Animal origin of 13th-century uterine vellum revealed using noninvasive peptide fingerprinting

Significance

This study reports the first use, to our knowledge, of triboelectric extraction of protein from parchment. The method is noninvasive and requires no specialist equipment or storage. Samples can be collected without the need to transport the artifacts; instead, researchers can sample when and where possible and analyze when required. The level of access we have achieved highlights the importance of this technique. For this study, we have extracted proteins from 513 parchment samples, used to resolve the long-standing question of the origin of “uterine vellum.” We find no evidence of unexpected species, such as rabbit or squirrel. We suggest that uterine vellum was often an achievement of technological production using available resources, and would not have demanded unsustainable agricultural practices.

Abstract

Tissue-thin parchment made it possible to produce the first pocket Bibles: Thousands were made in the 13th century. The source of this parchment, often called “uterine vellum,” has been a long-standing controversy in codicology. Use of the Latin term abortivum in many sources has led some scholars to suggest that the skin of fetal calves or sheep was used. Others have argued that it would not be possible to sustain herds if so many pocket Bibles were produced from fetal skins, arguing instead for unexpected alternatives, such as rabbit. Here, we report a simple and objective technique using standard conservation treatments to identify the animal origin of parchment. The noninvasive method is a variant on zooarchaeology by mass spectrometry (ZooMS) peptide mass fingerprinting but extracts protein from the parchment surface by using an electrostatic charge generated by gentle rubbing of a PVC eraser on the membrane surface. Using this method, we analyzed 72 pocket Bibles originating in France, England, and Italy and 293 additional parchment samples that bracket this period. We found no evidence for the use of unexpected animals; however, we did identify the use of more than one mammal species in a single manuscript, consistent with the local availability of hides. These results suggest that ultrafine vellum does not necessarily derive from the use of abortive or newborn animals with ultrathin hides, but could equally well reflect a production process that allowed the skins of maturing animals of several species to be rendered into vellum of equal quality and fineness.

Hamlet: Is not parchment made of sheepskins?

Horatio: Ay, my lord, and of calfskins too.

Hamlet, V.i

One of the outstanding controversies in the field of codicology concerns the origin and production of so-called “uterine vellum.” Researchers in the field of manuscript studies have long disputed the origin of this ultrafine writing material. Some older scholarship suggested that uterine vellum probably derived from the hides of smaller, more thin-skinned mammals, such as rabbits or squirrels (1, 2). However, most paleographers continue to view the notion of medieval uterine vellum as a myth, on the grounds that its production on the scale implied by extant manuscripts would have entailed an untenably high number of aborted fetuses (3). Other proposed solutions to the derivation of this material have involved specific production processes, such as the splitting of skins (4). The importance of analyzing parchment carefully to determine its origin species and, more specifically, the origin of uterine vellum is a priority research question for several disciplines. Such analysis might not only settle the long-standing debate over the source of supposedly “aborted” hides but would also provide valuable data about the localization and distribution of hides, and thus of membrane books and documents (5). The scholarly disputes notwithstanding, uterine vellum must represent either (i) the selection of specific animals whose skin was uniquely fine or (ii) the development of craft skills to work a wide range of skins into ultrafine sheets, or, of course, both at once.

The most frequently cited examples of uterine vellum are the 13th-century Paris Bibles (1, 2, 4, 6, 7). These books were single-volume Bibles (pandects) with a consistent organization, which meant that they were easily searchable, making them the ideal reference guide for study and sermon preparation (8). One of the most important subgroups of 13th-century Paris Bibles is the pocket or portable Bible, volumes sufficiently small to be easily transported. Ruzzier (6) suggests that the total output of portable Bibles in the 13th century could have exceeded 20,000 copies. The majority of surviving copies (∼54%) were made in France, most in Paris (6), although Bibles of this style were also produced in England at the same time and slightly later in both Italy (notably in the Veneto) and Spain.

To explore the question of uterine vellum, we examined the animal composition of ultrafine parchment (which ranged in thickness from 0.03–0.28 mm in our selected samples), sampled primarily from pocket Bibles, and compared these samples with other parchments that book-end their main period of production. If the skins of small animals (rabbits and squirrels) were used, their presence would be revealed by zooarchaeology by mass spectrometry (ZooMS). If the production of uterine vellum represented instead a specialized craft skill, then the selection of animals would be similar to the selection of animals evident in other coarser membranes from the same geographic region.

Species Identification of Parchment

To assess the origin of uterine vellum, the principal line of evidence is the identification of the skin. Skins have previously been identified by overall size, thickness, color, levels of grease, and patterns of follicles (9). A notable proponent of follicle pattern analysis was Ryder (10), who used the technique to identify animal species; however, not all parchments had discernible follicle patterns, not every pattern could be identified, and paleographers have often been overconfident in their ability to discern species origin from such patterns. Protein (11, 12) and DNA-based methods (13–17) potentially offer absolute determination of species but, until now, have had other limitations. Toniolo et al. (11) analyzed 5 mg of parchment from the 13th-century “Marco Polo Bible” at the Biblioteca Medicea Laurenziana, and used peptide sequences to identify a single leaf as calfskin, although if more leaves had been analyzed the results may have revealed different species as was the case with the family's archival documents (Table S1). Kirby et al. (12) also used destructive sampling techniques to analyze a Koran and other objects. Teasdale et al. (13) have recently reported genomic data from postmedieval parchment using, on average, 50 mg of parchment. In addition to the destructive nature of sampling, a further hidden cost of molecular analysis is adequate storage of extracted samples. It is usually necessary to freeze DNA and protein extracts to preserve sample integrity, which adds a further cost burden for libraries and archives in either shipping or storage.

Table S1.

Sample details for three of Marco Polo’s family archival documents

Sample no.	Library	Date	Type of document	Location of production	Species
VVK 26	Biblioteca Marciana	August 5, 1280	Testament of Marco Polo “il vecchio” (Marco Polo’s uncle)	Italy	Sheep
VVK 29	Biblioteca Marciana	August 31, 1300	Testament of Matteo Polo (Marco Polo’s step-brother)	Italy	Sheep
VVK 31	Biblioteca Marciana	January 9, 1324	Testament of Marco Polo “il viaggiatore”	Italy	Sheep

Triboelectric Extraction

Here, we describe a novel noninvasive molecular identification method to identify the origin of parchment, using electrostatic molecular extraction onto a solid-phase PVC polymer. PVC polymer erasers are widely accepted by the conservation community as a noninvasive measure for removing dirt. Indeed, every archive and conservation studio has access to and experience of the use of PVC erasers. A further advantage of the extraction method is that protein is preserved on the PVC polymer waste at room temperature with no further storage requirements (apparently indefinitely). It can therefore be retained at the point of collection without special preparation or storage until the researcher deems it appropriate to analyze a larger set of samples. The method is easily scalable and places the sampling in the hands of curators, codicologists, and conservators. Using this approach, we have sent kits (consisting of PVC erasers, nitrile gloves, and sampling tubes) to 14 archives and 40 libraries in Europe and North America. They have returned samples from 79 13th-century pandect Bibles for analysis, including 72 pocket Bibles.

The aim of this study was to identify the animal origin of the tissue paper-thin “uterine” vellum, used in 13th-century pocket Bibles, through the use of a novel noninvasive triboelectric extraction of skin collagen, subsequently analyzed by conventional peptide mass fingerprinting.

Results

Validation of Triboelectric Extraction.

To assess the quality of our results, we performed a comparative analysis of the same sample using conventional ZooMS techniques that require destructive sampling (12), as well as nondestructive electrostatic ZooMS (eZooMS). For the purpose of this experiment, we used two different documents: (i) a 16th-century manorial court roll (Fig. 1A) and (ii) an 18th-century seal tag (Fig. 1B). In case i, a fragment of parchment of ∼0.5 cm × 0.5 cm was used, and in case ii, a fragment of ∼0.1 cm × 0.3 cm was used. In both cases, an eZooMS sample was taken from the main document.

Fig. 1.

Comparison of peptide mass fingerprint from samples A and B using destructive sampling (lower spectra) and noninvasive eZooMS sampling (top spectra). (A) For one sample, the resulting peptide mass fingerprint detected fewer peaks than in its equivalent eZooMS sample. (B) In the other sample, the results obtained were very similar for both methods. eZooMS samples extracted and analyzed 12 mo apart, having been stored at room temperature, revealed no loss of signal.

The optimization of our eZooMS methodology has allowed us to obtain equal, if not better, results from PVC sampling than when using an actual fragment of parchment (Fig. 1). The observed difference in quality is probably due to an excessive amount of collagen being extracted during destructive analyses (Fig. 2). However, the eraser technique itself may also contribute to a cleaner signal by the eraser retaining contaminating molecules that interfere with the subsequent analysis on the PVC.

Fig. 2.

Comparison of peptide mass fingerprint from serial dilution of destructive sample and noninvasive eZooMS sampling. A destructive sample from historic calf parchment (0.5 cm × 0.5 cm) was extracted and digested using a serial dilution ranging from 1:2–1:1,280 and compared with a noninvasive eZooMS extraction. A cleaner signal is seen between the 1:16 and 1:128 dilutions, likely due to the dilution of excessive collagen molecules or contaminants that interfere with the subsequent analysis.

Species Variation.

Species identifications of all of the samples analyzed are included in Tables S2 and S3. A total of 220 folios were analyzed from the 72 pocket Bibles. Of these folios, 68% were calf, 26% were goat, and 6% were sheep. We found no evidence for the rabbit skin duodecimos suggested by Pollard (1). Of the 72 Bibles we sampled, 62 were sampled on multiple leaves. In most cases, the identification was consistent for each of the folios of one manuscript, including the 18 sampled leaves of the Hornby Bible (OSU.MS.MR.Frag.74). However, at least five Bibles were composed of parchment from multiple species. To confirm the presence of multiple animals, we were able to analyze 20 folios from Cambridge University Library Ee.6.26, six of which were identified as calf and 14 as sheep. The species distinction is mirrored by stylistic differences within the Bible, which suggest that this manuscript may, in fact, be a composite of two different Bibles. The first part of the Bible (fol. 1–108), which contained five folios identified as calf parchment, resembles a “proto-‘Paris’ Bible” model produced ca. 1200–1230, whereas the second part (fol. 109–459), containing the 14 folios made of sheepskin (and one of calfskin) parchment, is a far better fit for the “mature ‘Paris’ Bible” blueprint of ca. 1230–1280 as described by Light (8, 18–20). It is also worth noting that the attributed English provenance derives from inscriptions present in the second part of the Bible; no provenance indicators are recorded for the first part, so it is possible that the Bible’s first 108 folios were produced in France.

Table S2.

Sample details for 72 13th century pocket Bibles

Sample no.	Shelf-mark	Library	Date	Size, mm	No. of leaves	Thickness, mm	No. of pages sampled	Location of production	Species
CCC 31–32	CCCC MS 246	Corpus Christi College Cambridge	1200–1300	156 × 102	410	0.05–0.08	2	England	Calf
CL 001	Gg.6.15	Cambridge University Library	1225–1250	184 × 130	504	0.06–0.09	1	France	Calf
CL 002	Gg.6.45	Cambridge University Library	1225–1250	136 × 92	532	0.07–0.1	1	France	Calf
CL 003–005	Add.1848	Cambridge University Library	1225–1250	156 × 109	407	0.07–0.2	3	France	Goat
CL 006	Add.6159	Cambridge University Library	1225–1250	137 × 88	594	0.04–0.06	1	France	Calf
CL 057–059	Pembroke 268	Cambridge University Library	1200–1300	198 × 128	480	0.06–0.2	3	France	Calf
CL 064–067	Add.8438	Cambridge University Library	1200–1300	185 × 130	424	0.07–0.13	4	France	Calf
CL 082–086	Hh.1.3	Cambridge University Library	1250–1275	198 × 195	367	0.06–0.15	5	England	Calf
CL 087–091	Ii.6.22	Cambridge University Library	1250–1300	142 × 102	299	0.06–0.12	5	Italy	Goat
CL 092–094	Ii.6.12	Cambridge University Library	1200–1300	165 × 104	469	0.06–0.14	3	France	Calf
CL 100–101	Dd.5.52	Cambridge University Library	1200–1300	210 × 140	333	0.1–0.18	2	England	Calf and goat
CL 102–104	Gg.1.27	Cambridge University Library	1225–1250	224 × 153	406	0.07–0.22	3	England	Calf and goat
CL 105–106	Dd.10.29	Cambridge University Library	1250–1275	235 × 160	439	0.08–0.15	2	England	Calf
CL 107–108	Ee.1.16	Cambridge University Library	1200–1300	208 × 150	349	0.08–0.13	2	Northern France or England	Calf
CL 109–110	Ii.4.10	Cambridge University Library	1225–1250	242 × 170	346	0.1–0.2	2	England	Calf
CL 111–112	Ee.1.9	Cambridge University Library	1200–1300	210 × 150	285	0.08–0.18	4	England	Calf
CL 118–119
CL 113–114	Mm.1.2	Cambridge University Library	1200–1300	245 × 163	610	0.06–0.1	2	Northern France or England	Goat
CL 120–121	Gg.1.3	Cambridge University Library	1225–1275	211 × 140	375	0.08–0.11	2	England	Calf
CL 122–123	Ff.6.19	Cambridge University Library	1200–1300	146 × 97	567	0.04–0.09	2	—	Calf
CL 124–125	Ff.6.20	Cambridge University Library	1200–1300	146 × 98	453	0.03–0.06	2	—	Goat
CL 126–127	Dd.15.35	Cambridge University Library	1200–1300	148 × 94	476	0.06–0.1	2	—	Goat
CL 128–129	Dd.12.47	Cambridge University Library	1200–1300	150 × 106	502	0.04–0.08	2	Northern France or England	Calf
CL 130–131CL 179–196	Ee.6.26	Cambridge University Library	1220–1230	198 × 138	459	0.07–0.28	20	England	Calf and sheep
CL 132–136	Ff.6.45	Cambridge University Library	1225–1275	144 × 103	478	0.05–0.09	5	England	Goat
DCL 05–06	A.IV.37	Durham Cathedral Library	1200–1300	235 × 147	376	—	2	England?	Calf
DCL 07–10	A.IV.30	Durham Cathedral Library	1225–1275	159 × 106	576	—	4	France?	Calf
DUL 01–03	DUL MS Cosin V.V.17	Durham University Library	1200–1300	127 × 92	322	0.08–0.08	3	Southern France or Northern Italy	Goat
DUL 04–06	DUL MS Cosin V.V.18	Durham University Library	1200–1300	132 × 87	420	0.05–0.05	3	France	Calf
FLP 01–03	Lewis E 029	Free Library of Philadelphia	1230–1240	182 × 113	160	—	3	England	Calf
FLP 04–06	Lewis E 028	Free Library of Philadelphia	1270–1270	159 × 105	659	—	3	France	Calf
FLP 07–09	Lewis E 030	Free Library of Philadelphia	1225–1250	142 × 98	539	—	3	England	Goat
FLP 10–12	Lewis E 033	Free Library of Philadelphia	1250–1250	150 × 104	341	—	3	France	Calf
FLP 13–15	Lewis E 036	Free Library of Philadelphia	1260–1280	166 × 123	370	—	3	Italy	Goat
FLP 16–18	Lewis E 039	Free Library of Philadelphia	1250–1250	186 × 126	403	—	3	Italy	Goat
JRL 158–159	MSM0163	Morgan Library	1229–1229	220 × 150	454	0.09–0.15	2	France	Goat
KRPB	Ms. 56	University of St. Louis	1200–1300	150 × 100	236	0.11–0.18	1	France	Calf
MAGGS 02	14739	Maggs Bros. Ltd.	1200–1300	—	—	—	1	—	Goat
MAGGS 03	14754	Maggs Bros. Ltd.	1200–1300	—	—	—	1	—	Goat
MAGGS 05	12307	Maggs Bros. Ltd.	1200–1300	—	—	—	1	—	Goat
NRO W 14	PARIS BIBLE	Private collection	1225–1275	158 × 112	395	0.12–0.12	1	France	Calf
OSU 01–18	OSU.MS.MR.Frag.74	Ohio State University	1220–1220	220 × 145	440	0.1–0.2	18	France	Calf
RLB 01–03	Ms 2053	Royal Library of Belgium	1200–1300	223 × 157	443	—	3	France (north)	Goat
RLB 04–06	Ms 4911	Royal Library of Belgium	1200–1300	130 × 93	180	0.093–0.149	3	France (north), Low Countries, or England	Calf
RLB 07–09	Ms 5627	Royal Library of Belgium	1200–1300	140 × 94	588	0.061–0.078	3	France (north)	Calf
RLB 10–12	Ms 8428	Royal Library of Belgium	1200–1300	208 × 159	412	0.082–0.174	3	England	Calf
RLB 13–15	Ms 8882	Royal Library of Belgium	1200–1300	140 × 96	486	0.067–0.088	3	France (north)	Calf
RLB 16–18	Ms 10517	Royal Library of Belgium	1200–1300	150 × 95	664	—	3	France (north)	Calf
RLB 19–21	Ms 10518	Royal Library of Belgium	1200–1300	140 × 90	642	—	3	France (north)	Calf
RLB 22–24	Ms 10520	Royal Library of Belgium	1200–1300	140 × 85	790	—	3	France (north)	Calf
RLB 25–27	Ms 10521	Royal Library of Belgium	1200–1300	154 × 97	592	—	3	France (north)	Calf
RLB 28–30	Ms 10522	Royal Library of Belgium	1200–1300	178 × 122	481	0.074–0.126	3	France or Italy	Calf and goat
RLB 31–33	Ms 10523	Royal Library of Belgium	1200–1300	178 × 116	359	0.101–0.172	3	France (north)	Calf
RLB 34–36	Ms 10524	Royal Library of Belgium	1200–1300	169 × 117	418	0.087–0.147	3	France (north), Low Countries, or England	Calf
RLB 37–39	Ms 10545	Royal Library of Belgium	1200–1300	221 × 149	462	0.031–0.105	3	France (north)	Calf
RLB 40–42	Ms 10753	Royal Library of Belgium	1200–1300	211 × 147	550	0.06–0.083	3	France (north)	Calf
RLB 43–45	Ms 10860	Royal Library of Belgium	1200–1300	180 × 124	604	0.056–0.104	3	France (north) or Low Countries or England	Calf and goat
RLB 46–48	Ms 14680	Royal Library of Belgium	1200–1300	143 × 95	545	0.06–0.083	3	France (north)	Calf
RLB 49–51	Ms 21700	Royal Library of Belgium	1200–1300	125 × 91	583	0.05–0.084	3	France (north)	Calf
RLB 52–54	Ms 2663–64	Royal Library of Belgium	1200–1300	193 × 133	456	0.087–0.124	3	France (north)	Calf
RLB 55–57	Ms IV 290	Royal Library of Belgium	1200–1300	244 × 175	478	0.135–0.211	3	Spain?	Goat
UPenn 02	Ms. Codex 236	Schoenberg Institute, University of Pennsylvania	1235–1240	218 × 148	465	—	1	France	Calf
UPenn 03	Ms. Codex 1065	Schoenberg Institute, University of Pennsylvania	1240–1250	178 × 117	356	—	4	England?	Goat
DNA UPenn 04–06
UMLB 07–08	Bible Vulgata RARES 091 B47C	University of Melbourne	1250–1300	130 × 90	595	—	2	France or England	Calf
UVA 15	M.MS.N	Albert and Shirley Small Special Collections Library, University of Virginia	1200–1300	158 × 95	603	0.07–0.07	1	France	Calf
VV 17–18	Holkham Hall MS 10	Holkham Hall	1240–1260	226 × 152	—	—	2	France	Calf
VV 19–20	Holkham Hall MS 11	Holkham Hall	1200–1300	178 × 134	—	—	2	France	Calf
VV 21–22	Holkham Hall MS 12	Holkham Hall	1200–1300	185 × 123	—	—	2	France	Goat
VV 23–25	Holkham Hall MS 13	Holkham Hall	1230–1250	225 × 144	—	—	3	France	Goat
VV 26–27	Holkham Hall MS 1	Holkham Hall	1300–1400	163 × 108	—	—	2	England	Calf
YML 04–06	XVI.N.6	York Minster Library	1225–1275	203 × 143	333	0.1–0.14	3	England	Calf
YML 13–15	XVI.N.5	York Minster Library	1200–1250	200 × 140	420	0.07–0.11	3	France	Calf
YML 16–18	XVI.N.10	York Minster Library	1225–1275	151 × 99	665	0.06–0.1	3	France?	Calf

Table S3.

Sample details for seven 13th century nonpocket pandect Bibles

Sample no.	Shelf-mark	Library	Date	Size, mm	No. of leaves	Thickness, mm	No. of pages sampled	Location of production	Species
CL 07–08	Add.7801	Cambridge University Library	1225–1275	243 × 164	515	0.08–0.12	2	France	Calf
DCL 01–02	C.III.22	Durham Cathedral Library	1200–1300	342 × 233	438	—	2	England	Sheep
DCL 03–04	A.II.3	Durham Cathedral Library	1200–1300	415 × 265	415	—	2	Northern France	Calf
RLB 58–60	Ms 9111–9114	Royal Library of Belgium	1286–1286	421 × 314	972	0.125–0.28	3	Germany	Calf
RLB 61–63	Ms II 2523	Royal Library of Belgium	1200–1300	495 × 358	588	0.186–0.261	3	Low Countries	Calf
U Penn 04	Ms. Codex 724	Schoenberg Institute, University of Pennsylvania	1275–1299	375 × 246	330	—	1	France	Calf
YML 01–03	XVI.D.13	York Minster Library	1200–1250	277 × 195	326	0.09–0.22	3	England	Sheep

In addition, we explored the possibility of using liquid chromatography-tandem MS to determine if differential biomarkers for uterine skin could be identified (SI Materials and Methods). Modern uterine samples showed elevated expression of four proteins (Fig. S1); however, none of the peptide markers observed in eZooMS mapped to any of these proteins, so it is not possible to report the presence of these proteins in our Bible samples.

Fig. S1.

iTRAQ experiment. (A) Comparison of protein expression of uterine and nonuterine parchment samples. Uterine samples show higher expression levels for four proteins, including dolichyl-diphosphooligosaccharide-protein glycosyltransferase subunit 2 (RPN2), hemoglobin fetal subunit beta (HBBF), heat shock protein 90-alpha (HSP90A), and collagen alpha-1(III) chain (COL3A1). An error rate of 2 SDs is assumed. (B) Uterine calfskin indicating the four sampling points. (C) Historical legal document dated to 1776 used as one of the nonuterine samples.

Geographic Distribution.

Previous authors have identified geographic variation in the animals used for making parchment. Forbes (21), citing Wattenbach, notes that hides used for parchment were predominantly calf in the north of Europe and sheep and goat in the south. Ruzzier (6) agrees, and suggests that fine parchment north of the Alps was likely made from calfskins. Clarkson (4), noting the “warm, creamy tint” and “flexibility” of late-medieval Italian skins, attributes these qualities to the use of goatskin and possibly alternative methods of preparation. De Hamel (3) observes that Italian parchment is often prepared from goatskin, which contrasts with the poor representation of goatskin in England (22–24).

Our results are in agreement with most of these assessments. We see a predominance of calfskin being used in France, a pronounced use of goatskin in Italy, and a more mixed pattern emerging from England (Fig. 3).

Fig. 3.

Relative proportions of animals used to make parchment in each of the three regions studied (France, England, and Italy) during the 12th to 14th centuries. The size of the circle indicates the number of samples. Except for the figure describing exclusively pocket Bibles, data were obtained from all sources of parchment, including legal documents, secular codices, and Bibles, using the eZooMS method. Circles shown in paler colors indicate inconclusive provenance with respect to location.

To examine whether the geographic variation observed for ultrafine parchment differed from coarser membranes, we surveyed an additional 293 parchment objects from the 12th to 14th centuries. The selection of skins appears to reflect available livestock, and therefore a city’s or a region’s preferences: Sheepskin is most abundant in England, calfskin in France, and goatskin in Italy. In Italy, there is an absence of goatskin in the 12th century (contra 25), which possibly reflects regional variation. Six (of 12) 12th-century Italian sheep samples were sourced from Sicily, where a sheep-based dairy economy was practiced. Although there are differences between the two centuries spanning the period of interest, a more comprehensive investigation is required to study any long-term changes.

The pattern of selection of ultrafine parchment for Bible production seems to be broadly in line with the pattern of selection found in other parchment analyzed (Fig. 3). The English Bibles seem to be more varied in their composition, including multiple species in the same volume. Of the five Bibles that were composed of a mixture of animals (calf and goat or calf and sheep), three of them are of English provenance, one of possibly English provenance, and the fifth of either French or Italian provenance. Of the seven larger Paris Bibles (i.e., classified as nonpocket pandect Bibles), the three described as being of French production, as well as the two Bibles from the Low Countries and Germany, were all identified as calf; however, the two larger Paris Bibles of English origin were found to be made from sheep. All these nonpocket Bibles also seem to follow general regional preferences.

The scant use of sheepskin parchment for pocket Bibles is provocative and merits further investigation. A survey of English archival documents from the Borthwick Archive (York), using the same methods described in this paper (Table S4), reveal that all are written on sheepskin parchment. This selection of sheepskin parchment for legal documents may not merely reflect cost. The Dialogus de Scaccario (26) argues for the use of sheepskins as the medium for English legal documents due to the difficulty of erasure of text as a result of the propensity of sheepskin parchment to delaminate. This characteristic also means that sheepskin is the easiest skin to split, apparently making it ideally suitable for the thin parchment used in pocket Bibles. However, apart from the two larger Paris Bibles of English origin, we have found only one example of a pocket Bible made of sheepskin: This pocket Bible is of supposed English provenance (Cambridge University Library Ee.6.26).

Table S4.

Sample details of English archival documents held at the Borthwick Archive, York

Sample no.	Shelf-mark	Date	Country	Type of document	Species
BA 61	YM/D/CAMP 1	1322–1322	England	Legal deed	Sheep
BA 26	MOR 2	1326–1326	England	Legal deed	Sheep
BA 81	YM/D/PO	1330–1330	England	Legal deed	Sheep
BA 41	YM/D/ASK 1	1332–1332	England	Legal deed	Sheep
BA 141	YM/D/HAT 1	1376–1376	England	Legal deed	Sheep
BA 06	YM/D/BAL a. 1	1377–1377	England	Legal deed	Sheep
BA 27	MOR 3	1380–1380	England	Legal deed	Sheep
BA 01	YM/D/BAL(M) 1	1382–1382	England	Legal deed	Sheep
BA 28	MOR 4	1392–1392	England	Legal deed	Sheep
BA 21	BED 1	1393–1393	England	Legal deed	Sheep
BA 42	YM/D/ASK 2	1404–1404	England	Legal deed	Sheep
BA 29	MOR 6	1415–1415	England	Legal deed	Sheep
BA 62	YM/D/CAMP 2	1427–1427	England	Legal deed	Sheep
BA 30	MOR 7	1443–1443	England	Legal deed	Sheep
BA 82	YM/D/PO	1447–1447	England	Legal deed	Sheep
BA 31	MOR 8	1448–1448	England	Legal deed	Sheep
BA 32	MOR 10	1455–1455	England	Legal deed	Sheep
BA 63	YM/D/CAMP 3	1456–1457	England	Legal deed	Sheep
BA 136	YM/D/YOR 1	1466–1466	England	Legal deed	Sheep
BA 33	MOR 11	1481–1481	England	Legal deed	Sheep
BA 142	YM/D/HAT 2	1484–1484	England	Legal deed	Sheep
BA 43	YM/D/ASK 4	1492–1492	England	Legal deed	Sheep
BA 44	YM/D/ASK 7	1494–1495	England	Legal deed	Sheep
BA 83	YM/D/PO	1503–1503	England	Legal deed	Sheep
BA 64	YM/D/CAMP 4	1507–1507	England	Legal deed	Sheep
BA 34	MOR 12a	1511–1511	England	Legal deed	Sheep
BA 09	YM/D/BAL a. 5	1513–1513	England	Legal deed	Sheep
BA 161	YM/D/OUS 1	1518–1518	England	Legal deed	Sheep
BA 137	YM/D/YOR 2	1520–1520	England	Legal deed	Sheep
BA 10	YM/D/BAL a. 6	1525–1525	England	Legal deed	Sheep
BA 22	BED 2	1529–1530	England	Legal deed	Sheep
BA 35	MOR 13	1534–1534	England	Legal deed	Sheep
BA 139	YM/D/YOR 6	1537–1537	England	Legal deed	Sheep
BA 65	YM/D/CAMP 5 (b)	1540–1540	England	Legal deed	Sheep
BA 101	YM/D/LANC 1	1542–1542	England	Legal deed	Sheep
BA 102	YM/D/LANC 2	1546–1546	England	Legal deed	Sheep
BA 66	YM/D/CAMP 10	1547–1547	England	Legal deed	Sheep
BA 144	YM/D/HAT 5	1549–1550	England	Legal deed	Sheep
BA 121	YM/D/KIRK 1	1552–1552	England	Legal deed	Sheep
BA 67	YM/D/CAMP 13	1554–1554	England	Legal deed	Sheep
BA 46	YM/D/ASK 11	1556–1556	England	Legal deed	Sheep
BA 84	YM/D/PO	1557–1557	England	Legal deed	Sheep
BA 85	YM/D/PO	1558–1558	England	Legal deed	Sheep
BA 103	YM/D/LANC 3	1561–1561	England	Legal deed	Sheep
BA 68	YM/D/CAMP 16	1563–1564	England	Legal deed	Sheep
BA 02	YM/D/BAL(M) 4	1565–1565	England	Legal deed	Sheep
BA 11	YM/D/BAL a. 9	1567–1568	England	Legal deed	Sheep
BA 47	YM/D/ASK 13	1571–1571	England	Legal deed	Sheep
BA 70	YM/D/CAMP 21	1573–1573	England	Legal deed	Sheep
BA 107	YM/D/LANC 9	1574–1575	England	Legal deed	Sheep
BA 03	YM/D/BAL(M) 10	1576–1576	England	Legal deed	Sheep
BA 124	YM/D/KIRK 4	1577–1577	England	Legal deed	Sheep
BA 109	YM/D/LANC 14	1579–1580	England	Legal deed	Sheep
BA 127	YM/D/KIRK 8	1580–1581	England	Legal deed	Sheep
BA 48	YM/D/ASK 16	1581–1581	England	Legal deed	Sheep
BA 87	YM/D/PO	1582–1582	England	Legal deed	Sheep
BA 169	YM/D/OUS 9	1586–1587	England	Legal deed	Sheep
BA 170	YM/D/OUS 10	1589–1589	England	Legal deed	Sheep
BA 14	YM/D/BAL a. 18	1591–1591	England	Legal deed	Sheep
BA 130	YM/D/KIRK 12	1592–1593	England	Legal deed	Sheep
BA 153	YM/D/HEN 5	1595–1596	England	Legal deed	Sheep
BA 154	YM/D/HEN 6	1596–1597	England	Legal deed	Sheep
BA 115	YM/D/LANC 24	1598–1598	England	Legal deed	Sheep
BA 52	YM/D/ASK 21	1599–1599	England	Legal deed	Sheep
BA 116	YM/D/LANC 25	1600–1600	England	Legal deed	Sheep
BA 05	YM/D/BAL(M) 21	1601–1601	England	Legal deed	Sheep
BA 131	YM/D/KIRK 13	1603–1604	England	Legal deed	Sheep
BA 15	YM/D/BAL a. 28	1605–1605	England	Legal deed	Sheep
BA 16	YM/D/BAL a. 33	1607–1607	England	Legal deed	Sheep
BA 73	YM/D/CAMP 44	1608–1608	England	Legal deed	Sheep
BA 117	YM/D/LANC 27	1610–1610	England	Legal deed	Sheep
BA 74	YM/D/CAMP 46	1611–1611	England	Legal deed	Sheep
BA 118	YM/D/LANC 28	1611–1612	England	Legal deed	Sheep
BA 133	YM/D/KIRK 16	1614–1615	England	Legal deed	Sheep
BA 18	YM/D/BAL a. 44	1615–1615	England	Legal deed	Sheep
BA 75	YM/D/CAMP 50	1619–1619	England	Legal deed	Sheep
BA 20	YM/D/BAL a. 49	1621–1622	England	Legal deed	Sheep
BA 93	YM/D/PO	1624–1624	England	Legal deed	Sheep
BA 58	YM/D/ASK 40	1629–1630	England	Legal deed	Sheep
BA 77	YM/D/CAMP 56	1632–1632	England	Legal deed	Sheep
BA 59	YM/D/ASK 42	1633–1633	England	Legal deed	Sheep
BA 134	YM/D/KIRK 20	1635–1636	England	Legal deed	Sheep
BA 78	YM/D/CAMP 58	1650–1650	England	Legal deed	Sheep
BA 90	YM/D/PO	1661–1661	England	Legal deed	Sheep
BA 96	YM/D/PO	1669–1669	England	Legal deed	Sheep
BA 37	MOR 45	1677–1677	England	Legal deed	Sheep
BA 135	YM/D/KIRK 22	1678–1678	England	Legal deed	Sheep
BA 25	BED 14	1691–1691	England	Legal deed	Sheep
BA 147	YM/D/HAT 10(b)	1699–1699	England	Legal deed	Sheep
BA 148	YM/D/HAT 11	1717–1717	England	Legal deed	Sheep
BA 97	YM/D/PO	1726–1726	England	Legal deed	Sheep
BA 39	MOR 48	1744–1744	England	Legal deed	Sheep
BA 98	YM/D/PO	1754–1754	England	Legal deed	Sheep
BA 79	YM/D/CAMP 59 (a)	1765–1765	England	Legal deed	Sheep
BA 80	YM/D/CAMP 60	1767–1767	England	Legal deed	Sheep
BA 99	YM/D/PO	1775–1775	England	Legal deed	Sheep
BA 100	YM/D/PO	1790–1790	England	Legal deed	Sheep
BA 60	YM/D/ASK 44	1794–1794	England	Legal deed	Sheep
BA 149	YM/D/HAT 12	1818–1818	England	Legal deed	Sheep
BA 40	MOR 49	1819–1819	England	Legal deed	Sheep

Our results have shown that pocket Bibles, some of the first examples of commercial book production (8), were written on all three species used widely in parchment production. In our survey of 220 folios spanning three centuries, no unexpected species were recorded. The use of sheepskin in only one of 72 pocket Bibles indicates that it was not favored for these very thin membranes despite the facts that sheepskin delaminates and that the range of thicknesses measured for calf and sheep parchment are similar (0.09–0.28 mm for calf and 0.07–0.26 mm for sheep) (Fig. 4D). The presence of goatskin and sheepskin parchment in the sample set would seem to indicate that uterine calfskin was not necessarily used to produce very fine membranes. Indeed, similar thickness measurements can be achieved for all three species (Fig. 4A).

Fig. 4.

Thickness of folios. (A) Thickness of all pocket Bible folios by species. (B) Thickness of calf pocket Bible folios by total number of folios in Bible. (C) Thickness of goat pocket Bible folios by total number of folios in Bible. (D) Thickness of folios from pocket Bible Ee.6.26.

SI Materials and Methods

Four samples of uterine calfskin, each measuring 0.5 cm × 0.25 cm, were prepared, as well as four control “nonuterine” calfskins, also measuring 0.5 cm × 0.25 cm. Samples were incubated in 0.5 M triethylammoniumbicarbonate (TEAB) (pH 8.5) at 65 °C for 1 h to solubilize the collagen. The supernatant was quantified using a Qubit (Thermo Fisher Scientific). One hundred fifty micrograms of protein from each sample was digested overnight with trypsin at 37 °C at a ratio of 1:75 (trypsin to protein). The tryptic digest was transferred over C18 resin to desalt and concentrate peptides by washing with 0.1% TFA. Peptides were eluted in a final volume of 50 μL of 50% acetonitrile (ACN)/0.1% TFA (vol/vol). Digested samples were labeled with the iTRAQ reagents following the protocol provided by the vendor (Applied Biosystems). Briefly, one vial of iTRAQ labeling reagent was used for every 100 μg of protein. Ethanol was used to solubilize the iTRAQ reagent and then added to the peptide sample, ensuring a final organic concentration of at least 60% (vol/vol). The labeling reaction was performed by 2 h of incubation at room temperature. Samples were cleaned up before LC-MS/MS analysis using a cation-exchange cartridge system (SCIEX) and desalted using a C18 cartridge.

Dried peptides were resuspended in 2% (vol/vol) ACN and 0.1% TFA and analyzed by nano–LC-MS/MS as described previously (52). In brief, samples were separated using nano-ultraperformance liquid chromatography (Easy spray C18 column with a 75 μm × 500 mm, 2.1-μm particle size; Thermo Fisher Scientific) coupled to a Q Exactive tandem mass spectrometer (Thermo Scientific). MS data were acquired with a resolution of 70,000 at an m/z of 200 and by selecting the top 15 precursor ions. The ion target in MS1 mode was 3 × 10⁶, and it was 5 × 10⁵ in MS2 mode. Ions with an m/z between 380 and 1,800 were accumulated for up to 100 ms in MS1 and for up to 128 ms in MS2. We used a stepped normalized collision energy at 31%, 34%, and 37%; a precursor isolation window of 1.6 m/z with an offset of 0.3 m/z; a fixed first mass of 100 m/z; and a resolution of 35,000 (profile) for MS2 spectrum acquisition. The samples were loaded in 0.1% TFA in 1% CH₃CN and eluted with a gradient of 3–35% (vol/vol/vol) CH₃CN in 5% DMSO and 0.1% formic acid for 60 min at a flow rate of 250 nL⋅min⁻¹.

Identification/quantitation of iTRAQ signals was processed with PEAKS 7 (Bioinformatics Solutions, Inc.). Raw files were refined to correct precursor mass, and MS2 spectra were searched against the UniProt reference proteome of Bos taurus (April 7, 2014). Search parameters include trypsin with up to two missed cleavage sites, 10 ppm mass tolerance for the precursor, and 0.05 Da mass tolerance for fragment masses. Carbamidomethylation was a fixed modification, whereas we used the posttranslational modifications (PTM) search module to identify otherwise modified peptides. For quantitation of the eight-plex iTRAQ reporter signals, we used the default setting of the quantitation module.

DNA Analysis.

DNA extraction.

DNA was extracted from a 0.001-g cut piece of parchment UPenn 03 (Ms. Codex 1065), following a protocol based on the technique of Yang et al. (53), in a dedicated ancient DNA facility at Trinity College Dublin. Briefly, the parchment sample was incubated at 55 °C for 24 h in 0.5 mL of lysis buffer [1 M Tris⋅Cl (pH 7.4), proteinase K, 30% (vol/vol) sarkosyl NL, 0.5 M EDTA]. Following this incubation, the sample was centrifuged at 16,249 × g for 5 min to precipitate any undigested material; the supernatant was then removed, and DNA was recovered using the Qiagen Minelute kit. The first DNA binding step of this protocol (Qiagen Minelute) was repeated four times due to the large volume of starting material, and a final elution volume of 20 μL was used.

Library preparation and sequencing.

An Illumina sequencing library was produced for UPenn 03 from the 20 μL of extracted DNA following the protocol of Meyer and Kircher (54) with the previously described modifications of Gamba et al. (55). The 50-bp single-end sequencing of the constructed library was performed on the Illumina MiSeq platform at the Trinity Genome Sequencing Laboratory (TrinSeq), Trinity College Dublin.

Bioinformatic analysis.

Raw sequencing reads were trimmed of contaminating adapter sequences using Cutadapt (56). Identification of the source species of the parchment was performed from the trimmed reads using FastQ Screen (www.bioinformatics.babraham.ac.uk/projects/fastq_screen//) with default settings (13).

Numbers of Membranes.

Ruzzier (6) provides (conservative) estimates of the total volume of pocket Bibles produced during this period. If we assume that 20,000 Bibles (a number dependent upon estimates of survival rate) were produced, of which 54% were produced in France, and if we assume that the vast majority of these Bibles produced in France were produced in Paris (10,000), 44% of French production (∼4,700 Bibles) occurred in middle of the 13th century, which, if correct, would average more than 150 Bibles per year.

Using Ruzzier’s (6) data, we can estimate a maximum production of ∼200 Bibles per year. Assuming an average of 474 leaves (237 bifolia) per Bible, with an average height + width of 310 mm, the production of 200 Bibles would be the equivalent of 55,000 rabbits (assuming one rabbit per bifolia), ∼27,500 stillborn sheep/goats (slink), 18,000 stillborn calves (slunk) or young (20-d-old) kids/lambs, or 4,500 young calves or full-grown caprines.

After the early 11th century, when the population of Paris was estimated to be only a few thousand inhabitants (57), a period of rapid population growth occurred linked to the status of capital city conferred on Paris under the reign of Philippe Auguste (1180–1223). This population expansion continued until the early 14th century, when it reached 200,000–300,000 capita (58), falling following the plague (August 1348 through the winter of 1349/1350) (59). Le Menagier de Paris was written a century after the height of Bible production (1393), and the author estimates 306 veal calves were consumed per week (60) (15,000 per year). In modern-day herds, ∼2–5% of calves die at birth and a further 3–6% die before their first month of age (61). Therefore, even given alternative uses for calfskin parchment, our best estimates do not suggest that demand and supply are wildly at odds. As for sheep, Postan (62) estimates from taxation records that there were four to five sheep for every head of cattle in England in the 13th century, and perinatal mortalities are higher in sheep and goats than in cattle [10–12% in modern flocks (63)]. Because at the height of British production, only 9,000 stillborn lambskins would be required, and even fewer if parchment was being obtained from Italy (although this volume would be beyond the 10,000 skins required to satisfy annual Italian production at its height), once again, nothing indicates unsustainable herd management practices.

Discussion

Aside from local availability of livestock and preferences in meat consumption, parchment production in medieval Europe was limited by logistical challenges. Parchment production was presumably located close to the point of slaughter, because skins deteriorate within days, resulting in poor quality and spotty membranes (2). Hides can be preserved by salting, but this process would be prohibitively expensive in Northern Europe: The salt price in England rose from a base above 0.1 g⋅Ag⋅kg⁻¹ during this period.

In addition, the prestige and value of particular types of skin played an important role. Fourteenth-century accounts from Beaulieu Abbey (27) show that calfskin was more highly valued than sheepskin. The importance of the hide as a source of revenue is repeatedly evidenced by the abundance of skinning marks on animal remains, the range of decrees and bylaws restricting the flaying of animals that had died of diseases, and the lengths the authorities went to prevent these skins being used (28). The intact skeleton of a cow from a 14th-century burial at Téteghem Carlines 3 (Northern France) that died of a dystocic calving (29) gives a further insight into the importance (or otherwise) of uterine hides. Although the cow’s hide had been removed, the calf remained trapped and unflayed in the birth canal, despite the fact that it could have been easily released and flayed to obtain uterine calfskin for parchment.

However, most of the geographic variations can be explained by the local livestock availability and preferences in meat consumption in the different regions.

France.

The relatively low use of goat parchment in Parisian pocket Bibles may be explained, in part, by the patterns of consumption. In Le Viandier de Taillevent (a recipe book first produced in the early 14th century), of 140 recipes, veal appears in seven (5%). Kid and lamb are listed in only one recipe and are considered interchangeable depending on availability. Evidence from the 15th century suggests that sheep and goats were much more common in Southern France (30). Excavations in multiple sites from the Parisian area revealed 3,102 bone fragments dating to the 13th century, the majority of which were caprines (sheep/goat, 38.2%) followed by similar levels of ∼30% cattle and pig (31).

Although the species of animal used would reflect local livestock availability, the age of the animals was limited by the craft production. La Lande’s book Art de faire le parchemin, written in 1762, indicates that calfskins suitable for making parchment should not be taken from an animal older than 6 wk of age (32). The skins of young calves, although suitable for parchment production, are already several times thicker than the skin of adult goats or sheep. Although adult goats and sheep can be used for the production of certain types of parchment, adult cow skins are too thick for parchment production and are instead used to produce heavy, durable leather (33). However, Clavel (31) notes that calf bones are rarely found in 13th and 14th century sample collections and comprise only 2–5% of all cattle bones from excavated sites, appearing with greater frequency in later periods. Veal calves were typically slaughtered at around 6 mo of age, but the bones of very young cattle have rarely been found in French medieval archaeological sites (34).

England.

The presence of goatskin in the English pocket Bibles is intriguing, an assertion that is further supported for one sample with DNA evidence (Fig. S2). Goatskin folios produced in England are thinner than their Italian or French counterparts (Fig. 4C). This evidence would seem to indicate that goatskin was not routinely available and may have been acquired specifically for the production of pocket Bibles. However, goat horn-core accumulations are not unusual, particularly in the eastern part of the country and in urban sites, possibly the product of an international skin trade, because horns were routinely left within the transported skins (22). In both early and later medieval England, sheep are overwhelmingly more common than goats; when identifiable, the latter invariably represent less than 10% of the total sheep/goat bone and tooth fragments, and this proportion is often close to 1%. Goat bones occur on ∼50% of sites, but invariably in small frequency. In the late-medieval period, both documentary (35) and archaeological (34) sources attest to a further decline in the use of goat.

Fig. S2.

FastQ Screen analysis of DNA extracted from UPenn 03 (Ms. Codex 1065) following Teasdale et al. (13). A species identification of goat is confirmed, both by the alignment to the goat genome being the maxima for all species considered and by the most uniquely aligning sequences (red) belonging to goat. Cross-alignment of other reads is expected owing to the high homology, especially of repeated elements, among ruminant genomes.

Cattle were predominantly slaughtered as adults in medieval England, which reflects their major role as tractors. The gradual replacement of working cattle with horses toward the end of the period, at least in some regions, provided the opportunity for an increase in early culling. From the 15th century onward, several sites, mainly urban, have produced relatively high proportions of calf bones, probably a consequence of an increased demand for veal and milk, as well as a greater production of calfskins (34, 36), but this evidence does not explain the predominant use of calfskin in English pocket Bibles in the 13th century.

Italy.

At Italian archaeological sites, sheep bones also tend to predominate during the medieval period, but far less so than in England, with a typical sheep/goat ratio ranging between 3:1 and 2:1 (37–39). Italian zooarchaeological reports, however, rarely provide specific identifications for caprines, perhaps due to the difficulty in separating sheep from goat morphologically. The predominance of goat horn-cores is not a phenomenon known for Italian sites.

The use of cattle is clearly variable, with emphasis on meat production or traction, depending on the site. The presence of very young animals is, however, reported at several urban (39) and high-status sites (40). However, the nature of the current evidence is insufficient to allow us to identify any clear chronological trends.

Conclusion

This study reveals the value of the triboelectric eZooMS approach to the analysis of parchment. It requires no specialist equipment or storage, samples can be collected when appropriate without the need to transport the artifacts, and the samples can then be analyzed when required. As we have shown (Fig. 1), triboelectric extraction is more efficient than physical sampling for four reasons. First, much less material is required. Second, the molecular extraction is bulked onto a large volume of eraser waste, which means it is easy to subsample. Third, molecules are stabilized on the surface of the eraser waste. Fourth, eZooMS acts as a prepurification step, as evidenced by the results from whole samples. Following extraction, the pigmented extracts remain on the eraser crumbs and only the macromolecules are extracted.

We have been able to provide the first significant molecular evidence, to our knowledge, to resolve the long-standing question of the origin of uterine vellum. We find no evidence of unexpected species, such as rabbit or squirrel. Although this type of parchment derived mostly from calfskin in France, in other places, different skins were used subject to local availability. The production of ultrathin parchment was the result of a technological production process using available resources and, as such, would not have demanded unsustainable agricultural practices. We have further shown that manuscript documents made from animal skins are a valuable additional resource for archaeologists exploring livestock economies.

Although the use of genuine uterine vellum cannot be discounted, our results suggest that its availability was not a defining factor in medieval parchment production. Instead, our findings would seem to emphasize dependence on a highly specialized craft technique rather than the supply of a particular raw material. A more likely explanation for the production of fine parchment is the use of relatively young animals and the deployment of specific finishing techniques that enabled the corium to be ground to the desired thickness. The density of collagen fibrils in calf and goat parchment, compared with a more open weave and higher fat content in sheep parchment, favors the former two species; nevertheless, it is evident that parchment makers had the skills to make the finest parchments from all three.

Materials and Methods

All codices sampled were classified in their catalog entries as 13th-century Bibles. The 79 Bibles were subdivided, depending on their dimensions, into two groups based on criteria used by Ruzzier (6): pocket Bibles (height + width < 385 cm) and nonpocket Bibles (height + width >385 cm). Details of all Bibles sampled can be found in Tables S2 and S3.

Sampling.

Samples were extracted in the participating archives and libraries using kits sent by one of the authors (S.F.) consisting of 1.5-mL microcentrifuge tubes, nitrile gloves, acid-free paper, and nonabrasive conservator’s erasers. Sampling was performed using a Staedtler “Mars Plastic” eraser, rubbing the eraser in one direction and collecting the resulting eraser waste fragments in individual 1.5-mL microcentrifuge tubes. For each sample, a new individual piece of eraser and acid-free paper was used; the eraser and paper were thrown away once sampling of the folio was completed to avoid cross-contamination. Nitrile gloves were worn throughout the sampling process to avoid keratin (human skin) contamination. Sample collection was undertaken on areas of the document that had no writing and presented structural integrity (absence of holes or tears in the parchment). All samples were stored at room temperature until required, usually by the partner. Details of each document sampled were entered onto an online spreadsheet shared between the partner and the laboratories in York.

eZooMS.

Initially, eraser crumb samples [sometimes known as “erdu” (41)] were spun down at maximum speed on a benchtop centrifuge for 1 min and 75 μL of 0.05 M NH₃CO₃ (AmBic) buffer (pH 8) was added to each sample. Samples were heated at 65 °C for 1 h. Once cooled, 1 μL of trypsin (0.4 μg/μL) was added and samples were incubated at 37 °C for 18 h. However, at a later stage, the method of collagen extraction was optimized by removing the heating step and condensing the process into one incubation step at 37 °C for 4 h with both AmBic and trypsin added simultaneously. After incubation with trypsin, digests were spun down at maximum speed on a benchtop centrifuge for 1 min and 1 μL of 5% (vol/vol) TFA was added. Samples were desalted and concentrated using C18 resin (Millipore), following the manufacturer’s instructions. Peptides were eluted in a final volume of 50 μL of 50% acetonitrile (ACN)/0.1% TFA (vol/vol). One microliter of eluted peptides was mixed on a ground steel plate with 1 μL of α-cyano-4-hydroxycinnamic acid matrix solution [1% in 50% ACN/0.1% TFA (vol/vol/vol)] and air-dried. All samples were spotted in triplicate. Samples were analyzed using a calibrated Ultraflex III (NLD1; Bruker Daltonics) MALDI-TOF instrument in reflector mode. Spectral analysis was performed using the open-source cross-platform software mMass (www.mmass.org) (42), and individual peptides were identified manually according to Buckley et al. (43, 44).