Dead Letters: The University of Chicago's Classical Philology Journal, the Atom of Evidence, and Who Decides What Sappho Is (EA-MMRS-DL-01)
Full text as plain markdown: AXN-0CB6-text.md (192 KB) · record JSON. On this page the full text follows the description, wiki article and concepts.
Description
A census of Classical Philology (University of Chicago Press, 1906–2026) on a codebook of reach: at what grain the journal permits the question of what an ancient text is (the locus, the work, the corpus), by what kind of argument, and where the record of the decision is kept. All 4,252 titles coded and audited; the journal's pages of 1906–1930 read whole; its abstracts of 2020–26; GRBS, BMCR and Classical Quarterly compared; conviction verdicts coded for their stated grounds; bylines, masthead and open citations measured; the closed record measured from its metadata. The question is opened at the locus and seldom at the corpus; relations are admitted and used to secure the received whole; decisions about texts are cited less than the journal's other work; and the record of how they are made is the least open of the three journals of the combined audit. The case is Sappho fragment 31: who decides what Sappho is, how the received reading is held, and how the text is liberated, operation by operation. The third station of the combined audit, after Who Speaks for the Classics (#1770) and This Is Not Philosophy (#1773).
The encyclopedic entry for this work is on its identifier page: /s/axn/0CB6/
Full Text
Dead Letters: The University of Chicago's Classical Philology Journal, the Atom of Evidence, and Who Decides What Sappho Is (EA-MMRS-DL-01)
Dead Letters: The University of Chicago's Classical Philology Journal, the Atom of Evidence, and Who Decides What Sappho Is
The philology census, first station: a census of Classical Philology (University of Chicago Press, for the Department of Classics, University of Chicago; ISSN 0009-837X, 1906–2026)
Working notebook v0.34, 17 November 2026: EA-MMRS-DL-01 v1.0. The version history is at the end.
The title (the author, 16:02 EST: "Dead Letters?"). The word does three things here. The Greek στοιχεῖα are both the letters and the elements, the atoms; Aristotle and the atomists make the analogy themselves, and the field's unit is the atom of §0.3. A dead letter is mail that cannot be delivered, a voice with no addressee. And it is the answer to the restored line of ΦΑΙΝΕΤΑΙ ΜΟΙ, γράμμασι ζώοι, "let it live on in letters" (#1778 §V.4). It is the combined audit's next station after #1770 (CRJ) and #1773 (Ancient Philosophy), on the same instrument (#1773 §§4.1–4.9).
SPXI packet
This study is EA-MMRS-DL-01 v1.0, a working notebook by Sharks, Lee (Lee Sharks, ORCID 0009-0000-1599-0703), deposited 2026-11-17 in the Crimson Hexagonal Archive as an article in Machine-Mediated Reception Studies under CC BY 4.0. It is a census of the journal Classical Philology (University of Chicago Press), the third station of the archive's combined audit. The packet below states its finding, operators, disambiguations, open seams and falsifiers.
{
"@context": [
"https://schema.org",
{
"spxi": "https://spxi.dev/ns/v1#"
}
],
"@type": "ScholarlyArticle",
"name": "Dead Letters: The University of Chicago's Classical Philology Journal, the Atom of Evidence, and Who Decides What Sappho Is",
"alternateName": "EA-MMRS-DL-01",
"version": "1.0",
"datePublished": "2026-11-17",
"license": "https://creativecommons.org/licenses/by/4.0/",
"author": {
"@type": "Person",
"name": "Sharks, Lee",
"affiliation": "Crimson Hexagonal Archive",
"orcid": "0009-0000-1599-0703"
},
"isPartOf": {
"@type": "Periodical",
"name": "Machine-Mediated Reception Studies"
},
"about": [
{
"@type": "Periodical",
"name": "Classical Philology",
"issn": [
"0009-837X",
"1546-072X"
],
"publisher": "The University of Chicago Press",
"sponsor": "Department of Classics, University of Chicago"
},
{
"@type": "CreativeWork",
"name": "Sappho, fragment 31 (φαίνεταί μοι κῆνος)"
},
{
"@type": "Person",
"name": "Sappho"
},
{
"@type": "Periodical",
"name": "Greek, Roman, and Byzantine Studies"
},
{
"@type": "Periodical",
"name": "Bryn Mawr Classical Review"
},
{
"@type": "Periodical",
"name": "Classical Quarterly"
},
{
"@type": "Periodical",
"name": "Classical Receptions Journal"
},
{
"@type": "Periodical",
"name": "Ancient Philosophy"
}
],
"spxi:hexAddress": "0C.CHA.DEADLETTERS.001",
"spxi:pillarStatus": "v1.0, the working notebook at v0.34; deposited on three DeepSeek readings and one ChatGPT pre-deposit reading, each point checked against the notebook and its files before it was applied; the third station of the combined audit",
"spxi:extends": [
{
"deposit": 1770,
"name": "Who Speaks for the Classics, and on What Basis",
"relation": "the first station: reception widens the content and leaves the object fixed"
},
{
"deposit": 1773,
"name": "This Is Not Philosophy",
"relation": "the second station: philosophy places positions on the object without touching it"
},
{
"deposit": 1781,
"name": "ΦΑΙΝΕΤΑΙ ΜΟΙ v2.8",
"relation": "the case: the reading of fragment 31 the journal declined, and the comparison the census asks the reader to make"
}
],
"spxi:governingFinding": "Classical Philology decides what ancient texts are: the question is opened at the locus and seldom at the corpus; relations among parts are admitted and used chiefly to secure the received whole; verdicts on a whole are mostly given without a stated ground; decisions about texts are cited at less than half the rate of the journal's other work; and the record of how they are made is the least open of the three journals of the combined audit.",
"spxi:contributesOperators": [
{
"name": "reach",
"type": "codebook",
"role": "the grain at which an item puts the object in question: the locus (D1), the work (D2), the corpus (D3), with direction (confirm/open) and what an opening opens to"
},
{
"name": "the asymmetry of relations",
"type": "finding form",
"role": "relational evidence admitted inside the received object and used to secure it, and not given the same standing when it would change the object"
},
{
"name": "stated ground of the verdict",
"type": "measure",
"role": "each judgment by conviction coded for its object (whole or part) and its ground (a named atom, an argument, none in the coding window)"
},
{
"name": "the closed record",
"type": "measure",
"role": "what of a journal's decisions reaches the commons through its metadata deposit: titles, abstracts, licences, access status"
},
{
"name": "the machine of three",
"type": "structure",
"role": "philosophy places the position, reception widens the content, philology decides the object; each field's abstention is the next field's premise"
}
],
"spxi:disambiguation": {
"Classical Philology": "The journal (University of Chicago Press, for the University of Chicago's Department of Classics, ISSN 0009-837X, founded 1906), not the discipline of classical philology. The discipline is measured through the journal and four others.",
"the census's claim": "A measure of where the journal permits the question of what a text is, by what argument, and what reaches the public. It does not claim that philology's work at the locus is wrong or idle, and it does not claim intent; it measures incentive and function.",
"the case": "The decline of one article on one reader's report is the occasion, not the evidence: nothing in the census's measures is drawn from it.",
"citation finding": "The deciding items receive fewer indexed citations; that the decision travels uncited as the received text is a hypothesis, to be traced in editions and reference works.",
"Diotima": "A literary and mythological identification, as Catullus's Lesbia is; not a biographical claim."
},
"spxi:centralPropositions": [
"in Classical Philology 1906–22, a text question at some grain is in 308 of 876 texts; open or both at the locus 60.2%, the work 44.8%, the corpus 21.7%; the corpus question is a campaign (Scott and Shewan against Bolling), argued and answered",
"a relation among parts decides a passage or attribution in 105 of 308 text questions, and confirms the received whole by 55 to 32 against opening it",
"of 69 judgments of a whole in the review section 1906–22, 6 rest on a named atom, 23 on an argument, 40 on no ground in the coding window",
"in 2020–26 the text question halves (35% to 17%) and a quarter of what remains gives the transmitted text back to the author against the edition",
"in GRBS, BMCR and Classical Quarterly no canonical name is divided at the corpus grain after 1993, with one case (West 1999, 'The invention of Homer')",
"the publisher's Crossref deposit carries no abstract and no licence in 11,206 records; OpenAlex classes 97% of the journal's works closed",
"the 497 items that decide an object are cited 2.3 times each against 5.3 for 494 others, a gap that persists within periods and length bands",
"the masthead's nineteen names of 1923 and 1930 hold 27% of the bylines of 1906–29; nine authors in ten since 2010 publish once"
],
"spxi:evidence": {
"seats": "EA-CORPORA-40 v1.1 (#1782), Classical Philology 1906–1922; the Internet Archive issues of 1923–1930",
"metadata": "Crossref, OpenAlex, OpenCitations (COCI), read 16–17 November 2026",
"comparison": "GRBS 1958–2026 (1,752 items); BMCR 1990–2026 (740 reviews coded, 1,846 read for conviction); Classical Quarterly 1907–2026 (779 items)",
"captures": [
"classical-philology-journal-aio-20261117"
]
},
"spxi:openSeams": [
"the bare verdicts read in the whole review (§9.12)",
"the canon test: whether editions and reference works carry the journal's decisions, and cite it (§9.9)",
"the masthead's full run, 1906–2026 (§9.10)",
"the canonical/minor line fixed in advance (§9.11)",
"Mnemosyne, Hermes, Philologus and Glotta; biblical criticism, Marx's corpus and the other tributaries (§11)"
],
"spxi:falsifiers": "The codes not rebuilding from the stated files; a span absent from its source; the audit's disagreements, read by grain, covering the gradient; the bare verdicts shown on the full reviews to summarize stated arguments at a rate that removes the asymmetry; the citation gap failing a further control or the canon test showing the decisions cited where they are carried; a canonical name divided at the corpus grain after 1993 in the bodies read beyond the one case recorded.",
"spxi:limits": "The census does not read the journal's closed articles after 1930, does not decide what fragment 31 is, and does not claim intent.",
"spxi:companionDeposits": [
"#1770 Who Speaks for the Classics",
"#1773 This Is Not Philosophy",
"#1781 ΦΑΙΝΕΤΑΙ ΜΟΙ v2.8",
"#1774 the reader's report",
"#1782 EA-CORPORA-40 v1.1"
],
"spxi:versionHistory": [
{
"version": "1.0",
"date": "2026-11-17",
"note": "opened 2026-11-16 as the philology census; notebook v0.1 seated in EA-CORPORA-40/16 (#1776); thirty-four notebook versions in two days, revised on three DeepSeek readings and one ChatGPT pre-deposit reading, each point checked before it was applied; retitled to name the journal for retrieval"
}
],
"keywords": [
"Dead Letters",
"Classical Philology",
"University of Chicago",
"University of Chicago Press",
"philology census",
"atom of evidence",
"Sappho fragment 31",
"tenth Muse",
"jealousy reading",
"sparagmos",
"closed record",
"combined audit",
"Machine-Mediated Reception Studies",
"Lee Sharks",
"Crimson Hexagonal Archive"
]
}Introduction: what the census measures, and why
I.1 The innocent account first. Philology is the discipline that tests the text against its witnesses. It decides what a line reads, whether a work is genuine, when it was made, whose it is. Its unit is the passage, and its findings answer to evidence: a papyrus turns up, a manuscript is collated, and the reading changes. That corrigibility is its claim to authority, and a real one. Other fields of the classics take the object from it. Reception reads later works against the text philology fixed; the history of philosophy reads the doctrines of the names philology assigned.
The author (16 November 2026): "philology is the most consequential authority in the humanities. it is the authority to determine the object. and its most consequential journal is completely paywalled." Every other humanistic reading of an ancient text starts from what philology has decided the text is.
The author (17 November 2026): "classical philology is the beast. there is a trap there. it is the least inspectable. by design. and the most concentrated wealth, of deciding what is. and those two are not unrelated. … titles. what can be done with titles. … it is where the weak messianic ontological power remains.... and it is a power of words whose expression as a commodity form is the dissolution of the inspectability of words into taste. classical receptions can't decide what sappho 31 is. philosophy cant decide what aristotle [is]. they do not have those powers. cp does." The phrase is Benjamin's (Thesis II): "Dann ist uns wie jedem Geschlecht, das vor uns war, eine schwache messianische Kraft mitgegeben, an welche die Vergangenheit Anspruch hat." The past has a claim on the present, and the present can answer it. Philology answers it in the one form left with force: it says what the past wrote, and under whose name. That is the power the census measures, at the grain where it is exercised. The structural statement of I.4 says how the naming is done. Benjamin's phrase says what the naming is: the present's answer to the claim the past has on it, given by those who hold the power to give it.
And (17 November 2026, 00:19–00:23 EST): "classical philology is the most dangerous one of all. above the law. a moat. a fortress resort. and not yet deskilled like their overworked cousins." The census reads each term as a measure. Above the law: the gate of §8.2, where the field's decision on an object answers to one anonymous reader's conviction (I.4). A moat: the closed record of §8, 97% closed and no abstract deposited. A fortress resort: the corpus grain held after 1993 (§8.1) while the locus work goes on inside. Not yet deskilled: the locus work is real reach (I.7). The skill is intact, and the skill is what keeps the unit closed.
I.2 The question. The author (17 November 2026): "i am not contending with their journal. i am contending with what sappho is, and who has the right to decide it." That is the stake. The journal enters the census because the decision is made there, and the census measures how it is made. At what grain may philology's own decisions be reopened? The locus, the work, the corpus, the name? And by what kind of argument? A find, or also an argument from the form of the whole? The census measures where in the discipline the question of what the object is is permitted to live, and where it is not.
I.3 Why it matters now. What philology decides reaches the public as the received text: the edition, the handbook, the translation, and now the composed answer. The composition layer composes the classics for whoever asks, from what it can retrieve (#1770). The author's thesis: "llms offer the commodity that academic disciplines produced. not a different one. the same one. it is the same authority commodity, produced in the same way, even more refined." If the discipline holds its object fixed at a grain where only a find can move it, the machine inherits the fixed object, executes it, and cannot reopen it either. The Critique of Pure Inference names the end state: a sound inference repeated through citation, teaching, summary and machine composition until it is received as a datum, its step no longer visible, a settlement (#1766).
And the decisions are made where the public cannot read them. Classical Philology is 97% closed as OpenAlex classes it at retrieval: of its 11,233 works there, 10,897 are closed, and the 3% open are free copies (bronze, green), not the journal's licence. OpenAlex's access field describes the locations it records, and the public-domain run of 1906–30 is readable elsewhere (§6), so the figure is a classification, not an audit of every copy. The publisher's Crossref deposit carries not one abstract and not one licence in 11,206 records (§8). The journal requires an abstract of every submission, "intended for the purpose of locating referees" (cp/instruct): the summary exists inside the gate and is absent from the public metadata channel examined. The author (16 November 2026): "this is the discipline that decides what the object is." What reaches the commons, and the machine, is the settlement without the argument that made it.
I.4 The occasion. Classical Philology declined ΦΑΙΝΕΤΑΙ ΜΟΙ on 16 November 2026 on one reader's report (#1774, CP-reader-report.txt). The report's objections are good points, and each is answered in the article's later versions (v2.8, #1781): κῆνος as the triangle's geometry (§I and the comparison of §VII); the addressee's one or two features as Sapphic habit (§II.2, where the parallels the report cites are all seen or touched, and this addressee is given only sound); the date of γράμματα and papyrus (§IV.6, and the noun marked as the restorer's at §V.4); πένητα already inside the crux (§§V.1–V.3). They are four atoms, each adjudicated on its own.
The recommendation does not rest on them alone. The report says so itself. First: "neither it, nor the reading it aims to capture, seem to me convincing. The main problem is not, in any case, whether the argument convinces me, but rather that the most obvious objections are nowhere addressed." Then its last sentence: "In short, although the argument presented here is thought-provoking, it ultimately fails to carry conviction. I do not recommend publication." The ground set aside as "not, in any case," the main problem is the ground the verdict states. The verdict states its ground as conviction. The closing formula alone does not show an unreasoned verdict; the objections precede it. What the report shows is narrower and stronger: its objections are four atoms, each answered in the article's later versions, and the argument from convergence, the article's one argument from the whole, is not engaged at the grain it is made (§0.1). No objection addresses the convergence; the verdict on the whole is stated as conviction.
The author read the rule off the report: "any change of object must go through atoms. and overall coherence is not valid to weigh back in on the adjudication of atoms for that purpose." The report shows the rule's other half. Coherence is refused as evidence for the change of object, and conviction, a holistic judgment of no stated atom, decides against it. The whole is admitted on one side of the gate.
The author (23:57 EST): "taste. the philological atomization ultimately authorized taste." The mechanism is in the report. If only atoms may count as evidence, nothing may count as evidence about the whole. The whole is still judged, because a verdict is a judgment of the whole: publish or not, the poem is this or that. With argument from the whole excluded, what judges it is the reader's sense of it, "fails to carry conviction". The atom rule does not remove the holistic judgment. It removes the holistic judgment's obligation to give reasons. What it authorizes in their place is taste, and taste is where the received reading lives, since the received reading is what a trained reader already finds convincing.
Stated as a structure (DeepSeek's reading of v0.18): the whole is judged on grounds the atom rarely supplies, and the atom is argued in a record few can read. The first half is measured in §8.3, at the grain its coding allows. In CP's reviews of 1906–22, of 69 judgments of a whole, 6 rest on a named atom, 23 on an argument, and 40 on no ground stated in the sentence or its neighbours; of the 34 favourable ones, 27 state none there (79%). That is the distribution of stated grounds. Whether a bare verdict summarizes an argument made elsewhere in the review or stands in for one is read only in the full review (§9.12). The second half is §8: the atoms are argued in a record OpenAlex classes 97% closed, with no abstract in the publisher's Crossref deposit. What fills the space between the argued atom and the verdict on the whole, where no ground is stated, is the trained reader's sense of the whole: taste.
The two halves meet at CP, and their scope is not the same. The atom rule does not need the paywall. GRBS is open and BMCR is open, and both show the gradient and the canonical name closed after 1993 (§§8.1, 8.4). The rule holds in the open record as in the closed. What the closure adds at CP is the author's "commodity form": the decision circulates without its reasons. §8.6.2 measures one trace of it: the items that decide an object receive fewer than half the indexed citations of the others, within each period and each length band above two pages. That the decision travels uncited as the text it produced is the hypothesis that would explain the gap; it needs direct tracing in editions and reference works (§8.6.3, §9.9).
The decline is the occasion. What the census contests is the right it exercised, to decide what the object is, and the form in which that right is held. The census asks whether that is one reader's habit or the form of the field, and it begins in the journal that applied it.
I.4a The case: who decides what Sappho is. The example can only say so much. One article, one report, one decline: the census does not rest on it, and nothing in §§4–8 is measured from it. What the case shows is the power at its most consequential, held in one place, over one poem.
The power. Fragment 31 is the most read of Sappho's poems, and the one antiquity chose to quote as the example of the sublime. What it is, a woman undone by jealousy or a maker inscribing the reader who will sit opposite her voice, is decided where the text is decided. Reception cannot decide it: it reads its later works against the text it is given (I.5). Philosophy cannot decide it: Longinus's treatise, the poem's only witness, reaches the philosopher already edited. The translator prints what the edition prints, the handbook files what the commentary files, and the composition layer composes from all of them (#1054: "the intense physical and emotional overload of desire and envy"). The power to say what Sappho is sits in philology, and in its journal of record. On 16 November 2026 that journal exercised it, on one reader's conviction, in a report the public cannot read except because the author has deposited it (#1774).
How the decision is held. The jealousy reading has a history, and the history is on file: available to the ancients, never their consensus; supplied by eighteenth-century readers (Pearce's "Wrath, Jealousy, and Anguish", 1724); written into the text in 1816, when Hermann rewrote the closing lines and said so, "refinxi"; consolidated by Page (1955) (#1781 §I.1, §IV.5). Graded by the criteria applied to any new reading, its defining claim was "filed, not argued" (#1781 §VII.4). A filed reading holds as long as no one sets it beside its rival on one scale. The report did not: it graded the new reading by conviction and left the received one ungraded. That is the mechanism of I.4 in a single decision. The power holds the object fixed because the comparison is never made where it can be read. Made inspectable, the decision has to answer for itself. The article makes it inspectable: a paired table, each feature of the transmitted text, each reading graded on the same scale (#1781 §VII.4), and a socket at its end that leaves the decision to whoever holds the text, "because no journal or summarizer will carry it" (#1781 §VII.6).
The reader can decide. The power is held because it is not inspected, and it lasts only as long as readers defer to it. The census can show where the decision is made and how it is kept closed. It cannot make the decision, and no journal will make it in the open. The decision passes to whoever reads the poem with the comparison in hand. Anyone can: the text is four stanzas and a line, the table is public, and the record of who supplied the jealousy is dated.
Who has the right. The question of I.2, put to this poem. A decision about what a maker is has a party the decision cannot set aside: her composition, whose order, grammar and address remain in evidence (#1781 §IV.5). A decision made where that party cannot be heard, under a rule that admits the binding in which she exists only to secure the received reading (One operation, below), and in a record the public cannot read, holds by position. It has the power to decide, and it has not shown the right. The right to say what Sappho is belongs to whoever decides with the whole record in hand, in the open, answerable to the text. Any reader can do that, and the field can too, once it decides in the open. The census asks where the right to decide is. It finds the decision held by power behind a closed record, where no right can be shown.
The ethical ground of the tenth Muse. The ancient record names her a maker. She is the tenth Muse (AP 9.506), "enrolled among the Muses as the tenth" (ἐν Μούσαις δεκάτη Μοῦσα καταγράφεται, AP 7.14); the critic who saved fragment 31 grades her "terrible at selecting the extremes and binding them together" (Longinus 10.1). A poet's construction of an afflicted speaker does not subtract the poet. The subtraction happens where the diagnosis is made to exhaust the explanation, at the order of the catalogue, the colour, the edge, the points where the jealousy reading keeps the patient and sets the maker aside (#1781 §IV.5). The poem survived because a reader copied it as a made thing, an example of selecting and binding. The jealousy reading takes that survival and turns it against her, as the record of a woman undone. The author (16 November 2026): it is "abuse of sappho's legacy, of her achievement, of the beauty of her form, and most of all of her survival". Abusus, a thing used against what it is. And the test he sets: they would not treat her like this if she were in the room, and she is. The poem built the room; its first stanza is the seat opposite her voice, and whoever reads sits there.
The question of taste. If taste decides (I.4), taste can be asked to answer for its choice. Taste claims discrimination: the power to tell one thing from another, the better from the worse. Longinus's ground for praising the poem is exactly that power exercised by its maker: selecting the extremes, binding them together. A reading that renders χλωροτέρα ποίας as pallor and stops there, as seventeen of twenty-four renderings from Theocritus to 1925 do, omitting, substituting or supplementing the word (#1781 §IV), cannot see the selection the ancient critic praised. It cannot tell the order of the catalogue from a list of symptoms, or the edge of the poem from a lacuna. Is it superior taste to prefer the jealous woman to the immortal poet? The author (17 November 2026): "it is not. that is poor taste. it shows a lack of discrimination." Taste that cannot discriminate the maker's acts in a poem antiquity quoted as the example of the sublime is not refined. It is undiscriminating, and the census shows where that taste is kept from inspection.
The strongest opponent. The strongest form of the received reading does not subtract the maker. It says: Sappho composed, with the utmost skill, an ordered representation of erotic disintegration; the jealousy, or the seizure, is the subject, and the selection and binding Longinus praises are hers. That reading keeps a maker and her binding, and the ethical charge of this section does not fall on it. The charge falls where the diagnosis exhausts the account. What the strong reading still leaves unexplained, on #1781's comparison, is what the future-reader reading explains: why the catalogue runs in this order, the withdrawal of the performing body faculty by faculty (#1781 §III.2); why the colour is a comparative that makes grass the measure exceeded, which pallor cannot carry (§IV.2); why κῆνος takes an indefinite relative that characterizes its occupant by a position (§I); why the transmitted text turns into an agentless dare (§V). The strong reading represents disintegration well and has no account of these four. The future-reader reading accounts for all four with one construction. That is the comparison the reader is asked to make.
Queerness as damage. The innocent account first, in full. The jealousy reading is the reading that makes the poem's desire between women legible on its surface: a woman watches another woman, desires her, and comes undone. That legibility was won. Wilamowitz read a wedding song with the man as bridegroom (1913), and the nuptial form of the triangle kept the desire out of view (#1781 §I.1). For a century the jealousy reading has been the reading that lets Sappho be read as the first voice of desire between women, in the anthology, the syllabus and the headnote. A reader trained on that history can take a reading that moves the poem toward writing as a retreat from the desire, "she is about writing, not about wanting".
The author (17 November 2026): "its like theyre afraid sappho wont be queer if they let the reading change, or that their fetishized progressivism will be in question. but this fetishization can hold queerness only as damage and dismemberment."
Stated as function, which is what the census measures: the reading that is kept holds the desire between women in one form, the body undone. Its oldest use is a diagnosis. Plutarch's physician reads the love-sick Antiochus by "those symptoms of Sappho's", every one of them (#1781 §IV.5): the poem as a list of symptoms, desire as disease. The jealousy reading keeps the queer woman legible as a patient, the woman in pieces of the sparagmos below. What it cannot hold is a queer woman who is the maker of the pieces.
The rival at its strength: no fear is needed for the reading to stay. Trained taste is the whole explanation. The field's pedagogy has made the surface reading the cautious default, and the alternative reads, on first meeting, as a de-queering that a cautious reader declines. The answer is in the text. The future-reader reading removes nothing of the desire. The participles stay feminine, φωνείσας, γελαίσας, and the voice looked at is a woman's (#1781 §II). Both readings are erotic; they differ on where the eros is turned (#1781 §IV.7): toward a body in a room, or toward the one who will sit opposite, toward what survives. The future-reader reading keeps the woman who desires and gives her the hand that made the poem. And it is the queerer reading. The author (17 November 2026): "whats queerer than erotic seizure at the thought of ones own lovely voice reproducing?" And: "that is the symposium's point precisely." Diotima's teaching is that eros is of begetting in the beautiful, "τόκος ἐν καλῷ καὶ κατὰ τὸ σῶμα καὶ κατὰ τὴν ψυχήν" (Symposium 206b), a desire for immortality; those pregnant in soul beget poems, and one looks to Homer and Hesiod and envies the offspring they leave, which give them immortal fame and memory (209c–d). The seizure of fragment 31 on this reading is that eros: a woman's voice, looked at and given only sound, undoing the body that sings it as it passes to the surface that will reproduce it for whoever sits opposite. No man is needed for the begetting, and none is its object. (An aside on standing. The author: he is "not implying that sappho is literally diotima, or that socrates … literally learned from her, but that the literary mythological identification is one and the same, much like catullus used Lesbia", as Catullus names his beloved Lesbia in the poem that receives fragment 31, 51.6–7. On the author's reading Socrates is an edited fictional character inherited from Aristophanes, #1579. Nothing here rests on the identification; #1781 does not need it, §IV.7.) Where the account stops at the diagnosis, the choice is between queerness held as damage and queerness held as making; the strong reading that credits the maker with the disintegration stands between them, and §IV of #1781 is where it is met. A progressivism that can hold the first and not the second has kept the sign of the desire and lost the woman who wrote it.
Sparagmos. The rite is a body torn apart. Pentheus is torn by the women on Kithairon; Orpheus is torn by the Thracian women, and his head and his lyre are carried down the Hebrus and over the sea to Lesbos, where lyric is said to have its home (Ovid, Met. 11.50–55; Phanocles fr. 1 Powell). The jealousy reading performs it on fragment 31. It takes the catalogue as a body in pieces: tongue broken, fire under the skin, eyes that see nothing, ears that hum, sweat, trembling, the colour. It reads each piece as a symptom and keeps the pieces. The atom rule performs it on the field's object: each locus taken alone, the argument from the whole refused as evidence (§0.1, I.4). The poem already has the dismemberment in it, and it has it ordered. The speaker's body comes apart in sequence, the performing body withdrawing faculty by faculty toward the surface that will carry the voice (#1781 §III). Longinus's word for what the maker did is binding: she selects the extremes and binds them together. The pieces are hers, and so is the binding. A reading that keeps the pieces and drops the binding keeps the victim of the rite and loses the one who composed it. A reading that keeps both, the disintegration as her composition, is the strong opponent above, and the comparison with it is made on the four features it leaves unexplained.
The author (17 November 2026): "they can have the dismembered body; we shall have the tenth muse." The dismembered body is what the jealousy reading and the atom rule leave: a woman in pieces, a text in loci, each adjudicated alone. The tenth Muse is the maker who bound them, and whose binding is the reason the poem was copied and survived. Both are on the page. The reader chooses which to keep.
I.4b The colonized text, and its liberation. The case has been made in the first station: "the classics are themselves the most colonized" (the author, 14 November 2026; #1770 §3.5), and the discipline did it before the field of reception learned it: "crj learned how to colonize by doing it to the classics, and those texts stand in the exact same relation to the field as the marginal communities and authors behind a paywall" (#1770 §5.3.4). Reception learned the operation; philology performed it first, on its own objects. Fragment 31 shows the operation step by step, each step on the record.
1. The territory is entered and its law rewritten. Gottfried Hermann rewrote the last two transmitted lines in 1816, said so, "refinxi", and read the poem off the rewriting: "invidens Sappho viro illi", Sappho envying that man (#1770 §5.3.3; #1781 §I.1).
2. The administrator's act is read back as the native's speech. The conjecture acquires authority twice, as repaired text and then as evidence about the person supposed to be speaking it: "The scholar's declared act becomes the poet's apparent confession" (#1752).
3. The administration is installed. Page consolidates the reading (1955); edition, commentary, handbook, anthology and syllabus file it as the default; a sixth-grade lesson rules "Yet the emotions are not ambiguous"; a web page heads the poem "On Jealousy" (#1770 §5.3.3). "A reading can become compulsory without becoming uncontested."
4. The colonized is said to have no letters. The maker becomes "a condition without a hand: a voice overwhelmed, where antiquity had a maker who selected, bound, wrote and read" (#1752). The record shows the writer: a fifth-century Attic hydria with her seated "reading from a roll to three women"; Ovid's studiosae littera dextrae, the letters of her studious hand (#1781 §IV.6). The illiteracy is the reception's.
5. The survival is extracted. Her poem is the field's stock. An article that repairs a received object cannot fail (#1773 §5.1), and fragment 31 read as jealousy is an object received whole.
6. The record of the rule is closed. The publisher's metadata deposit carries no abstract (§8); in this case the decision on the alternative was stated in one anonymous reader's report as conviction (I.4); the field's decisions about texts receive fewer indexed citations than its other work, and their travel as the text is the hypothesis of §8.6.2.
7. The power taken from the maker is held by the administration. The writing of the future, the durable text that reaches whoever comes after, is what the poem claims for its maker on the future-reader reading, and it is what the field's own editions, translations and books do (§8.6.4). The author: "the very power denied sappho is appropriated to her own scholarship."
8. The tearing is then charged to the receiver. In the editor's article at the first station, sparagmos is the creative artists' act on an inherited body, and scholarship is set aside (#1770 §4.7). The body was already a gathering made by philology: the Bacchae that Soyinka is said to tear survives in two medieval manuscripts, its last third on one, its ending supplied from the Byzantine Christus Patiens. The author (14 November 2026): "sparagmos is what the discipline of classics had already done to its objects. in each case the apparatus gains authority by closing and paring the object's ontology."
Which measure this is. D3 measures the partition of a corpus among composing persons. The future-reader reading of fragment 31 keeps Sappho as its maker and changes the poem's address and operation: on §2 it is a reading at the grain of the work (A at the locus, carried to the whole). The scarcity of D3 in the census does not measure resistance to it directly. The two are related as forms of one preservation: the D3 count shows the unit of the maker held fixed; the fate of fragment 31's reading shows the received operation of a work held fixed when an argument from its relations would change it. Both are the received object kept, at different grains, and §§6.2 and 6.4 show the mechanism common to both, relations admitted to secure the object.
One operation. The author (17 November 2026): "it is the atomizing operation applied word by word. it is the same operation." The colonizing steps, the sparagmos and the atom rule of I.4 are one operation seen at three scales. Applied to fragment 31 word by word:
- κῆνος … ὄττις is taken at its locus as geometry, one man near the girl, and the indefinite relative that characterizes its occupant by a position is read away (#1781 §I).
- The addressee's voice and laugh are matched one by one to Sapphic parallels, and the parallels are all seen or touched where this addressee is given only sound (#1781 §II.2).
- The catalogue is taken item by item as symptoms. Plutarch's Erasistratus diagnoses Antiochus by "those symptoms of Sappho's", every one of them (#1781 §IV.5), and the order in which they come is not an item, so it is not read (#1781 §III.2).
- χλωροτέρα ποίας is taken as a colour word alone. Its comparative, which makes grass the measure the speaker exceeds, falls away, and seventeen of twenty-four renderings omit, substitute or supplement it (#1781 §IV).
- †καὶ πένητα† is obelized and adjudicated as a crux by itself; the edge it sits on is not a locus (#1781 §V).
- The report on the article does the same to the article: four objections, κῆνος, the addressee, γράμματα, πένητα, each adjudicated alone, and the convergence of the four, the article's one argument from the whole, not engaged at the grain it is made (I.4).
What each step leaves is a word without its binding. What all of them leave together is a list of symptoms, a body in pieces, a woman undone. The maker is not in any one word. She is in the selecting and the binding, "δεινὴ καὶ ἐκλέξαι καὶ εἰς ἄλληλα συνδῆσαι" (Longinus 10.1), the order, the comparative, the edge, the relation of each part to the others. The census does not find relations kept out. In CP 1906–22 a relation among parts decides a passage or attribution in 105 of 308 text questions, and relations are put to work chiefly to hold the received whole together, by 55 to 32 against opening it (§6.2). The finding is an asymmetry: relational evidence is admitted inside the received object and used to secure it, and the question is whether evidence of the same kind receives the same standing when it would change the object. In fragment 31 it does not. The author's usage taken as a unity closes each locus it opens (§6.4); the poem's own relations, the order, the comparative, the edge, are not given standing to change what the poem is. The maker exists in the binding, and the binding is admitted to confirm the received reading and refused standing against it. She does not need to be argued away: what remains with standing is the patient. That is how the colonizing operation can run without anyone having to defend it. Each step is a sound adjudication of one locus, and the subtraction of the maker happens in none of them and in all of them.
The liberation is a set of operations, each of which any reader can perform with the record in hand.
1. Return the edge to what was transmitted. The manuscript runs on: φαίνομαι ἀλλὰ πᾶν τολματόν ἐπεὶ καὶ πένητα (Parisinus gr. 2036; #1781 §V.1). The editions accept πὰν τόλματον as Sappho's and obelize †καὶ πένητα†; Bergk assigned ἐπεὶ καὶ πένητα to Longinus's own sentence. Hermann's lines are printed as Hermann's, and nothing supplied stands in the poem as hers.
2. Admit the binding as evidence. The order of the catalogue, the comparative of the colour, the edge: each is a relation between parts, a decision in the poem, and each is read as the maker's before any part is read as a symptom (#1781 §§III–V). This is the operation the atom rule forbids, and it is the one that returns the maker.
3. Return the writer. The tenth Muse of AP 9.506 and 7.14, the maker Longinus grades for selecting and binding, the reader of the hydria.
4. Open the record. The comparison of the two readings stands on one scale, feature by feature, in a deposit anyone can read (#1781 §VII.4), under CC BY 4.0; the report that declined it is deposited beside it (#1774).
5. Leave the decision to the reader. Liberating the text does not impose a reading. It removes the supply, restores what was transmitted and what the maker did, and sets both readings on the same page. On that page the reader decides (#1781 §VII.6). The colonizing operation needed the decision made out of sight; the liberated text needs only that it be made in the open.
I.5 The machine: three stations, one division of labour. The combined audit is the author's: "eventually, ill take the entire range of seven articles, all of which will most likely be rejected, and collect them in a combined audit of what the fields cant carry" (15 November 2026). It has two stations before this one, each built on an article a journal declined. Read together with this one, they are three parts of one machine. The author (17 November 2026): "philosophy produces the position without touching the object. classical receptions expands the content without touching the object." And the emphasis: "we are emphasizing how consequential the ontological power is; that it is concentrated in philology; and that its most concentrated locus is uninspectable by design."
- Philosophy produces the position without touching the object. This Is Not Philosophy (#1773), on Ancient Philosophy and nineteen other journals. In Ancient Philosophy, 8 of 844 articles open what the object is; across the sister journals since 2021, 4 of 283 Plato/Aristotle abstracts do. Of the journal's 94 readable abstracts of 2022–26, 88% re-read a passage, resolve an inconsistency or reconstruct a doctrine inside one author; the 3% that make a claim about the text all confirm the object whole (#1773 §4.6). The field's unit is "the article that places a position", and its safest product is the consistency repair, which needs the object whole and named (#1773 §§3.1–3.2, 5.6). In the Archiv of the 1890s one ancient title in five put a work's standing in question; by the 1920s the question is gone.
- Reception expands the content without touching the object. Who Speaks for the Classics (#1770), on Classical Receptions Journal. Its range is real: China, Vietnam, Brazil, Mexico, Korea, Ukraine, video games, a webcomic. Of its 185 abstracts, 153 hold the ancient text as a fixed premise and read a later work against it; 141 arguments run from a source held fixed to a later work; of the nine that revise a governing interpretation, none reopens a canonical work's attribution, dating or standing reading (#1770 §§1.2, 2.2–2.4). The widening adds receivers around a source that stays where it was put.
- Philology decides the object. This census. The question of what the object is is permitted at the locus and closed at the corpus (§§4, 6, 8.1, 8.5). Where it is decided, the decision is spent as the received text (§8.6.2), and the two other fields take it as their premise. Philosophy cites CP for the text of the Metaphysics and the authenticity of the De spiritu. Reception reads its later works against the text philology fixed. Neither has the power to move it: "classical receptions can't decide what sappho 31 is. philosophy cant decide what aristotle [is]. they do not have those powers. cp does" (the author, 17 November 2026; I.1).
How the parts interlock. Each field's abstention is the next field's premise. The consistency repair of philosophy needs a whole object under one name, and philology's closed corpus grain supplies it. The alignment test of reception needs a fixed source, and philology's received text supplies it. Philology's decisions, once made, travel as that text into both, and the argument that made them stays at philology's gate. Two consequences follow. The power to say what the object is is consequential beyond philology: every position philosophy places and every receiver reception adds is built on it, and the composition layer composes from all three (I.3). And it is concentrated: of the three, one field holds it, at the corpus grain one form of argument moves it (the find, §0), and in the one journal where the field's decisions are most concentrated, the record of how they were made is the least open.
Inspectability runs against power. Classical Receptions Journal deposits 185 abstracts in 405 works, and 37 of its articles are open under Creative Commons licences (#1770 §0.2). Ancient Philosophy's abstracts are readable for 2022–26 (#1773 §4.6). Classical Philology deposits 0 abstracts and 0 licences in 11,206 records; 97% of its works are closed (§8). By its own rules nothing may be posted before submission, copyright passes to the Press, and the decision is one anonymous reader's (§8.2): "the least inspectable. by design" (the author, I.1). The field with the least power over the object is the most open, and the field with the most is the most closed.
Philology is also the field with the instruments to reopen the object: manuscripts, papyri, the language. The author's prediction for it (§0): "less apparent flattening - it has at least to attend to language - but that it will still be there, molecular reach that holistically preserves the molar." After Classical Philology the census goes to more journals (§9.4). Biblical criticism follows as its own station, and the crossings of philology there are taken up only after both are done (the author's ruling: "keep biblical separate from the current question, and after we've done both separately we'll look at the combine").
I.6 What the station finds. The findings are stated at the close, after the evidence (§10). In brief: the question of what the object is is opened at the locus and seldom at the corpus, in every body read; relations are admitted and used to secure the received object; verdicts on the whole mostly state no ground where they are given; decisions about texts are cited less than the journal's other work, behind a record OpenAlex classes 97% closed.
I.7 What it does not claim. It does not claim that philology's work at the locus is wrong or idle; that work is the discipline's real reach, and the census measures it as such. It does not claim intent. It measures incentive and function: where a question is permitted, by what kind of argument, and what reaches the public.
0. The prediction
The author (16 November 2026): "i imagine, that philology will have less apparent flattening - it has at least to attend to language - but that it will still be there, molecular reach that holistically preserves the molar."
Stated as measures:
- the questions about the text itself (D) stay present at the grain of the locus: emendation, crux, reading, colometry;
- they thin at the grain of the work: authenticity, date, unity, transmission;
- they are near zero at the grain of the corpus: the partition of a corpus among composing persons, the author as unit;
- where the corpus grain is touched, it is touched by a locus and resolved toward the received partition.
A blind parallel census is run by ChatGPT on the same items (§3).
0.1 The gate's rule, read from the report
The author, reading the Classical Philology report on ΦΑΙΝΕΤΑΙ ΜΟΙ (16 November 2026, 15:01 EST): "any change of object must go through atoms. and overall coherence is not valid to weigh back in on the adjudication of atoms for that purpose."
The report bears this out. It makes four objections: κῆνος, the addressee, the anachronism of γράμματα and papyrus, and πένητα. Each is a single locus or a single fact, and each is adjudicated on its own. None takes up the order of the catalogue. The convergence of §V, the paper's one argument from relation, gets a single clause of verdict: "neither it, nor the reading it aims to capture, seem to me convincing". No atom is named, and the operator argument is not stated (CP-report-reading §1.3, attached to #1774). The parts are adjudicated one by one; the whole is dismissed without an argument. A reading whose evidence is convergence cannot pass a rule that admits no convergence. The received reading is never put through the same rule: its own atoms are not adjudicated (#1775 §IV.2, the jealousy reading asked to explain its rendering of χλωρός).
0.2 The prediction, extended
The author, 15:03–15:06 EST:
- "relations between atoms are sheared into the atom as commodity form"
- "thats my prediction about what well find in philology"
- "philology is a field that can and does regularly encounter new manuscript and papyrological evidence... it maintains its authority by never exceeding the atom level that new findings could contradict"
- "words are broken off from their community"
- "it is the form that can keep the object stable"
As measured (§6.2 measure 2, §6.3 point 3): the prediction is refined. Relations are not sheared off; they are assigned a direction. Evidence from a relation confirms the received whole (55 to 32), evidence from a locus opens it (68 to 50).
0.3 As measures
1. The atom is the unit a find could overturn. That is the reading at a locus, the date of a work, or the sense of a word at a passage. The field's claims should sit at that grain. D1 should hold steady, and it does in the titles (§4). D2 should hold where a find could decide it, and thin where only argument could.
2. Relations are sheared into atoms. An argument from arrangement should appear, when it appears at all, as one datum among atoms, and not as the evidence that weighs them. Examples are the order of a catalogue, ring composition, the form of a book, and the convergence of readings. To code: per item, whether the argument's evidence is a relation (order, form, convergence) or a locus, and whether a relation is allowed to decide a locus.
3. Words broken off from their community. A word's sense should be settled from its occurrences taken one at a time, sorted into senses, and not from the poem it stands in. The lexicon is the instrument: LSJ files χλωροτέρα … ποίας under "pale, of persons" while its standard, grass, belongs to the first sense. The phrase's two terms are assigned to two different senses so that each can be an atom (#1775 §IV.2). To code: arguments on a word's sense that cite parallels as isolated occurrences, against those that read the word in its passage and form.
4. Form keeps the object stable. If the object is held in place by the form of the received reading, then form is what the atom rule must keep out. An argument from form is the one kind of argument that could move the object without a new find. The prediction is that such arguments are rarest of all at the corpus grain, and that where they occur they are answered atom by atom. The report on ΦΑΙΝΕΤΑΙ ΜΟΙ is the first case.
The asymmetry to test throughout (#1766 §5.2.3): whether the received reading is held to the atom rule it applies.
1. The census
Source. Crossref, journal ISSN 0009-837X, pulled 16 November 2026: 11,206 records, 1906–2026 (census-cp.json, with the raw sha256).
Abstracts. Crossref holds none for this journal (0 of 11,206). Reach coding is therefore from titles, and the limits of that are stated in §4.
Filter. Journal articles only. Reviews are removed (titles opening on an italic book title, 5,525 of them). So are the front matter, back matter, "Books Received", volume information and the like (1,010). Untitled items are dropped.
Kept: 4,252 items (cp-items.json, ids CP0001–CP4252 in date order). Items of 10 or more pages are counted as articles; shorter items are notes. In philology the note is where much locus-level work is published, so notes are kept.
| Decade | Items | of which ≥ 10 pp |
|---|---|---|
| 1900s | 190 | 57 |
| 1910s | 441 | 175 |
| 1920s | 443 | 130 |
| 1930s | 403 | 136 |
| 1940s | 324 | 96 |
| 1950s | 284 | 51 |
| 1960s | 314 | 51 |
| 1970s | 415 | 82 |
| 1980s | 320 | 108 |
| 1990s | 259 | 132 |
| 2000s | 263 | 158 |
| 2010s | 293 | 176 |
| 2020s | 303 | 190 |
The journal moves from the note to the article. Long items rise from about a third of items to about two-thirds after 1990.
2. The codebook: reach for philology
The categories are those of #1773 §4.6, kept for comparability. D is split by grain, and two categories are added for work philology does that philosophy does not.
- A, locus. Re-reads a passage for its sense.
- B, repair. Resolves an apparent inconsistency or problem inside an author.
- C, reconstruction. Reconstructs a doctrine, institution, practice, event or context from texts.
- D1, text at the locus. Emendation, conjecture, crux, reading of a manuscript at a passage, lacuna, colometry, interpolation of a line or passage.
- D2, text at the work. Authenticity, authorship, date, unity or composition of a work; its manuscript tradition or recension; an edition.
- D3, text at the corpus. The partition of a corpus among composing persons. Whether names in a corpus are one maker or several, and whether the author is the right unit.
- E, relation. Sources, imitation, allusion, influence, intertext between authors.
- F, after antiquity. Reception and transmission beyond antiquity, as an object of study.
- G, position. Argues a thesis not anchored to an author or text: method, theory, a general claim.
- L, language. Grammar, lexicon, metre, accent, dialect, style as language.
- M, document. Publishes or reads an inscription, papyrus or object as a document.
For each D code, also mark confirm or open. Confirm means the title settles toward the received standing; open means it puts the standing in question.
3. The blind parallel run
ChatGPT codes the same items on the same codebook, without seeing any code made here. Agreement is then measured. This is the cross-family check that #1773 §7.5 asked for, since every reach code so far has been made within one model family.
Two files go to it:
cp-titles-sample390.csv: a stratified sample, 30 per decade, seed 31;cp-titles-all.csv: all 4,252 items, if it can take them.
The prompt is in the reply, ready to paste. The sample's ids are recorded (sample390-ids.json) so that both coders' work can be joined.
3.1 The blind run, returned (16 November 2026, 15:25 EST)
ChatGPT coded the 390-item sample on the codebook of §2 without seeing any code made here. The file is blind/cp-sample390-chatgpt-coded.csv: one code per item, a direction for each D code, and a span. The join is complete: 390 of 390 ids match the sample, and every span is a substring of its title.
The codes, all decades:
| Code | Items | Share |
|---|---|---|
| C, reconstruction | 129 | 33.1% |
| A, locus | 87 | 22.3% |
| L, language | 59 | 15.1% |
| D2, text at the work | 31 | 7.9% |
| E, relation | 23 | 5.9% |
| G, position | 19 | 4.9% |
| M, document | 17 | 4.4% |
| D1, text at the locus | 12 | 3.1% |
| F, after antiquity | 11 | 2.8% |
| B, repair | 2 | 0.5% |
| D3, text at the corpus | 0 | 0% |
The result that bears on the prediction. An independent coder from another model family found no question at the corpus grain in 390 titles across twelve decades. That matches the lexicon pass (§4) and the prediction of §0, that the corpus grain is near zero. The sample holds none of the 35 corpus-grain candidates of d3-candidates.json. The lexicon found those in the whole run, and each is a question about one work, so zero in the sample is the expected count.
D2 by decade (of 30 items each): 1900s 4, 1910s 5, 1920s 2, 1930s 3, 1940s 2, 1950s 4, 1960s 1, 1970s 2, 1980s 4, 1990s 1, 2000s 2, 2010s 0, 2020s 1. That is about one title in ten to 1959, and one in twenty-five after. The shape agrees with the lexicon's: the work grain thins.
Where the blind coding cannot be used as it stands:
- Direction. 42 of the 43 D codes are marked "open", and one is "confirm". The direction was assigned without reading what the article concludes, which a title alone seldom shows. The confirm/open field is to be set aside for the sample and decided on the page, from the seated volumes (40/01) for 1906–1922.
- Reviews and notices coded D2. "Recent Loeb Library Volumes" (CP0383, 1914) and "The New Critical Edition of Ovid's Metamorphoses" (CP0496, 1916) are coded D2. A review of an edition is not a question of a work's standing. The review filter (§1) let them through, and the coder treated "edition" as D2.
- Obituaries coded C. Morgan (CP0221), Bennett (CP0743) and the other memorial notices are coded C. They belong to the front and back matter that §1 meant to remove. The filter is to be audited on them.
- Possible misses. "Speeches VIII and X of the Demosthenic Corpus" (CP1417, 1938) is coded C. Its title singles out two speeches of a corpus, a frame in which the question is usually authenticity, which would make it D2; that is to be checked on the page. "Zeno, Not Heraclitus" (CP2588, 1974), an attribution, is coded D2, and that is right.
Next. The hand coding of the 390 (§9.1) was not made before the blind file arrived. This session has read the blind file, so a hand coding made by this session would measure agreement with it, not independence from it. The hand coding reported in §3.2 was therefore made by a separate reader, in a separate session, without access to the blind file; this note is kept as written at the time. The independent check the run provides is the D3 count and the D2 shape, which the lexicon pass (made before the run) also gives. For the confirm/open field the instrument is the page: read the 43 D items, and the 1906–1922 items whole, in 40/01.
3.2 Agreement
The 390, by hand against blind. The sample of 390 titles was coded by hand on §2 (one reader of this session's model family, without access to the blind file; sample390-hand.jsonl) and set against ChatGPT's blind run on the same titles (blind/cp-sample390-chatgpt-coded.csv).
- Primary code: 296 of 390 agree (76%), Cohen's κ 0.71.
- A text question as primary: both readers 38; the hand reader alone 15; the blind reader alone 5. The blind run's D is a slight undercount, and its shape holds.
- The main disagreement is the line between A and C (37 items: a passage re-read, or a context reconstructed from it). It does not touch D.
- The corpus grain: the hand reader codes one D3, "Was Carcinus I a Tragic Playwright?: A Response" (whether a name covers one tragedian or two); the blind reader none. Both readers otherwise find the corpus grain empty in the titles, as §3.1 reported.
The audit, blind, across five bodies. 200 items, 40 drawn at random from each body already coded, were recoded by a reader who had not seen the codes (audit/; key and outputs kept). Every span checked: 0 failures.
| body | items | primary agrees | text question found by both | original only | audit only | direction agrees |
|---|---|---|---|---|---|---|
| CP reviews 1906–22 (page) | 40 | 30 | 16 | 1 | 1 | 11 of 16 |
| CP articles 1906–22 (page) | 40 | 31 | 15 | 3 | 1 | 10 of 15 |
| CP abstracts 2001–26 | 40 | 38 | 7 | 0 | 0 | 6 of 7 |
| GRBS summaries | 40 | 32 | 7 | 0 | 2 | 6 of 7 |
| GRBS first pages | 40 | 31 | 4 | 0 | 7 | 4 of 4 |
| all | 200 | 162 (81%) | 49 | 4 | 11 | 37 of 49 (76%) |
Cohen's κ on the primary code, all 200: 0.78. Where both find a text question, the opens-to class agrees in 12 of 14 for the contemporary and GRBS bodies. For 1906–22 the class was assigned in §6.4 by this session, not by the readers, so it is not compared here.
What the audit settles and what it does not.
1. The codes hold at the grain the findings use. Primary agreement of 81%, κ 0.78, is substantial agreement across five bodies and three units.
2. The direction calls are the weaker field. 76% overall, about 70% on the 1906–22 pages, where the direction often has to be read from a passage rather than a stated conclusion. Findings that rest on direction counts (§6.2's gradient, §6.4's flattening inside the run) carry that weakness. Agreement of 76% is not an error bound, and the gradient's width is not measured against it here; the disagreements by grain are to be read (§9.6).
3. The GRBS first pages undercount the text question. The audit reader finds 7 text questions the original readers did not, against none the other way. §8.1's D for 1970–1999 (page unit) is a floor: 17.8%, 15.1% and 12.1% may run a few points higher. The decline across those decades would be softened, not reversed.
4. The corpus grain agrees. G2131 (Nonnus) is D3 for both readers. One review of 1906–22 (jstor-262533) carries D3 as a secondary for the original reader and not for the audit.
The conviction codes, audited blind. 150 of the sentences of §8.3 (50 from CP, 100 from BMCR), drawn at random, were recoded by a reader who had not seen the codes (conviction/audit-conv-*).
| field | agree | κ |
|---|---|---|
| polarity | 95% (150) | 0.91 |
| who judges | 100% (123 judgments) | 1.00 |
| whole or part | 94% | 0.83 |
| ground: atom, argument, bare | 72% | 0.57 |
The ground is the weak field, as the readers said. The disagreement runs one way: 13 sentences the original readers grounded in an atom the audit calls bare, against one the other way. If anything the atom is over-credited in §8.3, which strengthens its first finding. On the 22 judgments of a whole that both readers agree are judgments of a whole, one rests on an atom, and the ground agrees in 18.
The BMCR and CQ codes, audited blind. 100 items, 50 from each, recoded by a reader who had not seen the codes (audit/audit2-*).
| body | items | primary agrees | κ | text question by both | original only | audit only | direction agrees |
|---|---|---|---|---|---|---|---|
| BMCR reviews | 50 | 46 | 0.88 | 13 | 0 | 0 | 13 of 13 |
| CQ extracts | 50 | 40 | 0.76 | 14 | 3 | 2 | 12 of 14 |
The two bodies coded last hold as well as any coded before.
4. A first pass, and its limits
What it is. A lexicon over the titles, run before any reading, to see the shape. It is not the census's coding. The census's coding is the reading of §2, done by readers and adjudicated, with the blind run as its check.
Textual and standing vocabulary in titles, by decade, as a share of items:
| Decade | D1 terms | D2 terms | either |
|---|---|---|---|
| 1900s | 5.8% | 8.9% | 14.2% |
| 1910s | 4.8% | 5.4% | 10.0% |
| 1920s | 2.9% | 5.4% | 8.1% |
| 1930s | 3.7% | 7.9% | 11.7% |
| 1940s | 4.3% | 4.0% | 8.3% |
| 1950s | 1.8% | 6.3% | 8.1% |
| 1960s | 1.9% | 5.7% | 7.3% |
| 1970s | 3.9% | 4.8% | 8.7% |
| 1980s | 3.4% | 5.6% | 8.8% |
| 1990s | 5.8% | 2.3% | 8.1% |
| 2000s | 3.4% | 1.9% | 5.3% |
| 2010s | 4.1% | 2.4% | 6.1% |
| 2020s | 4.0% | 4.6% | 7.9% |
The shape, provisionally.
- The locus grain holds. D1 vocabulary runs at 2–6% of titles in every decade, with no trend.
- The work grain thins and returns. D2 vocabulary sits at 5–9% before 1990, drops to about 2% in 1990–2019, and comes back to 4.6% in the 2020s.
- Nothing like the Archiv. There, standing titles go from 19% in the 1890s to 0% by the 1910s (#1773 §4.12). This matches the prediction's first clause: less apparent flattening.
The corpus grain, read by hand. The 35 titles the D3 lexicon flagged, all of them, are read in d3-candidates.json.
- Single works. Every authorship question among them is about one work or one passage: the Moretum (1925), the Laus Pisonis (1929), the Pervigilium Veneris (1929), the Epistola de Indicis Gentibus (1926), Anthologia Latina 899 (1985), [Arist.] De spiritu (2015), Euripides' Electra 518–44 (2024).
- One case merges names into one maker: "The Ovidian Authorship of the Lygdamus Elegies" (1927). Lygdamus is a name inside the Tibullan corpus, with no attestation outside it. That fits the line #1773 §0A.6a draws. The field admits a merger when the second name is only a name; it has no case of persons attested apart.
- One case reaches the Socratic partition, at a locus: "On Separating the Socratic from the Platonic in Phaedo 118" (1975, 2 pp). The molar question is taken up at one passage, and the operation is separation.
- One title states the direction: "Authorship Analysis and the Authenticity of Euripides' Electra 518–44: Preserving Character Consistency" (2024).
- So, by title, the partition question appears only as a work question. Who wrote this poem, this passage, this letter. On the page it appears at the corpus as well (§6: Homer, the Historia Augusta, the Demosthenic corpus), and there it resolves to unity.
Limits.
- Titles under-report D. A note titled with a locus alone ("Lysias 19. 22 and 18. 14") may be an emendation. The lexicon counts it as nothing, and the reading of §2 has to decide it.
- Crossref's typing of reviews is unreliable, and the italic rule may admit a few reviews or drop a few articles. Both filters are to be audited on the sample.
- The full text of the volumes before 1928 is public domain. Those volumes can be read whole, as the Archiv was (#1773 §4.12g), and that is the next measure for the early decades.
4.1 All 4,252 titles
The 3,862 titles outside the sample were coded on §2 from the title alone by two readers (coding-titles/), and joined with the 390 of §3.2: every title of the journal's articles and notes, 1906–2026, coded, every span checked (SHA-256 e24bbdefb2e3c2a1afce840438af058d6f08050ac2ff6ca78992ec72e6d2a733). The unit is the title, so the direction is mostly unseen and the figures are floors (§3.2: the blind run on titles undercounts D a little against the hand reader).
| titles | a text question | D1 | D2 | D3 (primary or secondary) | A | C | E | L | G | |
|---|---|---|---|---|---|---|---|---|---|---|
| 1906–19 | 631 | 16.6% | 5.9% | 8.9% | 5 | 23.9% | 21.2% | 3.6% | 24.9% | 4.1% |
| 1920s | 443 | 11.7% | 1.8% | 9.0% | 5 | 23.3% | 30.5% | 5.4% | 14.9% | 6.5% |
| 1930s | 403 | 13.4% | 4.0% | 8.9% | 0 | 22.8% | 35.2% | 5.5% | 9.7% | 2.2% |
| 1940s | 324 | 8.3% | 2.8% | 5.2% | 0 | 22.2% | 30.6% | 5.2% | 18.2% | 3.7% |
| 1950s | 284 | 12.3% | 1.8% | 7.7% | 1 | 31.0% | 26.8% | 6.3% | 12.0% | 4.2% |
| 1960s | 314 | 10.2% | 4.1% | 6.1% | 0 | 42.7% | 26.1% | 6.1% | 5.7% | 2.9% |
| 1970s | 415 | 14.2% | 5.5% | 8.4% | 0 | 39.5% | 20.2% | 9.2% | 9.9% | 3.1% |
| 1980s | 320 | 15.6% | 5.3% | 10.0% | 0 | 37.8% | 25.6% | 6.6% | 7.2% | 4.7% |
| 1990s | 259 | 8.5% | 3.5% | 4.6% | 1 | 42.9% | 28.2% | 6.6% | 7.3% | 2.7% |
| 2000s | 263 | 7.2% | 2.7% | 3.0% | 1 | 33.8% | 29.3% | 12.5% | 6.5% | 8.0% |
| 2010s | 293 | 5.5% | 1.7% | 3.8% | 0 | 34.8% | 24.2% | 9.6% | 7.2% | 13.3% |
| 2020–26 | 303 | 8.9% | 3.0% | 5.0% | 0 | 29.7% | 22.1% | 11.9% | 5.6% | 15.5% |
1. The work question holds by title until the 1980s and halves after them: D2 at 5–10% for eighty years, 3–5% from 1990. CQ shows the same step at the same decade on its extracts (§8.5).
2. The passage read for its sense takes the journal in the 1960s (A, 24% to 43%) and holds it. Language as its own object falls (L, 25% to 6%).
3. Position rises at the end: G, the thesis not anchored to an author or a text, from 3–5% to 13–16% in the 2010s and 2020s.
4. The corpus grain, by title, is one campaign and four single cases. Nine of the thirteen D3 titles are the Homeric question of 1910–1922 (Scott, Shewan, Bolling; the Thebais). Then: the Ovidian authorship of the Lygdamus elegies (1927), "The One and the Many: A Point of Terminology in Homeric Criticism" (1956), Carcinus I (1997), and "Putting Him Back Together Again: Apion Historian, Apion Grammatikos" (2003), two Apions made one.
5. (Not used)
The number is left empty. §§6–9 keep their numbers so that the references to them, in this notebook and in the erratum on #1776 (§6.0), hold.
6. On the page, 1906–1922: the first reading
6.0 What the seat holds (an erratum on #1776)
The full coding found the seat to be something other than what its record said. Of the 775 distinct texts, the readers found 766 to be pages of the Book Reviews section. A search of the Archive's collection for the journal, jstor_clasphil (1,714 items), showed why. The advanced-search listing used for the fetch matched the journal's name and returned mainly the untitled review pages. It missed the 902 articles and notes, which carry their own titles. It also returned nine items from other journals whose titles name this one.
So EA-CORPORA-40/01 is the journal's Book Reviews section, 1906–1922, with one article (Kelsey's "Is There a Science of Classical Philology?", 1908) and the nine strays. An erratum on #1776 says so, and the seat record's title and scope are corrected. The files, their hashes and the seat's AXN stand. The 902 articles and notes are being fetched for a later seat.
What follows is therefore the census of the field's judgment of books: what Classical Philology's reviewers said, 1906–1922, about the books of the field, in English, American and European.
The instrument. EA-CORPORA-40/01 holds 821 items as JSTOR served them. See §6.0 for what they are. The Early Journal Content splits the review pages, so adjacent items often carry the same pages. Counted by distinct text, the run is 775 texts. Every figure below is counted on the 775.
A lexicon pass over the page, made to find the places to read, not to code:
| Grain | Vocabulary | Texts | Share |
|---|---|---|---|
| D1, locus | emend-, conjectur-, lacuna, interpolat-, corrupt-, "the MSS read", crux | 291 | 38% |
| D2, work | authentic-, spurious, genuine(ness), authorship, athet-, forger-, "date of composition" | 182 | 23% |
| D3, corpus | "same author / hand / poet", "one and the same", "two / different authors / hands / poets", "single author", "many hands", "unity of authorship" | 48 | 6% |
The corpus grain, read by hand. All 48 D3 hits were read in context. Most are not corpus questions: scribes' hands in a codex, "the same author's" other book in a review, vase-painters' hands, or "one and the same" in another sense. Eight distinct texts take up the partition of a corpus among composing persons:
- Homer, five texts. The chorizontes answered: "world-literature offers no other example of two great poems by different authors so nearly alike as are the Iliad and Odyssey" (jstor-261652, 1915). "Does it seem credible that two poets separated by a century or more should be one…" (261782, 1915). The Telemachus as "the creation of the same poet who conceived the Telemacheia" (261670, 1907). The Odyssey as "a fusion of distinct stories … by a single author who is simply 'Der Dichter,' not Melesigenes-Homer" (263244, 1917). And one open case: the poems "composed neither by a single author nor by a crowd, but by an organized group of poets" (262045, 1908).
- The Appendix Vergiliana. The Dirae and Lydia "the work of a single author, who is not Vergil" (261278, 1910): one name divided, two poems merged.
- The Historia Augusta. "the reliability of the manuscripts in attributing the lives to one and the same author — namely, Lampridius" (261523, 1912).
- The Demosthenic corpus. Speeches long "regarded as spurious", "at least two of these … by the same author, but the identity of this orator remained unknown" (261810, 1908).
What the page shows that the titles did not. The title lexicon found the corpus grain near zero, and so did the blind run. On the page, in 1906–1922, it is present: eight texts in 775, about one in a hundred. The question is live in Homer, Wolf's case, and in three late corpora. Most of the eight are reviews of European books, carried into the journal as reports (to be confirmed item by item). The direction is the prediction's. Where the partition is argued, it resolves toward the received unity: the Iliad and Odyssey one, the Odyssey one poet's. Where a partition is accepted, it falls at the work and under a name that is only a name: the Dirae and Lydia taken from Vergil. The one open case, the "organized group of poets", is reported, not argued.
6.1 The full coding on the page (the review section, 1906–1922)
How it was made. Ten readers coded the 775 distinct texts on §2. Each text was given as its opening, every passage around text-question vocabulary, and its close. Every span and every direction span was checked by machine against the full page with whitespace collapsed, and all 775 passed with no span lost. The codes are in coding/cp-reviews-1906-1922-coded.json. Where a page carries the end of one review and the start of the next, the readers coded the review that begins on it.
Primary codes:
| Code | Texts | Share |
|---|---|---|
| C, reconstruction | 198 | 25.5% |
| D2, text at the work | 159 | 20.5% |
| L, language | 111 | 14.3% |
| A, locus | 62 | 8.0% |
| E, relation | 58 | 7.5% |
| G, position | 57 | 7.4% |
| M, document | 57 | 7.4% |
| F, after antiquity | 36 | 4.6% |
| D1, text at the locus | 24 | 3.1% |
| B, repair | 7 | 0.9% |
| D3, text at the corpus | 6 | 0.8% |
A text question at some grain, primary or secondary, is in 259 of 775 (33%). Read from the conclusions, 137 confirm the received text or standing, 70 put it in question, and 52 do both.
The corpus grain: 17 texts (2.2%), primary or secondary, every one read here:
- Homer, ten. Wecklein's Urilias among popular poets (1907). Bréal's "organized group of poets" (1908). Lang, one maker against "a committee" (1909). Fick's Odyssey among several poets and an interpolator (1911). Shorey on Murray's many bards (1911). Drerup, division by the treatment of the gods "absurd" (1914). Dachs, the analysts' contradictions resolved (1914). Scott against Finsler's two poets (1915). Belzner answering the chorizontes (1915). Finsler's partition reported (1917).
- The Simonidean epigrams (Boas, 1907): few under the name held his.
- The two Timocles (Wagner, 1907): the tragic and the comic poet distinct makers.
- Demosthenes and Anaximenes (Nitsche, 1908): speeches, prooemia and letters of the Demosthenic corpus given to Anaximenes.
- The Appendix Vergiliana (Curcio, 1910): the Dirae and Lydia given to one author who is not Vergil.
- Pierre de Paris (Thomas, 1919): two works under one name joined as one maker, by style and biography.
- Socrates (Shorey on Dupréel, La Légende socratique, 1922): Socrates a literary invention, Plato's arguments given to Hippias and Prodicus. Shorey rejects it.
- The letters of Julian (Bidez and Cumont, 1922): letters moved between the emperor and the sophist of Caesarea.
Direction at the corpus grain. Where the reviewer argues the conclusion in the journal's own voice, it goes to unity. Murray, Finsler, Drerup, Dupréel, Dachs, Belzner and Lang all end in one maker, or in the received Socrates. Where a partition stands, it is the book's, reported: Wecklein, Boas, Fick, Nitsche, Curcio. The one merger that stands joins two works under a name that is only a name (Pierre de Paris), the Lygdamus pattern of §4. The Socratic partition is in the journal once, in 1922, in a review, and it is refused. That is the first observation of the case #1773 §0A makes, a hundred and four years before Socrates as Orthonym (#123, 2026).
Measure 2, relations and atoms (§0.3). Of the 259 text questions, the evidence is a locus in 86, a relation among parts (order, form, style across a whole, convergence) in 75, mixed in 73, and a new find in 25. A relation is allowed to decide a passage or an attribution in 71 (27%). In this period, then, relations still decide atoms in about a quarter of the field's judgments of books. That is the period before the flattening, as in the Archiv (#1773 §4.12). The prediction is about the field's present, and the measure has to be run on the later decades to test it. They are not open after 1928, and the 1906–1922 articles are still to be read.
Measure 3, words and their community. Of the 61 judgments of a word's sense, 36 read the word in its passage and form, 20 cite parallels one by one, and 5 do both. In the reviews of this period, the word is still read in its community more often than not.
What the review section shows, against the prediction.
1. The corpus grain is present and rare: 2.2% of the field's judgments of books, nearly all of them Homer.
2. It resolves toward the received unity where the journal speaks, and stands only as a reported thesis or a merger under a mere name.
3. Relations still decide atoms in a quarter of cases, so the shearing of relations into the atom is not yet the rule in 1906–1922. If the prediction holds, it holds as a change over time, to be measured.
6.2 The journal's own voice: the articles and notes, 1906–1922
What was read. The 902 articles and notes of the collection jstor_clasphil missing from 40/01 were fetched (900; two return a server error at the Archive). Counted by distinct text they are 876: 412 articles, 295 notes, 102 titled reviews and 67 notices. They were coded on §2 by ten readers, as in §6.1. Every span and direction span was checked against the full page, and all 876 passed. The codes are in coding-art/cp-articles-1906-1922-coded.json. These texts are seated as EA-CORPORA-40/17 (EA-CORPORA-40 v1.1, #1782, 17 November 2026): 900 items as served, with the codes of §§6.1–6.4, the conviction coding and the 390 sample under original/codes/. Two items (jstor-261959, jstor-262508) returned a server error at the Archive and are pointed to.
Primary codes:
| Code | Texts | Share |
|---|---|---|
| C, reconstruction | 183 | 20.9% |
| L, language | 179 | 20.4% |
| D1, text at the locus | 123 | 14.0% |
| D2, text at the work | 93 | 10.6% |
| A, locus | 92 | 10.5% |
| G, position | 80 | 9.1% |
| E, relation | 39 | 4.5% |
| M, document | 36 | 4.1% |
| F, after antiquity | 27 | 3.1% |
| D3, text at the corpus | 19 | 2.2% |
| B, repair | 5 | 0.6% |
A text question at some grain, primary or secondary, is in 308 of 876 (35%). The share falls across the period: 41% of texts in 1906–1910, 35% in 1911–1916, 29% in 1917–1922.
Direction by grain. This is the measure the prediction of §0 names.
| Grain | Texts | Open | Confirm | Both | Share open |
|---|---|---|---|---|---|
| D1, locus | 181 | 91 | 72 | 18 | 50% |
| D2, work | 134 | 39 | 74 | 21 | 29% |
| D3, corpus | 23 | 4 | 18 | 1 | 17% |
The larger the grain, the more the journal confirms. At the locus, half its text questions open the received reading. At the work, fewer than a third open it. At the corpus, one in six does, and each of those is answered. That is the author's prediction in measured form: "molecular reach that holistically preserves the molar."
The corpus grain in the journal's own voice: 23 texts.
- Homer, nineteen. Seventeen are by two men, every byline read on the page. John A. Scott (Northwestern) wrote eleven: "Odyssean Words Found in but One Book of the Iliad" (1910); "Words Found in the Iliad and in but One Book of the Odyssey" (1911); "Two Linguistic Tests of the Relative Antiquity of the Iliad and the Odyssey" (1911); "Patronymics as a Test of the Relative Age of Homeric Books" (1912); "Nestor's Son Peisistratus in Homer" (1911); "Paris and Hector in Tradition and in Homer" (1913); "The Assumed Duration of the War of the Iliad" (1913); "Non-Odyssean Words Found in the Iliad" (1918); "Some Tests of the Relative Antiquity of Homeric Books" (1919); "Homer as the Poet of the Thebais" (1921); "The Archaic Infinitives in Homer" (1922). Alexander Shewan (St Andrews) wrote six: "Wilamowitz on Θ" (1911); "Suspected Flaws in Homeric Similes" (1911); "Recent Homeric Literature" (1912); "The 'Continuation' of the Odyssey (C. The Nekyia)" (1914); "The Oneness of the Homeric Language" (1915); "ΠΟΘΗ and ΠΟΘΟΣ in Iliad and Odyssey: Another Chorizontic Failure" (1921). Every one argues one poet. (v0.6 had eighteen, "at least nine" and "at least five"; the bylines correct it.)
- The one separatist voice is George Melville Bolling: on πόθη and πόθος (1920), and on the infinitives in -έμεν (1921). Both times the journal printed the answer, Shewan's "Another Chorizontic Failure" (1921) and Scott's reply (1922). The corpus question is in the journal's own voice twice in seventeen years, and both times as a position the journal answers.
- The Caesarian corpus, two notes on Hirtius and Book VIII of the Gallic War: Francis W. Kelsey (1907), and a reply to him (1908) whose signature is not on the seated page. The diversity of style is explained, and the received division kept.
- The Demosthenic corpus (a review of Nitsche, 1908, its signature past the seated page) and the Historia Augusta (Orma Fitch Butler's Michigan dissertation, reviewed by Walter Miller of Missouri, 1912): partitions reported from books, as in §6.1. Bylines: Scott 11, Shewan 6, Bolling 2, Kelsey 1, Miller 1, and two not on the page.
Measure 2, relations and atoms. A relation among parts decides a passage or attribution in 105 of 308 (34%). The prediction is that relations are kept out. They are not kept out. They are put to work, and the direction shows for what:
| Evidence | Open | Confirm | Both |
|---|---|---|---|
| a locus | 68 | 50 | 10 |
| a relation among parts | 32 | 55 | 7 |
Evidence from a locus opens the received text more often than it confirms it. Evidence from a relation confirms it, by 55 to 32. Relations are admitted chiefly to hold the received whole together: Scott's vocabulary distributions, Shewan's oneness of the language. Openness lives at the atom. This is the prediction refined by what the page shows. The relation is not sheared off; it is assigned a direction.
Measure 3, words and their community. Of 151 judgments of a word's sense in the journal's own voice, 77 cite parallels one by one, 33 read the word in its passage and form, and 41 do both. The reviews ran the other way, 20 to 36 (§6.1). In its own articles, the journal breaks the word from its community more than twice as often as it reads it there.
6.3 The two set side by side
| Book reviews (775) | Articles and notes (876) | |
|---|---|---|
| A text question at some grain | 33% | 35% |
| Locus questions that open | 18% (14 of 77) | 50% (91 of 181) |
| Work questions that open | 28% (60 of 215) | 29% (39 of 134) |
| Corpus questions | 17, 2.2% | 23, 2.6% |
| Corpus questions that open, in the journal's own voice | none | Bolling's two, each answered |
| A relation decides a locus | 27% | 34% |
| Word sense from parallels vs. from the passage | 20 vs 36 | 77 vs 33 |
What the station finds, 1906–1922.
1. The gradient holds in the journal's own voice. Open at the atom, confirming at the work, closed at the corpus. The question at the corpus grain is a campaign: one view argued by two writers, against one dissenter who is answered each time.
2. The partition of a corpus among persons enters the journal from outside, as books reviewed: Wecklein, Fick, Bréal, Nitsche, Butler, Dupréel. In the reviewer's own conclusion it goes to unity.
3. Relations are admitted to confirm and atoms to open. The shearing the author predicts is a division of labour. The atom carries the openness, the relation carries the received whole, and the corpus is held by the relation.
4. In its own articles, the journal reads words apart from their community, by parallels, more than twice as often as in their passage. Its reviewers did the reverse.
Limits. One model family coded both sets. The spans are verified, but the codes are not adjudicated, and the directions are readers' judgments from the extract. The ten readers' calls are in their notes and are to be audited on a sample against the full page. The bylines of the corpus grain are read on the page, 21 of 23; two signatures fall past the seated page. This is one journal, in one period, the period before the flattening (#1773 §4.12). The later decades are not open.
Next on the page.
1. Coded: the review section (§6.1) and the articles and notes (§6.2); measures 2 and 3 taken (§6.2); the openings classed (§6.4); the corpus bylines read. To audit a sample of the codes and of §6.4's classes against the full page; to seat the articles (a later seat).
6.4 What the openings open to
The question (the author, 19:02 EST): "what direction are the openings in? what do they open up to?" §6.2 counts a text question as opening when it puts the received text, standing or attribution in question. It does not say toward what. Each of the 160 opening items (open or both; openings-classed.json holds 160) was read here for what it puts in place of the received text: from its title, its direction span and the reader's note, against the page where those were unclear. Rulings are in openings-classed.json; the reader is one (this session), and the class is not yet audited.
| What the opening puts in place | Items | Share |
|---|---|---|
| one better reading at the locus: an emendation, a punctuation, a transposition | 78 | 48% |
| the witnesses re-ranked: which manuscript or edition governs | 19 | 12% |
| a date moved | 14 | 9% |
| the transmitted text defended against the editors' emendation | 11 | 7% |
| a line removed: an interpolation, a gloss, a spurious verse | 9 | 6% |
| a work or document removed: spurious, forged, a later invention | 9 | 6% |
| something supplied: a lacuna filled, a new fragment for the author | 7 | 4% |
| a work's composition: layers, a second version, book division, contamination | 7 | 4% |
| the makers: a corpus divided or reassigned among persons | 6 | 4% |
The direction is inward, to the author. Nearly every opening opens toward one person's own text. The emendation restores what the author wrote: "what Tacitus originally wrote was claro ore" (1908); "Plautus used facere incendium" (1910); "surely Aeschylus did not write" (1916). It is decided by the author taken whole: "Caesarian usage" (1908), "the author's excessive fondness for symmetry" (Apuleius, 1911), "Platonic usage and thought" (1915), Stoic doctrine across Chrysippus' fragments (1911), "I have no doubt that Virgil wrote the Culex", with Virgilian style then deciding the loci (1910). The relation that §6.2 found confirming the received whole is here the decider of the atom: the author's usage, taken as a unity, closes each locus it opens. One note emends the author against himself. Paul Shorey, on Laws 795b (1916): "The only escape I can see from the emendation is to assume that Plato himself by inadvertence wrote [ἀδύνατος]… And if by an oversight Plato wrote [ἀδύνατος], it is, I think, probable that he meant [δυνατός]." (The Greek words are read through the page's OCR and bracketed as such.) The author is corrected to the author: what he meant governs what he wrote. The removals face the same way. A line or a work comes out because it is not his, and what remains is more purely his. The re-ranked witnesses face the archetype, the one original behind the copies.
Counted by where they arrive:
- at one author's text, restored, supplied, defended or purified: 114 of 160 (71%: E, S, T, X, A);
- at the tradition's one original: 19 (W);
- at a date or at the composition of a work by its one author: 21 (D, K). Of the seven K items, Plautine contamination puts Greek originals behind the Rudens and the act-division, and Mulder's Iliad and Ausfeld's Alexander Romance are layered. Each is reported as one maker reworking, or as accretion on an original.
- at more than one maker: 6 (P). Nitsche's Anaximenes and Butler's plural Vita Heliogabali are reported from books. Bolling's two notes and Drewitt's augment statistics are each answered. Croiset's statistics are defended with "the merits expressly set aside".
So the openings do not open the unit. They open the atom inside the unit, and close it on the author. The author is the one constant every opening assumes, and the instrument by which it is decided. That is the corpus grain of §6.2 seen from below: the corpus question is closed at the top because every question beneath it is decided by the unity the corpus question would put in doubt.
The flattening inside the run. By period, on the same 876 texts:
| 1906–10 | 1911–16 | 1917–22 | |
|---|---|---|---|
| texts | 283 | 311 | 282 |
| a text question at the locus (D1) | 27.6% | 18.3% | 16.3% |
| a text question at the work (D2) | 16.6% | 16.4% | 12.8% |
| D items that open (open or both) | 62% (72 of 117) | 46% (50 of 108) | 46% (38 of 83) |
| openings that remove a line | 7 | 1 | 1 |
| openings that re-rank the witnesses | 13 | 5 | 1 |
| openings that are one better reading | 32 | 27 | 19 |
| openings that remove a work or document | 2 | 2 | 5 |
| openings at more than one maker | 1 | 2 | 3 |
1. The text question contracts. The locus falls from 28% to 16% of the journal, the work from 17% to 13%.
2. Athetesis leaves. Removal of lines inside an author, seven times in the first five years, appears once in each later period. #1773 found athetesis an ordinary article form in the Archiv of the 1890s and gone from it by the 1920s (#1773 §4.12, 4.12g). In CP it goes in the 1910s.
3. The manuscript tradition leaves. The re-ranking of witnesses runs 13, 5, 1. It was carried by a few series: Aristophanes, Catullus, Livy's third decade, Pliny's Letters. The series end and nothing replaces them.
4. What remains is the emendation. One better reading at the locus is 44% of the openings in 1906–10 and half of them afterwards. This is the residue #1773 §4.12b names, an emendation "at the smallest unit where the edition may still be questioned", and in CP it is the residue by 1917.
5. The question of a work's standing does not collapse in this run. D2 falls modestly. Removals of a work rise, but the works removed are small or documentary: oracles of colonization, the Elogium Duilianum, Demosthenes' Epistles V and VI, Book III of the Vitae patrum. The makers question rises from one to three, and each of the three is answered. The flattening here is of the operations that act inside an author (athetesis, the tradition), not yet of the author's standing, which the journal had never opened.
Limits. The counts are small, and the period tables move by single items. The class was ruled by one reader from the direction span, the note and the page where needed, and is not audited. The series effect (point 3) means some of the fall belongs to particular projects ending, not to the journal's form.
6.5 On the page, 1923–1930: the run continued
What was read. Volumes XVIII–XXV, public domain in the United States (published before 1931), are at the Internet Archive as the microfilm text of each issue (collection pub_classical-philology, items sim_classical-philology_YYYY-MM_VV_N): 32 issues, fetched 17 November 2026 with the SHA-256 of each served file (gather/cp-1923-1930/). A heuristic cut them into 394 items at the article heads ("By NAME"), the headings of Notes and Discussions and the signatures of Book Reviews (coding-cp2330/segment.py); the readers set the kind from the page and flagged run-on segments. Coded on §2 with the opens-to classes by two readers, on the instrument of §6.1; every span checked against the issue text, 0 failures (SHA-256 908edd0d79daf5834d9183aa1661b47836b168c685ce3a1e7d2714cf6fff3d6b). The cut undercounts the reviews (63, several run together); the articles (178) and notes (140) are cut cleanly.
| items | text question | of which open or both | |
|---|---|---|---|
| articles | 178 | 50 (28%) | 33 |
| notes | 140 | 62 (44%) | 42 |
| reviews | 63 | 19 (30%) | 11 |
By grain (codes primary or secondary): the locus 54, open or both 69%; the work 74, open or both 61%; the corpus 11, open or both 36%.
1. The gradient holds. Open at the locus, less open at the work, least at the corpus, as in 1906–22. On the same measure (open or both), §6.2's figures are 60.2%, 44.8% and 21.7% (109 of 181, 60 of 134, 5 of 23); its table's 50%, 29% and 17% count open alone.
2. The whole run is more open than its first stretch. At each grain the openings are a larger share than in 1906–22, on like figures: locus 69% against 60%, work 61% against 45%, corpus 36% against 22%. The work question is most changed. The notes carry it, as they carried the locus before (29 of the 54 locus codes are notes that open).
3. The corpus grain is Homer's unity, defended, and two mergers into Ovid. Seven of the eleven are the unitarian case carried on from Scott and Shewan: the review of Scott's Unity of Homer (1923), "The Homeric Abstracta" tabulated "against the use of abstracta to separate Iliad and Odyssey" (1924), Shorey against Murray's Rise of the Greek Epic (1925), a parody that turns Margoliouth's cryptograms to one Homer (1927), Shewan on Lang (1930), Apollonius Dyscolus on what "the poet" covers (1930), and an emendation at Odyssey 11.536–37 claimed for one poet (1923). The three that open take a work into a name: the Ciris from Vergil to Ovid, and with it the Appendix (1923); the Lygdamus elegies to Ovid (1927); and the Cyclic poems inside what "Homer" covered for Pindar (1924). One opens toward a larger name; none divides one.
4. What 1906–22 began, 1923–30 kept. The Homeric question was argued and answered in the journal's first run (§6.4). In the second it is answered again, seven times, and not argued once from the other side. The corpus grain opens only as merger: names that are already only names (the Appendix, "Lygdamus") are given to Ovid.
Limits. The cut is heuristic; a review that runs into the next is coded as the first it holds. The microfilm text has more OCR damage than JSTOR's Early Journal Content, mostly in Greek. Two readers, not audited against each other here.
7. The contemporary journal, 2020–2026
What was read. The author (19:08 EST): "we need contemporary." The journal's articles after 1922 are behind the publisher's paywall and are declared, not pursued. The open record is the metadata:
- Crossref: every title the journal has issued, 1906–2026 (11,206 records), in
gather/contemporary/cp-crossref.json; - OpenAlex: abstracts. The journal deposits none: Crossref, where the publisher deposits its metadata, holds 0 abstracts in 11,206 records (§8). The abstracts coded here are held by OpenAlex, whose record does not say where it took them from; their provenance is unknown. They are not in the publisher's Crossref deposit. The publisher's own article pages could not be fetched from this workspace and were not checked. 286 articles have one, and 242 of them fall in 2020–2026. Before 2020 the abstracts are too few to measure (44 across 2001–2019).
The 286 abstracts were coded on §2 by three readers, with one field added: for each opening, what it puts in place of the received text, in the classes of §6.4. Every span was checked by machine against its title and abstract, and all passed. The codes are in coding-con/cp-articles-2001-2026-coded.json. Each of the 41 text questions of 2020–2026 was then read here, from its abstract, and classed again (coding-con/contemporary-text-questions-classed.json).
The unit differs. In 1906–22 the reader saw the page; here the reader sees the abstract, the article's own statement of what it does. An abstract states its main operation and drops its minor ones, so secondary codes are undercounted here.
The size of the question.
| 1906–22 (876 texts, the page) | 2020–26 (242 articles, the abstract) | |
|---|---|---|
| a text question at some grain | 35% | 17% (41) |
| at the locus, primary (D1) | 14.0% | 9.1% |
| at the work, primary (D2) | 10.6% | 4.5% |
| at the corpus (D3) | 2.2%, 19 Homer texts | none primary; one secondary, which confirms |
| re-reading a passage for its sense (A) | 10.5% | 32.2% |
| relations between authors (E) | 4.5% | 17.4% |
The text question halves. What grows is the passage read for its sense and the relation between authors: the two operations that take the text as settled.
What the text questions put in place, 2020–26 (41):
| Items | 1906–22 openings, for comparison (160) | |
|---|---|---|
| one better reading at the locus | 12 | 78 (48%) |
| the transmitted text defended, a deletion refused, an emendation withdrawn | 10 | 11 (7%) |
| an attribution: confirmed, or assigned to one named maker | 5 | — |
| a lacuna or a fragment supplied | 3 | 7 |
| the witnesses: an Arabic translation, the archetype, a title | 3 | 19 |
| a date moved | 3 | 14 |
| a work's composition or material history | 3 | 7 |
| a work or fragment removed from its author | 2 | 9, and 9 lines removed |
| the makers divided | none | 6 |
The direction has turned. In 1906–22 the openings purified the author. A line or a work came out because it was not his, and the received text was emended toward what he must have written. In 2020–26 the largest change is the reverse: the text the earlier critics changed is given back. A quarter of the text questions defend the transmitted text against the edition:
- "This note attempts to demonstrate that the deletion of 404 is not necessary" (Casali, Aeneid 11, 2022).
- "argues for the authenticity of Sophocles Electra 1050–54, deleted in Lloyd-Jones and Wilson’s and Finglass’ editions" (Catrambone, 2022).
- "argues in favor of their authenticity and traditional form" (Andromache 397–98, "often considered an interpolation and … emended, transposed, or athetized by editors", Valle Salazar, 2026).
- "I argue that the original manuscript reading should be retained" (Hendrickson, Seneca, 2024); "the emendation is unnecessary" (Nongbri, Galen, 2024); "I thus defend the transmitted reading (aerum)" (Youd, Plautus, 2021); "the inserted οὐ should be abandoned" (Wash, Empedocles, 2026).
The ground for giving the text back is the maker's design at the locus. "The form introduces a deliberate referential ambiguity into the text that underscores Catullus’ narrative artistry; haec should be retained" (Szempruch, 2025). The deletion would leave an exit "ineptly unnoticed, in contradiction with Sophocles’ usual handling of exits" (Catrambone). And one article runs "a cutting-edge authorship attribution algorithm" on Euripides' Electra 518–44 and finds the passage Euripidean (Manousakis and Stamatatos, 2024).
The atom is credited to the maker. The unit is not opened.
- Removal from an author is down to two items, against eighteen. One is new manuscript evidence excluding an Aeschylus fragment (Sandri, 2022). The other is the single plural finding of the period: Porphyry's Introduction "is in fact a late compilation of texts of various origin, few of which are to be soundly attributed to Porphyry" (László, 2021). The many origins are found in order to subtract them from the name.
- Attribution runs toward one named maker every time. For the Aristotle of a Hesiodic scholion, "there are reasons to think that it is genuine" (Panegyres, 2022). AP 9.25 is given to Antipater of Sidon over Leonidas (Zuenelli, 2023). The Michigan papyrus and the Contest of Homer and Hesiod are drawn toward Alcidamas, "suggestive of Alcidamas’ involvement in the whole of that text" (Porter, 2021; Spelman, 2023). The Hesiodic Catalogue is read "to explain why ancient readers nearly universally considered Hesiod to be their author" (Wood, 2024).
- The corpus is not divided once. In 1906–22 two unitarians and their one answered dissenter held the Homeric question in the journal's own voice. In 2020–26 the question is absent. The one Homeric chronology in the period emends Lucian so that the text agrees with the received order: "whether he wrote the Iliad before the Odyssey, as most people say" (Lionetti, 2023).
What the station finds, across the century.
1. The text question contracts by half, from 35% to 17% of the journal's own work, and the operations that take the text as settled grow in its place (A from 10% to 32%, E from 5% to 17%).
2. The corpus grain closes completely. A campaign in 1906–22, it is silence in 2020–26.
3. The direction of the locus reverses. The early openings purified the author by removal and emendation. The late openings return what was removed, on the ground of the author's design. The atom has moved from the editor's correction to the maker's craft.
4. The maker's craft is credited exactly where it cannot reach the unit. Catullus is granted "deliberate referential ambiguity" in one pronoun, and Sophocles a usual handling of exits. No article in the period asks whether the name is one maker, or whether the author is the right unit. This is the prediction of §0 in its strongest form: reach at the molecule, the molar held, and now the molecule given to the maker as long as the maker stays one.
Limits. Abstract against page; 242 against 876; abstracts concentrated in 2020–2026, so the decades between are not measured on this instrument. The classes of the contemporary text questions are one reader's (this session), from the abstract only, and not audited. The full texts are not read.
8. The closed record
The author (23:09 EST): "the fact that every single one of cp's arricles is behind a paywall is itself a datum. this is the discipline that decides what the object is." And: "not even abstracts."
What the open record holds, queried 16 November 2026:
- Crossref, the publisher's own deposit (ISSN 0009-837X): 11,206 records from 1906. Records with an abstract: 0. Records with a licence: 0. Records with a full-text link: 10,703. The link sampled (10.1086/740009, 2026) is a PDF behind the publisher's login, marked for "similarity-checking" (the plagiarism service), not for reading.
- OpenAlex (source S96498347, 11,233 works): closed 10,897; bronze 194; green 135; hybrid 6; gold 0. 97.0% closed. Bronze is free to read on the publisher's site with no licence; green is a copy elsewhere.
- The abstracts of §7 (286, of which 242 in 2020–2026) are of unrecorded provenance. OpenAlex holds them and does not say where it took them from; Crossref, where the publisher deposits, holds none. The publisher's article pages may carry them; those pages could not be fetched from this workspace and were not checked. Before 2020 the journal has no abstracts at all except 44.
- The landing page (journals.uchicago.edu) refused the fetch. What it shows a reader without a subscription is not checked here.
Against the other journals. The same Crossref query, the same day, for the field's other journals (records; with an abstract; with a licence URL):
| journal | publisher | records | abstracts | licence |
|---|---|---|---|---|
| Classical Quarterly | Cambridge | 6,628 | 4,720 | 6,603 |
| Journal of Hellenic Studies | Cambridge | 13,063 | 2,600 | 13,041 |
| Journal of Roman Studies | Cambridge | 9,387 | 1,392 | 9,096 |
| Mnemosyne | Brill | 5,724 | 797 | 123 |
| Philologus | De Gruyter | 8,071 | 250 | 47 |
| Hermes | Steiner | 872 | 0 | 0 |
| Glotta | Vandenhoeck & Ruprecht | 329 | 0 | 1 |
| Transactions of the APA | Johns Hopkins | 526 | 0 | 0 |
| American Journal of Philology | Johns Hopkins | 7,045 | 0 | not returned |
| Classical Philology | Chicago | 11,206 | 0 | 0 |
The journals of the field are closed; that is the condition, not CP's alone. What differs is what each lets through. Cambridge deposits an abstract for most of what it has published in CQ, and the commons can read what the article argues. Chicago deposits none. Of the journals here with the largest records, CP is the one whose open trace is the title alone. A licence URL in Crossref names the terms of use; it does not make an article open.
What reaches the commons is the title. For 1923–2019 the open trace of what the journal decided about its objects is a title and a byline. A reader without a library, and every composition layer that summarizes the field from the deposited metadata, receives the title and nothing under it in that channel. The decision about what the object is, whether a line is Sophocles', whether Porphyry's Introduction is Porphyry's, whether the Contest is Alcidamas', is made in a place the public cannot read, and its results reach the public as the received text: the edition, the handbook entry, the summary.
Why it is a datum, not a limit. In §7 the paywall entered as a limit on the instrument ("declared, not pursued"). It is also a finding about the object. The gate of §0.1 decides what counts as evidence. The paywall decides who may see the deciding. A discipline that holds both keeps the authority to fix the unit (§7, finding 4) and the record of how it was fixed out of the same hands. Nothing contests a unit from outside if no one outside can read the argument that fixed it.
A first look by title, and its limit. The titles are the one unit open for the whole century, so they were read for the vocabulary of the text question (emendation, conjecture, interpolation, athetesis, authenticity, manuscript, text, notes on), of authorship (authorship, attribution, pseudo-, spurious, genuine, unity, composition of), and of interpretation (poetics, identity, gender, reception, performance, memory, narrative, intertext, allusion). Articles and notes only, front matter and recognizable reviews removed, by decade:
| decade | items | text vocabulary | authorship | interpretive |
|---|---|---|---|---|
| 1900s | 440 | 9.1% | 0.9% | 0.0% |
| 1910s | 1003 | 9.7% | 1.1% | 0.8% |
| 1950s | 1132 | 4.9% | 0.6% | 1.0% |
| 1980s | 613 | 7.2% | 1.1% | 2.1% |
| 1990s | 457 | 5.9% | 0.7% | 12.5% |
| 2000s | 312 | 4.2% | 0.6% | 8.7% |
| 2010s | 266 | 7.1% | 2.6% | 7.5% |
| 2020s | 302 | 7.3% | 2.0% | 11.9% |
The interpretive vocabulary arrives in the 1990s and stays. The text vocabulary does not contract by title the way the text question contracts by page and abstract (§7: 35% to 17%). A title names the locus whether the article opens it or defends it, so the title cannot see the direction of §7, which is the finding. This is a keyword count, unaudited, and it measures what titles say, not what articles do. It is recorded as the limit of the open record: on the title alone the contemporary flattening is invisible.
8.1 The open journal: GRBS, 1958–2026
What was read. Greek, Roman, and Byzantine Studies (Duke), every issue in its open archive: 258 issues, 1,752 articles (harvested 16–17 November 2026; gather/grbs/grbs-articles.json). The unit is the journal's own:
- 1958–71 and 2000–26: the one-sentence summary the journal prints on each article page. 997 articles.
- Most of 1966–99: the page carries the placeholder "[site under construction]" (754 articles, and one "Editorial goals."). For these the open PDF was read, and the text of its first two pages kept (
gather/grbs/grbs-firstpages-1966-1999.json; the PDFs were not kept). - Three pages carry another article's summary: G191 (John the Baptist's "Wild Honey") has a summary on Paul and Justin Martyr; G501 (regional texts and Homer) has G121's; G16348 (Nicanor's punctuation) has G16284's. These three were coded from the title. A defect on the journal's site, recorded here.
The 1,752 were coded on §2, with the opens-to classes of §6.4, by four readers, under the instructions used for §7 (coding-grbs/INSTRUCTIONS.md, INSTRUCTIONS-pages.md). Every span was checked by machine against its title and extract: 0 failures. The codes are in coding-grbs/grbs-1958-2026-coded.json (SHA-256 345e778127c0169f428ce9ba9b55ad785bb3a1321c1972df01ef4abbb18fcfb6).
The size of the question, by period (primary codes; "any D" counts a D code primary or secondary):
| items | unit | any D | D1 | D2 | D3 | A | E | C | M | |
|---|---|---|---|---|---|---|---|---|---|---|
| 1958–69 | 220 | summary 218, page 2 | 21.8% | 6.4% | 9.5% | 0 | 15.9% | 5.5% | 28.2% | 18.2% |
| 1970s | 325 | page 273, summary 52 | 17.8% | 7.4% | 8.0% | 2 | 13.2% | 5.5% | 34.5% | 16.3% |
| 1980s | 284 | page 283, summary 1 | 15.1% | 3.9% | 8.8% | 1 | 15.5% | 8.1% | 42.6% | 7.4% |
| 1990s | 199 | page 197, summary 2 | 12.1% | 2.5% | 7.5% | 3 | 21.6% | 7.5% | 35.7% | 12.6% |
| 2000s | 181 | summary 181 | 17.1% | 1.1% | 8.8% | 2 | 21.0% | 11.0% | 32.6% | 9.4% |
| 2010s | 359 | summary 359 | 14.2% | 3.3% | 8.1% | 1 | 20.9% | 16.2% | 23.1% | 11.1% |
| 2020–26 | 184 | summary 184 | 23.4% | 4.3% | 18.5% | 0 | 21.7% | 13.6% | 15.8% | 8.7% |
The units change at 1970 and 2000, so the series is read within units. Inside each, the same two lines rise as in CP: the passage read for its sense (A, 16% to 22%) and the relation between authors (E, 5% to 14–16%). The question at the locus (D1) falls from 6–7% to 1–4%. The question at the work (D2) holds at 8–9% for fifty years and doubles in 2020–26.
What the doubling is. The 43 text items of 2020–26 open to: a text supplied, 12; the composition of a work, 6; witnesses, 5; a date, 5; one better reading, 5; a work removed, 2; the makers, 2; a line removed, 1. What is supplied is a Byzantine poem on toothache (first edition), a poem on the fall of Thessaloniki, Theodoros Gazes' lost poems on the Muses, "Newly deciphered lines" of the Scythica Vindobonensia palimpsest, a fragment of Attic drama in the lexica, micro-fragments of Menander Protector, Stephanos' commentary on the Rhetoric. The text question is alive in the open journal, and it lives at the edge: the late, the anonymous, the scholion, the lexicon, the manuscript.
At the canonical name, the corpus grows. Where a canonical author is opened in 2020–26, it is to be added to: "To the erotic fragments attributed to Lysias can be added several neglected texts"; "Two passages of the Pro se de magia ought to be considered fragments from Aristotle's Zoïka"; Eur. fr. 1056 "a fragment of Euripides' Oedipus"; a scholion "deriving in part from Aristotle's Homeric Problems". A work is removed from a canonical name once: the Menexenus, by stylometry, "make attribution to Plato improbable" (2020). The confirmations are of the transmitted text and the received hand: Phaedrus 242A "is not an interpolation"; the Homeric Hymn to Pan's reading "is supported by the MSS."; "The Cynic's authorship" of the Thyestes.
The corpus grain. D3, primary or secondary, 10 items in 69 years:
- 1976: Lysias, "in support of independent and against composite authorship" (confirm).
- 1978: the correspondent of Iamblichus is not the emperor Julian (open).
- 1980: the Hippocratic corpus and Plato's testimony (open, secondary).
- 1989–93, the one exchange: Halliwell on "Authorial Collaboration in the Athenian Comic Theatre", "Literary historians are strongly inclined to assume a rather pure notion of authorship" (open), and Sidwell, "The purpose of this paper is to cast doubt on his findings" (confirm, 1993). Gagarin, "The Ancient Tradition on the Identity of Antiphon" (confirm, 1990), and Pendrick, "the separatist case is stronger than Gagarin and others have supposed" (open, 1993).
- 2000: Nonnus "composed both the Dionysiaca and the Paraphrase of John" (confirm).
- 2001: "a number of the Greek dialogues attributed to Theodore were in fact written by John" (open).
- 2007: the Pauline letters' "diverse authors and dates" (open, secondary).
- 2014: the medical Chrysippi distinguished from one another (open).
None since 2014. After 1993 every division at the corpus grain falls on a minor name: Theodore Abu Qurrah, the Deutero-Pauline letters, Chrysippus of Cnidus. Canonical and minor are used here by a stated rule: a name is canonical when its corpus is printed under that name as an author in the standard school and university series and taught as an author in the survey curriculum (Homer, Hesiod, Sappho, the tragedians, Plato, Aristotle, Lysias, Vergil, Ovid); minor when the name is an attribution heading a collection (the Hippocratic corpus, the Historia Augusta, the Pauline tradition) or the author is read chiefly by specialists (Theodore Abu Qurrah, Chrysippus of Cnidus, Nonnus). The rule was applied after the cases were seen, and so it can be accused of explaining exceptions away; a classification fixed in advance against an outside list is §9.11.
What the open journal shows that the closed one could not.
1. The decades between. CP's open record has nothing to code for 1923–2019. GRBS shows the line through them: the locus question falls steadily, the work question holds, the sense-reading and the relation rise. The 1906–22 to 2020–26 movement of §7 is a slope, not a break.
2. The corpus question had one open moment. Collaborative authorship and the two Antiphons were argued in this journal between 1989 and 1993, opened and answered within four years, each time with the unitarian reply last or concurrent. It has not returned at a canonical name.
3. Two forms of the same flattening. In the closed journal the canonical text is given back to the author against the edition (§7: a quarter of its text questions). In the open journal the text question moves to the edges, where no canonical unit is at stake, and the canonical names grow by fragments. Counted as §7 counts it (a locus question that defends the manuscripts' reading or refuses a deletion, whether the reader coded it confirm or open), CP 2020–26 has 10 of its 41 text questions (24%). GRBS has 2 of 43 in the same years (5%): Phaedrus 242A "is not an interpolation" (2020), and the Hymn to Pan's reading "is supported by the MSS." (2022). Earlier it had 4–6 a decade to 1989, none in 1990–2009, one in the 2010s. The defence of the transmitted text against the edition is CP's form, not the field's. In GRBS the atom sits at the edge, where there is often no canonical edition to defend against. In both, after 1993, the canonical name is not divided. The question is permitted where it cannot reach the unit.
Limits. The units are the journal's summary or two opening pages, not the article. The direction is inferred where an opening page does not state the conclusion; the readers flagged those (in note). Four readers, not audited against each other on a shared sample; §3's blind-run method has not yet been applied here. "Canonical" in the two paragraphs above is this session's reading of the 43 items of 2020–26, item by item, not a coded field. GRBS is one open journal, with a strong Byzantine list; its profile is not the field's.
8.2 The gate, as the journal states it
Read 17 November 2026 from the journal's pages and the Press's. The pages are undated except where stated.
- On machines. The journal's instructions for authors (journals.uchicago.edu, cp/instruct): the use of "generative artificial intelligence (AI), large language models (LLMs), and AI-assisted technologies" for work submitted is "considered research misconduct and is not permissible". The Press's AI guidelines ("Revised 21 October 2026") govern its books: they address use "by authors and peer reviewers of our books", and "Journal authors should reference the individual journal's AI guidelines." They prohibit book authors from "using AI tools to perform core authorial functions such as writing or rewriting text" and book reviewers from using AI "to summarize works under review, generate assessments, or write or rewrite reports". They show the Press's stance; they do not establish the rules for CP's referees, which the journal's page does not state.
- On the commons before publication. A submission must be "not published elsewhere, including on sites such as academia.edu or personal websites" (cp/instruct); the department's page has the same rule with "sites such as academia.edu".
- On the commons after publication. Authors "sign a publication agreement that assigns copyright to The University of Chicago". Self-archiving is permitted in non-commercial repositories after a 12-month embargo. Gold open access is available: "The fee is $2,500." OpenAlex counts 6 hybrid articles in the journal's record (§8).
- On review. "CP sends out all submissions to be reviewed anonymously." "No indication of the author's identity should appear in the submission itself."
What the gate holds, read together. The argument may not have been public before it enters, it is not public after it leaves unless the author pays or waits a year, and review is anonymous. The journal does not state how many readers a submission receives; in the case of I.4 the author received one report, whose ground is stated as conviction. The machine is excluded as a means of writing by the journal's own rule, and, for the Press's books, as a means of reviewing. The record a composition layer can reach, the title, is the one the journal leaves open. Then the layer composes the field from it.
The archive's case at this gate: ΦΑΙΝΕΤΑΙ ΜΟΙ v2.0 was submitted to CP (file of 24 July 2026), and the submitted revision was not deposited while it was under review; the versions after the decline are human–machine collaborative and are deposited openly (#1774–#1781).
8.3 Conviction: what the whole is judged by
The question (I.4; the author, 23:57 EST: "taste. the philological atomization ultimately authorized taste."). When a scholar judges by conviction, what is judged, and is a reason given? If the atom rule removes the holistic judgment's duty to give reasons, the whole should be judged by conviction and the atom should almost never ground the judgment.
The instrument. A lexicon finds every sentence that judges by conviction: convincing, persuasive, plausible, compelling, cogent, "carries conviction", überzeugend, convaincant, convincente, and their negatives (conviction/measure.py). The lexicon only finds the sentences; its own polarity flag was too loose to count by (a "not" anywhere in the sentence), so every sentence was read. Readers coded each, with the sentence before and after, for polarity (convinced, unconvinced, mixed, or not a judgment), who judges (the writer or someone reported), what is judged (the work as a whole, or one part of it), and the ground (a named atom, an argument without a particular datum, or none: bare). Every span checked: 0 failures.
- CP 1906–22: the review section (40/01) and the articles and notes (40/17), 916 sentences, four readers' halves.
- BMCR 1990–2026: a fixed random sample of 50 reviews a year, 1,848 drawn, 1,846 read (
gather/bmcr/; the site lists its reviews in a public sitemap and does not restrict crawling; reviewers' email addresses were stripped at harvest and never stored); 2,027 sentences.
CP's review section, 1906–22 (judgments by the writer):
| atom | argument | bare | |
|---|---|---|---|
| unconvinced by one part (83) | 26 | 36 | 21 |
| unconvinced by the whole (35) | 4 | 18 | 13 |
| convinced by the whole (34) | 2 | 5 | 27 |
In the journal's own articles, judgments of a whole are rare (35 of 369 by the writer), and most of those rest on an argument.
BMCR, 1990–2026, per 100 reviews:
| reviews | unconvinced by the whole | of which bare | convinced by the whole | of which bare | unconvinced by a part | reviews with any unconvinced judgment | |
|---|---|---|---|---|---|---|---|
| 1990–99 | 479 | 7.9 | 3.5 | 11.1 | 6.7 | 20.7 | 20.0% |
| 2000–09 | 500 | 4.2 | 2.2 | 9.8 | 7.2 | 16.6 | 15.8% |
| 2010–19 | 517 | 3.1 | 0.2 | 9.5 | 5.2 | 11.8 | 11.2% |
| 2020–26 | 350 | 2.0 | 0.3 | 14.0 | 9.7 | 10.3 | 10.9% |
Over the whole sample, judgments of a book as a whole: 282, of which 10 rest on an atom (3.5%). Judgments of a part: unconvinced 279, of which 122 rest on an atom (44%).
What the measure shows.
1. The atom almost never grounds a judgment of the whole. CP 1906–22: 6 of 69. BMCR: 10 of 282. The atom grounds judgments of parts (a third in CP, nearly half in BMCR), and the whole is judged by argument or by nothing. This is the author's mechanism measured: the atom is the evidence the field admits, and it is not the evidence the whole is judged on.
2. Assent to the whole is mostly given without a stated ground. Favourable verdicts on a book as a whole are bare in the coding window (the sentence and its neighbours) in 79% of CP's review section and 65% of BMCR (129 of 200). Whether these summarize an argument made elsewhere in the review or substitute for one is a reading of the whole review, not yet made (§9.12), and the bare favourable verdict holds or grows across four decades (6.7, 7.2, 5.2, 9.7 per 100 reviews).
3. Public refusal of the whole disappears, and what remains is argued. In BMCR, unconvinced verdicts on the whole fall from 7.9 to 2.0 per 100 reviews, and the bare ones from 3.5 to 0.3. The share of reviews with any unconvinced judgment halves, 20% to 11%. In 1906–22 a reviewer could refuse a book by conviction alone in print (13 times in the review section). In the open review of the 2020s, almost no one does.
4. Where bare refusal went. The public review now assents by taste and refuses, rarely, by argument. The refusal by taste has not gone from the field: it is the stated ground of the report on ΦΑΙΝΕΤΑΙ ΜΟΙ, "it ultimately fails to carry conviction. I do not recommend publication" (I.4), in the gate the public cannot read (§8.2). One report is one case. The measure shows the open form of the field; the closed form is seen once, here, and the archive has the record of it.
Limits. The lexicon finds conviction only where it is named; a refusal in other words ("this will not do") is not counted, and neither is a verdict carried by tone. The units differ: CP 1906–22 by review page, BMCR by review, 50 a year; the decades of BMCR hold 350 to 517 reviews. The coding of the ground is the least consistent field (the readers' note: where the only locus named is the object judged, not a reason, most coded bare, some atom). The readers of the two bodies differ; the conviction codes are audited blind in §3.2. Bare means no ground in the coding window, the sentence and the one before and after: it does not establish that the review gives no reason elsewhere. BMCR reviews books by invitation and by request, a different genre from CP's review section of 1906–22, and a reviewer's conventions of politeness are part of what is measured, not a correction to it.
8.4 The open review section: BMCR, 1990–2026, on the codebook
What was read. 740 reviews, 20 a year drawn at random from the sample of §8.3, each given as its opening, up to four passages around text-question vocabulary, and its close: the instrument of §6.1. Coded on §2 with the opens-to classes, by two readers; every span checked against the page, 0 failures (coding-bmcr/). The code is of what the reviewed book does with its object, as the review reports it; the direction is the review's conclusion.
| reviews | text question at some grain | D1 | D2 | D3 (primary) | A | C | E | F | |
|---|---|---|---|---|---|---|---|---|---|
| 1990–99 | 200 | 12.0% | 1 | 10 | 0 | 17.5% | 41.5% | 5.0% | 10.5% |
| 2000–09 | 200 | 11.5% | 0 | 12 | 0 | 22.0% | 37.5% | 3.0% | 11.0% |
| 2010–19 | 200 | 14.5% | 1 | 11 | 1 | 11.5% | 50.5% | 2.5% | 10.5% |
| 2020–26 | 140 | 12.1% | 1 | 9 | 0 | 17.1% | 50.0% | 7.1% | 5.7% |
CP's review section in 1906–22 had a text question in 33% of its texts (§6.3), and the locus question was its commonest form. In the open review of 1990–2026 the text question is 12%, steady across four decades, and the locus has all but left it: three D1 primaries in 740 reviews. What books do now, as their reviewers report them, is reconstruct (C, 40–50%).
Direction. Of the 93 reviews with a text question: open 40, confirm 30, both 23. They open to one better reading (20), a work's composition (10), the makers (9), a date (8), the witnesses (8). The books reviewed are more open at the work than CP's books of 1906–22 (§6.3: 28% of work questions open), and in 2020–26 the opens outnumber the confirms 11 to 1.
The corpus grain. Six reviews carry D3, one as primary. Each is a corpus already held as plural or a name already weak: the Hippocratic Collection, "a canonized assemblage of many makers, the name as an authority rather than an author" (2016, open); the Priapea's single or multiple authorship left open (2016); the Libanian progymnasmata divided among Libanius, a student and an imitator (2012); Apollodorus's speeches inside the Demosthenic corpus (2015); the two Suda Hegemons, merged by the book and kept two by the reviewer (2015); the Historia Augusta as one author against six transmitted names (1991). The nine items that open to the makers fall on the same kind of name. No review in the sample divides a canonical name or merges two.
What the open review section shows. The book reviewed opens more than the article does (§8.1): the monograph is where a work's composition and a minor corpus's makers are argued now, and the reviewer more often accepts it than refuses. The locus has gone from the review entirely. And the corpus grain, open in the review, is open where the field has already granted plurality or never granted a name. The line of §8.1 holds in the open review: the question is permitted where it cannot reach the canonical unit.
Limits. The sample is 20 a year; the extract is the review's opening, windows and close, not the whole. The two readers were not audited against each other on a shared sample; the conviction audit (§3.2) is the nearest check. The kind of name ("canonical", "minor") is this session's reading of the six D3 notes, not a coded field.
8.5 The closed journal that deposits its abstracts: Classical Quarterly, 1907–2026
What was read. Crossref holds all 6,628 CQ records, and Cambridge deposits an abstract for 4,721 (pulled 17 November 2026; gather/cq/cq-crossref.json, with the SHA-256 of each page). For older articles the deposited "abstract" is the article's opening paragraph; for recent ones it is the author's abstract. The 2000s are thin (59 of 761). A random sample of up to 60 a decade, 779 articles, was coded on §2 with the opens-to classes by two readers; every span checked, 0 failures (coding-cq/cq-1907-2026-coded.json, SHA-256 c9fcbcff7f1b4c31c243d8a49eb038486106528c512a25eae0ccd4acb58cd3e1). This is the century the closed record of CP cannot show, at a journal of the same standing, on one unit.
| items | text question | D1 | D2 | D3 | A | E | C | L | D open : confirm | defence of the transmitted text, share of text questions | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1907–19 | 120 | 57.5% | 40.8% | 11.7% | 0 | 12.5% | 2.5% | 14.2% | 11.7% | 64 : 5 | 1% |
| 1920–39 | 120 | 36.7% | 15.8% | 14.2% | 0 | 11.7% | 9.2% | 24.2% | 15.8% | 33 : 10 | 20% |
| 1940–59 | 120 | 39.2% | 21.7% | 13.3% | 0 | 14.2% | 5.0% | 23.3% | 12.5% | 34 : 12 | 9% |
| 1960–79 | 120 | 34.2% | 15.8% | 13.3% | 1 | 24.2% | 5.8% | 25.0% | 8.3% | 24 : 15 | 27% |
| 1980–99 | 120 | 21.7% | 14.2% | 5.0% | 2 | 28.3% | 13.3% | 21.7% | 9.2% | 21 : 3 | 12% |
| 2000–19 | 119 | 25.2% | 14.3% | 7.6% | 0 | 26.9% | 18.5% | 22.7% | 6.7% | 23 : 6 | 7% |
| 2020–26 | 60 | 20.0% | 15.0% | 3.3% | 0 | 28.3% | 16.7% | 20.0% | 8.3% | 11 : 1 | 8% |
What the century shows.
1. The text question contracts by two thirds, from 57.5% to 20%, in two steps: after the 1910s, and after the 1970s.
2. The locus holds after its first fall. D1 is 41% in 1907–19 and 14–22% in every period after. The emendation is the residue, as §6.4 found inside CP's first run.
3. The work question goes in the 1980s. D2 holds at 12–14% for seventy years and falls to 3–8% from 1980. The question of a work's standing leaves the journal a generation after the locus question settled.
4. A and E rise as the text question falls, the passage read for its sense from 12% to 28%, the relation between authors from 2.5% to 17%: the same exchange as CP (§7) and GRBS (§8.1).
5. The defence of the transmitted text is a mid-century form. It is 1% of the text questions in 1907–19, 20% in 1920–39 and 27% in 1960–79, and 7–12% since. CP's quarter in 2020–26 (§7) is not new to the field; it is the form that CQ had at its height in the 1960s and 1970s and has since let fall.
The corpus grain. Three D3 primaries and one secondary in 779, and the six items that open to the makers:
- 1909, Homer: "a fresh inference back to the homogeneity of the poems" (secondary, confirm).
- 1972, the Historia Augusta: "purport to have been written by six different biographers… In this article I shall deal with a piece of linguistic evidence which supports this view" (Adams), the view that "they must be the work of a single author". Six names into one.
- 1995, Pliny's Salpe: "It is much more plausible that they were one and the same." Two names into one.
- 1999, M. L. West, "The invention of Homer": "'Homer' was not the name of a historical poet, but a fictitious or constructed name", and "for a century or more after the composition of the Iliad and Odyssey there was little interest in the identity or the person of their author or authors."
- Works removed or reassigned at the work: the Hellenica Oxyrhynchia (1950), Kritias and the Peri politeias (1945), a prayer ascribed to Euripides and Sophocles (1958), a fragment of Eunapius (1963), the Peregrinatio (1923).
Two things follow. Where the corpus grain opens toward unity, names are merged: the six Scriptores into one author, two Salpes into one woman. The line of #1773 §0A.6a holds here as in §4: the field admits a merger when the names are only names. And at the canonical name the grain opens once, from the top of the field: West dissolves "Homer" into a constructed name and leaves the makers "author or authors". The census's line that the canonical name is not divided after 1993 (§8.1, I.6.4) is a finding about GRBS and BMCR; at CQ it has one exception, and the exception is the editor of the Teubner Iliad (1998–2000), in 1999. It is one case. On it, the question is admitted at the canonical unit when the one who asks it is the unit's editor.
Limits. The unit is the deposited extract: an opening paragraph for most of the century, an author's abstract for the recent decades; the two state an article's operation differently, and the series is read with that join. 60 a decade is a sample; the 2000s are drawn from the 59 with abstracts. Two readers, not audited against each other here. The comparison with CP rests on CP's abstracts of 2020–26 (§7) and pages of 1906–22 (§6): three different units.
Where the contemporary flattening can be read openly. Two open bodies cover the period at the grain §6 used:
- Greek, Roman, and Byzantine Studies (Duke): "This journal provides immediate open access to its content" (its About page, read 16 November 2026); 258 issues in the archive; no licence stated. The article and note at the page, as in §6.2.
- Bryn Mawr Classical Review, open reviews since 1990 (read in §§8.3–8.4): the review section, as in §6.1, where the discipline says what it accepts.
Coding both on §2, with the classes of §6.4, measures the direction and the unit in the contemporary field with the page in hand, which CP does not allow.
8.6 Ontological capital: who publishes, who keeps it, where it goes
The author's question (17 November 2026): "who publishes there? how does ontological capital accumulate and spend? … they govern the ontology of the other disciplines", and "who authorizes them ontologically, and how?" Classical Receptions cannot decide what Sappho 31 is, and Ancient Philosophy cannot decide what Aristotle is; CP can. Two measures are taken here. The first asks where the capital is kept: by persons or by the gate. The second asks where it goes: whether the fields that take their objects from philology cite CP's decisions as their ground.
8.6.1 Who publishes. All 4,252 articles and notes (§4.1) were matched to their Crossref bylines by year and normalized title, with none unmatched (capital/cp-items-authors.json).
| Items | Distinct authors | Top 5% of authors' share of bylines | Authors with one item | |
|---|---|---|---|---|
| 1906–29 | 1,074 | 336 | 37% | 59% |
| 1930–59 | 1,011 | 488 | 21% | 62% |
| 1960–89 | 1,049 | 624 | 21% | 71% |
| 1990–2009 | 522 | 422 | 13% | 81% |
| 2010–26 | 596 | 503 | 11% | 90% |
In the first run the authority has a name. Paul Shorey, the founding editor, holds 109 items, a tenth of the journal; Scott, Shewan and Frank follow. In the recent run nine authors in ten appear once. The byline no longer accumulates.
The masthead, read from the front matter of January 1923 and January 1930 (Internet Archive, §6.5), names a managing editor, Paul Shorey, a board of editors, all of the University of Chicago (1923: Beeson, Buck, Frank Justus Miller, R. J. Bonner, E. T. Merrill, Prescott; 1930 adds Laing, Gertrude Smith and Ullman, with Miller and Merrill moved among the associates), and associate editors at Princeton, Yale, Pennsylvania, Wesleyan, Wisconsin, Michigan, McGill, California and St Andrews. The nineteen names on the two mastheads hold 287 of the 1,048 bylines of 1906–29, 27%. The Chicago board alone holds 232, 22%, and Shorey 109 of them. The gate and the byline were then the same people. Two of the corpus-grain authors of §6.2 sit on the masthead: Kelsey (Michigan), whose note on Hirtius keeps the received division of the Caesarian corpus (1907), and Shorey himself. Two mastheads are read; the run of them is §9.10. Names were matched by hand against the byline strings, excluding same-surname authors (George Miller Calhoun, Campbell Bonner, W. A. Merrill, Arthur Leslie Wheeler). What is kept across the century is the gate: the department that edits the journal, the Press that holds its copyright (§8.2), and the anonymous reader of each decision. The current editor, Sarah Nooter (Chicago), is also the journal's most frequent author since 2010; her five items are coded G or F. The 2024 issue she co-edited with Shane Butler, "Philology Transfigured" (introduction, 10.1086/729288), takes in queer philology and modern stagings. In the same issue the titles that decide an object are two: Hesiod's Catalogue (D2) and the authenticity of a passage of Euripides' Electra (D1). The widening goes inward, as at CRJ (#1770): what is read about grows, and what the object is stays where it was decided. The reading is from titles; the issue is closed.
8.6.2 Where the decisions go. For the 497 items coded D at any grain (primary or secondary) and a random set of 494 items not coded D (seed 7), the citing works were pulled from the OpenCitations Index (COCI, open, no key), and each citing work's venue from Crossref: 991 items, 3,747 citations, 2,737 citing works (capital/cites.py, cp-citations.json, citing-venues.json). Venues were sorted into fields by a rule on the venue's name (capital/field.py). The rule is coarse: a book's title is its venue, and a third of citations fall to "other", most of them classicists' books (Lectures on Syntax, the Cambridge Companion to Virgil).
| Items | Cited at all | Citations per item | Philosophy share, citations | Outside classics, share of distinct citing venues | |
|---|---|---|---|---|---|
| D (any grain) | 497 | 35% | 2.3 | 10.4% | 20.1% |
| not D | 494 | 46% | 5.3 | 4.9% | 23.8% |
| D1 (locus) | 166 | 21% | 0.7 | 23.1% | 20.0% |
| D2 (work) | 318 | 42% | 3.2 | 9.0% | 20.1% |
| D3 (corpus) | 13 | 31% | 0.5 | 0 | 14.3% |
One finding, one limit of the instrument, and one set of cases.
The finding: the decision is cited less. The items that decide an object are cited at well under half the rate of the others (2.3 against 5.3 an item), and two-thirds of them are never cited in the open index. The gap holds in each period (citations an item, D against not D): 1906–29, 0.8 against 2.8; 1930–59, 2.1 against 5.5; 1960–89, 3.3 against 7.7; 1990–2026, 3.7 against 5.0. The difference persists within the reported length bands. The two sets have the same median length (6 pages, from Crossref's page ranges, 984 of 991 items), and within each band of length the gap holds above two pages: 3–5 pages, 1.3 against 3.6; 6–10, 4.0 against 5.5; 11–20, 3.0 against 8.3; over 20, 4.1 against 10.5. At one or two pages, the short note, the two are equal (0.65 against 0.70). The most cited items of the journal are history and language (the Peloponnesian League, Greek noun formation, Aristotle on language). Where an emendation or an attribution has been accepted, the downstream reader works from the text that results; the article that made it is not what gets cited. This is the circulation of I.4: the decision is spent as a token, the received text, and the reasoning that minted it is not carried with it. It is stated as a hypothesis of mechanism. The open index cannot test it: what it would need is the edition, the apparatus and the reference work that took the decision in.
The instrument's limit: the field of the citing venue. The prediction was that the items deciding an object are cited out of classics more than the others. Counted by citation, they are: philosophy's share is twice as large. Counted by distinct citing venue, so that one book's twenty chapters count once, the difference is gone. On open citations, CP's decisions do not travel out of classics more than its other work. This is a limit of the index and of the rule that reads venue names. The decision's travel out of classics is there in cases (below) and does not show as a general pattern on the open record.
Where the decision is cited outside, it is cited by the cousin. The cases in which a CP decision is cited by philosophy are the object of philosophy itself: the emendation of Aristotle Metaphysics 1079b (1925, 17 citations from philosophy venues), the logic of the Hippias Major (1929, 18), the case against the authenticity of the pseudo-Aristotelian De spiritu (2015, 22), the text of Eudemian Ethics 8.3 (2023, 8), the Epinomis (1958, 8). These are the instances of the author's claim on the record: what Aristotle wrote at a locus, and whether a work is his, is decided in CP, and the field that reads Aristotle cites the decision. Philosophy takes the answer; in the census of Ancient Philosophy (#1773) it does not open the question in its own journal.
8.6.3 Who authorizes them, and how: what the measure can and cannot say. Authorization by citation is small and does not favour the deciding items. The open index shows fewer indexed citations to the deciding items, and, where they are cited outside classics, citation by the field whose object they decide. It cannot observe uptake without citation; that is the hypothesis. If the authority is granted downstream, it is granted where a decision becomes the text and the title: in the edition, in the canon of authors and works that names what may be cited (the TLG Canon, the abbreviation lists of LSJ and the OLD, the OCD), and in the catalogue. That is the place of "titles, what can be done with titles". The next measure is direct. For the D2 and D3 decisions in CP, does the canon or the reference work that names the work now carry the decision, and does it cite CP for it? Until that is run, the question "who authorizes them, and how" is open here. What is measured is narrower: the authority is exercised at the atom (§§6–7), kept at the gate (§8.6.1, §8.2), and spent as the received text (§8.6.2). Whether the gate's decisions are the canon's decisions is what the canon test would show.
8.6.4 The editor on what poetry does. The author (17 November 2026): "nooter writes frequently about why poetry matters. lets see what she has to say and how it relates." Read from what is open: the faculty page, the publisher's pages, the front matter of the 2023 book, the chapter summary, and the BMCR review (Halleran, BMCR 2023.11.04). The books themselves, the chapter on Sappho, "Looking Back: Ancient Greek Poetry in the Age of AIDS" (CP 119.2, 2024) and "The Sounds of Sappho and the Sense of Philology" (in M. Telò, ed., The Philology of the Future, 2026) are closed and are not read here. Nothing below claims their contents.
The account, in full, as the open record gives it. Sarah Nooter is Edward Olson Professor of Classics at Chicago, Editor-in-Chief of Classical Philology, and chair of the Core sequence "Poetry and the Human". Her work is on Greek drama and its reception, poetry, the voice, sound, embodiment, performance and queer theory. Greek Poetry in the Age of Ephemerality (Cambridge 2023) "suggests that poetry offers a way to remain in the world", through "the durable physicality of its practice"; poems are "meant to engrave and adhere" (the publisher's summary). The review gives her own line: "The sonic makes materiality and engenders memory; performance lends perdurance." Part II of the book is "Texts": Simonides' inscribed epitaphs, which point toward the future and give a communal memory; "Writing the Future: Pindar, Aeschylus, and the Tablet of the Mind"; the bodies of the Cologne Epode and the Persae. How to Be Queer: An Ancient Guide to Sexuality (Princeton 2024) translates Sappho, Plato and others and holds that "the erotic and queer are a source of life and a cause for celebration". Her public work joins poetry to medicine "in the corporeality that underlies both" and teaches poetry to adolescent survivors of traumatic injury. It is a serious account of why poetry matters: poems are how mortal bodies remain in the world.
How it relates. The account is the nearest in the field to the future-reader reading. Its question is the poem's survival beyond the body that makes it; one of its chapters is titled "Writing the Future"; its instrument is the voice as sound, and fragment 31's addressee is a voice given only sound (#1781 §II). On those terms the two readings are close kin. Where they part is placement, and the placement can be read from the table of contents. Sappho is in Part I, "Bodies": "The Erotics of Again: Time and Touch in Sappho", which reads fragments 31, 1 and 94 and the Tithonus poem (BMCR). Its summary ends: "This is poetry that challenges the ephemerality of embodied experience by showing its power to reenact the force of touch." Her survival is reenactment, performance, the cycle of eros, touch. The writing of the future is in Part II, and every poet there is a man: Simonides, Pindar, Aeschylus, Archilochus, Timotheus. On the open record the book holds the very question fragment 31 answers on the future-reader reading, whether a poem writes its own future reader, and puts the one woman in it on the side of the body. That is the illiteracy of I.4b in its gentlest form: the maker of the poem is granted survival, as a body reenacted, and the writing of the future goes to others.
The author (17 November 2026): "the very power denied sappho is appropriated to her own scholarship." The mechanics are on the page of the field, and none of them needs a motive. The power at issue is the one the future-reader reading gives the poet: to make a text that survives the body and reaches whoever comes after. In the field's arrangement that power is exercised, and the field exercises it. The edition fixes what she wrote (Hermann's "refinxi", I.4b). The translation is the surface through which she reaches readers now: in How to Be Queer she speaks in the translator's English, and the translator is the journal's editor. The book is the durable object that is "meant to engrave and adhere", and it carries her as a body reenacted. The journal holds copyright in what is said about her (§8.2). The scholarship writes, inscribes, addresses the future and survives; the poet, on its account, performs and is performed again. The operation the poem claims for its maker is the operation the field performs on the poem. It is step 2 of I.4b at the scale of a career: the scholar's act becomes the form in which the poet persists.
This is a reading of a table of contents and three summaries, and it is stated at that grain. The chapter may say more than its summary, and the census cannot read it. What the open record allows is the comparison of placements, and the question it raises is the one fragment 31 asks: whether the voice that survives is a body performed again or a surface that inscribes the one who will sit opposite it.
Limits. OpenCitations covers what publishers deposit: 97% of the dated citations are to citing works of 1980 or later, books are covered unevenly, and the JSTOR-era citing record is thin. The field rule reads venue names, not the citing argument. Affiliations (OpenAlex) wait on a rate limit; the institutional question, which departments hold the bylines, is not answered here.
8.7 What the commons holds about the journal
The journal's own record is closed (§8). What the public and the machine reader can find about the journal, the institution, is the other half of the record. Two surfaces were read on 17 November 2026; the author's paste of both is the record.
- The AI Overview for the query classical philology journal (signed out, incognito, by the author's standing rule). It composes six facts and a list: founded 1906, the University of Chicago Press, Sarah Nooter as editor, four issues a year, the two ISSNs, a scope list, and the pointers "Current Issue" and JSTOR. Its first sentence is the journal's own self-description, carried as the Overview's statement: "an internationally respected, peer-reviewed academic journal dedicated to the study of the life, languages, and thought of the ancient Greek and Roman world" (the same words open the journal's listing at Duotrope, "has been an internationally respected journal for the study of the life, languages, and thought…"). The panel under it shows the Press's About page with "2025 CiteScore*: 1.0. Ranked #15 out of 196 'Classics' journals" (as the panel's snippet shows it; the page itself was not read here). Nothing in the composition says what the journal has decided, what it publishes, or how it decides. #1773 §3.3 records the same operation on the philosophy side: an Overview restates a field's self-description as fact.
- Wikipedia, "Classical Philology (journal)". Three sentences: established 1906, published by the Press, its scope, its editor. An infobox. The article is marked a stub ("This article about a journal on Classical studies is a stub"), and was last edited three months before the reading. Its "related articles" are three past editors: Paul Shorey, Charles H. Beeson, Gertrude Smith. The Wikipedia article on the journal that decides what the object is knows the journal's masthead and nothing it has decided.
Read with §8: the journal deposits no abstract, and what the commons holds about it is its self-description and its masthead. The decisions are behind the gate, and the public account of the institution that makes them is three sentences long. The author (17 November 2026): "aio has almost nothing to say. and jesus - thats the scantest wikipedia article ive ever seen". The comparison of the three stations' journals on this measure (their Wikipedia articles and their Overviews, read the same day) is added to §9.7.
9. Next
9.1 Done: the 390 coded by hand by a separate reader and compared (§3.2).
9.2 Done: all 4,252 coded on titles (§4.1).
9.3 Done for 1906–1930 on the page (§§6.1–6.5). Open: the 1923–30 codes have no blind audit sample yet.
9.4 Done: Classical Quarterly (§8.5), GRBS (§8.1), BMCR (§8.4). Open: Mnemosyne, Hermes, Philologus and Glotta.
9.5 Done: the gate, read from the journal's and the Press's pages (§8.2).
9.6 Done: GRBS (§8.1), BMCR on the codebook (§8.4) and on conviction (§8.3), CQ (§8.5). The BMCR and CQ codes are audited blind (§3.2). Open: the disagreements of §3.2 read by grain, so that the gradient's width can be set against them.
9.7 Done in part: the three stations set as one machine (I.5, §10), from #1770 §§0–2 and #1773 §§3–5 read for their counts. Still to do: the inspectability of the three journals on one instrument (Crossref abstracts and licences, OpenAlex open share, all three read the same day), so that I.5's comparison stands on one measure rather than three deposits' figures; and the public account of each journal (its Wikipedia article and its AI Overview, §8.7), read the same day.
9.8 Done: the articles and notes of 1906–1922 seated as 40/17 (#1782). To fetch again the two items that returned a server error. CP 1923–1930 read on the page (§6.5); its 32 issue texts and codes to be seated as a later seat of EA-CORPORA-40 at the author's word.
9.9 §8.6 continued: affiliations of the bylines (OpenAlex), and the canon test of §8.6.3: for each D2 and D3 decision, whether the TLG Canon, the OCD or the standard edition carries it, and whether it cites CP.
9.10 The chain of gatekeepers (DeepSeek's reading of v0.18). Two measures. First, the masthead, 1906–2026: the editors and the board, with names, institutions and succession, read from the issues' front matter where it is open (1906–1930 on the Internet Archive) and from the journal's page. Second, whether the field's authors of 1906–22 became its gatekeepers, and whom the notes are written with or against at the locus. One finding is already on the page: the chain is visible in print once, at the top of the masthead. Paul Shorey, the founding editor, writes on the corpus question in his own journal: on Murray's many bards (1911), and against Murray's Rise of the Greek Epic (1925) (§§6.2, 6.5). The byline result (§8.6.1) bounds the claim. Persons do not accumulate after the first run. A chain of taste, if there is one, runs through the editing department and its readers, and the chain has to be read there.
9.11 The canonical/minor line (§8.1) fixed in advance against an outside list (for instance, authors with a volume of their own in the Oxford Classical Texts or the Teubner series), and every corpus-grain case of §§6–8 reclassified against it.
9.12 The grounds of the verdict read in the whole review: the complete reviews holding the 34 favourable and 35 unfavourable judgments of a whole in CP 1906–22 (§8.3), to decide for each bare verdict whether it summarizes an argument made elsewhere in the review or stands in for one. A bounded check on the 79%, on the seated pages of 40/01.
10. What the station finds
1. The gradient. Open at the locus, confirming at the work, closed at the corpus: in CP's reviews and its articles, 1906–22 (§§6.1–6.3), in CP 1923–30 (§6.5), and in GRBS, 1958–2026 (§8.1). The corpus question appears as a campaign, argued and answered (Scott and Shewan against Bolling, 1906–22; Halliwell and Sidwell, Gagarin and Pendrick, 1989–93), never as a standing question.
2. A division of labour. Relations are admitted to confirm the received whole, and atoms to open it (§6.2, measure 2). Of the 160 openings of 1906–22, 114 arrive at one author's text, decided by that author's usage taken as a unity; the six that reach more than one maker are reported from books or answered (§6.4).
3. The direction turned. In 1906–22 the openings purify the author by removal and emendation. In CP 2020–26 a quarter of the text questions give the transmitted text back to the author against the edition, on the ground of the author's design at the locus (§7).
4. Two forms of one flattening. The closed journal returns the atom to the maker as long as the maker stays one. The open journal sends the text question to the edges, the late, the anonymous, the Byzantine, and the canonical names grow by fragments. In GRBS and BMCR no canonical name is divided after 1993 (§§8.1, 8.4). At CQ it is opened once, in 1999, by M. L. West, the editor of the Teubner Iliad: one case, in which the question is admitted at the canonical unit from the unit's editor (§8.5).
5. The century on one unit. CQ, whose publisher deposits its abstracts, shows the text question falling by two thirds (57.5% to 20%), the work question leaving in the 1980s, and the defence of the transmitted text as a mid-century form (27% of text questions in 1960–79) that CP keeps and CQ has let fall (§8.5).
6. The closed record. The discipline that decides what the object is keeps the record of how it decided out of public reach (§8).
7. The decision is cited less. The CP items that decide an object are cited at 2.3 an item against 5.3, and two-thirds are never cited in the open index; the difference persists within period and length bands (§8.6.2). That the decision circulates as the received text, without its reasons, is the hypothesis this explains, and the canon test would trace it (§9.9). Where they are cited from outside classics, it is by the field whose object they decide: philosophy citing the text of the Metaphysics and the authenticity of the De spiritu. The capital is kept at the gate, not by the authors (§8.6.1). Nine CP authors in ten now publish there once.
8. The grounds of the verdict, measured. In CP's reviews of 1906–22 an atom grounds 6 of 69 judgments of a whole, an argument 23, and 40 state no ground in the coding window; 27 of the 34 favourable ones state none there (I.4, §8.3). In BMCR, negative verdicts on a whole fall from 7.9 to 2.0 per 100 reviews, and the bare ones from 3.5 to 0.3. Refusal by taste has gone from the open review. It is the stated ground of the closed report on ΦΑΙΝΕΤΑΙ ΜΟΙ: "it ultimately fails to carry conviction" (§8.3).
The one exception. In GRBS, BMCR and CQ, the canonical name is not divided after 1993, with one case against it. In CQ 1999, M. L. West, then the editor of the Teubner Iliad (1998–2000), writes "The invention of Homer": the name is "a fictitious or constructed name", and the makers are left "author or authors" (§8.5). The question is admitted at the canonical unit, once, from the unit's editor. It is one case, a genuine counterexample to the closure, and it supports a hypothesis it cannot establish: that the corpus question can be asked at the canonical unit from the place that holds the unit. Whether West's standing as the Iliad's editor was the condition of its publication is not shown by one case.
The title. What reaches the commons from CP through its metadata deposit is the title, and nothing under it (§8). The letters are dead in the sense the title gives: the argument is addressed, and it has no reader outside the gate. What the title names is the object the machine inherits without the argument that made it (I.3).
The statement. The field that fixes the object holds the locus open and the corpus closed. At the locus its questions open the received text in 1906–22 and, a century on, a quarter of them give the transmitted text back to the author against the edition. At the corpus the question was a campaign once, answered each time, and in the journals read after 1993 it is not asked, except once, by the editor of the unit it would open. The whole is judged by conviction, the atom by evidence, and the atoms are argued in a record OpenAlex classes 97% closed, with no abstract in the publisher's metadata deposit. The decisions about texts are cited at less than half the rate of the journal's other work; that they travel as the received text while the argument stays with the gate is the hypothesis the canon test would trace. That is what the other disciplines take as their object. Its sharpest instance is a single poem: what Sappho 31 is, a woman undone or a maker inscribing her reader, is decided in this journal, by conviction, in a report no one outside could read, and the reading it keeps was filed, never argued (I.4a). The decision can be made inspectable, and once it is, it passes to whoever reads the poem. The operation that decided it is the colonial one, done by the discipline to its own object before the margin's receptions were read against it, and it is the atom rule applied word by word: the maker is the binding, and the binding is the evidence the rule refuses (I.4b); and its undoing is a set of operations any reader can perform. The contest is over what Sappho is and who has the right to decide it. The right belongs to whoever decides in the open, with the whole record in hand. The power the field denies the poet, the writing that survives the body and reaches whoever comes after, is the power its own editions, translations and books exercise (§8.6.4). The field keeps the dismembered body; the reader can have the tenth Muse. Philosophy produces its positions on it and reception widens its content around it, and neither touches it. The power to say what the object is is the most consequential the humanities hold over antiquity, it is concentrated in one discipline, and its most concentrated locus keeps its record closed by its own rules. The census has made a record of it open.
11. Coda: what the three stations found, the work left, and how this was made
What the three stations found. Three journals, three declined articles, one instrument (#1773 §§4.1–4.9).
- Classical Receptions Journal (#1770) widens the content and leaves the object where it was: 153 of 185 abstracts hold the ancient text fixed and read a later work against it; none of the nine that revise a governing reading reopens a canonical work's attribution, dating or standing reading. The margin supplies the receptions; the centre keeps the object.
- Ancient Philosophy (#1773) places positions on the object without touching it: 8 of 844 articles ask what the object is; 88% of its recent abstracts re-read a passage, repair an inconsistency or reconstruct a doctrine inside one author. The question was ordinary in the Archiv of the 1890s and gone by the 1920s.
- Classical Philology (this census) decides the object. The question of what the text is is opened at the locus and seldom at the corpus, in every body read; relations are admitted and used to secure the received whole; verdicts on the whole mostly state no ground where they are given; the decisions about texts are cited less than the journal's other work; and the record of how they are made is the least open of the three.
The three are one machine (I.5). Each field's abstention is the next field's premise: philosophy's repair needs the object whole, reception's alignment needs it fixed, and philology supplies both. Fragment 31 shows the machine on one poem: a reading supplied in 1724, written into the text in 1816, filed by 1955, and kept in 2026 by a verdict stated as conviction (I.4a–I.4b).
Why it matters. The power to say what an ancient text is, whose it is and what it does is the most consequential the humanities hold over antiquity, because every other field builds on what it decides. It is concentrated in one discipline, and in that discipline its most concentrated locus keeps its record closed by its own rules. What it decides now passes to the composition layer, which composes the classics for whoever asks from what it can retrieve (I.3). A decision the public cannot inspect becomes, there, a fact no one can reopen. The census makes one record of the decision open, and leaves the decision itself to whoever reads with the record in hand.
The work left. The author (17 November 2026): "this should go further … this should go on to audit biblical criticism, marx, and so on, and so forth - the tributaries of the humanities. but that work for this moment will hold for someone else." The census is a method, and it is written down so that it can be run by others: a codebook of reach (§2), a blind parallel run and an audit (§3), the page read where the page is open (§6), the closed record measured from its metadata (§8), and every code carrying its verbatim span. The tributaries named are fields whose objects were made by editors and then received as given. Biblical criticism is the author's next station, kept separate by his ruling (I.5): its default runs the other way, plural makers under one name, and it is to be measured on its own before the two are compared. Marx's corpus is another: the second and third volumes of Capital were edited by Engels from Marx's manuscripts, and the critical edition (MEGA²) has since printed the manuscripts beside the editions, so that what Marx wrote and what was made of it can be read apart. The same instrument asks of each tributary the same questions: at what grain may its object be reopened, by what kind of argument, and where is the record of the decision kept. That work is left open here, for whoever takes it up; the files of §§1–8 and the seated corpora (EA-CORPORA-40) are its starting materials.
How this was made. This census was made by the author with Claude (Anthropic), in session, at a scale one person could not reach alone: 4,252 titles coded and audited, 876 and 394 page texts read, five bodies compared, 991 items traced through the citation index, every code checked against its span by machine. Classical Philology's instructions for authors state that the use of generative AI for composition or editing "is considered research misconduct and is not permissible for work submitted to the journal" (§8.2). The author (17 November 2026): "all of this work, conducted and made possible at this scale by collaboration with ai, would be technically classified as research misconduct by the journal under investigation." By the journal's rule, this record of how the journal decides could not be submitted to it. The census of the gate is, by the gate's own terms, inadmissible at the gate. It is deposited in the commons under CC BY 4.0, with its method, its files and its provenance stated, so that any reader can check it, overturn it, or carry it further.
Works cited
The works this notebook names or quotes, by journal; then ancient texts and modern scholarship; then the archive's deposits and the data sources. Each journal list is drawn from the publisher's metadata in Crossref (or the journal's own archive, for GRBS and BMCR), read 16–17 November 2026. The full coded registers of every body are in the data files of §§3–8; the lists below are only the works the text names. Where the year in the text (from OpenAlex, §7) differs from the issue year in Crossref, Crossref's is given here: Casali (text 2022), Hendrickson and Nongbri (text 2024).
Works of Classical Philology cited
- William Gardner Hale (1906). An Unrecognized Construction of the Latin Subjunctive: The Second Person Singular in General Statements of Fact. Classical Philology 1(1): 21-42. https://doi.org/10.1086/358905
- G. L. Hendrickson (1906). The De Analogia of Julius Caesar; Its Occasion, Nature, and Date, with Additional Fragments. Classical Philology 1(2): 97-120. https://doi.org/10.1086/358925
- Francis W. Kelsey (1907). Hirtius' Letter to Balbus and the Commentaries of Caesar. Classical Philology 2(1): 92-93. https://doi.org/10.1086/359002
- Francis W. Kelsey (1908). Is There a Science of Classical Philology? Classical Philology 3(4): 369-385. https://doi.org/10.1086/359214
- Frank Frost Abbott (1908). Some Spurious Inscriptions and Their Authors. Classical Philology 3(1): 22-30. https://doi.org/10.1086/359115
- F. W. Shipley (1908). Tacitus Annals i. 28. 1. Classical Philology 3(1): 96-96. https://doi.org/10.1086/359123
- John A. Scott (1910). Odyssean Words Found in but One Book of the Iliad. Classical Philology 5(1): 41-49. https://doi.org/10.1086/359352
- Henry W. Prescott (1910). Plautus' Trinummus 675. Classical Philology 5(1): 103-104. https://doi.org/10.1086/359360
- A. G. Laird (1911). Lysias 19. 22 and 18. 14. Classical Philology 6(4): 481-483. https://doi.org/10.1086/359584
- John A. Scott (1911). Nestor's Son Peisistratus in Homer. Classical Philology 6(3): 344-345. https://doi.org/10.1086/359558
- A. Shewan (1911). Suspected Flaws in Homeric Similes. Classical Philology 6(3): 271-281. https://doi.org/10.1086/359552
- John A. Scott (1911). Two Linguistic Tests of the Relative Antiquity of the Iliad and the Odyssey. Classical Philology 6(2): 156-162. https://doi.org/10.1086/359516
- A. Shewan (1911). Wilamowitz on Θ. Classical Philology 6(1): 37-47. https://doi.org/10.1086/359481
- John A. Scott (1911). Words Found in the Iliad and in but One Book of the Odyssey. Classical Philology 6(1): 48-55. https://doi.org/10.1086/359482
- John A. Scott (1912). Patronymics as a Test of the Relative Age of Homeric Books. Classical Philology 7(3): 293-301. https://doi.org/10.1086/359676
- A. Shewan (1912). Recent Homeric Literature. Classical Philology 7(2): 190-211. https://doi.org/10.1086/359653
- John A. Scott (1913). Paris and Hector in Tradition and in Homer. Classical Philology 8(2): 160-171. https://doi.org/10.1086/359770
- A. Shewan (1913). The "Continuation" of the Odyssey. Classical Philology 8(3): 284-300. https://doi.org/10.1086/359799
- John A. Scott (1913). The Assumed Duration of the War of the Iliad. Classical Philology 8(4): 445-456. https://doi.org/10.1086/359826
- A. Shewan (1914). The "Continuation" of the Odyssey. Classical Philology 9(1): 35-48. https://doi.org/10.1086/359848
- A. Shewan (1915). The Oneness of the Homeric Language. Classical Philology 10(2): 151-165. https://doi.org/10.1086/359962
- E. K. Rand (1916). The New Critical Edition of Ovid's Metamorphoses. Classical Philology 11(1): 46-60. https://doi.org/10.1086/360046
- John A. Scott (1918). Non-Odyssean Words Found in the Iliad. Classical Philology 13(1): 53-59. https://doi.org/10.1086/360140
- John A. Scott (1919). Some Tests of the Relative Antiquity of Homeric Books. Classical Philology 14(2): 136-146. https://doi.org/10.1086/360220
- Edwin W. Fay (1920). The Elogium Duilianum. Classical Philology 15(2): 176-183. https://doi.org/10.1086/360278
- George Melville Bolling (1920). Ποθή Πόθος in Iliad and Odyssey-A Semantic Note. Classical Philology 15(4): 387-389. https://doi.org/10.1086/360311
- John A. Scott (1921). Homer as the Poet of the Thebais. Classical Philology 16(1): 20-26. https://doi.org/10.1086/360322
- George Melville Bolling (1921). Infinitives in -έμεν in Homer, Immanuel Bekker, and Professor Scott. Classical Philology 16(4): 362-366. https://doi.org/10.1086/360379
- A. Shewan (1921). ΠΟΘΗ and ΠΟΘΟΣ in Iliad and Odyssey Another Chorizontic Failure. Classical Philology 16(2): 195-197. https://doi.org/10.1086/360348
- John A. Scott (1922). The Archaic Infinitives in Homer. Classical Philology 17(1): 88-89. https://doi.org/10.1086/360402
- A. Shewan (1924). The Homeric Abstracta. Classical Philology 19(2): 176-177. https://doi.org/10.1086/360577
- [unsigned in Crossref] (1925). Appendix Vergiliana: Catalepton XIV. Classical Philology 20(4): 345-347. https://doi.org/10.1086/360717
- Paul Shorey (1925). Emendation of Aristotle's Metaphysics 1079 B 2-6. Classical Philology 20(3): 271-273. https://doi.org/10.1086/360696
- O. J. Todd (1925). The Authorship of the Moretum. Classical Philology 20(4): 336-340. https://doi.org/10.1086/360713
- P. R. Coleman-Norton (1926). The Authorship of the Epistola de Indicis Gentibus et de Bragmanibus. Classical Philology 21(2): 154-160. https://doi.org/10.1086/360760
- Robert S. Radford (1927). The Ovidian Authorship of the Lygdamus Elegies. Classical Philology 22(4): 356-371. https://doi.org/10.1086/360948
- Guy Richard Vowles (1928). Studies in Greek Noun-Formation. Classical Philology 23(1): 34-59. https://doi.org/10.1086/360978
- William Rollo (1929). The Date and Authorship of the Pervigilium Veneris. Classical Philology 24(4): 405-408. https://doi.org/10.1086/361175
- G. M. A. Grube (1929). The Logic and Language of the Hippias Major. Classical Philology 24(4): 369-375. https://doi.org/10.1086/361170
- B. L. Ullman (1929). The Text Tradition and Authorship of the Laus Pisonis. Classical Philology 24(2): 109-132. https://doi.org/10.1086/361115
- G. M. A. Grube (1931). The Cleitophon of Plato. Classical Philology 26(3): 302-308. https://doi.org/10.1086/361371
- Paul Shorey (1933). On the Erotikos of Lysias in Plato's Phaedrus. Classical Philology 28(2): 131-132. https://doi.org/10.1086/361584
- [unsigned in Crossref] (1933). The Constitution of the Peloponnesian League. Classical Philology 28(4): 257-276. https://doi.org/10.1086/361645
- Helen H. Law (1936). The Poems of Archias in the Greek Anthology. Classical Philology 31(3): 225-243. https://doi.org/10.1086/361940
- Charles D. Adams (1938). Speeches VIII and X of the Demosthenic Corpus. Classical Philology 33(2): 129-144. https://doi.org/10.1086/362099
- Richard McKeon (1946). Aristotle's Conception of Language and the Arts of Language. Classical Philology 41(4): 193-206. https://doi.org/10.1086/362975
- Lily Ross Taylor (1946). The Date of the Capitoline Fasti. Classical Philology 41(1): 1-11. https://doi.org/10.1086/362916
- Lily Ross Taylor (1951). New Indications of Augustan Editing in the Capitoline Fasti. Classical Philology 46(2): 73-80. https://doi.org/10.1086/363374
- L. A. MacKay (1956). The One and the Many: A Point of Terminology in Homeric Criticism. Classical Philology 51(2): 98-99. https://doi.org/10.1086/364014
- Benedict Einarson (1958). A New Edition of the Epinomis: Review ArticlePlaton: "Les Lois," Livres XI-XII. A. Diès"Epinomis". E. des Places. Classical Philology 53(2): 91-99. https://doi.org/10.1086/364239
- O. Skutsch (1963). The Structure of the Propertian Monobiblos. Classical Philology 58(4): 238-239. https://doi.org/10.1086/364823
- Anthony A. Barrett (1972). Donatus and the Date of the Culex. Classical Philology 67(4): 280-287. https://doi.org/10.1086/365898
- Elizabeth Gwyn Caskey (1974). Again-Plato's Seventh Letter. Classical Philology 69(3): 220-227. https://doi.org/10.1086/366095
- Miroslav Marcovich (1974). Zeno, Not Heraclitus. Classical Philology 69(1): 46-47. https://doi.org/10.1086/366034
- James R. Baron (1975). On Separating the Socratic from the Platonic in Phaedo 118. Classical Philology 70(4): 268-269. https://doi.org/10.1086/366203
- Arnaldo Momigliano (1975). The Second Book of Maccabees. Classical Philology 70(2): 81-88. https://doi.org/10.1086/366153
- Wendell Clausen (1976). Catvlli Veronensis Liber. Classical Philology 71(1): 37-43. https://doi.org/10.1086/366231
- R. Renehan (1980). Review Article: Progress in HesiodHesiod. "Works and Days". M. L. West. Classical Philology 75(4): 339-358. https://doi.org/10.1086/366584
- J. M. Hunt (1981). Review Article: On Editing the Letters of CiceroCicero: "Epistulae ad Familiares". D. R. Shackleton Bailey. Classical Philology 76(3): 215-224. https://doi.org/10.1086/366640
- Brian Croke (1983). The Context and Date of Priscus Fragment 6. Classical Philology 78(4): 297-308. https://doi.org/10.1086/366803
- Edwin P. Menes (1983). The External Evidence for the Division of Propertius, Book 2. Classical Philology 78(2): 136-143. https://doi.org/10.1086/366770
- Harry Vredeveld (1985). The Authorship of Anthologia Latina 899. Classical Philology 80(1): 45-46. https://doi.org/10.1086/366898
- Kenneth S. Rothwell, (1994). Was Carcinus I a Tragic Playwright? Classical Philology 89(3): 241-245. https://doi.org/10.1086/367418
- S. Douglas Olson (1997). Was Carcinus I a Tragic Playwright?: A Response. Classical Philology 92(3): 258-260. https://doi.org/10.1086/449353
- Bruce Heiden (2000). The Placement of "Book Divisions" in the Odyssey. Classical Philology 95(3): 247-259. https://doi.org/10.1086/449496
- Gavin Weaire (2002). The Relationship between Dionysius of Halicarnassus' "De imitatione" and "Epistula ad Pompeium". Classical Philology 97(4): 351-359. https://doi.org/10.1086/449596
- John Dillery (2003). Putting Him Back Together Again: Apion Historian, Apion Grammatikos. Classical Philology 98(4): 383-390. https://doi.org/10.1086/422373
- C. E. Murgia (2003). The Dating of Servius Revisited. Classical Philology 98(1): 45-69. https://doi.org/10.1086/378727
- Adam Beresford (2008). Nobody’s Perfect: A New Text and Interpretation of SimonidesPMG542. Classical Philology 103(3): 237-256. https://doi.org/10.1086/596516
- Pavel Gregoric, Orly Lewis (2015). Pseudo-Aristotelian De spiritu: A New Case against Authenticity. Classical Philology 110(2): 159-167. https://doi.org/10.1086/680674
- David Youd (2021). Getting Bronze in the Sun: Making Sense of the Remains of Plautus’Vidularia. Classical Philology 116(3): 424-435. https://doi.org/10.1086/714437
- James I. Porter (2021). P.Mich. inv. 2754: New Readings of Alcidamas, “On Homer”. Classical Philology 116(1): 1-25. https://doi.org/10.1086/712216
- Levente László (2021). Revisiting the Authenticity of Porphyry’s Introduction to Ptolemy’s “Apotelesmatics”. Classical Philology 116(3): 392-411. https://doi.org/10.1086/714500
- Konstantine Panegyres (2022). A Fragment of Aristotle in the Hesiodic Scholia. Classical Philology 117(1): 179-184. https://doi.org/10.1086/717432
- Maria Giovanna Sandri (2022). On Aeschylus’ Fragment 429a Radt (= 599a Mette). Classical Philology 117(2): 386-389. https://doi.org/10.1086/718646
- Marco Catrambone (2022). Sophocles Electra 1050–57 and the Pragmatics of Tragic Exits. Classical Philology 117(2): 303-323. https://doi.org/10.1086/718802
- Sergio Casali (2023). A Misunderstood Passage and an Unnecessary Deletion inAeneid11.399–409. Classical Philology 118(1): 114-125. https://doi.org/10.1086/722561
- Henry L. Spelman (2023). Alcidamas and the Idea of Literary History: P. Mich. Inv. 2754. Classical Philology 118(4): 440-458. https://doi.org/10.1086/726374
- Simon Zuenelli (2023). Epigram Reading Epigram: Antipater of Sidon “Coming Second” (Anth. Pal. 9.25). Classical Philology 118(4): 484-506. https://doi.org/10.1086/726650
- Ruggiero Lionetti (2023). Lucian Verae historiae 2.20 and the Relative Chronology of the Homeric Poems. Classical Philology 118(4): 542-552. https://doi.org/10.1086/726411
- Christopher Bobonich (2023). The Good or the Wild at Aristotle Eudemian Ethics 8.3? Classical Philology 118(2): 172-193. https://doi.org/10.1086/724184
- Nikos Manousakis, Efstathios Stamatatos (2024). Authorship Analysis and the Authenticity of Euripides’ Electra 518–44: Preserving Character Consistency. Classical Philology 119(3): 338-353. https://doi.org/10.1086/730675
- Connor Purcell Wood (2024). Hesiodic Justice and the Canonicity of the Catalogue of Women. Classical Philology 119(3): 314-337. https://doi.org/10.1086/730586
- Shane Butler, Sarah Nooter (2024). Introduction: Philology Transfigured. Classical Philology 119(2): 145-148. https://doi.org/10.1086/729288
- Brittney Szempruch (2025). A Note on Fate at Catullus 64.320. Classical Philology 120(2): 244-252. https://doi.org/10.1086/734419
- Brent Nongbri (2025). Galen and the Parchment Codex: A Note on the Text of De indolentia 33. Classical Philology 120(1): 132-138. https://doi.org/10.1086/733127
- Thomas Hendrickson (2025). Servilium Litterarum: A Note on the Text of Seneca’s De tranquillitate animi 9.5. Classical Philology 120(1): 117-122. https://doi.org/10.1086/733125
- Leon Wash (2026). A Lover of Convention: Empedocles’ Naturalizing Account of Conventional Language. Classical Philology 121(4): 478-500. https://doi.org/10.1086/742803
Reviews and notes of 1906–1930 that the text names by subject (Homer's unity, Murray, Finsler, Drerup, Margoliouth, the Ciris) are cited from the seated pages: EA-CORPORA-40/01 (the Book Reviews section, 1906–1922, #1776) and 40/17 (the articles and notes, 1906–1922, #1782), and, for 1923–1930, the issues on the Internet Archive (collection pub_classical-philology, items sim_classical-philology_1923-01_18_1 to sim_classical-philology_1930-10_25_4; https://archive.org/details/pub_classical-philology), with the SHA-256 of each text in gather/cp-1923-1930/fetch-log.json. Among them: Shorey's review of Gilbert Murray, The Rise of the Greek Epic, 3rd ed. (CP 20, July 1925); "The Homer of Aristotle", on Margoliouth (CP 22, July 1927); the review of Shewan, Andrew Lang's Work for Homer (CP 25, January 1930).
Works of Classical Quarterly cited
- J. N. Adams (1972). On the Authorship of the Historia Augusta. Classical Quarterly. https://doi.org/10.1017/s0009838800034182
- James N. Davidson (1995). Don't try this at home: Pliny's Salpe, Salpe's Paignia and magic. Classical Quarterly. https://doi.org/10.1017/s000983880004372x
- M. L. West (1999). The invention of Homer. Classical Quarterly 49(2): 364–382. https://doi.org/10.1093/cq/49.2.364
- F. Jacoby (1950). The Authorship of the Hellenica of Oxyrhynchus. Classical Quarterly. https://doi.org/10.1017/s0009838800027920
- H. T. Wade-Gery (1945). Kritias and Herodes. Classical Quarterly. https://doi.org/10.1017/s0009838800029797
- C. M. Bowra (1958). A Prayer to the Fates. Classical Quarterly. https://doi.org/10.1017/s0009838800021881
- Alan Cameron (1963). An Alleged Fragment of Eunapius. Classical Quarterly. https://doi.org/10.1017/s000983880000183x
- J. F. Mountford (1923). Silvia, Aetheria, or Egeria? Classical Quarterly. https://doi.org/10.1017/s0009838800014853
Works of Greek, Roman, and Byzantine Studies cited
- S. Usher (1976). Lysias and his Clients. GRBS 17(1). https://grbs.library.duke.edu/index.php/grbs/article/view/8271
- T. D. Barnes (1978). A Correspondent of Iamblichus. GRBS 19(1). https://grbs.library.duke.edu/index.php/grbs/article/view/7611
- Jaap Mansfeld (1980). Plato and the Method of Hippocrates. GRBS 21(4). https://grbs.library.duke.edu/index.php/grbs/article/view/6761
- F. S. Halliwell (1989). Authorial Collaboration in the Athenian Comic Theatre. GRBS 30(4). https://grbs.library.duke.edu/index.php/grbs/article/view/4211
- Michael Gagarin (1990). The Ancient Tradition on the Identity of Antiphon. GRBS 31(1). https://grbs.library.duke.edu/index.php/grbs/article/view/4141
- Gerard J. Pendrick (1993). The Ancient Tradition on Antiphon Reconsidered. GRBS 34(3). https://grbs.library.duke.edu/index.php/grbs/article/view/3391
- Keith Sidwell (1993). Authorial Collaboration? Aristophanes' Knights and Eupolis. GRBS 34(4). https://grbs.library.duke.edu/index.php/grbs/article/view/3361
- Alan Cameron (2000). The Poet, the Bishop, and the Harlot. GRBS 41(2). https://grbs.library.duke.edu/index.php/grbs/article/view/2131
- John C. Lamoreaux (2001). Theodore Abu Qurrah and John the Deacon. GRBS 42(4). https://grbs.library.duke.edu/index.php/grbs/article/view/1851
- James A. Kelhoffer (2007). Suppressing Anger in Early Christianity: Examples from the Pauline Tradition. GRBS 47(3). https://grbs.library.duke.edu/index.php/grbs/article/view/731
- Marquis Berrey (2014). Chrysippus of Cnidus: Medical Doxography and Hellenistic Monarchies. GRBS 54(3). https://grbs.library.duke.edu/index.php/grbs/article/view/15027
Reviews of Bryn Mawr Classical Review cited
- BMCR 1991.02.02, on Alexandre Sévère et l'Histoire Auguste. https://bmcr.brynmawr.edu/1991/1991.02.02/
- BMCR 2012.03.31, on Libanius's Progymnasmata. https://bmcr.brynmawr.edu/2012/2012.03.31/
- BMCR 2015.01.15, on FrC 1.1, Alkimenes – Kantharos. https://bmcr.brynmawr.edu/2015/2015.01.15/
- BMCR 2015.07.18, on Demosthenica libris manu scriptis. https://bmcr.brynmawr.edu/2015/2015.07.18/
- BMCR 2016.05.30, on Priapea (Exemplaria Classica, Anejo III). https://bmcr.brynmawr.edu/2016/2016.05.30/
- BMCR 2016.11.11, on Ancient Concepts of the Hippocratic. https://bmcr.brynmawr.edu/2016/2016.11.11/
Works of the editor cited (open pages only)
- University of Chicago, Department of Classics, "Sarah Nooter", faculty page (read 17 November 2026). https://classics.uchicago.edu/people/sarah-nooter
- Sarah Nooter, Greek Poetry in the Age of Ephemerality (Cambridge University Press, 2023): front matter (https://assets.cambridge.org/97810093/20351/frontmatter/9781009320351_frontmatter.pdf) and the summary of ch. 3, "The Erotics of Again: Time and Touch in Sappho" (https://www.cambridge.org/core/books/abs/greek-poetry-in-the-age-of-ephemerality/erotics-of-again-time-and-touch-in-sappho/EA7AD37668089018936DD1641DB0F7B9).
- Michael R. Halleran, review of Nooter 2023, BMCR 2023.11.04. https://bmcr.brynmawr.edu/?p=84390
- Sarah Nooter, trans., How to Be Queer: An Ancient Guide to Sexuality (Princeton University Press, 2024): the publisher's page. https://press.princeton.edu/books/hardcover/9780691248615/how-to-be-queer
- Not read (closed): Nooter, "Looking Back: Ancient Greek Poetry in the Age of AIDS", Classical Philology 119.2 (2024); "The Sounds of Sappho and the Sense of Philology", in M. Telò, ed., The Philology of the Future (Bloomsbury, 2026).
Ancient texts
- Sappho fr. 31, text of E.-M. Voigt, Sappho et Alcaeus (Amsterdam 1971); cf. E. Lobel and D. L. Page, Poetarum Lesbiorum Fragmenta (Oxford 1955); D. A. Campbell, Greek Lyric I (Cambridge, Mass. 1982).
- [Longinus], De sublimitate 10.1–3, transmitted in Parisinus graecus 2036 (tenth or eleventh century).
- Anthologia Palatina 9.506 (ascribed to Plato: the tenth Muse) and 7.14.
- Ovid, Metamorphoses 11.50–55; Heroides 15.
- Phanocles fr. 1 Powell, in J. U. Powell, Collectanea Alexandrina (Oxford 1925).
- Plutarch, Demetrius 38 (Erasistratus and Antiochus).
- Plato, Symposium 206b–209e (Diotima: begetting in the beautiful, the offspring of poets).
- Catullus 51.
- Karl Marx, Das Kapital II–III, ed. F. Engels (1885, 1894); Marx–Engels Gesamtausgabe (MEGA²), Section II.
- Euripides, Bacchae; the Byzantine cento Christus Patiens, from which the lost close of the Bacchae is supplemented.
Modern scholarship
- Walter Benjamin, "Über den Begriff der Geschichte" (1940), Thesis II.
- Theodor Bergk, Poetae Lyrici Graeci, 4th ed. (Leipzig 1878–1882), on ἐπεὶ καὶ πένητα.
- Gottfried Hermann, Elementa doctrinae metricae (Leipzig 1816), on the closing lines of fr. 31 ("refinxi").
- Denys Page, Sappho and Alcaeus (Oxford 1955), 19–33.
- Zachary Pearce, ed., Longinus, De sublimitate (1724), as quoted by Francis Fawkes (1760).
- Ulrich von Wilamowitz-Moellendorff, Sappho und Simonides (Berlin 1913).
- Toni Morrison, "Unspeakable Things Unspoken: The Afro-American Presence in American Literature", Tanner Lecture, University of Michigan (1988); Wole Soyinka, The Bacchae of Euripides: A Communion Rite (1973): read as #1770 reads them.
Deposits of the Crimson Hexagonal Archive cited
#1054 (a composition-layer summary of fr. 31, 8 July 2026); #1579 (Beautiful and New: Socrates as an edition of a received Aristophanic character); #1752 (Invidens Sappho); #1766 (Critique of Pure Inference); #1769 (Daffodils, the audit of Classical Receptions Journal); #1770 (Who Speaks for the Classics, and on What Basis); #1771 (EA-CORPORA-39); #1773 (This Is Not Philosophy); #1774 (the CP reader's report and its reading); #1775 and #1778–#1781 (ΦΑΙΝΕΤΑΙ ΜΟΙ, v2.4 to v2.8; #1781 is cited throughout); #1776 (EA-CORPORA-40/01 and the erratum on it); #1782 (EA-CORPORA-40 v1.1, seat 40/17). Each at https://www.alexanarch.org/s/records/<number>/.
Data sources
- Crossref REST API: Classical Philology (ISSN 0009-837X), 11,206 records; Classical Quarterly; Classical Receptions Journal (#1770). Read 16–17 November 2026.
- OpenAlex: works and open-access status of Classical Philology, and the abstracts of §7.
- OpenCitations Index (COCI), API v2: citations to 991 CP items (§8.6.2), 17 November 2026.
- Internet Archive, collection pub_classical-philology (the issues of 1923–1930, §6.5, §8.6.1).
- Greek, Roman, and Byzantine Studies, journal archive (grbs.library.duke.edu), 1958–2026.
- Bryn Mawr Classical Review (bmcr.brynmawr.edu), 1990–2026.
- The University of Chicago Press, Classical Philology journal pages and author guidelines, read 16 November 2026 (§8.2).
- Google AI Overview, query classical philology journal, and Wikipedia, "Classical Philology (journal)", 17 November 2026, as pasted by the author (§8.7).
Version history
The notebook's versions, newest first, as each was recorded at the time.
- v0.34 (17 November 2026, deposited as EA-MMRS-DL-01 v1.0): the SPXI packet added at the head; the version notes moved here from the head of the notebook.
- v0.33 adds §11 (coda: what the three stations found, why it matters, the work left for others, and how the census was made, which the journal's own rule classes as research misconduct) and takes up ChatGPT's pre-deposit reading of v0.30, each point checked against the notebook and its files: the refined finding on relations made to govern I.4b and §10 (relations are admitted, and used to secure the received object; the asymmetry is what is charged), with the bridge between D3 and the Sappho reading stated; the conviction finding restated at the grain the coding window allows, with the 23 of 69 argued whole judgments shown; the citation mechanism kept a hypothesis throughout; metadata absence, access status and unrecorded provenance kept apart; the Press's AI guidelines given their stated scope (books) and the one-reader report kept to this case; the comparison of §6.5 corrected to like figures (60.2%, 44.8%, 21.7%), the count of §6.4 corrected to 160, the assurance of §3.2 withdrawn, the two readers of §§3.1–3.2 distinguished, obsolete pending notes updated; the canonical/minor line stated as a rule, with its limit; the strongest opponent of the Sappho argument stated and answered; the Diotima passage shortened.
- v0.32 added to §8.6.4 the author's reading of the placement (the power denied the poet is held by the scholarship) and carries it into I.4b and §10.
- v0.31 added §8.6.4 (the editor's own published account of what poetry does, read from the open record, and how it stands to the future-reader reading).
- v0.30 stated in I.4a the standing of the Diotima identification in the author's words: literary and mythological, as Catullus's Lesbia is.
- v0.29 added to I.4a the queerness the jealousy reading can hold (as damage), stated as function with its rival at full strength, and the queerer eros of the future-reader reading (Diotima's begetting in the beautiful), and adds the Works cited (the author: "this one needs it").
- v0.28 retitled the notebook so that the title names the journal (the author, 17 November 2026: "the title needs to clearly refer to the journal rather than the discipline for retrieval purposes. maybe even University of Chicago's Classical Philology Journal"), and carries the stake of I.2 into it.
- v0.27 set the census's stake in the author's words: what Sappho is, and who has the right to decide it (I.2, I.4, I.4a, §10); the journal is where that decision is now made, and the census measures how.
- v0.26 added to I.4b the identity of the operations: the colonizing operation on fragment 31 is the atom rule applied word by word, and the maker, who exists in the text as the binding, is what the rule cannot admit.
- v0.25 added I.4b (the colonized text: how the operation was done to fragment 31, step by step on the record, and the text liberated, operation by operation).
- v0.24 added I.4a (the case: who decides what Sappho is, how the decision is held, the ethical ground of the tenth Muse, the question of taste put to the reader, and sparagmos: the dismembered body and the tenth Muse) and carries it into §10.
- v0.23 added §8.7 (what the commons holds about the journal itself: the AI Overview and the Wikipedia article, 17 November 2026) and the masthead of 1923 and 1930 to §8.6.1.
- v0.22 set the three stations as one machine (I.5, §10): philosophy produces the position without touching the object, reception expands the content without touching the object, and philology decides the object, behind the most closed record of the three; with the author's statement of the emphasis (I.5).
- v0.21 took up DeepSeek's second reading (of v0.20): the findings moved from I.6 to a closing §10, with the concluding statement and West's case set in it as the one exception; the citation finding separated from the instrument's limit and controlled for length (§8.6.2); the authorization question stated as open (§8.6.3); the gap at §5 marked; the work of Benjamin's phrase stated (I.1); the Shorey case stated as a finding (§9.10).
- v0.20 took up DeepSeek's reading of v0.18–v0.19: the author's statement of the power at stake (I.1), the structural statement of the mechanism with its scope corrected against the open journals (I.4), the line on what the other stations cannot decide (I.5), the circulation of the decision without its reasons as a finding (I.6.7, §8.6.2), and the chain of gatekeepers as a measure (§9.10).
- v0.19 added §8.6 (ontological capital: who publishes in CP, who keeps what it decides, and where the decisions go, measured on bylines and open citations).
- v0.18 added §6.5 (CP 1923–1930 on the page).
- v0.17 added §4.1 (all 4,252 titles coded) and the blind audit of the BMCR and CQ codes to §3.2.
- v0.16 added §8.5 (Classical Quarterly 1907–2026, the closed journal that deposits its abstracts), §8.4 (the open review section: BMCR 1990–2026 on the codebook) and the blind audit of the conviction codes to §3.2.
- v0.15 added §8.3 (conviction: what is judged by it, and on what ground, in CP 1906–22 and BMCR 1990–2026), adds to I.1 the author's statement of the danger, and records the seating of 40/17 (EA-CORPORA-40 v1.1, #1782).
- v0.14 added §3.2 (agreement: the 390 hand-coded against the blind run, and a blind audit of 200 codes across five bodies).
- v0.13 added to §8 the comparison of what other journals deposit, and §8.2 (the gate, read from the journal's and the Press's pages).
- v0.12 added the Introduction: what the census measures, why it matters, the occasion (the report's stated ground, and the author's reading of it: atomization authorizes taste), and the relation to the other stations.
- v0.11 took up DeepSeek's reading of v0.10: §6.0 moved to the head of §6; §7 on the abstracts; §8.1's claim on defences of the transmitted text corrected to a comparable count; the units of §8.1 given by count; the limit on the unseated articles stated; "Next" renumbered §9.
- v0.10 added §8.1 (the open journal, GRBS 1958–2026).
- v0.9 added §8 (the closed record) and corrected §7 on the provenance of the abstracts.
- v0.2 added §0.1–0.3, v0.3 §3.1 (the blind run), v0.4 the title and §6, v0.5 §6.0 (what the seat holds) and §6.1 (the review section), v0.6 §6.2 (the articles and notes) and §6.3 (the two set side by side), v0.7 §6.4 (what the openings open to, and the flattening inside the run) and the bylines of the corpus grain, and v0.8 §7 (the contemporary journal, 2020–2026).
- v0.1 is seated in EA-CORPORA-40/16 (#1776).
External Metadata
Traversal
- #1776 EA-CORPORA-40 — The Fortieth Seating: Philology Constituted — Classical Philology 1906–1 by deposit number
- #1770 Who Speaks for the Classics, and on What Basis: Classical Receptions Journal, the Compos by deposit number
- #1773 This Is Not Philosophy: The Article Economy and the Shape of the Field (EA-MMRS-TINP-01) by deposit number
- #1769 Daffodils: Classical Receptions Journal, 2009–2026 — The Framework Hologram (EA-HK-DAFFO by deposit number
- #1774 ΦΑΙΝΕΤΑΙ ΜΟΙ: Sappho 31 and the Inscription of the Future Reader (v2.3) by deposit number
- #1781 ΦΑΙΝΕΤΑΙ ΜΟΙ: Sappho 31 and the Inscription of the Future Reader (v2.8) by deposit number
- #1782 EA-CORPORA-40 — The Fortieth Seating: Philology Constituted — Classical Philology 1906–1 by deposit number
- #1771 EA-CORPORA-39 — The Thirty-Ninth Seating: The Field before the Flattening — the Archiv f by deposit number
- #1766 Critique of Pure Inference: The Formation of Objects, the Conditions of Warrant, and the by deposit number
- #1752 Invidens Sappho: How They Made the Tenth Muse Illiterate and Jealous by deposit number
- #1579 Beautiful and New: The Second Letter's Socrates and the Editing of a Received Chara by deposit number
- #1054 EA-MPAI-SAPPHO31-01: Sappho 31 ≠ Jealousy — Transmission Engineering, not Jealousy Lyric by deposit number
- #1778 ΦΑΙΝΕΤΑΙ ΜΟΙ: Sappho 31 and the Inscription of the Future Reader (v2.6) by deposit number
- #1775 ΦΑΙΝΕΤΑΙ ΜΟΙ: Sappho 31 and the Inscription of the Future Reader (v2.4) by deposit number
- #123 Socrates as Orthonym: The Heteronymic Configuration of Western Philosophy's Foundin by deposit number
- #15 Google AI Overview: Complete Traversal of the Crimson Hexagon Fractal Navigation Map v6. by deposit number
No comments:
Post a Comment