标准猜想

Seven pages from Bombay

The bibliographic object is startlingly small. “Standard conjectures on algebraic cycles” occupies pages 193 through 199 of the proceedings of an international colloquium on algebraic geometry held at the Tata Institute of Fundamental Research in Bombay in 1968. Oxford University Press published the volume in 1969. A program intended to reorganize the relation among algebraic cycles, cohomology theories, and the Weil conjectures entered the record in seven pages.11Grothendieck, “Standard Conjectures on Algebraic Cycles”.

The brevity should not be mistaken for simplicity. Short mathematical papers can depend on a large shared language. By 1968 Grothendieck and his collaborators had spent years building schemes, étale cohomology, derived methods, duality, and the formalism of correspondences. The seven pages did not need to reconstruct that machinery. They identified structural assertions that the machinery still lacked.

Bombay also makes the program’s geography visible. The conjectures were shaped at IHÉS, connected to Serre’s ideas, discussed in correspondence and seminars, expounded by Kleiman, and presented to an international colloquium in India. A surviving photograph caption from the meeting places Grothendieck and Armand Borel among participants, but a photograph does not establish the content of a conversation. The proceedings do. They show algebraic geometry operating through centers that no national story can contain.

The title’s adjective is one of the first obstacles. In ordinary English, a standard problem is routine. These conjectures were not. “Standard” named the properties a satisfactory theory was expected to possess by analogy with classical Lefschetz and Hodge theory. The assertions were normative for the structure, not elementary for the solver. They proposed that the algebraic geometry of varieties over arbitrary fields should retain deep features known over the complex numbers.22Kleiman, “Algebraic Cycles and the Weil Conjectures,” 359–61.

Grothendieck’s strategy had a distinctive scale. The Weil conjectures concerned zeta functions and numbers of points over finite fields. One might attack their statements directly. He instead asked what general theorems about algebraic cycles would make the hardest parts of the Weil conjectures formal consequences. The route went upward in abstraction and downward toward geometry at once: upward to a universal cohomological architecture, downward to the question of whether specific linear operators were induced by actual algebraic correspondences.

The arithmetic target

Chapter 34 explains the zeta function, Frobenius, the trace formula, and the four Weil conjectures. This chapter uses only the conditional target needed for the standard conjectures: algebraic Lefschetz operators, degree projectors, and positivity would force the degreewise Frobenius weight bound $q^{i/2}$. Rationality and the functional equation do not require the full standard package, and Deligne’s 1974 proof reached the weight theorem without proving that package.33Kleiman, “Algebraic Cycles and the Weil Conjectures,” 359–61, 383–86.

Lefschetz as a ladder

Choose a hyperplane section of a smooth projective variety $X$ of dimension $n$, and let $L$ denote the operation of cup product with its cohomology class. Applying $L$ raises cohomological degree by two. Hard Lefschetz says that for $i\le n$,

$$L^{\,n-i}:H^i(X)\longrightarrow H^{2n-i}(X)$$

is an isomorphism. One may picture the graded cohomology groups as levels of a ladder, with $L$ climbing two degrees at a time. Enough applications carry a lower degree to its symmetric partner above the middle.44Kleiman, “Algebraic Cycles and the Weil Conjectures,” 359–60.

The ladder image captures the symmetry and misses the decomposition. Hard Lefschetz works together with primitive cohomology: classes killed by one further suitable power of $L$. Every class can be decomposed into powers of $L$ applied to primitive classes. On this decomposition, one defines a lowering operator, usually written $\Lambda$, that inverts the relevant action of $L$ up to known scalar factors.

As a linear operator on cohomology, $\Lambda$ is available once hard Lefschetz holds. The standard conjecture of Lefschetz type, conventionally called B, asks for more. It says that $\Lambda$ should be induced by an algebraic cycle with rational coefficients on $X\times X$. In the language of the previous chapter, the cohomological inverse should come from an algebraic correspondence.55Kleiman, “Algebraic Cycles and the Weil Conjectures,” 359.

This move is the heart of Grothendieck’s method. Cohomology may reveal that an operation exists, but a motive should not depend on a mysterious linear map visible only after realization. If the operation is algebraic, it belongs to the geometry before one selects a cohomology theory. Conjecture B is thus both a statement about cycles and a demand for independence.

The claim is stronger than the hard Lefschetz theorem. Hard Lefschetz says an operator is invertible in cohomology. B says an inverse is represented geometrically by a correspondence. An isomorphism in a realization need not automatically lift to the category from which the realization is supposed to come. The conjecture asks for precisely that lift.

In characteristic zero, Hodge theory supplies hard Lefschetz and related positivity. Over finite fields, étale cohomology was designed to reproduce formal properties but could not simply inherit every analytic proof. The program conjectured that the algebraicity assertion held across Weil cohomology theories. Calling it “standard” made a strong structural claim about algebraic geometry.

Cutting the diagonal

Conjecture B has a consequence often labeled C. The diagonal $\Delta_X$ in $X\times X$ acts as the identity correspondence on $H^*(X)$. The Künneth formula decomposes the cohomology class of the diagonal into pieces associated with each degree:

$$[\Delta_X]=\pi_0+\pi_1+\cdots+\pi_{2n}.$$

Each $\pi_i$ acts as the projector onto $H^i(X)$. Conjecture C says that these Künneth components are algebraic—that is, represented by algebraic cycles with rational coefficients. B implies C in Kleiman’s exposition.66Kleiman, “Algebraic Cycles and the Weil Conjectures,” 360.

The assertion matters twice. First, it separates the characteristic polynomial of Frobenius degree by degree using algebraic projectors, a key step in proving coefficient independence and integrality properties. Second, it lets the category of motives treat the cohomological degrees of a variety as algebraically defined direct summands. Without algebraic projectors, the grading seen by a realization may have no corresponding decomposition in the motivic source.

The diagonal can be visualized as identity before decomposition. In linear algebra the identity matrix is the sum of projectors onto invariant subspaces. Cohomology supplies such projectors. Conjecture C asks whether geometry supplies matrices for them in the form of cycles. This is why a motive is not a poetic “essence.” Its proposed pieces depend on highly nontrivial algebraicity statements.

Kleiman noted that B and C were stable in important ways and known in significant cases. Projective spaces and varieties with cellular decompositions behaved as expected; curves, surfaces, and abelian varieties supplied further evidence. Such cases matter historically because conjectures gain credibility not only from analogy but from survival under products, hyperplane sections, specialization, and every example then computable.

At the same time, low-dimensional success can mislead. A statement true for curves and surfaces may conceal new obstacles in higher codimension. Algebraic cycles become dramatically more complicated as dimension increases. Grothendieck’s wager was not that examples had made the general theorem obvious. It was that the examples disclosed the correct general structure.

Equivalence among cycles

The next conjecture concerns equivalence. Suppose two algebraic cycles determine the same class in a chosen Weil cohomology. They are homologically equivalent. Suppose instead that they have the same intersection numbers against every complementary cycle. They are numerically equivalent. Homological equivalence is defined through a measuring theory; numerical equivalence is intrinsic to intersection.77Kleiman, “Algebraic Cycles and the Weil Conjectures,” sec. 1.

Conjecture D asserts, in the standard framework, that homological and numerical equivalence coincide. The direction of implication should not be described casually. A cohomologically trivial cycle has zero intersections for formal reasons, so homological equivalence is finer than numerical equivalence. D says that numerical invisibility already forces homological invisibility. No extra classes disappear merely because one has passed from cohomology to intersection numbers.88Kleiman, “Algebraic Cycles and the Weil Conjectures,” 369–70.

The conceptual payoff is independence. If homological equivalence agrees with numerical equivalence, then a category defined using one Weil cohomology does not change when another cohomology is chosen, at least at that level. The common algebraic source of realizations becomes more plausible because the criterion for identifying correspondences is no longer attached to one realization.

Numerical equivalence also leads to a category with remarkable formal properties. Decades later Uwe Jannsen proved that the category of pure motives modulo numerical equivalence is semisimple abelian under a precise characterization. That theorem belongs to 1992, not to the Bombay colloquium. It confirms that Grothendieck’s choice of equivalence was structurally fertile while leaving the equality with homological equivalence unresolved in general.99Jannsen, “Motives, Numerical Equivalence, and Semi-Simplicity”.

The temptation to convert D into a philosophical slogan should be resisted. “What cannot be measured does not exist” is not its content. Numerical equivalence uses a very rich family of all intersection pairings, and homological equivalence uses a cohomology theory. The conjecture compares two exact mathematical notions. It does not license a general theory of knowledge.

It does, however, dramatize Grothendieck’s search for intrinsic language. A construction that depends on an arbitrary choice may still be useful. His preferred architecture asked whether the choice could be removed by passing to the right category. D identifies a condition under which the dependence disappears.

Positivity without analysis

The standard conjecture of Hodge type, or index conjecture I in Kleiman’s notation, concerns a quadratic form on primitive algebraic classes. After multiplying by the appropriate power of the hyperplane class and inserting a sign depending on codimension, the resulting pairing should be positive definite.1010Kleiman, “Algebraic Cycles and the Weil Conjectures,” 360.

Over the complex numbers, Hodge theory explains such signs and positivity through harmonic forms and the geometry of Kähler metrics. In positive characteristic there is no underlying complex manifold and no analytic metric available in the same way. The conjecture asserts that the algebraic classes nevertheless obey the same type of positivity.

Positivity is easy to understate because its formula looks like an inequality appended to a list of structural isomorphisms. In the route to the Weil conjectures it supplies force. Algebraic correspondences form an algebra with an involution analogous to transpose or adjoint. A positive trace pairing can then constrain operators that preserve it. Normalize Frobenius by the expected power of $q$; if it behaves as an isometry for a positive form, its eigenvalues have absolute value one. Undo the normalization, and the eigenvalues on $H^i$ have absolute value $q^{i/2}$.

This is the skeleton of the conditional Riemann-hypothesis argument. It is not a proof in the absence of the conjectures. It shows why positivity, Lefschetz, and algebraicity are exactly the missing properties. Rather than confronting each Frobenius eigenvalue individually, the program seeks a geometry in which the correct absolute values are forced by symmetry and a positive form.

The word positive can suggest optimism or moral value. Here it refers to a quadratic form taking positive values on nonzero primitive algebraic classes after specified signs. The moral resonance belongs to prose, not mathematics. Keeping that boundary clear is especially important in a life where ethical and mathematical narratives so easily become mirrors of one another.

The formal route

Kleiman’s “Algebraic Cycles and the Weil Conjectures,” dated 1967 in the table of contents and published in Dix exposés in 1968, laid out the route in detail. The text begins from a Weil cohomology satisfying Poincaré duality and the Künneth formula and receiving classes from algebraic cycles. It states the Lefschetz and index conjectures, studies their stability and equivalences, and then turns to Frobenius and the zeta function.1111Kleiman, “Algebraic Cycles and the Weil Conjectures,” 359–86.

Several stages must be distinguished. The trace formula gives point counts from Frobenius traces. From finite-dimensional cohomology one derives rationality of the zeta function. Poincaré duality yields the functional equation without all the unproved standard conjectures. Algebraic Künneth projectors control degree-wise characteristic polynomials. The Lefschetz and positivity conjectures then yield semisimplicity and the absolute-value statement under the hypotheses Kleiman specifies.

Popular summaries often compress this into the statement that the standard conjectures imply the Weil conjectures. The package-level implication appears in Kleiman’s contemporary exposition, but the precision window reveals more. Different Weil statements require different portions of the formalism, and some had already been established with étale cohomology. The standard conjectures offered a unified algebraic derivation and, above all, the positivity needed for the Riemann-hypothesis part.1212Kleiman, “Algebraic Cycles and the Weil Conjectures,” 359–61, 383–86.

The architecture is a characteristic Grothendieck move. Separate what is formal from what is geometric. Build the formal apparatus until the target theorem follows from a small set of exact missing statements. Then make those statements the subject of a broader theory. The famous problem becomes a corollary of understanding.

There is a risk in admiring the architecture too quickly. A reduction can move difficulty rather than dissolve it. The standard conjectures proved extraordinarily hard. To say that the Weil conjectures would follow “formally” does not mean they had become easy. It means the remaining difficulty had been concentrated in the algebraicity and positivity of cycles. Grothendieck regarded that concentration as explanatory progress.

The program also promised more than the Weil conjectures. If the projectors and correspondences were intrinsic, Betti numbers and characteristic polynomials would not depend on the chosen cohomology theory. Isomorphisms induced by algebraic cycles in one realization could be transported to another. Motives would acquire graded pieces and dualities. The conjectures supplied infrastructure for a universal cohomology rather than a detour to one theorem.

Degree by degree

The zeta function packages all cohomological degrees into one alternating product. Rationality of the total function does not by itself identify each degree factor intrinsically. To separate those factors, the formalism needs projectors onto $H^i(X)$. Algebraic Künneth projectors would supply them as correspondences rather than as operations visible only after choosing cohomology.

This distinction can be seen in the formula

$$Z(X,t)=\prod_i P_i(t)^{(-1)^{i+1}}, \qquad P_i(t)=\det(1-tF\mid H^i(X)).$$

The total product is determined by point counts. The individual $P_i$ express the action of Frobenius degree by degree. If the degree projectors come from algebraic cycles, traces of their compositions with powers of Frobenius can be interpreted through intersection theory. The factors then acquire an intrinsic route independent of the selected $\ell$-adic realization.1313Kleiman, “Algebraic Cycles and the Weil Conjectures,” 360, 382–86.

The precise integrality argument requires hypotheses and distinctions suppressed in the main narrative. A rationally algebraic correspondence leads first to rational characteristic data; integral cycles and additional control lead to integrality statements. The chapter therefore does not claim that every correspondence automatically has an integral characteristic polynomial.

This degree-by-degree issue explains why the Künneth conjecture is more than a desirable decomposition. It connects the arithmetic factorization of the zeta function to algebraic geometry on $X\times X$. It also supports the grading of the proposed category of pure motives.

Independence of $\ell$ sharpens the same problem. Étale cohomology uses an auxiliary prime $\ell\ne p$. Arithmetic geometry expects the degree factors to belong to $X$, not to that auxiliary choice. Algebraic projectors provide one proposed explanation for why different realizations agree.

Later theorems established broad independence results without proving the full standard conjectures. Historical sequence matters: the standard program first identified a geometric source that would make independence formal, while later mathematics found additional routes to parts of the conclusion.

For Grothendieck’s program, this was a recurring test of foundations. A theory could calculate a desired invariant and still leave its independence unexplained. The standard conjectures asked for the algebraic operations that would make independence structural.

Kleiman’s attributions

At the end of his introduction, Kleiman wrote that the study was essentially due to Grothendieck, then named three noteworthy exceptions. David Lieberman had proved Lefschetz-type conjectures for abelian varieties. Saul Lubkin had suggested a point concerning denominators in characteristic polynomials. Enrico Bombieri had independently observed that the Weil conjectures were formal consequences of the standard conjectures. Kleiman also acknowledged many others who had contributed to his notes.1414Kleiman, “Algebraic Cycles and the Weil Conjectures,” 361.

That paragraph is a small model of responsible mathematical history. It does not dilute the originality of Grothendieck’s architecture. It specifies where other people’s work enters it. The word independently is especially important for Bombieri. Two mathematicians can arrive at the same consequence within a shared context without one result being a derivative of the other.

Attribution here is not decoration added after the proof. It tells the reader how the program formed. Lieberman’s cases supplied evidence. Lubkin’s observation addressed arithmetic control. Bombieri’s independent insight clarified logical consequence. Kleiman’s own synthesis made the framework legible and citable. Grothendieck’s conjectures became a communal research object through those acts.

The history also warns against anachronistic ownership. A later textbook may refer simply to “Grothendieck’s standard conjectures.” That is an efficient label for origin, not a complete account of the 1967–1969 work. Precise prose states instead that Grothendieck formulated the program, Kleiman systematized it, Lieberman proved specified cases, and Bombieri independently observed a stated consequence.1515Grothendieck, “Standard Conjectures on Algebraic Cycles”; Kleiman, “Algebraic Cycles and the Weil Conjectures”.

The same grammar should govern SGA. A seminar directed by Grothendieck might contain an exposé by Katz proving a theorem inspired by Deligne’s or Lefschetz’s work. The director’s program and the expositor’s theorem are different kinds of contribution. When a field is moving quickly, failure to distinguish them can later produce incompatible memories of generosity and appropriation.

During the summit, Grothendieck’s reported practice was unusually open. He gave problems and ideas to students and colleagues. Serre later contrasted that generosity with the tone of Récoltes et Semailles. The contrast is real, but it cannot decide every credit dispute. A person may share freely and later believe the terms of memory have changed. What the contemporary Kleiman text provides is firmer: a published attribution written while the program was active.

Hodge, Tate, and the standard package

Three conjectural languages meet around algebraic cycles, but they do not ask the same question. The Hodge conjecture concerns specified rational cohomology classes on smooth projective complex varieties. The Tate conjecture concerns Galois-invariant $\ell$-adic classes in arithmetic settings. The standard conjectures concern algebraicity of Lefschetz operations, equivalence relations, and positivity.

The Hodge setting begins with the decomposition of complex cohomology into $(p,q)$-types. Algebraic cycles of codimension $p$ give classes of type $(p,p)$. The conjecture asks whether every rational class of that type is algebraic. Its formulation depends on complex geometry and Hodge structure.

The Tate setting replaces Hodge type with Galois action. Cycle classes are fixed by an appropriate Galois subgroup after the required twist. The conjecture asks whether the invariant classes are generated by algebraic cycles under its hypotheses. The exact statement varies with the base field and version.

The standard conjecture B instead asks for an algebraic correspondence inducing the inverse Lefschetz operator. D compares homological and numerical equivalence. The index conjecture imposes positivity on primitive algebraic classes. These claims can be formulated to address positive characteristic without importing a Hodge decomposition.

The conjectures interact. In characteristic zero, Hodge-theoretic results imply important Lefschetz and positivity properties, and a sufficiently strong algebraicity statement would produce needed correspondences. In positive characteristic, Tate-type results can supply algebraic classes but do not automatically replace every standard-conjecture assertion.

Divisors provide a true low-codimension anchor. The Lefschetz $(1,1)$ theorem controls integral Hodge classes of degree two on complex projective varieties. Divisor theory and the Néron–Severi group make codimension one much better behaved than arbitrary higher codimension. Success there does not prove the general Hodge, Tate, or standard conjectures.

The boundary appears with higher-codimension cycles, where intermediate Jacobians, Abel–Jacobi phenomena, and subtle equivalence relations enter. This chapter does not develop those later theories; it uses the boundary to show why a divisor example cannot stand for all cycles.

Historically, Grothendieck’s standard package extracted the structural properties needed for the Weil and motives programs. It did not merge every algebraicity conjecture into one universal statement. Kleiman’s exposition is especially valuable because it records implications and distinctions at the time.1616Kleiman, “Algebraic Cycles and the Weil Conjectures,” 359–61 and §§1–4.

Modern summaries often place Hodge, Tate, and standard conjectures in one diagram. That diagram is explanatory rather than a record of the exact April 1967 motives talk or a single 1968 formulation. This account states each conjecture’s target before describing relations.

The biographical return is limited but important. Grothendieck’s role was to formulate a package that connected cycles, cohomology, and motives across characteristics. The surrounding conjectures retained their own authors, histories, and later developments.

What the conjectures would give motives

The standard conjectures were not external tests attached to motives after the fact. They supplied properties needed for the desired category of pure motives. Algebraic Künneth projectors would split a variety into graded pieces. Equality of homological and numerical equivalence would reduce dependence on a chosen realization. Positivity would control correspondence algebras and weights.

Without the projectors, cohomology has a grading that the motivic category may not be able to express algebraically. Without D, two realizations may define different kernels on cycles. Without positivity, semisimplicity and Frobenius estimates lose the proposed geometric foundation. The conjectures therefore connect the category’s objects, morphisms, and realizations.1717Kleiman, “Algebraic Cycles and the Weil Conjectures,” 359–61, 382–86; Manin, “Correspondences, Motifs and Monoidal Transformations”.

This relationship also clarifies why motives could be defined unconditionally in several versions. A Chow-motive category can be formed using rational equivalence and algebraic projectors that are available. Numerical motives use numerical equivalence. Homological motives use a selected cohomology. The standard conjectures predict stronger comparisons among these constructions; they are not required for every definition.

The distinction prevents a common overstatement that motives do not exist because the standard conjectures are unproved. Many precise categories exist. What remains conjectural is the full pattern of identifications and properties Grothendieck expected.

The reverse overstatement is equally misleading. The existence of unconditional categories does not mean the original program is complete. A category can be defined while its realization functors, t-structures, weights, or relation to cycles remain only partly understood.

Kleiman’s exposition made these dependencies public in a form that other mathematicians could use. It identified which consequence followed from which assumption and named known cases. The program no longer depended on recovering every oral explanation at Bois-Marie.

This public dependency map also distributed proof obligations. A specialist in abelian varieties could establish B in that class. Another mathematician could study numerical equivalence. A cohomologist could prove Weil by a different route. Partial progress remained legible within the larger program.

The biographical effect was paradoxical only in appearance. Grothendieck’s authority helped establish the conjectures as central, but their precise formulation made them independent of his continuing leadership. Once stated and published, they could be proved, bypassed, or reformulated by mathematicians who did not share his later institutional choices.

That independence is one meaning of a successful mathematical program. It becomes common enough to survive disagreement with its originator while retaining a traceable history of origin.

A conjecture as a public object

The standard conjectures show how an unfinished theory can become public. Grothendieck’s short article stated the program. Kleiman’s long exposition developed its formal consequences. Seminar discussions and correspondence supplied context. Cases proved by others tested it. The name stabilized a family of questions sufficiently for mathematicians who had not heard every Bois-Marie conversation to work on them.

This differs from the motives program’s partly oral transmission. The two were inseparable mathematically, yet their archival forms diverged. Motives remained a broad vision with several possible categories. The standard conjectures became a relatively definite list of assertions. A researcher could attempt B for a class of varieties without possessing Grothendieck’s full philosophy of motives.

That independence is one measure of successful formulation. A conjecture ceases to belong only to its proposer when others can state it precisely, prove cases, find implications, or develop alternatives. Grothendieck’s authority helped the program command attention, but authority alone could not make B true or false. The conjectures submitted the vision to common mathematical judgment.

They also created a durable standard against which new cohomology theories could be compared. The tests included Poincaré duality, the Künneth formula, hard Lefschetz, cycle classes, and algebraicity of key projectors. These tests turned “cohomology” from a family resemblance into an axiomatized field of expectations.

The relation to philosophy is structural rather than mystical. Grothendieck repeatedly sought not just solutions but the natural habitat in which solutions followed. Here the habitat itself depended on conjectural foundations. The method therefore included intellectual risk. If the conjectures failed, parts of the proposed category and conditional proof would need revision. A broad theory did not shelter him from falsifiability; it multiplied the points at which geometry could resist.

Deligne’s different road

In 1974 Pierre Deligne proved the remaining Weil conjecture, including the Riemann-hypothesis statement, by methods that did not establish the full standard conjectures. That later theorem belongs beyond this chapter’s immediate chronology, but it must be stated because hindsight otherwise distorts the stakes. The target that had helped organize Grothendieck’s program was reached. The road he had proposed remained incomplete.1818Deligne, “La conjecture de Weil I”.

Two wrong conclusions often follow. The first is that Grothendieck’s program “failed” because Deligne used another route. This ignores the fact that étale cohomology, Lefschetz pencils, monodromy, weights, and the broader language in which the proof lived were deeply shaped by the Grothendieck school. The second is that Deligne merely completed a proof already contained in Grothendieck’s ideas. This erases the originality of Deligne’s argument and the mathematical difference between proving the Weil conjectures and proving the standard conjectures.

The responsible conclusion holds both truths. Deligne’s proof was made possible in a landscape transformed by Grothendieck and his collaborators. It also introduced decisive ideas and bypassed conjectures that remained open. Historical descent is not identity.

The standard conjectures survived the proof of the Weil conjectures because their purpose was larger. They concern algebraic cycles, equivalence, positivity, and the structure of motives. Even when the arithmetic corollary no longer needed them, the geometric explanation they promised remained compelling. A bridge is not useless because travellers found another route to one destination; it may connect other regions.

Their open status should be dated carefully. As of the writing of this account, the conjectures are known in important cases but not in full generality. An account should also resist the lazy drama of a sixty-year-old “failure.” Longevity can indicate that a conjecture sits at the intersection of several deep structures. It can also mean that its original formulation continues to organize partial results.

The pressure of the unfinished

Grothendieck was approaching forty when the conjectures entered print. Later commentators have wondered whether the incompletion of his enormous program contributed to exhaustion or disappointment. Cartier offered possibilities in this direction, and Scharlau surveyed several interpretations. None supplies direct evidence of Grothendieck’s state of mind while writing the Bombay paper. The biographer should not make the open conjectures into a psychological burden merely because a rupture followed.

What can be said is objective. The EGA plan remained incomplete. Motives lacked a comprehensive treatise. The standard conjectures concentrated major unresolved geometry. The Weil conjectures were not yet fully proved. SGA redactions lagged behind the seminars. Grothendieck’s program had become too large for one person to finish, if “finish” meant writing and proving every component himself.

But it had also become too collective to belong to one person’s completion. Kleiman’s exposition, Manin’s paper, Lieberman’s cases, Bombieri’s observation, Deligne’s and Katz’s seminar work—these were not signs that the program escaped because Grothendieck weakened. They were the normal form of success at that scale. A mathematical architecture becomes historical when other people inhabit and alter it.

That truth would later be difficult for everyone involved. Students and collaborators inherited unfinished work without inheriting a single instruction for how to order their lives around it. Grothendieck had given ideas generously, but recipients were mathematicians with their own judgment. They could choose different problems, styles, and standards of citation. The mathematical form already carried an ethical issue concerning the obligations attached to a freely given program.

The standard conjectures make the question especially sharp because they separate a named vision from many named acts of construction. Grothendieck formulated the package. Kleiman wrote the great contemporary exposition. Others proved cases and consequences. Later mathematicians would redefine motivic categories under relations for which unconditional theorems were available. The program persisted through departures from its founder’s preferred route.

What “standard” leaves open

The short Bombay article is sometimes described as prophetic. Prophecy is a dangerous category for mathematics. It praises foresight while hiding the exact propositions by which foresight can be judged. Grothendieck did more than predict that algebraic cycles would matter. He specified algebraicity of $\Lambda$, Künneth projectors, equivalence of homological and numerical relations, and a positivity condition. The content can be tested.

The propositions also delimit ignorance. One knows hard Lefschetz in certain realizations; one does not know that the inverse is algebraic in general. One knows how to define numerical equivalence; one does not know its equality with homological equivalence in general. One knows positivity from Hodge theory over $\mathbf C$; one does not possess its full algebraic analogue across characteristics. The list does not disguise conjecture as theorem.

In public reputation, the Fields Medal announced mastery. In mathematical practice, the standard conjectures announced ignorance made exact. The two announcements occurred within the same summit. There is no contradiction. The greatest command of a field may consist in seeing precisely what its existing language cannot yet prove.

That command also exposes its holder. A theorem, once proved, can be detached from the author. A conjectural program remains vulnerable to the future. Others may prove it, bypass it, rename parts, or cease to share the belief that it is the central road. Its proposer cannot control which outcome will define the historical memory.

By 1969 Grothendieck had made the standard conjectures public enough to survive him. They were connected to motives but did not depend on possession of his entire vision. They were rooted in the Weil program but not made obsolete by its later solution. They were called standard but marked a frontier.

The next conflict would concern a very different standard: what sources of support an institute devoted to pure research could accept. In one domain Grothendieck had asked whether the visible operations of cohomology descended from algebraic correspondences. In the other he would ask whether scientific freedom could be separated from the provenance of money. The analogy should not be pushed into identity. It is enough to note the repeated refusal to treat formal independence as proof of actual independence.

A map of the conjectures

Letters A, B, C, D, and I appear differently across historical and modern expositions. A reader can easily believe that the “standard conjectures” are one canonical numbered list. Kleiman’s 1968 text instead presents a web of implications under explicit assumptions. The safest history states the mathematical property before the label.

Property A concerns the hard-Lefschetz map restricted to algebraic classes: multiplication by a power of the hyperplane class should carry codimension-$p$ algebraic classes isomorphically to the complementary degree. B says the inverse Lefschetz operator is algebraic as a correspondence. C says the Künneth components of the diagonal are algebraic. D says homological and numerical equivalence agree. I is the positive-definiteness statement for primitive algebraic classes.

B implies C. B also provides algebraic adjoints and projectors needed for the correspondence algebra. Under strong Lefschetz and the index conjecture, A and D are closely related; Kleiman gives precise equivalences under the stated hypotheses. The package often called standard can be formulated as B plus I, or through equivalent combinations once background Lefschetz statements are assumed.1919Kleiman, “Algebraic Cycles and the Weil Conjectures,” 359–61, 382–83.

This web matters historically because Grothendieck was minimizing and rearranging assumptions, not compiling a list of hopes. If one property can be derived from another, the true geometric obstacle may lie in a smaller statement. If several formulations are equivalent, researchers can attack the one best suited to their methods.

Conjecture C is sometimes separated as the “Künneth standard conjecture,” even though Kleiman derives it from B. Modern usage may call an algebraicity statement C and group it among the standards. The integration editor should preserve notation local to each cited source or explicitly announce a convention. Switching labels silently creates factual error even when the underlying mathematics is right.

The same caution applies to the “Hodge standard conjecture.” It is not the classical Hodge conjecture. The adjective indicates a Hodge-index-type positivity property. In characteristic zero the Hodge–Riemann bilinear relations provide the model. In arbitrary characteristic the conjecture asks algebraic cycles to carry the corresponding sign and definiteness.

Conjecture D’s relation to a chosen cohomology theory also needs qualification. “Homological equivalence” depends on a Weil cohomology. If D holds and the standard framework connects theories, the resulting equivalence becomes intrinsic. One should not state at the outset that there is one theory-independent homological equivalence and then praise D for proving independence.

The map of implications is one reason the seven-page primary article deserves page-level study. A popular account can convey the architecture, but a precise one must match Grothendieck’s own formulation and Kleiman’s contemporary expansion, including hypotheses that later shorthand suppresses.

What was known in the late 1960s

Kleiman’s introduction surveyed cases available then. Projective spaces and other varieties with cellular decompositions satisfied the relevant Lefschetz-type statements. Curves and surfaces were under control in important respects. Abelian varieties formed a deep test case, with Lieberman’s work supplying key algebraicity results. The index theorem was known for surfaces through work associated with Hodge, Segre, Bronowski, and Grothendieck.

These successes were not a random catalog. Cellular varieties have cohomology generated by explicit algebraic cells, so algebraic projectors are more accessible. Curves have limited codimension. Surfaces possess intersection forms governed by the Hodge index theorem. Abelian varieties have group structure and rich endomorphism algebras. Each case tests a different part of the proposed architecture.

The general difficulty concentrates in higher-codimension cycles. Divisors—codimension one—benefit from Picard varieties, line bundles, and established equivalence theorems. Codimension two and beyond exhibit phenomena invisible to divisor theory. A conjecture extrapolated from divisors must survive a world where cycle groups can be enormous and subtle.

This is why phrases such as “it holds for low-dimensional examples” require caution. Dimension of the variety and codimension of cycles interact. Some standard conjectures are known for all varieties up to a certain dimension; others for special classes in arbitrary dimension; closure under products can extend cases. A mathematically reviewed version should state only cases checked against a current survey.

In 1968 the point was not to advertise a nearly solved program. Kleiman explicitly said that only limited equalities and index results were known. His exposition balanced confidence in the structure with candour about evidence. That balance is a historical fact about the school’s style at its best.

The article also records a moving frontier. A footnote notes Griffiths’s announced counterexample to a conjectured equality involving equivalence relations stronger than numerical equivalence. The correction did not destroy the standard package; it distinguished which equivalences could plausibly coincide. Mathematical programs evolve by losing overly strong hopes as well as gaining theorems.

This detail resists the myth that Grothendieck’s generality consisted in choosing abstractions immune to counterexample. The program made assertions precise enough for examples to refine them. A universal language can be empirically disciplined within mathematics.

Why positivity was the hard bridge

Algebraicity and positivity play different roles. Algebraicity says an operation visible in cohomology comes from a cycle. Positivity says a pairing on certain algebraic classes has a definite sign. One can imagine proving algebraicity without positivity, or positivity in a realization without knowing its algebraic origin. The conditional Weil argument needs them to cooperate.

Suppose $\Lambda$ is algebraic. One can construct an adjoint operation on correspondences compatible with Lefschetz decomposition. Suppose the index form is positive. Then the trace pairing $u,v\mapsto\operatorname{Tr}(u'v)$ behaves like an inner product on an appropriate algebra of correspondences. An operator $g$ satisfying $g'g=1$ acts orthogonally and has eigenvalues on the unit circle after extension to $\mathbf C$.

Normalize Frobenius on degree $i$ by dividing by $q^{i/2}$. The geometric relation between Frobenius, its transpose, and powers of $q$ makes the normalized operator satisfy the required adjoint identity. Positivity then puts its eigenvalues on the unit circle. Multiplying back yields absolute value $q^{i/2}$.2020Kleiman, “Algebraic Cycles and the Weil Conjectures,” 383–86.

This account suppresses the coefficient-field extension needed to speak of $\sqrt q$, the grading, and exact transpose formulas. It retains the logic: an analogue of a positive inner product turns a spectral estimate into formal linear algebra.

The method resembles the proof that an orthogonal or unitary matrix has eigenvalues of absolute value one. The difficulty is not that linear-algebra lemma. It is constructing the correct positive adjoint structure algebraically. Once again, the program moves difficulty from a famous statement to a foundational geometric source.

Deligne’s later proof established the weight estimate through a different network of ideas, including monodromy, Lefschetz pencils, and induction. It produced the needed positivity-like control without proving that all the operators and pairings in Grothendieck’s proposed route arose from algebraic cycles. This is why one proof could settle Weil and leave the standard conjectures open.

The distinction also has motivic consequences. Deligne proved a statement about realizations and weights. The standard conjectures would assert that the structure descends to algebraic correspondences and becomes independent of realization. A correct numerical answer can precede the universal explanation sought by the motivic program.

Conjectures as coordination

A conjecture coordinates research by doing more than predicting truth. It selects examples, suggests intermediate lemmas, and creates a vocabulary for partial progress. The standard conjectures told geometers that algebraic cycles, Lefschetz operators, and positivity were not peripheral technicalities but a connected foundation.

They also coordinated people. Kleiman could write a systematic exposition. Lieberman could establish a class of cases. Bombieri could identify the formal consequence independently. Specialists in abelian varieties, surfaces, and cohomology could locate their results within the same map.

Such coordination depends on trust. A young mathematician may spend years on a conjecture because a program’s originator and community judge it central. Grothendieck’s prestige amplified that trust. The ethical responsibility of a conjecturer is not to guarantee truth—no one can—but to state evidence and implications plainly. The contemporary exposition’s explicit limits served that responsibility.

Prestige can also narrow a field if one program monopolizes attention. Jackson records that Grothendieck’s school dominated algebraic geometry strongly enough for some other approaches to feel displaced. The standard conjectures’ long life may reflect their depth; it may also reflect the institutional power that made them standard. These explanations are not exclusive.

Historical evaluation should therefore separate mathematical fertility from sociological dominance. The conjectures produced concepts and results even where unsolved. That is mathematical evidence of fertility. The concentration of visitors and seminar attention at IHÉS is evidence of dominance. Neither alone proves the other deserved or harmful.

The term standard itself coordinates by naming a package as normal expectation. Once the name is accepted, a new cohomology theory can be tested against it. Naming creates a research checklist. It can also make dissent appear deviation. A historian should ask whether alternatives existed without inventing a suppressed school where sources do not show one.

For motives, the coordination was indispensable. Different equivalence relations yielded different categories. The standard conjectures predicted bridges. Researchers could define Chow motives or numerical motives unconditionally while keeping track of what would follow if D or B held. Conditional architecture allowed work before final foundations.

Afterlife without closure

The later history divides the package into proved cases, reductions, and still-open statements. Algebraicity of the Lefschetz involution is known for many important classes. Positivity is established in settings where comparison to characteristic zero or other methods apply. Numerical motives have the semisimplicity Jannsen proved. No single sentence can inventory the current frontier responsibly.

For this biography, the durable point is that the conjectures remain mathematically meaningful after their original arithmetic target was reached. They ask whether the symmetries seen by cohomology are geometric and whether numerical intersection contains all homological information. Those questions concern the ontology of motives, not only zeta functions.

The open status changes how Grothendieck’s foresight should be praised. One cannot say he saw a theorem that later mathematicians confirmed in full. One can say he isolated statements that continue to connect central structures and proved powerful consequences from them. Foresight appears in the quality of the questions, not retrospective truth alone.

The unresolved conjectures also place a limit on biographical metaphor. It is tempting to compare Grothendieck’s departure to an inverse operator not known to be algebraic, or the institute’s ethics to numerical equivalence. Such wordplay adds no evidence and risks trivializing both mathematics and conflict. This chapter’s final analogy—formal independence versus actual origin—remains a rhetorical transition, not a deduction.

The seven pages from Bombay survive because they made a frontier portable. A mathematician did not need to attend the April motives lecture or share Grothendieck’s philosophy to know what B or D demanded. That portability was one of the summit’s greatest institutional successes.

It also meant the program could continue without him. When the door at IHÉS closed, the conjectures did not follow him out. They had entered common mathematics, where their future would be decided by proofs no founder could command.

Independence from the prime

One of the Weil program’s recurring demands was independence of the auxiliary prime $\ell$. Étale cohomology uses coefficients built from a prime $\ell$ different from the characteristic $p$ of the base field. The construction appears to introduce an arbitrary choice into invariants expected to belong to the variety.

For a variety over a finite field, the characteristic polynomial of Frobenius on $H^i_{\mathrm{\acute et}}(X,\mathbf Q_\ell)$ should have rational or integral coefficients independent of $\ell$. The zeta function’s point counts are intrinsic, but its cohomological factorization separates degrees. Without algebraic Künneth projectors, independence of the total alternating product does not automatically give independence factor by factor.

Conjecture C supplies algebraic projectors onto degrees. An algebraic correspondence has intersection-theoretic traces that are not tied to one $\ell$. The projectors therefore help show that each factor is intrinsic rather than an artefact of a chosen realization. This is another reason motives and standard conjectures belong together.

The issue illustrates Grothendieck’s intolerance of unexplained auxiliaries. A useful construction may choose $\ell$; the final object should reveal why results do not. The standard conjectures locate the explanation in algebraic cycles.

Later work proved broad independence results by methods not requiring the full conjectures. That success does not make the original demand misguided. It separates the statement of independence from one proposed geometric cause.

In general-reader prose, “independence of $\ell$” should not be rendered as “independence from prime numbers.” It concerns invariance across coefficient systems indexed by auxiliary primes. The base field’s characteristic remains central.

The same distinction appears in motives. Different $\ell$-adic realizations should come from one motivic object. Agreement among characteristic polynomials is evidence for common origin, not a definition of the origin.

What a conditional proof accomplishes

Mathematicians sometimes disparage conditional proofs because the hypothesis may be as hard as the conclusion. In Grothendieck’s program, conditional derivation had several independent values.

It tested sufficiency. If B and I yielded the Frobenius estimate through a correct formal argument, then the conjectures captured enough geometry. A proposed foundation that did not recover Weil would be inadequate.

It localized difficulty. Rationality, functional equation, integrality, semisimplicity, and absolute values could be assigned to different inputs. Researchers could see which obstacle was algebraic and which formal.

It unified cases. A proof of standard conjectures for a class of varieties would automatically produce Weil consequences for that class without a new point-count argument each time.

It connected programs. The same hypotheses improved the category of motives and explained zeta functions. This economy suggested that the conjectures were not ad hoc devices invented for one target.

It generated diagnostics. If a stronger equivalence conjecture failed, one could ask whether the conditional route actually needed it. Kleiman’s map of implications made revision possible.

Conditional proof also disciplined rhetoric. Grothendieck could not claim the Weil conjectures proved. He could claim that a specific algebraic-cycle package would prove them. The difference was public and checkable.

Deligne’s later unconditional proof changed the value of the conditional route without erasing it. The route ceased to be necessary for establishing the target theorem. It remained explanatory and motivic.

This is a recurring pattern in mathematics. A conjectural structure may first justify itself by yielding known or desired results. Later, those results receive independent proofs. The structure then survives only if it organizes more. The standard conjectures did.

What the explanations leave out

The mathematics in this chapter has used symbols because some distinctions are shorter in notation. It has also omitted categorical and cohomological machinery that specialists need. The three-layer method requires stating where intuition stops.

The ladder image for hard Lefschetz hides scalar factors and primitive decomposition. The prism image for projectors hides algebraicity. The inner-product image for positivity hides signs and degrees. The orthogonal-matrix image for Frobenius hides transpose correspondences and coefficient embeddings.

These omissions are legitimate if marked. They become errors when a metaphor is used to infer a theorem. A specialist window in the final book should give exact definitions and source locators; the main narrative should preserve the problem each formula solves.

Historical attribution needs the same layered method. General readers may first learn that Grothendieck formulated the conjectures. A second layer names Kleiman, Lieberman, Lubkin, and Bombieri. A precision layer maps exact propositions and proofs.

The page should not overwhelm a reader with every letter label before the central problem is clear: cohomology sees symmetries and positivity that the standard program asks geometry itself to supply.

That problem joins his mathematical style to this moment without psychologizing. He preferred intrinsic structure to unexplained coincidence. The standard conjectures are documents of that preference in exact form.

Their seven printed pages also impose humility on narrative proportion. The book may spend thousands of words explaining what the paper compressed. Explanation is not evidence that Grothendieck said all these words. Grothendieck’s paper establishes the program and its bibliographic identity; exact technical statements in this chapter follow Kleiman’s contemporary exposition or later mathematical sources. No direct quotation from Grothendieck’s paper appears.

A biography can make hard mathematics readable without making the biographer a coauthor of the conjectures.

Why seven pages were enough

The shortness of Grothendieck’s paper reflects a mature specialist network. Terms such as algebraic cycle, correspondence, Weil cohomology, Lefschetz operator, and numerical equivalence already carried bodies of literature. Readers at the Bombay colloquium could supply them. Kleiman’s much longer exposition shows what compression presupposed.

Compression also expressed program rather than textbook. The paper’s purpose was to formulate and connect conjectures, not teach étale cohomology from first principles. Its authority depended on exact placement of a few assertions inside machinery built elsewhere.

This form creates a biographical danger. A seven-page paper can look like a casual announcement and its consequences be attributed to charisma. In mathematics, shortness often means the author has identified a small interface controlling a large system. The intellectual work lies in choosing the interface.

The reverse danger is canonization. Once later readers know the conjectures were influential, every compressed sentence can be treated as revelation. The scan must be read historically, beside Kleiman, not as scripture.

The article’s venue matters. Proceedings place a statement inside a meeting and a field’s current conversation. They can circulate less widely than a journal and preserve less of oral context. The Grothendieck Circle’s hosted scan improves access but does not change copyright or supply missing discussion.

The notes cite the printed pages 193–199 rather than only the website. The website is provenance for the consulted scan. Bibliographic identity belongs to the Oxford proceedings.

The chapter’s extended explanation is thus commentary on a compact primary source. Every technical claim must resolve either to that paper, Kleiman’s contemporary account, or a modern verification source. The narrative should never imply that Grothendieck used the metaphors introduced here.

Seven pages were enough to make the conjectures public. They were not enough to make them true. The distinction is the intellectual drama of the chapter: formulation reached completion where proof did not.

A durable frontier

The standard conjectures remained attached to Grothendieck’s name because they continued to mark an exact boundary. A boundary is durable when work on either side changes while the missing passage remains recognizable.

On the proved side stood expanding cohomology theories, cases of Lefschetz algebraicity, comparison theorems, and eventually Deligne’s Weil proof. On the conjectural side stood general algebraic projectors, positivity, and equality of equivalences. Motive categories could be built under different relations while the hoped-for identifications stayed open.

This arrangement made the conjectures a reference point even for mathematicians not attacking them directly. A theorem could be described as evidence for B, a consequence of D, or a result unconditional under numerical equivalence. The frontier organized language.

Its durability also revised the meaning of Grothendieck’s summit. He had not only solved problems and built tools; he had drawn boundaries future work would still recognize. That achievement is different from proof and should be praised differently.

The chapter has therefore avoided “almost proved Weil.” The conditional route was profound, but the missing conjectures were not technical residue. Deligne’s later road confirms that distinction.

It has also avoided “failed grand program.” A program that produces precise cases, alternative constructions, and lasting questions has mathematical content independent of total completion.

The fairest final verb is formulated. Grothendieck formulated a standard by which algebraic geometry could ask whether cohomological structure was truly algebraic. The answer remains distributed among theorem, case, and conjecture.

Formulation carried reputational risk. A theorem enters the record with proof; a conjecture can remain open, be bypassed, or turn false. Grothendieck attached his authority to claims stronger than what his machinery had established. Their exactness made future correction possible.

That risk distinguishes the program from retrospective myth. A vague prophecy can always be declared fulfilled. B, D, and the positivity statement have definite content and status. Their survival is mathematical, not rhetorical.

The summit thus included an act of exposure. The field’s leading architect published where his architecture still depended on unbuilt supports. The honesty of the boundary is part of the achievement.

When institutional conflict arrived, those supports remained common tasks. Leaving IHÉS could interrupt direction, but it could not retract conjectures already stated for anyone to test.

That public test is the final contrast with reputation. Authority brought attention to the conjectures; only proofs, counterexamples, and reductions could determine their mathematical future. Grothendieck had used the summit to make claims that no summit could settle.

The conjectures therefore belong in a life story not because biography can solve them, but because they show the form of his ambition at the moment of greatest authority. He did not present one more difficult problem. He proposed conditions under which several theories would disclose one geometry. Their unresolved status preserves the scale of that proposal better than retrospective triumph could.

The frontier remained open, exact, and shared by mathematicians who no longer needed its founder beside them.