University of Virginia Library

For most of the English Lollard writings we know publicly, we owe a considerable debt to two Victorian editors, Thomas Arnold and F. D. Matthew.[1] Both, in the course of their work, devoted considerable attention to an important codex which collects a wide range of such tracts, Corpus Christi College, Cambridge, MS. 296. And both acknowledged that they were flummoxed by one feature of this book. Matthew remarked, in his headnote to the tract called "Of Feigned Contemplative Life" (hereafter FCL):

And Arnold, who had earlier edited "A Petition" (hereafter PKP), noted the same textual mishandling and commented, "How it occurred, it is now impossible to say."[2]

While I am certain that Matthew was correct in suggesting that the Corpus scribe copied from a misarranged exemplar, the error seems to me particularly significant. I hope, in the course of examining how this misadventure befell the Corpus scribe, to indicate some constraints which underlie his book-production. These, in turn, will allow some consideration of known features of Lollard production generally.

As Arnold and Matthew indicate, Corpus includes full texts of both FCL and PKP, but the texts occur in fused forms. In each case, the brief conclusion of the other text has been consecutively attached to the lengthy head of the tract.[3] In Corpus, FCL begins on p. 165a (sig. k3ra); the text continues, in due order, until line 34, well down the column, of p. 170a (sig. k5va). At that point, which corresponds to EETS 74:195/7, the scribe simply begins copying PKP, at a point which corresponds to Arnold's p. 520/18. The text reads (I insert a solidus at the point of juncture): "Lord, where þes worldly prestis wisere þan / distroied þe false feiþ tauʒte of anticrist and of his false cursede disciplis." The remainder of what is presented as FCL, ending on p. 172a (sig. k6va), merely transcribes the concluding portion of PKP.

The text of PKP occurs at the very end of the codex, beginning on p. 288a (sig. s9va). At just the precise point in that tract where the final pages (already copied in the manuscript) should occur, the scribe ceases to copy PKP and instead finishes the text with that portion of FCL that he should have included earlier. This juncture occurs at p. 297b (sig. t4rb), line 20; the text reads: "by cursed ypocritis and heretikis and worldly prestis vnkunnynge in goddes lawe / ben alle þe apostlis of crist." The remainder of the manuscript, which ends at p. 298b (sig. t4vb), line 12, gives the conclusion of FCL, the portion for which the scribe substituted the conclusion of PKP earlier in the codex.

At this point, I want to draw attention to several features of the Corpus text which neither Arnold nor Matthew noted. They were interested, after all, in making texts available, not in codicological examination. But there appears to me ample evidence on which to argue that the Corpus scribe never planned to provide a text of PKP. A number of manuscript features indicates that the text has been added to an otherwise complete and consecutively produced book.

The quiring of Corpus 296 at this point suggests that PKP may be a late supply. The entire manuscript, except for the last two quires, was produced in regular eights. But the penultimate quire s (pp. 271-290) is a ten, and the final quire t (pp. 290-298) a four. This particular confirmation of quires looks to me suspiciously like what I have elsewhere described as "a superseded fascicle boundary."[4] The Corpus scribe saw that the text with which he intended to end his copying could not be completed in a normal eight-leaf quire and constructed a ten-leaf one instead. This unique text, "Þe grete sentence of curs expouned," begins at p. 239a and ends on p. 288a (sig.

s9va). Had the scribe copied it straight through in his usual eights, he would have ended quire s at p. 286 and then would have finished transcribing the work on a loose singleton. Like a great many medieval book-producers, he rejected this solution—small quires, singletons, and bifolia frequently get misplaced, as we will see; his solution, to produce an extra-long quire to accommodate text runover, is a commonplace one.

But the concluding pages of quire s and all those of quire t, which bear PKP, look markedly different from all earlier sections of the volume. I do not think the text was copied by the same scribe responsible for the rest: the hand is not so carefully formed and shows different habits of letter-formation and -distribution. And especially in the final quire of the book, the page format differs in slight ways from the remainder. The main Corpus scribe has copied within a regular forty-line frame for something in excess of 140 folios; here, as the writing sprawls out, the text is typically presented in a 37- or 38-line column.

When what I believe a second scribe decided to copy PKP, he practiced economies. Rather than starting a new quire, he simply began to add his text of PKP on the blank pages left at the end of the heretofore concluding ten-leaf quire s. The text was longer than the blank portions of this existing quire, yet not so long as to require an additional full eight; the short quire t, in which this scribe concluded his work on the final verso, proved just right for his purposes and conserved parchment. But these decisions indicate that the plan for the book changed at some point very near the end of production and that, with the exception of the odd bit he had inadvertently copied earlier, the main scribe had apparently not planned to copy PKP.

This interpretation of the codicological evidence, of course, immediately raises problems. The materials the second scribe eventually copied at the end of the codex, given the exactness of the textual fits, must have been derived from the same exemplar from which the main scribe had already transcribed a portion of PKP in apparently the wrong place. Thus, the evidence suggests that the scribes responsible for Corpus might simultaneously have and yet not have access to portions of that single exemplar containing two of their texts. From the absence of the full text, one might surmise that the main scribe initially did not make use of this portion of the exemplar at all, perhaps for editorial reasons, perhaps because it was unavailable to him. Yet his colleague could access a portion of the same exemplar, presumptively contiguous, when he later decided to extend the apparently completed codex.

At this point, I want to add a further complicating detail, which should have occurred to both Arnold and Matthew. For Corpus 296 is not, for much of its course, a unique manuscript: it has a partial twin, another early fifteenth-century manuscript, Trinity College, Dublin, MS. 244 (C.3.12). The evidence Trinity provides is particularly disconcerting, for, excepting some leaves now lost which contained the conclusion of the work, Trinity includes all of PKP in one place. Moreover, that codex includes its version of the text at a provocative point: after providing the complete tract FCL with its sections in proper order (ff. 136-141v), it originally contained a full version

of PKP (ff. 141v-148v, after which a leaf and a full quire have been lost).[5] Even in this truncated form, the textual presentation in Trinity suggests that the exemplar had full and continuous texts of FCL and PKP as adjacent units. Thus, the fragmentary conclusion of PKP in Corpus has been placed correctly (or at least, in the intended position in their common exemplar); what appears, from this perspective, the after-the-fact provision of the opening sections of this text in Corpus is that manuscript's aberrancy.

So I return to Arnold and Matthew's befuddlement. What kind of error is this? How could such an exchange of textual materials have occurred?

I want first to note the nature of those textual portions exchanged in Corpus 296. In both cases, the misrepresentation of the texts is discrete and affects only these texts while respecting their boundaries, their junctions with adjacent items in the manuscript. Moreover, only the ends of the tracts have been exchanged. On this basis, it seems reasonable to suppose that, whatever else is true of the exemplar behind Trinity and Corpus, it was not, for some portions at least, a continuous manuscript, in the sense that each text was immediately followed by its successor on the same leaf. Had texts appeared in this consecutive fashion in the exemplar, one would hardly expect to find the exchange of material between the two tracts text-bounded: one might reasonably assume that some portions of subsequent texts, included on the same leaves as the conclusions of FCL and PKP, would also show dislocation. But the order and extent of texts in Trinity, otherwise identical to Corpus so long as the manuscripts are true twins, insure that no such dislocation has occurred.

If one ignores the exchange of passages, Corpus has accurate textual boundaries. If pieces of other texts have been dislocated, the main Corpus scribe has corrected these failures and reestablished accurate textual boundaries. But if he was capable of such correction, he should have discovered his initial mistake, and might well be expected to have amended it as well. One should conclude, I think, that the textual exchange I describe occurred through inadvertence (Matthew's "happy unconsciousness") and that it reflected properties of the lost exemplar.

Further, identifying one of these properties as the discontinuous presentation of texts places the lost exemplar in a well-recognized class of codices. It gathered together a set of fascicles or of "booklets," short self-contained groups of quires which formed a whole only by aggregation. Given the nearly universal agreement of Corpus and Trinity in contents and order for a very protracted period (224 pages in Corpus, 186 folios, discounting losses, in Trinity), one must assume that both scribes acquired this bunch of fascicles in some relatively fixed form. But the state of PKP in Corpus is telling: it represents, as I have suggested above, a separate piece of production, derived from an apparently separable textual unit. This feature supposes that the exemplar was capable either of being disbound or of shuffling of textual portions; in any event, the booklet which included PKP was somehow mobile, perhaps even detachable from the codex at large (and thus added only belatedly by the second Corpus scribe).

In an effort at specifying how the endings of texts might become exchanged in Corpus, I want to consider the size of the textual portions mishandled. In actual terms, the bit of PKP on pp. 170-172 runs for nearly four columns, 152 lines in the scribe's format here. The isolated piece of FCL is roughly half that size, nearly two columns, 68 lines in the second scribe's comparable format in the added quire at the end. If the extent of the textual fragments is considered in terms of the full texts of the works, just about one seventh of FCL (forty-one lines in Matthew's text out of 279 total) appears apart from the body of the text. For PKP the corresponding figures differ: something over one sixth of the text (eighty lines in Arnold's text out of 494 total) has been copied out of sequence.

In trying to visualize the exemplar behind both manuscripts, PKP provides the most useful place to begin. Given its placement in Corpus 296, this text must have begun an independent fascicle. From this fact, and from the relative extents of the separate portions in which the text was copied, one can make some rough estimates about the form of this textual unit. The two separated bits of copy include the whole text; the shorter piece essentially represents one sixth of the whole. On this basis, as a separate fascicle, the text should have been divided between production units related to one another in a ratio of 5:1: the simplest hypothesis would postulate a text divided between a ten-leaf quire and a continuation in a single bifolium.

In passing, one should note that in this minimal size for the units involved—any multiple which maintained a five-to-one ratio is theoretically possible—the folios of the exemplar would have been roughly comparable in textual volume to the pages in the surviving copy, Corpus 296. In fact, a modest amount of evidence implies that such a minimal unit was in fact at issue here.

Such evidence that a folio of the exemplar bore approximately the same amount of text as a page in Corpus emerges from one shift of scribal stints in Trinity. The manuscript was written by two or three scribes, both or all involved in copying materials from the exemplar shared with Corpus. At the end of f. 91v, the scribe responsible for the entire text to this point apparently gave a segment of the exemplar to a colleague: this step was probably undertaken to expedite copying through allowing simultaneous work in two hands. Scribe 1 resumed his work at the end of the segment he had given scribe 2, and each left his partner, on the page where the other was to begin his stint, a partial sentence to indicate where copying was to start.

The section of text copied by scribe 2 was not a fascicle: it includes part or all of three texts, corresponding to EETS 74:112/25-143/22. But it seems likely to have been two eight-leaf quires, or 32 pages. The presumption for such a sizing depends upon the correspondence of this portion to seventeen plus pages in Corpus (pp. 106a, line 13-123b, line 9). Since Trinity scribe 2 did not include almost a column of Latin unique to Corpus (pp. 106b-107a) and arguably not part of his split examplar, the textual block he was copying corresponds to slightly more than sixteen pages of Corpus. (Scribe 2 nonetheless miscalculated the number of leaves necessary to hold the text, and on the

final leaf of his stint had to reduce his writing to about half its normal size so as to match the point at which his partner's copying resumed.)

This sizing of portions of PKP also allows some inferences about the exemplar's presentation of FCL. If exchanged portions of PKP occupied most of two full leaves in the exemplar, the exchanged portions of the other text, less than half the length, would have taken up less than a full leaf. However, they should have been extensive enough to cover more than just the recto of this leaf. One can then imagine that, in the minimum possible proportions, the displaced part of FCL was written on a single folio, the concluding one in its fascicle.

However, one can only with difficulty speculate about the shape of that fascicle. PKP stood at the head of its textual unit, but one may not be similarly assured about FCL. The text may have been situated at the end of a very lengthy fascicle, and in such a fascicle, it could have begun at the middle of a quire, or of a leaf. But the misplaced section most probably occupied the final leaf of this unit, a leaf which was somehow mobile—perhaps because single and sewn to its predecessors, perhaps because detached from its conjugate. Were FCL to have comprised a single booklet, it could have formed a six-leaf quire with a sewn-on singleton; or it might have been copied into an eight-leaf quire with a cancelled blank eighth leaf. These two situations could produce the alternative hypotheses I have mentioned—a singleton becoming unsewn, a weakened quire structure from which a leaf might become detached. Although I adopt the first of these hypotheses simply as an explanatory procedure in what follows, it remains only a suggestive possibility: one cannot know to the same degree as with PKP the original shape of this portion of the exemplar.

However, this information suffices to explain the textual mismanagement in Corpus 296. One must, first of all, assume that the sensible copying of two full texts in Trinity indicates that this scribe received the exemplar intact, with the texts appropriately ordered. I would schematize this ordering of the exemplar:

     

booklet a	booklet b
. . . FCL	PKP
16 + 1 leaf	110 22

After the Trinity scribe had used this textual source, but prior to the activity of the main Corpus scribe, the exemplar had been subjected to displacement. The late provision of PKP in Corpus may indicate that, in addition to displacement, minimal separation of contents had occurred. I would schematize the form in which the Corpus scribes received the exemplar:      

booklet a1	booklet b2 . . .	booklet b1 + a2
. . . FCL	PKP	PKP + FCL
16	22	110 + 1 leaf

(I insert an ellipsis simply to indicate the removal of materials from their original station: I do not believe the nature of the ellipsis can be specified, and the removed materials may have had no set place within the exemplar.) The loose concluding leaf of FCL had become detached from the quire to which it properly belonged (booklet a, quire 1). Disbound, it may have been inserted into booklet b, quire 1, travelled with that quire as it was moved from its earlier placement in the exemplar (the form as available to Trinity), and (as an obviously final leaf, because partially blank) removed to a position following that quire as the apparent conclusion of the now atelous PKP. Moreover, this movement of the quire containing the greater portion of PKP left the bifolium bearing the end of that text to stand as the apparent conclusion to the preceding, now also atelous text, FCL.

This reconstruction seems to me more likely than an apparently simpler alternative Anne Hudson and David Vander Meulen, who have read this article in draft, have suggested—the exchanged texts disposed in two consecutive full quires.[6] In such a view, the first quire after one comprising most of FCL would have contained the end of FCL (f. 1) and head of PKP (ff. 2-8), and the following one the end of PKP (ff. 1-2) and subsequent materials (ff. 3-8). The initial anomaly—the conjunction of the start of FCL and the end of PKP—might then have resulted from the misplacement of the middle quire. But in such a view, the textual dislocation I describe becomes more, not less, difficult to comprehend, inasmuch as it would require two acts of motivated scribal intervention.

In this account, one must assume that the first Corpus scribe chose deliberately to ignore codicological signals in his exemplar. In its unbound state, proper ordering of the exemplar would have been insured by either conventional catchwords at quire endings, clear quire signatures, or both. For the scribe to miss out a full quire implies that he did not heed either noncongruence of catchword and subsequent text, or non-consecutive quire signatures, or both.

Moreover, the disposition of texts in Corpus must, in this account, presume a further set of erratic activities by the second scribe. Having acquired the intact quire 1 and having determined to copy it, this scribe would have had to ignore the construction of the quire (as well as any signatures) and to have deliberately refolded it, to make its current f. 1 its f. 8 so as to provide a clear conclusion to the text of PKP. Further, unless the conclusion of FCL was entirely page-bound (limited to quire 1, f. 1, with no runover and with PKP beginning at the head of f. 2), the scribe should have been acutely aware that he was tampering in a way which did not respect the exemplar at all. This seems particularly unlikely, since his sole reason for adding materials would have been fidelity to the exemplar, a sense that the first Corpus scribe had not reproduced it in its appropriate fullness. Such a concatenation of erratic procedures is possible, but seems to me inherently less probable than the booklet situation which I argue here.

This reconstruction allows two conclusions in passing. First, both Arnold and Matthew misdescribed the textual disruptions of Corpus 296. Only metaphorically did anything like a "transfer" or "transposition" of texts occur, and the main scribe in fact copied the conclusion of PKP precisely where he should

have done, in the position it always, within our knowledge, had in the exemplar. Second, both Corpus scribes seem relatively blameless figures in the procedure: the first palpably did not recognize that his exemplar had been disrupted by removal of pieces of text, and both copied faithfully and straight on the apparently consecutive materials they had at hand.[7]

One might wonder why such a disruption occurred at all and especially why, in the period between these two copyings of this exemplar, a piece of that source text might have been separated from the remainder. Such considerations will move the discussion beyond the narrower topic of the precise construction of Corpus and Trinity toward a social siting of this variety of book-production. Although there exists no evidence to support the supposition (these are the unique copies of PKP),[8] the first quire of PKP may have been removed for a separate copying. And this copying may have occurred in a locale removed from the central place where the body of the exemplar was retained (and where it was available, more or less whole, to the group of scribes responsible for Corpus and Trinity).

This exemplar may have been produced in booklet form for a reason rather different from the usual forces which impel this type of production and perhaps reflects uniquely Lollard modes of production. Most normally, one assumes booklets exist to facilitate production of the text in hand—to take account of difficulty of access to exemplars, for example.[9] But in the case of the Corpus/Trinity exemplar, which contained illegal texts, heretical ones, and which may consequently have had to be copied in quasi-clandestine circumstances, producing the exemplar in booklet form may have been a deliberate effort at facilitating subsequent book-production. Corpus may initially have lost access to the first quire of PKP because it was on loan, out to be copied—not as part of a production which necessarily involved other portions of the exemplar volume, but perhaps as a unique text. In effect, the fascicular exemplar may have constituted a separable lending library—a series of smallish booklets, held in relatively fixed order in a central location, but available in broken form to a larger reading community.

Two different considerations lend some credence to this argument. First, as I have indicated above, although twins for much of their length, Corpus and Trinity do diverge. Their identical contents originally included twenty-six items (some no longer extant in Trinity, owing to textual damage). At this point, the main booklet of Trinity concludes with a brief filler item (apparently both acephalic and atelous); what is perhaps the same scribe (but in any event, certainly one involved elsewhere in the portion of Trinity copied from the common exemplar) then adds another booklet, comprised, with a minor exception, of unique Lollard texts.

Following p. 224b (but preceding the late supply of PKP at the end), Corpus offers a different series of Wycliffite texts, all but one unique. Conceivably this continuation represents the end of the common exemplar: Trinity either may have chosen not to reproduce these portions or may have lacked access to the whole. Such a view would explain the second Corpus scribe's decision to supply PKP: he may have felt that the manuscript in

production was intended as a complete rendition of the exemplar, checked off rubrics against the booklets at hand, and extended the manuscript on discovering it was short an item.

Whatever the case, both groups of scribes show a substantial interest in gathering large groups of Lollard materials. Beyond copying a huge sequence of heretical documents intact, they also appear capable of exercising independent initiative: they are able to acquire a rich assortment of items similar to, but probably exclusive of, their prepared exemplar. One could well see both groups as attached to some center of Lollard copying and, through that association, in a borrower-lender relationship with other Lollard writers and copyists.

Moreover, both manuscripts inscribe centralized cooperation. In sum, we are discussing four or five different scribes who alternate and finish one another's work. (If there is a third Trinity scribe, he succeeds scribe 1 in mid-leaf and mid-text on f. 111v.) And all had access to a partially fixed canon in a common exemplar. Both manuscripts stand as monuments to an organized cooperative form of book-production with, at the least, links to a single textual source, and thus center.

To approach the second consideration I have mentioned, I want to examine briefly the one non-unique text included in Corpus but lacking in Trinity. This fairly popular work, "Of weddid men and wifis," represents a rather mild form of Lollardy and, indeed, appears in at least one manuscript well removed from sectarian taint, Westminster School 3. Moreover, only one of the other six manuscripts which contain the work shares further contents with Corpus. If one wonders about the mode by which such a text escapes its original sectarian circulation to enter a wider religious culture, at least one possible answer is that it circulated alone, without sign of its attachment to particularistic views. Such separable circulation would have been facilitated by the existence of the text in loose booklet form, and in fact, "Of weddid men" would fit neatly within a ten-leaf quire of the same format as PKP.

But one need not simply speculate about such circulation of bits from the Corpus/Trinity exemplar. For isolated manuscripts share with the two full copies groups of texts, reproduced in the exemplar order. Bodley 647, for example, contains a sequence, found in both manuscripts, of three antifraternal texts (Corpus, items 3-5, pp. 29b-65a). Similarly, Bodley 540 provides two texts which appear consecutively in both manuscripts (Corpus, items 12-13, pp. 145b-157b).[10] And Bodley 938, one of the codices providing "Of weddid men," shares with Corpus and Trinity, although in a different order, three other texts (Corpus, items 14, 15, and 19, pp. 157b-165a, 175b-79a-very interestingly surrounding but skirting FCL and PKP). Such fragmentary, but usually ordered, parallels look as if they reflect access to individual constituent booklets of the Corpus/Trinity exemplar, as if that codex most normally circulated not as the whole, but the fragment.

Moreover, such evidence of partial circulation, by putatively marking off free-circulating portions of the exemplar, can be used to identify other, perhaps separable blocks of text. Although they achieved no circulation visible

today, one might tentatively suggest these as booklets in the lost exemplar. Thus the contents which Bodley 647 and Bodley 540 share with Corpus and Trinity bracket an extensive group of items common to these two manuscripts but not reproduced elsewhere. These would appear, on the basis of subject matter (prelates, priests) potentially divisible into two groups, each arguably the material of a fascicle (Corpus, item 6, pp. 65a-103b; items 7-11, pp. 103b-145b).

And even the unique contents of Corpus and Trinity are occasionally susceptible of similar analysis. In the original conclusion to Corpus, before the scribe regained access to the full PKP, the lengthy unique text, "Þe grete sentence of curs expouned" (pp. 239a-288a) at least appears potentially a candidate for fascicular circulation. Of the preceding texts not shared with Trinity, I have already suggested that "Of weddid men" may have achieved a relatively wide circulation through a separate booklet. The other two texts in this portion, both unique, are extremely brief and may represent filler copied while awaiting arrival of the much more substantial work.

This discussion should help fill in features which have been studied on several occasions by the modern doctor evangelicus of Lollardy, Anne Hudson.[11] Hudson's extensive survey of Lollard trial records has drawn attention to the variety of Lollard book-production: the bishops who sought to extirpate the heresy referred to three specific forms in which such writings were found—"schedulae," "quaterni," and "libri." Ample references to bills or schedules (small parchment sheets or rolls) occur, and full books, like Corpus and Trinity, survive with very great frequency. But to date, Hudson seems to have found but a single "quaternus," a dialogue in a Cosin manuscript at Durham University Library. My discussion should suggest that the bishops' distinction between "libri" and "quaterni" may have been factitious, as is, in many circumstances, any modern distinction between quires and fascicles. Production by such fascicles, comprised of a single "quaternus" or multiple "quaterni," is inscribed in two well-known Lollard codices—and there it was very likely undertaken in an effort at generating yet additional "quaterni" for consumption by a partisan audience.

Such production might induce us to examine more thoroughly book-production associated with the origins of Lollardy, the Oxford milieu.[12] For the behavior I have been describing—the use of fascicular exemplars to facilitate multiple and selective copyings of texts deemed canonical in an at least partially dispersed situation—smacks of a procedure one can well imagine in a university environment. In such a situation, the persistent need for numerous copies of various canonical texts should have produced such a system. And given standard medieval complaints about the "penuria studencium," such a production system could well have incorporated exemplar dispersal as well: affordability might best be served by a rule of "each student, his own scribe," with copying proceeding by textual portions lent out to transcribers/readers/owners. But the production-system which underlies Corpus 296 and Trinity 244 differs in one appreciable manner from any putative Oxonian forebearer. The notion of canon at work here remains, for the most

part, unfixed: the two Lollard codices, and related manuscripts, testify to an allowed audience selectivity, a possible appeal to a topical interest—that very individuality which underlies heterodox belief itself.