Published rights research for a work the Archive cites but does not reproduce.

Rights dossier — Clement of Alexandria, Stromata (ANF II, Wilson 1885)

Researched: 2026-08-12, RIGHTS-DOSSIER drip night 8. Ticket LS-320, queue position 5.

Verdict: CLEAR on rights — but the English is INCOMPLETE, and the missing part has no public-domain English anywhere. Stromata Book III is printed in Latin, not English, by the deliberate decision of the edition's own editors. That is a coverage gate the ticket did not anticipate and the single most important thing in this dossier.

1. The edition

The Ante-Nicene Fathers: Translations of the Writings of the Fathers down to A.D. 325, Volume II — Fathers of the Second Century: Hermas, Tatian, Athenagoras, Theophilus, and Clement of Alexandria (Entire).

PD basis in one sentence: a US publication of 1885, reprinted 1913, both well before 1930, so the translation is public domain in the United States. Identical in kind to the basis already accepted for ANF vol. I (Barnabas, Justin) and vol. VI (Alexander of Cappadocia).

Read by eye

Title page — leaf n12: "THE ANTE-NICENE FATHERS. TRANSLATIONS OF The Writings of the Fathers down to A.D. 325. THE REV. ALEXANDER ROBERTS, D.D., AND JAMES DONALDSON, LL.D., EDITORS. AMERICAN REPRINT OF THE EDINBURGH EDITION. … BY A. CLEVELAND COXE, D.D. VOLUME II. … NEW YORK: CHARLES SCRIBNER'S SONS. 1913."
Copyright page — leaf n13: "COPYRIGHT, 1885, BY THE CHRISTIAN LITERATURE PUBLISHING COMPANY."
Contents of Volume II — leaf n10: Clement's works begin at p. 163; The Stromata, or Miscellanies at p. 299; Fragments 569; Who is the Rich Man that shall be saved? 589.
Printed page 381 — leaf n394, the opening of Book III, read in full (see §2).

Leaf offset for this item: printed page + 13. Verified at p. 381 (leaf n394). Cached (gitignored) at data/raw/rights-2026-08-12/anf02_n{6,8,10,12,13,394}.jpg.

Contents of Clement's section, with page ranges from the source's own markers: Introductory Note 165–170 · Exhortation to the Heathen 171–206 · The Instructor 207–298 · The Stromata 299–588 (Bk I 300–346, II 347–380, III 381–408, IV 409–443, V 444–479, VI 480–522, VII 523–557, VIII 558–568) · Fragments 569–588 · Quis dives salvetur 589–605.

2. THE COVERAGE FINDING — Book III is in Latin

At printed page 381 the book heads "THE STROMATA, OR MISCELLANIES. BOOK III." and the text beneath it opens "Ac Valentiniani quidem, qui desuper ex divinis emissionibus deduxere conjugationes…" — Latin, with Latin chapter heads ("CAPUT I.—BASILIDIS SENTENTIAM DE CONTINENTIA ET NUPTIIS REFUTAT."). It stays Latin for the whole book.

This is not a scanning accident or an OCR misread. The editors say so themselves, in a footnote hung off the book title and read by eye on the page image:

"After much consideration, the Editors have deemed it best to give the whole of this Book in Latin. [In the former Book, Clement has shown, not without a decided leaning to chaste celibacy, that marriage is a holy estate… He now enters upon the refutation of the false-Gnostics and their licentious tenets… this Book, though necessarily offensive to our Christian tastes, is most useful as a commentary upon the history of monasticism… The resolution of the Edinburgh editors to give this Book to scholars only, in the Latin, is probably wise. I subjoin a succinct analysis, in the elucidations.]"

Extent. The Latin runs pp. 381–401. Coxe's own English "ELUCIDATIONS" — a chapter-by- chapter summary, not a translation — follow at pp. 402–408. English resumes with Book IV at p. 409. The volume's title page claim "CLEMENT OF ALEXANDRIA (ENTIRE)" refers to the corpus, not to the language: every work is present, one book of one work is not in English.

The same treatment reaches into the Instructor. Book II, chapter X — "QUÆNAM DE PROCREATIONE LIBERORUM TRACTANDA SINT" — is Latinised the same way at pp. 259–263, with English paragraphs interleaved between Latin ones, and a Coxe footnote at the far end marking the resumption: "He now continues the argument in a form which our translators do not scruple to English." So the exposure is two sites, not one, and the second is easy to miss because it is mid-chapter rather than a whole book.

Is there a public-domain English Book III? No.

The first complete English translation of Stromateis III is Oulton & Chadwick, Alexandrian Christianity (Library of Christian Classics vol. II, SCM / Westminster Press, 1954), pp. 40–92. 1954 is decades past 1930; J. E. L. Oulton d. 1957 and Henry Chadwick d. 2008, so UK life+70 runs to 2078. Not usable on any theory.
John Ferguson's Stromateis Books 1–3 (Fathers of the Church vol. 85, CUA Press, 1991) is likewise modern and in copyright.
The Edinburgh Ante-Nicene Christian Library (T&T Clark, 1867–69) that ANF reprints is the same text with the same Latin — the footnote above names "the Edinburgh editors" as the ones who made the decision, so going back to the Edinburgh printing does not recover an English Book III. It never existed there.

This is the Origen, On Prayer shape rather than the Bernard shape: not "the queued copy is the wrong translation, find a better scan", but no public-domain English text exists to find. The difference from Origen is that here it costs us one book of eight, not the work.

3. The CCEL text — same edition, same Latin, no defect of its own

CCEL's https://ccel.org/ccel/schaff/anf02.xml transcribes this edition faithfully, Latin and all: its Book III carries the identical editors' footnote and page marker <pb n="381">. A CCEL-sourced build inherits the gap; it does not cause it.

Header, for the standing tally: anf02's ThML header is populated with authors, DC.Rights: Public Domain, and DC.Publisher: Grand Rapids, MI: Christian Classics Ethereal Library — CCEL's own imprint. It names no translator (Wilson is absent), no print publisher, no print year. Same refinement as recorded for npnf201 tonight: populated ≠ evidence, because none of the populated fields is a print fact. Wilson's attribution here comes from the scan ("[TRANSLATED BY REV. WILLIAM WILSON, M.A.]", printed at the head of Clement's works and again at the Fragments), not from CCEL.

Transcription audit — RUN, CLEAN. Same instrument and calibration as the Eusebius dossier (§4 there; method and script in the 2026-08-12 addendum to npnf110-transcription-defect.md), CCEL anf02 against the antenicenefather02robe OCR:

| Division | CCEL words | chained | suspect holes | | :-- | --: | --: | --: | | Exhortation to the Heathen | 34,062 | 89% | 0 | | The Instructor | 82,774 | 82% | 0 | | Stromata I | 43,516 | 83% | 0 | | Stromata II | 31,617 | 84% | 0 | | Stromata III (Latin) | 19,709 | 54% | 0 | | Stromata IV | 32,291 | 85% | 0 | | Stromata V | 31,320 | 87% | 0 | | Stromata VI | 37,982 | 82% | 0 | | Stromata VII | 32,779 | 84% | 0 | | Stromata VIII | 9,183 | 93% | 0 | | Quis dives salvetur | 13,979 | 90% | 0 |

Book III's lower chain rate is the OCR of Latin being noisier, not a signal. Positive controls on Book V: a 700-word excision at word 15,000 is flagged at 14,960 (12/12 shingles absent), a 1,200-word excision at 8,000 is flagged at 7,960 (12/12). The negative is therefore an instrument-checked negative, bounded as in the Eusebius dossier at roughly the ≥500-word scale.

One minor CCEL defect found and worth recording: Book VII contains a page marker <pb … n="436"> inside chapter vii, where the surrounding sequence runs 523–557. A stray, almost certainly a typo for 536. Harmless to us (we do not address by print page), but a converter that trusts page markers for ordering would trip on it.

4. What I could not establish

Whether the Instructor II.10 Latinisation has further sites elsewhere in Clement. The volume was swept with a Latin-density detector at 8,000-character resolution, which found exactly two regions (Instructor II.10, Stromata III). A shorter Latinised passage — a sentence or two inside an otherwise English chapter — would fall below that resolution. Before publishing, run a per-paragraph Latin check, not a per-page one.
What the study Bible actually cites Book III for, if anything. If its citations concentrate in Book III we would be shipping a work that cannot answer them; if they avoid it the gap is cosmetic. Unanswerable from here — we do not hold that apparatus as data.
Nothing was published, transcribed, or written to any database.

5. Recommendation

Proceed, with the gap disclosed rather than papered over. Record provenance in work.yaml as: translator Rev. William Wilson, M.A., editors Alexander Roberts and James Donaldson, American editor A. Cleveland Coxe, Ante-Nicene Fathers vol. II, Christian Literature Publishing Company, 1885 (this printing New York: Charles Scribner's Sons, 1913); rights Public Domain; PD basis pre-1930 US publication. Cite IA antenicenefather02robe. titleEnglish: Miscellanies, unitNoun: book.

Gates:

The provenance note must state that Book III is Latin in this edition, and why — the editors' own decision, quoted or paraphrased. A reader who opens Book III and finds Latin with no explanation will reasonably think our transcription broke.
Decide Book III's disposition explicitly, and record it. Three defensible options: (a) publish the Latin as printed, labelled; (b) publish Coxe's English Elucidations (pp. 402–408) in its place, labelled as a summary and not as Clement's text; (c) omit the book with a stated gap. Do not silently drop it — the contents would then show Books I, II, IV… and look like a transcription failure. This is a founder-facing call, not a converter default.
The Instructor II.10 Latin is the sleeper. Whatever is decided for Book III must be applied there too, and it is mid-chapter with English on both sides.
Never translate the Latin ourselves, machine or otherwise. It would be a new translation of a Father published under our name, which is exactly the class of claim the standards forbid.
The transcription audit is done and clean — do not re-run it as a gate, but honour the per-paragraph Latin sweep in §4.
Citation-index note: the Stromata has no entry in the 308-row cited-sources manifest; Clement of Alexandria appears there only as paedagogus-clement-of-alexandria (2 citations). LS-320's "heavily cited by the study-Bible apparatus" is a claim about the founder's photographed abbreviations list, not about our index. If Clement is built, the Paedagogus is the work with citations attached — and it is in this same volume, on this same basis, pp. 207–298, already audited clean above. That is a free second work, and it is the one our own data asks for.

Sources

https://archive.org/details/antenicenefather02robe — the dated scan; title page (n12), copyright page (n13), contents (n10) and p. 381 (n394) read by eye
https://ccel.org/ccel/schaff/anf02.xml — CCEL ThML; header read, text audited, Latin Book III confirmed present with the editors' note
Oulton, J. E. L. & Chadwick, H., Alexandrian Christianity (Library of Christian Classics II, 1954) — identified as the first complete English of Book III; in copyright, not usable

2026-10-01 (content night) — BOOK III's DISPOSITION IS DECIDED, AND A SECOND BLOCKER WAS FOUND THAT THIS DOSSIER DID NOT RECORD

Rights are unchanged and not re-researched. Two things settled, one of which is new.

1. THE DECISION ON STROMATA BOOK III: publish the Latin, labelled, AND Coxe's Elucidations labelled as a summary

The gate offered three options (publish the Latin labelled / publish the Elucidations labelled as summary / omit with a stated gap). Take the first two together and reject the third.

Why not omit. The gate's own objection stands: a contents list reading I, II, IV… looks like a broken transcription, and the reader cannot tell a deliberate editorial gap from a defect in our pipeline.

Why not the Latin alone. An English reader gets nothing usable for one book of eight.

Why not the Elucidations alone. Coxe's Elucidations are a chapter-by-chapter summary, in his own framing ("I subjoin a succinct analysis"). Shipping a summary in the slot where the book belongs misrepresents it, and it is the only option of the three that states something false.

Both, each labelled for what it is, is the only disposition that is complete and honest: Book III is present in the single language a public-domain text of it exists in, and Coxe's English analysis is present beside it as an analysis. This also follows the ladder the founder already ruled for devotional texts in DECISIONS.md ("Neither an English witness nor a second original-language printing → the original displays alongside the English, always", so that the reader becomes the witness of last resort). We never translate it ourselves, which the entry already forbids and this disposition never requires.

It is implementable with machinery that already exists, and the shape matters

work.json already carries a per-edition lang field and a reader-facing coverage: { status, detail } block (see apps/web/content/works/ignatius-to-the-romans/render/ work.json). coverage.detail is the purpose-built place for this gap, so the honest statement rides in the manifest rather than in prose somebody has to remember to write:

ONE edition (ANF, Wilson tr., 1885 / this printing 1913) with coverage.status NOT complete, and coverage.detail naming Book III's Latin and the Instructor II.10 interleaving explicitly.
Book III's nodes carry the Latin with lang="la" in their content and a heading that says so. There is no language field on WorkNode (checked against the schema), so the marker has to be in the content HTML or the heading. Do not invent a column for this without deciding it deliberately.
The Elucidations ship as their own clearly-labelled unit, never as Book III's text.
The Instructor II.10 is the harder half and must not be treated as a smaller copy of Book III. There, Latin and English paragraphs INTERLEAVE mid-chapter, so the labelling has to work at PARAGRAPH granularity. A book-level mechanism will silently mislabel it.
The dossier's existing requirement stands and is a gate on publishing, not advice: sweep for Latin per PARAGRAPH, not per page. The two known sites were found at 8,000-character resolution and a one-sentence Latinisation falls below it.

2. NEW BLOCKER, NOT PREVIOUSLY RECORDED: the Paedagogus's citation cannot be served from ANF

**This is queue item 11's blocker (Barnabas), Justin Martyr's, and the Augustine Sermones blocker found earlier tonight, for the fourth time. It is now the most common failure shape in this backlog.**

Measured tonight, against the database and against anf02.xml:

paedagogus-clement-of-alexandria is CITATION_ONLY with 1 DS stub, heading 1, 6, 27, carrying 2 incoming Connection edges. That is book 1, chapter 6, section 27, the Stählin section numbering.
ANF publishes the Instructor at BOOK and CHAPTER only. div3 is the book and div4 is the chapter ("Chap. I.—On Eating."), and Book I chapter VI contains 79 paragraphs with no section series at all — a numeral probe over that chapter returns two strays and nothing resembling a continuous 1..N. So the 27 has nothing in ANF to resolve against.
The Stromata is unaffected and this is the useful asymmetry: it has NO citation row at all (confirmed: no Stromata work exists in the database). So the Stromata is a brand-new work with no stub and no edges, takes load.ts, and must not fake a before/after edge assertion. It can ship at book/chapter granularity with nothing at risk.
The exposure is therefore 2 edges on the Paedagogus alone, not the whole pass. Options, none of which should be chosen unattended given the founder's Barnabas ruling: find a PD English Paedagogus carrying Stählin sections; or publish at book/chapter and let loadAdditive keep the 1, 6, 27 stub alive beside the text (nothing strands, but a reader arriving from the Catechism lands on a stub inside a work that now contains the passage); or leave the Paedagogus CITATION_ONLY and ship the Stromata alone.
The dossier's "build it in the same pass" recommendation above was made on coverage grounds and did not know this. It is still probably right, but it is now a decision with an edge cost attached rather than a free add-on.