Scholars Have Been Vastly Mistaken in Filling the Holes in Our History

The standard way to estimating the missing length of ancient scrolls is wrong by as much as 1800 percent, Israeli scholars find. Why it matters if an ancient text is missing 6 inches or 6 feet

Send in e-mailSend in e-mail
A fragment of the Dead Sea Scrolls.
A fragment of the Dead Sea Scrolls. Credit: Ohad Zwigenberg

Ancient manuscripts thrill for the peep they give us into life and beliefs thousands of years ago. Roughly speaking, at least. Many scriveners of yore were less committed to the naked truth than to the scowling king. Worse, ancient scrolls are almost always deteriorated, leaving frustrating gaps in the story.

But how much of a gap? The standard method for estimating missing scroll lengths is sorely unreliable, warns a study by Eshbal Ratzon of Ariel University’s Department of Land of Israel Studies and Archaeology and Nachum Dershowitz of Tel Aviv University’s School of Computer Science, published recently in PLOS One. (Ratzon began working on the topic while at the University of Haifa.) 

The method is based on a mathematical model used by papyrologists since the 19th century and standardized by the German theologian Hartmut Stegemann (1933-2005), which is theoretically sound. But empirically, the margin of error the model produces is too wide, Ratzon and Dershowitz conclude in the first-ever quantitative study of the methodology for estimating missing scroll length.

To empirically check how well the theory jibes with the reality, Ratzon applied the Stegemann method to fragmented copies of three of the Dead Sea Scrolls that survived relatively intact (so we know their length): Serekh Hayahad (“Community Rule”), the Great Psalms Scroll and Apocryphal Psalms.

“The best results were found for [the Great Psalms Scroll] with approximately 40 percent error and up to 250 percent in the worst-case scenario,” the paper says. “The poorest results were found for 1QS [“Community Rule”] with an average of around 240 percent error and deviation from the actual length of up to 1800 percent!”

Fragments of the Dead Sea Scrolls. How much is missing?Credit: Ohad Zwigenberg

Heartstopping. Say you estimate that 10 percent of a manuscript is gone but your margin of error is 100 percent – that is not helpful. Worse, many scholars simply ignore the problem.

Vexation aside, why does missing length actually matter to scholarship? Take the Torah, Ratzon explains to Haaretz.

“We have no idea when people began to copy the five books together as we know them today. Did they do it back in the Second Temple period? We don’t have complete Torahs from that time, but the material must have existed somewhere before,” she says. So if a 2,000-year-old scroll fragment contains bits from Genesis and Deuteronomy, and a researcher assesses that enough is missing to have contained the whole Torah, that would be a dramatic conclusion. But, using the standard method, it would likely not be a reliable conclusion, Ratzon notes.

For another example, length can cast light on the relative importance of a text. The Bible as we know it hadn’t been canonized when the Qumran scrolls were penned (from the third century B.C.E. to the first century). Presumably, some books were more important than others. Absent a list of important books, researchers hypothesize how sacred a book was held to be based on clues: how many copies were found; other books quoting it; and how much material was used to make the book. “Parchment was expensive. A long scroll must have been very important,” Ratzon says.

Also, pre-canonization, biblical texts differ from one manuscript to another, and compared with accepted testament text. Missing scroll length may indicate that parts were left out or lost. For example, there is a 20 percent difference in length between Jeremiah in the Old Testament and in the Septuagint (the Old Testament’s Greek translation), Ratzon says. The Septuagint version is shorter. “But, if an estimated difference is 20 percent and the margin of error is 100 percent – this will not work,” says Ratzon, driving home the problem.

The methodology was also tapped to estimate the total length of the Scroll of Hôr – a Book of Breathing (ancient Egyptian guide to the afterlife) that had been buried with a priest named Hôr. It’s a long story, but the bottom line is that the scroll was found on Hôr’s mummy and wound up being sold to the self-proclaimed prophet Joseph Smith for $2,400 in 1835, explains a study in the Journal of Mormon Thought. Smith claimed to have a scroll with a record of the patriarch Abraham, but he unfortunately died before finishing its purportedly God-inspired translation.

The point is that some scholars think the Hôr scroll was longer at Smith’s lifetime, in the 19th century, and that part was lost. The question is whether the undamaged scroll of Hôr was ever long enough to accommodate a hieratic Book of Abraham source text. Inquiring minds applied the Stegemann model – and still can’t tell, Ratzon and Dershowitz contend.

If the model is off, the problem is immense because the same basic methodology is widely used to estimate missing scroll lengths.

Hidden scrollsCredit: Alex Levac

How not to estimate a missing scroll length

Proto-writing began over 6,000 years ago; papyrus and parchment were invented in Egypt more than 5,000 years ago and were used through to the early centuries C.E. That is a lot of material. Over 2,000 scrolls were found in a villa in Herculaneum, another town destroyed in the same eruption that obliterated Pompeii in 79. The hundreds of scrolls found at Qumran date from the second century B.C.E. to the first century C.E.

All these are of historic significance but sadly, most are in terrible shape: decayed, carbonized, crushed, gelatinized, moldy, worm-eaten. Clumsy early efforts to unroll scrolls did posterity no favors, either.

In this first quantitative critique of the Stegemann method, Ratzon and Dershowitz identified problematic assumptions used in scroll reconstruction and conclude: highly significant errors are “quite frequent.”

Do they have a new method for estimating missing scroll length? They do not. They just warn that the margins of error produced by the present methodology are too wide to render the estimates trustworthy.

“The method works in theory, but the results are only as good as the data and assumptions that go into it,” says Drew Longacre, a postdoctoral researcher at the University of Groningen in the Netherlands. He isn’t involved in this research but he has worked on the Dead Sea Scrolls and he applauds the admonition the two Israeli researchers provide.

We can’t know if material is missing from a scroll’s outside. The methodology applies to missing material on the inside part. Ratzon and Dershowitz helpfully provide the example of a wallpaper merchant who didn’t note how many meters he sold off a roll. But knowing the circumference of the core and the thickness of the wallpaper, he can reliably calculate how much is left on the roll.

The scrolls of Herculaneum. Unfortunate resemblance to charcoal.Credit: DIGITAL RESTORATION INITIATIVE

Like the resourceful vendor, scholars have been estimating absent scroll length by measuring its circumference at a certain point, then estimating how many rolls could fit inside it based on the thickness of the parchment or papyrus. If there are fewer rolls, the inference is that material is missing. The snag is, unlike wallpaper, papyrus and parchment are uneven in thickness and rolling tightness.

If a large unrolled fragment shows repetitive points of damage, measuring the distance between them (at least three points) could theoretically shed light on how tightly the scroll was rolled i.e., the calculation is based on a model of concentric cylinders. It assumes that the circumference of consecutive rolls shrink linearly, at a constant rate.

In this context, another weak assumption is that repeating damage visible now happened while the scroll was still rolled. That need not be so. In reality, scrolls weren’t rolled with uniform tightness; the distance between repeating patterns of damage is not a reliable indicator for scroll circumference – not least because the scrolls continued to deteriorate before and after being unearthed. So, sadly, methodology based on spiral mathematics is no more accurate, the authors say.

Stegemann himself estimated that his method resulted in a 20 to 30 percent margin of error. Subsequent work estimated the error to be as great as 50 percent. As said, Ratzon and Dershowitz say it can reach 1,800 percent.

“The high amount of error they report is an important caveat for scholars who tend to propose overly precise reconstructions,” Longacre tells Haaretz. “For example, I have seen scholars reconstruct over 5 meters of a scroll and then report results in millimeters. This is clearly unrealistic.”

Unhappy resemblance to charcoal

At Qumran, three scrolls were found more or less complete and several hundred more were found in about 25,000 fragments, which researchers are trying to piece together like a puzzle. But estimates of missing length would be crucial to researchers – if only there was a reliable methodology. In this context, Longacre does quibble with the two researchers’ pessimism, suggesting that while “some circumstances could lead to wildly inaccurate results, others may allow more robust reconstructions.”

In Herculaneum, the scrolls were carbonized by pyroclastic flows then covered by volcanic ash. They look like charcoal and unfortunately a number were mistaken for that very thing and thrown away. Here, however, the scrolls – crushed, moldy, etc. – were not fragmented, just damaged, and were still rolled up. Often fragments detached from outside of the scroll. Herculaneum researchers estimate the gap between the preserved inner part and separated fragments.

HerculaneumCredit: ho visto nina volare

Briefly reviewing twists on the Stegemann method, the researchers conclude they’re no better.

Longacre notes that Ratzon and Dershowitz measured an outer layer with a smaller circumference than the layer inside it, which is physically impossible. Ratzon points out that’s exactly the point: the methodology can lead to absurd results. Ultimately, Longacre feels that in some cases, the Stegemann methodology can produce reliable results: “The devil – as usual – is in the details,” he says.

The authors, however, suggest that if the underlying method is unreliable, as they have shown that it is, all these estimations of missing scroll length will have to be redone. It will have to be done using a methodology that doesn’t exist yet, though.