University of Virginia Library

Among the most important recent contributions to codicology, the descriptive bibliography of manuscript books, has been the identification and discussion of a unit called the "booklet" or "fascicle." This unit, associated particularly with the work of P. R. Robinson,[1] may be described as a group

of leaves forming at least one quire, but more likely several, and presenting a self-contained group of texts. Booklets thus form units intermediate in extent between the quire and the full codex. In Robinson's handling, the booklet has proven a powerful analytical tool; I offer the following qualifications and extensions of her views in an effort at achieving an instrument at once more precise and flexible.

An initial problem has been created by Robinson's attempt, perfectly comprehensible in her effort to elaborate her findings, to show booklets in a variety of manuscript contexts. However, this extension of the concept has sometimes, I think, obscured important distinctions. Robinson, especially in her important contribution to Codicologica, conflates different bibliographical situations as if they reflected the same forms of fascicular production.

Two major classes of booklet manuscripts, one of these divisible into two subclasses, should, I think, be distinguished. One can consider booklets from two perspectives—that of the vendor or owner of books and that of the producer of books.[2] From the first perspective, a booklet basically exists as an object found at a stationer's or a bookseller's; it presents a desired text, and can be purchased, perhaps to be joined with other texts in a "composite manuscript." Here one would, I think, want to consider two possible subclasses distinguished on a basis purely historical. On the one hand, some such "composite manuscripts" have been joined together by private citizens, rather than persons involved in book-production, during the Middle Ages; others, in contrast, might be considered binding accidents, collections acquired after the Middle Ages and bound together for reasons of convenience alone.[3] The first subclass might include a variety of products, perhaps emanating from the same scribe or scriptorium: potentially, such a volume may be indistinguishable from the second large class of fascicular manuscripts.

From this second perspective, a booklet is an object produced. Rather than merely something found or desired by a purchaser, it is an object intended, and perhaps one intended to be joined with other booklets in the same or a similar format. Booklets provided medieval booksellers, who typically produced works to order in what is considered a "bespoke trade,"[4] a way to have some ready stock, especially of popular texts, without the major investment inherent in producing a full codex "on spec." Booklets further

provided a useful way of building up such large codices out of a series of small sections—a procedure which may have been useful both in terms of marketing and in terms of flexibly planning out the production work. This perspective on the booklet seems to me potentially more interesting than an emphasis on composite texts (which are primarily problems of cataloguing): rather than what may have been largely accidental collocations conceived by a purchaser or an owner several centuries after the fact, this view of the booklet involves actual bibliographical interest in production itself.

These two perspectives—the booklet as purchased/the booklet as produced—need to be clearly distinguished. For only the second is of real importance for a pure descriptive bibliography. In the first instance, one cannot with any safety talk about anything other than the taste of the collector of a composite manuscript or the diverse paths of transmission along which the pieces of the book may have passed. Only in the second instance does one deal with the real object of descriptive bibliography, a work with a single mode of production which must, in its more or less problematic nature, be defined. For this reason, it is especially important to distinguish with booklet texts of medieval provenance whether these have been conceived and produced as single volumes or merely reflect the eclectic tastes of an owner.

Robinson defines the booklet in the context of other codicological units, such as the quire and the pecia. A booklet is a textual unit which "could circulate independently and at the same time provide a complete copy of a text." Or again, "the existence of a 'booklet' is established only if its content forms a self-sufficient unit. The beginning and end of a 'booklet' always coincides with the beginning and end of a text or group of texts."[5] The definition is designed to separate this unit from both the quire and the pecia, which frequently present only pieces of texts and cannot have an independent life.

Again, certain refinements seem necessary. These may be suggested by converting Robinson's definition into questions: "Are all self-sufficient textual units in manuscripts 'booklets?'" Or alternatively, "Are all textual units not self-sufficient in manuscript not 'booklets?'" The first question asks whether groups of quires where the first text begins early in quire 1 and the last text concludes near the end of quire n must necessarily form a booklet. This version of the issue is perhaps best deferred until one considers Robinson's procedures for the recognition of a booklet. The second question asks whether booklets are necessarily self-sufficient and independent units. Here, it seems to me, Robinson's definition of the booklet requires some revision: it seems out of touch with some facts of book-production.

Robinson's identification of the booklet directs attention to a fundamental way of forming books in the middle ages, the creation of larger codices from relatively small textual units. Reasons for the prevalence of such a procedure are not far to seek, and I have suggested some of them above. In sum, the booklet occurs with such frequency because of its cheapness and its flexibility: this unit involves a rather minimal commitment of resources while still allowing ongoing book-production. Moreover, the use of the booklet in

production forestalls or delays quite indefinitely any very absolute decisions about the form of the final product. That is, for its producer, the booklet creates an infinitely flexible situation in which the codicological form of the resulting book may never be fixed during production. In this case, the booklet will be sold as a single booklet, and it will thus reflect the producer's interest in marketing. The subsequent fate of such a booklet becomes, as I've already suggested, a function of the purchaser's resources and tastes.

But other scenarios are possible, if not equally or more probable, and these serve to qualify Robinson's definition. In contrast to the situation I've just described, in which the booklet is simply sold as a small limp bunch of bifolia, at various points in production booklets can become contextualized. That is, a plan for the incorporation of several such units into a single codex may emerge. In many, perhaps most, cases, this plan involves no disruption of the booklet format: a group of booklets is taken up and joined end to end, put additively together, into a single codex. These efforts are ambiguous: they may be indistinguishable from a series of separate booklets bought from a single bookseller and joined by the purchaser. And they offer no challenge to Robinson's definition.[6]

But on other occasions, substantial changes may accompany contextualization or codicizing of booklets. In effect, such changes reflect revisions in plan: rather than conceiving of the booklet as a basic unit, the producer begins to conceive of the unit of the whole codex. And when this step occurs, although production proceeds (or has proceeded) within booklet format, the "self-sufficiency" of the booklet is frequently lost. There are, I should think, two models for such a procedure, although both share the same ultimate effect, the obscuring of the existence of booklets altogether. First, booklets of less than a single quire in length can be imbedded into other textual units, typically at their middle, sometimes at their end. Second, larger booklets can be integrated or joined in ways which obscure the original separateness of their production. Such procedures, by readjusting textual linkages, in effect

disrupt the appearance of "self-sufficiency" without changing the fact that the codex has been produced by means of booklets.

A pair of examples will clarify the point. Robert Thornton's miscellany, Lincoln Cathedral 91, contains at least one hidden single-quire booklet. This codex has been recognized as a fascicular production ever since the manuscript was rebound a decade ago.[7] However, recently John J. Thompson has argued persuasively that what has been taken as the second booklet (quires 4-10) of the manuscript was in fact originally two booklets later joined by adding intermediate quires.[8] Originally, quire 9, folded in reverse so that f. 154, now the eleventh sheet of twenty but originally the twelfth of twenty-two, stood at its head, opened a booklet. This textual unit began, on f. 156, with The Awntyrs off Arthure, which ends on f. 161. At a later date, the quire was refolded and its blank second half, now made its head, used to accomodate runover from the intermediate texts copied from the end of quire 6 through quire 8. To join the two sets of texts, some brief filler items were copied into quire 9. And in the new second half of the quire, an additional text was added, beginning at f. 161 and running over to the end of quire 10 (and of the apparent fascicle). As a result of this sequence of actions, the boundaries of the original have been virtually lost in the extant manuscript.[9]

Although one-quire booklets may seem, on an a priori basis, especially useless in the production of manuscripts, the reverse is in fact the case. One's presumption that the single quire would be of minimal aid in preparing a codex is essentially based upon a content-oriented view of production. A single quire quite simply appears a unit of too brief a compass to carry any very viable group of literary texts. Such a view is probably incorrect,[10] and

it overlooks a more persuasive reason for using the one-quire booklet, flexibility in production.

As I've suggested above, a fundamental reason for booklet production is the possibility of delaying any step which would absolutely determine the shape of the resulting codex. The producer of a codex, in a single-quire booklet, possesses a bibliographical unit which can potentially be fitted into nearly any context. First, so long as the quire remains less than half filled, the size of the single-quire booklet can be adjusted: the scribe has only to fold additional sheets and insert them at the center of the quire to expand his writing space. Even at a later date, booklet size may be expanded by "quiring" the extant booklet within other bifolia. With equal ease, the booklet can be reduced in size. Obviously, the scribe may either remove leaves from the center of the quire, or, if he has begun by leaving one or more outer bifolia as a blank cover to protect his writing, he may reduce booklet size from the outside. And the single-quire booklet can either be utilized "as is"—inserted into the codex alone or at the head of a larger booklet of variable length—or it can be reversed, folded inside out, for use in the middle or at the end of a larger unit. But in this last case, the integrity of the booklet may apparently be lost within the codex which ultimately results.

Other processes which seem of tolerably frequent occurrence also render Robinson's effort to associate the booklet with a clear textual unit moot. One might, in this regard, consider the case of Lambeth Palace 491, a manuscript produced by a scribe who elsewhere utilizes booklets.[11] The Lambeth text certainly contains at least parts of two booklets, the materials of quire 21 (ff. 275-290) being the fragmentary opening of a second textual unit. But somewhat less evident is a further booklet division, between lines 632 and 633 of The Siege of Jerusalem, at the end of quire 16 (f. 216v).

This juncture of two originally separate units may be recognized by four features—excessive wear on some leaves of quire 16, some difficulties of textual presentation within the quire, a change in the signing of the manuscript, and a change in materials at the start of quire 17. First, the scribe finished a copy of the prose Brut on f. 205v: this leaf and the facing f. 206r are more heavily worn than the outside leaves of the quire, a fact which suggests that the quire was left for some time folded open to the last copied text page. Some time lapse, apparently fairly protracted, separated copying of ff. 205v and 206r.

Second, the disposition of the text of The Siege in ff. 206-216 suggests that this portion was copied independently of the remainder of the poem. The scribe apparently began by copying Siege (and conceivably other materials) into subsequent quires, beginning with line 633. He thus had to

perform some "casting off" procedure, estimating the amount of copy necessary to fill ff. 206-216; but, unfortunately, he underestimated the number of lines the blank leaves could hold. As a result, these folios show an effort at splaying out the text in hopes of filling the too extensive codicological unit. This effort fails: the scribe writes only sixteen text lines on f. 216v, rather than his normal twenty-eight to thirty. In addition, this leaf carries a double juncture with the following quire. Originally, the scribe wrote a catchword "As rathe" in his normal fashion at the foot of the then-blank page; but when this last page ended up short of material, as a sign that the textual unit was complete, even though the leaf remained one-third blank, he added after the last text-line the notation "As rathe in proximo folio."

Two other features support the notion of independent copying. First, quire 17 is written on a different paper stock than is quire 16, a sign that copying may have proceeded only after a period in which the scribe exhausted his original paper stock.[12] Second, the method of signing leaves changes in the course of quire 16. The scribe has two forms of signing. One is a set of predictable page-foot signatures for the first halves of quires only; these are comprised of a quire letter plus leaf number. The second system is made up of two numbers, one for quire and one for leaf, and appears in the upper right corner of all rectos. The first system runs through the entire manuscript down to f. 274 (quires b-v); the second covers only the single text, the prose Brut, and ends at the top of f. 205 (running 2.1-16.5). The variation in form, the loss of the second method of signing, again testifies to the separateness of copying.

Lambeth 491 thus provides another example of codicizing which obscures booklet boundaries. The procedure here might be contrasted with what Robinson presents as the normal booklet procedure, the cancellation of blank leaves at the end of a booklet.[13] In Lambeth, the scribe simply refuses to waste a substantial amount of material and finds an alternative way to fit his booklets together into a sequential text. But in this process, he did begin by considering ff. 1-216 and 217-274 as separate units; only the sloppiness of his efforts at codicizing them, fusing them into a single manuscript text, allows one to see them as originally separate. And that separateness reveals that booklet production, in this case, does not respect the integrity of textual units.[14]

Robinson's effort at defining the booklet then leans too much on the notion of textual self-sufficiency, and not enough on the notion of separately conceived production. In part, this imbalance of presentation should be viewed as not simply pardonable but perhaps almost inevitable. That is, Robinson, in defining the booklet as a unit in a codex must, of course, prove that such a unit in fact exists. Because of this logical necessity, identification of the booklet is crucial: Robinson must work from extant texts and must point out to the reader how to recognize ways in which these texts divide into separable units. In such a context, the evolution of a new codicological tool, the stance of the researcher, seeking examples, rather than of the scribe, producing books for specific purposes, is inescapable. Thus Robinson seems to overemphasize content at the expense of form, and she tends to overlook that feature which I would think most basically distinguishes the booklet from other forms of production, the postponement of any overall plan for a finished book, in some cases until after production has ceased.

Since Robinson is interested in defining a codicological unit, a great deal of her work identifies those features which may aid in establishing the existence of a booklet. In her fullest listing, Robinson provides ten such features, not all of which are necessarily present in any single booklet:[15]

• 1. Variation in size of leaves in different parts of a manuscript.
• 2. Variation in scribal hand or in page format in different parts of a manuscript.
• 3. Variation in style of decoration or illumination in different parts of a manuscript.
• 4. Absence of catchwords at ends of quires (which may indicate once independent sections of a manuscript).
• 5. Independent sets of quire signatures in different parts of a manuscript.
• 6. Soiled or rubbed outer leaves of a quire.
• 7. Quires formed of varying numbers of leaves in different parts of a manuscript.

---

108




• 8. Variation in size of possible final quires of a textual unit—either an excessively large quire or a quire containing very few leaves so as to exactly accommodate the end of a text.[16]
• 9. Blank leaves at the end of quires, often cut away.
• 10. Short texts, filler, added, sometimes in later hands, in originally blank spaces at the end of quires.

To these features, one can add three others, one of which has already emerged in previous discussion:

• 11. Variation in the material from which different parts of a manuscript are made: shifts between paper and vellum, shifts (insofar as these are recognizible) among kinds or qualities of vellum, shifts among different paper stocks.
• 12. Variation between sources from which different parts of a manuscript have been copied.
• 13. Variation in subject matter in different parts of a manuscript.

Feature number 12 directs attention to one of the most basic causative forces underlying booklet production. That is, if one adopts the view that the most basic definition of the booklet would insist upon the lack of overall planning of a codex until late in production, one reason for desiring such openendedness is simply the limited availability of manuscript sources. Not all texts are universally available to the compiler. As a result, the literary manuscript becomes produced like a commonplace book, a gathering of materials which meet, for whatever reason, the interests of a compiler. Such products, excessively miscellaneous in their contents, frequently reflect, not a single parent or source copied in its full sequence, but an eclectic melding of works derived from different sources.

This procedure, the creation of the eclectic miscellany, the volume which testifies to individual and idiosyncratic taste, is one goal easily served by the booklet. In this model, the compiler acquires an exemplar containing some desired texts. These he copies off, perhaps supplementing them with additional texts, heretofore unknown to him but present in the exemplar. This material he copies, very likely retaining the contents order of his exemplar, into a series of quires which may be tailored to form a booklet. Acquisition of a second exemplar leads to the production of a second block of material, which again may be self-contained. At some later point, these various booklets may be bound into a single volume, with or without utilizing those procedures I have earlier called codicizing.

In terms of feature 12, verifying apparent booklet limits turns out to be a protracted process. The booklet becomes that manuscript unit most or all

of which is based upon use of a single exemplar or archetypal manuscript.[17] It will be identified, first, by noting other codices with which the booklet shares some or all contents, preferably in a similar order. The approach to relatively definitive proof will require collation of the texts in the booklet with those texts putatively related and identification of which textual version is most proximate. This step only approaches identification, since in most cases, at the very best one is only identifying a text which shares with the booklet a common archetype.[18]

Feature number 13 calls attention to another principle commonly used in organizing a miscellaneous manuscript. That is, rather than group together materials on the basis of source, all desired texts from the same archetype or exemplar being copied in sequence, a compiling scribe may attempt some rough differentiation by literary type or genre. Robert Thornton is perhaps the most well-known compiling scribe who devides booklets in this way: in his Lincoln 91, for example, items of religious instruction, medical recipes, and romances each appear in different booklets. Similarly, in British Library, Additional 31042, the two alliterative dream debates, Winner and Waster and Parliament of the Three Ages, occupy a separate fascicle.

All these features, the ten adduced by Robinson and the three which I add here, share common debilities. For none of them necessarily appears in any given booklet and thus, a priori, the absence of any one neither confirms nor denies the possible presence of a booklet. Moreover, none of the features seems the exclusive property of the booklet as a codicological unit; rather, each equally typifies other forms of manuscript production. These facts create a substantial methodological problem, signalled above by the question, "Are all self-sufficient textual units in manuscript 'booklets'?" That is, ignoring for the moment the issue addressed above of whether booklets are limited only to self-sufficient groups of texts, on finding such a unit how can one be certain that it is in fact a booklet? Can the thirteen criteria mentioned be in any way prioritized? Or, can one identify any necessary criteria for a booklet?

First, some of Robinson's criteria can be pretty much eliminated as necessary to the existence of this codicological unit. Some of these features, those

which reflect Robinson's effort at explaining "composite" manuscripts, are in effect only useful in separating the products put into shape by a purchaser, someone joining a variety of texts with separate genesis. Of the criteria suggested by Robinson, numbers 1 and 3 and number 2 insofar as it refers to scribal hand (but not to page format) may be rejected from consideration rather quickly. To these, one should probably add numbers 4 and 7, not because they typify "composite," rather than "booklet," manuscripts, but because they form so usual a part of medieval scribal behavior that they cannot be taken in isolation as indicative of any specific bibliographical situation. In fact, catchwords at textual boundaries of any sort are infrequent to rare in occurrence, and odd-sized quires occur frequently and unpredictably in codices clearly planned as full units from their inception and written by single scribes without any apparent interruption.[19]

Those criteria which rely on disposition of texts may also be eliminated from consideration. First, if booklets which are textually non self-sufficient occur, some question should arise as to whether disposition of texts can in actuality identify the presence of such a unit. Moreover, the two criteria, numbers 10 and 13 above, which are predicated upon textual contents do not necessarily typify fascicles.[20] Manuscript books of all types can be organized on a subject matter basis; given that some literary productions, lyrics for example, may be by their nature brief, their appearance in no necessary way identifies the conclusion of a textual unit. In fact, some scribes may seek the neatness of beginning, where possible, new major works at the tops of blank rectos or at the head of blank quires; they may thus reserve some leaves to pick up short works without considering or consciously planning their works as fascicles.

Following these eliminations, one is left with a group of criteria which deal with the physical formation of quires. The discussion turns, in this case on simply a logical basis, toward items concerned with the basics of book-production. Left for consideration are categories 2 (so far as it relates to page format), 5, 6, 8, 9, 11, and 12. These can be ranked in terms of the persuasiveness

of the information each offers, the degree to which each necessarily presents evidence for the presence of a fascicle.

In these terms, two of the criteria indicate strongly the probable existence of such a unit; one should consider the remaining features as at best sources of evidence to confirm the first two. Most persuasive, indeed practically definitive, is criterion 5: independent systems of signatures in different parts of a manuscript, even in a codex written by a single scribe, if the scribe is careful, argue strenuously for production as units conceived as independent.[21] Criteria 8 and 9 define booklets with somewhat less, but still a high order of, clarity. These features reflect, in fact, two possible responses to a single production problem, the conclusion of a text. Feature 8 directs attention to the use of casting off procedures to conserve copying materials; feature 9, alternatively, to the willingness to retain a consistent quiring format even at the expense of some wastage. But both procedures clearly mark conclusions and thus probable fascicular boundaries.

In this context, the remaining features are of secondary importance. Use of criteria 5, 8, and 9 provides "best evidence" for identifying fascicular boundaries: these limits may be confirmed and supported by recourse to these additional features. Here especially category 12 may be of particular importance: certainly, its inclusion suggests the desirability of rapprochement between the descriptive and analytical bibliography of the manuscript book. One cannot, with propriety, be concerned only with the texts contained in a codex or only with the physical composition of the vehicle for communicating texts. The two studies may properly meet and support one another in the discussion of booklet production.

I hope that this treatment of booklet problems draws attention to a particularly valuable and provocative tool for manuscript study. Robinson's insistence upon the piecemeal quality of a great many literary manuscripts has important implications for literary history, insofar as it suggests the independent circulation of texts in very small chunks. This information, when coupled with the findings of such studies as dialectology, may allow for quite precise localizations of medieval literary works and give depth to our sense of the local literary scene in the later middle ages.