Published rights research for a work the Archive cites but does not reproduce.
Rights dossier — Justin Martyr: Dialogue with Trypho, First and Second Apology (ANF I)
Researched: 2026-08-12, RIGHTS-DOSSIER drip night 8. Ticket LS-321, queue position 6.
Verdict: CLEAR. Same volume and same basis as the already-published Epistle of Barnabas. The cheap build the ticket promised is real — and it is now cheaper than promised, because the mandatory CCEL transcription audit was run tonight against the dated scan and comes back clean across all eleven divisions of Justin's corpus.
On our own citation index this is the highest-value of tonight's three works, not the lowest — 9 citations against Clement's 2 and Eusebius' 0. See §5.
1. The edition
The Ante-Nicene Fathers, Volume I — The Apostolic Fathers with Justin Martyr and
Irenaeus. The volume's rights basis is already on record from the Barnabas dossier
(rights-epistle-of-barnabas.md) and is unchanged: Christian Literature Publishing Co.,
1885, verified against IA antenicenefather01robe (Charles Scribner's Sons reprint, 1913).
PD by pre-1930 US publication.
Translators of the three works LS-321 names — read off the scan, not assumed: the
First Apology, Second Apology and Dialogue with Trypho carry the printed attributions
"[TRANSLATED BY THE REV. M. DODS, M.A.]" and "[TRANSLATED BY THE REV. G. REITH, M.A.]" —
Marcus Dods and George Reith, under the Roberts–Donaldson editorship. This matters:
work.yaml must not name Roberts and Donaldson as the translators of Justin. They are the
series editors; Dods and Reith did this English. The Barnabas provenance line (Roberts and
Donaldson as translators) is correct for Barnabas and would be wrong here.
Contents and page ranges (from the source's own page markers)
| Division | Printed pages | Chapters | | :-- | :-- | --: | | Introductory Note to the Writings of Justin Martyr | 159–161 | — | | The First Apology | 163–187 | 68 (+ 3 appended imperial epistles) | | The Second Apology | 188–193 | 15 | | Dialogue with Trypho | 194–270 | 142 | | The Discourse to the Greeks | 271–272 | 5 | | Hortatory Address to the Greeks | 273–289 | 38 | | On the Sole Government of God | 290–293 | 6 | | On the Resurrection, Fragments | 294–299 | 10 | | Other Fragments from the Lost Writings of Justin | 300–302 | 19 | | Introductory Note to the Martyrdom | 303–304 | — | | The Martyrdom of Justin Martyr | 305–306 | 5 |
Chapter counts match the standard printed counts (68 / 15 / 142). The First Apology's apparent 71 divisions are the 68 chapters plus the three appended documents — the Epistle of Hadrian to Minucius Fundanus, the Epistle of Antoninus to the Common Assembly of Asia, and the Epistle of Marcus Aurelius to the Senate. Those are not chapters of the Apology and should not be numbered as such.
2. The build is already half-done
anf01.xml is cached locally at data/raw/anf01_irenaeus.xml from the Irenaeus build
(4.5 MB, 55,922 lines), and ingestion/works/transcribe-anf.ts already parses this exact
file's ThML shape. The Justin corpus sits at div1 id viii, between vii (Papias) and
ix (Irenæus), with one div2 per work and one div3 per chapter — the same structure the
Irenaeus converter walks.
No download, no new parser, no new rights question. This is the cheapest remaining item on the whole acquisition list.
3. The CCEL transcription audit — RUN, and CLEAN
The standing rule from npnf110-transcription-defect.md is that every CCEL build runs
the audit first, and that the already-shipped ANF builds (Irenaeus, Alexander) should get a
retroactive pass "when convenient". Both were done tonight, since the files and the scan OCR
were already in hand.
Method and calibration: as in rights-eusebius-historia-ecclesiastica.md §4 and the
2026-08-12 addendum to npnf110-transcription-defect.md. CCEL text with notes included,
6-word probes, longest-increasing-subsequence chaining, >250-word gap threshold, ≥60% of
twelve OCR shingles absent to confirm.
Justin — no body-text loss in any division:
| Division | CCEL words | chained | suspect holes | | :-- | --: | --: | --: | | Introductory Note | 1,686 | 90% | 0 | | The First Apology | 24,975 | 80% | 0 | | The Second Apology | 5,099 | 84% | 0 | | Dialogue with Trypho | 75,123 | 87% | 0 | | The Discourse to the Greeks | 1,757 | 70% | 0 | | Hortatory Address to the Greeks | 16,024 | 87% | 0 | | On the Sole Government of God | 3,228 | 88% | 0 | | On the Resurrection, Fragments | 4,982 | 91% | 0 | | Other Fragments | 1,929 | 88% | 0 | | The Martyrdom of Justin Martyr | 1,217 | 77% | 0 |
Positive control: excising 700 words from the Dialogue at word 40,000 is flagged at 39,960 with 12/12 shingles absent. The negative is instrument-checked.
Retroactive passes — both clean, debt discharged:
- Irenaeus, Against Heresies (published as
adversus-haereses) — Books I, II, III, V and the Fragments all clean; Book IV produced one flag at CCEL word 40 spanning 391,133 OCR words, i.e. more than the volume's entire body, which is the chain-start artifact class described in the Eusebius dossier. Re-running Book IV from word 400 gives 0 suspects at 83% chaining. Structure also confirmed: Book IV is Preface + 41 chapters (42 divisions), matching the print. - Alexander of Alexandria (published as the
anf06epistles) — "Epistles on the Arian Heresy and the Deposition of Arius" clean at 70% chaining; the only flag is on the 1,019-word Introductory Notice at word 0, again a chain start. Positive control on the epistles fires correctly at a 700-word excision.
So the npnf110 defect does not extend to the ANF volumes. It remains what it was found to be: a defect in CCEL's NPNF series 1 volume 10, not a property of CCEL transcriptions in general. That is worth knowing precisely, because the opposite belief would have made every future CCEL build look unaffordable.
What the audit does not prove, restated so it is not overclaimed: it detects missing spans at roughly the ≥500-word scale. It does not detect word-level silent edits, and it does not detect a loss that preserves structure and falls under ~300 words.
4. The CCEL header, for the standing tally
anf01's header is the same shape as tonight's other two: it names editors and asserts
DC.Rights: Public Domain, with DC.Publisher giving CCEL's own Grand Rapids imprint. No
print publisher, no print year, and — as with anf02 — no translator, so Dods and Reith
would have been invisible had we trusted it. The refinement recorded tonight holds: a CCEL
header can be populated and still contain no print fact, and it is print facts that decide
rights and provenance.
5. Decision value — this is the one our own data asks for
apps/web/content/cited-sources/manifest.json carries two Justin rows:
apologiae-st-justin— 7 citationsdialogus-cum-tryphone-judaeo-st-justin— 2 citations
Nine in total, against Clement of Alexandria's 2 (Paedagogus; the Stromata has none) and Eusebius' 0 (no entry of any kind). The queue order 4–5–6 puts Justin last of tonight's three; our own index puts it first. The queue order reflects the founder's photographed study-Bible list, which is a legitimate basis — but where the two disagree it is worth saying which is which rather than letting "most-cited" do work it cannot support.
This also creates the night's one hard build gate, and it is the if-then.md rule that
Augustine's Confessions is the standing scar for:
Both Justin rows already exist as citation-only rows. Publishing under new slugs would orphan all 9 citations onto stubs while the full text sits unreachable beside them — exactly the
Confessionsoutcome (466 nodes, zero connections; the citation-only twin holding all 25 CCC citations). CheckSELECT slug, text_status FROM works WHERE author ILIKE '%justin%'first; givework.yamlthe existing slug and load withloadAdditive.ts, notload.ts; then runnpx tsx ingestion/works/verify-citation-twins.tsfromapps/api, which exits non-zero if a confirmed pair is left unjoined.
Note also that the manifest's two rows do not map one-to-one onto the printed works:
apologiae-st-justin is both Apologies under one slug, while the print (and the ThML)
separates the First and Second. Decide whether we host one work with two books or two works,
and make the join match — a mismatch here is how a citation lands nowhere.
6. What I could not establish
- Whether to host the seven minor pieces (Discourse to the Greeks, Hortatory Address, On the Sole Government of God, On the Resurrection, Fragments, the Martyrdom). All are in the volume, all are PD on the same basis, and all are cleared by this dossier — but several are of disputed authenticity, which is an editorial question for the founder rather than a rights question. The three works LS-321 names are the safe core.
- CCEL's text was audited for completeness against the scan but not proofread word-for-word; no fingerprint of a specific sentence was made against a page image here, since the volume's title-page facts were already read for Barnabas and the OCR-level audit is a stronger completeness check than a single spot-fingerprint.
- Nothing was published, transcribed, or written to any database.
7. Recommendation
Proceed — and if only one of tonight's three is built, build this one. Provenance in
work.yaml: translators Rev. Marcus Dods, M.A. and Rev. George Reith, M.A., editors
Alexander Roberts and James Donaldson, American editor A. Cleveland Coxe, Ante-Nicene
Fathers vol. I, Christian Literature Publishing Company, 1885 (this printing New York:
Charles Scribner's Sons, 1913); rights Public Domain; PD basis pre-1930 US publication;
source IA antenicenefather01robe.
Gates: the citation-twin join in §5 (hard); the First Apology's three appended imperial
epistles must not be numbered as chapters; titleEnglish for the Latin-titled rows; and a
founder call on the seven minor pieces.
Sources
https://archive.org/details/antenicenefather01robe— the dated scan; volume title-page facts already recorded inrights-epistle-of-barnabas.md, translator attributions for Justin read from the printed textdata/raw/anf01_irenaeus.xml— the cached CCEL ThML, audited against the scanapps/web/content/cited-sources/manifest.json— citation countsdocs/ops/if-then.md§"IF you publish a WORK THAT ALREADY EXISTS AS A CITATION_ONLY ROW" — the rule requiresloadAdditive.tswith the EXISTING slug, neverload.ts