University of Virginia Library

Search this document 


expand section 
expand section 
expand section 
expand section 
collapse section 
expand section3. 
expand section4. 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 

expand section 

Westminster School MS. 3 is a squat and thick volume, in a late fifteenth-or early sixteenth-century binding of brown leather. The codex, written on vellum, contains eighteen main text items, with two later additions; excepting one of these additions, a lyric, the contents are entirely pieces of Middle English religious prose.[1] The volume belongs within a large but amorphous class of Middle English codices which provide one-volume libraries of the religious life for lay use. But some of its more interesting codicological features provide clues to its modes of production, and these may render it interesting within a more general context.

To facilitate an analysis of the production, I begin by presenting a modified description of Westminster 3. At the left of the page, I note the quires which comprise the volume and the number of leaves in each; toward the right, I identify the texts which fill these quires.

quires 1-88   1) "þe Pater Noster of Richard Ermyte" (Jolliffe M.3b, IPMEP 150)[2]  
98   "þe Pater Noster" concluded
2) (ff. 68-72) the Wycliffite commentary on the Ave Maria (Wells Rev. III:58, IPMEP 276)
f. 72v is blank. 
108   3) the so-called "orthodox" commentary on the Decalogue (IPMEP 48)[3]  
118   the Decalogue tract concluded
4) "Beniamin" (i.e., Benjamin Minor, IPMEP 4) 
12-138   "Beniamin" concluded
5) (ff. 103v-104v) "How men schulden be pacient in tribulacioun . . ." (Jolliffe J.8) 
148   6) "How men þat beþ in heele schulde visite seke men" (Jolliffe L.5b, IPMEP 460)


Page 198
7) "þe chartir of heuene," a tract from Pore Caitif (Jolliffe B, IPMEP 166) 
158   "þe chartir" concluded
8) a translated excerpt from ps.-Bonaventura, Meditationes Vitae Christi (IPMEP 22)[4]
9) "How lordis and housbondemen schulden teche . . ." (Jolliffe I.1)
10) a brief tract on love (Jolliffe G.26, IPMEP 155) 
168   the tract on love concluded
11) "A tretis of weddid men and wymmen and of her children also" (Wells Rev. III:26, IPMEP 521) 
178   "A tretis of weddid men" concluded
12) (ff. 132v-135v) "A schort reule of lyf for eche man in general . . ." (Wells Rev. III:28, IPMEP 203) f. 136rv is blank. 
18-198   13) "A noble tretys of maydenhode" (Jolliffe G.16b) 
2010   "Maydenhode" concluded
14) (ff. 153-162v) "Tretys in fyue schort chapitres," the concluding tract from Pore Caitif (Jolliffe B) 
21-228 232   15) "þe ten comaundementis of god"[5] f. 180v is blank; the work of scribe 1 ends here. 
24-268   16) "þe myrour of seynt Edmound," i.e. one of the translations of St. Edmund Rich, Speculum Ecclesie (Wells VI:6 in supplement 3, IPMEP 706, 799, 800)[6]
f. 204v originally mostly blank, now with some added lists (item 17 of Ker's description); the work of scribe 2 ends here. 
27-288   18) Richard Rolle's Form of Living (Wells XI:5, IPMEP 351)[7]  
298   Form of Living concluded
19) Rolle's Ego Dormio (Wells XI:6, IPMEP 160), following the preceding without rubric 
304   Ego Dormio concluded
ff. 231v, 232rv originally blank, now with a lyric (IMEV Supplement 754.5) written in on f. 231v (Ker's item 20). 
The manuscript lacks signatures (three partial ones occur in quire 28) but has catchwords at the ends of quires where the text is consecutive.

On the basis of features identified by Pamela Robinson as typifying production in booklets or fascicles,[8] the text appears to divide into a series of seven separate units:

  • I = quires 1-9 (identifiable by the remaining blank side at the end and by the possibility that the short text 2 has been copied in as a filler)
  • II = quires 10-13 (identifiable by possibility that the brief unique text 5 has been copied in as a filler)
  • III = quires 14-17 (identifiable by the remaining blank leaf at the end)
  • IV = quires 18-20 (identifiable by the extension of the final quire to accommodate text runover)
  • V = quires 21-23 (identifiable by the remaining blank side at the end and by the small final quire to accommodate text runover)
  • VI = quires 24-26 (identifiable by the originally blank final leaf)
  • VII = quires 27-30 (identifiable by the blank leaves at the end)
Further evidence not associated with quiring supports such a view. The booklets reflect the activity of three scribes, all of the early fifteenth century: the


Page 199
first copied five booklets in anglicana formata, the second Booklet VI in bastard anglicana, and the third Booklet VII in textura.[9] In addition, scribe 1's Booklet IV differs in format from the remainder of his copying: within a text area 130 X 90 mm. he writes thirty-two lines per page in Book IV, as opposed to his normal 130 X 80-85 mm. frame containing twenty-four lines.[10]

This information, while it does not exhaust what may be gleaned from a physical inspection,[11] gives the essential data about the preparation of the text. But although it explains the book's production artifactually, it leaves unanswered a variety of questions. Why does Westminster 3 have the shape it does? How did the materials provided by the three separate scribes come to be joined? Why did the first scribe prepare his work in separate fascicular units? And where did all three scribes get their materials? What kinds of sources did they rely upon? Ultimately, these questions about textual production implicate one in questions about basic medieval literary history, the availability of works and the form(s) in which those works circulated. In the remainder of this essay, I want to address these questions briefly, using the tools of analytical bibliography and dialectology.[12]

One set of questions can be answered very easily—at least as regards certain separable parts of the volume. That is, even rather perfunctory investigations will quickly convince one that the materials copied by scribes 2 and 3 are of a highly localizable sort; put otherwise, they reflect two different narrow traditions. Both sets of texts—item 16, The Mirror of St. Edmund; and items 18 and 19, the fused Form of Living and Ego Dormio—turn out to belong to limited recensions of the texts they represent. In the case of The Mirror, one is dealing with a translation not widely attested, at least in its full form; with the Rolle epistles, a peculiar scribal redaction of relatively narrow dispersal.

I would suggest an immediate inference to be drawn from this fact: scribes 2 and 3 each worked on texts of limited circulation—and a circulation quite separate from other portions of Westminster 3. One should, I think, conclude that a substantial cause of the discontinuous structure of the Westminster manuscript reflects simply this fact, that booklets were generated only as exemplars became available. Further, excepting some minor tailoring, not all a matter of economics but some of it an effort at placing thematically related works together, booklet divisions in the text correspond to distinctions of source.

As an example of the kind of study which underlies this abbreviated account, I want to examine the transmission of one of the Rolle epistles in some detail. Booklet VII, as has been previously recognized,[13] provides evidence for a well-known deviant text of Rolle's epistles. In Rawlinson A 389, two copies of the work Ego Dormio occur. One of these (ff. 77-81), published by C. Horstman in the late nineteenth century, introduces numerous variant readings into the apparent Rolle original, as found in Cambridge Univ. Library Dd.v.64. The second copy, found later in Rawlinson (ff. 96v-99) fused to the end of a deviant version of Rolle's Form of Living, plainly comes from


Page 200
a version like the first Rawlinson text; however, it was not transcribed from the first Rawlinson copy. This second text seems to be a verbatim transcript, the foot of each page being marked "corrigitur," a sign that the text has been rechecked against its archetype. Given some substantial variations in the two Rawlinson copies, the archetype of the second Rawlinson text cannot likely have been the first Rawlinson text.[14]

This version of Ego Dormio fused with Form occurs in a very few codices of limited geographical provenance. Rawlinson A 389 (R) can be associated firmly with Lichfield, both by its language and by the fact that it belonged to canons of Lichfield at the turn of the sixteenth century. The same version of Ego Dormio also appears in a pair of manuscripts from a nearby West Midland locale, north Worcestershire—Bodleian Eng. poet. a.1 (the Vernon ms., V) and British Library Additional 22283 (the Simeon ms., S). These two mammoth miscellanies, both the size of monastic lectern bibles, share common exemplars for most of their contents. In addition to these three codices and to Westminster 3 (W), the text also occurs in a manuscript which Smith and McIntosh tell me was copied by a north Warwickshire scribe—Paris, Bibliothèque Ste. Geneviève 3390 (G).[15]

The relations of these five manuscripts are fairly easy to untangle, and the variants show W rather distantly connected to the remainder of the manuscripts. Agreements of RSVG in readings probably erroneous occur extremely frequently, perhaps seventy times in the sample I have collated, about two-thirds of the Ego Dormio.[16] And within this grouping, S and V, probably copied from a common exemplar, share a number of unique readings; more frequently, they share readings with R (nearly thirty examples). Only the G text agrees in error with W to any extent—about eight readings.[17] G's agreements with W, a few provocative unique readings, and its avoidance of numerous specifically RSV errors suggest one of two situations. On the one hand, G may simply resemble W because it provides readings of a better version of the archetype than that available to RSV. On the other, its scribe may have had access to two archetypes—one which provided the readings shared with RSV, as well as a second and better copy which, like W, avoided many of these readings and provided some readings shared with W. One can represent this situation by the following stemma, only a single part of the complicated descent of Rolle's epistles:



Page 201
The dotted line indicates the possible G conflation. In any event, the text of W appears completely isolated: its only close congener G is more likely to have borrowed from an archetype resembling W than to be primarily descended from it.

This material is provocative, because western connections for surviving manuscripts probably from exemplars related to those used in composing Westminster 3 pop up at every turn. Here collation of texts demonstrates that the Westminster Ego Dormio is most closely related to the Warwickshire Ste. Geneviève copy. Further examinations of textual relations, exercises like that I have described above but refrain from repeating in detail, indicate that the nearest textual relatives of Westminster 3 works almost invariably come from the southwest midlands. Excluding short items, Booklets I, II, V, VI, VII, and the first item of IV share this bent.

Booklet I has been investigated more thoroughly than any other portion of the manuscript. The Pater Noster exposition attributed to Richard Rolle (item 1) is one of only two Westminster works to appear in an edition which analyzes all the manuscript evidence. In the case of this work, the most extensive in the entire codex, Florent Aarts has presented quite unequivocal conclusions: "Textually these two manuscripts [Westminster and Sidney Sussex 74] form a very close group. Since they avoid each other's errors, they cannot have been copied from each other, but there can be little doubt that they derive from a common ancestor, which must have contained a fairly accurate text."[18] The Sidney Sussex copy, again, is identified by the Edinburgh dialect project as a Warwickshire codex.

The other fully edited text in Westminster 3 is the Benjamin Minor (item 4) which takes up most of Booklet II. Here the preponderance of indirect evidence suggests some connection between Westminster and Arundel 286. The evidence remains indirect, since Westminster is a twin of Cambridge Univ. Library; both these texts and Arundel belong to the same large branch of generally inferior copies.[19] But two factors point toward connections with the Arundel copy—a pair of other texts shared with Westminster and the sometimes erratic behavior of the scribe who copied relevant portions of the Arundel manuscript.

Arundel provides two other texts which also appear in Westminster, for both of which it appears the closest textual relative. This manuscript contains a copy of the standard decalogue (item 3), which shows some correspondences with Westminster in what, so far as I can tell, are unique readings.[20] And the manuscript also includes the only other known copy of "A noble tretys of maydenhode" (item 13). In both cases, the connections may remain tenuous, but this reflects certain peculiarities of the Arundel scribe.

This scribe is perhaps better considered an editor than a mere copyist. Virtually all the texts I have examined have been subjected to considerable rewriting. Most frequently, this takes the form of excisions and abridgement; somewhat less often, the scribe provides summarizing paraphrase to keep the argument of his texts afloat.[21] Thus the decalogue tract has been severely condensed,


Page 202
a factor which makes even the small number of agreements with Westminster particularly significant; the tract on virginity has been subjected to similar summarizing procedures. In this context, one might consider that the Arundel version of Benjamin Minor, a reasonably accurate reproduction by this editing scribe's lights, differs from Westminster, not because copied from a different archetype, but because of the Arundel scribe's freedoms. In any event, the preponderance of the evidence suggests that Westminster relied on an archetype like that available to Arundel for the contents of two booklets (II and IV). This common archetype again suggests western connections, since the Arundel scribe is another Warwickshire copyist.[22]

Booklet V is devoted to a single text, a rather strange decalogue tract. Instead of offering sustained moral analysis of the precepts of Exodus 20, this discussion typically gives a translation of the biblical injunction and then illustrates that injunction by biblical translation and paraphrase alone. Unusually, this text relies not simply upon concordance of parallel precepts elsewhere in Scripture but upon lengthy biblical narratives designed as examples to illustrate the sins the decalogue names. So far as I know, there are only two other copies of the work: Laud Misc. 656, a late fourteenth-century collection of biblical histories, comes from Oxfordshire, another potential western connection for a Westminster 3 text, while the later Trinity copy is in the hand of Stephen Doddesham, monk of Sheen.[23]

Booklet VI proves relatively easy to localize, since the full text it transmits does not appear widely. However, this text does appear to have been at least marginally popular in some fifteenth-century circles. Only three full manuscripts of this translation of St. Edmund's Mirror are known; however, there are several excerpted versions—one, a pair of lengthy extracts from the total text; others comprised of small segments chosen for their instructional purport.[24]

Collation of sample passages from the full versions allows fairly unambiguous conclusions. The Additional ms. appears, on the basis of a number of unique additions and the avoidance of errors shared by the other texts, to be independent. Although this codex shares a number of erroneous readings with Westminster, on inspection these reflect independent coincidence, rather than genetic relationship; they include stripping prefixes from verbs and eliminating apparently repetitive phrases. In contrast, Westminster agrees considerably more frequently with Bodley 416 than with Additional (a ratio of about 2.5 to 1), and many of these shared errors are apt to have been genetically transmitted. The Huntington fragments clearly have been edited and rewritten; although a number of readings not in other copies appear potentially authorial, they may equally be examples of inspired scribal editing.[25] Because it lacks any signs of the additions, Huntington appears copied from an exemplar resembling Bodley and Westminster. Again, the closest surviving text is western: the forms of Bodley 416 are mixed, but, as Smith tells me, "certainly West Midland, perhaps Warwickshire." Thus a quite substantial portion of the materials available to the copyists of Westminster 3 seems to have come to them from a southwest midland source.


Page 203

In discussing these "western" portions of the codex, I have ignored two short texts which appear interspersed with these materials—item 2 (in Booklet I) and item 5 (in Booklet II). It is now time to bring them from the periphery to the center of discussion. These texts do not share the western antecedents of the materials which surround them, and indicate some minimal tailoring of independently conceived booklets to form the completed codex.

I have already noted the apparent reliance of Westminster 3 and Sidney Sussex 74 upon a common exemplar for the Pater Noster tract at the head of the volume. Sidney also contains a copy of the Wycliffite Ave Maria (item 2): it occurs there twenty-odd folios after the conclusion of Pater Noster. However, if one examines the manuscript variants for this text, one finds that the Westminster 3 copy is not only closely related to that found in Sidney Sussex; in fact, the Sidney Sussex is the least closely related of all the copies.

Three of the texts—Westminster, Bodley 938, and Corpus Christi College Cambridge 296—agree in a variety of readings against Sidney; of these eleven appear significant. Moreover, within the three-manuscript group, Bodley and Westminster agree against the other texts in a substantial number of erroneous readings.[26] In fact, the Westminster copy seems to have been derived from the same exemplar as was Bodley 938, some ancestor of which lies behind four texts in Booklet III (items 6, 8, 11, and 12). Apparently Westminster scribe 1 felt that analysis of basic prayers belonged together and used an exemplar which provided the central contents of another booklet to fill out a partially blank quire elsewhere.

Similar decisions appear to occur elsewhere in the volume. Thus, the same supply of filler to avoid a blank occurs at the end of Booklet II, where the text is unique and perhaps even composed for the occasion.[27] And the joining of like thematic material also explains the provision of a section of Pore Caitif which discusses virginity to conclude booklet IV, otherwise devoted to a lengthy tract on the subject.

The treatment of this Pore Caitif excerpt indicates one of the favored procedures of scribe 1. Much of the form his work takes appears predicated upon copying texts into booklets which depend upon available exemplars. But factors involving the potential joining of booklets into a full codex also operated.

The example of the three texts which may have come from an exemplar shared with Arundel 286 is here instructive. Although he was working from a single source, the scribe divided materials from it between two different booklets. This looks like a content-oriented choice: two of the texts were of general relevance, the third concerned with a more specific topic, virginity. Moreover, the two texts which comprise Westminster Booklet 2 appear in Arundel 286 as adjacent but in reversed order—although that may not have been the original order in Arundel or in its (and Westminster's) shared exemplar. By reversing them, scribe 1 managed to create a situation in which the common decalogue tract might be brought into juxtaposition with Booklet 1: he thus established a codex-opening sequence which analyzed proverbially


Page 204
basic documents of the spiritual life: pater noster, ave, and decalogue.

Moreover, this content-driven arrangement may also explain the separate appearance, at the end of scribe 1's portion, of a second decalogue tract. The very different rhetorical approach of this text, with its emphasis on connecting individual morality with salvation history through exemplary biblical anecdotes, may have rendered it intrusive in the earlier part of the codex. Thus, even as production went on piecemeal, dependent largely upon exemplar availability, the scribe appears to have been careful not to foreclose absolutely some important options: working through booklets, he was permitted a certain flexibility which eventually enabled him to impose a logical and thematic ordering on contents, even at a point before the codex might be considered as some unified whole.

Booklet III is the messiest portion of the codes, but the issue of its sources is potentially soluble. As I have noted above, four of the texts are shared with Bodley 938, and for three of these (8, 11, 12), collation shows that Bodley and Westminster 3 share an archetype not available to other copies.[28] However, for the fourth of the texts, the popular Visitacio Infirmorum (item 6), Bodley is not clearly the source, and Westminster shows connections with a group of manuscripts which have this text and others in booklet III, including Bodley, Laud Misc. 210 (one of two surviving copies closest to Westminster for item 10, which does not occur in Bodley) and University College 97.[29]

In addition to this shifting of closest textual relatives, contents of the fascicle do not neatly correspond with the contents of any surviving manuscript. I have already noted Bodley's failure to provide item 10 (while the text does occur in UCO and Laud). But given the fifty-odd surviving copies, item 7 is untraceable (and because it is a common extract, not necessarily from the same source as item 14).[30] Item 9 survives elsewhere in only two manuscripts otherwise unrelated to Westminster 3. The appropriate inference seems that one here sees the remains of a rather large miscellany of texts, now only selectively extant (according to the choices of individual patrons or editors) in a congeries of manuscripts.

At this stage in the argument, however, no one will be surprised to discover that these texts could have come to the Westminster scribe from a western source. UCO, as Doyle has shown, is most probably from western Worcestershire and from exemplars also available to the Worcester scribe of the Simeon manuscript.[31] Most (ff. 20-93v, 98-168) of the composite Laud MS. again shows the forms of Warwick, and others perhaps those of Worcester itself. (However, other portions are by Cambridge and Lincolnshire scribes.) In contrast, Bodley displays mixed dialect forms and is not precisely localizable; but, the spellings of Cambridge Univ. Library, Part II, ff. 43-99v, copied by the same scribe, show less contamination from his archetypes and are those of Huntingdonshire.

The Huntingdon forms are provocative, since they generally accord with the dialect of Westminster 3. That is, the main scribe of the manuscript has been placed by the Middle English Dialect Survey in the Soke of Peter-borough.


Page 205
Although hand 2 is a mixture which has as yet resisted analysis, hand 3 is similar to the first and is described by McIntosh as "extreme southern Lincs on the edge of the Soke." Thus these scribes write that kind of East Midland which the dialect project has elsewhere associated with the forms of early version Wycliffite bible manuscripts.[32]

At this point one faces, I should think, something of a quandary. The dialectical similarity of the language of two of the three Westminster scribes should, I think, conduce one to think of all three as a team, although one with unequal shares: scribes 2 and 3 essentially provided piecework to accompany a much greater amount of work by scribe 1. Thus the booklets of the manuscript could reflect a bit of production economy, shared work, perhaps of a temporary team, making the copying proceed apace. And this division of labor would explain the diversity of exemplars used—the subsidiary scribes got archetypes unneeded for the copying of other, more extensive portions. But the apparent nature of these archetypes is surprising: the localism of the Westminster scribes' language does not accord with that of their sources, especially the persistent western flavor of many manuscripts most closely related to their product.

Here I think that one may run up against a problem inherent in all dialect studies, whether medieval or modern. When I read a version of this paper at a scholarly conference I amused my audience by pointing out to them that I have never learned to speak anything remotely confusable with standard English and have persisted for years with a local dialect.[33] I also pointed out that through vicissitudes of family history, this is not the dialect of my birth-place; and that through further vicissitudes, I happen now to speak it near my birth-place. But, because of other accidents (in this case, large-scale social movements), my speech is no longer thoroughly outré in its current context: indeed, certain features gain support from a variety of dialects spoken in inland southern California and parts of the southern San Joaquin Valley (generally misdenominated "Okie"). I happen to speak this way, not because of geography, but because of birth: this happens to be the dialect of my mother's birth-place and has been reinforced by spending a substantial portion of my childhood in that locale.

I would suggest that the various disparities uncovered in this study of the origins of Westminster School 3 might be resolved by considering similar situations. First, maps prepared by the Middle English Dialect Survey clearly show that what is being identified are phonetic variants transcribed by copyists: the isoglosses produced closely resemble those on mappings of modern phonetic variation.[34] In consistent cases, these, I would suggest, reflect birth dialects, or (what comes to much the same thing) spelling systems inculcated during scribal training which impose on apprentice writers preferred spellings corresponding to the phonetic forms of the local dialect. Scribes vary in spelling because they follow a general rule to "spell phonetically," but what they are spelling is the English they hear themselves speaking.

However, as my personal anecdote suggests, the spellings only identify


Page 206
the scribe's birth dialect. They localize family origin. Dialect maps do not recognize migration of any sort: scribes may practice their craft in areas far removed from their place of origin, and their books may wander far from the place where they were copied. In effect, source does not identify provenance.

This view allows, I think, a convenient hypothesis about the origins of Westminster 3. Any hypothesis does have to account for an impressive coherence of data: that this manuscript was copied by scribes apparently of central East Midland origin and that they had access to a succession of texts apparently of western extraction. The easiest hypothesis is that these two provincial groups had some kind of loose association in some single center, not one necessarily associated with either dialect.

Here one might conjecture three different possibilities. Agreement with Sidney Sussex 74 for the Pater Noster tract (item 1), as well as the inclusion of some texts often considered vaguely Wycliffite, might place the composition of the codex in that central East Midland Lollard manuscript factory of which Anne Hudson keeps providing hints.[35] Alternatively, the binding stamp on the Westminster 3 boards, first used in the 1480s in Cambridge, might suggest that provincial center;[36] such placement might take into account the large number of texts shared with Bodley 938. But the best likelihood is provided by the known provenance of the manuscript: f. 231 bears the inscription "Amen per Ricardo Cloos the wiche is owner of this bouke anno 1472."

As Doyle has pointed out, Richard Close is known from the accounts of St. Mary at Hill, London, 1483-1502, and was a frequent warden of the parish.[37] And although his name does not appear in Paul Christianson's lists of persons associated with the book trade, it is barely possible, as Doyle noted, that he was involved in book production.[38] There is one unfortunate limitation on such speculations: the inscription on f. 231 was probably written no earlier than the 1530s. However, given its meticulous copying of a Richard Close monogram (or some variety of merchant's mark) it appears inexplicable unless done by a family member, perhaps copying from another of Close's manuscripts.

However uncertain this information, it nonetheless implies the simplest explanation of the manuscript and its origins. It allows one to suppose the Westminster scribes were participants in that well-documented migration from the central East Midlands to London, and that they executed the codex there. Their texts could then have come either from a western customer, who had access to and provided the copyists with exemplars from home, or from some geographically equivalent route—an equally provincial western scribe or stationer, say, who could supply exemplars.

Certain aspects of this discussion should be of general relevance in approaching other Middle English manuscripts, especially religious miscellanies. First, Westminster School 3 exhibits two underlying causes of booklet production which have yet to receive analysis equivalent to that accorded features purely codicological. In this codex, booklets reflect exemplar supply:


Page 207
typically the scribes began copying in new booklets when they switched from one source to another. Further, booklets in this manuscript also show efforts at grouping materials within thematic or rhetorical ranges. Second, this codex illustrates another, insufficiently examined feature, that booklets, without extensive physical tailoring, may receive adjustments designed to "codicize" them, to fit them neatly together into larger, more coherent sections so as to create a full codex. Finally, my discussion of scribal dialects should provoke further thought on the migration of copyists and codices.

A further conclusion strongly suggested by my discussion is the need in Middle English studies of a new type of critical edition. Customarily editions are text-centered: that is, they choose for central objects of study single works which they abstract from manuscript context.[39] I do not wish to argue that anyone should want to print a bad text, simply because it appears in one manuscript or another. But I hope that I have suggested that the combination of works which a single manuscript offers could be a perfectly sound basis for a series of critical texts with full apparatus. Such an edition would, I think, reveal a good deal about the nuts and bolts of literary history—a history which is not exhausted simply by study of source-receiver relations or of themes and styles but includes transmission and reception as well. Particularly useful would be what such an edition would have to tell about the very raison d'être of manuscript copies, for all manuscripts are involved, however covertly, in procedures of canonization, of creating literary tradition by providing (groups of) "accepted works."