EA-CORPORA-02 is the gathering record of the second corpus-seating run of the Crimson Hexagonal Archive's primary-source library, executed in a single session on 22 August 2026 under the EA-OMEGA-BUILD-01 program. The run added nine corpora to the fifteen seated by EA-CORPORA-01, bringing the library to twenty-four, and is notable both for its scale — 61,692 flat ref-tagged lines emitted, 6,236 further verse units seated verbatim, roughly 700 MB of originals secured with per-file checksums — and for the four structural rulings it added to the library's seating doctrine.
The largest seat converted the prior gathering's largest declared absence: the Dead Sea Scrolls entered as the Abegg–Bowley–Cook electronic transcriptions (1,001 scrolls, 52,878 lines) via the ETCBC Text-Fabric dataset, whose Creative Commons BY-NC license is declared inside every data file and whose source line — 'Martin Abegg's data files, personal communication' — places the grant in the hands of the scholar who broke the corpus's forty-year publication monopoly in 1991. This seat motivated the first new ruling, the NC payload/record split, under which a NonCommercial payload license is carried forward unmodified while the record text remains CC-BY. The second ruling, the image-seat class, was established by the Apocryphon of John seat: the complete Codex II witness (thirty-two plates at native resolution, from the Robinson project's own negatives served openly by Claremont) was secured as images with transcription queued, converting the archive's oldest declared absence into custody. The third, the verbatim-seat class, arose from the Qur'an seat: the Tanzil Uthmani text arrives already reference-tagged under terms requiring verbatim distribution, so the original is seated as the machine-readable layer with no derived text emitted. The fourth, the sparse-tradition translation-witness rule, permits translations to be seated as witnesses, named as such, where a tradition survives in no other open form; the Mandaean corpus — five Lidzbarski and Brandt volumes, with the 1925 Ginza the only complete open Ginza in any accessible form — is its motivating case.
The run also made the library's comparative-witness program operational. The Communist Manifesto was seated as a witness pair (London 1848 against Cologne 1850/51, the latter's false imprint — applied with Marx's consent and exposed only by MEGA philology in 1996 — documented as an 1851 case of provenance inversion). The Greek Gospel of Thomas fragments (P.Oxy. 1, 654, 655) formed the first Greek–Coptic pairing against the Nag Hammadi Coptic text, and the Lucretius and Greek-atomists seats jointly hold the atomist transmission chain — Democritus and Leucippus fragments (Diels 1912), the Diogenes Laertius doxography with the complete Epicurus dossier, and De Rerum Natura — end to end, a case the record describes as machine-mediated reception studies avant la lettre. Richter's 1883 Leonardo transcriptions and the Kapital first edition of 1867 were secured with transcription queued.
The record documents its refusals as carefully as its seats: no Trismegistos number was guessed for the queued Gospel of Mary and Sophia fragments; the Häberl–McGrath open-access Mandaean edition, unreachable through its own publisher at fetch time, was pinned rather than seated from a provenance-murky copy; and Petermann's 1867 Ginza was declared absent with search terms and date. Talmud, Dante, and Ethiopic routes were verified and queued for dedicated sessions.