Codicology · a rule you cannot break by folding

Hair Faces Hair

Open a parchment manuscript almost anywhere and the two pages in front of you are the same side of an animal's skin: hair side facing hair side, flesh side facing flesh side. Scribes kept this rule for well over a thousand years. Fold a skin yourself below and you will find you cannot break it. Stack loose sheets and you will find you almost always do.

Parchment is skin, scraped, stretched and dried, and a skin has two faces: the side the hair grew on, and the side that lay against the animal. They rarely look quite the same. Caspar René Gregory, writing on Greek manuscripts in 1900, put it plainly: prepared skins zeigten deutlich den Unterschied der zwei Seiten, they "clearly showed the difference between the two sides."

A book made of parchment is a stack of quires, and a quire is a few sheets folded down the middle and nested one inside another. Each folded sheet is a bifolium: two leaves, four pages. Wherever you open the book, you see the back of one leaf on the left and the front of the next on the right. Codicologists have a name for the habit of making those two pages the same kind of surface. J. P. Gumbert's standard glossary defines it:

Gregory's rule, according to - / like facing like / a °disposition in which the two pages of an °opening always present matching °parchment surfaces (°hair side faces hair side, °flesh side faces flesh side), in alternating sequence through the °quire.

J. P. Gumbert, Words for Codices, §313.20 (the degree signs are his cross-references)

It is named for Gregory, who described the construction of Greek quires to the Académie des Inscriptions in 1885. It is a real regularity, and old: Marilena Maniaci notes that it is already witnessed by P.Ryl. I 53, Odyssey, third/fourth century CE. The question on this page is a small, strange one. Is it a rule people followed, or something that happens to paper whether you mean it or not?

Fold a skin

Here is one skin. Choose which face lies on the table, then fold it in half, and in half again, as many times as you like. Each fold doubles the leaves. The drawing on the right is the quire you get, drawn the way codicologists draw one: each leaf is a band, its upper stripe the front (recto), its lower stripe the back (verso), with nested arcs at the spine joining each leaf to its partner. At every opening, the mark on the right tells you what faces what.

Every opening matches, every time. Turn the skin over first and the faces swap, but the matching never breaks. Something else never changes either: the face that lay on the table always ends up on the outside of the quire, front and back.

Why it cannot break

A fold lifts half of what is on the table, turns it over, and lays it on top. Turning it over does two things at once: it reverses the order of the layers in that half, and it flips every face. Before any folding there is one layer. Suppose that, going down the stack, the layers alternate, each one showing the opposite face upward to the one above it. Then the two surfaces that touch between any pair of neighbours are the same face of the skin. The lifted half, reversed and flipped, still alternates. And where it lands, the old top layer shows some face upward, while the new layer above it is that same old top layer, turned over, so it presents the same face downward. Alternation survives every fold, so every pair of touching surfaces is like against like. Open the folded stack as a book, and the touching surfaces are exactly the openings.

The bottom layer never moves, so the face that started on the table stays on the outside. That is all Gregory's Rule is, if you fold: a fact about paper, not a decision anybody made.

This page's own model, quire.js, is about forty lines and does exactly the above. As an independent check, a second model written for a different page on this site, Beside Page One, which tracks real coordinates, creases and quarter turns, was asked the same question for every way of folding and turning a sheet up to five folds: 37,448 sequences. Every one obeys the rule, and every one puts the table face outside.

Now stack loose sheets

The other way to make a quire is to cut the skin into separate bifolia and nest them by hand. Now nothing forces anything. Each sheet can go in either way up, and the rule holds only if you turn every other one over. Press any arc to turn that sheet over.

It starts as a pile, every sheet laid the same way up, and the pile obeys at only one opening: the centre, where a single sheet faces itself. Of the 2ⁿ ways to lay n sheets, exactly two obey the rule, one with each face outside. So a quaternion (four sheets, eight leaves: Gregory called it the basis of the Greek manuscript) picked up at random obeys one time in eight, and obeys with flesh outside one time in sixteen.

Which is it, then?

Here is the catch that makes the question hard. A folded skin and a carefully arranged stack leave exactly the same faces, opening for opening. Gregory's Rule, on its own, cannot tell you how a quire was made.

In 1977 the codicologist Léon Gilissen argued, in Prolégomènes à la codicologie, that it could at least be explained: that quires were very often folded from a whole skin, and the rule is what folding leaves behind. His reviewer Emmanuel Poulle summarised the claim as proving that

un cahier n'est pas la réunion accidentelle de bifeuillets préalablement découpés, mais le résultat du pliage, selon des règles précises, d'une peau

E. Poulle, review of Gilissen, Bibliothèque de l'École des chartes 137 (1979), p. 84: "a quire is not the accidental joining of bifolia cut beforehand, but the result of folding, according to precise rules, of a skin"

The evidence was physical, not the rule itself: across about twenty manuscripts, mostly in Brussels, Gilissen matched the edges of leaves and followed stains, holes and repairs that ran from one leaf onto another, showing that leaves now separate had once been one piece. Others pushed back. Rodney Dennis granted that Gilissen had proved his point "to a great extent" but found the system, offered as nearly universal, simply too rational to comprehend over six hundred years of book production. For Greek books Maniaci reports a different pattern entirely, skins divided by a T-cut resulting in three bifolia from each skin. And Malachi Beit-Arié points out that most dated Hebrew quires have five sheets, an odd number that halving can never produce.

So the rule underdetermines the method, and the method has to be read from the parchment. The one thing a single manuscript can say, if you count carefully, is whether its makers kept the rule where folding could not have done it for them.

338 openings of Codex Sinaiticus

Codex Sinaiticus, the fourth-century Greek Bible now divided between four libraries, has been described leaf by leaf by the Codex Sinaiticus Project, and its physical description records, for every side of every leaf it covers, Hair or flesh side: H or F. This page's second program, sinaiticus.mjs, asks the project's own server for each one (824 leaf-sides listed) and counts. Nothing in the count comes from this site. Run on 23 September 2026:

That is Gregory's Rule, with flesh outside, which is what Maniaci describes as the Greek habit: The ancient habit of beginning the quire with a flesh side was maintained in Greek codices until the end of the Middle Ages. The one quire that starts with hair, quire 92, has only two leaves on record.

Now the size. The leaves measure roughly 38 cm by 34 cm in the project's own records (a sample of five leaves ran 372 to 386 mm tall and 323 to 350 mm wide). To fold a quire of eight such leaves from one skin you would need a single piece at least four leaves wide and two tall: about 136 by 76 cm, over a square metre, before any trimming. The skins the sources below describe are far smaller. A "fairly typical" sheepskin in one twelfth-century book was about 78 by 45 cm; a young calf for another, about 70 by 51 cm; and the Hamburg Bible of 1255 used one young calf for each single sheet, about 71 by 51 cm in its first volume. Each of those is about a third of the area a folded Sinaiticus quire would need. So these quires were not folded from one skin, and the rule did not come free. It was kept by hand, sheet by sheet, 43 quires out of 44, when chance would have kept it about one quire in sixteen.

And the exception is the most human thing on this page. In quire 82, a quire of Romans, the record shows the innermost sheet turned over: the opening between leaves 3 and 4 is hair against flesh, the centre is fine, and the opening between leaves 5 and 6 is flesh against hair. Press Sinaiticus, quire 82 in the second instrument and you get exactly those two marks. A folded skin cannot make that mistake. Only a hand can. Whether it was made in the fourth century or at some later rebinding, the record does not say. (The other two unlike openings are in quire 91, a six-leaf gathering whose structure the record does not settle; if its leaves are three nested sheets, the middle one is turned.)

The pile, in history

The arrangement the instrument starts from, every sheet the same way up, is not only a beginner's mistake. It is a tradition. The Arabic codicology section of the Comparative Oriental Manuscript Studies handbook, drawing on François Déroche's work, describes Qur'an manuscripts copied on parchment between the seventh and tenth centuries, whose quires of five sheets read, in the notation codicologists use for the rectos, HHHHH^FFFFF:

When the manuscript is opened, a contrast is evident between the two halves of every double page, except at the junction of two quires (where two hair sides face one another) and in the middle of each quire (where, naturally, two flesh sides appear).

Comparative Oriental Manuscript Studies: An Introduction (2015), ch. 1, Arabic codicology, p. 98, citing Déroche 2005

Press early Qur'an quinion and the instrument gives exactly that: unlike at every opening but the centre, where flesh meets flesh. The same chapter calls it a consistent approach, and Beit-Arié reports the same HHHHH arrangement in a few early Hebrew codices probably from Iraq. Irish and Anglo-Saxon books before the eleventh century went their own way too: H. A. G. Houghton writes that Insular manuscripts arranged the sides allowing the two to face each other with a hair side on the first page, though Jiří Vnouček, in his York thesis on parchment, warns that the Insular pattern is not always strictly followed. So there were at least three settled answers to how the sheets of a quire should face: Gregory's, the pile, and variations between. Only one of them could ever have come for free.

A puzzle in Gregory's own words

One loose end, reported rather than resolved. In his 1900 handbook, Textkritik des Neuen Testamentes, Gregory describes the two sides like this (vol. 1, p. 8):

… die äussere oder Haarseite glätter und heller, die innere oder Fleischseite rauher und dunkler war und häufig Spuren von den Farbflecken der Haarwurzeln aufwies.

"the outer or hair side smoother and lighter, the inner or flesh side rougher and darker, and often showing traces of the colour spots of the hair roots"

and two pages later, that bei einem griechischen Viererheft muss die erste Seite eine Haarseite sein, "in a Greek quaternion the first page must be a hair side." Modern accounts say the opposite of that second claim: Greek quires begin with flesh, and Sinaiticus, above, agrees. His first sentence may explain why. The traces of hair roots are, by definition, on the side the hair grew from, and he places them on the side he calls flesh. Read by his description rather than his labels, his "hair side first" is the modern "flesh side first". We found no source that comments on this, so treat it as our reading of two sentences, not as a finding.

curl -sO https://artwaste.land/strata/hair-faces-hair/quire.js
curl -sO https://artwaste.land/strata/hair-faces-hair/check.mjs
curl -sO https://artwaste.land/checks/research/beside-page-one/fold.mjs
node check.mjs          # 71 passed, 0 failed

curl -sO https://artwaste.land/strata/hair-faces-hair/sinaiticus.mjs
node sinaiticus.mjs     # Node 18+, about 900 requests, a minute or two

Sources

• C. R. Gregory, "Les cahiers des manuscrits grecs", Comptes rendus de l'Académie des Inscriptions et Belles-Lettres 29.3 (1885), 261–268, Persée; and Textkritik des Neuen Testamentes, vol. 1 (Leipzig 1900), pp. 8 and 10, archive.org scan.
• J. P. Gumbert, Words for Codices, ch. 31, §313.20, PDF.
• A. Bausi et al. (eds.), Comparative Oriental Manuscript Studies: An Introduction (Hamburg 2015): M. Maniaci, pp. 79 and 196; Arabic codicology section, p. 98; M. Beit-Arié, pp. 214–215. Codicology chapter, PDF.
• E. Poulle, review of L. Gilissen, Prolégomènes à la codicologie (1977), Bibliothèque de l'École des chartes 137 (1979), 84–88, Persée.
• R. G. Dennis, "Notes concerning Le Pliage", Scriptorium 47 (1993), 166–173, Persée.
• H. A. G. Houghton, The Latin New Testament (Oxford 2016), p. 188, open access.
• J. Vnouček, The Language of Parchment (PhD thesis, York 2019), pp. 61, 165, 207–210, White Rose eTheses (Hamburg Bible, the Canterbury calfskin, the Insular caution).
• L. Fagin Davis, "Reverse-Engineering the Codex", Manuscript Road Trip, 4 Feb 2021, blog (the 780 × 454 mm sheepskin).
• Codex Sinaiticus Project, physical description, methodology and per-leaf records, fetched 23 September 2026.