1. A Man Overboard

The story usually arrives already finished. The Pythagoreans believed that everything was number. One of their own discovered that the side of a square and its diagonal could not be measured by any common unit. He disclosed what the brotherhood needed to conceal. His companions threw him from a ship.

It is an unusually efficient piece of history. A creed produces a discovery; the discovery produces a conflict; the conflict produces a punishment. Remove any link and the story falls apart. That very neatness should make us hesitate. Events rarely arrange themselves so considerately for the people who will tell them later. The parts we can establish are much less complete than the story suggests.

No early surviving source records Pythagoras himself proclaiming that everything is number. Aristotle discusses people he calls the Pythagoreans, and what he attributes to them is not simply the modern assertion that every measurable magnitude is a rational number. Nor do we possess an early account naming Hippasus as the discoverer of incommensurability. Later traditions connect mathematical disclosure, Hippasus, and death at sea in different combinations; they do not give us a contemporary report of murder aboard a boat.

The familiar foundational crisis belongs to another layer again. A particularly influential formulation appeared in a 1928 study by Helmut Hasse and Heinrich Scholz, though earlier historians had already advanced related ideas. The named discoverer, the shattered foundations, and the murderous brotherhood did not enter the record together. Treating their eventual conjunction as one ancient testimony manufactures evidence that we do not have.

Yet underneath these additions lies something exceptionally secure. The side and diagonal really are incommensurable. By the fourth century BCE, Greek writers knew a proof and could assume that educated readers recognized the example. Plato and Aristotle support this mathematical layer. The identity, emotions, and fate of the first discoverer stand on a very different evidential footing.

Separating those layers is the purpose of this essay. The question is not whether a gripping story ought to be forbidden. It is what changes when a story's architecture is mistaken for the architecture of the evidence. Who supplied the protagonist? Who supplied the catastrophe? How much later were these additions made, and what did they help their authors explain?

In the preceding essay, a claim could often be set beside a clay tablet: something that could be weighed, photographed, measured, and assigned a museum number. Here we seldom have anything comparable. We must ask when a surviving statement was composed, how it reached us, and how far its author stood from the event described. Dates become evidence about evidence.

This is a deliberately unglamorous procedure. A late witness is not automatically false; a lost earlier source may lie behind it. But the interval allows more opportunities for selection, rearrangement, and retrospective attribution. A narrative that grows more vivid as it recedes from its subject requires explanation, not gratitude for the extra detail.

The previous essay ended with a child smoothing a practice tablet and beginning again. The writing could disappear while the object survived to be excavated thousands of years later. This episode begins in a world where the surfaces themselves generally disappeared. We are not merely missing the decisive sentence. Often we are missing the very kind of object on which such a sentence might have been written.

2. Names Without Objects

Pythagoras, Hippasus, Theodorus, Theaetetus, Eudoxus: behind these names we possess no mathematical document that can be identified as written in its author's own hand. Not a damaged original awaiting decipherment, but no securely identified original at all. Part of the explanation is material.

A wax tablet was an excellent working surface. A shallow recess in wood or ivory held wax; a pointed stylus made marks, and its flattened end erased them. Several leaves could be fastened together. Such an object was designed for repeated use, which also made it an efficient instrument for destroying its own history. Surviving examples vary greatly in size. Later ivory writing tablets in museum collections demonstrate the durability of the format, not the precise contents of a fifth-century mathematician's desk. A specimen roughly thirty-four by sixteen centimeters should not quietly become Pythagoras's notebook.

Papyrus creates a different problem. It survives remarkably in dry Egyptian conditions and poorly in many of the wetter environments of Greece and southern Italy. The geographical distribution of surviving Greek documents therefore does not reproduce the geographical distribution of Greek writing. The place where an activity flourished and the place where evidence can survive need not coincide.

Dust or fine sand offered an even more disposable surface for drawing and calculation. A demonstration could disappear at the sweep of a hand. Such practices are known from ancient traditions, although assigning a particular apparatus to a particular early mathematician is another matter. The broader point does not depend on reconstructing anyone's furniture: intense mathematical activity can leave a very thin archaeological trace.

That absence is easy to misunderstand. A surface inexpensive enough to erase and reuse is especially good for exploratory work. It is especially bad for giving future historians something to find. The archaeological silence measures conditions of preservation as much as the amount of mathematics performed.

Transmission accounts for the other half of the difficulty. The securely dated complete Greek manuscript of Euclid's thirteen books known as Bodleian MS D'Orville 301 was copied in Constantinople in 888 CE. Stephen the Clerk produced it for Arethas of Patras. More than eleven centuries separate that act of copying from Euclid's conventional date, around 300 BCE.

Earlier witnesses are fragmentary. The Oxyrhynchus papyrus preserving part of Book II, including a proposition and diagram, has been dated by paleographical arguments to the early centuries of the Roman Empire; some assessments place it around the late first or early second century CE. Even the earlier dating leaves several centuries between fragment and author. Dating a hand is also not the same as reading a date supplied by its scribe.

A manuscript family tree can identify relationships among surviving witnesses. It cannot count all the copies once made between antiquity and the present. Most branches are gone. The surviving leaves permit reconstruction, but they do not restore every vanished intermediate stage.

Each stage afforded opportunities for change: an omitted line, a duplicated phrase, a redrawn diagram, an explanation promoted from margin to text, a proof replaced by an easier proof. None of this means that the transmitted mathematics is worthless. It means that the date of a mathematical idea and the date of its surviving wording must be distinguished.

The route was not one straight relay. Greek copying continued through the editorial tradition associated with Theon in the fourth century CE and into Byzantine manuscripts. Arabic translations and revisions associated with al-Hajjaj, Ishaq ibn Hunayn, and Thabit ibn Qurra developed alongside it. From the twelfth century, Arabic materials entered Latin through translators including Adelard of Bath. Renaissance editors then compared Greek and Latin traditions before print and modern criticism created further layers.

There are medieval Greek witnesses on the scale of hundreds, but ancient papyrus witnesses on the scale of fragments. Abundance at the late end does not eliminate scarcity at the early end. Mesopotamia and Greece thus confront us with nearly opposite imbalances: many objects whose makers are unknown, and many celebrated names without the objects their hands touched.

Neither contrast should be exaggerated. Cuneiform scribes sometimes signed texts, while Greek attributions can be late and unreliable. An anonymous tablet at least belongs to its excavated material world. A reassuring name in a book may have been attached centuries after the work. Familiarity is not provenance.

3. The Earlier the Witness, the Less Familiar the Mathematician

The three surviving full ancient lives of Pythagoras were written under the Roman Empire, roughly seven or eight centuries after his death. He left no securely identifiable writings, and we have no continuous contemporary biography. When the testimonies are arranged chronologically, the familiar schoolroom portrait begins to dissolve.

Xenophanes, near the end of the sixth century BCE, mocks a man who hears a beaten puppy and recognizes a dead friend's soul in its cries. The subject is transmigration, not geometry. Heraclitus, around 500 BCE, attacks Pythagoras's accumulation of learning. That tells us that Pythagoras was famous as a seeker or collector of knowledge; it does not establish a theorem. Herodotus, writing in the fifth century, associates Pythagorean practice with ritual restrictions.

Plato's explicit reference to Pythagoras concerns a distinctive way of life maintained by followers. It supplies no proof of the theorem bearing his name, no experiment on a monochord, and no discovery of irrational magnitudes. Aristotle discusses Pythagoreans extensively in connection with number, harmony, and cosmology, but is much less willing to assign mathematical achievements directly to the founder.

As the sources become later, the mathematician becomes more accomplished. Imperial biographies give him discoveries, demonstrations, teaching programs, and experiments. Walter Burkert made the direction of this growth central to his study, published in German in 1962 and in revised English form in 1972. His point need not be reduced to the claim that Pythagoras knew no mathematics. It concerns how mathematical achievements accumulated around a founder's name.

One much-disputed passage survives in Proclus's fifth-century CE commentary. Proclus drew on an earlier history of geometry associated with Eudemus, a pupil of Aristotle. His account credits Pythagoras with transforming geometry into a liberal education and with investigations involving irrational magnitudes and regular solids. Burkert argued that the attributions cannot simply be backdated to Eudemus: the passage contains layers, and later traditions may have supplied precisely the claims modern readers most want to extract.

Leonid Zhmud and other scholars have resisted so skeptical a reconstruction. Their objection is methodological as well as substantive. When the loss of sixth-century material is overwhelming, silence cannot be treated as proof that scientific work did not occur. Between the early traditions surrounding Thales and the demonstrative geometry of later fifth-century figures, substantial development plainly took place. A blank in the archive is not necessarily a blank in history.

The disagreement matters because both cautions are legitimate. Founder-attribution requires scrutiny; absence of surviving documentation does not establish absence of activity. We should neither turn every late report into fact nor convert an austere evidential standard into certainty that nothing happened.

A thinner outline remains defensible. Pythagoras acquired an extraordinary reputation for wisdom. A named way of life persisted. He was associated with Samos, moved to Croton in southern Italy, and ended his life in the region of Metapontum. Conventional dates for these moves are reconstructed from synchronisms and later accounts, not recovered from civil registration. Their apparent precision can be misleading.

The same warning applies elsewhere. Hippasus's period of activity remains disputed. Theodorus's chronology depends heavily on the age assigned to him in Plato's dramatic setting. Theaetetus is often dated approximately 415–369 BCE, but the military episode framing his death has competing interpretations. Euclid has a conventional period of activity around 300 BCE, not a securely documented birth and death.

Descriptions of five-year silences, admission tests, teaching behind a curtain, and communal property are fullest in Iamblichus, writing around 300 CE. They cannot be imported unqualified into the early community. One more tangible dimension of that community is its political vulnerability.

Later traditions preserve conflicts in southern Italy, often reconstructed as more than one episode: hostility surrounding Pythagoras's departure from Croton and subsequent attacks on Pythagorean associations. A tradition attributed to Aristoxenus describes a gathering in Milo's house being burned. Polybius gives a broader account of meeting places destroyed across cities, followed by killings and political disorder.

Exact dates, casualties, and survivor lists remain insecure; the supposedly burned house has no accepted archaeological identification. What survives supports a politically consequential network exposed to violence, not a modern university quietly discussing numbers. Croton, Metapontum, and Tarentum were competitive communities with their own social conflicts. A private association intertwined with political influence is a safer description than either a modern party or a monastery. The names lost in that violence are a historical absence of a different kind.

4. A Classroom

The closest thing to a scene from mathematical life is not a biography. It is a classroom recollection embedded in Plato's Theaetetus, composed in the fourth century BCE. The young Theaetetus recalls Theodorus drawing demonstrations about the sides of squares: the square on three units, the square on five, and further cases up to seventeen, where the lesson stopped.

Afterward, Theaetetus and another boy tried to collect the apparently endless examples into a classification. The scene gives the inquiry a remarkably ordinary texture. A teacher proceeds case by case. Students look for a unifying distinction. The discovery is not presented as a forbidden object whose disclosure threatens their lives.

Plato gives Theaetetus a family, an Athenian deme, a recognizable appearance, and exceptional mathematical promise. The dialogue's framing story places him on the road back from campaigning near Corinth, injured and ill. Theodorus appears as an older teacher from Cyrene. These are literary presentations, not stenographic records, but they belong much nearer the people described than the imperial biographies belong to Pythagoras.

Neither figure is identified in our early evidence simply as a Pythagorean. Automatically enrolling everyone who studied incommensurability in that school expands a label until it ceases to explain anything. The scene establishes an intellectual practice more securely than an institutional affiliation.

Why did Theodorus stop at seventeen? Many reconstructions have been proposed. Some invoke arithmetic remainders; others arrange successive right triangles into a spiral, making seventeen a suggestive stopping point. Another possibility is that Plato supplied no recoverable clue to the exact method. A reconstruction's elegance cannot make it eyewitness evidence. The text tells us the range of examples, not the complete procedure.

The omission of the diagonal of a unit square is usually taken to suggest that this elementary case was already familiar. Aristotle provides firmer evidence for its proof. In the Prior Analytics he invokes the incommensurability of side and diagonal as an example of reductio: assume the contrary and arrive at an impossible consequence involving odd and even.

Here is a modern presentation of that argument, not a claim to reproduce the discoverer's wording. Suppose a common measure exists. The side and diagonal would then consist of whole numbers of copies of that measure. Let those integers be s and d, and cancel common factors until the pair is in lowest terms. The square's geometry gives d² = 2s².

It follows that d² is even. Since the square of an odd integer is odd, d must itself be even. Write d = 2k. Substitution gives 4k² = 2s², hence s² = 2k². The same reasoning makes s even. Both integers therefore share a factor of two, contradicting the choice of a pair with no common factor. The assumed common measure cannot exist.

Notice what has not happened. Nobody has tried increasingly fine rulers and failed to obtain a satisfactory fit. Nobody has calculated a sufficiently long decimal expansion. The argument rules out every possible common measure at once. Failure to find something has become proof that there is nothing of the required kind to find.

Aristotle supplies the proof's logical skeleton, not every step of this modern reconstruction. A related complete demonstration appears as Proposition X.117 in part of the manuscript tradition of the Elements. Heiberg treated it as an interpolation rather than as part of Euclid's original text. The relationship to Aristotle's example is important, but it does not entitle us to project the surviving wording unchanged into the fifth century BCE.

The original conceptual language also matters. Incommensurable means lacking a common measure. It concerns a relation between magnitudes. It is not yet the modern classification of a real number by its decimal expansion, and its most straightforward Greek terminology carries no necessary implication of a religiously forbidden secret.

The Old Babylonian tablet with its extremely accurate approximation to the square diagonal, discussed in the preceding essay, answers a different question. It offers a numerical value useful for calculation. The Greek proof establishes the impossibility of an exact ratio of integers. More digits, however impressive, do not automatically become a proof of impossibility.

This is not a contest in which accurate computation loses to abstract reasoning. The questions differ. How can we calculate this length? Does an exact common measure exist? The same square can sustain both inquiries, and the surviving evidence should not be forced to answer a question its maker never claimed to ask.

5. The Shame of Not Knowing

If incommensurability provoked a crisis, one might expect to find some trace of the reaction in fifth- or fourth-century evidence. Plato's Laws contains a passage that initially looks promising. The Athenian speaker says that he learned embarrassingly late about commensurability and incommensurability among magnitudes. His language about widespread Greek ignorance is emphatic and deliberately insulting.

The direction of the emotion is decisive. He is ashamed not because incommensurability exists but because he did not know about it. The scandal is ignorance. The speaker does not propose concealing a discovery whose consequences are unbearable; he wants young people taught the subject.

The surrounding discussion reinforces that direction. Such knowledge is described as appropriate to education, not inherently dangerous or inaccessible. Problems may be set playfully. A topic later narratives surround with secrecy appears here as something children should learn and educated adults should be embarrassed to have missed.

We should not exaggerate the inference. A fourth-century educational recommendation cannot establish what every earlier Pythagorean felt. It does, however, resist being recruited as evidence for the opposite attitude. A source praising instruction is not testimony that knowledge was suppressed.

Aristotle is calmer still. The side-and-diagonal proof serves the practical purposes of a discussion of logical form. It is a standard example, sufficiently familiar to be mentioned without a full exposition. There is no narrated panic, no institutional breakdown, and no appeal to divine punishment.

Set the securely attested activities beside one another. Theodorus proves individual cases. Theaetetus classifies them. Plato makes knowledge of them part of an educational ideal. Aristotle uses a proof as an illustration of reasoning. These witnesses do not provide a complete chronology of discovery, but they do show the subject being taught, generalized, and absorbed.

Euclid's Book X extends that work into an enormous classification. In the standard count it contains 115 propositions, almost a quarter of the Elements' 465 propositions and more than any other individual book. These numbers vary when later additions are counted, which is one reason the textual qualification matters.

Book X is not a chapter about nonterminating decimals. It distinguishes commensurability in length from commensurability in square and studies magnitudes relative to a chosen reference. It constructs and identifies kinds of irrational lines and establishes relations among them. Much of its labor consists precisely in refusing to treat all troublesome magnitudes as one undifferentiated remainder.

The scale of the work suggests cumulative research. It would be misleading to imagine that Euclid created every part unaided. How individual contributions should be assigned is debated, but the finished book testifies to sustained technical interest, not a decision to avoid the phenomenon.

This is where the argument must remain proportionate. Silence cannot prove that nobody was ever shocked. Nor does eventual incorporation into a curriculum prove that an earlier crisis was impossible; mathematics has often taught material that once caused deep controversy. The point is narrower: the surviving early evidence does not establish the psychological or institutional catastrophe that the popular story confidently supplies.

Compare the atmospheres without collapsing their dates. In one set of sources there are diagrams, definitions, classifications, lessons, and educational games. In much later narratives there are secrets, expulsions, symbolic graves, divine anger, and drowning. The second set may preserve traditions worth investigating, but it cannot simply overwrite the first.

A discovery may be momentous without producing the particular crisis a historian expects. Incommensurability changed the available mathematical relations and encouraged new techniques. That structural importance is not itself evidence of collective terror. To infer the terror from the importance is to supply an ancient emotional response from a modern estimate of what ought to have been upsetting.

6. Several Stories, Not One

The Hippasus legend is better understood as a cluster of motifs than as six copies of a single report. In authors writing under the Roman Empire, secrecy, unsuitable recipients, particular geometric discoveries, personal credit, and death circulate in combinations that do not remain fixed.

Plutarch, in the late first or early second century CE, refers to Pythagorean reluctance to publish secrets and to divine displeasure when hidden geometrical knowledge was disclosed to the unworthy. This does not furnish the name of a discoverer or an account of companions committing murder. The chronological distance from an early Pythagorean event is already several centuries.

In Clement of Alexandria, around the end of the second century, Hippasus is associated with revealing the construction involving twelve pentagons and with a fate at sea. Iamblichus, writing around 300, offers several versions. In one, Hippasus makes public the construction of the figure formed from twelve pentagons and suffers the impious person's maritime fate. The credit for the discovery nevertheless belongs, in this account, to Pythagoras.

Elsewhere, the first person to disclose the theory of commensurable and incommensurable magnitudes is unnamed. The community rejects him and erects a tomb, treating him as symbolically dead. That is not the same event as an actual drowning. Another passage juxtaposes the disclosure of the inscribed dodecahedron with a report that others assigned the same punishment to the disclosure of incommensurability.

A commentary on Book X associated with Pappus, preserved through a later transmission, also links disclosure of irrational magnitudes with drowning. Here too the relation between an original historical event, a moralized tradition, and the transmitted text requires care.

A plausible model is that a general story about secret knowledge and divine punishment acquired more specific mathematical contents. One version linked Hippasus, the dodecahedron, and death at sea; another linked an unnamed discloser, incommensurability, and exclusion. Juxtaposition made combination easier. A later reader could identify two figures who had never been explicitly identified in the earlier wording.

That is a model of how the surviving evidence can be related, not a demonstrated linear genealogy. Different authors may have drawn independently on a lost source. A sequence of surviving dates does not establish a sequence of inventions. The very caution used against the popular legend must also govern an account of how the legend grew.

Four variables deserve to remain visible. What was revealed: geometry generally, the dodecahedron, or incommensurability? Who revealed it: Hippasus or someone unnamed? What happened: threatened disaster, exclusion, a symbolic burial, or death at sea? What was wrong with the act: impiety, indiscretion, disclosure to outsiders, or claiming personal credit for something reserved to the founder?

The last possibility is especially revealing. In a tradition that assigns every important discovery to Pythagoras, independent credit can itself become a transgression. The punishment then expresses a late community's account of authority, not necessarily the consequences of an early mathematical contradiction.

Etymology supplies another tempting shortcut. Greek words connected with the unsaid or the unutterable can carry religious associations; mathematical terminology has its own history of overlapping senses. But the existence of those associations does not prove that incommensurable magnitudes were named as forbidden secrets. The ordinary meaning of lacking a common measure remains a mathematical relation. A theological story cannot be recovered merely by translating a suggestive word.

Hippasus himself is not thereby erased. Aristotle's Metaphysics names him alongside Heraclitus in connection with fire as a first principle. Another tradition, attributed to Aristoxenus, associates him with differently proportioned bronze discs and musical intervals. Accounts of such experiments have been assessed for physical plausibility, but a feasible apparatus still does not verify every transmitted detail.

His origins, affiliations, and writings remain disputed. A historical natural philosopher stands behind the name; the detailed discoverer-and-victim biography is another matter. That is why disentanglement is harder than simply exposing a fictional person. The name is real enough to anchor later additions. What must be tested is each attachment, not merely whether its bearer once existed.

7. When the Crisis Acquired a Date

The modern foundational-crisis narrative has a documentary history of its own. Paul Tannery in 1887 proposed that early Pythagorean geometry assumed universal commensurability and characterized the discovery of its failure as a logical scandal. Hasse and Scholz's 1928 study gave the crisis model an especially systematic form.

Its architecture is recognizably modern. Begin with an arithmetic worldview in which measurable relations are ultimately controlled by integers and their ratios. Introduce a magnitude that cannot be expressed as such a ratio. Interpret the result as the collapse of the foundation. Then present the general theory of proportion associated with Eudoxus as the repair.

Belief, disaster, crisis, reconstruction: the sequence is powerful partly because twentieth-century mathematicians had a crisis of foundations close at hand. Problems concerning infinity, consistency, and the legitimacy of mathematical methods supplied a vocabulary through which antiquity could be read. That context does not automatically refute the reconstruction, but it explains why the choice of framework requires independent testing.

Wilbur Knorr challenged the move from what ought to have happened to what the sources show did happen. One can imagine being a Pythagorean and feeling devastated. An imagined reaction is not a recovered historical reaction. Nor is the technical inadequacy of one proposed framework necessarily evidence that working mathematics became paralyzed.

The record of fifth- and fourth-century geometry does not display the clear interruption the strongest crisis account would lead us to expect. Work associated with Hippocrates of Chios and Archytas continued to employ proportional reasoning. The historical reconstruction of their methods remains difficult, but the burden lies with anyone who claims that the discovery disabled mathematical practice.

Burkert's criticism attacks an earlier premise. Aristotle's account of the Pythagoreans does not establish the explicit axiom that every ratio of lengths equals a ratio of integers. A number ontology, in which numbers are pluralities composed of units and structures are understood numerically, is not automatically the modern doctrine that every magnitude is rational.

To make incommensurability refute the supreme creed therefore requires a translation before the refutation even begins. The historian first converts a complex ancient view into a modern universal assertion about rational ratios. Only then can the diagonal become its decisive counterexample. The absence of a source for that first conversion weakens the whole drama.

Late stories of disclosure and punishment do not repair the gap. Their subject is often religious conduct, community boundaries, or entitlement to credit. Translating divine anger into a philosophical crisis is itself an interpretation. It should not be presented as the ancient story's plain content.

The association with Hippasus has its own modern landmark. Kurt von Fritz's 1945 paper, The Discovery of Incommensurability by Hippasus of Metapontum, developed an influential reconstruction involving the regular pentagon and the dodecahedron. Its mathematical ingenuity does not supply a missing ancient sentence explicitly joining all the proposed steps.

It would nevertheless be wrong to say that von Fritz invented the attribution in 1945. Earlier scholarship had discussed Hippasus, and ancient disclosure traditions made him a candidate. The paper gave an already available hypothesis a prominent and detailed formulation. Publication dates can locate influential interventions without becoming fictitious birth certificates for every idea they contain.

Later surveys could compress the reconstruction into a declarative fact. Popular retellings went further, converting death at sea or divine punishment into a crime scene: the discovery aboard ship, a verdict from the master, companions throwing the offender overboard. No secure early account supports that precise scene.

The chronological comparison is thus not a schedule of proven inventions. It is a map of evidential distance. The mathematics is securely attested by the fourth century BCE. Imperial sources narrate secrecy and punishment centuries later. Modern scholars formulate foundational crises and reconstruct discoverers. Popular accounts fuse the layers.

Jens Høyrup and other historians have examined such frameworks as objects of historiography in their own right. That change of focus is useful. A narrative once employed to explain Greek mathematics now also needs explanation as a product of modern historical writing. The discovery remains important when the crisis becomes a question rather than a fact.

8. Absorption Without a Proven Catastrophe

Incommensurability was not removed from mathematics. A way was found to reason rigorously in its presence. Book V of the Elements preserves a general theory of proportion conventionally associated with Eudoxus, though Euclid's text does not sign the definitions with his name.

Its central move is to avoid demanding a common measuring unit. It also avoids treating every ratio as a number. To decide whether A:B and C:D are equal, compare integer multiples. For any positive integers m and n, compare mA with nB and mC with nD. The two ratios are equal when the greater-than, equal-to, or less-than relation always agrees.

The order of the quantifiers matters. A few successful comparisons do not suffice. The definition concerns every allowed pair of integer multiples. It does not tell us to find a ratio's numerical value; it tells us what must hold for one ratio to be the same as another.

This can feel cumbersome because it exchanges a desired object—a single numerical expression—for a complete pattern of relations. The gain is generality. Commensurable and incommensurable magnitudes fall under the same criterion. A common unit is no longer the admission ticket to proportional reasoning.

The procedure should not be described as a practical instruction to perform infinitely many tests. It is a mathematical definition: a universal condition to be established by proof. One can prove that it holds without physically running through an endless list. That distinction keeps the ancient theory from being confused with an impossible experimental program.

Did Eudoxus invent it in order to rescue a shattered system? We have no surviving personal statement explaining that motive. Functionally, the theory handles incommensurable magnitudes. Historically, they form an important background to the need for a more general account. Those claims are stronger than a confident account of his inner intentions.

Knorr's reconstruction emphasizes development rather than dramatic rescue: earlier methods, including approaches involving successive subtraction, encountered technical limits; classification and proportional theory acquired wider scope. Other reconstructions differ. Their disagreement should remain visible wherever the evidence does not decide between them.

This also changes the familiar story that Greek arithmetic collapsed and its practitioners fled into geometry. Geometry already existed as a vigorous tradition. An alleged flight cannot explain why mathematicians moved into a domain they already occupied. Books VII through IX of the Elements, meanwhile, preserve substantial arithmetic concerning divisibility, primes, relative primality, parity, and greatest common divisors.

Greek number and geometrical magnitude were also not simply two names for our real-number system. Saying that the Greeks discovered a new kind of number and then refused to recognize it silently imports modern categories. In many contexts they were studying relations among continuous magnitudes while reserving number for discrete plurality.

A more modest conclusion is defensible. Incommensurability made certain distinctions between number and magnitude mathematically consequential and increased the importance of methods that did not presuppose a shared unit. It helped shape the expression and organization of Greek mathematics. That is a claim about structures and techniques, not an established account of psychological flight.

The difference returns us to the man on the ship. The proof is firm; the protagonist is not. We have evidence of teachers, students, classification, and demonstrations. We have a historical Hippasus, but not a secure report identifying the first moment of discovery or recording the murder imagined by the popular tale.

Even the essay's own ending must resist excess. We cannot prove that none of these people ever died by violence or drowning. We can say that the evidence examined here does not establish that someone was thrown overboard for discovering incommensurability. Refusing an unsupported event is different from claiming complete knowledge of every person's fate.

A structural fact makes an awkward story. It remains true without a hero, a turning point, or a body. Later narration supplies a discoverer to give it a face and a crisis to give it a plot. Recognizing that impulse does not make history less human. It brings other people back into view: an elderly teacher from Cyrene, an ill young Athenian, and members of southern Italian associations dispersed by political conflict.

The mathematical response deserves attention on its own. Nobody found the common measure, because no such measure exists. Nobody therefore had to announce that the magnitudes could no longer stand in rigorously comparable relations. What changed was a third thing: the criterion for saying that two ratios are equal.

That is the philosophical thread this series will follow. It is an interpretation of a mathematical achievement, not an ancient declaration of purpose. Again and again, what cannot be obtained under one standard becomes accessible after the standard is reformulated. The object need not disappear. Sometimes the decisive invention is a new account of what it means to have reached it.