Saving a Thousand Years of Knowledge: The Race to Digitize Arabic Manuscripts

posted on: Sep 2, 2026

By Shayla Frank / Arab America Contributing Writer

In 1831, a man born near the Gambia River sat down in North Carolina and filled 15 pages with Arabic. In it, he writes a story that is not widely known. Omar Ibn Said had been a scholar in Senegal until his capture in 1807, when slavers shipped him to Charleston. What he set down is the only known narrative in Arabic by an enslaved person in the United States. Unlike most Arabic manuscripts in the Library of Congress collections, it was written here and he wrote it while in captivity.

His story was passed from owner to owner and dropped out of sight for nearly half a century, until the Library of Congress bought them in 2017. By the time they arrived, the paper had gone weak, holed and torn along the creases. Conservators mended and reinforced the sheets, working, as Book Conservation Section head Shelly Smith put it, to strengthen the manuscript while letting its previous history remain evident.

Now, his story lives in the Library of Congress. 42 documents for anyone to read online for free. His story not only marks the importance of preserving Arab history in the United States, but the race to capture it before it’s gone.

Why Arabic Manuscripts Survive in Single Copies

The African and Middle Eastern Division holds more than a million items in the languages of 77 countries. Its Near East Section, which opened in 1945, keeps custody of the Arabic material. Most of those Arabic manuscripts date from the twelfth century to the early twentieth. So far 1,230 of them are online, gathered in the Library’s Arabic Language Rare Materials Collection.

More than most languages, dates matter in Arabic. Europe had movable type from the 1450s, but the first official press in the Arab world, the Bulaq Press in Cairo, didn’t  open until 1821. For four centuries scholars in Damascus or Fez were only abe to preserve knowledge in one way – by hand. Thus, European records often exist in editions of hundreds, whereas Arabic records only survive in single copies. The Library says outright that some of its own holdings are the only surviving copies anywhere.

What the Library Actually Holds

The collection is a set of archives assembled by very different routes, and the routes are the interesting part. The largest single body is the Mansuri Collection, the personal library of the Azhari scholar al-Imam Muhammad al-Mansuri. The Library puts it at 1,500 to 2,000 Arabic manuscripts, bought in the mid-1940s. However, the question of how a Cairo scholar’s library ended up in Washington years after the Second World War, isn’t a question the catalog answers.

Other material arrived as diplomacy. In 1884, Sultan Abdul-Hamid II sent 325 volumes of Ottoman Turkish, Persian, and Arabic works. He had them embossed with an inscription naming the recipient as the national library of the United States of America. The Ottoman empire was addressing a republic barely a century old as a peer institution. Less ceremoniously, the New York dealer Kirkor Minassian sold the Library Arabic works in the 1920s and Persian ones in the 1930s. Currently, 888 of these Persian literary works are now online. Furthermore, the Baker Collection arrived as a donation in the late 1970s.

Newer digitization has pushed well past the medieval. The Eltaher Collection runs to more than 2,000 items. It documents the Arab world from 1912 to 1974 through the papers of Muhammad Ali Eltaher, a Palestinian journalist and editor. A calligraphy collection adds another 273 pieces, Ottoman and Persian alongside Arabic. More than 30 manuscripts came from the Mamma Haidara Commemorative Library and the Library of Cheick Zayni Baye in Timbuktu. These articles aren’t owned by the Library but instead are digitized in partnership.

Inside the Digitization Process

Contrary to what many believe about digitizing, the process is slower than it seems. In order for articles to first be digitized it goes through a process in which the division picks out works of high research value and routes them through the wider digitization program and its outside partners. Before anything goes near a camera it needs a conservation assessment, and the items that fail go to conservators for treatment or rehousing. Cradles hold the bound volumes that pass. Rights and privacy clearance proceeds on a separate track entirely.

The bigger change of the past five years has been institutional rather than technical. The Library opened a dedicated Digital Scan Center in 2021 and adopted federal imaging guidelines whose current edition dates to 2023. Its strategy for the following four years commits the Library to steady, planned scanning of rare and unique holdings in place of one-off projects. That treats digitization as core preservation instead of an add-on, which for most of life only existed as such.

Cataloguing is the quieter constraint, and it binds tighter than the scanning does. A scanned page still needs someone who can read a fifteenth-century hand well enough to say what the book is. Unfortunately, that pool of specialists is small and shrinking. The division has said it is exploring machine learning to generate metadata. The aim is to connect holdings to the people searching for them. Until that works, many of the Arabic manuscripts already online sit in a state that is preserved but barely findable.

What the Race Is Actually Against

The race is not mainly a race against fire, though fire is part of it. Militants took Timbuktu in 2012, and reports of mass manuscript burning drew global headlines. Scholars later established that those reports were greatly exaggerated. Even so, the scare pushed international money toward digitization in a way decades of quiet appeals never had.

The Islamic State then destroyed manuscript collections in Mosul and across the Nineveh Plain beginning in 2014. In Yemen, as the Princeton historian Sabine Schmidtke documents, political turmoil and war damaged, looted, or destroyed many private libraries in the north across the twentieth century. The current conflict threatens what survived. Long before any of that, foreign collectors had bought up between 10,000 and 20,000 Yemeni manuscripts, which now sit in libraries abroad.

The Al-Furqan Islamic Heritage Foundation estimates that 3 million Islamic manuscripts survive worldwide. Against that figure the wartime losses are real but bounded. Nobody can say precisely, because no one has catalogued more than a fraction of the world’s smaller and family libraries.

A bigger problem is at hand. Arabic manuscripts degrade through handling, humidity, and storage without climate control, and funding cycles end before the work does. The Library has been plain about its own reasoning. Demand for these items has long outstripped what a fragile object can survive, and every visit costs the manuscript something.

The Limits of a Scan

A scan settles less than it appears to. A digital image captures how a page looks and nothing about what it says. Readers can find it only through whatever a cataloguer typed into the record, and for much of the world’s Arabic manuscripts that record holds very little. Machine reading could close the gap, but Arabic script has been the hard case. 

From cursive joins, to the range of individual hands, and uneven fading all defeat attempts to digitize. Progress, while real, is narrower than the coverage suggests. In a 2026 shared task, the best systems reached single-digit character error rates on mid-twentieth-century Palestinian handwriting. 

Raha Rafii, an Islamic studies scholar, warns against treating digital preservation as permanent, since it depends on funding and hardware that nobody can guarantee. She raises a harder question too. The communities whose Arabic manuscripts go online are often the least able to reach them, because destroyed infrastructure and economic collapse sit between them and the file.

Access Still Runs One Way

Public domain status is what makes that inheritance usable. The Omar Ibn Said Collection carries no reuse restrictions. A teacher in Dearborn can put the autobiography on a screen, and a graduate student can quote from it without clearing permissions. Scholars call the practice digital repatriation, the return of access, if not the object, to the people it came from. It is the closest thing the field has to a remedy for a century and a half of Arabic manuscripts leaving the places that produced them. European libraries holding Yemeni collections have adopted it too.

The remedy is partial. A manuscript scanned in Washington reaches a reader in Michigan easily and a reader in Sanaa with far more difficulty.

Want more articles like this? Sign up for our e-newsletter!

Check out our blog here!