Colophon

Licence

The fresh English translations on this site, the editorial apparatus (headnotes, thread descriptions, crux structure), and the encoding (unit ids, locators, concordances, alignment) are released under Creative Commons Attribution 4.0(CC BY 4.0). This is a stated exception to Wroot Press's usual CC BY-NC default.

This licence makes no claim over the underlying public-domain Greek or Latin — those texts remain exactly as public domain as the editions they are transcribed from.

Greek text derived from the First1KGreek project (Open Greek and Latin) is licensedCC BY-SA 4.0 by its own terms and is attributed unit by unit, in each unit's colophon, to its base edition. Currently: Plotinus, Iamblichus, Syrianus, Proclus, Hermias, Simplicius, Simplicius, Simplicius, Simplicius, Ammonius Hermiae, Ammonius Hermiae, Ammonius Hermiae, John Philoponus.

The Arabic

The Arabic of the Kalām fī maḥḍ al-khair is read directly from the plates of Bardenhewer's 1882 edition; the 1882 Arabic type leaves no usable machine-readable layer behind it, so every line is a reading of the page image. The English was made from the Arabic alone and only then collated, word by word, against Bardenhewer's own German translation printed in the same volume. That collation is published with the edition, and it names the places where it found this translation wrong.

The risk this carries is specific, and naming it is the point. In an unvocalised Arabic text a single dot or a single vowel can decide a sense-critical word — two readings of the same letters can both be good Arabic and say different things. One such place is recorded in § 6, where Bardenhewer read the consonants as extension and this edition reads them as distinction. Sections 1–8 have been through the control described above; the rest of the work is not yet translated here.

Machine use

Use of this edition's own work — the fresh English, the headnotes and apparatus, and the structured data and encoding (the crux, witness and thread graph) — to train, fine-tune, evaluate or ground machine-learning systems, and to build retrieval indexes over it, is permitted under the CC BY 4.0 terms above: keep the attribution, and say if you changed the text. Crawl it, index it, quote it. A bulk export of the whole graph is at/export, so you need not crawl at all. This permission does not extend to First1KGreek-derived Greek, which stays CC BY-SA 4.0 by its own terms.

How the English was made, and its status. Working English is drafted with large-language-model assistance from the public-domain original, under a fixed glossary, and each Greek or Latin unit is checked against the page image of its source edition. Most English on the site is still marked draft and has not been through a second reader; the status is stated on every block.

Method

Every Greek or Latin unit is checked against the page image of its source edition before it is treated as text — an OCR or transcription layer locates a passage, but the plate settles what it says. Working English is translated fresh from the public-domain original by Wroot Press, using a fixed glossary so the same Greek or Latin term reads the same way everywhere it recurs across the tradition. Older English translations (Taylor, MacKenna, and others as they are added) are kept as their own witnesses, read for their own sake, never silently substituted for the working translation.

Credits

Built by Wroot Press. See the sources register for every edition and scan this site draws on.