Most of Galen has never been translated into English.

Galen of Pergamon (129 – c. 216 CE) is the largest surviving author of Greco-Roman antiquity: roughly three million words, by a common scholarly estimate about a tenth of all ancient Greek literature that survives from before 350 CE. Much of it can only be read in ancient Greek, medieval Arabic, or the Latin of an 1820s edition.

–works tracked

–no English translation known

–status unverified — help us check

–have at least one English version

The works without English are not the leftovers — they include most of Galen's largest books: the eleven-book On the Powers of Simple Drugs, the seventeen books of compound-drug recipes, the sixteen-book treatise cycle on the pulse, On Antidotes, On Crises, On the Differences of Fevers. Counted by words rather than titles, the untranslated share of Galen is far larger than the numbers above suggest.

How we read Galen

One text is not enough. We let the witnesses vote.

Much of Galen survives in more than one line at once — the Greek, and the ninth-century Arabic of Ḥunayn ibn Isḥāq, made from Greek manuscripts centuries older than any we still have. These are independent witnesses to a lost original, not a hierarchy. So we translate the Greek with the Arabic open beside it: where they agree we can be confident, and where they diverge we learn something the Greek alone would hide.

Take the most famous sentence in Greek medicine — “life is short, the art long.” Its third clause, ὁ καιρὸς ὀξύς, is a genuine puzzle: “the moment is sharp,” or “the time is short”? Ḥunayn's Arabic reads “the time is narrow” — and Galen, glossing his own words, calls the moment ὀλιγοχρόνιον, “of little duration.” The independent Arabic and the author himself agree, and the puzzle resolves. See the verse, triangulated →

What this site is

Three things, all in the open:

An atlas. A tracking table of the surviving corpus: each work, the language it survives in (Greek, Arabic, Latin, fragments), where the text is digitized, and whether an English translation exists. Each entry is flagged by how well its status has been verified, because half the battle is simply knowing what's missing. Coverage today: the 97 works digitized by the First1KGreek project plus 11 added by hand; roughly two dozen more surviving works are still to be entered.
A roadmap. A plan for translating the untranslated, starting with texts whose Greek is already digitized, and a public vote on which work gets the first full campaign.
Working translations. Not just a plan. A triangulated verse from the Commentary on the Aphorisms — the Greek read against Ḥunayn's independent Arabic — and the opening of the hair-and-scalp drug book, which had waited about 1,800 years for any English at all.

Honesty first. This is an amateur, AI-assisted project, not a critical edition. Translations are made from the Greek, checked against every independent witness we can reach — Kühn's facing Latin, and Ḥunayn's Arabic where it survives — and published with the source alongside so anyone can catch our mistakes. Catalog entries marked “unverified” are exactly that. Corrections are welcome — that's the point of working in public.

How to help

Vote on the first full translation campaign in the open poll.
Verify an entry. Pick any work marked “unverified” in the corpus table, find out whether an English translation exists, and file an issue or pull request with your source.
Translate a chunk. If you read Greek (or Arabic, or want to review AI drafts), see CONTRIBUTING.

How to contribute via pull request

The whole site is a public repository of plain HTML and one JSON file — no build step, no framework. If you can edit a text file, you can contribute:

Fork posix4e/pergamap on GitHub and make a branch.
Fixing the catalog? Edit data/works.json — update the work's english / english_ref fields, and set confidence to checked only if you cite where you checked in the pull request.
Adding a translation? Use the harness: one command generates a translation packet (the Greek with Kühn anchors, ready for your English), or grab an already-open packet from data/chunks.json. Fill in the translation plus a trace — your working, shown LLM-style, so the reasoning can be checked and not just the result. First claim the chunk in an issue so two people don't translate the same pages.
Preview locally by opening the HTML file in your browser — what you see is what deploys.
Open the pull request — one work or one chunk per PR, sources cited. Review happens in the open, arguing from the Greek. Merged means live.