收获与播种

There was no single morning on which Récoltes et Semailles became a finished book. There were beginnings, endings that reopened, typed pages, fair copies, added notes, mailed fascicles, replies, corrections, and new prefatory texts composed after much of the work already existed. A reader could receive one configuration while another received more. Later digital files would omit material that belonged to the expanding whole. Decades later, publishers would issue portions and editions in languages and formats unavailable during composition. The first fact about the work is therefore not its size, though it is enormous. It is its changing state.

Grothendieck began writing in June 1983. According to the chronology reconstructed from the manuscript and correspondence, he initially expected a relatively short testimony. The first movement, Fatuité et Renouvellement—Vanity and Renewal—grew to roughly 170 numbered pages, with footnotes added in 1984. By March 1984 he could describe it as a meditative review of his past life as a mathematician and imagine it as the introductory volume to a projected series. He was already having it typed and reproduced for friends and former students. Yet what seemed an ending became the threshold to three large movements called L’Enterrement, “The Burial.” 11Scharlau, Spirituality, ch. 23, pp. 1–2.

The dates show the expansion. L’Enterrement (I), ou la robe de l’Empereur de Chine was drafted mainly in spring 1984 and received notes and supplements during the first half of 1985. L’Enterrement (II), ou La Clef du Yin et du Yang runs from September 22, 1984, to January 14, 1985. The main text of L’Enterrement (III), ou Les Quatre Opérations was written between February 16 and April 7, 1985, with some notes later. A prefatory letter is dated May 1985. In January 1986 came Promenade à travers une œuvre ou L’enfant et la Mère. In March and April 1986 he added Les Portes sur l’Univers. Corrections and complements responded to an extended correspondence, including exchanges with Pierre Schapira and Christian Houzel. 22Alexandre Grothendieck, “Complements and corrections to Récoltes et Semailles,” 1986, https://webusers.imj-prg.fr/~leila.schneps/grothendieckcircle/G-RS-Compl.pdf; Scharlau, Spirituality, ch. 23, p. 2.

To list these dates is not to imply that each block was sealed before the next began. Grothendieck’s method allowed a later realization to enter as a new note rather than a silent revision of an earlier paragraph. Cross-references multiplied. The text preserved the path of reflection, including returns and contradictions, because the path was part of the testimony. A critical edition may help a reader navigate that history, but an editor who regularizes it into one seamless argument risks destroying evidence about how the work was made.33Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Alexandre Grothendieck, Récoltes Et Semailles, Partial Japanese Translation, trans. Yūichi Tsuji, 3 vols. (Gendai Sūgakusha, 1989–1993); Scharlau, Spirituality.

The title joins two agricultural actions. Reaping gathers what has grown; sowing entrusts something to a future one cannot control. In the manuscript, the harvest includes the mathematical work of earlier decades, the relationships and institutions that enabled it, and the consequences—desired or not—that followed. The sowing includes memories, concepts, accusations, warnings, examples of creative work, and questions offered to readers. Harvest and seed are not cleanly separated. What is reaped scatters seed. What appears to be a weed in one judgment may prove fertile in another. The title resists a ledger that counts and assigns past achievements.44Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

This agricultural language differs from the architecture that dominated Grothendieck’s published mathematics. EGA and SGA evoke elements, seminars, foundations, structures, and operations. Récoltes et Semailles invokes season, growth, ripeness, cutting, ground, and uncertain germination. The change does not remove mathematics from the text. It changes the relation between mathematics and time. A theorem can be stated as if detached from its discovery; a harvest cannot be detached from weather, labor, inheritance, and the interval between planting and gathering.55Grothendieck, Récoltes et Semailles (Université des Sciences et Techniques du Languedoc and Centre National de la Recherche Scientifique, 1986); Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, “Complements and corrections to Récoltes et Semailles”; Scharlau, Spirituality.

The subtitle presents the work as reflections and testimony on a past as a mathematician. Testimony is not identical to autobiography. A conventional autobiography promises a chronological life. Récoltes et Semailles moves by themes, memories, documents, disputes, and recursive commentary. Grothendieck says that it is a reflection on himself and his life, but also a witness to the immediate present in which the pages are being written. Past event and current act of remembering meet on the page. 66Grothendieck, Récoltes et Semailles, ed. Carmona (1985).

That double temporality is one reason the text can be treacherous as a biographical source. It gives first-person access to judgments, language, and remembered experience. It does not grant automatic accuracy to every recollection or accusation. When Grothendieck describes what he thought in 1960, the document proves that in the 1980s he described the earlier thought in that way. Other records are needed to establish dates, actions, and other people’s perspectives. The manuscript is indispensable precisely when read as testimony rather than converted into an omniscient chronicle.77Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Its voice repeatedly uses second-person address. The intended reader is not always a named colleague and not a demographic category. Grothendieck imagines one person reading alone. He says he is not addressing a public but the person who can be alone, the child in the reader, the being not completely absorbed by rank and collective habit. This second person gives the work intimacy, but it also gives its judgments pressure. The reader is not allowed the neutral position of observing a quarrel at a distance; the text asks what vanity, fear, creativity, and silence look like in an individual life.88Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The address can be hospitable and coercive in adjacent pages. It may invite someone with little mathematics to follow an account of discovery. It may imply that resistance demonstrates the very fear or inertia under examination. That structure makes disagreement difficult: a critic risks being placed inside the diagnosis offered by the text. A responsible biography need not choose between reverence and dismissal. It can recognize the generosity of the invitation and the argumentative trap.

The material passage from desk to reader

The physical production of the work mattered. Grothendieck typed drafts on ordinary paper, using the available space densely and adding handwritten material. A secretary at the Montpellier mathematics institute—identified by initials in accounts that preserve her privacy—prepared fair copies. Completed portions and originals traveled through the post. Grothendieck proofread the typescripts and returned corrections. When enough material was ready, copies were made, and he supplied lists specifying who should receive which parts. 99Scharlau, Spirituality, ch. 23, p. 3.

This division of labor deserves notice because Récoltes et Semailles is intensely concerned with creation and transmission. The author’s solitary voice reached readers through another person’s sustained technical work and through university reproduction facilities. Typing was not authorship of the claims, but neither was it negligible. A 1,000-page meditation does not become legible, duplicated, consulted, and mailed by metaphor. The material infrastructure of testimony included paper quality, typewriter habits, copy preparation, postage, and address lists.

Grothendieck reported in a later preliminary fascicle that a provisional printing had been sent in October 1985. Secondary accounts estimate the circle at roughly 200 friends, acquaintances, and colleagues. “Roughly” is necessary, and so is “some configuration.” The work was changing; mailing lists differentiated recipients; later complements did not retroactively enter every packet. To say that 200 people “received the book” suggests a uniform bound volume that did not yet exist. It is more accurate to say that about 200 recipients were sent typescript or reproduced fascicles of the developing work.

Circulation produced new writing. Replies from mathematicians became occasions for clarification, defense, escalation, or correction. Grothendieck copied, dated, and archived correspondence with care. Notes could answer an objection without erasing the text that prompted it. The resulting object resembles a conversation preserved inside one speaker’s continuously expanding frame. Other voices appear, sometimes through quotation, but the architecture and final placement remain his.1010Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality.

The 1986 complements associated with Schapira and Houzel show why “completed in 1985” is inaccurate. The main movements may have been substantially present, and provisional copies may have gone out, but the manuscript kept receiving additions. Les Portes sur l’Univers, written March 17–April 11, 1986, belongs to that later phase. The work’s date should therefore be expressed as a span—1983–1986—with local dates for particular parts.

Digital circulation later created another deceptive unity. A searchable PDF can look like the work itself, but the widely hosted volunteer transcription explicitly omits Les Portes sur l’Univers and some 1986 complements. It is extraordinarily useful for locating phrases and following much of the text. It is not a critical edition and should not be used to prove completeness. The scan of the original first-part typescript offers different evidence: page layout, corrections, material sequence. One surrogate supports text search; the other supports object history. Neither should be silently treated as the 2022 Gallimard edition. 1111Grothendieck, Récoltes et Semailles (Université des Sciences et Techniques du Languedoc and Centre National de la Recherche Scientifique, 1986); Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, “Complements and corrections to Récoltes et Semailles”.

A promenade through the mathematical work

The Promenade à travers une œuvre, written in January 1986, is the most accessible map into Grothendieck’s mathematical memory. Its alternate title, “The Child and the Mother,” establishes a vocabulary of discovery. The child represents a direct, patient curiosity that asks elementary questions without shame. The mother represents an embracing vision or fertile surrounding reality, not a biological claim about mathematics. The metaphors are his, and they should be interpreted through the text rather than turned into a theory of his family.1212Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The promenade begins with school memories. He recalls liking school and being taught reading, arithmetic, song, prehistory, and fire by the same teacher. What mattered in later mathematical development, as he tells it, was the freedom to examine what others treated as settled. At Montpellier he had been dissatisfied by definitions of length, area, and volume and attempted to reconstruct measure for himself. The episode had already become part of his personal account of creative innocence: rather than first mastering the literature, he pursued the question in isolation.

This recollection can easily feed a myth of the self-created genius. The rest of the mathematical life complicates it. Nancy supplied problems, mentors, seminars, and a community capable of recognizing what he did. The IHÉS years depended on collaborators, students, editors, and seminar participants. Récoltes et Semailles knows that mathematics is collective even when its rhetoric celebrates solitude. Its strongest tension is not between a lone creator and a useless crowd. It is between solitude as a condition of discovery and community as a condition of testing, transmitting, and continuing discoveries.

The promenade names major “master themes” of his earlier work. Among them are topological tensor products and nuclear spaces; continuous and discrete duality; Riemann–Roch and $K$-theory; schemes; topoi; étale and $\ell$-adic cohomology; motives; anabelian geometry; Galois–Teichmüller ideas; tame topology; and the yoga of stacks. The exact list varies with how one counts intertwined themes. Its purpose is not a curriculum vitae. It displays a movement from solving given problems to changing the language in which whole families of problems are posed.1313Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Consider schemes. A terse heroic narrative says that Grothendieck invented schemes and transformed algebraic geometry. The promenade’s logic is more instructive. The concept allowed geometry over arbitrary commutative rings, nilpotent structure, arithmetic fibers, and base change to inhabit one language. It did not win by being maximally abstract; it made operations that mathematicians already needed systematic. In the agricultural metaphor, a language is fertile if it produces unforeseen work in other hands.1414Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Topoi offer another example. They generalize spaces as settings in which sheaves live, allowing geometric and logical structures to be carried by a category. The word can intimidate a general reader because its definition is abstract. Grothendieck’s retrospective emphasis is relational: a topos is a world organized by how local data cover, restrict, and glue. The same instinct later appears in stacks and in attempts to recover geometry from categories or groups. The promenade is therefore not a museum tour of disconnected inventions. It identifies recurring motions of thought.1515Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Motives occupy a special place as an unfinished unification. Different cohomology theories—Betti, de Rham, $\ell$-adic, crystalline—seem to realize aspects of a more universal object. Grothendieck proposed motives as that underlying source. In Récoltes et Semailles, motives become both mathematical horizon and evidence in the dispute about continuation. Later mathematicians used motivic ideas in ways he sometimes admired and sometimes regarded as evasions. The biographer must keep the mathematical conjecture distinct from the ethical conflict.1616Grothendieck, Récoltes et Semailles (Université des Sciences et Techniques du Languedoc and Centre National de la Recherche Scientifique, 1986); Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, “Complements and corrections to Récoltes et Semailles”; Scharlau, Spirituality.

The promenade also names late programs. Dessins and the Teichmüller tower are not appended as an eccentric coda; they are presented as renewed access to themes of Galois theory and moduli. Pursuing Stacks belongs to the continuing concern with homotopy and descent. The retrospective map thus refuses the familiar periodization in which mathematics ends in 1970. It does not prove that all late projects succeeded. It shows how their author located them inside one work.1717Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Discovery, writing, and redaction

One of the manuscript’s most valuable distinctions is between discovering and presenting. A discovery may arrive as a new concept, a relation, a proof strategy, or a field of questions. Writing can extend discovery because the demand for exact expression reveals gaps. Redaction, in the French mathematical tradition, can mean the sustained preparation of seminar material or notes into a form others can use. Teaching can supply examples and questions. Editing can stabilize references, repair errors, and bring a text to publication. None of these roles is identical, and several people can occupy them in one work.1818Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

Grothendieck had unusual experience of all of them. Dieudonné transformed rapidly developing foundations into EGA’s formidable written architecture while contributing mathematical judgment of his own. SGA volumes grew from spoken seminars, written exposés, circulated notes, revisions, and later editorial decisions. A theorem might be proposed by one person, proved by another, generalized in discussion, and written by a third. Standard title-page authorship cannot encode every contribution. Yet abandoning standard attribution altogether leaves memory to prestige and anecdote.1919Grothendieck, Récoltes et Semailles (Université des Sciences et Techniques du Languedoc and Centre National de la Recherche Scientifique, 1986); Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, “Complements and corrections to Récoltes et Semailles”; Scharlau, Spirituality.

In Récoltes et Semailles, this complexity becomes personal. Grothendieck believed that ideas he had offered generously were later detached from their origin, narrowed, renamed, or used without adequate continuation of the vision that gave them meaning. He also examined his own vanity and failures. The harvest metaphor lets these coexist: he can claim a field as the fruit of his labor while asking whether the very wish to possess it corrupts creation.2020Grothendieck, Récoltes et Semailles (Université des Sciences et Techniques du Languedoc and Centre National de la Recherche Scientifique, 1986); Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, “Complements and corrections to Récoltes et Semailles”; Scharlau, Spirituality.

The reader encounters a long oscillation between gift and ownership. Mathematical ideas are given so that they can live in other minds. Once given, they change. If the giver demands control over every use, the gift becomes a license. If recipients erase the conditions and people from which it came, circulation becomes dispossession. Récoltes et Semailles never turns that tension into a stable rule. Its force lies in making the tension impossible to ignore.

Length is not a defect in every instance, though it can become one. Repetition enacts the inability to reach a final account. A name, event, or publication is revisited from another angle because the previous account failed to settle it. The process can reveal self-correction. It can also amplify grievance. Readers who praise every recurrence as depth disregard the burden placed on attention; readers who call the whole work unreadable disregard the discoveries made through return.

Pierre Samuel’s response, as reported in correspondence, was blunt: the text was far too long and could say what mattered in a fraction of its extent. René Thom, by contrast, recognized an author in the French prose. These judgments are not mutually exclusive. A writer can possess an unmistakable voice and need severe editing. The text’s resistance to ordinary editing arises because cutting repeated passages might remove the evidence of meditation that Grothendieck considered essential. The problem is literary and archival as well as stylistic.2121Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality.

The child, the boss, and the work of attention

The manuscript develops figures rather than psychological case labels. The child is curiosity, openness, and the capacity to see what convention has made invisible. The “boss” or patron is the internalized authority that seeks recognition, controls appearances, and fears exposure. These are textual personae used to examine creativity. They should not be converted into clinical entities or treated as transparent maps of the author’s mind.2222Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The child’s solitude is not social abandonment. It is the interval in which a question can be heard before the group supplies the approved answer. Grothendieck links this solitude to mathematical work: textbook presentation does not settle a definition, fashionable problems do not dictate attention, and an apparently naive object may become central. Yet the child needs language, examples, and interlocutors. Pure isolation cannot explain how a private intuition becomes a theorem that others can check.2323Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The “boss” introduces social recognition into the same act. A mathematician may choose a problem for prestige, protect territory, cite selectively, or confuse a school’s habits with the structure of mathematics. Grothendieck applies the critique to himself as well as others, though not evenly. Vanity and renewal form the first major movement because renewal requires seeing how identity has attached itself to accomplishment. This ethical language can illuminate professional conduct. It becomes unreliable when used to infer motives without evidence.2424Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

His style often turns a local experience into a general claim about the mathematical world. A missing response becomes evidence of indifference; an editorial choice becomes a sign of burial; a professional convention becomes a symptom of collective decline. Sometimes the generalization identifies a real structural problem. Sometimes it outruns the available facts. The chapter that follows will examine those claims in detail. Here the important point is formal: Récoltes et Semailles moves continually from incident to moral world.

The movement explains its unusual mixture of technical and nontechnical pages. A discussion of motives or cohomology can be followed by a meditation on fear, then by a letter, then by a note written months later. The boundaries among mathematical content, sociology, and spiritual reflection are porous because the work asks how the quality of attention shapes all three. For a reader seeking a clean intellectual autobiography, this is frustrating. For a historian of mathematical practice, it is unusually rich.2525Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The text also challenges the easy opposition between reason and feeling. Grothendieck describes mathematical beauty, trust, disappointment, indignation, and shame as forces in the production and reception of mathematics. A proof remains subject to formal verification; emotion does not make it true. But what a community notices, who receives time to develop an idea, and whether a young researcher is heard are not determined by proof alone. The social life of rigor includes affect, hierarchy, and memory.

Mailing a meditation

When copies began arriving, recipients faced claims and an implied relationship. Many had known Grothendieck as teacher, collaborator, or dominant figure. Some were named inside an account of decline. To answer meant entering a correspondence that could be quoted or answered in later notes. Not to answer could be interpreted as confirmation of silence. This asymmetry helps explain why Grothendieck’s claim of receiving little response cannot be assessed by counting letters alone.

There were responses. Serre wrote in July 1985. Other mathematicians commented on style, allegations, and publication. Grothendieck himself archived letters and expanded his text. His repeated statement that he had heard almost nothing expresses his experience of the quality or adequacy of response, not literal proof that envelopes did not arrive. The difference is crucial. A community can respond and still fail, in an author’s judgment, to answer the question he believes he asked.2626Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Scharlau, Spirituality.

The mailed text altered relationships. It revived correspondence after long silence and deepened breaks. It made private assessments available to a bounded but substantial circle. Because the copies were not a public commercial edition, recipients inhabited an ambiguous space: the work was addressed to them and already circulating, yet it was not a published book open to impersonal review. Later online access would widen the audience without erasing that first mode of address.

The word diffusion covers several states that English often compresses as publication. A typescript can be copied. A university can reproduce it. An author can mail it. A translator can prepare a version. A publisher can accept or decline it. A press can issue a volume. Libraries can catalog it. A website can host a scan. These acts produce different public objects and different evidence. Part X preserves them because the history of Grothendieck’s late work has often been distorted by treating “available” as “published.”

In April 1987, Grothendieck made a day trip to Paris in connection with a possible French publication. He mentioned bringing the final typescript to a publisher and hoped it might appear that year. Accounts identify contacts with Éditions Kimé and perhaps later Odile Jacob, but no French book resulted. The journey demonstrates intention, not publication. It also shows that the absence of an edition was not always a simple refusal to publish; negotiations occurred and failed. 2727Scharlau, Spirituality, ch. 23, p. 7.

Partial formal publication did occur before the twenty-first century. Yuichi Tsuji translated portions into Japanese, and three volumes appeared with Gendai-Sugaku-sha in 1989, 1990, and 1993. A 1998 fourth volume is also cataloged, with the publisher shown as Tsuji himself, indicating a limited/private rather than a commercial-series issue. These records prevent the common overstatement that Récoltes et Semailles “was never published until 2022.” The accurate statement is narrower: the developing French typescript circulated privately and institutionally in the 1980s; partial Japanese volumes were formally issued beginning in 1989; a major French edition appeared in 2022. 2828Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; National Institute of Informatics, “CiNii Books: Records of the Japanese Translation of *Récoltes Et Semailles*,” https://ci.nii.ac.jp/books/.

The Japanese history requires its own bibliographic audit, including exact titles, contents, and print runs. It should not be treated as a curiosity in a footnote. Translation changes the work’s possible audience and may select or reorder material. A partial edition can become the first stable encounter for readers in one language while the source remains unstable in another. This is not a lesser form of reception; it is a different one.

Gallimard issued two French volumes in January 2022: publisher metadata gives January 13, 2022, for Volume II and January 20, 2022, for Volume I. That issue-record sequence does not reverse the manuscript’s composition or reading order. The edition transformed the work’s public status by creating books with ISBNs and stable catalog records; it did not transform the 1983–1986 drafting and 1985–1986 provisional mailings retroactively into a 2022 text. 2929Alexandre Grothendieck, Récoltes et semailles, I: Réflexions et témoignage sur un passé de mathématicien (Gallimard, 2022), https://openlibrary.org/books/OL39209113M; Alexandre Grothendieck, Récoltes et semailles, II: Réflexions et témoignage sur un passé de mathématicien (Gallimard, 2022), https://openlibrary.org/books/OL39209114M.

The 2022 date also raises editorial questions about the governing typescript state, placement of supplements, supplied corrections, and treatment of names and repeated passages. Those issues belong to edition history and require page-by-page consultation before quotation. For this account, the online copy of the 1980s typescript supplies short phrases, and the Gallimard records establish publication facts. No long passage is reproduced.3030Grothendieck, Récoltes et semailles I; Grothendieck, Récoltes et semailles II.

Reaping a mathematical past

The work’s account of the earlier decades is not a list of discoveries followed by praise. It asks what made discovery possible. Grothendieck remembers long periods of concentrated labor, the willingness to begin from naive questions, the role of images and names, and the moment when many facts align under a concept. He contrasts force with patient attention. A difficult problem can be attacked directly; a larger theory may dissolve it by changing the surrounding language. The famous image of a rising sea belongs to this general method: rather than break a nut with a hammer, one raises the water until the obstacle is submerged. The metaphor must not replace the actual proofs, which still demand exact work.

Naming is part of that work. A good term can gather phenomena and make a new question possible rather than label an object already understood. Scheme, topos, site, motive, and crystalline cohomology each carry histories more collective than a single baptism. Récoltes et Semailles pays intense attention to names because names mediate memory. If a concept circulates under a name detached from its origin, history changes; if a founder’s name is attached to everything, collaborators disappear. There is no automatic ethical solution.

The manuscript’s history of EGA and SGA makes redaction visible as creative labor. Seminar mathematics is born in speech, blackboard argument, notes, and questions. A polished volume can conceal that process. Grothendieck sometimes writes as though his vision generated an entire world and others developed details. Elsewhere he recognizes an adventure that no one owns. The contradiction is not incidental. The central drama of reaping lies in acknowledging the field one planted without claiming every plant that grew in it.

His own departure in 1970 creates the unavoidable counter-question of responsibility for continuation. Former students had careers, questions, and methods of their own. Some used Grothendieck’s tools in directions he did not value. Some edited seminar material because leaving it unpublished would waste collective work. He could see such acts as appropriation or diminution. They could see them as mathematical life. The next chapter holds these accounts beside the bibliographic record.3131Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

In this chapter, the emphasis remains on the form that made the conflict public. Récoltes et Semailles did not deliver a short petition for corrected citations. It made attribution a symptom of spiritual and communal health. A name omitted or included revealed, for Grothendieck, how people regarded creation, gift, and power. That scale gives the text grandeur and danger. It allows an editorial dispute to expose hidden hierarchy; it also allows disagreement over a title page to become evidence of moral catastrophe.

Four movements, several entrances

The work’s organization gives readers more than one entrance. The numbered four-part body begins with Fatuité et Renouvellement and proceeds through three phases of L’Enterrement. The later “Prelude in Four Movements” stands in front of material written earlier. A prefatory letter, detailed contents, prologue, promenade, epilogue, postscripts, and appended “doors” surround the main blocks. Chronology and reading order do not coincide.

This architecture resembles a building whose entrance hall was added after many rooms were occupied. The hall helps a visitor orient herself, but it reflects the architect’s later understanding of the structure. A statement in the 1986 Promenade may summarize a 1983 section more lucidly while also changing its emphasis. Both versions belong to the evidence.3232Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; Scharlau, Spirituality.

The first movement’s title is itself double. Fatuité can mean vanity, conceit, or fatuity; renouvellement is renewal. The pair refuses a simple confession followed by cure. Vanity recurs within renewal, and the ability to see vanity is one sign of renewal. Translation must preserve the ethical charge without pretending one English word covers every occurrence.

The repeated title L’Enterrement gives Parts II–IV continuity, but each subtitle shifts the lens. The emperor’s robe invokes collective pretense; yin and yang suggest complementary forces; the four operations turn elementary arithmetic into a schema for relations and acts. The metaphors are not technical definitions. They organize attention and can change meaning across hundreds of pages.

The detailed table of contents functions almost as a parallel book. Section titles create a compressed itinerary through names, events, and recurring images. In a work difficult to read linearly, the table is an intellectual map. It is also authorial interpretation: choosing a title after a section is written can make one strand appear dominant.

Later editions face a choice between preserving the sequence in which readers of 1985 encountered fascicles and presenting the author’s eventual preferred order. A historical edition might need both: a reading text and a composition map. Without the latter, “Part I” can falsely imply “written first and distributed with everything that now precedes it.”3333Grothendieck, Récoltes et Semailles (Université des Sciences et Techniques du Languedoc and Centre National de la Recherche Scientifique, 1986); Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, “Complements and corrections to Récoltes et Semailles”; Scharlau, Spirituality.

A close reading of the promenade’s method

The Promenade earns its name by moving rather than cataloging. It begins from a child’s dissatisfaction and follows recurrent gestures across distinct fields. A catalog says what Grothendieck produced. A promenade asks how one landscape opened into another.

In functional analysis, he had encountered spaces through continuous linear maps, tensor products, and duality. The work on nuclear spaces reorganized when infinite-dimensional operations behave as if finite-dimensional regularity persisted. The later geometry is mathematically different, but the habit of seeking the class in which natural operations become well behaved is recognizable.

Riemann–Roch provides another transition. The classical theorem relates analytic and topological invariants of a curve. Grothendieck’s reformulation uses $K$-theory and functorial operations, turning a formula into a structural statement applicable in families. The promenade treats this not as one isolated victory but as evidence that the right language changes what counts as a problem.

Schemes extend geometry over arbitrary base rings. The relative point of view then asks how an object varies over a base rather than what it is over one fixed field. Representable functors turn classification into a criterion: a geometric object is characterized by the families of maps into it. These moves prepared moduli and descent.

A site replaces ordinary open subsets with a category of objects equipped with a notion of covering. A topos is the resulting sheaf world. Étale cohomology uses this flexibility to construct invariants over finite fields and arithmetic bases. The path from schemes to sites to topoi is not ascent into abstraction for its own sake. Each stage repairs a specific failure of older topology in algebraic geometry.

The six operations organize direct and inverse images, extraordinary variants, tensor products, and internal Homs. Duality, trace formulas, and cohomology become parts of one formalism. Again the ambition is relational. A list of cohomology groups is less important than the operations connecting them.

Motives then appear as a hoped-for source beneath several cohomology theories. The promenade’s retrospective power comes from placing realized formalisms beside unrealized visions. It does not disguise the difference. Schemes and étale cohomology had extensive published foundations; the general category of mixed motives remained conjectural. An accurate map contains both roads and projected bridges.

The late themes repeat the movement at smaller dimension and greater combinatorial visibility. Curves and dessins become an entry to absolute Galois actions. Weak higher groupoids become an entry to homotopy types. Tame forms become an entry to topology. The promenade argues for continuity of method without claiming equal completion.

This method can be summarized as enlargement followed by recognition. Enlarge the category until obstructed operations become possible; then recognize the classical objects inside it as special cases. Schemes include varieties, topoi include ordinary spaces’ sheaf categories, stacks include spaces with automorphisms, derived settings include ordinary exact phenomena. The danger is endless enlargement. The success condition is new theorem-level control.

For a general reader, the promenade’s value is not that it teaches all these theories. It shows why one person regarded them as one work. That unity is historical evidence even where modern mathematicians draw boundaries differently. A biography should present his map and then mark where the field’s current map diverges.

Time written into the page

Grothendieck’s refusal to rewrite earlier sections silently preserves changes of mind. A note may say that an earlier judgment was incomplete. A later complement may intensify rather than soften it. Page dates allow readers to reconstruct movement. This is why the text sometimes resembles a laboratory notebook, a legal dossier, and a letter at once.

The laboratory analogy illuminates iteration but fails at public verification. A scientific notebook records observations that can be checked against experiments. Here many observations concern relationships and motives, available only through testimony. The legal-dossier analogy illuminates documents but fails because the author is also advocate, witness, and arranger. No neutral court assembled the file.

Footnotes become temporal hinges. In ordinary exposition, a note supplies a reference or qualification subordinate to the main text. In Récoltes et Semailles, a note can grow for pages, acquire its own date, and redirect the argument. The hierarchy of main text and note ceases to measure importance.

Cross-references create a web. A reader may be sent backward to an earlier memory and forward to a later note. The web enacts Grothendieck’s conviction that a life cannot be told once in linear sequence. It also produces practical exhaustion. Digital links can help navigation, but they risk making a recursive structure seem effortless when the original reader handled piles of paper.

Repetition sometimes preserves exact language, sometimes revises it slightly. A textual scholar can use those differences to see whether a thought hardened, broadened, or changed object. An editor who deletes “duplicates” may erase that evidence. Conversely, a trade edition that retains all repetition without guidance may lose readers before they reach the variation. There is no purely mechanical solution.

The work’s “present” also stretches. A section written in 1983 can be present-tense meditation; a 1986 note can comment on it; a 2022 edition can place both on adjacent pages. Every edition therefore contains at least three presents: composition, supplementation, and reading. Dates printed beside sections are not clutter. They keep those presents apart.3434Grothendieck, Récoltes et semailles I; Grothendieck, Récoltes et semailles II.

The ethics of the singular reader

Addressing one reader changes responsibility. Second-person address can make an abstract idea concrete by turning attention toward work, fear, and the capacity for solitude. It can also assign a role the actual reader did not choose. The text sometimes interprets a reader’s resistance before that reader has spoken.

This rhetorical structure resembles a letter, but the recipient is open. A real letter presupposes shared history and permits reply. Printed second-person address can reach strangers who cannot enter the manuscript except through later commentary. The private tone becomes public technique.

Grothendieck’s hope was that singular address would bypass the anonymous collective. The mathematical establishment could dismiss a book; one attentive person might receive it. The agricultural metaphor fits: sowing is not broadcasting an argument to a voting body but entrusting seed to particular ground.

Yet a community is made from singular readers. The text’s opposition between person and public can become too sharp. Libraries, seminars, translation groups, and publishers enable solitary encounters. Even the most private reading depends on collective preservation. The history of the manuscript demonstrates what its rhetoric sometimes denies.

The imagined reader is also often innocent of technical mathematics. Grothendieck explains that some passages may pass overhead, but the central account of creative work should remain accessible. This produces an unusual layering: technical names without full definitions, followed by metaphors intended to carry the manner of discovery. A specialist and non-specialist do not read the same book, though the pages are identical.

The biographer can preserve this layering by resisting two temptations. One is to remove all technical detail and leave only personality. The other is to turn every named theory into a textbook inset and bury the meditation. The “human problem → faithful intuition → precision window” method follows the manuscript’s own effort while adding clearer boundaries.

From testimony to public object

The provisional 1985 mailing had a bounded audience, but the text anticipated wider diffusion. The 1986 preface mentions people who encouraged publication. The 1987 Paris visit records another attempt. These facts refute the claim that Grothendieck intended permanent secrecy during composition.3535Grothendieck, Récoltes et Semailles (Université des Sciences et Techniques du Languedoc and Centre National de la Recherche Scientifique, 1986); Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, “Complements and corrections to Récoltes et Semailles”; Scharlau, Spirituality.

They do not establish that he accepted every later form of publication. Authorial intention can change. A person can circulate a typescript, seek one publisher, reject another edition, and later restrict reproduction. Rights analysis must identify the date and scope of each act instead of deriving perpetual consent from initial mailing.

The Japanese volumes show translation as editorial selection. A translator must decide how to render wordplay, mathematical terminology, and the second-person voice. The title alone requires choices: “reaping and sowing,” “harvests and sowings,” or another agricultural rhythm. Each shifts emphasis between acts and products.

The French Gallimard edition presents a different challenge: stabilizing a French work whose provisional states circulated for decades. Its ISBNs and dates establish formal issue. They do not tell a historian which manuscript leaves underlie every page. That requires consultation with the Montpellier holdings and authorial copies.

Online files created a de facto international readership before 2022. Search engines made phrases easy to quote, often without the warnings on the hosting page. The same technology that rescues an inaccessible text can flatten its state history. A quotation copied from a volunteer transcription may migrate until no one remembers that supplements were missing.3636Grothendieck, Récoltes et semailles I; Grothendieck, Récoltes et semailles II.

This account counters that flattening with source labels. “Original typescript scan,” “volunteer transcription,” “1986 complement,” “Japanese partial edition,” and “Gallimard edition” are not redundant descriptions. They identify different witnesses. A citation should name the witness actually checked.3737Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; National Institute of Informatics, “CiNii Books: Records of the Japanese Translation of *Récoltes Et Semailles*”.

Reception beyond admiration or rejection

Readers have often divided around the work’s tone. Some value its account of mathematical creation and community more than any conventional autobiography. Some cannot accept its allegations or scale. Some read only the Promenade. Others approach it as literature, spiritual testimony, or source for the history of SGA.

These receptions can coexist because the work is composite. A mathematician may find a precise memory about motives valuable and reject the moral inference drawn from it. A literary critic may admire the voice and question the documentary method. A former participant may recognize events but contest their arrangement.

The work’s status as testimony makes disagreement part of its afterlife. Testimony asks to be believed but also invites corroboration. The ethical reader neither grants belief automatically nor treats inconsistency as reason to discard the witness. She asks what this person could know, what documents survive, what interest shaped the account, and what other voices say.

The manuscript’s greatest contribution to biography may be its refusal to separate mathematical life from modes of relation. It asks how a seminar transmits trust, how a name recognizes dependence, how silence injures, and how a brilliant result can coexist with institutional failure. Even when its verdict is excessive, the questions expose what prize lists omit.

Its greatest danger may be the conversion of hurt into total explanation. Once “burial” governs the field, a missing letter, a new theorem, an editorial choice, and a student’s independence can all confirm the same story. No possible event disproves it. A historical argument that cannot be tested needs another form.

For that reason the work should be read both slowly and externally. Slowly, to hear changes and qualifications. Externally, against title pages, letters, dates, and other testimony. Reverence reads only inward; dismissal reads only outward. Scholarship moves between.

Editions and witnesses

A reliable critical apparatus would begin with a table for every fascicle: authorial title, section numbers, composition dates, typescript witness, fair-copy witness, first mailing, supplements, and later edition pages. Without such a table, citations to “p. 100” may point to incompatible states.

The original first-part scan is valuable because corrections and spacing reveal authorial process. The volunteer transcription is valuable because text can be searched. The Gallimard edition is valuable because it is a formally issued reading text. These values are complementary, not hierarchical in every research question.

For wording, a facsimile or critically checked edition controls. For the history of circulation, envelopes, mailing lists, and recipient copies control. For publication date, publisher metadata controls. For reception, letters and reviews control. One “best source” cannot answer all four.

A future digital edition could align witnesses without pretending to merge them. A reader might select “1985 mailed state,” “1986 complemented state,” or “2022 edition,” with variants visible. Such infrastructure would make the work’s temporal method legible rather than normalize it away.3838Grothendieck, Récoltes et semailles I; Grothendieck, Récoltes et semailles II.

Translation needs its own alignment. The Japanese volumes should be mapped section by section. Existing English fragments vary in freedom and completeness; none should be silently treated as an authorized complete translation. This biography therefore translates only short phrases and paraphrases arguments.

The title itself has variants in English. “Reaping and Sowing” emphasizes actions in progress. “Harvests and Sowings” would emphasize products or occasions. Grothendieck used “Reaping and Sowing” in correspondence, according to the source account, making it preferable here. Translation remains part of reception, not the French original’s date.

Length as historical fact, not spectacle

Page totals vary because preliminary material, notes, appendices, and typesetting differ. A number such as 1,500 can communicate scale but becomes misleading if detached from edition. The chapter therefore names page ranges only with a witness.

Word-count analogies are even less secure. Dense French typescript, handwritten notes, and a modern printed book have different page capacities. Calling the work “millions of words” without a computational source turns scale into legend. The accepted fact is that it spans well over 1,000 typescript or transcription pages across its main architecture and additions.

Length affected who could respond. A former colleague with teaching and research obligations could not quickly verify hundreds of pages of allegation and mathematics. Silence may therefore reflect burden rather than contempt. The format shaped the social event.

Length also distributed power. The author could revisit every reply through another note; a correspondent had to decide whether to enter an expanding system. A shorter review could appear superficial because it did not match the manuscript’s scale. This asymmetry helps explain the impasse.

Yet compression carries its own violence. Reducing the work to “Grothendieck accused his students” erases the promenade, self-examination, and mathematical history. The biographer must perform selective compression openly: state which themes are omitted and preserve routes back to the text.

The work as a source for oral mathematics

Published mathematics often hides speech. Récoltes et Semailles recalls seminars, conversations, questions, and the moment an idea was named. These memories can help historians reconstruct oral transmission, provided they are checked against seminar programs and other accounts.

Oral knowledge has uncertain ownership. A phrase can circulate for years before appearing in print. Students may no longer remember who first said it. The person with the strongest archive may appear to be the originator. The manuscript’s attention to oral life is therefore both evidence and warning.

Grothendieck often learned through conversation rather than systematic reading, especially in intense collaborative periods. He also supplied others with large stores of unpublished ideas. This made the school fertile and created later provenance problems. Informality speeds discovery and weakens the paper trail.

Modern practices—dated preprints, version control, recorded seminars, contribution statements—can improve provenance but cannot capture every influence. They also create surveillance and information overload. The goal is not to bureaucratize every thought; it is to leave enough evidence for fair attribution.

Récoltes et Semailles is itself an attempt to rebuild that evidence from memory. Its failures of proportion do not cancel its archival impulse. The text names what polished papers leave out: who spoke, who listened, who typed, who answered, and who did not.

Sowing toward an unknown reader

The future reader is essential to the title. Grothendieck could not control whether the mailed pages would be read, dismissed, recopied, translated, or archived. He could choose names, arrange lists, add notes, and answer letters. After that, the seed left his hand. The limited 1980s circulation did not prevent later influence; the absence of a French book did not preserve the text unchanged; online surrogates did not make the source history simple.

Some seeds were mathematical. The promenade offered later readers a conceptual map of his work. Historians and mathematicians have used it to identify the internal connections among schemes, topoi, cohomology, motives, and late programs. Some seeds were literary: the child, the mother, the boss, burial, harvest, and sowing became a vocabulary through which readers discuss creation. Some were polemical and continued old injuries. A harvest can carry all of these at once.

The text’s refusal of a final conclusion is therefore accurate in more than one sense. Composition remained open for years. Disputes elicited supplements. Publication came in partial and delayed forms. Readers continue to disagree about the balance of insight and accusation. A critical biography cannot impose finality by declaring the work either masterpiece or breakdown. It can specify what kind of evidence each page offers and where corroboration ends.

The absence of a final conclusion does not mean that every claim remains equally plausible. Dates can be checked. Publications can be examined. Letters can be compared. Mathematical attribution can be separated into proposal, theorem, proof, exposition, and editorship. Allegations can be identified as allegations. A disciplined history preserves openness at the level of interpretation while closing avoidable factual gaps.

Quoting a vast text

The scale of Récoltes et Semailles can tempt a biographer to build the chapter from long extracts. That would outsource narration to a copyrighted source and obscure the work’s architecture. This account instead uses two short sentences with exact locators and paraphrases the rest.

Shortness alone is not sufficient. A fragment can distort context. The six-word French statement about mathematics as collective work sounds harmonious if detached from the surrounding contrast with solitary meditation. The chapter restores the tension in prose while keeping the French quotation brief.

Translation adds risk. A lyrical English rendering may intensify or soften accusation. Working translations here stay close to the syntax of the original. A publication edition should either commission and credit a translator or obtain rights to an existing translation.

Names inside accusatory passages require special handling. Reproducing a vivid charge can renew harm even when introduced as quotation. Paraphrase allows the historian to identify the allegation, supply counterevidence, and avoid making invective the most memorable language.

The work’s style still needs representation. Rhythm, direct address, agricultural imagery, and recursive notes can be described formally. A few licensed samples in a later edition could demonstrate voice, but they should be selected for literary method rather than sensational force.

Quotation policy is thus part of source ethics, not only copyright compliance. Exact locator, limited length, faithful context, translator credit, and right of reply all matter. A text about failed attribution should not be used through careless attribution.

This restraint also protects future editions. The Gallimard text may differ from the online transcription in punctuation or wording. Quotations should be checked against the chosen edition, with the older witness noted where it differs.

The manuscript also sows a question about biography itself. A life of a mathematician is often written as a sequence of results, appointments, honors, and influences. Récoltes et Semailles insists that such a sequence misses the inner quality of work and the communal conditions of transmission. Yet its own inwardness can obscure people around the author. The biographer must do what the text both demands and resists: take the first-person meditation seriously, then restore the plurality of witnesses.

This restoration begins with small grammatical choices. “Grothendieck alleged” is not evasive when evidence is contested; it is accurate. “The volume credits” describes a title page; it does not settle who originated an idea. “Serre replied” records a countervoice without making him judge of all facts. “The secretary prepared a fair copy” acknowledges labor without inventing authorship. “Copies circulated” does not pretend there was a published edition. Precision in verbs is an ethics of credit.

The same precision governs any timeline of the work. “Drafting” occupies 1983–1986; “provisional circulation” begins during that span; “Japanese partial publication” begins in 1989; “French Gallimard publication” belongs to 2022. Separate rows prevent one colored bar from implying that the whole work was continuously in print.3939Grothendieck, Récoltes Et Semailles, Partial Japanese Translation; National Institute of Informatics, “CiNii Books: Records of the Japanese Translation of *Récoltes Et Semailles*”.

An image of stacked pages would communicate magnitude but not state; a timeline shows what readers most often get wrong. A facsimile, likewise, must be identified as original typescript, fair copy, or edition; “manuscript page” is not enough.

This documentary discipline answers the title’s metaphor. A seed has a date of sowing, a plant a period of growth, and a harvest a later occasion. Publication history is not botany, but the ordered stages explain why composition and reception cannot share one year.

The work itself repeatedly crosses those stages, returning to seed what had seemed harvested. The biography need not force completion where the author denied finality. It needs only to mark each surviving state so another reader can follow.

That principle extends to word count. The 50,000-word target for Part X measures this biography’s English main text, not the size of Grothendieck’s French manuscript. It should never imply equivalence, an abridgment ratio, or completeness of coverage.

A reader seeking the whole source must go to the edition and archive. The chapter is a sourced narrative route, not a surrogate publication. Its omissions are governed by relevance, evidence, privacy, and copyright rather than a claim that omitted pages do not matter.

The route also preserves disagreement. It does not ask readers to choose in advance between Grothendieck’s testimony and his correspondents. It supplies dates and categories of claim so that the act of reading can remain an inquiry rather than an oath of allegiance. That freedom is part of what sowing toward an unknown reader should mean.

Inquiry also permits later correction. A living apparatus is the scholarly counterpart to the manuscript’s open-ended form.

The same precision protects the work from caricature. To say that Récoltes et Semailles is only a settling of accounts ignores its mathematical autobiography, theory of creative work, self-criticism, and experiment in address. To say it transcends every factual objection because it is a meditation excuses harm and makes source criticism impossible. Its form contains both generosity and accusation. The reader must remain free to receive one without being conscripted into the other.

By April 1986, the principal late complements had been written. By April 1987, a French publication was again being discussed and again did not materialize. Grothendieck was already writing in another register. La Clef des Songes would begin at the end of that month and move from the ethical and creative life of mathematics toward dreams, God, childhood, and mission. The transition was not a clean abandonment of Récoltes et Semailles. Themes of listening, solitude, creativity, testimony, and fear passed into the new manuscript.4040Grothendieck, Récoltes et Semailles (Université des Sciences et Techniques du Languedoc and Centre National de la Recherche Scientifique, 1986); Grothendieck, Récoltes et Semailles, ed. Carmona (1985); Grothendieck, “Complements and corrections to Récoltes et Semailles”; Scharlau, Spirituality.

Before following that passage, however, the community dispute must be examined on its own terms: the roles assigned to authors, editors, students, heirs, and alleged appropriators; the title pages of the published volumes; Serre’s answer; Grothendieck’s account of Mebkhout; and the line between durable problems of credit and unsupported accusation. Reaping had produced a contested account of ownership. Sowing required a community in which an idea could leave its origin without losing its history.