Dating the Dead Sea Scrolls to decode their place in biblical history

Dead sea ancient scroll on black background with man silhouette
Dead Sea Scrolls At Reagan Library Simi Valley, CA. Credit: Sarah Reingewirtz / MediaNews Group / Los Angeles Daily News via Getty Images.

The Dead Sea Scrolls were discovered by Bedouin shepherds and archaeologists in the Qumran Caves in the Judaean Desert between 1946 and 1956. Since then, they have changed perceptions about Jewish, Christian and Islamic origins.

But, until now, many individual scrolls – almost 1,000 manuscripts made up of about 100,000 fragments – were not accurately dated.

Previously, individual manuscripts were dated by palaeography alone – the study of ancient handwriting. This method, however, has no empirical foundation – it is up to interpretation.

The Dead Sea Scrolls include some of the oldest surviving fragments of the Hebrew Bible, as well as other religious writings, hymns, prayers and legal documents. Some of the texts found among the scrolls do not form part of the accepted canon of the biblical scripture.

Some of the scrolls bear dates on them in Aramaic and Hebrew. They date, roughly, to the 5th and 4th century BCE, and another lot from the late 1st and early 2nd century CE.

Between these groups is a 400-year gap, presenting a problem.

A new project has sought to solve this problem by using radiocarbon dating and machine learning models to provide dates for individual manuscripts of the Dead Sea Scrolls.

The project, The Hands That Wrote the Bible, is a European Research Council-funded collaboration.

Researchers in the collaboration have attempted to date 24 scroll samples using radio-carbon dating in combination with palaeographic analysis that is based on a machine-learning model.

Flowchart showing dating of ancient scrolls
For this study, writing styles in digitized manuscripts were analyzed using BiNet, a previously developed deep neural network for detection of handwritten ink-trace patterns. By combining writing styles with carbon-14 dates of manuscripts using artificial intelligence, the date-prediction model Enoch is able to produce an accurate date for the manuscript. Credit: Maruf Dhali, University of Groningen.

The model is named Enoch – referencing the Book of Enoch which is one of the Dead Sea Scrolls, a text supposedly written by Noah’s great-grandfather Enoch. The Book of Enoch is considered “pseudepigraphal” – that is, written by someone who is not the purported author – because it is inconsistent with the teachings of the Torah. It is, therefore, not considered part of the Jewish Bible.

Radiocarbon dating of the scroll samples showed they were produced in the period 300–50 BCE.

But applying the machine-learning-based palaeography of Enoch is even more accurate. Cross validation shows that Enoch can improve accuracy to plus or minus 30 years.

Enoch is the first complete machine-learning-based model which can use raw image data to date handwritten manuscripts. Now, the model can be used to support, refine or modify date estimates for specific manuscripts to with higher-than-ever precision.

Two men in front of computer monitors looking at ancient writing
This photo shows Mladen Popovic, Professor of Hebrew Bible and Ancient Judaism and Director of the Qumran Institute (front) and his colleague Dr. Maruf Dhali, Assistant Professor in Artificial Intelligence, working with Enoch to date a manuscript from the Dead Sea Scrolls. Credit: University of Groningen.

The initial results suggest that many Dead Sea Scrolls are older than previously thought. Archaeologists also need to rethink how they have interpreted 2 ancient Jewish script styles called “Hasmonaean” and “Herodian”.

Hasmonaean manuscripts may be older than the current estimate of 150–50 BCE. Herodian-type scripts also emerged earlier than previously thought.

The results don’t just relate to biblical texts. They shift archaeologists’ understanding of political and intellectual development in the eastern Mediterranean around 2,000 years ago.

The Hands That Wrote the Bible research project’s results are in a paper published in the peer-reviewed PLOS One journal.

Sign up to our weekly newsletter

Please login to favourite this article.