University of Virginia Library

The Two Scribes of the Cardigan Manuscript and the "Evidence" of Scribal Supervision and Shop Production
by
Daniel W. Mosser [*]

The desire to establish a definitive ordering of the Canterbury Tales has generated a number of ingenious and sometimes hard-to-follow arguments based both on the internal evidence of the Tales themselves and on the ordering found in those manuscripts which have been determined to be most authoritative (the growing consensus favors the Ellesmere a order[1]). Too often, the scholarly debate ignores the MSS, but even when the focus is on MS evidence, the risk incurred by assuming too much can undermine the resulting argument. Larry D. Benson describes the usual tendency: "The still flourishing debate over the order of The Canterbury Tales, which has narrowed to arguments for or against the 'Ellesmere order' and the 'Bradshaw shift,' has been carried on pretty much free of any necessity to consult the mss, and proponents on both sides argue almost purely on the basis of literary analysis." However, in trying to shift the focus to manuscript study, Benson commits the error of remaining at least one step removed from the MSS: "I shall therefore try to avoid elaborate speculation and depend instead on the evidence of the mss (drawing almost entirely on Manly and Rickert for evidence; one assumption I do make—and this seems safe enough—is that their analyses of the contents of the mss are generally correct)."[2] While Manly and Rickert probably are "generally correct," my analysis of the Cardigan Manuscript shows that Manly and Rickert's analysis is flawed on

at least two critical points and that the conclusions consequent upon those points are in need of revision.

One assumption that arguments about the unity and intended order of the Canterbury Tales must make is that there was a highly structured and orderly system of fifteenth-century manuscript production, an assumption which has been challenged by recent scholarship.[3] Manly and Rickert assert that Cardigan contains evidence of shop production: "The MS was certainly made in a shop and under careful supervision, of which unusual traces survive."[4] Therefore, it is of particular importance, given Cardigan's status among Chaucer MSS, either to confirm or refute this assumption. The present study will show that Manly and Rickert's evidence, while highly significant in establishing Cardigan's textual authority, in fact points toward the diametrically opposite conclusion: that the MS was not made in a shop. This conclusion is contingent upon the refutation of Manly and Rickert's proposed "supervisor," a figure who, like the "corrector" referred to by H. S. Bennett, would have been "a familiar figure in any big scriptorium."[5] While Manly and Rickert are correct in observing that Cardigan was carefully "proofread," it was in fact corrected by the scribes themselves. In order to establish this it will first be necessary to determine the number of scribes copying the text and the distribution of their work throughout the book.

Cardigan contains a nearly complete text of the Canterbury Tales, lacking the first 152 lines of the Prologue which were presumably contained on two of the three missing folios at the front of the MS.[6] The text of the Tales occupies fols. 1-244 (present foliation). In this portion of the MS, one scribe

(referred to in the present study as Scribe A) wrote from fol. 1 through half of fol. 175 verso (Qq 1-16iv: fig. 1). A second scribe (Scribe B) wrote from fol. 175 verso (l. B2 3109 of the Monk's Prologue) to fol. 244 recto (the end of Chaucer's Retraction), where he finishes Q 24 with a sketchy chronicle (fols. 244 recto through 245 verso). This same scribe (though Manly and Rickert believe it was a third scribe) wrote an inserted leaf, fol. 109 in Q 10, and made some corrections in Scribe A's portion of the MS. Scribe A apparently resumes copying at the beginning of Q 25 (fol. 246 recto) and completes the MS through Q 30, copying Lydgate's Siege of Thebes (fols. 246-304 recto) and Churl and Bird (fols. 304 recto-308 verso; three folios and part of the text of Churl and Bird are missing at the end of Q 30). The text is written on parchment, with an average of 40 lines of text, in single columns, to a page. A page averages 12" x 8½" in size (the parchment is trimmed unevenly).

Despite Cardigan's interest and importance, of the published accounts I have found which deal with the MS at any length, only two focus on some of the many paleographical and codicological problems that the MS presents. The lack of attention paid to the MS results at least in part from its previous owners' (the Brudenells) refusal to allow scholars to examine it.[7] Cardigan came into the "public domain" only after it turned up at Vassar College following its mysterious disappearance from the library of G. L. Brudenell (see note 6). Before the return of the MS to the Brudenell estate, Manly and Rickert were able to obtain photostats for use in their study of the text of the Canterbury Tales. And it is through that study that the first detailed description of the codicological features of the MS, and to a lesser extent, of the paleographical characteristics of the scribes, was made generally available.[8] More recently, George Keiser has addressed a curious problem presented by the tenth quire of the MS, which I will discuss in the latter portion of this study. In an unpublished dissertation, Robert Lovell presents an edition of the Lydgate material contained in Cardigan and describes the MS in his introduction.[9] Both Keiser and Lovell repeat, and presumably,

accept the conclusions of Manly and Rickert concerning the number of scribes and the evidence of supervision.

Manly and Rickert (I, 72) propose that the manuscript was executed by three professional scribes. Their scribe 1, whom I will refer to as Scribe A in order to distinguish my references from those of Manly and Rickert, is described as a "somewhat formal, stiff, though cursive hand." Manly and Rickert express some uncertainty about the hand between fols. 246 and 249, where they say "it is probably the same [as scribe 1], though smaller." This scribe's name may have been Ware, since this word appears in the crosshatching of the page headings on fols. 71 verso (fig. 8), 139 recto, and 261 verso. Scribe 2 is characterized as a "freer, more flourished hand." Manly and Rickert say the third scribe has a "still looser hand" (fol. 109 and "probably some of the corrections"). A careful paleographical analysis of the hands in the MS will demonstrate that Manly and Rickert's distribution of work among scribes needs clarification and revision: specifically, that their scribes 2 and 3 are in fact one scribe, Scribe B.

The scripts employed by the scribes are cursive and suggest, along with the lack of ornament and decoration, that this vernacular manuscript was made at the request of a family or individual who desired the work for the purposes of reading and reflecting upon its contents; it was clearly not intended to be a "display copy" in the sense that elaborately decorated Books of Hours or the Ellesmere Chaucer Manuscript were. M. B. Parkes has suggested that these types of scripts, which allow for more rapid copying, were developed in response to the increased demand for books.[10]

Scribe A uses many of the characteristic graphs of a script labelled by Parkes as anglicana, but also exhibits some tendencies of what would at the time have been a script recently introduced from France, referred to as secretary. Anthony G. Petti, adopting Parkes' terminology, states that the anglicana script was the "chief literary hand at the time of Chaucer . . . formerly known as a court hand."[11] To achieve the degree of cursiveness necessary to the speeding up of production, letter forms were developed which reduced the number of strokes and changes of pen direction to a minimum. Scribe A's book hand comes closest to the general category of scripts labelled anglicana formata, though it is a somewhat bastardized form, employing some of the graphs and characteristics of secretary. The hand preserves the circular e (fig. 2, l. 1: "gone") and regular e (fig. 2, l. 2: "Where"), and sigma s in initial and final position (fig. 2, l. 2: "so"; l. 3: "hys"), characteristics of "normal" anglicana, but evidences a "squatter and squarer" appearance associated with formata (derived from textura, a more formal, less cursive script) as well as penstrokes which are thicker and more angled and ascenders which are taller and more arched than is characteristic of normal anglicana (fig. 2: the "squat

and square" characteristics are particularly apparent in the graphs for o and d, which, in addition to the regular e graph, are formed by thicker and more angular penstrokes; the ascenders of the graphs for l, d, and h all reflect the tendency toward taller and more arched forms; cf. Petti, p. 14, figs. 11 and 12, anglicana and anglicana formata).

The hand also exhibits, but does not exaggerate, tendencies of the ductus of the secretary script, described as an "antithesis of broad strokes and hairlines placed in different diagonals according to the angle of the slanted pen," which gives the hand a slight "splayed" appearance (Parkes, p. xix). Especially characteristic of this tendency in Scribe A's work are the graphs for u, m, and n, formed with far fewer strokes than the more laboriously formed graphs of anglicana formata (fig. 2, l. 3: "thrugh" "strynges"; l. 5: "man"). The hand also tends to emphasize the taper of the descenders, one of the first secretary features to appear in anglicana hands (Parkes, p. xxii). The scribe's habituation to the anglicana graphs is apparent in the retention of the two-compartment a and "double-diamond" g (fig. 2, l. 2: "that"; l. 1: "kynge"). The d graph remains looped, unlike the hybrid anglicana form (fig. 2, l. 3: "and") (Petti, p. 15, fig. 14: hybrid anglicana). The scribe's confidence in the execution of the script is evident in the consistency of the vertical upright strokes, in the regular spacing and shape of the letters, and in the execution of the complex form of w (a pair of rounded loops with a backwards 3 affixed finally: fig. 2, l. 1: "with").

A hand which is similar to that of Cardigan A appears in a copy of John Gower's Confessio Amantis, dated to the early part of the fifteenth century (Cambridge, Trinity College MS R. 3. 2) (Doyle and Parkes, p. 163, pl. 44 [Trinity A]). While this manuscript is dated at least twenty-five years earlier than the Cardigan MS, the A scribe's hand in the Trinity MS is representative of the direction that the anglicana script was moving in the years between 1425 and 1450. Thus it would not be unusual to find this style of anglicana script in Cardigan around 1450. Indeed, characteristics of the script continue to surface as late as 1475, in the hand of Stephen Doddesham (Parkes, pl. 6ii).

A very general distinction can be made between the overall feeling of "roundness" displayed by the hand of Scribe A and the angularity of the hand produced by Scribe B. Petti notes that this is a key distinction between the anglicana and secretary scripts, and Scribe B much more closely approximates the secretary script than does A, not only in producing this feeling of angularity but also in the use of characteristic secretary graphs. On the other hand, the wide variety of forms for the individual letters that can be found in Scribe B's script is evidence of the difficulty with which the transition to the secretary script was made. Cardigan B uses both sigma and B-shaped s finally, the latter being especially characteristic of secretary (fig. 3, l. 3: "his"; l. 5: "Maximus"). Also notable is the alternation between a "double-v" shaped w (fifteenth-century secretary) and the "two-looped l and 3" anglicana form (fig. 3, l. 16: "was" (inserted); l. 15 "trow"). Scribe B also uses two other forms of w which fall somewhat in between these extremes. The distribution of these forms appears to be idiosyncratic. Forms of g are also mixed, with both

the single compartment tailed secretary and the double compartment anglicana graphs in evidence. The secretary g is less common, but again the distribution follows no set pattern (secretary: fig. 1, l. 16: "god"; l. 17: "gentyll" "goste"; anglicana: fig. 3, l. 1: "Image"). A very distinctive final r graph, long and forked with a finishing stroke forming a leftward loop, is often apparently used to abbreviate a final -re (fig. 3, l. 3 "her").[12] Both circular e and regular e are used (fig. 3, l. 12: "prechyng"; l. 1: "Image") and double compartment anglicana a is regular (fig. 3, l. 1: "Image"), though single compartment secretary a appears occasionally. The looped d of anglicana alternates with the unlooped form of the hybrid script (fig. 3, l. 1: "and"; l. 2: "and"). Scribe B uses a number of abbreviations, especially in the prose sections, but his use of them, as well as his use of þ (thorn) for th, which he does often, is inconsistent. A comparison of portions of Cardigan with a microfilm of the Manchester Chaucer MS (Chaucer and Manchester were copied from a common exemplar) suggests that Scribe B was often influenced by his exemplar. This also holds true for the variation between I/y for the first person singular pronoun, which varies in Manchester in many of the same places where B varies his use in Cardigan.

The scribe's hand is cursive, though the letters are frequently not joined, and the thick-and-thin stroke alternation is characteristic of secretary (fig. 3, l. 13: note the contrast between the horizontal and diagonal strokes in the z-shaped r in the word "fro" and between the vertical and diagonal strokes in the m of "hem" in l. 7), as are the very tapered descenders (more tapered than Cardigan A's), as in the long s and f graphs (fig. 3, l. 2: "sacrefyce"). The upward strokes of the graphs do not lead the eye in one direction consistently, unlike the hand of Scribe A whose graphs tend to slant to the right, and the tension produced by the conflicting directions might convey an initial impression of haste and carelessness (fig. 3: note for instance the way the upright strokes of a and d in "and" at the beginning of lines 1 and 2 slant leftwards, while in "sacrefyce" the upright strokes of the s and f graphs are either vertical or rightward-slanting). The hand is characteristic of the middle of the fifteenth century and bears some resemblence to the hands of William Ebesham and some members of the Paston family.[13]

This paleographical analysis shows that two scribes writing with scripts of clearly separate and distinct character copied this manuscript. However, Manly and Rickert also identify a third scribe who copied only one folio, 109, and who made "some corrections." Analysis of this hand reveals the same identifying features found in the script used by Scribe B. A comparison of a sample of the "third scribe's" hand with that of Scribe B shows a clear

correspondence between graphs, for example Th at the beginning of a line (fig. 4, l. 11: "This"/fig. 3. l. 13: "They"), w (fig. 4, l. 1: "warne"/fig. 3, l. 2: "who" "wolde"), and v (fig. 4, l. 3: "lovyst"/fig. 3, l. 4: "devyse"). A further comparison between Scribe B's hand and the corrections Manly and Rickert state were "probably made" by scribe 3 confirms the presence of the same identifying characteristics (fig. 3, l. 3: "her"; l. 2: "who"/fig. 7, marginal correction: "wher").

A comparison of the inks used provides additional confirmation that Manly and Rickert's scribes 2 and 3 were in fact one scribe, Scribe B. Scribe A uses what is most likely an iron paste base for his ink, which is darker in some places where a fresh batch results in a higher concentration of particulate matter. The usual color of this ink in the MS is a russet brown. This ink has the quality of producing a wash-like effect (gradations of color within individual strokes). Scribe B uses a carbon base ink from fol. 175 verso, where he takes over from A, through fol. 177, where he switches to an iron paste base. The carbon base ink produces a much more uniform stroke (not wash-like) and ranges from black to a brownish-gray, depending on the consistency of the batch. This same carbon base ink is used by the scribe who copied fol. 109 and made some corrections in A's portion of the MS. With the knowledge that Scribe B copied fol. 109 as well as fols. 175 verso through 245 verso, we can now also speculate that he executed fol. 109 shortly before he began work on fols. 175 through 177, using the same batch of ink.[14]

It is also now clear from this paleographical analysis that Scribe A was responsible for fols. 246 through 249, which Manly and Rickert had stated was only "probable" and which their description seems to indicate could have been copied by either of the two main scribes. It was largely the difference in letter size which contributed to their confusion on this point, and since both Scribe B and Scribe A are using an iron paste base in their ink at this point in the MS (where A resumes copying) it is not possible to differentiate on the basis of ink analysis. However, despite the difference in letter size (an average body height of 2 mm in fol. 246 compared to an average of 2.5 to 3 mm elsewhere in the MS, except in some of the prose sections where the letter size approximates that in fol. 246) the identifying characteristics of Scribe A's script are clearly present in this disputed section.

Thus far I have shown that there are only two scribes copying in Cardigan, Scribes A and B, rather than the three scribes proposed by Manly and Rickert and assumed by others who have faithfully accepted Manly and Rickert's analysis. We have also seen that the scribe who copied the first four folios of the Lydgate material (fols. 246-249) was indeed Scribe A and that some of the corrections made in portions of A's work (i.e. those which A did not make himself) were made by Scribe B. A more detailed examination of the codicological evidence will establish that these two scribes were also the "supervisor" or "examinatur" whose traces led Manly and Rickert to assume that this MS was produced in a shop. It is clear from their description of this evidence

that they distinguished the person who left these traces of supervision from the scribes:

We already know that Scribe B made some corrections in portions of A's work, using black ink. Other corrections in black ink can be attributed to Scribe A by making comparisons similar to those above.

Manly and Rickert's assertion that the "ex'" notation does not occur with any "regularity" must also be challenged. The notation can be found at the end of Qq 2, 3, 4, 5, 26, 27, and 28 (fols. 18v, 28v, 40v, 52v, 268v, 280v, and 292v). In addition, the notation can be found within quires, usually beside corrections. Manly and Rickert observe this in fols. 47v (Q 5), 69r (Q 7), 212r (Q 20), 214r (Q 21), 248v (Q 25 [see fig. 2]; they are unsure of this one, but it is there), 278r (Q 27), and 289r (Q 28). I have confirmed four additional occurrences not previously noted by Manly and Rickert in fols. 219v (Q 21), 258r (Q 26), 264v (Q 26), and 268v (previously noted above). The "ex'" can thus be found thirteen times in Scribe A's portion of the MS and three times in Scribe B's. The long hiatus between Q 7 and Q 20 interrupts the fairly regular pattern where we find the notation on the last verso of each quire. The regularity is not reestablished until Scribe A resumes copying in the Lydgate portion of the MS. This is a puzzle which can, I believe, be satisfactorily unravelled, and the first step in this process involves a comparison of the "ex'" notation with the hand of the scribe who is copying at those points in the MS where it is found. In each case, the characteristics of ductus and graph formation make an identification with the scribe possible. The usual position of the notation can be seen in fol. 28 verso (fig. 5) where it occurs adjacent to the final catchword of the quire. A comparison of the notation with the first two letters of "exortacyon" (fig. 2, l. 3) reveals a clear correspondence between the graphs for x: a bold downward stroke, slashing from left to right, crossed by a rounded 2-shaped stroke which has an elongated tail crossing the initial stroke at the bottom from left to right. The angle of the initial, circular e's differs, but an e written at the same angle can be found in the word "one" (fig. 2, l. 6), and a quick inspection of the variety of circular e's appearing in fig. 2 confirms that the angle at which the letter is placed is not at all regular.

While there are fewer examples of the notation occurring in Scribe B's portion of the MS, a comparison of the x graph of the notation which appears in the gutter of fol. 212 recto (fig. 6) with the x's which occur in fig. 3, executed by Scribe B (ll. 5 and 9: "Maximus") shows the similarity of graph formation: a single stroke, crossing downward from left to right to the level of the bottom

(unruled) line, and then curving back upward and left in a very rounded stroke before crossing itself and finishing in a rightward-curving tail. In Qq 10 and 11, where we find evidence of B proofreading A's work, no examples of the notation made by either A or B can be found. Scribe A apparently discontinued that kind of correcting and proofreading (there are still numerous examples of concurrent corrections) following Q 7, only to resume the procedure in the Lydgate portion of the MS. Quite possibly the absence of a notation by A in these intervening quires alerted B to the need to reexamine that material. In Q 10, he must have discovered something which caused him to insert fol. 109 (established above as the work of Scribe B). Manly and Rickert propose that in The first problem with this analysis is that it is unlikely that a scribe would cut away a perfectly good leaf for the sake of uniformity, which, as Manly and Rickert observe above, is not rigidly maintained anyway. A second problem is that if the scribe's exemplar had lost a folio, why was he able to predict that the first words of that folio would be "I warne the," which he has signalled by the catchword placed at the bottom of fol. 108 verso? Manly and Rickert state that the placing of catchwords on the last 6 folios of a 12-leaf quire is "unusual" (I, 74), so we cannot easily ascribe this knowledge of missing material to a faithful copying of the exemplar.

• Diagram A: The state that Keiser proposes the supervisor found Q 10 to be in with the i-xii bifolium having been misplaced after the text was copied onto i, but before it was copied onto xii. Thus xii became x and in moving i back to its correct position in the gathering "the order vii-viii-ix-xi-xii-x would have been created."
• Diagram B: The present ordering of Q 10. The original xii has been excised, leaving
  
  a large stub between fols. 111 and 112. The stub between fols. 102 and 103 has, presumably, been left to allow the inserted folio, 109, to be more easily bound into the gathering, the stub being part of the sheet which forms fol. 109.

George Keiser offers a different explanation for the appearance of Q 10. Diagram A illustrates Keiser's version of the disarrangement of the quire. He believes that a "supervisor" discovered that "the outer bifolium was accidentally moved so that the order became ii-iii-i-iv-v-vi. (Such a disarrangement would be easily possible in view of the fact that there are no catchwords used in the first part of the quires in this manuscript)." Keiser goes on to say that the proofreader then "discovered the disarrangement and moved i back to its proper position, thereby creating the order vii-viii-ix-xi-xii-x in the second part of the quire. To remedy this problem, he cut away the right half of the i-x bifolium near the inner vertical margin line of x, leaving a stub to facilitate binding of the remaining leaf (f. 100) with the quire." Diagram B shows the present state of the gathering, with the stub left by the cutting away of the gemel of i (fol. 100). Keiser believes that the supervisor assigned the copying of fol. 109, the inserted leaf, to "another scribe," leaving a stub "to facilitate its binding into the quire." That scribe, as well as the supervisor, was Scribe B, as established above.

Keiser says that the proof that this explanation is correct is provided by ink marks on the stub between fols. 111 and 112 (fig. 9). This stub is the gemel of i, the out-of-place text having been excised. Not only do the ink marks establish that the folio contained text before it was cut away, pace Manly and Rickert, but, according to Keiser, they also establish that the text contained on the missing folio was the same as that found on the present fol. 109. Of the two ink marks, Keiser says that the first is "a symbol regularly used by the scribes to aid the reader by indicating the beginning of a new stanza in works written in stanzaic form, or in other cases, to indicate shifts in subject or speaker." The key here is that if the stub is held against fol. 112 verso it

Keiser's "proof" is problematic. The top ink mark on the stub, a stanza marker, does indeed line up with the tenth line of text on fol. 112 recto. But in fig. 4 we can see that the tenth line of text on fol. 109 recto reads: "And namely vpon the wyffys syde." The line cited by Keiser as the "natural place" for such a marker is in fact the eleventh line. A comparison of the text with that in Manchester, the MS sharing a common exemplar with Cardigan, reveals that such a marker is placed in the margin next to the line (E 1393: "This olde knyght . . ."). But there is also a marker in the margin next to the line which corresponds to the first line on fol. 109 of Cardigan, E 1383: "I warne the yf þou wysely wolt worch." Why this marker does not also appear on the stub is a problem. What is important to this study, however, is that it is at this point in the book that we first find evidence of Scribe B at work, while the evidence of Scribe A proofreading his own work, which we find in the earlier and latter quires of the MS, is lacking.

When Scribe B does enter the project as the main scribe, he does so rather abruptly in fol. 175 verso (fig. 1), in the middle of a page and in the middle of a quire. This intrusion deviates from the usual practice, when multiple scribes were engaged in the production of a single manuscript, of changing over between gatherings. While Scribe A's gatherings are customarily twelves, Q 16 becomes a gathering of eight, which is usual for B. We may never know what happened to Scribe A at this point, but we can speculate that when B entered the project, he may have noted that Qq 8-16 had not been proofread and so began the task himself, perhaps to acquire a "feel" for the plan and execution of the codex. He then found the problem in Q 10, executed fol. 109, and proceeded to make some other corrections in black ink. After finishing Q 16, as well as the rest of the Canterbury Tales through the Retraction, he then appended the chronicle to complete the gathering of eight which makes up Q 24.

The question that must be asked at this point is why Scribe A would have reentered the project in Q 25 (fol. 246 recto, the Siege of Thebes) without proofreading Qq 8-16, instead turning that work over to Scribe B? If the MS was made in a shop, as Manly and Rickert state, then certainly he would have had access to the material. But what if it was not executed in a shop, but was instead "farmed out" to independent craftsmen? Doyle and Parkes (pp. 196-197) cite the failure of various craftsmen to interact as evidence of a "bespoke" book trade. Scribe B left blanks for initial capitals with guide letters in the margins (fols. 176r, 186r, 188v, 194r (twice), 201r, 203r, 213r, 216r, 217r, and 244r) which were never filled in. In a shop, a limner should have been on hand to complete these initials, the only places in Cardigan where any evidence of intended decoration appears. This evidence, coupled with the absence of a "shop supervisor" (the scribes provided their own supervision), points to the manuscript's having been produced by independent craftsmen. We must therefore conclude that Scribe A either abandoned the work or became unable to finish. A similar case is that of Trinity Scribe A, whom Doyle and Parkes (p. 166) point out was replaced by another scribe before the end of a gathering, in the middle of a line of verse. This change, they say, "suggests that A had abandoned his task rather than a transition from one portion of an exemplar to another." If Cardigan A did abandon the job at fol. 175, it would seem most unlikely that he would then, as the present order of the MS suggests, have resumed the project at a later date. But what if the present order does not reflect the original order of execution and/or intention? While the Siege of Thebes does begin a new gathering, there is neither title nor incipit preceding it, as is characteristic in the MS before and after all the Tales and Prologues. An explicit does occur at the end of the poem, just as one would expect, preceding Churl and Bird. Recall also the confusion caused Manly and Rickert by the relatively smaller

letter size of Scribe A's hand at this point, which gradually becomes larger through fol. 249, where it achieves the size characteristic of the rest of A's work in the MS.

There peculiarities—the missing incipit, the change in handwriting size, the peculiar changeover of scribes in fol. 175 verso and again at fol. 246 recto—suggest that Scribe A began his work on the MS with the Lydgate material, writing in a smaller hand, gradually acquiring a feel for the constraints of the writing space (where there are space limitations, as in some of the prose tales, both Scribe A and B write much smaller). Then after finishing the Lydgate material, Scribe A began work on the Canterbury Tales, continuing until something interrupted his work on fol. 175 verso.

The codicological evidence is inconclusive. The sheet of parchment which forms the outer folios of Q 24 (fols. 238 and 245) is a full thickness of sheepskin, while the Cardigan scribes write primarily on sheets of split sheepskin. The splits result in two types of hair sides (an inner and an outer) as well as two types of flesh sides (an inner and an outer), making it difficult at times to determine whether a given page is hair or flesh (Yeager). While at least two other gatherings, Qq 6 and 27, are also enclosed by full-skin sheets (comprising fols. 53 and 64, and fols. 269 and 280 respectively), Q 24 also has several sheets of full skin within the quire. This may suggest that the last quire of Scribe B's work was especially reinforced as an intended final gathering of the codex, which would be subject to greater wear and strain.

The present final gathering (Q 30), consisting of most of Churl and Bird, which begins on the recto of the final folio of the preceding quire, was apparently intended to be a gathering of six, since there is a catchword on iv (fol. 308v) and Scribe A customarily places catchwords on all the versos of the last half of a gathering. One can also see sewing between fols. 307 and 308. Manly and Rickert, overlooking this evidence, suggest that it may have been an "original 12" (I, 71). Churl and Bird would require only part of one more folio, and thus what the remaining sheets contained, if anything, remains a mystery. The fact that the gathering was intended to be one of six rather than twelve indicates that the scribe either anticipated the end of the MS or desired to keep the material discrete from whatever was to follow. Even if the material was intended to follow the Chaucer material, it might very well have been copied first, and the popularity of Lydgate in the fifteenth century makes it at least possible that this portion was considered to be more important than the Canterbury Tales.[15]

Another reason why the compiler of Cardigan might have chosen to place the Lydgate material before the Canterbury Tales in the manuscript is the special affinity between the Siege of Thebes and the Knight's Tale: both deal with the Theban story, a story though of today as legendary, but which in the

fifteenth century was often referred to as "historical."[16] Given the inclusion of the chronicle in the codex, it would seem that the compiler had an interest in history, and an ordering which placed the Siege of Thebes before the Knight's Tale would present the story chronologically. Admittedly, though, the intervention of the Churl and Bird does pose a problem for this theory. Besides Cardigan, four other MSS contain the Siege of Thebes and the Knight's Tale: Additional 5140, Christ Church CLII, Egerton 2864, and Longleat 257. Of these, only Longleat 257 is presently ordered to place the Siege of Thebes before the Knight's Tale (there called Arcite and Palamon) (Manly and Rickert, I, 339). In Cardigan, a marginal notation, a flourished n or nota marker, can be found at several points in the Siege of Thebes and the Knight's Tale (and only in those two works) at points of sententia or narratorial intrusion, creating a further bond between the two "historical" tales.[17]

The worn, well-thumbed condition of the first full page of the Knight's Tale, coupled with the worn condition of the first page of the Siege of Thebes, suggests not only that these tales were frequently handled and read, but that they may have been kept separate from the other materials in an unbound state, with the two worn pages serving as "cover-sheets" and with the present, bound ordering of Cardigan reflecting a later owner's sense of priorities. That at least fol. 10 (the first leaf of Q 2 and the first full page of the Knight's Tale) once served as a cover-sheet is strongly suggested by the appearance in the upper right corner of the signature of Thomas Brudenell (cf. note 6).

While none of the pieces of the puzzle discussed above offers a conclusive solution to the problem of the unusual interaction between the scribes or to the problem of the "correct" ordering of Cardigan, taken together they do offer a reasonable explanation for the apparent absence of evidence of regular "supervision" of the scribes' work between Qq 7 and 20, that is, the lack of the "ex'" notation in those quires which does occur with great regularity at the end of gatherings at the beginning and end of the MS. First of all, there was no separate supervisor who proofread the entire MS. This task was undertaken by the scribes themselves, as the paleographical evidence presented above so clearly shows. The paleographical and codicological evidence also

suggest very strongly that Scribe A began his work on the MS with the present fol. 246 (Q 25) and ended on fol. 175 verso of Q 16, apparently unexpectedly. Scribe A carefully proofread his work in Qq 25-28 and 1-7, but not in what must have been the last portion he worked on, Qq 8-16. These portions were proofread by Scribe B when he took over the project, and while they contain evidence of B's editorial and scribal concern, they do not contain evidence of the usual "ex'" notation. But Scribe B exhibits no regular pattern in his use of the symbol and, indeed, it occurs only three times in his portion of the work.

None of this evidence supports Manly and Rickert's assertion that the MS was "certainly made in a shop"; indeed, had the Cardigan MS been executed and compiled in a scriptorium, we would expect to find a regular pattern of evidence of supervision independent from that of the scribes. The evidence suggests instead that independent craftsmen produced the book, that Cardigan was originally intended for one scribe—Scribe A—and that only when he was unable to finish did another scribe begin work on it. I remain hopeful that further work on this and other MSS, especially those containing similar materials, will produce some answers to the question of the intended order of the very important compilation.[18]

Notes