The most-read book on Source Library is Robert Fludd’s Utriusque Cosmi Historia — his 1617 history of the two worlds, the macrocosm and the microcosm, the source of that famous engraving of Nature as a chained woman standing between God and the ape of art. Thousands of people opened it here over the past months.
Here is the strange part. It has never been translated into English in full.
Not for lack of trying. In 1979, Patricia Tahil rendered the first two books of the macrocosm for Adam McLean’s Hermetic Sourceworks series; a recent edition adds part of the rest. But the whole work — five folio volumes — has never been carried into English. A reader who wants to follow Fludd’s actual argument, end to end, still cannot, four hundred years on.
I assumed, when we started measuring this, that the famous esoteric authors were the done part — that the gap was all obscure humanists nobody has heard of, and the Fludds and Kirchers and Ficinos had been safely translated long ago. I was wrong, and the way I was wrong turns out to be the whole story.
“Translated” hides more than it reveals
When you ask a catalog whether an author has been translated, it answers at the level of the author, not the work. Has anything by Fludd appeared in English? Yes — so the box gets ticked. But a sourcebook excerpt is not the book. A 1993 anthology that prints six pages of Fludd on music does not make the Utriusque Cosmi readable.
Athanasius Kircher is the sharper case. The seventeenth century’s most famous polymath — Oedipus Aegyptiacus, Mundus Subterraneus, Musurgia Universalis — thousand-page folios that shaped how Europe imagined Egypt, the underworld, and the harmony of the cosmos. What exists in English? A translation of one minor work, China Illustrata; some biographies; an anthology. His monumental works — the ones that made him Kircher — have never been translated at all. And his Mundus Subterraneus is, as I write this, the fifth most-read book on Source Library. Two of our ten most-read books have no English edition.
Even Marsilio Ficino, the man who started the Renaissance by translating Plato and the Hermetic writings into Latin, was largely untranslated until our own century. His Platonic Theology, his Plato commentaries, his letters — these reached English only in the 2000s, through Harvard’s I Tatti Renaissance Library, one beautiful volume at a time. The translator of the Renaissance waited five hundred years for his own translation.
This is why “first full translation” is a meaningful thing to claim, even for a famous author with his name in a few anthologies. The unit that matters is the work made whole — the thing a reader can actually sit down and read from beginning to end. By that measure, the giants are mostly still waiting.
The shape of the gap
Zoom out from the giants and it gets vertiginous. As the UCLA scholar Debora Shuger put it, “ninety percent of the Latin texts from the Renaissance have never been available in translation.”
We wanted to know what that means in books, so we counted — and the counting is most of the story. Start with the Universal Short Title Catalogue, the union catalog of nearly everything printed in Europe before 1700: about 1.6 million editions, of which roughly half a million are in Latin. Editions are not works — a popular text was reprinted dozens of times — so collapse them down to distinct titles, and you are left with on the order of four hundred thousand distinct Latin works.
Now the other side of the ratio. We gathered every English translation we could find — thirty catalogs, from the UNESCO Index Translationum to Harvard’s I Tatti library to library holdings worldwide — and matched them, work by work, against that corpus. The number that lands a translation is fewer than nine thousand. About one in fifty. You can search the whole census, author by author, at /census.
We have tried hard to break that figure, in both directions, and it keeps its shape. The count of translations is surely too low — catalogs are incomplete, and we already know we are missing volumes of I Tatti and Brill. And the corpus is surely a little too high — some of those “works” are reprints of antiquity that slipped the filter, and our edition-to-work clustering is conservative. But push every lever as far as it will honestly go and the translated share still comes out at a percent or two. The order of magnitude does not move. (The full method, and exactly where it is shaky, is written up at /research/translation-gap.)
And the work is slow. We measured the pace of genuinely new translations — distinct works opened for the first time, with retranslations of Plato and Cicero stripped out — across the late twentieth century: roughly twenty to thirty a year. It is accelerating, which is the hopeful part. But divide the hundreds of thousands still untouched by a few dozen a year and the arithmetic is brutal: at the human pace, finishing the Latin Renaissance alone would take on the order of ten thousand years.
Ten thousand years to read our own inheritance.
Why we built this
That number is the reason Source Library exists. We use AI to produce a first full translation of a work, and we place it beside a scan of the original page, so any line can be checked against the source — because a translation you cannot verify is not scholarship, it is a rumor. The machine is fast and fallible; the original is the ground truth; the reader is the judge.
It does not replace the scholars. The I Tatti volumes are masterpieces, and a careful human edition of Fludd will always be worth more than a machine draft. But ten thousand years is ten thousand years. The choice is not between a perfect translation and a quick one. It is between a quick one and nothing at all, for centuries.
The first full English translation of Fludd’s Utriusque Cosmi Historia is still ahead of us. So are tens of thousands of others. That gap — not the famous tip of it, but the whole dark mass below — is the work.
Counts drawn from the Universal Short Title Catalogue and a thirty-source federated translation census; method and caveats at /research/translation-gap. The Shuger quotation: UCLA, 2012. Fludd and Kircher translation status verified against WorldCat and the published record, June 2026.