The Herd Inside the Book

The field in brief

Codicology is the archaeology of the book, the study of a manuscript as a made object rather than a text: how the skins were prepared, folded into gatherings, ruled and bound[1]. Until roughly the 2010s that study stopped at the surface, because the material under investigation was also a priceless artifact and the standard conservation rule was blunt, do not sample[1].

Parchment is animal skin, scraped and limed and stretched dry under tension, and skin carries the biology of the creature it came from[1]. Tim Stinson, a medievalist at North Carolina State University, argued from 2009 that every parchment leaf is therefore a faunal archive, a record of the sheep, goats and calves of a medieval economy waiting to be read[6].

Two kinds of molecule survive in old skin. Collagen, the structural protein of hide, is tough and abundant, and its species-specific fragments can be identified by a method borrowed from archaeology called ZooMS, zooarchaeology by mass spectrometry, which reads a peptide ‘fingerprint’ to name the animal[2][1]. DNA carries far more information and is far more fragile, breaking into short fragments and mixing with the DNA of everyone who ever handled the page[1].

The breakthrough that turned a forbidden subject into a discipline was a way to sample almost nothing. In 2015 Sarah Fiddyment and Matthew Collins showed that gently rubbing a page with a PVC eraser lifts enough collagen for a ZooMS identification, a trick that also settled an old codicological legend[2]. The gossamer-thin parchment of thirteenth-century pocket Bibles had long been credited to aborted calf fetuses or rabbit skin, so-called uterine vellum; the eraser crumbs showed ordinary calf, sheep and goat, made thin by craft rather than by slaughtering the unborn[2].

The fight

The first fight was cultural, and Collins described it, borrowing C. P. Snow’s phrase, as a ‘two cultures problem’ between laboratory scientists who wanted material and humanities curators who existed to deny it[6]. When Fiddyment first proposed shaving a sample at the Borthwick Institute in York, the conservators refused outright[6]. The eraser method was the peace treaty, because a conservator could do the rubbing herself, in place, and approve every touch[1].

That treaty made ZooMS the field’s workhorse. It is cheap and fast, succeeds on better than nine samples in ten, and needs only a pinch of eraser dust[1]. Fiddyment and Collins, working across Cambridge, Copenhagen and York, built much of the discipline on it, reading everything from the animal of a book to a fifteenth-century birth girdle’s traces of the women who once wore it against the pains of labour[6]. Their caution is consistent: DNA from parchment is, in their own words, of variable quality and abundance, and one should always begin with the least invasive and least ambiguous method[1].

Stinson wanted more than a species name. DNA can separate sex, breed and individual animals, and can match two leaves cut from the same hide, which offers a way to reunite manuscripts long ago broken up and scattered across the world’s libraries[4][6]. The obstacle was that reading DNA meant taking real material, the very thing conservators had forbidden. In the spring of 2026 Stinson’s group, working with the geneticist Matthew Breen, published its own non-destructive answer in the journal Manuscript Studies, a soft cytology brush borrowed from cervical screening that lifts DNA from a page without leaving a mark[3][8].

Across 91 manuscripts at Duke, spanning the eighth century to the twentieth, the brush recovered enough DNA to map regional habits, English books mostly sheepskin, Ethiopian ones mostly goat[7]. It also recovered the discipline’s central anxiety. One manuscript returned pig DNA that almost certainly came not from its pages but from the pig-derived glue of a later repair[7]. In the wider survey behind the recent coverage, DNA could be pulled from only about 58 percent of 351 samples, and some of the most tantalizing hits, a Greek New Testament that nearly matched red deer, fell just short of a confident call[6].

The stakes are ordinary academic ones, which is to say real. Stinson has lost a United States federal grant, while European research councils have committed more than twenty million euros to biomolecular heritage work, so the question of which method the field standardizes on is also a question of whose program survives[6]. For now the two approaches are converging on the same non-destructive ethic while disagreeing on what to chase, robust but limited protein, or rich but unreliable DNA[1][3].

What the fight reveals

On the narrow question the field began with, which animal, the evidence currently favours the protein camp. ZooMS is more reliable, cheaper and higher-yielding than DNA, and it settled real codicological disputes, the uterine-vellum myth among them, without ever putting the object at risk[2][1]. DNA has not matched that reliability, because recovering it from a page handled by a thousand hands across a thousand years means every result must be defended against contamination, as the pig-glue and near-miss red deer cases show[7][6].

Yet the protein camp’s reliability comes partly from asking less. Only DNA can tell a calf from its sibling, or prove that a leaf in Durham and a leaf in North Carolina were once the same animal, and those are exactly the questions that could rebuild dismembered medieval books[4][3]. Whether that richer signal can be made trustworthy enough to bear such conclusions is genuinely unresolved, and it is the fault line the field will spend the next decade arguing over[6].

The deeper thing the quarrel exposes is how tempting it is to over-read a molecule. When a twelfth-century manuscript switches to goatskin just after its text mentions a goat, even the DNA method’s own champion allows that it could simply be coincidence[6]. A field built to escape the softness of connoisseurship now has to guard against a harder-looking softness, the story told too confidently from a single peak on a mass spectrum[1]. The molecules are real; the history read out of them still has to be argued, one contaminated, degraded, gloriously informative sample at a time[3].