The Library · a supplemental page

Why we have
the manuscripts

One million and thirty-four thousand pages of the oldest writing humanity managed to keep. Nineteen collections. Held, not borrowed.

1,034,177

manuscript page images held in our own vault

measured 2026-08-15 · 19 shelves

separate collections, from five traditions

each shelf sized individually

1.53TB

of page imagery — the manuscripts alone

1,530,832,097,044 bytes

20,760/hr

still arriving, right now, around the clock

65 GB per hour · measured live

What this actually is

Not a subscription. Not an index of things held elsewhere. The page images themselves, on our own shelves, ours to read, ours to keep, and ours to teach a machine with.

The scale, made real

A person could not read it in a lifetime of trying

A million pages is a number nobody can feel. So here it is in hours: reading one page every minute, eight hours a day, without a day off — turning every page in this library exactly once takes nearly six years.

That is one pass. No study, no cross-referencing, no translation. Just turning the pages.

5.9 years

of continuous eight-hour days to turn every page once

1,034,177 pages ÷ 60/hr ÷ 8 hr/day = 2,154 days ÷ 365

Two-thirds of
the Eiffel Tower

if the leaves were stacked as physical folios

1,034,177 × 0.2 mm ≈ 207 m · Eiffel Tower = 330 m
0.2 mm = typical parchment leaf — stated assumption, not a measurement

Why it is ours and not rented

Nobody can take it back

Every one of these pages sits in storage we control. Not an API key that expires. Not a licence that gets revoked. Not a viewer that a library can switch off. If every institution on this list closed its doors tomorrow, the pages would still be here.

That mattered more than it sounds. Institutions restrict, reorganise and retire their digital collections constantly, and the terms change without notice. What is downloaded is settled. What is merely linked is on loan.

The shelves

Nineteen collections, grouped by the tradition they come out of. Every count below was measured directly from the vault, shelf by shelf, not estimated.

The Hebrew and Jewish world

349,319 pages · 177 GB

Overwhelmingly the Cairo Genizah — the storeroom of a Cairo synagogue where, for roughly a thousand years, nothing bearing the name of God was thrown away. Letters, contracts, children's writing exercises, medical notes, and scripture, all preserved by accident of reverence.

Genizah	The Cairo Genizah core — everyday documents and sacred text from the 9th century forward	321,461	125.4 GB
Cambridge Genizah	The Taylor–Schechter holdings at Cambridge	12,482	4.2 GB
JTS Genizah	Jewish Theological Seminary Genizah fragments	11,101	45.9 GB
Hebrew manuscripts	Hebrew codices outside the Genizah collections	4,275	1.4 GB

The Christian west

504,704 pages · 1.29 TB

The monastic and cathedral libraries of Europe — the institutions that copied scripture and classical learning by hand, generation after generation, and are the reason any of it survived at all.

Vatican Library	The Vatican Apostolic Library — one of the oldest libraries on earth	185,433	179.5 GB
e-codices	The medieval manuscripts of Switzerland — the highest-resolution imagery we hold	126,411	699.8 GB
Bavarian State Library	Munich — one of Europe's great manuscript holdings	70,159	116.8 GB
Parker Library	Corpus Christi, Cambridge — the Anglo-Saxon survival, including some of the oldest books in England	45,063	201.8 GB
Bodleian	Oxford	32,096	73.6 GB
Treasures	Curated flagship items across institutions	22,890	7.5 GB
Cambridge	Cambridge University Library, general manuscript holdings	18,053	4.9 GB
Walters	The Walters Art Museum — illuminated manuscripts	4,599	3.0 GB

The east

152,035 pages · 47.6 GB

The traditions that were writing, and preserving what they wrote, at the same time as everyone else — and which almost every western collection leaves out.

Japanese	Japanese manuscript and early printed holdings	57,629	24.8 GB
Sanskrit	The Cambridge Sanskrit collection — Indian religious and scientific texts	53,195	10.4 GB
Chinese	Chinese manuscript holdings	24,969	7.3 GB
Islamic	Arabic and Persian manuscripts — science, medicine, philosophy, scripture	16,242	5.1 GB

Science and medicine

16,679 pages · 4.8 GB

Small in page count, and the shelf most people would not expect us to hold.

Newton papers	Isaac Newton's own manuscripts — the notebooks, not the published books	16,370	4.5 GB
Historical medicine	Early medical texts	309	0.3 GB

The American record

11,440 pages · 14.6 GB

Library of Congress

Manuscript holdings from the Library of Congress

11,440

14.6 GB

Why we went and got them

Four reasons, and they compound. None of them is "because it is interesting."

One — the source, not the translation

Every answer traces back to what was actually written

Every layer between a reader and an original text is a place where a human decision got made — a translation choice, a doctrinal preference, an editorial cut. Most people never see those layers, because they only ever meet the finished product.

Holding the original manuscripts means a question can be taken all the way down. Not to what a commentary says the text says. To the page.

God said put me at the center — the original manuscripts without man's interpretation.Carter · August 3

Two — it is built into the machine

The system cannot finish a piece of work without going back to them

This is not an intention. It is a gate. Every task the system completes has to pass a final check that ties it back to the original manuscripts before it can be called done — a mechanism that no session, and no person, can step around.

That gate was built because intentions decay and mechanisms do not.

Every single thing we do goes back to the original manuscripts and the construction of the Kingdom of God and His Spirit… we need a mechanism that is mechanical that you or none of the sessions can get beyond unless that's part of it.Carter · August 7

Three — what we teach it with

Anyone can train on the internet. Almost nobody can train on this.

The frontier systems are all fed roughly the same thing: the public web, scraped. They are converging, and they are starting to sound like each other, because they are eating each other's output.

A library of a million original pages that predate the internet by centuries is not a bigger pile of the same material. It is different material — and it is the part that has already been tested by time. Texts that survived a thousand years of copying survived because people kept deciding they were worth the labour.

If we just took our brain and we got all the books… everything that's ancient forward that we can get our hands on and the original manuscripts, and designing it according to the Kingdom of God.Carter · August 1

Four — because they are disappearing

Preservation is not a side effect. It was a motive.

Physical manuscripts degrade. Institutions lose funding, reorganise, restrict access, and quietly take digital collections offline. Material that is freely reachable this year is routinely gone the next.

Every page pulled into our own vault is a page that stops depending on somebody else's budget. The collection is ordered oldest-first for exactly that reason — the most ancient material is the most fragile and the least replaceable.

What it leads to

The library is not the destination. It is the fuel, and it is already in the tank.

The next step

A measure of outcome the manuscripts define

Money coming out the other side of this system is easy to count, and everybody counts it. The harder question is whether the work is any good by a standard that does not move — and that is what a library like this makes possible.

Not our opinion of whether something was worthwhile. A judgement traceable to the oldest written record humanity kept.

We can have a king to watch the kingdom outcome based upon the original manuscripts.Carter · S1501

Still arriving — stated plainly

The collection is not finished, and this page does not pretend it is.

375,112 further page images are being re-acquired after the loss of the old server. At the current measured rate that is under a day of running.
Cambridge Sanskrit is mid-collection — roughly half its manifests fed so far.
Whole traditions are not represented yet — the Sinai monastery, the Oxyrhynchus papyri, and the great national libraries beyond those listed above.
Of everything that has been digitised on earth, what we hold is a sliver. That is the honest framing, and it is also the opportunity.