1. Size
2. Binding
3. Writing surface and structure of gatherings
3.1 Gathering 1
3.2 Gatherings 2-7
3.3 Gathering 8
3.4 Gatherings 9-10
3.5 Gatherings 11-12
3.6 Gatherings 13-16
3.7 Gathering 17-18
3.8 Gathering 19
3.9 Gatherings 20-22
3.10 Signatures
3.11 Are the gatherings in order?
4. Ruling
5. Framing
6. Mise en page

1. Size

The leaves of MS 17 measure, on average, 340 x 250 mm. This makes it a large volume by comparison with most computus manuscripts. With few exceptions, contemporary English computi never exceed the 280 x 195 mm of Cambridge St John's College A.22; indeed, apart from the great Bibles, few English manuscripts of any kind from this period are larger than 300 x 190 mm.1 English computi on average fall into the range of 190/230 x 120/150 mm, and some, notably Oxford Bodleian Library Bodley 614 (142 x 98 mm) and Durham Hunter 100 (170 x 120 mm) are literally hand-books. Continental manuscripts are also comparatively modest in size. Only four can be compared to MS 17 in grandeur. Two are Carolingian: Cologne 83(II) (380 x 270 mm) and the Fleury manuscript Paris BNF nouv. acq. lat. 1615 (320 x 225 mm). A 10th century volume from St Martial in Limoges, Paris BNF lat. 5239, is similarly proportioned (330 x 255 mm). Finally, Bodley 309 measures 343 x 238 mm.

The two English computus manuscripts which most closely approximate MS 17 in size are the Winchcombe computus (British Library Cotton Tiberius E.IV), and Egerton 3088, possibly a copy of the Winchcombe volume. The Tiberius manuscript measures 310 x 230 mm, but it was damaged in the Ashburnham House fire of 1731, so this probably represents a diminution of 3 or 4 cm from its original size. The Egerton volume is 320 x 225 mm. Their resemblance to MS 17 is significant because many of their texts and tables (particularly of Bede's computistical treatises and their glosses) belong to the same family as those in MS 17. The three manuscripts also share a splendour and finesse of execution which set them apart from other contemporary volumes of this genre.

The size of MS 17 plays an important role in its overall codicological character. It resonates with the fine parchment, the script, decoration and lay-out to create a striking impression of beauty and dignity. The Thorney scriptorium deliberately set out to present computus in ceremonial dress, and in so doing, revealed something of the value they placed on this subject.

2. Binding

MS 17 was rebound at least twice: once, probably in the 13th century, when gathering [19] was inserted, and again in the 17th century. The 13th century boards were incorporated into this new binding.2 The terminus a quo for this binding is 1623, when Robert Cotton was still in possession of the manuscript. Were it in its present binding then, it is doubtful that he would have been able to slice out the five leaves which now comprise Cotton Nero C.VII fols. 80-84 without leaving a more obvious wound: the two stubs which witness to this mutilation are now almost invisible in the tight binding. Moreover, it was foliated in a seventeenth century hand after it was restored to the College, for there is no break in the numbering where the Cotton leaves should be. This foliation therefore probably took place in conjunction with re-binding.

The present binding's metal corner-plates and boss, as well as its elegant gold-stamped centre-piece and thin gold fillet, confirm that it was made in the 17th century. The volume is sewn on five thongs. Two clasps survive on the back cover, but the straps that once were inserted into the grooves on the front cover have disappeared. The paper fly-leaves came from two manufacturies. One used a watermark with the maker's name, possibly "Robert Crivet," surmounted by a fleur-de-lys.3 The manuscript's shelf-mark "17" is in gold at the head of the spine, and the designation "old" has been written in black ink on the leading edges. The bookplate of St John's College is attached to the front pastedown.

3. Writing surface and structure of gatherings

The parchment of MS 17 is of very high quality -- thin, supple, smooth, white, and with very few holes or blemishes. The skins were prepared so well that their hair and flesh sides can often only be distinguished by touch. The parchment is frequently thin to the point of translucency, so that painted areas will show through on the other side: for example, the frame of Byrhtferth's Diagram is visible on fol. 7r, while fol. 39r carries the "ghost" of the syzygia elementorum on its verso. The occasional small hole was tolerated and written around, but larger defects had to be repaired before writing could begin. For example, on fol. 109, a gap was patched by gluing and stitching on a second piece of parchment; the text travels across the sewing. A suture of salmon-pink and green silk was used to close a rend on fol. 168, and beige linen thread served the same purpose on fol. 177.

The parchment used for the gathering of twelve leaves between fols. 144 and 155 is noticeably thicker, coarser and darker. In fact, it was inserted into MS 17 a century after the original codex was completed.

The excellence of the parchment makes it difficult to determine whether the gatherings were cut after folding, or built up of pre-cut bifolia.4 There are no stains or holes at the edges of the leaves which can be matched with others to show that whole sheets had been folded and then slit open. Moreover, given the dimension of MS 17, this is inherently unlikely. English scriptoria apparently favoured bifolia construction of gatherings, as is evident from the number of manuscripts in English Caroline which have hair side facing flesh side across an opening. This is the normal pattern in early Insular codices, and in continental manuscripts from centres under Insular influence. In England, the practice declined in the 10th century, but was revived in the 11th.5 In MS 17, however, with one notable exception, hair side faces hair side, and flesh flesh. The outside of a gathering is always the hair side.

The Site Map shows the arrangement of the gatherings in MS 17, together with the distribution of the texts and the fields of activity of the various scribes. Gatherings 1-17, 20-21 were numbered in the Middle Ages as I-XIX. Gathering 18 consists of two 12th century leaves, continuing the text from gathering 17 and written by the Thorney scribal team. Gathering 19 is a complete gathering of twelve leaves inserted in the 13th century. These have no signatures. From this we may conclude that the signatures antedate the insertion of gathering 19.

The binding of MS 17 makes it difficult to see the top and bottom of the spine, and thus to trace a folio to its conjoin. Comparing the colour and texture of leaves suspected of being conjoins reveals many anticipated regularities, but also some irregularities which suggest that on occasion two single leaves were employed instead of a bifolium. In some instances, this in confirmed by a stub. An added complication is presented by Cotton Nero C.VII fols. 80-84. These five leaves were excised from MS 17 by Robert Cotton, and eventually mounted on binding strips, together with an extra blank leaf at the end, to form an artificial gathering of six. This extra leaf (fol. 85) is difficult to explain. Was it originally in MS 17, or provided from elsewhere to make up this gathering? The parchment is fine, creamy, and suspiciously similar to that used in the other Cotton folios, but if it came from MS 17, why would Cotton have bothered to remove it? And why was it not removed when MS 17 was re-bound in the 13th century to insert gathering 19? To suggest that it came from MS 17 is to invoke too many improbabilities: moreover, fol. 85 is slightly wider than the other leaves, and was never pricked or ruled. It is therefore likely that it was supplied from elsewhere.

3.1 Gathering 1 (fol. 1-3)

Gathering 1, signed at the foot of fol. 3v, is composed of three leaves which at some point broke loose from the binding and were mounted on strips for re-binding. The flow of the texts proves that the leaves are in correct order, but was this always a gathering of three? The binder's cord runs between fols. 2 and 3, but it is improbable that there was ever a fourth folio conjoined to fol. 1. Fol. 3v and the present fol. 4 contain the same type of text -- prognostica -- written by the same scribe. However, a missing leaf could fall between fols. 2 and 3, or just possibly before fol. 1. At present, gathering 1 looks like this:

1r 1v2r 2v 3r 3v

If it originally was a gathering of four, it could have looked like this:

hair|flesh hair|fleshflesh|hairflesh|hair
?r ?v1r 1v 2r 2v 3r 3v

or this:

hair|flesh hair|fleshhair|fleshflesh|hair
1r 1v2r 2v?r ?v3r 3v

Alternately, it could have been a bifolium (fols. 1 and 3) with fol. 2 as an inserted singleton.

None of these options, however, threatens the position of gathering 1 as the original first gathering in the codex. Fol. 1r was blank until the 15th century when the geometry problem was copied onto it, which suggests that it had always been the guard-page.

3.2 Gatherings 2-7

Gathering 2 in twelve (fols. 4-15), gatherings 3-5 in eight (fols. 16-23, 24-31 and 32-39), and gathering 6 in twelve (fols. 40-51) are unproblematic. At gathering 7 (fol. 52-61) we encounter for the first time the economical habit of the Thorney scriptorium of using single sheets. Sometimes, as in gatherings 9 (fols. 70 and 73) , 11 (fols. 86 and 89) and 12 (fols. 94 and 97), a pair of single leaves does duty as a bifolium. In gatherings 7 (fols. 54 and 56) and 18 , however, sheets are inserted to accommodate the end of a text, thus producing an asymmetrical gathering. This expedient is only necessary if the text following the piece to be accommodated had already been written, thus leaving the scribe no choice but to insert an extra folio. Gathering 7 thus demonstrates that the compilers of MS 17 did not copy their manuscript tel quel from a single exemplar, but rather were selecting materials from a variety of sources. They seem to have had a plan in mind, but the plan occasionally ran into space-allocation problems. What disturbed the orderly structure of gathering 7 and necessitated the insertion of additional leaves was the acquisition of a full text of Bede's De temporibus: see the Overview of Scientific Works of Bede for details.

3.3 Gathering 8 (fol. 62- 67)

The first two leaves of gathering 8, containing the opening chapters of Bede's De natura rerum, were cut out before the manuscript was foliated in the 17th century. At the top of fol. 62, a note penned at St John's College in 1611 signals this omission. The stubs of the two missing leaves, with fragments of glosses and chapter rubrics, are visible.

3.4 Gatherings 9-10

Gathering 9 (fol. 68-75) is in eight, the third "bifolium" (fols. 70 and 73) being composed of two single leaves (the stubs are visible between fols. 70 and 71, and fols. 72 and 73). Gathering 10 (fol. 76-83) is an ordinary gathering of eight.

3.5 Gatherings 11-12

Gathering 11 is in eight (fols. 84-91), but the colour and texture of fols. 86 and 89 very strongly suggest that these are also two single leaves doing duty as a bifolium. However, only one stub is visible, namely between fols. 89 and 90. The most likely explanation is that in this gathering the stubs are turned outwards rather than inwards. In gathering 12 (fol. 92-99), fols. 94 and 97 are also, to all appearances, single sheets, but the location of the stubs suggests that both stubs were folded towards the front of the gathering. There are thus four different ways of handling single sheets: lapping the stubs around the inner bifolium (gathering 7), folding the stubs towards one another, but not overlapping (gathering 9), folding the stubs away from each other (gathering 11), and folding them in the same direction (gathering 12).6

3.6 Gatherings 13-16

Gatherings 13 (fol. 100-107), 14 (fols. 108-115), 15 (fols. 116-123) and 16 (fols. 124-131) are regular gatherings of eight.

3.7 Gathering 17-18

Gathering 17 (fols. 132-143 + Cotton fols. 82-82) is a mine of information about the practices of the Thorney scriptorium, and the behaviour of later users of the manuscript. At first glace, gathering 17 looks like a gathering of fourteen with two leaves removed at the end, for which two stubs remain. However, when we turn to the Cotton Nero C.VII leaves, we discover that signature XVII is on the third leaf, i.e. fol. 82v. Therefore three leaves, and not two, were removed from the end of gathering 17. One of the first three Cotton leaves was originally a single: but which?

The evidence points to Cotton fol. 80 as the inserted single. Up until MS 17 fol. 143v, each page of the Paschal table covers 43 or 44 years, plus two heading lines giving the tituli of the various columns (annus domini, indictio etc.). On the next leaf, Cotton fol. 80r, this figure jumps to 46 years plus two heading lines. Then from Cotton fol. 81r to fol. 82r, it drops back to 44. At Cotton fol. 82v, the last folio of gathering 17, Scribe A takes over writing the tables from Scribe B. Scribe A is the supervising scribe, responsible amongst other things for the proper positioning and spacing of the texts. By eliminating the heading lines, Scribe A enlarged the page to accommodate 46 years. At the beginning of the new gathering on Cotton fol. 83r, this increases to 48. This remains the pattern until Cotton fol. 84r, the last tables we have from the original 12th century scribes.

MS 17's Paschal tables began in AD 532, the year when Dionysius Exiguus' table started: Dionysius' letters precede the tables. Bede had extended Dionysius' table to a full 532-year Great Paschal Cycle, but that cycle ended in 1063, before MS 17 was begin. The close connection of the Paschal tables with works by Dionysius and Bede, and that fact that the scribes had chosen to begin them at the beginning of the Dionysian-Bedan Great Cycle, strongly suggest that they intended to continue copying until the end of the current cycle, namely 1595. We may suppose that Cotton folios 83 and 84 (gathering 18) were originally the first two folios of a gathering of four allotted for finishing this cycle. Were they continue at the rate of 43 years per page, as was the case up to MS 17 fol. 143v, they would require slightly more than seven leaves to finish the job: fol. 143v ends at AD 960, leaving 633 years to go, which divided by 43 yields 14 pages or 7 folios. However, only six folios were available: two more in gathering 17, and four in gathering 18. The solution was, as usual, to insert a single leaf. That fol. 80 was this single leaf is suggested by the fact that a few extra years were squeezed into the columns -- a little additional effort of compensation.

However, it was not sufficient. At the end of Cotton fol. 82r, there were still 413 years to enter. At 44 years per page, this would require over nine pages, and there were exactly nine pages left. At this point, Scribe A took over. He immediately began his economies on Cotton fol. 82v by eliminating the headlines. At the beginning of the new gathering, he increased the number of years per page to 48. Going at that pace, he would have completed the cycle nineteen lines from the end of gathering 18. Scribe A was being over-cautious, but this was one of his traits.

3.8 Gathering 19 (fol. 144-155)

This reconstruction raises the question of the fate of the last two folios of gathering 18. Following gathering 18 is gathering 19, in twelve. Both the quality of the parchment and the form of the script proclaim that this gathering was not part of the original confection of the volume, but was inserted later. The scribe who wrote out this section of the Paschal tables is "Gothic I", a scribe active in the last quarter of the 13th century. Gothic I also appears on the Cotton leaves, making corrections and additions to the tables there. About halfway down Cotton fol. 82v, Scribe A made the mistake of omitting 1200 from the annus domini column. Therefore from that point onwards, the other columns were out of phase by one line. Lines were drawn between the number for the year and the correct line of computistical data, but the year at the bottom of the table would have no figures for indictions etc. at all, since this information would be attached to the first AD on the next page. Gothic I mended this by filling in these figures at the foot of Cotton fols. 83v and 84r, and canceling or erasing the redundant figures on the following pages. On Cotton fol. 83r, following AD 1236, Gothic I has erased and re-written the cyclus column so that it records the 19-year Paschal cycle, and not the lunar cycle. On Cotton fol. 83v, he begins writing the dominical letters as well. On Cotton fol. 84v, he picks up the Paschal full moon column from AD 1399 onwards, and the Easter Sunday column from AD 1406. On this page, he did not add the data for the final year on the page, but instead he erased the annus domini. At this point, his own new gathering -- gathering 19 -- begins.

From this we may conclude that Scribe A may have completed the AD column of his table to 1595, but that he left the other columns in various stages of incompleteness. This is not improbable, because there are many parts of MS 17 which are incomplete, and which would have been Scribe A's responsibility. At first, Gothic I simply adjusted the tables as described above, but by the end of Cotton fol. 84v, he was writing or re-writing virtually half of them. At this point, he simply decided that it would be less trouble to discard the last two folios of gathering 18 and re-write the tables correctly himself. However, he did not simply replace the last two folios with a bifolium of his own. Rather, he inserted a whole gathering of twelve, and wrote in tables not simply to the end of the current cycle in 1595, but until his parchment ran out in AD 2612.7 In short, he added almost two entire Great Paschal Cycles. There is something intriguing about the imagination and faith that lay behind this feat. A Paschal table is by definition oriented to the future -- its purpose it to tell the date of Easters yet to come -- but the notion that for Gothic I, computus could give some kind of reasonable reality to such unimaginable expanses of future time speaks eloquently to the symbolic dimensions of this science.

3.9 Gatherings 20-22

Gathering 20 contains seven folios (fols. 156-162), the odd single leaf being fol. 158. Apparently the verb list on fol. 159 had already been written, but the prognostica took up more space than had initially been allotted. Gatherings 21 (fols. 163-170) and 22 (fol. 171-177) are in eight, but the last leaf of gathering 22 has been cut off. It is doubtful that any 12th century text was lost, for the medical treatise ends about two-thirds of the way down fol. 177v.

3.10 Signatures

The unusual structure of gatherings 17-19 is a clue to the date at which the signature marks were added. They could not have been added before gathering 19 was bound in, for in that case gathering 18, were it a binion as suggested above, or simply a bifolium, would have been included in the numbering, in which case the present signature XVIII would have been numbered XIX and so forth. Nor could the signatures have been added after gathering 19 was bound in, for were that the case, both gatherings 18 and 19 -- or at least 19 -- would have been numbered. The evidence points to the signatures having been added at the time gathering 19 was bound in. They may have been added for the express purpose of controlling the structure of the codex during the re-binding process. This doubtless took place in the last quarter of the 13th century when gathering 19 was written; the hand that wrote the signatures, Gothic II, does not challenge this dating.

3.11 Are the gatherings in order?

Could the gatherings of MS 17 have been re-arranged at any time after the volume was completed? The distribution of the texts across the gatherings, visible in the Site Map, renders this unlikely. Between gathering 3 and gathering 19, no text begins or ends a gathering with the exception of Bede's De temporibus, which ends with gathering 7 at fol. 61v, and De natura rerum which starts off gathering 8 on the next folio. However, at the close of De natura rerum (fol. 65r), Scribe A notes as part of his explicit that De temporibus may be found "in anteriori quarternione" ("in the preceding gathering"). Gatherings 20-22 likewise form a block. Finally, there are solid arguments for the integrity of gatherings 1-3. Fols. 3v and 4r were both written by Scribe A, and the texts of both these pages are about prognostication. The computistical poetry on fols. 14-15 is closely connected to the calendar verses and metrical martyrology which appear in the calendar in gathering 3. There is therefore little to support any argument that the medical materials on the first and last folios are a later addition, or that they were artificially split into two segments.8 All the codicological evidence supports the conclusion that MS 17's present order represents its original order.

4. Ruling.

Following a practice common in pre-Conquest English scriptoria, but which would gradually die out in the 12th century, MS 17's scribes ruled their folios in dry point, and on the hair side only.9 It is possible, however, that they adopted the new fashion of ruling the sheets after they had been made up into gatherings.10 There are two ruling widths, and several styles of framing in MS 17; each is adapted to the demands of different types of text, and only rarely does a gathering display only one style of ruling and framing. Conjoined leaves are frequently pricked and ruled in different modes: for example, fols. 5 and 6 are ruled to 5 mm, and their conjoins fols. 14 and 15 to 7 mm. The simplest procedure under these circumstances would be to construct a gathering, decide where the texts would go, and custom-rule each leaf. But practical as it may seem, this method may not have been adopted by the Thorney scriptorium. Ruling on one side only generally implies ruling before folding; when pages are ruled individually after folding, this is usually done on both sides.11

Two units of ruling are employed in MS 17: a wide line of 7 mm for texts with interlinear glosses (the computus tracts and poems on fols. 14r-15v, the treatises of Bede, Helperic and Dionysius on fols. 58v-138v, and the verb list on fols. 159-167v), and a narrower ruling of about 5 mm for the remaining texts. Where ruled, glosses are written in a space 4 mm high. One can observe on fol. 63 and its conjoint, fol. 64, how a sheet originally pricked for a ruling of 5 mm was re-pricked and ruled to 7 mm for this glossed text. If the leaves were ruled before the gatherings were assembled, this would argue for a high degree of forethought and control in the execution of this complex manuscript. At a very early stage in its planning, the scribes would have to know where each table and text would fall in order to set out the appropriate ruling and framing.

5. Framing

The type of framing -- single column or double column -- was also selected in accordance with the type of text. There is a single case of triple-column ruling: the Coena Cypriani on fols. 4v-5r. These leaves were originally ruled for two columns and then altered to three. It should be noted that the lunaria on fol. 4r and the runic and exotic alphabets on fol. 5r are both set out as long, narrow lists; possibly this influenced the choice of a triple column format for the text on the reverse sides of these leaves.

The general rule is that unillustrated texts like the medical tracts, Byrhtferht's Proemium, Bede's De temporibus and De natura rerum, and the treatises of Helperic and Dionysius, were written in double columns. So also were texts whose illustrations were subordinate to the text, for example the abacus treatises of Gerlandus with its worked examples, or the text of Bede's De temporum ratione with its adopted family of graphic glosses. But where a diagram or table is the primary object of interest, and the text explains the table, this text, regardless of length, is written in a single column across the page. A good example of this procedure is the section of the manuscript we have called "Computus Tables and Texts II" (fols. 22r-34r). The focus here is on the tables, and in every case the text is in a single column. Another good illustration is provided by the cosmographical anthology (fols. 35v-40v), where single and double columns alternate. The texts from Isidore written in single columns on fols. 37v and fols. 39v-40v are essentially commentaries on the rotae. These rotae are large, splendidly decorated, and visually dominate the page. In the double-column passages from Abbo and Isidore on fol. 38v-39r, the schemata -- small, plain, and limited to the column width or the margin, are visual footnotes to a primary text. It is interesting to contrast Isidore's rather casual reference to the figura solida illustrated on fol. 39r as an extra concession to clarify the text ("Haec itaque ne confusa minus colligantur..."), with his text on fol. 39v introducing the syzygia elementorum, where the studied verbs of tension and interconnection ("complecti", "coniungitur", "compugnantia") strive to translate into words what the rota makes immediately clear. The figura solida is subordinate to the text, while the syzygia is the rationale of its text.

MS 17's constellation of cosmographical materials appears in the Peterborough computus and in Baltimore, Walters Art Gallery 73. In both these volumes, virtually all the texts are in single column format, and where double columns are used (e.g. Walters fol. 4r), the illustration is not subordinated to the column as in MS 17.

This policy for relating text and picture throws an interesting light on the loyalty of the Thorney scribes to the ancient traditions of book design. From the earliest pictured scrolls of ancient Greece, the law obtained that an illustration which was functionally subordinate to its text should also be visually subordinate. Since the unit of writing was a column, illustrations of this type were confined to the width of the column and distributed through the text as its sequence dictated.12

The Thorney scriptorium's concern for visual harmony is also expressed in the geometrical proportions of the text frames of MS 17. Gilissen has identified four basic modules: square; Pythagorean rectangle (4:3), Golden Section rectangle (approximately 1:0.618) and a rectangle of a x a√3, a√4, or a√5.13 These also served as the canons of proportion in MS 17. Throughout MS 17, the height of the writing area is consistently between 232 and 238 mm. The width varies between 102 and 194 mm, depending on the nature of the text. In the verb list on fols. 158v-167v, it is only 102 mm wide, since generous margins were required to accommodate the glosses. The writing area is 153 mm wide for Gerlandus' treatise on the abacus, since the worked examples with their elaborate frames are set into the margins. For text such as the medical tracts or Bede's De temporibus which have no glosses, the width increases to 184 mm. Elsewhere the scribes favour for the most part a writing area about 175 mm wide, which for a single column text produces two Golden Section rectangles laid side to side. If divided into double columns, the proportions are determined by the width of the dividing margin. For example, in Byrhtferht's Proemium (fol. 12v-13r), the columns of 238 x 76 mm are two Golden Section rectangles on top of one another. The writing area of Gerlandus' abacus treatise is a double Pythagorean rectangle, while the columns of 235 x 69 mm are two rectangles of a x a√3 on top of each other. The verb list writing area is a rectangle of a x a√5.

6. Mise en page

Finally, the behaviour of J's scribes towards the page space at their disposal appears at first sight to be somewhat contradictory, but on closer inspection proves to be further evidence of their concern that the volume preserve a certain appearance of continuity. Sometimes they appear to be willing to waste a good deal of space, leaving as much as a whole column blank; at other times, we discover them ruling extra lines to crowd more text onto a page. The motive for the former appears to be a preference for beginning a new text on a fresh opening, or at least on a fresh page: for example, the grammar anthology ends half-way down the last column of fol. 175r, but Scribe A waits for the new opening, fol. 175v, before beginning the medical materials. On the other hand, the scribes were sensitive to the intellectual continuity of their material, and to the visual plenitude of the page, and took pains to ensure that some texts connected seamlessly with others. For example, in the course of writing fol. 138v, Scribe B gradually began to stretch out his n's and the bars of his t's so that a paragraph requiring three lines at the top of the first column required six by the bottom of the second. He did this in order that Dionysius Exiguus' text explaining the Paschal cycle would link up without a gap to the tables based on that cycle, and which start on the following page, fol. 139r. Where there was no intellectual need to carefully mortise a text into its neighbour, he was content to leave empty lines, for instance at the end of the Coena Cypriani on fol. 5r. Crowding, on the other hand, is evidence that the materials were planned to appear in a certain order, but unexpected spacing crises occurred post facto, as was the case of the final page of De temporibus, discussed above.

1 Ker 1960, 40-41.

2 Hanna and Griffiths 2002, 32-33.

3 This mark does not appear in the repertories of Briquet 1968, Churchill 1935 or Heawood 1950.

4 Criteria for this are set out in Gilissen 1977, chs. 1-2. Gilissen finds evidence of the folding method in continental manuscripts dating from as early as the 11th century.

5 Bishop 1971, xii.

6 See MacCarthy 1901, iii-iv re: structure of gatherings in Bodleian Library Rawlinson B.489, where the insertion of single sheets is handled in two different ways.

7 Hart 1970 32 and n. 4 argues that since 2612 is two 19-year cycles short of a full 532-year Great Paschal Cycle, there must have been another gathering of tables following gathering 19. There is no codicological evidence to support this assertion.

8 These are the positions of Singer 1917, 118 and Hart 2003, 423-445 respectively. For detailed discussion of the rationale behind the disposition of the medical materials as "brackets" for the computistica, see Wallis 1995a, 122-127.

9 Ker 1960, 41.

10 Ker 1957, xxiii; Lowe 1934-1966, 2 (2nd ed. 1972), x.

11 Ker 1960, 41-43.

12 Weitzmann 1970, 52-53.

13 Gilissen 1977, part II.