一门未来几何的原理

Open chapters

EGA I’s 1960 general plan listed thirteen open chapters; Chapter 25 preserves the full sequence, while this chapter tracks publication and migration.11Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”.

At the end of 1961, three large instalments existed: EGA I, The Language of Schemes, published in 1960; EGA II, Global Elementary Study of Some Classes of Morphisms, published in 1961; and the first part of EGA III, Cohomological Study of Coherent Sheaves, also published in 1961. Together they had already changed the available grammar.22Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”.

What “elements” meant

The title Éléments de géométrie algébrique echoed Euclid and Bourbaki. It did not promise elementary exercises. “Elements” meant foundational components from which a field could be constructed: definitions, operations, exact dependencies, and theorems reusable across problems.33Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”.

This explains the density. EGA is not a chronological record of discoveries. It suppresses false starts and arranges results by logical need. A definition may appear hundreds of pages before its most dramatic use. A proposition proves stability under base change because later chapters will use the fact without stopping.44Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”.

The human problem was to produce a reliable reference while the research program was still growing. If the authors waited for completion, nothing might appear. If they published disconnected articles, the language might fragment. Serial “elements” offered a third path.

An operating-system analogy is useful with limits. EGA supplied common operations on which later mathematical applications could run: fiber products, relative spectra, coherent sheaves, proper morphisms, cohomology. The analogy fails if it implies central design controlled every later theorem or that users merely executed predefined commands. Algebraic geometers modified the language, found counterexamples, and built new categories.

The title also declared scale. A paper can solve a problem; elements aim to make many papers possible. The ambition attracted readers and collaborators, but it imposed a steep entry cost and placed unusual authority in the authors’ notation.

The first volume: language

EGA I runs from pages 5 to 228 of Publications Mathématiques de l’IHÉS number 4. Its legal deposit records the third quarter of 1960. On the title page, Grothendieck appears as author and Dieudonné under a separate line crediting collaboration in redaction.55Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”.

The chapter builds affine schemes from spectra, defines preschemes and schemes in the historical terminology, develops products and separation, treats finiteness conditions and rational maps, and gives an extended account of sheaves of modules, quasi-coherence, coherence, and formal schemes.66Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”.

The order begins with spaces but moves quickly to sheaves. A scheme’s structure sheaf makes local algebra geometric; modules over that sheaf make linear and ideal-theoretic data geometric. An ideal sheaf defines a closed subscheme. An invertible sheaf encodes a line bundle. A coherent sheaf can describe a subvariety, divisor, vector bundle, or more singular object.

On an affine scheme $X=\operatorname{Spec}A$, an $A$-module $M$ gives a quasi-coherent sheaf $\widetilde M$. Localization agrees with restriction:

$$\widetilde M(D(f))\cong M_f.$$

This dictionary lets module calculations glue across affine opens. Coherence adds finite generation under Noetherian hypotheses and supplies the finiteness needed for cohomology.

Serre’s FAC is the direct predecessor EGA names. FAC developed coherent sheaves and cohomology on algebraic varieties. EGA I extends the local module–sheaf architecture to schemes over general rings. The historical sentence is not “Grothendieck invented coherent sheaves.” It is that he and Dieudonné made them structural in the scheme foundation, building on Serre.77Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”.

The second volume: maps

EGA II appeared as number 8 in 1961 and occupied pages 5–222. Its title identifies a global elementary study of classes of morphisms. Chapter 27 gives the definitions and examples; the publication consequence here is that relative morphism language had become a numbered reference layer before the coherent-cohomology chapter appeared.88Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”.

The volume’s position made later construction and cohomology arguments shorter to state. Authors could cite established classes and stability results instead of rebuilding them inside every exposé. The title’s word elementary marks the volume’s place before the cohomological methods of Chapter III, not an assessment of difficulty.

The third volume opens the cohomological chapter

EGA III.1 appeared as number 11 in 1961, pages 5–167. It continues the general preliminaries with a section on representable functors and then begins the chapter on coherent cohomology. Chapter 28 treats $h_X$, Yoneda, and full faithfulness; this chapter records only the placement and publication date.99Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”.

The first part did not complete Chapter III; the second part appeared in 1963. That open edge is the dependency consequence: by the end of 1961 a substantial coherent-cohomology layer was public, while its continuation and many later formulations remained future work. The exact modern theorem statements are not reassigned to this fascicle without proposition-level locators.

Dieudonné’s pages

The title page continued to identify Dieudonné’s role as collaboration in redaction. The correspondence gives snapshots of the work: Grothendieck sent notes and commentaries; Dieudonné drafted chapters; Serre read and corrected; chapters went to press; Grothendieck developed material farther ahead.1010Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Serre, Correspondance Grothendieck–Serre.

It is common to say Grothendieck supplied vision and Dieudonné details. The contrast is useful only if “details” receives its mathematical weight. Turning a new theory into thousands of linked statements requires selecting lemmas, ordering dependence, checking edge cases, and enforcing notation. A gap in a detail can collapse a later theorem.

At the same time, the asymmetry should not be blurred. EGA identifies Grothendieck as the author and Dieudonné as collaborator in redaction. The research program, principal new concepts, and direction belonged to Grothendieck. The precise sentence-level division cannot be reconstructed from the printed title alone.

Serre’s contribution remains external and direct: foundational earlier papers, correspondence, draft criticism, examples, and corrections. He declined or never completed proposed historical notes and joint pieces for EGA. The book must not use “Grothendieck–Serre school” to dissolve authorship.

The treatise was thus neither solitary nor anonymously collective. It was a directed collaboration embedded in a critical network.

The general plan and its migrations

Chapter V was to develop elementary construction of schemes. Chapter VI was to treat descent and general construction. These did not appear as completed EGA chapters. Material circulated through FGA Bourbaki lectures: formal geometry, descent, quotients, Hilbert schemes, and later Picard schemes.1111Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck, “Géométrie formelle et géométrie algébrique”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. I. Généralités. Descente par morphismes fidèlement plats”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. II. Le théorème d’existence en théorie formelle des modules”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. III. Préschémas quotients”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. IV. Les schémas de Hilbert”; Grothendieck, Revêtements étales et groupe fondamental (SGA 1); Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. V. Les schémas de Picard: Théorèmes d’existence”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. VI. Les schémas de Picard: Propriétés générales”.

Chapter IX was to treat the fundamental group, but that material moved into later seminar work. Chapter VII’s group schemes and other cohomological and duality themes also developed outside the planned EGA sequence.

Migration reflects more than excess length. Some subjects benefited from multiple speakers and rapid seminar development. The seminar could respond to current problems while EGA maintained a slower systematic form.

But migration complicates citation. A theorem anticipated in EGA I’s plan may have its proof in an SGA volume edited years later from a seminar held earlier. The date of idea, lecture, manuscript, and publication can differ. This Part uses the date appropriate to each claim rather than one undifferentiated “Grothendieck period.”

The acronym FGA is likewise retrospective as a collection of Bourbaki exposés on foundations. The individual exposés have specific dates: May 1959, December 1959, February 1960, February 1961, May 1961, and the Picard lectures in February and May 1962. Grouping them should not erase chronology.1212Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck, “Géométrie formelle et géométrie algébrique”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. I. Généralités. Descente par morphismes fidèlement plats”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. II. Le théorème d’existence en théorie formelle des modules”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. III. Préschémas quotients”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. IV. Les schémas de Hilbert”; Grothendieck, Revêtements étales et groupe fondamental (SGA 1); Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. V. Les schémas de Picard: Théorèmes d’existence”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. VI. Les schémas de Picard: Propriétés générales”.

The future named too early

The 1960 plan names Weil cohomology as Chapter XIII. Grothendieck hoped for a theory strong enough to address Weil’s conjectures. At that point, the desired étale cohomology had not been constructed.1313Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck, Revêtements étales et groupe fondamental (SGA 1); Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. V. Les schémas de Picard: Théorèmes d’existence”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. VI. Les schémas de Picard: Propriétés générales”.

It is tempting to read the chapter title as proof that the future was already known. It was a research target. The program identified properties a cohomology should have and arranged foundations thought necessary to reach it. The route would require étale maps, sites, topoi, derived methods, and collective work.

Similarly, naming Picard schemes did not prove representability in all cases. The 1961 exposés set up construction; Picard existence lectures followed in February and May 1962 under hypotheses. Naming motives later would not construct their full category.1414Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck, Revêtements étales et groupe fondamental (SGA 1); Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. V. Les schémas de Picard: Théorèmes d’existence”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. VI. Les schémas de Picard: Propriétés générales”.

Grothendieck’s named research destinations coordinated work and exposed missing bridges. Historical writing should preserve the subjunctive: would, should, hoped, projected.

The difference between program and theorem does not diminish vision. It makes vision legible as a force acting before proof.

The architecture of cross-reference

EGA’s numbered propositions create a map through the corpus. A later proof can cite (I, 5.5.8) or (II, 4.5.2) rather than restating a theorem. Chapter, section, and proposition numbers make dependencies inspectable.1515Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”.

This system also reveals the modular ideal. Definitions and results should be reusable outside their first motivation. A lemma about quasi-coherent algebras supports relative Spec; relative Spec supports affine morphisms; affine morphisms enter descent and construction.

The cost is that reading becomes non-linear. A page may refer backward through several fascicles and forward to promised material. Changes to early chapters propagate. The authors’ declaration that chapters remain open acknowledges this versioning problem.

Serial publication gave each fascicle a stable page range but not an eternally final text. EGA I was later revised in a new edition. Modern readers often use electronic retypings or summaries whose numbering follows one version. Exact quotations and claims must cite the 1960 fascicle used here.

The publication architecture belongs to the intellectual history. It made a growing theory navigable enough to become communal infrastructure.

An unfinished monument

EGA’s scale and endurance can make the text look like a finished stone object. Its publication policy instead treated chapters as revisable and open to supplementary fascicles.1616Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”.

The published corpus is incomplete relative to the plan, yet individual fascicles are complete mathematical works. The distinction prevents two errors. One is to say EGA was abandoned and therefore failed. The other is to speak as if all thirteen planned chapters exist somewhere in the eight fascicles.

Incompleteness also had creative effects. Seminars took over open regions. Students and collaborators became authors. Later books reorganized the foundations pedagogically. The absence of a completed canonical treatment left room for alternative expositions.

There were costs too. Important results remained dispersed or difficult to access. SGA volumes appeared years after seminars. Readers needed French, extensive prerequisites, and patience with numbering. A universal language did not mean universal accessibility.

The biography should allow both judgments. EGA was a transformative achievement and an unfinished project. These are not opposing verdicts.

Chapter I as a dependency tree

EGA I begins from prime spectra and builds outward. Basic opens and localizations establish the affine topology. The structure sheaf turns rings into locally ringed spaces. Morphisms reverse ring homomorphisms. Affine pieces then glue into preschemes.1717Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”.

Products appear early because every relative construction needs them. A fiber of $X\to S$ is a fiber product. The diagonal defining separatedness maps into a fiber product. Base change is a fiber product. Without existence and associativity up to canonical isomorphism, later notation would have no foundation.

Subschemes and immersions follow. Quotient rings define closed affine subschemes; quasi-coherent ideal sheaves globalize them. Open subschemes arise from restriction. Locally closed immersions combine the two.

Finiteness conditions then classify manageable objects. Quasi-compactness, Noetherian conditions, finite type, and finite presentation measure different kinds of size. The distinctions become essential once base rings are arbitrary.

Rational maps preserve contact with birational geometry. A rational map is not a morphism defined at generic points by magic; it is an equivalence class of morphisms on dense opens, with hypotheses ensuring useful domains. Integral schemes and function fields recover classical language inside schemes.

The long treatment of modules provides the linear layer. Quasi-coherent sheaves localize like modules. Coherent sheaves add finite presentation in the Noetherian setting. Tensor products, Hom, pullback, and pushforward prepare cohomology.

Formal schemes close the volume by passing from powers of ideals to completed local structures. This ending points directly toward the 1959 formal-geometry lecture and the existence theorems.

The sequence shows a dependency tree rather than a list. Remove products and relative geometry fails. Remove quasi-coherent sheaves and relative Spec fails. Remove coherence and finiteness theorems lose their objects. Remove formal completion and one route to representability closes.

This architecture is part of Grothendieck’s conceptual contribution and Dieudonné’s redactional achievement. A definition’s place determines which later proof can cite it.

Historical terminology inside the tree

EGA’s préschéma and schéma distinction places separation inside the noun. Modern convention calls both schemes and adds “separated” to the latter. A reader using a modern translation must track this shift.1818Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”.

Other terms moved too. A morphism called simple in early correspondence became smooth. “Complete” in older algebraic geometry is often rendered proper in the relative language. Universal domains and geometric points received more exact treatment.

Terminology is not cosmetic. If a quoted theorem says “scheme” in EGA’s restricted sense, translating it as modern “scheme” may drop separatedness. If a letter says “unramified” where the author later distinguishes étale and unramified, a silent update can distort chronology.

Main prose uses current vocabulary for comprehension and flags the historical difference at first use.

Notation also evolved between editions. EGA I’s first edition was later revised. Cross-references in EGA II and III can target the original Chapter I numbering. A digital typesetting based on a revision may not match the 1960 page.

This account treats the Numdam fascicles as primary objects. Modern references verify interpretation but do not replace exact locators.

This textual discipline is one reason the corpus cannot be summarized solely from later textbooks. Textbooks present the successful mature language; EGA’s first editions preserve the language becoming stable.

Relative Proj in the volume sequence

EGA II’s sections on homogeneous spectra, projective bundles, and ample sheaves show how the relative language absorbed classical projective geometry. Chapter 27 has already explained the relative map and base change; Chapter 28 has shown why projectivity matters for Hilbert representability. Here the documentary point is the order in which EGA made those later uses citable.1919Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes,” secs. 2–5.

The volume begins with affine morphisms and then moves through relative projective constructions before reaching proper and projective morphisms. That sequence turns coordinate methods into reusable statements over a base. It also keeps distinctions visible: projective implies proper under the stated hypotheses, while proper need not mean projective.

For the future geometry, the important consequence was a shared reference layer. A later exposé could invoke a projective bundle, an ample invertible sheaf, or a projective morphism without rebuilding the definitions. EGA II therefore supplied infrastructure for the Hilbert construction rather than a second independent route to it.

Blow-ups as evidence of an operative language

EGA II closes with blow-ups, cones, and projective closure. Their presence matters because the scheme language was being tested by transformations, not only by classification. The chapter does not repeat the Rees-algebra construction already available in standard algebraic geometry; it records what the construction demonstrated in the EGA sequence.2020Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes,” sec. 8.

A blow-up modifies a scheme along a specified center and produces a projective morphism back to the original scheme. This allowed local and birational questions to be stated over general bases while retaining the center’s scheme structure. The construction therefore linked the inherited geometry of modification with the new relative vocabulary.

Its historical role is bounded. Blow-ups did not solve resolution of singularities in every setting, and their universal property did not make every modification canonical. They showed that the new foundations could carry a classical operation into a broader category without reducing it to a picture.

Serre’s projective results as load-bearing precedents

Serre’s projective cohomology and vanishing results predated EGA and supplied essential control over coherent sheaves after sufficiently positive twists. The exact theorem and its hypotheses belong to Serre’s work; EGA used and extended that infrastructure in a relative setting.2121Serre, “Faisceaux algébriques cohérents”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”.

The consequence for the construction program was finite presentation. High twists can make coherent sheaves globally generated and suppress higher cohomology under the standard hypotheses, allowing families to be encoded by finite linear data. Chapter 28 has explained how this enters the Grassmannian route to the Hilbert scheme.

This placement keeps attribution and chronology separate. Serre’s theorem is not retroactively an EGA invention, and EGA’s relative organization is not reduced to a citation of Serre. The later construction depends on both the prior theorem and the new framework.

EGA as a machine for hypotheses

The repeated adjectives in EGA—Noetherian, quasi-compact, separated, finite type, proper, flat, coherent—can look like barriers between reader and idea. Their function is diagnostic.2222Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”.

Each adjective supports an operation. Noetherian conditions give finiteness. Separation controls diagonals and intersections of affines. Properness controls limits and cohomology. Flatness controls variation. Coherence keeps modules finite.

A proof’s hypothesis list records which mechanisms it uses. Removing an adjective becomes a research question. Counterexamples identify whether the mechanism was essential or merely convenient.

This style made later generalization systematic. One could extend a theorem from Noetherian to quasi-compact and quasi-separated schemes by replacing the finiteness argument, rather than redefining the whole subject.

It also encouraged overstatement in summary. Saying “Grothendieck proved cohomology behaves in families” hides the hypotheses that give the phrase meaning. This chapter keeps the adjectives even at a cost to pace.

The chapter plan, item by item

Chapter I appeared in 1960. Chapter II and the first half of III appeared in 1961. The second half of III would appear in 1963. Chapter IV would occupy four fascicles from 1964 to 1967.2323Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”; Grothendieck, “Géométrie formelle et géométrie algébrique”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. I. Généralités. Descente par morphismes fidèlement plats”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. II. Le théorème d’existence en théorie formelle des modules”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. III. Préschémas quotients”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. IV. Les schémas de Hilbert”; Grothendieck, Revêtements étales et groupe fondamental (SGA 1); Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. V. Les schémas de Picard: Théorèmes d’existence”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. VI. Les schémas de Picard: Propriétés générales”.

Chapters V and VI did not appear as completed EGA chapters. Related construction and descent material circulated through the dated Bourbaki exposés and later seminar publications; this chapter makes no claim here about surviving or translated EGA V manuscripts.

Group schemes, projected for Chapter VII, became a major SGA 3 program. Differential topics and fundamental groups migrated into seminar work. Residues and duality developed through later seminars and publications.

Intersection theory, Riemann–Roch, abelian schemes, Picard schemes, and Weil cohomology each generated bodies of work, but not the intended sequence of EGA chapters.

This itemized ledger prevents a rhetorical move in which “the EGA program” means the total later Grothendieck school. A program can influence work without being its bibliographic container.

The ledger also respects Dieudonné. His collaboration is attached to the EGA fascicles actually written, not automatically to every seminar theme in the original plan.

A plan revised by its own success

No single cause for the treatise stopping during Chapter IV should be asserted without broader archival evidence. The observable facts include expansion of material, growth of seminars, competing research directions, and the immense labor of writing.2424Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Serre, Correspondance Grothendieck–Serre; Grothendieck, Revêtements étales et groupe fondamental (SGA 1).

Success itself enlarged the task. Schemes generated new questions; representability required quotients; cohomology required sites; local study of morphisms became hundreds of pages. A concise foundation was no longer plausible.

The seminar form offered faster collective development. It also dispersed the corpus and made later publication dependent on editors. Some results no longer fit the linear EGA order.

Grothendieck’s published work continued to move into new problems and seminar forms while the planned EGA sequence remained incomplete. The record establishes that trajectory; it does not support a general claim about his temperament or ability to finish a program.

Dieudonné’s finite capacity also mattered in the plain sense that one collaborator could write only so much. The chapter does not speculate about conflict or fatigue without dated sources.

The unfinished plan is therefore evidence of scale and changing method, not a psychological diagnosis.

Three title pages, three stages

Placed side by side, the title pages of EGA I, II, and III.1 form a restrained timeline. Number 4 in 1960 names the language. Number 8 in 1961 names global classes of morphisms. Number 11 in 1961 names coherent cohomology.2525Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”.

The numbering shows other IHÉS publications appearing between EGA instalments. The journal was not a private series for one author. Grauert and other mathematicians published major work there. EGA occupied the journal without defining its entire scope.

The page ranges—5–228, 5–222, and 5–167—show more than six hundred published pages by the end of 1961, excluding FGA and seminar manuscripts. Quantity does not prove quality, but it makes the redactional scale concrete.

Each Numdam scan adds a modern digitization notice and rights statement before the historical pages. Image use must distinguish scan wrapper, original title page, and current host. A facsimile reproduced in the book requires rights review even when freely viewable.

The first title page explicitly contains the collaboration line. Numdam metadata often indexes the article under Alexander Grothendieck alone. Bibliographies variously list Grothendieck with Dieudonné or note Dieudonné’s collaboration. The bibliography preserves the exact title-page wording rather than forcing one modern citation convention to carry the history.

The three pages also give the institute a textual portrait. Before a photograph of a seminar is found, they document what IHÉS was publishing, where, and when.

What publication fixed by the end of 1961

The three fascicles fixed different layers of the new geometry. EGA I established a public language and a numbered system of reference. EGA II organized major classes of morphisms and projective constructions. EGA III.1 opened the coherent-cohomology chapter and carried representable-functor preliminaries into the treatise. The sequence matters because later work could cite these layers rather than rely only on circulating notes.2626Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”.

Publication fixed wording and pagination, but it did not freeze the research program. EGA I itself declared chapters open to supplementary sections. Later editions altered early material. EGA III continued in a second part, and EGA IV grew across several later fascicles. A 1961 citation therefore names a real object while leaving the larger architecture incomplete.

The published order also makes dependencies visible. Classes of morphisms precede relative cohomology because properness, projectivity, and finiteness conditions control what cohomological statements can be made. Representable functors appear in the preliminaries because universal constructions and limits recur throughout the theory. These are documentary observations about placement; the precise theorem statements remain in their mathematical sources and cards.

The Bourbaki exposés formed a second publication channel. Their dates record formal geometry, descent, quotients, Hilbert schemes, and the beginning of the Picard-scheme sequence. The retrospective label FGA is convenient only if it does not erase those dates or imply a single contemporaneous volume. The exposés moved faster than EGA and could state open problems directly.

The two channels distributed authority differently. EGA’s title page named Grothendieck and Dieudonné under a recurring asymmetric formula. A Bourbaki exposé carried Grothendieck’s name as lecturer and author of that treatment. Later SGA volumes would add directors, exposé authors, and editors. The same mathematical program therefore left several kinds of authorship record.

By the end of 1961, construction had advanced without becoming universal. The Hilbert scheme supplied a major projective representability theorem. Quotients still produced explicit obstacles. Picard-scheme results were moving into the 1962 exposés. The chapter can state this balance without teaching the constructions again: Chapter 28 gives the full representability ladder.

The coherent-cohomology layer was similarly real and incomplete. EGA III.1 placed finiteness and formal methods into a relative architecture, but modern semicontinuity and base-change formulations require their own exact locators. This chapter therefore describes their consequences as modern explanatory windows rather than assigning every present theorem sentence to the 1961 fascicle.

Institutional publication was part of the achievement. Publications Mathématiques de l’IHÉS gave the long fascicles a stable venue while the institute itself still worked from temporary Paris premises. Dieudonné’s editorship joined mathematical redaction to the journal’s early operation. The text and institution grew together without becoming the same history.

This ledger also clarifies what Part VI inherits. It does not inherit a completed cohomology theory or a finished seminar corpus. It inherits schemes, relative morphisms, coherent methods, construction techniques, open quotient problems, and a habit of recording dependencies. The next Part must explain how a seminar transformed those resources; Part V ends before that explanation begins.

At this boundary, incompletion is factual rather than rhetorical. The planned chapters were not all published, the seminar books were not yet the later revised volumes, and the Weil program remained open. What existed was sufficient to reorganize research and insufficient to close the problems that had motivated it.

The ledger has one further use: it separates availability from eventual fame. A result printed in 1961 could be cited immediately even if later textbooks presented it in another notation. A seminar topic could be active without a revised volume. A chapter title could announce a destination without supplying its proof. These differences determine what a mathematician in 1961 could actually read, cite, or assume.

For EGA, availability meant fascicles with fixed pages and cross-references. For the Bourbaki sequence, it meant shorter dated expositions that could move a construction program faster than the treatise. For correspondence, it meant proposals and corrections whose circulation was narrower. The same mathematical idea could therefore have several documentary states without those states being interchangeable.

This distinction helps explain why later summaries compress so much. Once a definition is standard, readers no longer see the interval between the lecture and the reference text. Once a theorem has a familiar modern formulation, they may not see which hypotheses or terminology were added later. The Part V ledger restores those intervals without claiming that delay itself produced the mathematics.

It also provides the correct interface to Part VI. The next Part receives an unfinished program, available foundations, and a seminar already in motion. It does not receive a completed SGA corpus or a final cohomology theory. That boundary keeps later books, editors, and proofs from moving backward into 1961.

The chapter can then close with exact dates and open dependencies instead of again repeating the definitions assigned to earlier owner chapters.

Theorem, method, and “yoga”

Grothendieck and later algebraic geometers often used the informal word yoga for a coherent body of methods, analogies, and expected compatibilities. A yoga is not one theorem. It guides theorems and suggests proof patterns.2727Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”.

The scheme yoga includes localization, gluing, base change, and functoriality. The proper-cohomology yoga includes finiteness, formal functions, and semicontinuity. The representability yoga joins deformation, descent, and algebraization.

The word can help describe mathematical practice between formal statements. It can also become a substitute for explanation. This book uses it sparingly and always names the operations involved.

A historical yoga can be stronger than its proved range. Mathematicians may expect a statement before hypotheses are settled. Letters capture this prospective level. EGA records the proved level.

Confusing the two makes Grothendieck either infallible or vague. Keeping them distinct reveals how a broad expectation generated precise work.

EGA and Bourbaki

EGA’s title and style invite the label “Bourbaki for algebraic geometry.” The comparison identifies real features: structural organization, foundational ambition, numbered propositions, collective standards, and personnel shared with Bourbaki.2828Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck, “Géométrie formelle et géométrie algébrique”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. I. Généralités. Descente par morphismes fidèlement plats”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. II. Le théorème d’existence en théorie formelle des modules”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. III. Préschémas quotients”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. IV. Les schémas de Hilbert”; Grothendieck and Serre, Correspondance Grothendieck–Serre.

The comparison is incomplete. EGA was attributed to Grothendieck with Dieudonné’s collaboration, not anonymously to Nicolas Bourbaki. Its chapters arose from a live research program and included results near the frontier. Its planned culmination in Weil cohomology answered a specific conjectural agenda.

Bourbaki supplied language in topology, algebra, and category theory, and its seminar gave FGA a venue. Grothendieck’s 1959 correspondence also records category writing undertaken for a Bourbaki obligation.

Serre, Chevalley, Cartier, and Dieudonné connected the two worlds, but none made them identical. Disagreements within Bourbaki and between exposition styles require sources beyond this Part.

The safest formulation is institutional and textual: EGA drew on Bourbaki habits and people while remaining a distinct directed project in algebraic geometry.

Readers who become collaborators

A foundational corpus creates roles for younger mathematicians. One can verify a lemma, present an exposé, apply a theorem, or discover that the category must be enlarged. The text becomes a research syllabus.2929Grothendieck and Serre, Correspondance Grothendieck–Serre; Grothendieck, Revêtements étales et groupe fondamental (SGA 1).

Grothendieck’s seminar would turn this pathway into collective production. Participants learned the language by using it on fundamental groups, descent, cohomology, and group schemes.

The process should not be romanticized as effortless initiation. EGA’s prerequisites and pace were formidable. Authority over the program was concentrated. Specific participants’ experiences require testimony.

Yet the corpus’s modularity made contribution possible. A seminar speaker could own an exposé inside an architecture. A student could pursue one base-change problem. A visitor could connect a theorem to another field.

This is how an open chapter becomes social. It is not blank space waiting for anonymous completion; it is a defined problem that other mathematicians can enter and alter.

The next Part will distinguish Grothendieck’s direction, named exposé authors, and later editors rather than attributing every page of SGA to one hand.

SGA 1 would turn finite étale covers into a fundamental group; the seminar mechanism, precise construction, and publication history belong to Chapter 31.3030Grothendieck, Revêtements étales et groupe fondamental (SGA 1).

Toward a cohomology for Weil

Weil’s conjectures required a cohomology theory for varieties over finite fields with features analogous to singular cohomology: finite-dimensional groups, functoriality, cycle classes, a Lefschetz trace formula, duality, and an action of Frobenius.3131Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck, “The Cohomology Theory of Abstract Algebraic Varieties”; Grothendieck, Revêtements étales et groupe fondamental (SGA 1).

Zariski cohomology of coherent sheaves did not supply that Weil cohomology. EGA III was essential infrastructure but not the final theory.

Étale cohomology would use torsion sheaves on the étale site and later $\ell$-adic limits. Building it required homological algebra, sites, enough points or substitute arguments, derived functors, and finiteness theorems.

The original EGA plan’s Chapter XIII names the destination. The next seminar years construct roads. Deligne’s proof of the final Weil conjecture in 1974 lies far beyond this Part and has its own authorship.

It is historically important to say both sentences: Grothendieck designed a program centrally aimed at Weil’s conjectures; he did not prove all of them. Program and proof are different achievements.

The unfinished EGA ending before Chapter XIII makes the distinction visible on the shelf.

Reading the open sentence

The sentence used for the proposed epigraph can sound like a declaration of endless creation. In its immediate paragraph, however, it states a publication policy: supplementary sections could appear in separate fascicles to reduce inconvenience. The French wording therefore stays outside the body.3232Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”.

The practical locator should govern interpretation. The authors needed a way to update chapters while publishing serially.

The sentence nevertheless has a broader resonance because the policy matched the program’s growth. Definitions were stable enough to use and revisable enough to extend.

Its limit is equally important. An open chapter is not permission for later editors to alter the historical text silently. Editions, supplements, and translations need version identifiers.

Nor does openness eliminate authorship. A contribution added by another mathematician remains that mathematician’s work even if it fulfills a planned slot.

The sentence therefore closes Part V with disciplined openness: future work is invited, but dates, versions, proofs, and names remain exact.

The state of the work in 1961

The achievements can be stated without superlative. The category of schemes over arbitrary bases had an extended published foundation. Nilpotents, generic points, local rings, and sheaves were integrated. Fiber products and base change made families internal.3333Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”; Grothendieck, “Géométrie formelle et géométrie algébrique”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. I. Généralités. Descente par morphismes fidèlement plats”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. II. Le théorème d’existence en théorie formelle des modules”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. III. Préschémas quotients”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. IV. Les schémas de Hilbert”; Grothendieck, Revêtements étales et groupe fondamental (SGA 1); Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. V. Les schémas de Picard: Théorèmes d’existence”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. VI. Les schémas de Picard: Propriétés générales”.

Global classes of morphisms had a systematic treatment through EGA II. Relative Proj, ample sheaves, properness, finiteness, valuative criteria, and blow-ups were available in one notation.

Formal geometry and proper existence had been presented through Bourbaki. Faithfully flat descent unified earlier field cases and supplied a construction method. Representable and pro-representable functors had become standard questions.

The Hilbert scheme represented flat projective families with fixed polynomial. Quotient theory had important results and explicit open problems. Picard constructions were announced, but the dedicated February and May 1962 exposés lay beyond this Part.3434Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”; Grothendieck, “Géométrie formelle et géométrie algébrique”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. I. Généralités. Descente par morphismes fidèlement plats”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. II. Le théorème d’existence en théorie formelle des modules”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. III. Préschémas quotients”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. IV. Les schémas de Hilbert”; Grothendieck, Revêtements étales et groupe fondamental (SGA 1); Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. V. Les schémas de Picard: Théorèmes d’existence”; Grothendieck, “Technique de descente et théorèmes d’existence en géométrie algébrique. VI. Les schémas de Picard: Propriétés générales”.

EGA III.1 had begun coherent cohomology and installed categorical preliminaries. The second half of the chapter was still forthcoming. Cohomology-and-base-change architecture was active but not every later theorem can be assigned to the first 1961 fascicle.

IHÉS had permanent mathematicians, visitors, a journal, and a purchased future site. It had not yet moved to Bois-Marie, and a new seminar program was under way.

The unresolved work is equally specific. No completed Weil cohomology existed. The quotient obstacles had not disappeared. The thirteen EGA chapters were not written. General moduli with automorphisms lacked the later stack language. Important terms and hypotheses were still changing.

This balanced ledger is more informative than saying Grothendieck had “revolutionized geometry” by 1961. It names the mechanisms of change and the limits of their date.

Toward the seminars

Part VI will inherit an unfinished seminar program. Its first obligation is documentary: separate seminar dates, revised books, directors, exposé authors, and later editors.3535Grothendieck, Revêtements étales et groupe fondamental (SGA 1).

Grothendieck often supplied the program, definitions, and decisive theorems. Seminar speakers proved and wrote named exposés. Artin, Verdier, Giraud, Raynaud, Deligne, and others contributed work with distinct scope. Serre’s earlier conjectures and constructions remained active inputs.

Dieudonné’s role also changes. EGA collaboration continues, while SGA has different editorial and authorial structures. The book should not extend the EGA title-page formula automatically to every seminar volume.

The mathematical three-layer method will remain, but its definitions and examples belong to the next Part. This chapter passes forward only the need for a new cohomological setting.

The chronological method will remain too. A seminar date, exposé revision, and publication date are separate. A conjectural letter is not a theorem. A later proof is not an implicit 1961 possession.

Part V has built the objects and arrows required for that discipline. Its final lesson is not that foundations settle a future, but that they make future questions exact enough for many people to work on them.

The relative world in one view

The Part’s mathematics can be arranged around the map $X\to S$. At the left, rings and spectra construct $X$. Above, base change sends it to $X_T$. Below, quasi-coherent sheaves attach algebraic data. At the right, cohomology pushes data toward $S$. A functor $F(T)$ asks whether a parameter scheme represents families. Descent tests whether local objects return from a cover.

No arrow in that arrangement belongs to one person without qualification. Spectra and local rings have predecessors. Sheaf cohomology carries Serre, Cartan, and others. Yoneda supplies the categorical lemma. Weil and Cartier precede general descent.

The arrangement also has a temporal edge. Beyond it lie sites, topoi, and étale cohomology. They belong after 1961.

Such an arrangement cannot show the institute, journal, correspondence, and seminars that produced the pages. A chronology can. Mathematical and material architectures should meet without one being presented as proof of the other.

The result summarizes this Part: a network with a directing intelligence, inherited nodes, named collaborators, and open outgoing arrows.

What the elements made possible

After EGA I, a mathematician could begin with a scheme over an arbitrary base without redefining its local functions. After EGA II, the map could be classified as affine, projective, finite, or proper with known stability properties. EGA III began supplying coherent cohomology and direct images.3636Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”.

FGA added construction methods. Formal geometry approached an object through every infinitesimal order. Descent asked whether local objects came from the base. Hilbert schemes turned embedded families into a representing scheme.

These tools compose. A projective family gives coherent sheaves. Their cohomology supplies finite modules. Formal completion probes a parameter near a point. Descent glues local constructions. Representability identifies the result through a universal family.

Composition is the measure of a foundation. An isolated definition can be elegant; a system lets one theorem become input to another.

The system also multiplies failure modes. A family may not be flat. A quotient may not be a scheme. Formal data may not algebraize. Cohomology may not commute with base change. The elements make those failures statements rather than surprises.

This precision opened research to correction. Artin could broaden representability. Mumford could address quotients. Giraud could systematize nonabelian descent. Later derived geometry could retain homological intersections lost by ordinary schemes.

Their work is not an appendix to an already complete vision. It shows how a powerful foundation generates questions its original authors do not finish.

The future geometry of the title is therefore plural. Grothendieck supplied an organizing architecture of exceptional scope. Dieudonné made much of it a durable text. Serre and predecessors supplied indispensable beams. Subsequent mathematicians redesigned rooms and added wings.

The architectural metaphor ends where proof begins. Every addition has its own definitions, hypotheses, and authors. The book’s final responsibility is to keep those names attached.

By December 1961 the elements were sufficient for a new collective phase. They were not sufficient for the final conjectures. That distance is the narrative space of Part VI.

Credit and chronology

By the closing date, EGA I, EGA II, and EGA III.1 were published achievements. Formal geometry, faithfully flat descent, quotient results, and Hilbert representability had dated Bourbaki exposés. The Picard exposés, most mature SGA developments, algebraic spaces, stacks, geometric invariant theory, Artin representability, and Deligne’s proof of the Weil conjectures belong later. Grothendieck directed the architecture and proved central results. Dieudonné collaborated in EGA’s redaction. Serre, Weil, Zariski, Chevalley, Nagata, Cartier, Yoneda, Hilbert, Cartan, Eilenberg, Godement, and others supplied distinct prior theorems and languages. Seminar authors and later mathematicians retain their named contributions. The checkpoint is not a disclaimer appended after celebration. It is part of the explanation: the future geometry became possible because several kinds of work joined without becoming identical. A plan can guide results it does not contain, and a foundation can enable authors it does not own. The chronological separation protects both vision and later achievement. Grothendieck could name a destination without possessing its proof; a seminar could begin a theory before its edited volume appeared; a collaborator could write an exposé inside a directed program without surrendering authorship. These distinctions will govern Part VI. They also protect readers from a false choice between single-author myth and featureless collectivity. The history is more exact: one mathematician set an exceptional direction, several predecessors supplied indispensable structures, collaborators converted plans into texts and theorems, institutions sustained the work, and later authors changed the architecture. The outgoing arrows belong to a future whose names must remain attached.3737Grothendieck and Dieudonné, “Éléments de géométrie algébrique I: Le langage des schémas”; Grothendieck and Serre, Correspondance Grothendieck–Serre; Serre, “Faisceaux algébriques cohérents”; Grothendieck, “Sur quelques points d’algèbre homologique, I”; Grothendieck, Revêtements étales et groupe fondamental (SGA 1).

The Part ends with no claim of completion. Its achievement is a verified platform: enough language, relative structure, and existence machinery for a larger collective program to start, and enough recorded limits to show why that program was still necessary. Work would continue.

1961 at the edge

By May 1961, the Hilbert-scheme exposé could say that a satisfactory technique of projective construction was available, while noting missing quotient methods and the absence of a nonprojective general construction technique.3838Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”; Grothendieck and Dieudonné, “Éléments de géométrie algébrique III: Étude cohomologique des faisceaux cohérents, première partie”; Grothendieck, “Techniques de construction et théorèmes d’existence en géométrie algébrique. IV. Les schémas de Hilbert”; Grothendieck, Revêtements étales et groupe fondamental (SGA 1).

By the end of the year, EGA II and III.1 had carried relative morphisms and coherent cohomology into print. New seminar work was under way, and IHÉS was preparing to move to Bures.

This is a summit only in retrospect. From inside the documents, major components remained under construction. Grothendieck was expanding programs in correspondence. Dieudonné was writing. Serre was correcting and pursuing his own work. Quotient questions remained open.

The next phase would make the seminar the primary engine for a cohomological architecture. Its mechanisms, dates, and named contributors belong to Part VI.

Those achievements should not be credited to 1961 before their proofs. What 1961 supplied was an unusually prepared threshold: schemes, relative maps, representability, formal existence, projective parameter spaces, and an institution capable of sustaining collective work.

A future organized, not owned

Grothendieck’s program gave future algebraic geometry a vocabulary of tasks. Build a cohomology. Represent a functor. Descend an object. Study a morphism relative to its base. Replace a coordinate construction by a universal one.3939Grothendieck and Serre, Correspondance Grothendieck–Serre; Grothendieck, Revêtements étales et groupe fondamental (SGA 1).

The program did not own every answer. Michael Artin would transform representability and algebraic spaces. David Mumford would develop geometric invariant theory. Pierre Deligne would later prove the Weil conjectures using an architecture to which many had contributed. Jean Giraud would systematize nonabelian cohomology. Students and collaborators would write SGA exposés and new theories.

Historical balance is not achieved by appending these names to Grothendieck’s vision as executors. Their results have independent creativity. The program’s power can be measured by the number of strong mathematicians who could enter it and alter it.

Dieudonné’s independent mathematical identity also persists beyond his collaboration. Serre’s work was never contained by EGA. Motchane’s institute hosted physics and other mathematics. The future geometry grew in a community broader than one program even when that program dominated.

The philosophical theme is therefore openness with boundaries. EGA opened chapters because research was alive, not because every later idea was already latent and attributable to its author.

Foundations become a research program

Part V began with a demand for foundations and ends with foundations becoming a research machine. The scheme retained prime, generic, arithmetic, and infinitesimal information. The relative viewpoint placed the map $X\to S$ at the center. Representability asked whether families themselves formed geometry. IHÉS supplied time, publication, and encounters. EGA stabilized enough of the language for a collective program to proceed.4040Grothendieck and Dieudonné, “Éléments de géométrie algébrique II: Étude globale élémentaire de quelques classes de morphismes”.

No one element explains the others. Schemes without descent would be a broad category with fewer construction tools. Representability without formal existence would remain a question. A seminar without Dieudonné’s and others’ written labor would leave a thinner record. An institute without Grothendieck’s program would have had a different early identity.

The joining is the achievement. It is also where myth tends to concentrate. One name comes to stand for a language built from many sources; one estate for work begun elsewhere; one completed-sounding acronym for open chapters. This Part has kept the seams visible because the seams carried the work.

The last printed destination in EGA I’s plan was Weil cohomology. Reaching it would require changing what “open set,” “cover,” and even “space” could mean. The relative world had made the base variable. The next architecture would make topology variable too.

Part VI begins from that unfinished work.