Preface

Alexander Grothendieck enters this book several times before he is born. He enters under uncertain surnames in the borderlands of the Russian Empire; in the unfinished education and radical commitments of his mother; in photographs, identity papers, a novel written from memory, and family stories that do not always agree. Secondary and institutional sources place his birth in Berlin on March 28, 1928; the Berlin civil entry and exact registered name have not yet been inspected. Law, kinship, domestic life, and later recollection describe the same infant differently. The discrepancy is the first lesson in how that life can be known.

Heroic biography looks backward for signs: a childhood question anticipates a theorem, exile becomes the source of abstraction, the solitary child already contains the mathematician who will remake geometry. This book takes a different route. It follows documents before symbols, chronology before destiny, and relations before essence. Few twentieth-century mathematicians changed so many working assumptions or proposed structures of comparable reach. Their making required teachers, students, correspondents, institutions, money, editorial labor, and the freedom to remain in—or refuse—a shared world.

Greatness here resides in the scale of the structures Grothendieck made possible and in the exact history of their making. Statelessness, bereavement, collaboration, and solitude belong to the life without serving as hidden causes of the mathematics. The lived and mathematical histories illuminate one another when comparison remains comparison rather than destiny.

Across twelve Parts and seventy-two chapters, this biography follows an unusually long change of scale. It begins with a family archive whose smallest facts are contested. It passes through camps, schools, provincial examinations, postwar seminars, and the collective construction of a new geometry. It reaches an institute at the height of international prestige, then follows a documented rupture over military-origin funding into ecological and antimilitarist activism. It returns to university classrooms, private mathematical manuscripts, autobiographical indictment, spiritual writing, retirement, and a village in the Ariège. It ends after the death, among two distinct archival collections, new editions, inherited rights, unfinished theories, and mathematical communities whose work no longer requires the founder’s presence.

The book tells one chronological life through three inseparable universes. The lived universe: statelessness, war, migration, institutions, friendship, conflict, activism, withdrawal, and the archive left behind. The mathematical universe: the movement from isolated problem solving toward languages capable of reorganizing whole fields. The philosophical universe: simplicity, childhood, creation, community, responsibility, rupture, solitude, dreams, and the limits of possession. The word has a technical sense in his own mathematics: the universes of SGA 4 are sets large enough to hold the constructions of a theory without leaving set theory, and Chapter 33 describes them.11Michael Artin et al., eds., Théorie des topos et cohomologie étale des schémas (SGA 4), Tome 1, Lecture Notes in Mathematics 269 (Springer, 1972), https://doi.org/10.1007/BFb0081551, exposé I, appendix. The subtitle borrows the word as a figure; it makes no claim about that mathematics.

One movement holds this Preface together: a change of scale can make relations visible, but every enlargement meets a boundary. That movement governs how the book explains mathematics, collective work, refusal, silence, and the archive. It also governs how the narrative stops when its evidence stops.

Names before birth: evidence and relation

One of the deepest changes associated with Grothendieck can be stated in ordinary language. Instead of asking only what a mathematical object is by itself, ask how it varies, what maps into it, what maps out of it, how it appears after a change of base, and which constructions remain compatible across those changes. The object does not disappear. It becomes intelligible through a system of relations.

This shift takes several technical forms. A scheme is studied over a base, so that fibers and changes of ground belong to the problem. A moduli question follows whole families and asks whether one geometric object represents them. Descent asks whether compatible local objects come from a global one. A topos gives sheaves and cohomology a setting in which such local-to-global traffic can occur. Later chapters unfold the distinctions; here the governing point is that an object becomes richer when the maps, tests, and transformations around it are retained.

Grothendieck’s life also passed through names, countries, households, institutions, and communities. The resemblance is suggestive, but mathematical relations do not disclose their maker’s psychology. Schemes, descent, and cohomology arose from technical pressures: classical varieties were too narrow, sheaves had to be transported and glued, fibers had to be compared, and the Weil conjectures demanded new forms of locality. Biography can place those pressures beside the conditions of a life while leaving cause to the evidence.

Relation over object nevertheless provides this book with an ethical as well as a mathematical discipline. A celebrated name should not absorb the network in which it operated. Grothendieck’s results have relations to the work of Riemann, Roch, Lebesgue, Zariski, Weil, Cartan, Eilenberg, Mac Lane, Buchsbaum, Serre, Schwartz, Dieudonné, and many others. EGA has a relation to Dieudonné’s sustained redaction. SGA has relations to its individual expositors, co-directors, students, and later editors. The Weil conjectures have a relation to Grothendieck’s cohomological program and to Deligne’s proof of the remaining central estimate. Later motives, stacks, derived geometry, anabelian geometry, and higher categories have relations to Grothendieck’s programs and to the independent constructions of those who made them rigorous in forms he did not supply.

A related discipline of attention applies to the nonmathematical life. A civil record places a person in relation to a state; a memoir places later memory beside an earlier event; a letter records an address from one person to another; a resignation records an institutional act. These historical relations have no common proof structure with morphisms or descent. Their value lies in keeping source, date, speaker, and claim visibly connected.

This method also prevents the book from becoming either prosecution or defense. Grothendieck’s later accusations about credit must be heard because they are part of his intellectual and moral history. They must also be tested against publications, correspondence, bylines, seminar records, and the independent testimony of the people accused. His ecological warnings deserve historical seriousness without being made infallible by mathematical authority. His spiritual writings deserve description without devotional endorsement or retrospective diagnosis. The relative view refuses the false choice between reverence and dismissal. It asks: relative to which document, which date, which speaker, which institution, which standard of proof?

Long before the research institutions that made Grothendieck’s name familiar, later biographical and memorial sources identified his father as Alexander or Sascha Schapiro using Tanaroff papers and reported an amputation, imprisonment, and anarchist activity. Authenticated civil and prison records remain open, as do the alleged sentence and commutation and the scale of his civil-war activity. The evidence does securely distinguish him from the better-documented anarcho-syndicalist Alexander Schapiro. That correction removes an attractive borrowed biography and leaves the father’s own record thinner.

Public memorial, family, and biographical sources give his mother’s name as Johanna Frida Elisabeth Grothendieck and record the later forms Hanna and Hanka. Those sources place her in a northern German household whose economic security deteriorated and describe training, artistic ambition, reform movements, political experiment, writing, performance, and precarious work. Her autobiographical novel, Eine Frau, is literary self-representation: evidence of how she shaped a life in prose rather than civil proof of each episode.

Jackson and later biographies report that Sascha and Hanka met within the political and artistic geography of Weimar Berlin and lived amid irregular photographic income, prior relationships, children across households, and the vulnerability of Tanaroff papers. Secondary and institutional sources place Grothendieck’s birth in Berlin on March 28, 1928. The legal birth-name, paternity, legitimacy, and marriage sequence remains unresolved pending inspection of Berlin civil and family-court records. Naming here is an unsettled series of family and administrative acts, not a label silently attached at birth.

After 1933, National Socialist power made an already precarious household dangerous. Biographical sources report that the young Alexander lived with Wilhelm and Dagmar Heydorn near Hamburg from late 1933 or early 1934 until 1939 and place the household in an anti-Nazi milieu; underlying family and placement records remain unchecked. His parents went toward Spain and the anti-fascist struggle. Distance divided political commitment from daily care, while correspondence and later recollection preserve a relationship across it.

By 1939 the arrangement could no longer hold. Grothendieck traveled from Germany to France and rejoined his parents during the collapse of the Spanish Republic and the approach of another war. French policy toward foreigners and enemy nationals fractured the household again. Inspected biographies report Hanka and her son at Rieucros, Sascha’s internment at Le Vernet, his deportation in August 1942, and his death at Auschwitz. The intermediate Drancy, convoy, arrival, and death-record chain awaits direct archival inspection.

The father’s death remained part of the son’s life without becoming the moral capital of his career. Mathematical eminence offers no compensation for murder. The documentary sequence carries all the weight the narrative can ask of it.

Grothendieck and his mother survived through another network. Named biographies and institutional histories place him in the Le Chambon-sur-Lignon orbit and at the Collège Cévenol, where refuge and schooling depended on teachers, religious communities, relief organizations, hosts, administrators, and other refugees. The route after Rieucros, the possible Gurs or Brens layers, the residential facility, enrollment dates, and labor details remain unclosed. Collective protection is secure at the level of the network; its exact itinerary is not.

War did not stop education. The surviving record supports study and examinations, while Grothendieck’s later recollection identifies dissatisfaction with the way length, area, and volume had been taught. Those questions acquired importance through the work that followed, not as prophecies staged in a classroom.

Liberation brought neither a restored family nor an uncomplicated nationality. Grothendieck entered the University of Montpellier in the 1945–1946 academic year. Accounts describe a modest scholarship, agricultural work, and material difficulty, but exact household claims remain unevenly documented. The legally precise onset and continuity of statelessness are among the questions that require records not yet opened. It is safest to describe the condition as it appears in the available biographical and administrative evidence without inventing a single juridical turning point.

Institutional and scholarly biographies describe Grothendieck as stateless through 1971 and as obtaining French citizenship that year. They also connect particular travel and employment constraints to documentary status, although the onset, continuity, naturalization decree, and effective date await archival confirmation. Statelessness was a concrete vulnerability among his many relations as student, son, partner, parent, employee, and colleague; it was never a total identity.

At Montpellier, Grothendieck later recalled pursuing questions about length, area, volume, measure, and integration that the local curriculum did not satisfy. Secondary accounts characterize the work as an independent reconstruction of measure theory. No authenticated student manuscript has yet been checked against Lebesgue theory, so its exact definitions, scope, and omissions remain open.

This episode establishes a habit without establishing a destiny. Grothendieck wanted something beyond a formula that returned the right number. He wanted to know what made the formula an instance of a coherent construction. He later recalled declining, for a time, to consult the established solution because he wished to let the question disclose its own necessities. The method could waste effort, miss counterexamples, and produce gaps; his recollection acknowledges as much. It could also reveal why a definition had the shape it did. Understanding came to mean rebuilding the setting in which a result became unavoidable.

Montpellier could not offer a research community equal to that inquiry, but it supplied examinations, a degree, books, and a professor who saw that Paris offered what the local faculty could not. Dependence and independence coexisted. This will recur throughout the life. The institution that fails one need can still supply another; the person who refuses its intellectual limit can still require its certificate.

From solitary questions to a relative world

Grothendieck later described his 1948 arrival in Paris as an encounter with unfamiliar vocabulary—groups, fields, rings, modules, complexes, and homology—and biographical accounts emphasize gaps between his Montpellier training and the language assumed in research seminars. He arrived with an unusual measure manuscript as well. The mismatch shows the distance between reconstructing one subject alone and entering a community in which many subjects have already been organized.

The Cartan seminar and the Parisian mathematical network exposed both the strength and the limit of solitary reconstruction. Henri Cartan, André Weil, Jean Dieudonné, Laurent Schwartz, Jean-Pierre Serre, and others represented different ages, institutions, and mathematical temperaments. They should not be arranged retrospectively as a pantheon assembled to recognize a future master. At that moment Grothendieck was a newcomer with uncertain professional prospects. Statelessness complicated ordinary routes to appointments and travel. What mattered was not ceremony but the identification of a problem and a place where he could work on it.

Nancy provided that place from October 1949. Schwartz and Dieudonné were developing functional analysis beyond the established terrain of Banach spaces. Spaces of smooth functions, test functions, and distributions required topologies governed by families of seminorms rather than one norm. Duality, tensor products, boundedness, completeness, and infinite-dimensional operations did not line up automatically. Schwartz and Dieudonné offered a list of hard questions. Grothendieck’s rapid progress altered his status from inadequately prepared student to an active contributor.

Functional analysis can look remote from the algebraic geometry for which he became known. It already displayed several enduring demands. A distribution is a continuous linear functional on a specified space of test functions, so a generalized object is understood through all admissible probes. A topology can be chosen to make the right operations continuous. Tensor products require a setting in which bilinear relations can be represented linearly. Nuclear spaces identify conditions under which competing completions and tensor constructions behave with unusual harmony.

Topological vector spaces arose because ordinary normed spaces were too narrow for central objects of analysis. Nuclearity controlled operators, tensor products, and duality in settings tied to distributions and partial differential equations. These were live questions posed by mentors and predecessors, not symbols waiting to be decoded as personality.

Nancy also corrects the myth of intellectual self-creation. Schwartz and Dieudonné selected the terrain, offered problems, read work, and connected Grothendieck to a professional community. Grothendieck supplied solutions of extraordinary range and speed. Mentorship became collaboration without ceasing to be asymmetric. Hanka remained part of the material and emotional perimeter of these years, but the available record does not license an intimate domestic reconstruction. The dedication associated with his thesis and later accounts establish gratitude more securely than imagined conversations.

The doctoral work appeared across publications whose dates and exact administrative sequence require care. This book keeps the memoir, journal publication, degree process, and later reputation distinct. A result can be completed, presented, accepted, and published at different times. The common habit of assigning one year to an entire intellectual transition creates false simultaneity. The same caution will matter for EGA and SGA, where seminar year, fascicle circulation, revision, and book publication often diverge by a decade.

After Nancy came appointments and travel in São Paulo and Kansas, followed by a return to France. These years were not decorative bridges between analysis and geometry. In São Paulo, a different mathematical community and teaching setting widened the professional world. In Kansas, Grothendieck worked on the foundations of homological algebra that would become the 1957 paper commonly called Tôhoku. The nickname refers to the journal, not to a visit to Japan. Writing belongs to the Kansas period; publication belongs to a later year; canonical status belongs to the subsequent use of the paper.

The paper’s central move can be explained without requiring the reader to calculate a derived functor. Algebra often studies operations that preserve part of an exact sequence and fail to preserve the rest. Homological algebra records the failure as structured information. Higher Ext, Tor, and cohomology groups do more than announce that something went wrong; they measure the obstruction and arrange it across degrees. Cartan and Eilenberg had systematized this machinery for modules. Buchsbaum had already sought a broader exact setting. Grothendieck developed an axiomatic environment in which the method worked for categories including sheaves, proved major existence results for injectives under stated conditions, and established machinery for composing derived functors.

An abelian category is not a mist in which objects disappear. It is a setting where arrows can be added, kernels and cokernels behave well, and exact sequences retain the formal properties needed for homological arguments. The achievement was portability. A proof that used only this grammar could move from modules to sheaves and future examples. The limits also matter. Infinite sums, filtered colimits, enough injectives, generators, and size conditions require explicit hypotheses. A universal language is useful because it names its conditions, not because it makes conditions vanish.

The Tôhoku paper also displays the discipline behind universal language. A universal property specifies a class of candidate objects and a unique morphism in a stated direction; initial and terminal constructions point in opposite directions, and each theorem must say which one it uses. Universality therefore means exact relational characterization, not applicability to everything. Grothendieck repeatedly sought this form of understanding, with generality following from the conditions rather than from a taste for altitude.

During the same period, sheaves and cohomology drew him toward geometry. Serre’s work on coherent algebraic sheaves and the relation between algebraic and analytic geometry supplied major foundations. Category theory, introduced by Eilenberg and Mac Lane, supplied language already in use. Weil’s arithmetic program and Zariski’s local methods supplied problems and predecessors. The transition was not an analyst waking one morning as an algebraic geometer. It was a convergence of tools whose compatibility had become visible.

Grothendieck’s reformulation of Riemann–Roch offers an early, decisive example. The classical theorem had developed from questions about meromorphic functions on curves. Hirzebruch had carried it to nonsingular complex projective varieties through characteristic classes. Grothendieck changed the subject from an invariant attached to one variety to compatibility under a proper morphism. K-theory encoded additivity from exact sequences; direct images transported sheaf-theoretic information; characteristic classes translated between kinds of data. When the target was a point, the earlier theorem reappeared as a special case.

The published history prevents a single-author fable. The new theorem was presented in 1957 and expounded in print by Armand Borel and Jean-Pierre Serre in 1958, with explicit attribution to Grothendieck’s unpublished results. Grothendieck supplied a separate appendix on Chern classes. Authorship of the exposition and origin of the theorem were related but distinct. The episode is a compact model for this biography’s account of collective mathematics: name each act at the resolution the documents allow.

By 1958 the questions surrounding algebraic geometry had outgrown the available objects. Varieties over fields did not comfortably hold arithmetic families over the integers, infinitesimal structure, degeneration, specialization, and change of base within one grammar. Cohomology suitable for the Weil conjectures would require constructions stable across characteristics. Moduli problems required a way to treat families, not only isolated examples. Foundations were not a retreat from research. They were what research had begun to demand.

Grothendieck’s address at the International Congress in Edinburgh announced part of this need. That same year, correspondence records the beginning of the writing project that became Éléments de géométrie algébrique, or EGA. The timing is useful because it catches the project before later monumentality concealed its provisional character. There was an ambitious program, a collaborator beginning sustained redaction, notes and letters moving among mathematicians, and an expected schedule that would prove radically optimistic.

A scheme is the central object in the new foundation, but its significance lies in the relations it permits. For a commutative ring $A$, the prime ideals form the points of $\operatorname{Spec} A$; a topology and structure sheaf record how algebra behaves locally. Schemes are built by gluing such affine pieces. The definition allows ordinary varieties, arithmetic bases, generic points, special fibers, and nilpotent infinitesimal information to occupy one category. It also admits objects that resist ordinary visualization.

The familiar phrase generalized space is useful for less than a page. It tells a general reader that geometry has been enlarged. It becomes false if it suggests that every scheme is a hidden shape in three-dimensional space. $\operatorname{Spec}\mathbb Z$ has closed points associated with prime numbers and a generic point associated with the zero ideal; changing base to a finite field selects an arithmetic fiber. Nilpotents may carry deformation information without adding visible points. The faithful intuition is not a strange picture. It is a setting in which algebraic variation, localization, and base change can be performed coherently.

The relative viewpoint changes the unit of attention. Instead of a scheme $X$ alone, consider a morphism $X\to S$. The base $S$ may parameterize a family. Pulling back along $S'\to S$ produces $X\times_S S'$, the same construction viewed after a change of ground. A property worth using in families must have an explicitly stated behavior under this operation. Geometry becomes a study of transport.

This is where relation over object acquires technical force. A point is one kind of test, but a whole scheme can be another. The functor of points asks, for every test scheme $T$, which morphisms $T\to X$ exist. Under the Yoneda principle, the compatible system of all such maps determines $X$ up to unique isomorphism. An object is recovered through every admissible relation into it. The principle does not mean that objects are illusions. It means that identity in a category can be expressed without opening an object to inspect hidden substance.

Representability turns this into a research method. Suppose a rule assigns to each test scheme $T$ a set of families, line bundles, divisors, curves, or other data, and pullback makes the assignment functorial. Is there a scheme $M$ such that maps $T\to M$ correspond naturally to those families? If so, $M$ represents the functor and carries a universal family. A classification problem has become geometry.

Representability is never guaranteed. Automorphisms may require a stack rather than a scheme. Separation can fail. A parameter object may be infinite-dimensional, or no acceptable object may exist. Grothendieck’s Fondements de la géométrie algébrique, associated with a sequence of Bourbaki expositions, developed representability and formal methods in concrete, difficult settings. FGA should not be described as a third book parallel to EGA and SGA. It is a publication history built from expositions and later collections. Its role was to show how the functorial question could yield Picard schemes, Hilbert schemes, formal existence results, and general criteria under precise hypotheses.

The relative world also changed how old objects were valued. Nilpotents, once easy to dismiss because they do not create new ordinary points, could record an infinitesimal thickening and hence a possible deformation. A fiber over one prime could be compared with fibers over others. A family could reveal which features persist and which jump. A singular member was no longer treated as a defective isolated object; it might be a boundary point of a moduli problem and a source of information about the whole family.

The intellectual power of this shift should not obscure its ancestry. EGA itself names Zariski, Weil, Serre, Chevalley, Nagata, commutative algebra, homological algebra, and sheaf theory among the sources from which it worked. Bourbaki contributed standards, personnel, and a structural discipline, but EGA was not Bourbaki mechanically applied to geometry. The driving problems came from geometry and arithmetic. The language changed because those problems resisted inherited categories.

The making of EGA belonged equally to an institutional transition. Léon Motchane founded the Institut des Hautes Études Scientifiques in 1958. Jean Dieudonné became its first permanent mathematician and, in the institution’s later account, made Grothendieck’s recruitment a condition of his own. The early institute worked from Paris before moving to the Bois-Marie property at Bures-sur-Yvette in 1962. Publications Mathématiques de l’IHÉS provided a vehicle for work whose length and serial form exceeded ordinary journal expectations.

This sequence matters because institutions are often described as neutral containers for genius. IHÉS was made through fundraising, negotiation, appointments, buildings, journals, visitors, and administrative choices. It gave Grothendieck unusual freedom and a concentrated international community. His work gave the new institute extraordinary scientific authority. Neither side can be reduced to benefactor and beneficiary. Their later conflict arose inside a relation of mutual construction.

The new geometry soon required measuring systems equal to its change of scale. Cohomology often begins when locally available data fail to produce a global solution. Functions that agree on overlaps do glue; the obstruction lies in finding compatible choices, or in the cocycles left when local choices differ. Many such obstructions can be encoded cohomologically in graded groups whose structure can be transported and compared.

For algebraic varieties over the complex numbers, topology supplied familiar cohomology. Over finite fields and in positive characteristic, there may be no underlying complex manifold whose ordinary topology can be borrowed. Weil’s conjectures concerning point counts and zeta functions demanded an algebraic cohomology with properties analogous to those of a good topological theory: finite-dimensionality, functoriality, products, duality, cycle classes, and an action of Frobenius whose traces encode arithmetic counts.

The first requirement was a new notion of local covering. Zariski-open sets are often too coarse. Étale maps behave, in one useful sense, like local isomorphisms without requiring inclusion as an open subset. The analogy with covering spaces helps explain why finite étale covers can support an algebraic fundamental group. It stops where topological paths are assumed. In arithmetic geometry the group is reconstructed from the category of covers and a fiber functor, not from loops drawn in an ambient manifold.

A site abstracts this choice of locality. It is a category equipped with families of morphisms declared to cover, subject to compatibility conditions. A sheaf on the site is data that restricts and glues according to those covers. This separates the idea of locality from a predetermined set of open subsets. The étale site of a scheme permits local observations that the Zariski topology cannot see.

Descent asks the complementary question. If objects are given over a cover and isomorphisms identify them on overlaps, with the identifications satisfying a cocycle condition, do they come from one object downstairs? The slogan local data glue is insufficient. The overlaps, arrows, and compatibility on triple overlaps are part of the substance. Effective descent is a theorem under hypotheses, not an automatic feature of any attractive collection of pieces.

Descent carries philosophical resonance because it refuses to regard a global object as a pile of local appearances: overlap data determine whether a genuine whole exists. The analogy stops before mathematical diagrams are treated as social proofs.

Topoi enlarge the setting again. A Grothendieck topos is, in one standard description, a category equivalent to sheaves of sets on a site. Different sites can present equivalent topoi, so the world of sheaves is more intrinsic than any one presentation. Ordinary topological spaces produce examples, but a topos cannot be reduced to a space with unusual open sets. It is a category with limits, colimits, exponentials, a subobject classifier, and a distinctive internal logic. In geometry, it provides a habitat for sheaves and cohomology; in logic, it supports variable mathematical universes.

The word world is useful if its limit is stated. A topos supplies rules for how objects, truth values, and local sections behave internally. It does not create an alternative physical universe. Nor should its logical applications be credited entirely to Grothendieck: the theory developed through Lawvere, Tierney, Giraud, and many others, in mathematical conversations whose branches had their own authors.

The seminars known collectively as SGA were the engine of this construction. The first numbered seminar treated étale coverings and the fundamental group. Later programs addressed local cohomology, group schemes, topoi and étale cohomology, $\ell$-adic methods, L-functions, intersections, Riemann–Roch, and monodromy. EGA stabilized foundational grammar while SGA tested, extended, and taught it against active problems. The separation is imperfect because each depended on the other.

SGA was not a transcript of one person speaking. Seminar years, oral expositions, typed fascicles, revisions, and publication dates form distinct layers. SGA 1 carries the seminar year 1960–1961 and a 1971 publication date in its revised edition. The three volumes of SGA 4 appeared in 1972 and 1973 after the 1963–1964 seminar. To cite only the book year can make the research late; to cite only the seminar year can transport later redaction backward. This biography gives both when they matter.

The collective form was real and unequal. Grothendieck chose programs, supplied concepts and prenotes, assigned expositions, corrected drafts, and exercised exceptional authority. Participants proved results, filled gaps, proposed generalizations, wrote expositions, and later edited volumes. Individual bylines, co-directorships, and editorial formulas should survive summary. The phrase Grothendieck’s seminar identifies an organizing center. Used possessively to claim every theorem, it becomes false.

The architecture of cohomology grew through this system. Derived functors translated failures of exactness into higher groups. Spectral sequences organized calculations across layers. Base-change theorems controlled the relation between fibers and families; duality related complementary degrees; $\ell$-adic constructions supplied coefficients suitable for Frobenius. The Grothendieck–Lefschetz trace-formula program connected fixed points and arithmetic counts through work spread across SGA 4 and SGA 5. SGA 4 credits Artin, Grothendieck, and Verdier as directors; SGA 5 records the 1965–1966 seminar and appeared in 1977 under Illusie’s editorship, while theorem-level credit remains with named expositors. This collective architecture advanced the Weil program without completing the analogue of the Riemann hypothesis.

The distinction is crucial for attribution. Grothendieck proposed and directed a cohomological architecture in which the Weil conjectures could be attacked. Deligne later proved the remaining central estimate. Saying that Grothendieck proved the Weil conjectures is false. Saying that Deligne’s proof occurred in an unrelated landscape is equally false. History needs a grammar capable of naming program, framework, theorem, proof, and continuation separately.

Motives arose from the plurality of cohomologies. Betti, de Rham, étale, and crystalline theories seemed to reveal different realizations of something shared. Grothendieck proposed motives as a universal source for such realizations, linked to algebraic cycles and equivalence relations. The musical metaphor of one motif sounding through several arrangements is memorable and incomplete. It does not construct a category, settle the standard conjectures, or supply comparison functors.

There is no single finished category of motives possessing every expected property. There are rigorous categories of pure motives under specified choices, triangulated and mixed constructions, motivic cohomology, and motivic homotopy theories, alongside conjectural structures. The standard conjectures were intended to secure deep positivity and equivalence properties for algebraic cycles, with consequences for a satisfactory theory. Their openness was not a minor gap beneath a completed vision. It marked the boundary between program and theorem.

This period culminated publicly in the 1966 Fields Medal. The International Congress of Mathematicians met in Moscow, and Grothendieck did not attend. The absence is documented; the exact motive is not fixed by a contemporaneous public explanation in the present record. Later biographies place it within his pacifism and political refusals, but consistency with later action cannot become a missing acceptance speech. It should not be called a refusal of the medal itself. His explicit refusal of a prize came in 1988 under different circumstances.

The absent laureate is an apt but dangerous emblem. It can make political withdrawal appear inevitable years before the institutional break. In 1966 Grothendieck remained intensely engaged at IHÉS, completing large texts, directing seminars, corresponding, and opening further programs in crystals, motives, and monodromy. The summit was not already the fall. Chronology requires that later knowledge be withheld from the earlier moment.

The house, its many hands, and the price of refusal

By the mid-1960s the mathematical school stood near its height, while the institution that sustained it had become a house in more than one sense. IHÉS consisted of an estate and seminar rooms, but also schedules, trains, visiting appointments, salaries, publication budgets, secretarial work, governance, and money from sources that did not carry the same political meaning for every participant.

At Bois-Marie, Grothendieck’s Tuesday seminars combined preparation and openness. Witnesses remember lucid exposition, careful recall of prerequisites, severe standards for written work, and long correction sessions. Students could receive questions that determined years of research. Being chosen could feel like recognition and like an obligation. A leader able to see a field’s architecture could give unusually fertile problems; the same authority could narrow the space in which another person’s priorities appeared.

The biographer should not settle this tension with one adjective. Generous and dominating can describe different effects of the same act. Patient with a basic question and exacting about a comma can belong to one teacher. A collective enterprise can be shared in fact and strongly hierarchical. The purpose of naming contradictions is not to turn a person into an enigma. It is to avoid making one witness cancel another.

The printed record makes the plurality visible. EGA attributes the work to Grothendieck with the collaboration of Dieudonné. Retrospective accounts gathered by Borel and Jackson, including Illusie’s testimony, describe Dieudonné organizing and rewriting material, completing proofs, controlling notation, and contributing mathematics. SGA title pages and tables of contents identify directors, editors, collaborators, and exposé authors. Hartshorne’s Residues and Duality emerged from Grothendieck’s program and prenotes while remaining Hartshorne’s book. Posters name Deligne and Grothendieck together as directors of the local monodromy seminar before the breach. Katz, Illusie, Berthelot, Verdier, Raynaud, Artin, and many others entered the work under distinct bibliographic forms.

Four kinds of making repeatedly overlap: proposing a concept or program, constructing definitions and proofs, writing or redacting an exposition, and editing or producing a stable publication. Standard authorship compresses them. Biography should expand them again without pretending that every lemma can be assigned from surviving evidence. Typists and production staff did not become coauthors of theorems, but their skilled work made thousands of pages navigable. Grothendieck’s 1967 instructions for manuscript preparation reveal a precise infrastructure of notation, punctuation, references, correction, and return to authors when doubt arose. The mathematical monument had a workflow.

The ethic of this system depended partly on abundance. Problems and ideas could be given because a large program would benefit from their development in other hands. A student’s theorem could become part of a shared architecture. Tacit knowledge within a close community might preserve origin even when every paper did not repeat a name. Once the community dispersed, those assumptions became unstable. A printed byline could record the expositor without recording the person who proposed the problem; a program director could remember an idea as part of his construction while the person who proved it remembered the labor that made it true.

Grothendieck’s later writings would transform these tensions into accusations of burial and appropriation. They must not be imported backward as secret causes already governing the 1960s. During the summit years, many witnesses remember openness, excitement, and shared work. The later grievance is evidence of a changed judgment, not proof that generosity was false. Conversely, earlier generosity does not make all later complaints imaginary. The chronology allows both records to exist and asks that each claim about credit be tested at its proper date.

The mathematics at the summit also remained unfinished. EGA had reached a vast published extent without completing its original plan. SGA volumes required years of redaction. Motives and the standard conjectures opened a horizon rather than closing one. The scale of the work makes it tempting to interpret incompletion psychologically, as exhaustion or despair. The documents establish expansion, competing programs, editorial burden, and unfinished texts. They do not establish one private cause.

Political action had already entered the life before the IHÉS dispute. Grothendieck traveled to North Vietnam in 1967 and taught during a war in which air raids were a material condition. May 1968 altered the French political landscape and the relation between universities, authority, and public dissent. Neither event alone explains the 1970 resignation. They belong to an accumulation of encounters in which mathematical work and organized violence could no longer be held in separate moral compartments.

The house at Bures had enabled extraordinary independence from routine university demands. That independence required support. Motchane assembled a mixed financial structure from public, private, French, European, and American sources. Its negotiated plurality of patrons allowed research freedom while leaving open who could set an ethical boundary for the institution. Here a method successful in mathematics reached its limit: enlarging the context clarified the conflict but could not dissolve it.

The immediate rupture can be narrated without a psychological theory. An audited statement for 1969 lists 90,000 francs received from the Ministry of the Armies. The amount formed a small but nontrivial part of the institute’s finances, roughly four percent of contributions and subsidies and a smaller share of total income depending on the denominator used. In a budget under pressure, it was neither the foundation of IHÉS nor an amount without practical consequence.

The evidence does not show that Grothendieck or his seminar performed a weapons project. The dispute concerned the origin of the money and the moral meaning of accepting it. This distinction should not be used to trivialize the objection. For Grothendieck, provenance itself crossed a boundary. For colleagues seeking continuity, the absence of directed military research and the possibility of assigning the funds to visits in physics could matter. The disagreement was partly a disagreement over whether use could contain origin.

The chronology separates several acts often compressed into one dramatic departure. Conflict entered the scientific committee on April 6, 1970. The board minutes of April 22 record Grothendieck’s condition that such funding be refused if he were to remain, the position of physicists Louis Michel and David Ruelle, and the board’s decision to accept a proposed 90,000-franc grant while giving priority to invitations in physics. Grothendieck’s letter of May 25 announced his decision and a probable October 1 departure. A June 9 letter confirmed October 1 as the effective date.

The break was a process. On June 26, Grothendieck spoke at Orsay on the responsibility of scientists and the military apparatus. In July, he lectured on current mathematics at a summer school in Montreal while making space for a public address about science and military institutions. The first Survivre bulletin dated the movement’s formation to July 20 and its first general meeting to July 24; the issue appeared in August, while later accounts disagree over founder and editorial labels. At Nice, during the International Congress of Mathematicians, political distribution and congress rules came into conflict. Mathematical and public work therefore overlapped the documented departure.

The military-funding sequence has the strongest documentary claim as the proximate cause of departure from IHÉS. Broader accounts name other factors: political transformation, the Vietnam War, May 1968, accumulated conflict with Motchane, work intensity, dissatisfaction with the profession, and changing priorities. Some may belong to a fuller explanation, but their evidentiary status differs. A later self-interpretation cannot be made contemporary by smooth prose. An institutional cause need not become a total explanation of a life.

The ethical force of resignation came from consistency between a stated condition and an enacted consequence. Its ethical difficulty came from dependence. Grothendieck did not leave work belonging only to himself. Students and collaborators were writing inside programs he had directed. Seminar volumes remained unfinished. Colleagues had commitments to visitors and institutions. Asking others to resign with him made solidarity materially consequential, but it also asked people whose careers and mathematical choices were connected to his authority to accept his political judgment. Deligne and the other permanent professors did not leave. Their refusal was not a theorem against Grothendieck’s conscience; his disappointment was not proof of their betrayal.

Professional continuity after the breach further complicates the word exile. Grothendieck held a temporary position at the Collège de France, described in his own contemporary account as proposed and later renewed through Serre’s initiative. He also taught at Orsay. Technical mathematical activity continued. These positions did not recreate the permanent post, visitor network, publication machinery, and seminar authority of IHÉS. They show continuity without cancelling rupture.

Survivre, later Survivre et vivre, widened the argument beyond one budget line. Its early program joined nuclear danger, militarism, pollution, ecological imbalance, industrial society, resource destruction, scientific authority, and the separation of science from life. Grothendieck was a principal initiator, but the movement was not his private instrument. Gordon Edwards and a broader founding and editorial group participated from the Montreal phase; Claude Chevalley and Pierre Samuel became important in France; later contributors altered the bulletin’s emphasis and form.

The name supplied a minimum rather than a complete politics: survival of humanity and life. That breadth allowed weapons, pollution, technology, population, food, energy, and professional responsibility to appear in one field. It also created disagreements about priority, expertise, and action. Should scientists begin by refusing military money? Should the movement first analyze capitalism, state power, and industrial production? Could specialized knowledge be used without reproducing the hierarchy being criticized? Was catastrophic urgency compatible with slow collective deliberation?

Grothendieck’s answer often began with individual refusal. A scientist implicated in military systems should decline support and warn others; analysis that indefinitely postponed action risked becoming accommodation. The argument was morally clear and practically incomplete. Removing one grant could not dismantle an armaments system. Technical knowledge, public institutions, and collective organization remained necessary. The movement’s history records attempts to move from personal purity toward a broader ecology of responsibility.

Public lectures became assemblies rather than one-way transmissions. Mathematical reputation drew audiences, while the talks questioned the social value of the activity that had produced the reputation. This contradiction was productive and unstable. Grothendieck could challenge scientific hierarchy from a platform made available by scientific distinction. Listeners could agree about nuclear danger and reject his prescription. Younger militants could regard the recent dissident from IHÉS as an established mandarin. Leaving an elite institution did not erase the social authority accumulated within it.

The activism also changed the meaning of expertise. Grothendieck’s mathematical achievements gave him no special authority to predict ecological timelines or settle political economy. He learned from activists, scientists, pamphlets, travel, and public debate. Some warnings were prescient in direction; some forecasts and formulations were uneven. The book neither borrows certainty from the Fields Medal nor uses error in one domain to diminish achievement in another.

The years 1969–1971 therefore form no simple conversion from pure mathematics to politics. They show a refusal of the separation itself. Mathematical work, institutional finance, military systems, environmental danger, pedagogy, and public authority had become connected questions. The connection did not yield one durable organization or a complete social theory. It altered what Grothendieck was willing to treat as outside the mathematician’s responsibility.

By 1973, the most intense phase of his direct involvement in the movement was passing, though the bulletin continued until 1975 under changing leadership. He had neither returned to the old research life nor left institutions altogether. The next phase would unfold in the public university where his own formal mathematical education had begun.

Work beyond the center: reaping, sowing, pursuing

Grothendieck took up a professorship at Montpellier during the 1973–1974 academic year. The appointment is an ordinary administrative fact with unusual corrective power. Three years after the IHÉS resignation, he had not ceased to be employed as a mathematician. He entered a different academic ecology: regular courses, examinations, degree structures, students of uneven preparation, fewer advanced interlocutors, and far less control over the surrounding research agenda.

The return to Montpellier creates a biographical symmetry that should be distrusted before it is used. The young student had found the curriculum inadequate and pursued his own foundational question. The professor now had to teach within a curriculum and decide how another person’s question could become serious work. A university cannot require every student to rediscover a theory alone. It must transmit common language, establish standards, and still leave room for problems that the syllabus has not authorized. Grothendieck’s later teaching experiments can be read as efforts to hold these demands together. There is no evidence that he arrived in 1973 with the answer already formed.

The setting was marginal only in relation to the research center he had left. Montpellier was an established university city, and Grothendieck held rank, salary, and international reputation. Students knew that their professor had received a Fields Medal. Yet he stood at a distance from the seminar circuits, journals, visitor networks, and concentrated school through which research prestige was reproduced. In Esquisse d’un programme, he later said that he had ceased publishing scientific articles after 1970. That self-report concerns ordinary article publication: private manuscripts continued to circulate, and earlier collective EGA and SGA work continued to appear in print. Geography, teaching, conflict, and narrowing contact added other forms of distance.

Teaching brought mathematics to a smaller scale. Courses and student projects turned toward surfaces, elementary geometry, groupoids, and research initiation. The word elementary did not mean easy or trivial. A surface can be cut, marked, glued, and deformed in ways that expose topology before a large technical machine is introduced. Several mathematical languages can describe the same surface: topological, differentiable, conformal, algebraic. Passing among them raises exact questions about what each language remembers.

Grothendieck’s own retrospective assessment was mixed. He regarded the Montpellier encounter as fertile for his mathematical imagination and often judged his research-level teaching unsuccessful. The judgment belongs to him and should not become an objective grade assigned to every student. It does establish that the later period was not a pastoral success story in which a famous mathematician chose ordinary teaching and found uncomplicated renewal. Preparation gaps, institutional requirements, scarce positions, different expectations, and the difficulty of communicating an enormous private program limited what could become collective work.

The classroom also displaced the idea of first principles. At Nancy, Grothendieck had entered a field through hard questions posed by established mathematicians. At Montpellier, some students needed a question whose surface could be seen before the technical language was mastered. Research initiation therefore had to distinguish productive ignorance from simple lack of equipment. A teacher could honor a student’s intuition while still showing where proof, counterexample, and inherited theory changed the problem.

These pedagogical years formed the foreground of an extensive private workshop. Letters, course papers, manuscripts, and later catalogs show mathematical production across the 1970s and 1980s. The works have different states. Some are complete documents with stable publication histories; some are long investigations circulated by copy; some are programs; some are editorially reconstructed much later; some remain blocked from strong claims because their manuscripts and provenance have not received adequate inspection. The mere quantity of paper proves activity, not correctness or completion.

La Longue Marche à travers la théorie de Galois, associated with 1981, belongs to this workshop. Its complete extent and critical status remain insufficiently verified for the strongest bibliographic claims, so the title should not be made to carry a finished hidden theory. Its subject returns to the absolute Galois group and to the problem of understanding an immense symmetry through more accessible geometric and combinatorial structures. The route would lead toward dessins, moduli, fundamental groups, and anabelian reconstruction.

The movement back toward algebraic geometry was neither a repudiation of activism nor a restoration of the IHÉS period. The objects were often lower-dimensional and visually elementary; the mode of circulation was personal rather than institutional; the intended architecture remained broad. Curves and surfaces sit at a junction where topology, complex analysis, algebraic geometry, and arithmetic meet. Their apparent smallness concentrates relations rather than reducing them.

An event in the late 1970s illustrates the new communication system. Through Deligne, Grothendieck learned of Belyi’s theorem, which connects algebraic curves over the algebraic closure of the rational numbers with finite maps to the projective line ramified only over $0$, $1$, and $\infty$. The theorem traveled through publications, letters, and personal report rather than through a seminar directed by Grothendieck. Its consequences became central to his later program.

For a Belyi map $\beta:X\to\mathbb P^1$, the inverse image of the interval $[0,1]$ is a bipartite graph embedded on the oriented surface $X$: a dessin d’enfant. The vertices lie over $0$ and $1$, and the embedding supplies cyclic order data. This structured dessin, rather than an arbitrary drawing, records the branched covering up to the appropriate equivalence. The absolute Galois group acts on Belyi pairs and hence on dessins; Belyi’s theorem and Grothendieck’s later dessins program remain distinct achievements.

The charm of the picture can conceal the difficulty. A drawing does not visibly reveal its field of definition. Similar combinatorial features need not determine a Galois orbit. One curve may carry many Belyi maps. The theorem normalizes branching; it does not make arithmetic transparent. The program’s force lay in making invisible symmetry act on finite objects without pretending that the action had thereby been classified.

Grothendieck also turned toward the Teichmüller tower: the interconnected moduli spaces of curves with marked points, together with operations of gluing, forgetting, degeneration, and the actions on their fundamental groups. The popular image of Lego pieces captures assembly from elementary forms. It fails if it suggests rigid blocks, no automorphisms, or effortless compatibility. Curves deform; boundary strata intersect; gluing in different orders requires relations among relations. The hoped-for reconstruction from low levels was a program, not a theorem already completed in the sketch.

Anabelian geometry asked a related but distinct question. For suitable arithmetic varieties, how much geometry can be reconstructed from a nonabelian profinite fundamental group equipped with Galois action? The phrase far from abelian signals the refusal to discard commutator information that may encode geometry. Later work separated isomorphism, homomorphism, birational, and section problems under different hypotheses; the chapters name individual authors and versions rather than folding that theorem history into a claim that every space is determined by a group.

The Montpellier years also returned Grothendieck to stacks. A moduli problem involving objects with automorphisms is not faithfully represented by a set of isomorphism classes. Local objects may be isomorphic rather than equal; their isomorphisms must agree on overlaps; the global result retains stabilizers. A stack is a category-valued form of sheaf satisfying descent for objects and arrows. Higher stacks add further levels of coherent comparison.

This is not only an ascent in abstraction. It is a refusal to throw away symmetry to obtain a simpler answer. A coarse moduli space can be useful, but it forgets information. The stack keeps the automorphisms because they affect deformation, intersection, and gluing. Once again an object is understood through relations, including relations among its relations.

The private workshop did not develop in total isolation. Students, correspondents, former colleagues, and mathematicians in Bangor, Paris, and elsewhere exchanged pages and questions. Selective contact is not the same as a seminar community, but it refutes the image of twenty years without mathematical relation. The hidden work was hidden from ordinary professional visibility, not created outside all social infrastructure.

One later document makes the continuity visible. Esquisse d’un programme accompanied Grothendieck’s effort to leave regular teaching for a CNRS research position. Its printed terminal date is January 1984, while a curated catalog assigns a 1983 label whose basis has not been fully reopened. Composition had antecedents in 1983 letters and manuscripts; formal publication came only in 1997. Writing date, circulation, and publication are separate facts.

The Esquisse takes stock by proposing work. It moves among dessins, the absolute Galois group, marked curves, the Teichmüller tower, tame topology, higher structures, and questions arising from teaching. Some statements are theorems, some conjectures, some desired foundations, and some heuristic images. To praise the document by turning every task into an achievement would destroy the programmatic tense that gives it meaning.

Its administrative setting sharpens the mathematics. Grothendieck argued that research-level teaching at Montpellier had become unproductive for him and that no local colleague or successor seemed likely to develop the perspectives he was describing. He also acknowledged the scarcity of research positions and the cost of appointing an established mathematician. The CNRS application was neither the manifesto of an unknown outsider nor the restoration of the Bois-Marie school. It sought public support for a Fields medalist who described himself as mathematically isolated.

The sketch’s proposal for tame topology reveals an enduring method. General topology admitted phenomena too wild for the geometry he wanted to control. Rather than exclude pathologies one by one, he sought categories of spaces stable under gluing, stratification, tubular neighborhoods, fibrations, and isotopy. The proposal remained incomplete. Its significance lies in the attempt to choose a world whose native operations match the questions, as schemes and topoi had done in different settings.

Récoltes et Semailles began in June 1983 and expanded through 1986. There was no single finished book suddenly present at the end of composition. Movements acquired notes; provisional fascicles circulated; letters produced supplements; later preliminary texts reorganized the entrance; digital surrogates omitted portions; published editions created new textual objects. The work’s changing state is part of its history.

The title joins harvest and seed. Reaping gathers mathematical work, relationships, conflicts, institutions, and the consequences of earlier choices. Sowing addresses a future the writer cannot control. The agricultural language changes the temporal form of mathematical autobiography. A theorem can be printed as if detached from discovery; a harvest carries labor, season, inheritance, and the uncertainty of what will grow from the seed.

The work calls itself reflections and testimony on a past as a mathematician, and its genres exceed ordinary autobiography. Mathematical history, personal recollection, ethical meditation, polemic, self-analysis, correspondence, and recursive commentary occupy the same structure. This makes it indispensable and hazardous. It is direct evidence of Grothendieck’s language and his judgment in the 1980s. It is not independent proof of every event recalled or accusation made.

One of its most valuable contributions is an account of mathematical creation that resists both command and prestige. The figure of the child names the capacity to ask an elementary question without shame and to remain with it before convention supplies an answer. The contrasting figure of the boss names the internalized demand for rank, possession, and control of appearances. These are textual figures, not psychiatric entities. They organize moral reflection and should not be converted into diagnoses.

The child can become another heroic myth if removed from community. Montpellier supplied the story of solitary reconstruction, but Nancy supplied mentors and questions. EGA required Dieudonné. SGA required expositors and editors. A private insight becomes mathematics through definitions, examples, counterexamples, proof, criticism, and transmission. Solitude may protect attention; it cannot by itself create public validity.

The manuscript’s controversy concerns what happens after ideas are given. Grothendieck believed that former students and colleagues had failed to credit, publish, or continue parts of his work and had participated in a burial of his mathematical person. He also examined his own vanity and the wish to possess creation. The result oscillates between gift and ownership. If a giver retains control over every future use, the gift becomes conditional. If recipients detach an idea from the people and conditions that produced it, circulation can become erasure.

There is no universal accounting rule that resolves every case. A person who proposes a theorem and a person who proves it have different claims. A student who turns prenotes into a rigorous exposition performs authorship even when the program came from a teacher. A field may stop repeating a name because an idea has become grammar, but tacit attribution can disappear for later readers. A recipient remains free to choose another research life. Intellectual gratitude requires honesty; it does not establish permanent obedience.

This biography therefore separates credit from continuation and both from friendship. A missing citation, an abandoned program, a delayed seminar volume, and a broken personal relation are not the same ethical act. Récoltes et Semailles often joins them under one moral weather. Historical analysis must take the judgment seriously enough to disaggregate it.

The book’s length and recurrence produce further difficulties. Return can expose a new layer of self-correction or demonstrate that an earlier account failed to settle the writer’s concern. It can also amplify grievance and pressure the reader toward the author’s diagnosis. Severe editing might make the work more accessible while removing the temporal evidence Grothendieck valued. A critical edition has to decide how to preserve sequence, notes, variants, and contradiction without treating every repetition as sacred.

The material labor again matters. Typescripts, fair copies, handwritten correction, duplication, address lists, and mailing carried a solitary voice to roughly two hundred recipients in differing configurations. Secretarial and institutional reproduction work did not authorize the claims, but without it the testimony would not have circulated in legible form. The work that criticizes collective mathematics was itself transmitted through a small collective infrastructure.

Pursuing Stacks offers another form of expansion. Its prehistory lies in correspondence with Ronald Brown, Philip Higgins, Larry Breen, Tim Porter, and others concerned with higher groupoids and forms of the van Kampen theorem. A request for contact and an exchange of letters became a manuscript distributed in installments during 1983, last dated in November of that year. It was written in English because the correspondence through which it grew was conducted in English.

The title is partly misleading. Some sections directly address stacks, but much of the manuscript pursues homotopy types, higher groupoids, test categories, localizers, model structures, and coherence. Grothendieck himself described the foundational investigation as an unforeseen digression. The digression became the principal work because higher stacks require trustworthy models for the higher homotopical data they are meant to glue.

A groupoid retains several base points and invertible paths among them. A higher groupoid also retains homotopies between paths and further layers of comparison. Composition then holds through coherent higher data rather than strict equality. Grothendieck’s homotopy hypothesis and his search for test categories asked which such models faithfully present homotopy types. In 1983 these were guiding programs requiring adequate definitions and comparison theorems, not one completed formalism.

The manuscript did not contain all of modern higher or derived geometry. Later mathematicians supplied models, coherence theorems, representability results, and applications in frameworks with their own histories. Grothendieck articulated powerful questions and strategies. Georges Maltsiniotis, Denis-Charles Cisinski, and many others developed parts of the homotopical program. Modern higher categories and higher topoi were built through further work associated with numerous authors. A precursor is honored most accurately when the future is not placed inside his pages as completed possession.

The mode of circulation shaped reception. Photocopies moved through permission-based personal networks. Definitions remained provisional; errors remained visible; recipients could answer. There was no publisher fixing pagination or guaranteeing that two readers held the same state. Informal circulation can move ideas quickly and preserve the openness of discovery. It can also make priority, correction, access, and citation fragile.

During the same years, spiritual attention and mathematical work overlapped. Grothendieck’s interest in dreams did not begin after mathematics ended. La Longue Marche, Pursuing Stacks, the Faltings letter, Esquisse, and the beginning of Récoltes et Semailles occupied years in which he was also interpreting dreams and enlarging a personal religious vocabulary. The chronology resists a clean division between rational work and spiritual turn.

A scholarly chronology dates La Clef des Songes to 1987, and the BnF reports a corrected typescript. Exact place and start, item-level sequence, completeness, chapter count, pagination, and quotation rights await examination of the authenticated archival object. The publicly described text presents Grothendieck’s conviction that dreams came from a figure he called the Dreamer and identified with God; this is evidence of his stated belief, not verification of its supernatural claim.

The opposite reduction is diagnosis. Religious conviction, long writing, polemic, seclusion, and unusual cosmology do not by themselves establish a medical condition. No retrospective clinical label should replace description of the text. Claims can be criticized, contradictions shown, and standards of warrant compared without pretending to examine a patient through prose.

Dream interpretation and mathematical discovery share formal descriptions in Grothendieck’s writing: attention to material not yet understood, patience with resistance, the arrival of a relation that reorganizes details, and writing as an instrument of discovery. Their warrants remain different. A proof is public and rule-governed; another competent reader can inspect each inference. A revelation may be personally compelling and not reproducible by another reader. Similarity of experience does not make dreams proofs or proofs revelations.

CNRS and BnF accounts, read with Grothendieck’s April 1988 letter, place the research position from October 1984 to pension retirement in October 1988; personnel-file confirmation of the effective dates remains open. Retirement removed teaching and then institutional research status while writing continued. That year brought a final major public refusal. The Royal Swedish Academy of Sciences awarded the Crafoord Prize jointly to Grothendieck and Deligne. On April 19, Grothendieck declined it in a letter later published in Le Monde.

His stated reasons should govern the account. He considered his salary and approaching pension sufficient, criticized the concentration of money and prestige among established researchers, argued that time rather than honors tests an idea’s fertility, and condemned what he saw as a decline in scientific ethics. The letter proves that he made these arguments. It does not prove every accusation about the community. Deligne accepted his share. Their different acts require no invented exchange.

The contrast with 1966 is real. Grothendieck did not attend the Moscow ceremony but did not then issue an explicit refusal of the Fields Medal. In 1988 he refused the Crafoord Prize with reasons. The intervening twenty-two years changed his relation to recognition. Those reasons are documented; a hidden inner conversion is unnecessary.

Mathematical correspondence and large projects continued after administrative retirement. The manuscript known as Les Dérivateurs is reported by editors as an extensive work from 1990–1991, but exact extent, chapter structure, priority claims, and author-final status remain dependent on editorial descriptions that require further archive-level verification. Open letters from 1991 independently establish continuing technical thought and selective exchange. This book prefers the narrow fact to the maximal claim.

Silence, archive, and afterlife

Grothendieck moved to Lasserre in the Ariège in 1991. The exact arrival day is not established. Public and professional accessibility diminished drastically. “Seclusion” accurately describes his withdrawal from conferences, ordinary publication, broadly available correspondence, and a public address, while documented visits, letters, and practical relations continued.

Later documents make the limit visible. Papers were transferred. Letters were sent. A substantial manuscript sequence continued through the 1990s. In 2010, a declaration reached Luc Illusie with a request that it be circulated. Every act implies material and human infrastructure without restoring public life. Withdrawal controlled channels; it did not suspend dependence on roads, postal service, archives, and other people.

This book calls the place a village rather than a hermitage. A village is an inhabited civic place with roads, services, property, and neighbors. Hermitage imports a spiritual role; disappearance denies documented contact. The exact house, household routine, and local witnesses are not narrative trophies. Privacy remains relevant after death, especially where family members, neighbors, care workers, and correspondents would be exposed only because they were near a famous person.

The move did not cause the enlarged late cosmology. A scholarly chronology places La Clef des Songes among the late manuscripts of Grothendieck’s retirement period. Questions of dreams, psyche, creation, God, mathematics, activism, and solitude were already coexisting. Lasserre altered conditions of access and circulation. Geography cannot be made to explain every idea that followed.

The Bibliothèque nationale de France dates the large sequence Réflexions sur la Vie et le Cosmos from 1992 to 1999 and describes forty-one volumes. Named components include writing on elementary schematic geometry, psyche, Maxwell’s equations, creation, and the problem of evil. The scale is extraordinary and epistemically neutral. Forty-one volumes do not constitute one conclusion; thousands of pages do not make an argument true.

The crossing of disciplines is itself a historical fact. Mathematical diagrams and structural vocabulary enter reflections on psychology, physics, cosmology, and theology. The crossing must be read page by page. A formal pattern used as analogy does not provide mathematical proof in another domain. A physicist, historian of religion, philosopher, or clinician may ask questions different from those of an algebraic geometer. Grothendieck’s competence in one field does not automatically transfer authority to all others.

Dismissal is equally inadequate. A mathematical habit can shape serious thought outside mathematics without turning it into theorem. The historian must identify whether a diagram serves as analogy, proposed model, notation, or deduction; which premises support it; and what kind of criticism applies. The late corpus deserves readers competent in its several domains, not a single verdict borrowed from fame.

The cataloged sequence reaches 1999. Grothendieck lived for another fifteen years. That endpoint belongs to a surviving archival body, not necessarily to writing or thought. The record used here does not authorize a daily chronology after the catalog stops. Press anecdotes, sightings, private testimony, and retrospective interpretations must meet a higher threshold because the subject had withdrawn and the people around him did not volunteer to become scenery.

Silence is relational. To the mathematical profession, there were no regular papers, seminars, or conference interventions. To a manuscript page, there could be sustained work. To a correspondent, there might be a letter. To a biographer, silence means the absence of adequate evidence for a claim, not a discovered emotional state. It cannot be translated without support into shame, hostility, serenity, illness, holiness, or despair.

According to a publicly circulated facsimile and draft transcription, a covering letter dated January 3, 2010, transmitted a Déclaration d’intention de non-publication. The public text opposes unauthorized dissemination across scientific and personal writing, print and electronic media, letters, translations, and substantial extracts, while allowing brief quotation. The original has not been authenticated or compared phrase by phrase for this account, so this summary is attributed rather than offered as a close translation.

The declaration concerns intention, authorization, and withdrawal from circulation rather than a command to destroy unique manuscripts. Its temporal language cannot by itself decide posthumous rights, and its legal effect depends on records for particular works, contracts, succession, and transfers.

An earlier institutional finding aid attributes to Jean Malgoire the statement that Grothendieck destroyed some nonmathematical papers around the first transfer of mathematical material. It supplies neither a complete inventory nor a dramatic scene. This is distinct from the 2010 request to remove editions from circulation, and both are distinct from later archival preservation. Combining them into a wish to burn everything falsifies the record.

The declaration also exposes a paradox of public refusal. To limit circulation, Grothendieck had to circulate the limit. To oppose publication, he produced a text intended to be copied through an old mathematical network. The paradox requires no diagnosis. It is the ordinary dependence of refusal on an audience capable of honoring it.

Libraries can preserve an object while placing different conditions on consultation, reproduction, quotation, and publication. The layers matter: preservation keeps the page; a catalog makes it findable; permission governs a proposed use.

Grothendieck died on November 13, 2014, in the Ariège. The exact civil and medical details beyond the established public record remain subject to unresolved documentation, and no final scene is supplied. A death date does not authorize a hospital-room reconstruction, last thought, or reconciliation. The life ends where the evidence permits. The archive continues because custody has changed.

Death altered the papers before it altered their interpretation. Materials that had belonged to a living author with strong and sometimes prohibitive intentions became the concern of children, legatees, custodians, archivists, librarians, editors, mathematicians, rights holders, and readers. These actors do not possess the same authority. A repository can number a leaf without settling whether its argument is correct. An editor can make a text readable without acquiring the right to expose every person named in it. A mathematician can verify a proof without resolving the ethics of publication.

There is no single Grothendieck archive. One substantial mathematical corpus, covering much of the period from 1949 to 1991, traveled through Jean Malgoire to the University of Montpellier. Official accounts differ on whether Grothendieck entrusted material in 1990 alone or in stages in 1990 and 1995. They agree on institutional deposit in 2010. After inventory and conservation, digitization began in 2016. In May 2017, the university opened approximately 18,000 pages from a corpus described as approximately 28,000.

The difference between those approximate totals should not be called the secret remainder. Correspondence may involve third-party rights; materials may await processing; counting rules may change. A page can be a recto, a verso, a blank, an enclosure, or a reused sheet carrying two chronologies. Archival number, authorial page number, and digital image number are different things. The value of a digital portal lies not in visibility alone but in stable identifiers, sequence, provenance, and the ability to recover context.

The second major corpus entered the Bibliothèque nationale de France in late September 2023. The BnF describes the accession as a bequest by Grothendieck completed by a gift from his children. It includes the corrected typescript of La Clef des Songes and the forty-one-volume Réflexions sur la Vie et le Cosmos, estimated in public reports at roughly 30,000 pages. These figures and holdings must not be merged with Montpellier’s. The two collections have distinct provenance, genres, catalog structures, and access regimes.

An archive does not speak when opened. Handwriting must be transcribed; dates distinguished as authorial or assigned; physical order compared with later arrangement; versions collated; mathematical claims checked; references opened; and private statements assessed for relevance and harm. A plan must not become a completed theorem. A polemic must not become an impartial account of another person’s conduct. A corrected typescript may still be unfinished.

Archival order is itself evidence and intervention. Grothendieck organized some dossiers; Malgoire preserved material; university staff inventoried and scanned; BnF staff appraised and cataloged; digital interfaces sort; researchers request some units and ignore others. None of this discredits the record. It makes the chain of custody and selection part of what the historian must describe.

The posthumous editions add another layer. Editions of Récoltes et Semailles and La Clef des Songes make difficult texts available in new forms. The Société Mathématique de France’s 2022 Pursuing Stacks is Volume I, covering sections 1–91 and 95–98; Volume II remains announced rather than independently cataloged as published. Each edition has a distinct scope and history. They have appeared through editorial and estate processes whose work-specific authorization records have not been inspected here, so their legal status is not inferred from publication. The 2010 declaration remains part of that history.

Public study has widened. Montpellier’s archive supported exhibitions and research on mathematical work after 1973. The BnF and the Société mathématique de France convened a 2025 study day bringing activism, teaching, mathematics, autobiography, film, and late spiritual writing into one program. A mobile exhibition selected twelve panels from thousands of pages. Selection is not a defect if its purpose is stated. An exhibition, like a biography, is an argument about a route through material that remains larger than the route.

Mathematics meanwhile continued without waiting for a complete archive. Scheme theory became infrastructure. Sites, sheaves, descent, and six-operation formalisms traveled into new geometries. Stacks became routine in moduli and acquired higher and derived forms. Motives developed through several rigorous theories and persistent conjectures. Dessins, anabelian geometry, and Grothendieck–Teichmüller theory drew from the Esquisse while acquiring independent builders. Homotopy-theoretic programs from Pursuing Stacks entered research through later proofs, comparisons, and models.

The strongest evidence of legacy is not the atmospheric claim that Grothendieck is everywhere. It is a traceable relation. The Stacks Project provides a collaborative digital reference built on schemes, algebraic spaces, and stacks with its own maintainers and contributors. Voevodsky, Morel, and others constructed motivic homotopy and cohomology theories with theorems the earlier program did not contain. Toën, Vezzosi, Lurie, and many collaborators developed forms of derived and higher geometry. Scholze’s work on perfectoid spaces, diamonds, and $v$-stacks uses sites, descent, étale cohomology, and functorial operations in settings whose mathematics belongs to Scholze and his contemporaries.

An influence sentence needs at least two verbs. Grothendieck proposed, formulated, organized, or supplied a language; another mathematician defined, proved, constructed, corrected, or extended. Giving the later verb to its proper subject does not diminish the earlier one. It prevents legacy from becoming annexation.

The same rule applies backward. Belyi’s theorem remains Belyi’s. Deligne’s proof remains Deligne’s. Serre’s FAC and GAGA remain Serre’s. Dieudonné’s redaction remains intellectual labor. Weil’s conjectures remain Weil’s questions. Grothendieck’s achievement is large enough to survive accurate boundaries.

How this book knows what it knows

A biography of Grothendieck faces an unusual imbalance. Administrative records give some dates and amounts exactly; family history may travel from oral report to novel, late meditation, and scholarly biography; a mathematical result may have separate oral, manuscript, seminar, fascicle, edition, and reception dates. The prose marks those differences instead of smoothing them away.

Five levels guide its assertions. A documentary fact rests on a record speaking within its competence. Attributed testimony names the person remembering. Corroborated reconstruction brings independent sources together and preserves disagreement. Interpretation identifies itself as a reading. Speculation directs research but stays outside factual narration. Documents and memories answer different questions: a government file may classify accurately and understand badly, while a memoir may preserve an experience no office recorded.

Unknown means that the warrant has reached its boundary. Family testimony about Sascha remains testimony where prison files are missing. An absent reply cannot prove that no reply existed; a catalog endpoint cannot date the end of writing; Moscow without a public explanation supplies an act but no certain motive. Scenes follow the same rule. Supported actions and objects can make a vivid narrative; invented weather, dialogue, interior thought, and final glances cannot.

Grothendieck’s late writings carry two times at once. Récoltes et Semailles directly records his judgments in the 1980s and his retrospective account of earlier decades. La Clef des Songes records a spiritual self-understanding assigned by its scholarly chronology to 1987. Neither converts recollection into a contemporary record of childhood. A source chain also remains a chain: three books drawing from one interview do not become three witnesses.

The mathematical explanations begin with a problem and a concrete anchor, then give the minimum exact structure and the point where the image fails. A general reader should be able to identify the change of viewpoint and its consequence; a mathematician should be able to locate every suppressed condition. Metaphor opens a door, while definition determines what may pass through it. Mathematical ideas illuminate documented habits of work without diagnosing personality.

Historical language receives the same care. Modern definitions are identified as modern when period terminology differed. Seminar year, circulated fascicle, revised edition, and later reception keep separate dates. Program, theorem, proof, exposition, and editorship keep separate verbs. This is why the life cannot be organized as one rise and one renunciation: the IHÉS resignation, cessation of ordinary article publication, move to Montpellier, CNRS retirement, move to Lasserre, and 2010 declaration were different acts with different objects.

An immense archive sharpens rather than removes the limit. Page count measures paper, not the duration or importance of every relationship. Grothendieck selected, discarded, entrusted, retained, and ordered; Malgoire, university staff, BnF archivists, heirs, editors, and researchers added further acts of custody and selection. People named in those pages retain voices and privacy independent of the protagonist.

AI tools assisted with drafting, with the polishing and proofreading of the whole text, and with the search for, the verification and the organisation of the literature: they can compare wording, locate repetition, and test a sentence against the evidence supplied to them. They supply no human witness, mathematical expertise, archival authority, translation approval, or legal judgment. The conception of this biography, the design of its Parts and chapters and of their content, and the direction of the writing are the author’s; the author has reviewed the whole book and is responsible for its text.

The rising sea and its shore

Near the end, this book returns to an image associated with Grothendieck’s description of mathematical work. The present route is Colin McLarty’s secondary translation and analysis; the French passage and its mapping to an authorized edition have not been reopened, so no quotation or independent translation is printed here. In paraphrase, a resistant problem may be approached by enlarging the surrounding theory until the obstacle stands within a more natural landscape and the desired statement follows from the changed level. The rising sea belongs to that attributed mathematical image. The shore is this biography’s counter-image.

The image illuminates patience, universality, and conceptual enlargement. Abelian categories made separate homological constructions instances of one grammar. Schemes made arithmetic and geometric fibers part of one relative world. Topoi made locality and sheaf theory portable beyond ordinary spaces. Motives proposed a source beneath several cohomologies. Higher structures sought to retain coherence that lower categories discard. In each case, a broader setting changed which questions appeared natural.

The image also conceals labor if left alone. A theory rises because named people write definitions, check a base-change hypothesis, find a counterexample, correct an exposé, type a fascicle, secure a journal budget, or maintain a digital reference. Its apparent inevitability is usually visible only afterward.

Biography can use context to dissolve a false opposition, but it cannot make a person the consequence of a theory. The shore marks where one kind of explanation stops and privacy, contradiction, or singular responsibility begins.

The state assigned names; family memory carried what registers omitted; seminars joined unequal forms of authority; resignation made principle costly; manuscripts survived their author; later mathematicians built beyond his programs. Grothendieck’s mathematics enlarged the worlds in which problems could be seen, while his life exposed the limits of worlds made by states, universities, disciplines, prizes, movements, families, and archives.

After his death, access widened unevenly. Editions appeared under different conditions, repositories acquired distinct duties, mathematicians continued along paths with their own authors, and disputes about credit, privacy, and permission remained. The last movement of the book therefore returns to small, accountable acts. A leaf receives a shelfmark. A date is marked as authorial or assigned. A conjecture is separated from a theorem. A contested memory keeps the witness’s name. A translation names its translator. A later proof keeps its author.