Putting the names back in

Naql is an atlas of carried books. It records, for 27 classical texts, every crossing from one language to another that scholarship can support: who made the translation, where, in which decade, for whom, and on what evidence.

The history of ideas tends to get told as weather. Greek science "flowed" into Arabic; learning "passed" to Europe; influence "spread". Nothing flowed. A specific physician memorized a specific book in India because his king sent him for it. A specific monk in a specific scriptorium chose one text over another because a storm had soaked the rest. Every crossing in this atlas was a decision by a person, usually a person with a name, and the atlas exists to put the names back in.

Why "naql"

Naql نقل is the Arabic word for carrying across: transport, conveyance, transmission. The Islamic tradition uses it for the transmitted sciences, the knowledge that arrives by being handed down. And it is the word the great age of translation used for its own work; a translator was a nāqil, a carrier. English hides the same idea in plain sight: translate descends from the Latin translatus, carried across. Both languages agree that a book does not move. Someone moves it.

The isnad standard

Hadith scholarship built the most demanding transmission discipline any civilization has produced. A report is only as credible as its isnad, the chain of named people who passed it along, and the science of grading those chains filled libraries. This atlas borrows the standard and applies it to books. Each chain here is an isnad for a text. Every link names its carriers and cites its evidence, and every link carries one of three grades. Attested means the medieval record names the act, or modern scholarship has settled it. Probable means the standard view rests on real but incomplete evidence. Disputed means the experts disagree, and the disagreement is part of the record, so it is recorded.

The grades do real work. Michael Scot's Latin De Anima is probable, because only a few of fifty manuscripts name him and the rest is stylometry. The Dutch Hayy ibn Yaqzan of 1672 is disputed, because the book is anonymous and the Spinoza circle attribution is an argument, however good. An atlas that graded everything "attested" would be lying about how history is known.

What no one planned

The striking thing about the whole story is that nobody designed it. A caliph wanted star tables that worked. A patriarch needed Aristotle's manual of dialectic for a staged debate. A Cordoban court could not read its own diplomatic gift and had to write to Constantinople for a tutor. An Italian wanted one astronomy book so badly he learned Arabic and stayed in Toledo for forty years. Each acted on local knowledge of a local need, and the sum of those decisions moved the libraries of three civilizations.

Hayek argued that knowledge always works this way: it lives dispersed in particular minds in particular places, and no central plan can gather it. The translation movement is a thousand-year demonstration. There was no ministry of translation. Even the one institution legend supplies, a caliphal House of Wisdom directing the work, mostly dissolves on inspection; what the record shows instead is patrons, prices and craftsmen (Gutas 1998). The books crossed because many hands found many separate reasons to carry them.

Method, and what the checking changed

The dataset is small on purpose. 27 works is few enough to verify properly and enough to show the shape of the thing. Every chain was checked against the scholarship on the sources page before it entered the dataset, and the checking changed it: dates got hedged, attributions got demoted from fact to probable, and one good story about a translation made at a royal court turned out to be an anachronism and was cut. Where the atlas says c. 1220-1224 rather than 1220, the vagueness is the accuracy.

Some rules the data must pass on every build: every crossing cites at least one source. Every chain is a tree with no cycles. Nothing is dated before the version it descends from. The validator that enforces this runs before the site can build, so the deployed atlas and the checked atlas are the same thing.

Naql was built by Adnan Abbasi with heavy AI assistance, including the research pass: independent verification runs checked each draft chain against published scholarship, and their corrections were applied before anything shipped. Errors that survive are mine. The repository takes issues and pull requests, and corrections that come with citations get merged with thanks.

Colophon

The atlas began as Naql, a standalone open-source project, and now lives inside Falsafa as its historical justification; the dataset is six JSON files and one validator, no database. Original titles are set in Amiri and the Noto faces for Devanagari, Hebrew and Syriac, with Greek in Source Serif 4. Code is MIT, the dataset is CC BY 4.0. An open-source project by Thothica.