I would like to point out that the sub-
more to it, but this remark expressed well how we
felt Cartan exemplified Bourbaki’s approach and title Le choix bourbachique in [9] is extremely misleading.
Bourbaki members gave many talks at the seminar and
how fruitful the latter was. At the time Cartan’s in-
had much input in the choice of the lectures, so it is fair
fluence through his seminar, papers, and teaching to say that most topics discussed were of interest to at least
was broadly felt. Speaking of his generation, R. Bott some members, but many equally interesting ones turned
said of him, “He has been truly our teacher,” at the out to be left out, if only because no suitable speaker ap-
colloquium in honor of Cartan’s seventieth year [4]. peared to be available. So the seminar is by no means to
The fifties also saw the emergence of someone be viewed as a concerted effort by Bourbaki to present a
who was even more of an incarnation of Bourbaki comprehensive survey of all recent research in math-
in his quest for the most powerful, most general, ematics of interest to him and a ranking of contributions.
and most basic—namely, Alexander Grothendieck. Such conclusions by Dieudonné are solely his own. He says
that much in his introduction, p. xi, but it seems worth re-
His first research interests, from 1949 on, were in
peating. Of course, like most mathematicians, Bourbaki
functional analysis. He quickly made mincemeat members had strong likes and dislikes, but it never oc-
of many problems on topological vector spaces put curred to them to erect them as absolute judgments by
to him by Dieudonné and Schwartz and proceeded Bourbaki, as a body. Even when it came to his strong be-
to establish a far-reaching theory. Then he turned lief in the underlying unity of mathematics, Bourbaki pre-
his attention to algebraic topology, analytic and al- ferred to display it by action rather than by proclamation.

MARCH 1998 NOTICES OF THE AMS 377

 the mathematical landscape had changed consid- we had to be more systematic and provide first
erably, in part through the work of Bourbaki, and foundations for this topic itself. His counterpro-
it became clear we could not go on simply follow- posal was to have as the next two books:
ing the traditional pattern. Although this had not
Book VII: Homological Algebra
been intended, the founding members had often
Book VIII: Elementary Topology
carried a greater weight on basic decisions, but they
were now retiring10 and the primary responsibil- the latter to be tentatively subdivided into:
ity was shifting to younger members. Some basic
Chap. I: Topological categories, local categories,
principles had to be reexamined.
gluing of local categories, sheaves
One, for instance, was the linear ordering and
Chap. II: H 1 with coefficients in a sheaf
the system of references. We were aiming at more
Chap. III: H n and spectral sequences
special topics. To keep a strict linear ordering
Chap. IV: Coverings
might postpone unduly the writing of some vol-
umes. Also, when that course had been adopted to be followed by
at the beginning, there was indeed a dearth of
Book IX: Manifolds
suitable references. But Bourbaki had caught on,
some new books were rather close to Bourbaki in which had already been planned.
style, and some members were publishing others. He also added a rather detailed plan for the
To ignore them might lead to a considerable du- chapter on sheaves that I shall not go into.
plication and waste of effort. If we did not, how This was surely in the spirit of Bourbaki. To op-
could we take them into account without de- pose it would have been a bit like arguing against
stroying the autonomous character of the work? motherhood, so it had to be given a hearing.
Another traditional basic tenet was that everyone Grothendieck lost no time and presented to the
should be interested in everything. As meritorious next Congress, about three months later, two
as it was to adhere to, it had been comparatively drafts:
easy while writing the “Éléments”, which consist
Chap. 0: Preliminaries to the book on mani-
of basic mathematics, part of the baggage of most
folds. Categories of manifolds, 98 pages
professional mathematicians. It might, however, be
harder to implement it when dealing with more spe- Chap. I: Differentiable manifolds, The differen-
cialized topics closer to the frontier. The prospect tial formalism, 164 pages
of dividing up, of entrusting the primary respon-
sibility of a book to a subset of Bourbaki, was lurk- and warned that much more algebra would be
ing but was not one we would adopt lightly. These needed, e.g., hyperalgebras. As was often the case
questions and others were debated, though not con- with Grothendieck’s papers, they were at points
clusively for a while. There were more questions discouragingly general, but at others rich in ideas
than answers. In short, two tendencies, two ap- and insights. However, it was rather clear that if
proaches, emerged: one (let me call it the idealis- we followed that route, we would be bogged down
tic one) to go on building up broad foundations in with foundations for many years, with a very un-
certain outcome. Conceived so broadly, his plan
an autonomous way, in the tradition of Bourbaki;
aimed at supplying foundations not just for ex-
the other, more pragmatic, to get to the topics we
isting mathematics, as had been the case for the
felt we could handle, even if the foundations had
“Éléments”, but also for future developments to the
not been thoroughly laid out in the optimal gen-
extent they could be foreseen. If the label “Chap-
erality.
ter 0” was any indication, one could fear that the
Rather than remain at the level of vague gen-
numbering might go both ways, Chapters -1,
eralities, I would like to illustrate this dilemma by
-2,…being needed to give foundations to founda-
an example.
tions, etc.
At some point a draft on elementary sheaf the- On the other hand, many members thought we
ory was produced. It was meant to supply basic might achieve more tangible goals in a finite time,
background material in algebraic topology, fibre not so fundamental maybe, but still worthwhile.
bundles, differential manifolds, analytic and alge- There were quite a number of areas (algebraic
braic geometry. However, Grothendieck objected11: topology, manifolds, Lie groups, differential geom-
etry, distributions, commutative algebra, algebraic
10It had been apparently agreed early on that the re-
number theory, to name a few) in which they felt
tirement age would be fifty (at the latest). However, when
the Bourbaki approach might produce useful ex-
the time came to implement that rule, from 1953 on,
there was little mention of it until 1956, when Weil wrote positions, without needing such an extensive foun-
a letter to Bourbaki announcing his retirement. From dational basis as a prerequisite.
then on it has been strictly followed. The ideal solution would have been to go both
11At the March 1957 Congress, later called “Congress of ways, but this exceeded by far our possibilities.
the inflexible functor’’. Choices had to be made, but which ones? The

378 NOTICES OF THE AMS VOLUME 45, NUMBER 3

question was not answered for some time, result- 130 pages, was submitted, one member remarked
ing in a sort of paralysis. A way out was finally ar- that it was all right, but really Bourbaki was spend-
rived at a year later: namely, to write a fascicle of ing too much time on such a minor topic, and oth-
results on differential and analytic manifolds, thus ers acquiesced. Well, the final outcome is well
bypassing, at least provisorily, the problem of known: 288 pages, one of the most successful
foundations, at any rate for the main topics we had books by Bourbaki. It is a truly collective work, in-
in mind. After all, as far as manifolds were con- volving very actively about seven of us, none of
cerned, we knew what kind of basic material was whom could have written it by himself. Bourbaki
needed. To state what was required and prove it had developed a strong technique to elicit a col-
for ourselves was quite feasible (and was indeed laboration on a given topic by specialists and peo-
carried out rather quickly). ple with related interests looking at it from dif-
This decision lifted a stumbling block, and we ferent angles. My feeling (not unanimously shared)
could now set plans for a series of books which we is that we might have produced more books of that
hoped would essentially include commutative al- type but that the inconclusive discussions and
gebra, algebraic geometry, Lie groups, global and controversies, and the difficulties in mapping out
functional analysis, algebraic number theory, and a clear plan of activity had created a loss of mo-
automorphic forms. mentum from which Bourbaki never fully recov-
Again, this was too ambitious. Still, in the next ered. There is indeed a tremendous amount of un-
fifteen years or so a sizable number of volumes ap- used material in Bourbaki’s archives.
peared: This approach was less ambitious than the
Grothendieck plan. Whether the latter would have
Commutative Algebra (9 chapters) been successful, had we gone fully in that direc-
Lie Groups and Lie Algebras (9 chapters) tion, seems unlikely to me, but is not ruled out. The
Spectral Theory (2 chapters) development of mathematics does not seem to
besides preliminary drafts for several others. have gone that way, but implementation of that
plan might have influenced its course. Who knows?
In 1958 a decision had also been made to solve Of course, Bourbaki has not realized all its
in principle a problem which had been plaguing us dreams or reached all of its goals by far. Enough
for quite a while: additions to the “Éléments”. On was carried out, it seems to me, to have a lasting
occasions, while writing a new chapter, we would impact on mathematics by fostering a global vision
realize that some complement to one of the first of mathematics and of its basic unity and also by
six books was in order. How to handle this? Some- the style of exposition and choice of notation, but
times, if a volume was out of print, it was possi- as an interested party I am not the one to express
ble to include these complements in a revised edi- a judgment.
tion. If not, one could conceivably add an appendix What remains most vividly in my mind is the un-
to the new chapter. But this threatened to create selfish collaboration over many years of math-
a lot of confusion in the references. In 1958 it was ematicians with diverse personalities toward a
resolved to revise the “Éléments” and publish a common goal, a truly unique experience, maybe a
“final” edition, not to be tampered with for at least unique occurrence in the history of mathematics.
fifteen years. Unfortunately, it took more time and The underlying commitment and obligations were
effort than anticipated. It is in fact not quite fin- assumed as a matter of course, not even talked
ished as of now, and (I feel) it slowed down progress about, a fact which seems to me more and more
in the more innovative parts of the treatise. But it astonishing, almost unreal, as these events recede
was certainly in the logic of Bourbaki and hardly into the past.
avoidable.
Of the three books listed above, Commutative References
Algebra was obviously well within Bourbaki’s [1] E. ARTIN, Review of Algebra (I–VII) by N. Bourbaki, Bull.
purview; it could, and did in fact, proceed inde- Amer. Math. Soc. 59 (1953), 474–479.
pendently of the resolution of the dilemma we [2] L. BAULIEU, A Parisian café and ten proto-Bourbaki
had faced. But the fascicle of results on manifolds meetings (1934–35), Math. Intelligencer 15 (1993),
was an essential prerequisite for the book on Lie 27–35.
groups and Lie algebras. The latter also shows [3] ———, Bourbaki: Une histoire du groupe de math-
that the more pragmatic way could lead to useful ématiciens français et de ses travaux, Thèse. Uni-
work. A good example is provided by Chapters 4, versité de Montréal, 1989.
[4] R. BOTT, On characteristic classes in the framework
5, and 6 on reflection groups and root systems.
of Gelfand-Fuks cohomology, Colloque Analyse et
It started with a draft of about 70 pages on root Topologie en L’honneur de H. Cartan, Astérisque
systems. The author was almost apologetic in pre- 32–33 (1976), 113–139; Collected papers vol. 3,
senting to Bourbaki such a technical and special Birkhäuser, 1995, pp. 492–558.
topic, but asserted this would be justified later by [5] N. BOURBAKI, L’architecture des mathématiques, Les
many applications. When the next draft, of some Grands Courants de la Pensée Mathématique (F. Le

MARCH 1998 NOTICES OF THE AMS 379

 Lionnais, ed.), Cahiers du Sud (1948); English trans.,
Amer. Math. Monthly 57 (1950), 221–232.
[6] H. CARTAN, Nicolas Bourbaki and contemporary math-
ematics, Math. Intelligencer 2 (1979–80), 175–180.
[7] ———, Oeuvres Vol. 1, Springer, 1979.
[8] J. DIEUDONNÉ, The work of Nicholas Bourbaki, Amer.
Math. Monthly 77 (1970), 134–145.
[9] ———, Panorama des mathématiques pures, Le choix
bourbachique, Bordas, Paris, 1977.
[10] D. GUEDJ, Nicholas Bourbaki, collective mathematician:
An interview with Claude Chevalley, Math. Intelli-
gencer 7 (1985), 18–22.
[11] P. HALMOS, Review of Integration (I–IV) by Bourbaki,
Bull. Amer. Math. Soc. 59 (1963), 249–255.
[12] A. WEIL, Review of “Introduction to the theory of al-
gebraic functions of one variable by C. Chevalley”
Bull. Amer. Math. Soc. 57 (1951), 384–398; Oeuvres
Scientifiques II, Springer, 1979, pp. 2–16.
[13] ——— , Organisation et désorganisation en math-
ématique, Bull. Soc. Franco-Japonaise des Sci. 3
(1961), 23–35; Oeuvres Scientifiques II, Springer,
1979, pp. 465–469.
[14] ———, Notice biographique de J. Delsarte, Oeuvres de
Delsarte I, C.N.R.S., Paris 1971, pp. 17–28; Oeuvres
Scientifiques III, Springer, 1980, pp. 217–228.

380 NOTICES OF THE AMS VOLUME 45, NUMBER 3