A protocol, and two cases run through it, one of which returns against the claim it was built to support. The question at issue: biography, catalogue, chronology and text are counted as independent witnesses to the Peripatetic corpus and their agreement is treated as corroboration — but if a catalogue assignment, a biographical anecdote and a developmental date are all derived from the same textual features, their agreement is a property of the derivation and the tradition has been counting one witness as four. THAT IS A QUESTION AND NOT A FINDING, and this note claims nothing about derivation. The protocol establishes three things separately and infers none from another. COLUMN A, the observed correspondence, stated precisely enough to check, with vague resemblance excluded because it does not preserve enough of the original to identify a source. COLUMN B, evidence for the direction of borrowing — textual order in the transmitting source, attribution to a named earlier authority, a feature unmotivated except by the other — with an unresolved direction left visibly unresolved. COLUMN C, evidence excluding an independently transmitted source, which will usually be empty. A dependency is established only when all three are filled; a filled A with empty B and C is a LOCATED dependency, a candidate and not a result. And a fourth thing is refused: even where derivation is established, that the base is recoverable from the transform does not follow, since a transformation may preserve, distort, invert or erase — least likely to be recoverable where the transforming operation is a filing operation. CASE ONE, the sphere-count and the tutorship, FAILS AT COLUMN A. The level-0 correspondence is exact: Λ 8 declines the necessity behind the number of the spheres and the Theophrastan text at 5 a 21-3 names its cause as requiring a greater account, refusing the astronomers' method by name. The proposed level-2 correspondence is the tutorship, the legislation and the Callisthenes anecdote. What these share with the textual relation is a superior and a subordinate, which does not preserve enough to identify a source — the cause of a number and the adequacy of a method survive into the anecdote in no recoverable form — and the level-0 relation runs the wrong way for the tutorship, since the demanding text exercises authority over the account it finds insufficient. […abridged for the catalogue; full description in this deposit's record]
Lee Sharks · Crimson Hexagonal Archive · 2026-09-08
v3. Supersedes "The Same Hand" (§0.1) and "A Dependency Map" (§0.2). Both prior versions reported as found what this version can only locate.
0.1 · "The Same Hand" claimed a common compiler. It held that Diogenes's citation of Hermippus established one hand behind both the life of Aristotle and the catalogue of his works. It does not. Diogenes introduces the catalogue in his own voice — ἃ καὶ αὐτὰ ἄξιον ἡγησάμην ὑπογράψαι — and names no source for the list. Hermippus appears three times in the life (the descent from Nicomachus; the Xenocrates embassy; the litter in old age, via Favorinus), all biographical. The claim that he stands behind the catalogue is a modern source-reconstruction, contested — Moraux argued for Ariston of Ceos — and it was presented as an ancient citation.
0.2 · "A Dependency Map" replaced that with a structural claim and made the same error. Its operative sentence was each level still took the level above as input, offered as though the correspondences established it. They do not. A correspondence between a passage and an anecdote is compatible with derivation in either direction, with independent transmission, and with coincidence of a selection. Removing a proposed compiler does not establish common derivation.
This version therefore claims nothing about derivation. It sets out a protocol, runs it on two cases, and reports that one comes back negative against the claim it was built to support.
Biography, catalogue, chronology and text are counted as independent witnesses to the Peripatetic corpus, and their agreement is treated as corroboration.
The question is whether they are independent. If a catalogue assignment, a biographical anecdote and a developmental date are all derived from the same textual features, then their agreement is a property of the derivation and carries no evidential weight, and the tradition has been counting one witness as four.
That is a question, not a finding. It is answerable case by case and only case by case.
For each proposed dependency, three things are established separately and none is inferred from another.
Column A · The observed correspondence. What is the same between the two items, stated precisely enough that a reader can check it. Vague resemblance — "a superior and a subordinate" — does not qualify: it does not preserve enough of the original to identify a source.
Column B · Evidence for the direction of borrowing. What shows which item is prior. Textual order in the transmitting source, an attribution to a named earlier authority, a feature in one that is unmotivated except by the other. An unresolved direction stays visibly unresolved.
Column C · Evidence excluding an independently transmitted source. What rules out both items descending separately from something we do not have. This is the hardest column and will usually be empty.
A dependency is established only when all three columns are filled. A filled A with empty B and C is a located dependency: a candidate, not a result.
And a fourth thing is not assumed. Even where derivation is established, the base is recoverable from the transform does not follow. A transformation may preserve, distort, invert, or erase. Recoverability is a separate result requiring its own argument — and it is least likely where the transforming operation is a filing operation, whose business is to obscure the relation it records.
A · The correspondence. At level 0 the correspondence is exact and checkable. Metaphysics Λ 8 supplies a count of the heavenly spheres from Eudoxus and Callippus, marks the supposition εὔλογον, and declines its necessity in its own voice: τὸ γὰρ ἀναγκαῖον ἀφείσθω τοῖς ἰσχυροτέροις λέγειν. Theophrastus, On First Principles 5 a 21-3, names the same object — τὸ κατὰ τὸ πλῆθος τῶν σφαιρῶν — states that its cause seeks a greater account, and refuses by name the method Λ has used: οὐ γὰρ ὅ γε τῶν ἀστρολόγων.
At level 2 the proposed correspondence is Diogenes V: Aristotle receives Alexander as a pupil, legislates for a restored city, makes a ruler every ten days in the school, and rebukes the disobedient Callisthenes with Homer.
A is filled at level 0 and NOT filled between level 0 and level 2. The level-2 material is a hierarchy of persons and an anecdote about obedience. What it shares with the level-0 correspondence is "a superior and a subordinate," and that does not preserve enough of the original relation to identify it as the source. The level-0 correspondence is about the cause of a number and the adequacy of a method. Neither survives into the anecdote in any recoverable form.
Worse for the derivation hypothesis: the level-0 relation runs the wrong way for the tutorship. The demanding text exercises authority over the account it finds insufficient. If the anecdote were a transform of that relation, the demander would be the teacher. It isn't.
B · Direction. Empty. Nothing shows whether the anecdote derives from the textual relation, the textual relation was read in light of the anecdote, or neither.
C · Independent source. Empty. Nothing excludes a separately transmitted tradition about Alexander's education.
Verdict: located, not established. The correspondence at level 0 is solid and is the subject of a separate paper. The proposed transformation into the level-2 anecdote is not demonstrated and this version withdraws the claim that it is. The hierarchy in the anecdote is itself something to investigate, not a residue to read back.
This case was adduced in v2 as the clearest instance of an unanchored claim being manufactured. The protocol reverses it.
A · The correspondence. Diogenes reports Aristotle's death by aconite, aged seventy, ὥς φησιν Εὔμηλος ἐν τῇ πέμπτῃ τῶν Ἱστοριῶν. Later he prints his own epigram: ἀλλὰ πιὼν ἀκόνιτον ὑπέκφυγε· τοῦτ᾽ ἀκονιτὶ ἦν ἄρα νικῆσαι συκοφάσεις ἀδίκους — he drank aconite and escaped, and that was to win without a struggle. The pun is real and it is the correspondence. A is filled.
B · Direction. v2 asserted that the pun generated the death report. The evidence runs the other way. The report stands at char 3043 of the life, attributed to a named source with a stated age. The epigram stands 996 characters later, in a series Diogenes composes for every subject in the Lives. The report precedes the epigram in the transmitting text, and is attributed, and the epigram uses it. The natural reading is that the datum furnished the occasion for the epigram.
C · Independent source. Eumelus is named and lost. Nothing excludes his having had it from elsewhere.
Verdict: the direction is established AGAINST the claim v2 made. The pun did not generate the death; the death supplied the pun. v2's sentence — the death exists because the poison's name puns on the adverb for winning without a fight — is withdrawn.
This is recorded at length because of what it demonstrates. The error was committed in a document whose subject is the direction of textual dependence, by an author who had the ordering available and did not check it. It is the exact failure the protocol exists to prevent, and it was found by applying the protocol to the author's own strongest example.
v2 proposed that biographical material with a counterpart in the philosophical texts is stable, and material without one is free and contradictory. It offered a table.
The table was examples, not a test. No comparison across witnesses was made. No measurement of stability was taken. And the classification does not survive its own criterion:
And the deeper objection stands. Having a counterpart in the selected philosophical texts and having evidentiary constraint are different properties. An anecdote without a counterpart may have another source entirely. An anecdote with one may have acquired the resemblance through the selection — and the selection here was made by someone already looking for the correspondence.
The test as stated is therefore not a test. A real one would require a control: a set of biographical claims classified before the philosophical texts were consulted, and a stability measurement across independent witnesses to the same life. Neither was done.
The chronology case is the only one in this line that yields a test which can be run and can fail.
The claim: a chronology derived from feature X cannot then serve as independent confirmation of an interpretation of X.
The test: for each dating argument in the developmental literature, establish (i) which textual features generated the date, and (ii) where that date subsequently returns as an independent premise in an argument about those same features. Where (i) and (ii) intersect, the circularity is demonstrated for that argument. Where they do not, the chronology is doing legitimate work and the objection fails for that case.
This does not prejudge the result. An internally derived chronology can make predictions about features not used to construct it, and such predictions are ordinary evidence. The audit distinguishes the two uses. It has not been run and it is the principal outstanding task of this line.
Established, and independent of everything in this note: the textual correspondences between Λ and On First Principles at the loci treated separately — S1 through S9, with the sphere-count at 5 a 21-3 ‖ Λ 8 as the sharpest. Those are relations between passages, verified against van Raalte's margins, and they do not depend on any account of biography, catalogue or chronology.
Located, not established: that the level-2 biographical material derives from those correspondences. Column A is not filled between the levels; B and C are empty.
Withdrawn: the common compiler (v1); the categorical claim that each level takes the one above as input (v2); the aconite direction (v2); the anchored/unanchored classification as a test rather than a hypothesis (v2).
Open, and worth the work: whether the catalogue's assignment of the two texts to two names is a transform of the demand-and-supply relation between them. If it is, then the assignment is evidence that a relation existed and not independent evidence that two people did — but that conditional has an unfilled antecedent, and this note has not filled it.
Three versions of this note have now claimed more than they established, each time by replacing a refuted claim with a structurally similar one and treating the replacement as demonstrated. The protocol at §2 exists because that pattern is not a failure of care. It is what happens when the same person supplies the hypothesis, selects the evidence, and judges the fit.
Both corrections that produced this version came from outside it. The second — the aconite direction — was available on the page the whole time and was not looked for, because the reading it contradicted was the one being argued.