功劳的伦理

The brief French epigraph defines mathematics through collective undertaking. In Récoltes et Semailles, it sits inside a work consumed by the question of how a group remembers. Ideas pass from conversation to blackboard, from seminar to exposé, from handwritten note to edited volume, from a teacher’s program to a student’s theorem. At each passage something is gained: clarity, proof, generality, accessibility, a new application. At each passage something can also be lost: context, uncertainty, the name of an originator, the labor of an expositor, or the freedom of a successor.11Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

Grothendieck had worked inside this system. His greatest period of mathematical production was inseparable from collaboration. Dieudonné helped turn a swiftly changing architecture into EGA. The Séminaire de Géométrie Algébrique depended on speakers, note takers, students, visitors, and later editors. Deligne, Verdier, Artin, Illusie, Raynaud, Giraud, Demazure, and many others received developing doctrines and also proved results, supplied counterexamples, developed formalisms, wrote exposés, and carried techniques into new fields. Any history that makes them anonymous assistants is false. Any history that erases Grothendieck’s organizing vision is false in another direction.22Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

The conflict of the 1980s arose partly because ordinary bibliographic forms could not carry this many-layered history. A title page asks for authors and editors. A citation points to a publication. A theorem may acquire a personal name. These devices are necessary; without them, readers cannot locate work or assign responsibility. But they are lossy. “Author” can mean originator of a program, writer of this proof, speaker of a seminar, or person who approved the final text. “Editor” can mean collator, mathematical reviser, rescuer of an abandoned manuscript, or architect of a new volume. The ethics of credit begins by refusing to treat those roles as interchangeable.33Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

In Récoltes et Semailles, Grothendieck described what he saw after years of distance. He believed that parts of his work had been used while their source and unifying purpose were minimized. He wrote of a “burial”: a process by which the living work and, at times, its author were treated as belonging to a completed past. The metaphor was not a neutral description. It transformed publication choices, research directions, citations, and personal silences into evidence of a communal rite. 44Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality, ch. 23.

The metaphor had real explanatory power. Mathematical communities do canonize some texts and let others vanish. They can celebrate a founder while ignoring an unfinished program. They can use a language so thoroughly that its origin becomes invisible. A person outside current networks may learn late that ideas have circulated under altered names. Prestige affects whose account is heard. “Burial” makes these mechanisms visible as acts rather than natural forgetting.

It also had a grave defect. It implied coordination and intention where the evidence might show divergence, editorial necessity, or independent choice. A student who pursued modular forms rather than motives had not necessarily betrayed a program. An editor who published seminar notes could be preserving work rather than appropriating it. A proof that bypassed Grothendieck’s preferred conjectures could be mathematically fruitful even if it disappointed him. The metaphor organized heterogeneous events under one moral verdict.55Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

For that reason, this chapter treats the allegations as claims made by Grothendieck, not facts established by the force of his prose. The distinction is not a polite compromise. It is the only way to protect all parties. If an accusation of theft is true, it requires evidence specific enough to identify what was taken, when, by whom, and how it was represented. If it is false or exaggerated, repeating it without attribution renews the harm. Great mathematical authority does not turn autobiographical judgment into proof.

Six kinds of contribution

A useful audit separates at least six roles. The first is problem or program formation: identifying a field of questions and relations worth pursuing. The second is concept formation: defining the objects and language in which those questions become tractable. The third is proof: establishing a precise statement under stated hypotheses. The fourth is exposition or teaching: making the argument intelligible and transmitting techniques. The fifth is redaction and editing: preparing unstable notes for reliable use. The sixth is application or continuation: taking concepts into domains the originator may not have anticipated.

One person can perform all six, but often no one does. Credit should follow the actual configuration. A program is not a proof. A proof of one case is not authorship of every later case. A lecture’s expositor may deserve visible credit even when the theorem is another’s. An editor who corrects and reorganizes a text can contribute mathematics without becoming the original speaker. An application can be new while depending on inherited language. 66Grothendieck, Récoltes Et Semailles, Partial Japanese Translation.

The distinction also clarifies ownership. Mathematical ideas are not property in exactly the sense of a physical manuscript or copyrighted prose. A theorem, once known, may be reproved and used. Professional ethics nonetheless demands attribution, honest description of dependence, and consent where someone’s unpublished text or name is used. These duties concern truth and trust, not a permanent right to control a field.77Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Grothendieck’s earlier practice had encouraged rapid circulation. Ideas were offered in letters and seminars before formal publication. Problems were given to students. Definitions changed collectively. This generosity helped create a school, but it also left a complex record. Later, when relations had broken, there was no simple boundary between what had been “given,” what had been collaboratively developed, and what remained a personal project. Retrospective claims of exclusive possession could not restore a boundary that had never been formalized.88Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

SGA after the seminar

The SGA volumes provide a concrete test. They arose from seminars at the Bois-Marie in which Grothendieck set much of the architecture and many participants delivered exposés. Publication often came years after the spoken event. Some volumes were edited after his 1970 departure. The lag created practical problems: notes needed correction, references needed stabilization, and important mathematics risked remaining inaccessible.99Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

SGA 4½, Cohomologie étale, appeared in Lecture Notes in Mathematics 569 in 1977 under Pierre Deligne’s name, with collaborators credited in the volume. It was not literally another half-year of the original SGA 4 seminar; its title announced a supplement or bridge. It supplied an efficient entry to étale cohomology and results needed by working mathematicians. Its formal record is public and checkable. 1010Deligne, Cohomologie étale (SGA 4½).

SGA 5, Cohomologie l-adique et fonctions L, also appeared in 1977, in Lecture Notes in Mathematics 589. The volume gathers work from the 1965–1966 seminar and visibly associates Grothendieck, Illusie, and others through exposé authorship and editorial history. Again, bibliographic presence does not settle every question about consent or adequacy. It establishes what the published object says. 1111Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

Grothendieck objected to aspects of post-IHÉS redaction and to uses of his name. One legitimate issue is consent: attaching a living person as coauthor or presenting a text under a collective label without consultation can misrepresent responsibility. Another is historical completeness: a volume may make results usable while narrowing the original program. A third is quality: editors can correct errors, but major intervention may create a new work whose authorship must be described.1212Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

Those issues require separate findings: whether Grothendieck was consulted, what appeared under his name, which exposés were his, what editors added, whether a new publication claimed originality for ideas traceable to earlier notes, and whether the result was mathematically correct and useful. A single word—“theft” or “burial”—cannot answer all of them.1313Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The usefulness of SGA 4½ is relevant but not exculpatory in itself. A useful book can have defective attribution. Conversely, imperfect consent does not make every theorem in a volume illegitimate. Ethical and mathematical evaluation overlap but are not identical. Grothendieck often fused them because he regarded the manner of doing mathematics as part of its truthfulness. A historian must preserve his concern while keeping the questions analytically distinct.1414Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

Deligne: succession without obedience

No relationship in the dispute carried more weight than the one with Pierre Deligne. Deligne had been Grothendieck’s student and became one of the mathematicians most capable of using the languages of schemes, cohomology, weights, and motives. His proof of the Weil conjectures followed a route that did not first prove Grothendieck’s standard conjectures. To Grothendieck, this could look like bypassing the architecture that would have explained the result in the desired generality. To others, including Serre, it was an extraordinary mathematical achievement precisely because it found a viable route.1515Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

The word “successor” is dangerous here. It can mean that Deligne occupied a position in an intellectual lineage. It must not mean that he accepted an obligation to complete every item in his teacher’s program. A student receives methods and problems but remains free to choose questions. Creative succession includes deviation. If every departure counts as betrayal, a school becomes hereditary administration.1616Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

Grothendieck’s criticism combined several complaints. He believed Deligne had used motivic ideas without adequately sustaining their source. He objected to publications and redactions. He lamented that former students did not continue the foundational program as a whole. These are not the same allegation. Using a known idea with citation, publishing seminar material, and choosing a different research agenda require different evidence and ethical standards.1717Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

Serre’s letter of July 23, 1985, answered with unusual directness. He defended Deligne as one of the most honest mathematicians he knew. He argued that the mathematical world recognized Grothendieck’s role in introducing the yoga of motives and that others were right to use it, provided they distinguished conjecture from proof. He admired Deligne’s method of going around the standard conjectures to prove the Weil conjectures. 1818Scharlau, Spirituality, ch. 23, pp. 4–5.

“Going around” a difficulty marked a difference of mathematical temperament. Grothendieck often sought a theory in which a theorem became natural, perhaps inevitable. Serre valued an argument that reached the result through available structure. Neither disposition can be made a universal rule. A foundational program can unify decades of work; a strategic detour can solve a problem and reveal new structure. The history of mathematics needs both.1919Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

Serre’s response also identified a missing question. On February 8, 1986, he asked why Grothendieck himself had abandoned the work whose noncontinuation he deplored. The question was not a denial of the 1970 ethical breach or of later mathematical activity. It challenged an asymmetry of responsibility. If no one else possessed the whole vision, then leaving it created a gap that could not be blamed on students alone.

There is no easy answer. Grothendieck left IHÉS over military funding and a broader conflict of values. He redirected enormous energy to ecological activism and later to teaching and private reflection. He had the right to leave. Former students had the right not to reconstruct the abandoned institution around him. The cost of both freedoms was that a collective project ceased to have its former center.2020Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The last meeting between Grothendieck and Deligne during this period, as reported in biographical testimony, did not resolve the conflict. Deligne visited Les Aumettes before moving to the Institute for Advanced Study. The documentary record does not authorize an invented conversation or a cinematic farewell. It establishes a visit and an enduring break. The silence afterward should not be filled with imagined motives. 2121Scharlau, Spirituality, ch. 23, p. 6.

Serre’s countervoice

Serre occupied a distinctive position. He had known Grothendieck before the IHÉS school and could understand both the mathematics and the scale of his claims. His letters do not read as a neutral verdict; they are interventions by a participant with his own values and loyalties. They nevertheless prevent Récoltes et Semailles from becoming a one-voice archive.2222Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

He distinguished recognition from obedience. The world could know that Grothendieck introduced motives while mathematicians used the idea in ways he disliked. Deligne could be deeply indebted to the earlier work and still originate new mathematics. Students could fail to continue one program without committing an ethical wrong. A founder’s disappointment was real but did not create a debt enforceable against other careers.2323Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

Serre also insisted on the difference between conjecture and theorem. A broad vision often includes statements that guide work before proof. Later authors who carefully mark the boundary do not diminish the vision by proving conditional or adjacent results. They give the community reliable ground. If they silently convert conjecture into attributed fact, that is a failure. If they solve a different theorem, it is not.2424Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

The correspondence eventually reached an impasse. That fact should not be narrated as proof that one man was incapable of dialogue or that the other lacked sympathy. Their conceptions of the dispute were incompatible. Grothendieck’s letters demanded acknowledgment of a moral and spiritual collapse; Serre answered claims about mathematics, agency, and fairness. Each text treated the other as missing the central issue.2525Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

Mebkhout and the problem of the unheard researcher

Zoghman Mebkhout stands at the center of another strand. Born in Algeria and working in France, he developed major results in the theory of $D$-modules and the Riemann–Hilbert correspondence. Grothendieck believed that Mebkhout’s contributions had been systematically denied proper recognition. He dedicated a major part of L’Enterrement to him and treated the case as evidence of institutional corruption. 2626Scharlau, Spirituality, ch. 23, pp. 8–9.

The historical question is difficult because priority in a fast-moving field can involve independent work, preprints, seminar circulation, partial results, and differences in theorem formulation. A fair account requires the original papers, submission and publication dates, correspondence, and specialist mathematical comparison. Récoltes et Semailles is evidence that Grothendieck advocated fiercely for Mebkhout. It is not sufficient by itself to establish a conspiracy or to adjudicate every priority claim.2727Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality.

Mebkhout’s relationship with Grothendieck was not created by the dispute. He had encountered Grothendieck’s methods through SGA 5 and sought him out. After an unsuccessful visit in 1979, they corresponded; a meeting took place in 1980. In April 1983 Grothendieck proposed work on a geometric topology of forms, and Mebkhout visited for days of extended mathematical discussion. Mebkhout’s move to a CNRS position helped prompt Grothendieck to consider applying there. These contacts gave the later advocacy a personal and mathematical history. 2828Scharlau, Spirituality.

The case exposes a real structural issue even before its disputed details are resolved. Recognition is uneven. Researchers outside dominant networks, from marginalized backgrounds, or without influential advocates can be overlooked. At the same time, advocacy by a celebrated figure can overwhelm the person ostensibly defended, making his work an instrument in another’s indictment. A biography should restore Mebkhout as a mathematician, not leave him only as Grothendieck’s example.2929Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality.

The safe wording is therefore exact. Grothendieck alleged systematic injustice toward Mebkhout. He presented extensive arguments and correspondence. Independent assessment requires a dedicated specialist study beyond this chapter. The existence of the allegation, its prominence, and its effect on relationships are facts. Its most expansive causal claims remain contested.3030Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality.

Self-accusation and asymmetry

It would be wrong to represent Récoltes et Semailles as accusation directed only outward. Fatuité et Renouvellement examines Grothendieck’s own vanity, appetite for recognition, and failures of attention. He revisits how the identity of a great mathematician formed and how creative power could coexist with blindness. The text’s credibility for many readers comes from this willingness to place the author under examination.

Yet self-criticism does not automatically balance criticism of others. A person controls the terms, depth, and eventual meaning of his own confession. He can present a fault as the prelude to renewal. The named colleague has no equivalent control inside the text. Even when letters are quoted, Grothendieck frames their position through placement and reply. Ethical reading must notice that asymmetry.3131Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The word “vanity” can also blur institutional and personal scales. Desire for recognition may influence behavior, but a citation dispute can be settled through documents without deciding anyone’s inner vanity. A hiring structure may be unjust even if every participant has good intentions. Conversely, bad feeling does not prove institutional theft. The manuscript often moves rapidly among these levels. Source criticism separates them.3232Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Grothendieck’s self-portrait as one who gave ideas freely is supported by many memories of seminars and letters. Generosity, however, can coexist with expectations never stated as conditions. A gift may carry the hope that the recipient will remain loyal to its spirit. When the recipient treats it as unrestricted, the giver experiences betrayal. That emotional structure is understandable; it does not create a retroactive contract.3333Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

The text’s most productive insight is that credit is not exhausted by citation. A footnote can name an originator while a surrounding narrative makes the idea appear inevitable or marginal. A student can cite a teacher and still fail to recognize an expositor, collaborator, or less prestigious predecessor. Conversely, abundant ceremonial praise can coexist with neglect of the actual unfinished work. Credit includes attention, historical narration, and opportunities for continuation.

Its least reliable move is to treat inadequate recognition as proof of coordinated moral decline. Communities are capable of theft and exclusion. They are also composed of overlapping memories, independent discoveries, limited time, and changing standards. A claim of conspiracy needs more than a pattern perceived by an injured participant. Responsible narration neither sanitizes the community nor grants infallibility to its critic.

Authorship as a chain of accountable verbs

One way out of the binary of possession and erasure is to write history with verbs. Grothendieck proposed schemes in a broad new form, organized programs, lectured, conjectured, proved, assigned, corresponded, and withdrew. Dieudonné redacted, checked, systematized, and coauthored. Deligne learned, proved, bypassed, edited, applied, and originated. Illusie lectured, developed, edited, and preserved. A title page cannot contain this syntax; a biography can.3434Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

Verbs impose accountability. “Inspired” should identify evidence of contact. “Anticipated” should compare precise statements, not moods. “Founded” should not absorb collaborators. “Completed” should not be used for an open manuscript. “Stole” requires proof of misrepresentation rather than dependence alone. “Published” requires a public edition, not a packet of typescript pages. This grammar is not pedantry. It is a method for distributing credit without converting intellectual history into property law.3535Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The same method protects successors. Drinfeld’s Grothendieck–Teichmüller group is related to the Teichmüller-tower program, but Drinfeld’s definitions and theorems are his. Mochizuki and Tamagawa proved anabelian results under specific hypotheses rather than executing a script. Maltsiniotis and Cisinski developed the homotopy ideas of Pursuing Stacks with extensive new work. Acknowledging the program clarifies their achievement; claiming their achievement for the program’s author diminishes everyone.3636Grothendieck, “Brief an G. Faltings”; Schneps, The Grothendieck Theory of Dessins d’Enfants; Tamagawa, “The Grothendieck Conjecture for Affine Curves”; Mochizuki, “The Profinite Grothendieck Conjecture for Closed Hyperbolic Curves over Number Fields”; Alex Heller, Homotopy Theories, Memoirs of the American Mathematical Society 383 (American Mathematical Society, 1988).

It also protects Grothendieck from ceremonial emptiness. To call him a visionary without naming the pages, dates, and unresolved questions reduces his work to aura. Precise priority can be more generous than praise. The Faltings letter of June 27, 1983, contains an anabelian program. Esquisse, terminal-dated January 1984 while carrying an unresolved 1983 catalog label, sets out dessins and the Teichmüller tower. Pursuing Stacks develops weak higher structures and test categories in 1983. These claims are strong enough without saying he proved what later authors proved.3737Grothendieck, “Brief an G. Faltings”; Schneps, The Grothendieck Theory of Dessins d’Enfants; Tamagawa, “The Grothendieck Conjecture for Affine Curves”; Mochizuki, “The Profinite Grothendieck Conjecture for Closed Hyperbolic Curves over Number Fields”; Heller, Homotopy Theories.

A concept-by-concept credit test

For dessins d’enfants, Belyi receives credit for the theorem bearing the criterion; Grothendieck receives credit for the programmatic recognition of dessins and Galois action in Esquisse; later authors receive credit for invariants, computations, and proofs.3838Grothendieck, “Brief an G. Faltings”; Schneps, The Grothendieck Theory of Dessins d’Enfants; Tamagawa, “The Grothendieck Conjecture for Affine Curves”; Mochizuki, “The Profinite Grothendieck Conjecture for Closed Hyperbolic Curves over Number Fields”; Heller, Homotopy Theories.

For anabelian geometry, Grothendieck receives credit for the named program and conjectural dictionary in the 1983 letter and related notes. Mochizuki, Tamagawa, and others receive theorem credit under their exact hypotheses. Earlier étale fundamental-group theory remains part of the foundation.3939Grothendieck, “Brief an G. Faltings”; Schneps, The Grothendieck Theory of Dessins d’Enfants; Tamagawa, “The Grothendieck Conjecture for Affine Curves”; Mochizuki, “The Profinite Grothendieck Conjecture for Closed Hyperbolic Curves over Number Fields”; Heller, Homotopy Theories.

For the Grothendieck–Teichmüller group, the hyphen matters. Grothendieck’s tower vision and Drinfeld’s defined group are connected but not identical acts. A sentence should name both rather than use the compound term as proof of one-person authorship.4040Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

For weak $\infty$-groupoids, Pursuing Stacks is a major programmatic source. Brown, Higgins, Loday, Bénabou, Street, and many others belong to the higher-categorical and higher-groupoid history. Modern definitions require their own attribution.

For derivators, the term and vast late development belong to Grothendieck’s manuscripts, while Heller provides an independent line and later authors establish modern formalisms and applications. “Grothendieck derivator” can identify a lineage but not erase parallel work.4141Grothendieck, “Brief an G. Faltings”; Schneps, The Grothendieck Theory of Dessins d’Enfants; Tamagawa, “The Grothendieck Conjecture for Affine Curves”; Mochizuki, “The Profinite Grothendieck Conjecture for Closed Hyperbolic Curves over Number Fields”; Heller, Homotopy Theories.

This test is deliberately repetitive. General declarations about fairness become useful only when applied to specific nouns.

The result is not dilution. Grothendieck’s documented contribution remains extraordinary across all five cases. Precision prevents the extraordinary from becoming unbelievable and lets other creators remain visible.

The same test should be applied to negative credit. If a published approach failed, identify the theorem and failure; do not attach failure to a whole school. If a conjecture was bypassed, say so without calling the bypass cowardice. If an editor introduced an error, separate it from the speaker’s manuscript.4242Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Credit is symmetrical only at the level of evidence, not praise and blame. The same exactness that assigns a discovery must assign a mistake. Reputation should neither absorb collaborators’ successes nor deflect errors onto them.

This discipline changes narrative rhythm. It replaces a heroic rise and betrayal with many accountable acts. The result may be less dramatic sentence by sentence, but the real drama—how a collective builds durable truth—becomes clearer.

For Grothendieck, that collective drama was inseparable from personal pain. The chapter can acknowledge pain without letting it decide priority. Compassion and verification are not rivals; each prevents a different kind of erasure.

One further category deserves credit: maintenance. Mathematicians who correct references, preserve notation tables, teach standard constructions, and keep old texts usable may introduce no named theorem, yet fields depend on them. A history centered only on conceptual invention repeats the hierarchy Grothendieck criticized.4343Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Maintenance also describes archives. Scanning a fragile leaf, recording its folder, and preserving the original filename are intellectual infrastructure. These acts should be credited in image and source cards, not hidden behind “available online.”

Collective mathematical work therefore includes discovery, proof, redaction, application, criticism, and care. Its ethics is not a perfect distribution of fame. It is a recoverable account of who did what, under what conditions, with what evidence.

Recoverability has a practical test: another researcher should be able to reach the cited page, identify the edition, and see why the verb was chosen. If the path ends in “everyone knows,” the claim is not ready for print.4444Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The standard applies even to praise. “Revolutionized,” “created,” and “founded” need defined objects and dates. Superlatives without scope can erase predecessors as effectively as a missing footnote.

Grothendieck’s life rewards strong language, but strong language becomes credible through narrow claims. The architecture of schemes, topoi, and étale cohomology needs no inflated exclusivity. The late programs need no borrowed proofs.

A seminar volume is not one kind of book

The letters SGA abbreviate Séminaire de Géométrie Algébrique. That title signals an event before it signals a shelf of books. A seminar has dates, speakers, listeners, questions, preliminary notes, and a local sequence. The later Lecture Notes volumes are documentary descendants of those events, not stenographic transcripts.

An exposé can contain several layers. Grothendieck may have supplied the problem and main argument. A speaker may have found a proof, repaired a gap, or chosen examples. A redactor may have rewritten the talk after discussion. An editor years later may have compared versions and added references. The printed exposé bears one or more names, but its mathematical genealogy can be wider.4545Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

This is why “Grothendieck’s SGA” is useful shorthand and dangerous history. It identifies his extraordinary organizing role and the common program. It can also make each theorem and paragraph appear authored by him. Conversely, treating SGA as an anonymous collective obscures the asymmetry: the seminar’s direction, vocabulary, and relentless expansion depended heavily on one person.4646Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

The appropriate unit of credit may differ by claim. For the architecture of étale cohomology, broad programmatic credit is appropriate. For a theorem in one exposé, the exposé’s author and proof history matter. For the final printed text, editors matter. For a corrected re-edition, revisers matter. No one citation sentence can do all this, but notes can.4747Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Post-1970 editing added another layer: absence of the central organizer. Editors could preserve notes, leave them unpublished, or reconstruct them. Every choice had a cost. Preservation risked using his name without consent; nonpublication risked losing a collective investment; reconstruction risked introducing later mathematics. The ethical question was real even if Grothendieck’s account of motives was disputed.4848Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

SGA 4½ illustrates the difference between continuation and supplement. Deligne’s volume made a usable route through étale cohomology and incorporated collaboration visibly. It was not the completion of every project Grothendieck associated with SGA 4. A reader can call it mathematically successful without calling it faithful to the total vision.4949Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

SGA 5 illustrates the problem of delayed authorship. Seminar work from 1965–1966 appeared in 1977. During 11 years, notation and the surrounding field changed. Editors could not reproduce a blackboard without intervention. The published result belongs simultaneously to the seminar date and publication date. Citation should preserve both when chronology matters.5050Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

Grothendieck’s objection to being made an author without consultation, where the record substantiates it, deserves independent weight. Authorship entails responsibility for correctness and consent to presentation. Admiration does not authorize attaching a person’s name. An editor can instead say “based on lectures by,” identify manuscripts, and state interventions.5151Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

At the same time, withdrawal cannot create a veto over the independent work of every seminar participant. Speakers own their contributions; students can publish proofs; archives can preserve documents under lawful and ethical conditions. The boundary must be drawn object by object. A general claim that “the work was his” is too broad for a collective seminar.

Motives and the proof that chose another road

The Weil conjectures gave the dispute a mathematical center. They predicted deep relations among point counts of varieties over finite fields, topology-like invariants, and eigenvalues of Frobenius. Grothendieck’s étale cohomology supplied the cohomological setting, and his standard conjectures on algebraic cycles promised a route to the remaining hard bounds.5252Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Deligne proved the last Weil conjecture in 1974 without first proving the standard conjectures. This was not a rejection of étale cohomology; it used the world Grothendieck had helped build. It was a rejection, or suspension, of one required route. The achievement therefore exemplifies dependence and independence at once.5353Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

Grothendieck could see the detour as a lost opportunity. A proof through the standard conjectures would have established a more general structure and explained why the result belonged to motives. Solving the target without that foundation might leave the deeper task untouched. This is a legitimate mathematical evaluation of scope.5454Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Serre could see the detour as elegance and effectiveness. A conjectural bridge does not become available because it would unify the landscape. Finding another bridge proves the theorem and may disclose structure the original program did not anticipate. This is a legitimate evaluation of proof.5555Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

Credit becomes distorted if either view swallows the other. A claim that Deligne completed Grothendieck’s plan erases the new argument. A claim that Deligne’s proof superseded Grothendieck’s program erases the cohomological infrastructure and the continued significance of standard conjectures. The accurate sentence needs two clauses.5656Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

Motives create a similar problem. Grothendieck proposed a universal cohomological vision; later authors constructed categories under differing equivalence relations, proved conditional results, developed mixed motives in several frameworks, and used motivic language strategically. No single later theory is identical to “the motives Grothendieck invented,” yet the organizing idea and name have a traceable source.5757Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The ethics of inheritance therefore requires scope labels. “Following Grothendieck’s motivic program” acknowledges direction. “Proved by Deligne” assigns theorem credit. “Edited by” assigns textual responsibility. “Using a construction of” identifies technical dependence. Rich syntax prevents a lineage from becoming a dynasty.5858Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality; Deligne, Cohomologie étale (SGA 4½).

Priority when statements are not identical

Priority disputes are hardest when two results resemble one another but differ in hypotheses, methods, or date of circulation. A theorem may be announced before proof, proved locally before globally, submitted after an independent preprint, or published later because of editorial delay. “First” has no meaning until the object of comparison is specified.5959Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

For Mebkhout and the Riemann–Hilbert correspondence, a serious account would place theorem statements side by side and compare the category of $D$-modules, regularity hypotheses, form of equivalence, local or global scope, algebraic or analytic setting, circulation date of a complete proof, and citation history. A biographical paragraph cannot replace that comparison.

Grothendieck’s defense is still important. It records that a mathematician of Algerian origin working in France felt, or was seen as, insufficiently recognized. It identifies institutional power as part of priority. It preserves letters and questions that a history written only from later canonical citations might miss.

But advocacy can introduce its own power. When Grothendieck framed Mebkhout’s experience as the decisive proof of a generalized fraud, readers could become more interested in Grothendieck’s rage than in Mebkhout’s mathematics. Restorative history must reverse that displacement by examining Mebkhout’s papers on their own terms.

The provisional verdict “disputed” is not indecision. It is a status with work attached: collect manuscripts, establish dates, compare statements, consult specialists, and give all parties response. Until then, the prose should say who claimed what. This protects the possibility of a later firmer conclusion.

Naming, eponyms, and invisible collectives

Mathematics often stores history in names: Grothendieck topology, Grothendieck group, Grothendieck–Riemann–Roch, Grothendieck–Teichmüller group. Eponyms are compact and memorable. They are also poor contribution statements.

One name can attach to different objects because a person’s work was broad. Two names can hide a dozen collaborators. A theorem can bear the name of someone who conjectured it rather than proved it. Independent discoverers may be omitted because one tradition dominated textbooks. Eponyms should be treated as vocabulary, not final historiography.

Grothendieck’s own name illustrates the paradox of burial. He wrote repeatedly about erasure, yet few twentieth-century mathematicians have a name attached to more concepts. Ceremonial visibility can coexist with selective invisibility. The public may know the name while mathematicians neglect a particular manuscript or misremember a collaborator.

This is why counting eponyms cannot answer his complaint. Nor can the complaint be proved by showing that a later paper uses a different term. One must ask how the idea is defined, cited, and historically narrated. A renamed concept may be independently developed; an eponym may conceal substantive neglect.

Joint names can acknowledge convergence but produce false symmetry. “Grothendieck–Teichmüller” marks the relation between the tower vision and Drinfeld’s group. It should not imply equal contribution to one jointly written theory. “Grothendieck–Riemann–Roch” locates a reformulation in a longer theorem history. The hyphen is a prompt for explanation, not the explanation.

The economics of redaction

Credit has a material economy. Someone must have time and salary to turn notes into a book. Secretaries, typists, junior researchers, and editors often perform labor that title pages understate. The mathematical community’s ideal of pure ideas can make this infrastructure invisible.

At Bois-Marie, institutional support allowed seminars, duplication, and publication. At Les Aumettes, Grothendieck relied on postal networks and individual help. After 1991, editors devoted thousands of hours to manuscripts without the author’s routine participation. The form in which ideas survive depends on this labor.6060Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

Redaction is sometimes treated as secondary because the “real” creation occurred earlier. Yet a proof unavailable in intelligible form has limited communal life. Redaction can reveal gaps and force definitions. Dieudonné’s role in EGA is the classic case: systematization and writing were themselves mathematically creative.

The opposite mistake is to let a polished redaction appropriate conceptual origin. A clear textbook may become the standard citation while the less accessible source disappears. Ethical bibliography can cite both: the source of the idea and the source actually used for the proof or exposition.

Employment structures matter too. A secure professor can finish a long project; a precarious researcher may publish fragments quickly. A famous author receives patient editorial help; an unknown one may be rejected for rough presentation. Priority rules that ignore these conditions can reproduce inequality while appearing neutral.

Grothendieck’s refusal of prize money later targeted this distribution. His generalizations were severe, but the structural problem remains that recognition often adds resources to those already secure. A community committed to discovery should support unproven potential as well as reward established achievement.

Letters as evidence and as action

The 1985–1986 correspondence changed Récoltes et Semailles rather than serving only as commentary. A reply could be quoted, answered, and incorporated through notes. The boundary between source and event collapses: the document describing the dispute becomes another act in it.6161Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

This requires caution when quoting Serre. His letter is evidence of his view and of the arguments available to Grothendieck. It is not an external peer-review report. Its authority comes from specificity, mathematical knowledge, and contemporaneity, not neutrality.

Grothendieck’s preservation of letters gives historians an advantage and a problem. The archive may retain the messages he considered significant while missing telephone calls or letters others kept. His annotations can date receipt but also frame reading. An archive assembled by a participant is never a transparent totality.

Telephone conversations pose a further limit. Accounts report hours-long calls with Mebkhout and others. Unless notes survive, exact words cannot be reconstructed. A later witness can describe themes and impressions, with the ordinary limits of memory. Quotation marks should be reserved for documented text.

Silence is hardest of all. No reply may mean rejection, overwhelm, lost mail, illness, uncertainty, or a choice not to enter a conflict. Grothendieck often interpreted silence morally. Biography can record that interpretation but cannot infer one motive for every nonresponse.

This chapter’s own debts

The ethics advocated here applies reflexively. Grothendieck authored the testimony and programmatic texts. Named editors and translators made the surrogates consulted. Scharlau collected interviews and correspondence, though this chapter rejects his diagnostic framing. Herreman supplied an analysis of collective mathematics and redaction. Publisher metadata fixes edition dates. This account synthesizes these sources but does not become the discoverer of the facts.

The notes make that chain inspectable.

Historical interpretations require comparison with archives. An ethics of credit applies to the biographer as well: claim no credentials or certainty the account does not possess.

The role map proposed as an image follows the same principle. It does not rank people or assign percentages. It shows verbs—proposed, proved, lectured, redacted, edited, published—so readers can see why one word, “author,” is insufficient. A diagram that made Grothendieck the sun and collaborators orbiting bodies would reproduce the mythology under criticism.

Contested claims by type

The disputes in Récoltes et Semailles can be sorted by the evidence needed. A citation claim asks whether a publication names a source. The publication itself can answer. An authorship claim asks who wrote or approved text; drafts and correspondence are needed. A priority claim asks which precise result existed first; dated manuscripts and theorem comparison are needed. A motive claim asks why someone acted; letters may help but rarely settle it. A community claim asks whether conduct forms a pattern; multiple cases and selection criteria are needed.

Grothendieck often moves from the first three to the last two. A title page becomes evidence of motive; several incidents become a theory of the community. The matrix slows that movement. It permits a finding at one level without granting all levels.

Suppose a later article cites an earlier seminar but not a private letter. Bibliographically, the public seminar citation may be adequate. Historically, the private letter may show a more specific source. Ethically, adequacy depends on whether the author actually used the letter and whether it was available. Similarity alone proves little.

Suppose an editor places Grothendieck’s name on a volume drawn from his lectures. The title page establishes representation. Correspondence establishes consent or objection. Mathematical comparison establishes the editor’s intervention. No claim about theft is needed to say consent was absent.

Suppose two mathematicians prove related Riemann–Hilbert statements. Publication dates alone may mislead if one proof circulated earlier. Preprint dates alone may mislead if a crucial gap remained. Specialist comparison must determine whether the statements are equivalent. Priority is a theorem about documents, not a contest of reputations.

Suppose former students choose other topics. That fact cannot prove ingratitude because professional independence supplies an ordinary explanation. Evidence of explicit disparagement or erasure would be needed for a stronger claim. Noncontinuation is not misconduct.

These cases give one rule for fact-checking: each severe adjective must point to the kind of evidence it requires. “Uncited” can be verified from a text; “appropriated” requires dependence; “conspired” requires coordination. Precision protects criticism from easy dismissal.

The right to be wrong in public

A generous intellectual community needs room for conjecture and error. Grothendieck’s seminars thrived partly because unfinished ideas could be offered. If every speculation permanently attaches liability or priority, people will withhold them.

The same freedom requires correction. A conjecture later proved false should remain attributable as conjecture, not as disgrace. An error in an unpublished manuscript can be marked by editors. Prestige should not make correction impolite.

Grothendieck’s late manuscripts deliberately preserve errors as part of discovery. Lawvere reportedly urged correction before publication of Pursuing Stacks, while Grothendieck preferred students to see that a great mathematician makes mistakes. Both concerns are pedagogical: readability and honesty.

A critical edition can satisfy both by separating text and apparatus. The author’s wording remains; the margin or note identifies an error and later result. Students see process without being abandoned to it. This solution depends on editors receiving explicit credit.

The right to be wrong does not include the right to make unsupported accusations without response. Mathematical speculation concerns objects; allegations concern persons. The evidentiary and ethical threshold is higher when reputation can be harmed.

Récoltes et Semailles blurs the genres by making community judgment part of meditation. Its publication therefore requires more than mathematical errata. Named individuals need contextual notes and, where appropriate, preserved counterstatements.

Generosity, dependence, and power

Teacherly generosity is not power-free. The person who gives a student a problem may influence career, recommendation, and access. A gift can create gratitude and dependence even without stated conditions. Later conflict may reveal expectations both parties understood differently.

Students also exercise agency. They can refuse a problem, alter it, or take methods elsewhere. Their dependence does not make every result the teacher’s. A mature school is measured partly by whether students become authors rather than custodians.

Grothendieck’s school produced mathematicians whose independence is evidence of successful transmission. That same independence could feel like fragmentation of a unified vision. Fertility and loss were two names for the same event depending on viewpoint.

Power changed after 1970. Grothendieck retained enormous symbolic authority but lost daily institutional control. Former students held positions, journals, and active networks. Each side could experience the other as powerful: one through legend and accusation, the other through institutions and publication.6262Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

This reciprocal perception does not create symmetry. Access to a press differs from charismatic prestige; a private mailing differs from a journal article. Analysis should specify the resource at issue rather than call everyone powerful.

The fairest community would make expectations explicit. A teacher would say whether a problem is open for independent publication. A student would cite conversations when they materially shaped work. Editors would request consent. Institutions would document contributions. None of these practices eliminates conflict, but each leaves evidence.

Credit after withdrawal

Late programs create a special problem of credit when an originator no longer participates in standard exchange. Excessive deference can freeze a field. Insufficient attention can erase a source. The answer is neither to wait for permission to think nor to treat circulated pages as ownerless.

Researchers can cite the dated manuscript, state which ideas they adopt, and define their modifications. They can publish proofs under their own names while distinguishing conjecture from theorem. They can invite comment without making response a condition of work.

Editors can preserve facsimiles and produce reading editions with intervention logs. If authorial permission is contested, legal and ethical review should be explicit. The edition should not claim authorial approval it lacks.

Historians can resist the phrase “rediscovered” unless earlier access is demonstrated. An independent construction may resemble an unpublished page. If no path of transmission exists, the correct history has parallel origins.

These principles apply directly to dessins, anabelian geometry, and derivators. Later mathematicians did not need to subordinate their work to Grothendieck. They did need, once the sources were known, to cite accurately. His priority in naming a program and their priority in proving results can coexist.

Community after the center

The IHÉS seminar had once made transmission immediate. A question could be tested in a room. Students could learn a language before publication. Disagreement could alter the next exposé. After 1970, the network continued but its center was gone. Grothendieck experienced later publications from outside the processes that produced them. The delay intensified alienation: what participants saw as development could reach him as a finished object bearing familiar ideas in unfamiliar arrangements.6363Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Deligne, Cohomologie étale (SGA 4½); Illusie, Cohomologie ℓ-adique et fonctions L (SGA 5).

His later chosen distance had consequences. Invitations were often refused. Contact was irregular. He did not participate in ordinary editorial negotiation. This does not excuse using his name without consent. It does mean that collaboration required forms compatible with his withdrawal, and such forms were hard to sustain. A community cannot consult someone who makes consultation impossible; it can still document its attempts and avoid presumptive attribution.

The phrase “mathematical community” can itself conceal plurality. There was no single committee deciding to bury a body of work. Publishers, institute directors, former students, young researchers, seminar editors, and distant readers had different knowledge and power. Some decisions were coordinated; many were not. Grothendieck’s critique is strongest when it reveals shared norms that permit neglect. It is weakest when it turns the plurality into one moral actor.

At the same time, collective diffusion can produce structural responsibility without conspiracy. If everyone assumes someone else will credit an origin, the origin disappears. If prestigious testimony is trusted while a junior researcher’s preprint is ignored, exclusion can occur without secret agreement. If title-page conventions routinely hide redaction, the resulting history is systematically incomplete. The absence of a mastermind does not make the pattern harmless.

Récoltes et Semailles demands an account of what sort of community mathematical practices create: whether generosity is rewarded by erasure, unfinished programs can be criticized without ridicule, errors are corrected without humiliation, seminar speakers receive durable credit, students can differ from teachers without being cast as disloyal, and founders can recognize successors’ independence as evidence of fertility. These issues outlast the particular accusations.

Prize, prestige, and the economy of recognition

Recognition is not only a name in a footnote. It is employment, invitation, editorial attention, and prize money. Grothendieck’s 1988 refusal of the Crafoord Prize would connect the credit dispute to a wider economy. He argued that prestigious researchers already possessed material security and status while others lacked both. He also claimed that prizes could not measure the fertility of ideas. These views belong to Chapter 60, but their roots are visible here.6464Scharlau, Spirituality; Alexandre Grothendieck, “Lettre ouverte refusant le prix Crafoord,” July 9, 1988, https://webusers.imj-prg.fr/~leila.schneps/grothendieckcircle/Letters/crafoord.pdf.

Prizes individualize work that was collective. A medal names one or two people even when an achievement depended on a school. This does not make every prize dishonest; institutions can state a precise citation and recognize exceptional contribution. The danger is retrospective simplification. A prize can make one person the author of a field and everyone else supporting cast.

Grothendieck had accepted the Fields Medal in 1966, though he did not travel to Moscow. By 1988 his relation to honors had changed. The contrast should not be explained by invented inner conversion. His own letter supplies reasons: material sufficiency, inequality, time rather than honors as the test of fertility, and his condemnation of scientific ethics. Whether those reasons persuade is left to the reader.6565Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Toward a usable ethics

An ethics of credit adequate to this history would begin with provenance. Preserve drafts, dates, letters, seminar programs, and editorial states. Do not rely on prestige to settle priority. Name uncertainty. Distinguish recollection from contemporaneous record. Keep rejected or superseded formulations when they explain development. Archive is not an administrative afterthought; it is the infrastructure of fairness.

Second, it would distribute roles. A publication should state who originated results, who proved them, who wrote each section, who edited, and what changed from prior versions. Modern contribution statements attempt this in scientific fields, though they too can become formulaic. Mathematics can adapt them without pretending that creative influence is fully quantifiable.

Third, it would obtain consent for the use of a living person’s name on a new work. If consent cannot be obtained, editors should describe the historical materials and their interventions rather than manufacture coauthorship. Preservation and respect need not conflict if the edition is candid.

Fourth, it would protect independence. Students owe accurate acknowledgment, not lifelong obedience. Teachers should not convert generosity into territorial claim. Later authors should neither erase programs nor surrender the right to alter them. A fertile idea proves its life partly by producing work its originator did not foresee or approve.

Fifth, it would distinguish error from bad faith. A missing citation can be corrected. A priority dispute can be investigated. Not every mistake proves moral collapse. Repeated or strategic erasure may establish a pattern, but the case must be made with documents. Moral language should follow evidence rather than replace it.

Finally, it would preserve criticism itself. Communities often defend celebrated figures by treating uncomfortable testimony as embarrassment. Récoltes et Semailles should not be neutralized in that way. Its accusations require scrutiny, but its questions about power, prestige, and transmission are legitimate. The fair response to an overstatement is not silence; it is a more exact account.

Editing a disagreement fairly

A note attached to a contested passage should do four things. It should identify Grothendieck’s claim and date. It should cite the object at issue. It should summarize a documented counterposition. It should state what remains unresolved. The note should not announce a verdict the sources cannot sustain.

Where Serre’s letter directly answers a claim, a short quotation or precise paraphrase can give him voice. Where Deligne declined public engagement, the biography should not invent his defense. Silence is not consent, confession, or contempt.

Living or recently living third parties require sensitivity even when papers are open. Allegations involving private conduct should meet relevance and corroboration tests. Mathematical priority can usually be discussed without publishing intimate correspondence unrelated to the theorem.

Technical disputes need specialists. A literary editor cannot determine whether two $D$-module theorems are equivalent by comparing titles. A mathematical reviewer must examine statements and proofs. Historical confidence should follow that review.

The notes should distinguish edition intervention. If an editor supplied a theorem label, corrected notation, or rearranged sections, readers need to know. Otherwise later clarity can be mistaken for authorial form.

Indexes can also distribute credit. A general index that lists only Grothendieck under every concept repeats erasure. Names of exposé authors, editors, correspondents, and later theorem provers should lead to substantive pages, not ceremonial mentions.

The image program must avoid visual verdicts. Placing one person in shadow and another under light would turn design into accusation. Title pages, correspondence diagrams, and contribution maps provide documentary alternatives.

A response appendix could collect concise positions where disputes are too complex for narrative. But it should not reproduce hundreds of pages or imply equal evidence through equal space. Editorial fairness is proportional and sourced, not mechanically balanced.

Finally, the edition should preserve the date of its own judgments. New archives may change priority findings. A note can say “unresolved in the present edition” and invite correction. This openness is not weakness; it is the scholarly form of renewal.

The ethical balance can be expressed in one pair of duties: let ideas travel, and let their histories travel with them. If ideas cannot travel, mathematics becomes private property. If histories do not travel, collective creation becomes anonymous extraction. Neither duty is easy. Attribution can grow cumbersome; historical influence can be diffuse; simultaneous discovery is real. Difficulty is not an excuse for mythology.

Grothendieck’s French line about mathematics as collective work does not resolve his grievance. It makes the grievance sharper. No one owns the landscape, but paths have makers. Maps have authors. Bridges have builders. Travelers inherit roads and open new ones. A truthful account can say all of this without awarding the territory to its most famous explorer.

The chapter’s documentary verdict is therefore deliberately limited. Grothendieck’s foundational and programmatic contributions were immense and sometimes inadequately captured by later shorthand. Post-1970 publications involved real issues of redaction, consent, and historical narration. His allegations against Deligne and others were contested and cannot be repeated as established theft. Serre supplied a substantial counterargument. Mebkhout’s priority history deserves specialist study independent of Grothendieck’s advocacy. The broader ethical questions remain open and productive.

In 1983, while these conflicts were beginning to take written form, another network demonstrated a different mode of circulation. Ronald Brown, working in Bangor, exchanged letters with Grothendieck about higher groupoids and homotopy. With permission, Brown duplicated pages and sent them to interested mathematicians. The manuscript was unfinished, personal, sprawling, and technically difficult, yet it traveled. Its title was Pursuing Stacks. Its history shows both the power of informal transmission and the decades of editorial labor required to turn circulated thought into a formal publication.6666Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.