Dead Sea Scrolls: dating, scribes and provenance

The corpus is held mainly by the Israel Antiquities Authority and the Shrine of the Book at the Israel Museum in Jerusalem; smaller groups are in other museums and in private collections⁠[8]⁠[7]. Hebrew is the most common language, with about 15 per cent of the scrolls in Aramaic and a few in Greek; about 230 manuscripts are biblical⁠[2]. The Leon Levy Dead Sea Scrolls Digital Library publishes high-resolution spectral images of the fragments and scans of the 1950s infrared negatives, but its terms forbid reproduction without the IAA's written permission⁠[8]⁠[9].

Radiocarbon dating has fixed the broad range. Accelerator dates published in 1992 for 14 scrolls and in 1995 for further scrolls and linen fell between the third century BC and the first century AD and broadly agreed with the palaeographic sequence⁠[10]⁠[11]. A proposal in 2004 that the radiocarbon results allowed later, first-century dates for key sectarian texts was answered by a radiocarbon specialist who found that the calibrated dates did not support it⁠[12]⁠[13]. Some fragments were treated with castor oil in the 1950s, and tests in 2001 showed that standard cleaning did not remove all of it, so some earlier dates may be too young by an unknown amount⁠[14].

The 2025 study re-sampled 30 manuscripts from Qumran, Masada, Wadi Murabba'at and Nahal Hever, with chemical screening and cleaning for castor oil and other contaminants; 27 gave valid dates and 24 were used to train Enoch⁠[3]. The model binarises an image of the writing with a neural network (BiNet), extracts features of stroke angle and letter shape, and fits a Bayesian ridge regression to the radiocarbon dates. Its mean absolute error against the radiocarbon ranges was 28 to 31 years. On 13 unseen test images its predictions overlapped the radiocarbon distributions by 85 per cent, and palaeographers judged 107 of 135 predictions for undated manuscripts (79 per cent) realistic⁠[3]. The new radiocarbon range for 4Q114, a copy of Daniel, is 230 to 160 BC; Enoch dates 4Q109, a copy of Ecclesiastes, to the third century BC; both halves of the Great Isaiah Scroll (1QIsaa) come out at about 180 to 100 BC⁠[3].

Scribal identity has been tested with the same image tools. A 2021 study found that the columns of the Great Isaiah Scroll fall into two distinct clusters split at the scroll's midpoint, across several independent feature sets, which supports two scribes with very similar hands rather than one⁠[4].

Provenance has been tested with DNA. A 2020 study sequenced the animal skins of fragments and grouped pieces by species and by individual animal. It found that some Jeremiah fragments were on skins that point to an origin outside Qumran, that patterns in non-biblical scrolls such as the Songs of the Sabbath Sacrifice fit a wider Judaean milieu, and that some debated fragments do not belong with the Qumran scrolls at all⁠[6].

Authenticity has been tested on the market fragments. In 2017 Kipp Davis and colleagues argued on material and palaeographic grounds that nine fragments sold after 2002 were modern⁠[7]. In 2020 an investigation commissioned by the Museum of the Bible concluded that all 16 of its "Dead Sea Scroll" fragments were forgeries on ancient leather⁠[15].

On dating, three positions are in play. The first, from the 2025 authors, is that many manuscripts are older than the traditional palaeographic chronology, sometimes by a century, and that the chronology needs rebuilding on radiocarbon and model dates⁠[3]. The second is that palaeography remains broadly right and that Enoch is a useful aid but too lightly trained to overturn it: 24 training dates span a period of three centuries, and the 79 per cent figure depends on palaeographers' own judgement of what is realistic⁠[16]. Christopher Rollston told Science that such tools should not be the basis for firm conclusions because handwriting is "a deeply human thing"⁠[17]. The third, that key texts are later than the consensus places them, has not survived radiocarbon review⁠[13].

On provenance, the majority view holds that a sectarian community at Qumran collected and partly copied the library, with some scrolls brought in from outside⁠[1]. Norman Golb argued that the scrolls were the libraries of Jerusalem, hidden in the caves during the war with Rome, and that Qumran was not a sectarian settlement⁠[5]. The DNA results fit a mixed library in which some manuscripts were imported⁠[6].

More radiocarbon dates, sampled with the 2025 cleaning protocol across all script styles, would test Enoch properly, especially on manuscripts it dates earliest. A blind test, in which palaeographers and the model date manuscripts whose radiocarbon results are withheld, would show which is more accurate. Open images under a reuse licence would let other groups train competing models. For provenance, DNA and isotope profiles of skins across the whole corpus, matched against local and imported livestock, would show how much of the library was made at Qumran. Every unprovenanced fragment needs material tests before it is used as evidence.