Machine learning helps researchers decipher the Dead Sea Scrolls

Machine learning helps researchers decipher the Dead Sea Scrolls
Researchers used machine learning to identify textural patterns within the text of a Dead Sea Scroll. Photo by Maruf A. Dhali/University of Groningen

April 21 (UPI) -- It's been seventy years since the Dead Sea Scrolls were unearthed, but most of the questions about the scrolls' authorship remain unanswered.

That's because the ancient scrolls, featuring some of the oldest manuscripts from the Old Testament, are unsigned.


Scientists have tried to find handwriting clues that can be used to identify the scribes responsible for the different scrolls but success has been limited, and uncertainty and disagreement persists.

In a new study, published Wednesday in the journal PLOS One, researchers used machine learning to study the Great Isaiah Scroll, which some researchers hypothesized was penned by two scribes with similar handwriting styles.

RELATED New set of Dead Sea Scrolls recovered, first Biblical artifacts found in 60 years

"This scroll contains the letter aleph, or 'a,' at least five thousand times. It is impossible to compare them all just by eye," study co-author Lambert Schomaker said in a news release.

"Computers are well suited to analyze large datasets, like 5,000 handwritten a's," said Schomaker, a professor of computer science and artificial intelligence at the University of Groningen in the Netherlands.


Digital imaging and advanced computer algorithms can identify unique textural features, like the way parts of letters curve, as well as allographic characteristics, or the shapes of whole letters.

RELATED Museum of the Bible says 5 Dead Sea Scrolls are forgeries

Almost unconsciously, the human eye and brain can recognize differences in handwriting, but the cognitive mechanics behind this process aren't well understood.

"Furthermore, it is virtually impossible for these experts to process the large amounts of data the scrolls provide," said lead author Mladen Popović, a professor of the Hebrew Bible and ancient Judaism at Groningen.

Not surprisingly, handwriting experts often arrive at different interpretations of the same source material.

RELATED Researchers decipher ancient Dead Sea Scrolls with help of advanced imaging

To build their new machine learning algorithm, researchers had to start by training it to separate the ink from the papyrus background -- and not just the basic shapes, but the entirety of the ink lettering.

"This is important because the ancient ink traces relate directly to a person's muscle movement and are person-specific," Schomaker said.

When scientists used the algorithm to search for textural and allographic patterns within the text of the Great Isaiah Scroll, they found commonalities were clustered into two distinct columns, not distributed randomly throughout the text.

The discovery suggested there was, in fact, more than one author.


Scientists separated the two columns and reanalyzed the text, which confirmed the textural distinctions. Followup tests provided additional certainty.

"When we added extra noise to the data, the result didn't change," Schomaker said. "We also succeeded in demonstrating that the second scribe shows more variation within his writing than the first, although their writing is very similar."

Additionally, researchers reprinted the text using a computerized average of the different letters, so to minimize some of the textural noise. This allowed researchers to more easily conduct a visual analysis of the scroll's different columns.

"Now, we can confirm this with a quantitative analysis of the handwriting as well as with robust statistical analyses," Popović said. "Instead of basing judgment on more-or-less impressionistic evidence, with the intelligent assistance of the computer, we can demonstrate that the separation is statistically significant."

Moving forward, researchers plan to use their algorithm to analyze all of the Dead Sea Scrolls.

"We are now able to identify different scribes," Popović said. "We will never know their names. But after seventy years of study, this feels as if we can finally shake hands with them through their handwriting."

Latest Headlines


Follow Us