by robert j. meyer-lee

Transcription

1 Fragments IV and V of the Canterbury Tales Do Not Exist by robert j. meyer-lee Over the last three decades or so, epistemological and methodological critiques of Middle English editorial theory and practice have rendered relatively common the complaint that modern editions of Chaucer obscure the actual uncertainties pertaining to the surviving texts and misrepresent how those texts would have been received by Chaucer s contemporary audience. 1 Dovetailing with iterations of long-standing editorial debates, such as how to respond to the variant tale order in surviving witnesses of the Canterbury Tales, this skeptical attitude toward the modern edition prompted Derek Pearsall s famous suggestion for an alternative presented partly as a bound book (with the first and last fragments fixed) and partly as a set of fragments in folders, with the incomplete information as to their nature and placement fully displayed. 2 A less radical concession to tale order uncertainty is, of course, already evident in the term fragments the term by which editions such as The Riverside Chaucer explicitly delineate the contents of Pearsall s would-be folders (e.g., Fragment I ), even if the folders themselves remain bound in fixed order. For most Chaucer critics today, these fragments are a basic given of the interpretive landscape of the Tales : they mark off that which we can be I thank A. S. G. Edwards for his encouraging response to material from this paper presented at Kalamazoo, For an early version of this complaint, see Derek Pearsall, Editing Medieval Texts: Some Developments and Some Problems, in Jerome J. McGann, ed., Textual Criticism and Literary Interpretation (Chicago, 1985), More recently, see, for example, Tim William Machan, I endowed thy purposes : Shakespeare, Editing, and Middle English Literature, Text 13 (2000): For some general reflections on the implications of this critique for Chaucer criticism and pedagogy, see my Manuscript Studies, Literary Value, and the Object of Chaucer Studies, Studies in the Age of Chaucer 30 (2008): Derek Pearsall, The Canterbury Tales (London, 1985), 23. the chaucer review, vol. 45, no. 1, Copyright 2010 The Pennsylvania State University, University Park, PA. CR45.1_01Meyer_Lee.indd 1 4/19/10 11:01:27 AM

2 2 The Chaucer Review virtually certain Chaucer intended (the order of the tales within fragments) from that for which we have varying and contested degrees of uncertainty (the order of the fragments). Indeed, the concept of the fragment has become so familiar that its specific editorial deployment bears revisiting. As Larry Benson describes in the Riverside s introduction to the Tales, The work survives in ten fragments, labeled with Roman numerals in this edition (the alphabetical designations added in parentheses are those of the Chaucer Society, adopted by Skeat in his edition). These fragments are editorial units determined by the existence of internal signs of linkage bits of conversation or narrative that explicitly refer to a tale just told or to one that immediately follows. 3 That the ten fragments are, as Benson puts it, editorial units and do not in fact possess surviving codicological reality (at least, not neatly corresponding codicological reality, despite continued speculation that individual tales or sequences circulated prior to their collection in the surviving manuscripts) 4 is important to remember. To state that the work survives in ten fragments is not to assert that it survives physically thus, as a reader new to Chaucer studies might suppose, but rather to explain in shorthand that Chaucer editors have identified, through internal evidence, ten (relatively) stable sequences of tales in the existing manuscript collections. As an editorial construct rather than a physical reality, then, the constitution of the fragments themselves, in addition to their order, is a product of interpretation and hence subject to debate. Benson s reference to the alternative Chaucer Society fragment designations hints as much, as does, more explicitly and specifically, the parenthetical exceptions of his following statement: There are no explicit connections between the fragments (save for IX X and, in the tradition of the Ellesmere manuscript, IV V) and, consequently, no explicit indication of the order in which Chaucer intended the fragments to be read. 5 In this present article I focus on the second of Benson s exceptions, the connection between Fragments IV V, which comprises the so-called Merchant s Endlink (MerE ) and Squire s Headlink (SqH ). I argue that Benson s recognition of this exception need not be as limited or provisional 3. The Riverside Chaucer, ed. Larry D. Benson, 3rd edn. (Boston, 1987), 5. Henceforth cited as Riverside. 4. As argued, for example, by Charles A. Owen, Jr., The Manuscripts of the Canterbury Tales (Cambridge, U.K, 1991), esp. 103 and Riverside, 5. CR45.1_01Meyer_Lee.indd 2 4/19/10 11:01:28 AM

3 robert j. meyer-lee 3 as his remark implies, but rather the fragment break dividing MerE and SqH, as it occurs in the Riverside and most other modern editions, is not, in any legitimate way, for whatever manuscript tradition, a defensible editorial decision. Instead, the original nineteenth-century rationale for the break which has not been superseded but simply forgotten by most critics was founded not on manuscript evidence but rather on artistic assumptions now almost universally rejected. Hence, the designation of two sequences as Fragment IV (or Chaucer Society Group E, consisting of the Clerk s and Merchant s Tales ) and Fragment V (or Group F, consisting of the Squire s and Franklin s Tales ) realizes precisely the opposite of the recognition of textual uncertainty supposedly signaled by the term fragment. Rather than a marker of uncertainty about Chaucer s intention, the term in this case is a covert editorial imposition of assumptions about that intention. The IV V fragment break is thus exactly the sort of anachronistic interpretive editorial distortion about which Pearsall and others complain. For this reason, a properly critical attitude toward the modern Canterbury Tales edition or, more specifically, its inherited editorial tradition demands the elimination of fragments in this instance (or, less plausibly, a multiplication of them) rather than their recognition. As will be evident in the ensuing discussion, my argument here is not a new one and rests on readily available evidence. Nonetheless, this argument does not seem to have stuck, as the fragment break in question continues to mislead readers and consequently, I believe, has an adverse effect on Canterbury Tales criticism. My aim in this article is to bring renewed and wider attention to the dubiousness of this fragment break by offering detailed and, I hope, therefore decisive considerations of, first, the manuscript evidence for the IV V link and, second, the editorial interventions that resulted in its current representation. I then conclude with an example suggestion for how Canterbury Tales criticism which remains highly dependent on the Riverside representation of the Tales, despite generally increased awareness of manuscripts might respond to the elimination of this fragment break. 6 M a n u s c r i p t s As has been frequently observed, often paradoxically in the textual notes to these very lines, the fragment break occurring between the twenty-two lines 6. For an account of the interpretive possibilities pendant upon practical editorial decisions that is congruent in a general fashion with my concerns in this article, see Susanna Fein, The Epistemology of Titles in Editing Whole-Manuscript Anthologies: The Lyric Sequence, in Particular, Poetica 71 (2008): CR45.1_01Meyer_Lee.indd 3 4/19/10 11:01:28 AM

4 4 The Chaucer Review of MerE and the eight lines of SqH is with just two inconsequential exceptions that I describe below nowhere attested in the manuscripts. The textual uncertainty that seemingly calls for recognition in this case is not then a gap in the dramatic tissue between tales. It is rather (given the mobility of the tales in these fragments, and especially the Squire s Tale, in the surviving witnesses) the question of which specific tales that tissue should bind together or, alternatively, whether the thirty lines of the unbroken passage (henceforth referred to by the admittedly tendentious label MerSqL ) should appear at all. As Ralph Hanna s textual note in the Riverside states, Lines IV [ MerE ] and V.1 8 [ SqH ] constitute a single unit in all MSS in which they occur, although they do not always join the same tales (in Hg and some other copies they link the Merchant and Franklin). 7 In w hat follows, I amplify this brief comment and spell out its ramifications, drawing throughout on the treasure trove of data collected by John M. Manly and Edith Rickert in the The Text of the Canterbury Tales. 8 Fifty manuscripts survive that contain enough of the tales to be relevant to this case. In eighteen, MerSqL is written to join the Merchant s and Squire s Tales : Ad 1, Ad 3, Bo 2, Ch, Cn, Dd, Ds, El, En 1, En 3, Ha 4, Ha 5, Ld 1, Ma, Ps, Ry 1, Se, To. In addition, the deliberately mutilated Gg ( M R, 1:173) almost certainly had this version of MerSqL, and so makes a nineteenth member of this group. (In Appendix A, I supply the text of this version from San Marino, Huntington Library El.26.C.9, the Ellesmere manuscript.) Twelve of these manuscripts comprise all those of Manly and Rickert s type a tale arrangement that have enough tales to be relevant; six (Ch, Ha 4, Ld 1, Ps, 7. Riverside, John M. Manly and Edith Rickert, The Text of the Canterbury Tales: Studied on the Basis of All Known Manuscripts, 8 vols. (Chicago, 1940), henceforth cited in the text as M R by volume and page. Unless otherwise stated, all information about manuscripts comes from this source, and especially from the authors account of MerE and SqH (2:284 87, 3:479 82); their discussion of tale order (2:474 94) and the three pages of tables that follow; and their catalog of variants for MerSqL (6: ). The well-known problems in Manly and Rickert s procedures for assembling their text are not here my concern; see, e.g., the trenchant critique of George Kane, John M. Manly ( ) and Edith Rickert ( ), in Paul G. Ruggiers, ed., Editing Chaucer: The Great Tradition (Norman, Okla., 1984), I have also benefited from the relevant discussion in Donald C. Baker, ed., A Variorum Edition of the Works of Geoffrey Chaucer, Volume II: The Canterbury Tales, Part Twelve: The Squire s Tale (Norman, Okla., 1990), henceforth cited as Variorum. Throughout, for convenience, I list manuscripts simply by M R s sigils, providing full information for only those of special significance. Readers familiar with the topic I examine here might expect also a consideration of the Man of Law s Endlink (MLE ), which serves as a prologue to SqT in twenty-six of MLE s thirty-three instances and hence likely influenced the variation of MerSqL (or vice-versa). But, as I hope this article demonstrates, I believe that we may achieve more clarity about MerSqL by considering it independently of MLE. Indeed, this clarity, once obtained, may perhaps then extend to the status of MLE, although in this present article I must leave this topic unexplored. For Variorum s discussion of MLE, see CR45.1_01Meyer_Lee.indd 4 4/19/10 11:01:28 AM

5 robert j. meyer-lee 5 Se, To) fall into Manly and Rickert s anomalous tale arrangement category. 9 The remaining manuscript, Ry 1, superficially like [type] d, but differing as to links and in reality unique ( M R, 1:478), is unlike the rest in that MerSqL does not actually join the Merchant s and Squire s Tales, though it is written this way. Instead, MerSqL follows the Man of Law s Endlink, and, as in most other type d manuscripts, the Squire s Tale and Merchant s Tale follow next, in that order. Fourteen manuscripts (not counting the damaged Gg, included above, and En 2 and Ra 2, included below) lack MerSqL altogether, although they possess all four tales of Fragments IV V. These manuscripts include all eight of Manly and Rickert s type b that have the four tales: Ha 3, He, Ln, Ne, Py, Ra 3, Tc 1, Tc 2 (although Ln is one of the exceptional cases discussed below); all three of type c : Cp, La, Sl 2 ; two of type d : Gl, Ph 3 ; and one anomalous manuscript: Hk. In three manuscripts, MerSqL joins the Merchant s Tale to the Franklin s Tale rather than to the Squire s Tale, with minor but important differences, such as Sire Frankeleyn com neer in place of Squier come neer in the first line of SqH (V 1). Among these manuscripts is the anomalous Hg (Aberystwyth, National Library of Wales Peniarth 392D, the Hengwrt manuscript), assumed to be the earliest surviving copy of the tales; in Appendix B, I supply Hg s text of this version of MerSqL. The other two manuscripts are both type d : Ht and Ii. In addition, since Ht is the twin of Ra 2 at this point in the Tales ( M R, 1:455), the defective Ra 2 almost certainly had this version of MerSqL, and so brings the membership of this group to four. Notably, Ha 4 and Ld 1 listed above in the group with MerSqL joining the Merchant s and Squire s Tales appear to have had an exemplar with the Hengwrt version, which was then adapted to its new disposition. Thus, for example, the first line of SqH begins not, like Ellesmere, with Squier but with Sir Squyer, presumably on the model of the Hengwrt-like exemplar s Sire Frankeleyn (see M R, 2:284 85, 3:481, 6:507). In two manuscripts, Mm and Pw (both type d ), MerSqL joins the Clerk s and Franklin s Tales (thereby putting a severe strain on the applicability of my label for the link!). This version of MerSqL, like the previous one, mentions the Franklin rather than the Squire in the first line of SqH, and Pw has, among other differences in MerE, By mony ensamples instead of the 9. Throughout this article my references to Manly and Rickert s manuscript types refer to tale arrangement categories, of which they describe five: a, b, c, d, and anomalous. I do not make reference to the similarly lettered categories of textual affiliation, although, as M R (see esp. 2:41 44) and many others have noticed, there are obviously correspondences, of a rather complex but consistently attestable nature, between them. CR45.1_01Meyer_Lee.indd 5 4/19/10 11:01:28 AM

6 6 The Chaucer Review Ellesmere version s By this Marchauntes tale at IV 2425 (see M R, 3:481, 6:502 3). However, Mm, despite the preceding Clerk s Tale, still refers here to the Merchant (see M R, 2:286, 6:502 3). Ten manuscripts contain a short form of MerSqL : Bw, Dl, Fi, Ha 2, Lc, Ld 2, Mg, Nl, Ry 2, Sl 1. This version appears, with minor variation, as two rhyme royal stanzas closely resembling, respectively, lines 9 14 of MerE (i.e., IV ) and SqH. (For the text of this version from the Lc manuscript, see Appendix C.) All these manuscripts are type d except for the anomalous Nl. In addition, the defective type d manuscript En 2 most likely had this version (see M R, 1:137), bringing the members of this group to eleven. In all but one of these manuscripts, this version joins the Clerk s and Franklin s Tales ; in Fi it joins the Merchant s and Franklin s Tales. Among the more significant variants, the copies in Dl and Fi crowd in two additional lines (the final two of the long form of MerSqL ) at the end of the link, and Nl substitutes these two lines for the last two lines of the short form (see M R, 2:285 86, 3:482, 6:507 8). The two exceptions to the rule of MerSqL always appearing as a single passage occur in Oxford, Trinity College Arch. 49 (To) and Lincoln, Lincoln Cathedral Library 110 (Ln), but neither is truly exceptional. As indicated above, in To the full MerSqL occurs between the Merchant s and Squire s Tales, but the scribe also inserted just SqH before the Merchant s Prologue (MerPro ), that is, IV , with the first line of SqH reading Marchaunt instead of Squyer (see M R, 1:537, 6:507). 10 According to Manly and Rickert, To, which they date to (2:48), is full of errors (1:537) and shows many signs of amateurish work (1:538); its entirely anomalous arrangement of tales seems to be due in part to accidental shifting of quires in a battered ancestor and in part to picking up bits from different sources (as the textual affiliations also indicate) (1:539). It seems most likely, then, that To s obviously very hurried and careless scribe (1:539) carved out this anomalous version of SqH from a copy of the full MerSqL perhaps because of some misperceived need for additional introductory material for the Merchant. The b type Ln, dated by Manly and Rickert to (2:47), did not initially possess any portion of MerSqL. As in the case of most other b manuscripts, the Squire s Tale follows the Man of Law s Endlink (here with sompnour in II 1179), and is in turn succeeded by the Merchant s Tale and the sequence of Wife of Bath s, Friar s, and Summoner s Tales (i.e., Fragment III or Group D). Yet, as Manly and Rickert observe, Ln is a 10. See also Variorum, note to lines 1 8 (125 26). CR45.1_01Meyer_Lee.indd 6 4/19/10 11:01:28 AM

7 robert j. meyer-lee 7 striking example of a MS made up from several sources, with clear directions, apparently written by more than one supervisor, not only for its present arrangement but for a rearrangement of some of the tales according to a different type of order, which must have resulted from careful comparison with other MSS. Some passages are added in the margin, and others are indicated. (1:332) Among these later marginal insertions is a copy of SqH, which appears next to the so-called Host Stanza, which in several manuscripts separates the Lenvoy de Chaucer concluding the Clerk s Tale (IV ) from MerPro. Here, however, the Host Stanza links the Clerk s Tale to the Franklin s Tale, to which latter speaker the marginal SqH refers; apparently, this later scribe discovered in SqH a more appropriate introduction to the Franklin s Tale and wrote it in the margin. Yet even more curiously, additional marginal comments plainly indicate knowledge of the version of the entire MerSqL joining not Clerk and Franklin but Merchant and Squire. Following are Manly and Rickert s transcriptions of these scribbled comments : opposite the beginning of the Host Stanza, [Here?] sholde þe marchantz tale ; opposite lines 4 5 of the Host Stanza, þis prolog is fals for it is [for þ]e skquiers tale ; above SqH, skquier [ frankleyn stroked out] þis longeth ; and below SqH, [At] Sarray in þe londe of Tarterie, that is, V 9, the first line of the Squire s Tale (1:334). Supporting this supposition is, then, the manuscript s placement of a uniquely independent MerE. After the Merchant s Tale the first scribe seems initially to have left a space on the recto and verso of folio 91 for about seventeen lines, perhaps expecting the short form of MerSqL (even though the Wife of Bath s Prologue follows). But instead a later scribe, making use of the margins, squeezed in the entire twenty-two lines of MerE (1:333 34). The most likely explanation of both of these additions, as Manly and Rickert suggest, is that the later scribe had access to an exemplar with the full MerSqL in its Ellesmere form. Given Ln s already-copied b order of tales, this scribe shrewdly decided to use the MerE portion in the space left after the Merchant s Tale, keeping the SqH portion until he could find a place for it (1:334) as an improved introduction to the Franklin s Tale to the chagrin of the later annotator, who was aware of the disposition of MerSqL in its exemplar. The divisions of MerSqL that appear in To and Ln do not, therefore, provide evidence of independent MerE and SqH textual traditions, but rather of scribal responses to particular copying exigencies, with these scribes just as aware as their nineteenth-century editorial successors would be of exactly CR45.1_01Meyer_Lee.indd 7

8 8 The Chaucer Review where MerSqL is best split. We have solid grounds, as well, to put aside the short form of MerSqL as a scribal invention an adaptation of the long form, as Manly and Rickert argue, to match better a preceding rhyme royal Clerk s Tale rather than the couplets of the Merchant s Tale. Given the plain relation of the short form s wording to that of the long form, the strongest claim that could be made for the short form is that (ignoring for the moment all protests on aesthetic grounds) it is Chaucer s original rhyme royal epilogue to the Clerk s Tale, which someone later expanded, reformatted, and reworded into the long forms of MerSqL to follow the Merchant s Tale rather than the Clerk s Tale. But this supposition seems untenable on the internal evidence that, with the Host s allusion to his vice-ridden wife, the first stanza of the short form is, despite its changes from the long form, still much more suitable following the Merchant s Tale of January and May than the Clerk s Tale of Walter and Griselda. 11 This evidence, moreover, receives external support from the fact that, as mentioned above, Mm, though one of the two manuscripts in which the long form of MerSqL joins the Clerk s and Franklin s Tales, nonetheless preserves the reading this marchauntes tale in IV 2425 further suggesting the short form to be a scribal adaptation in response to the apparently misplaced position of MerSqL (see M R, 2:286). And if we do entertain aesthetic considerations, then any scenario that makes Chaucer the author of the rough rhyme royal stanzas of the short form, and at the same time not the author of the witty thirty-line supposed revision of these stanzas, strains all credulity. Minus this aesthetic argument, this same evidence (i.e., the reading of marchauntes in Mm and poor suitability for a preceding Clerk s Tale ) applies also to the two manuscripts containing the long version of MerSqL following the Clerk s Tale. Hence, despite the early date and importance of the other of these manuscripts, Pw, we can safely assume that this version of MerSqL is also a scribal adaptation, perhaps the version on which the short form was based. If we thus drop these two manuscripts and the eleven containing the short form from consideration, and ignore the use of SqH and MerE in To 11. Although, as W. W. Skeat has argued (in respect to the long form of MerSqL in Pw that joins Clerk and Franklin), the Host s comments do make some sense as a response to the Clerk s concluding Lenvoy de Chaucer ; see W. W. Skeat, The Evolution of the Canterbury Tales, Chaucer Society Publications second series, no. 38 (London, 1907), 33. But Skeat s argument depends on the proposition (rejected by editors from Manly and Rickert on) that the order of Pw represents Chaucer s own earlier attempt to arrange the tales, with Ha 4 standing as his final revision in this regard. Thus, for Skeat, despite the putative authenticity of a MerSqL joining Clerk and Franklin, the MerSqL variant joining Merchant and Squire has the most authority. And in any event, Skeat later decides that, even though Chaucer was responsible for the order of Pw, the alteration of MerSqL in Pw was a scribal attempt to salvage a piece of text that Chaucer temporarily had decided to cancel; see W. W. Skeat, The Eight-Text Edition of the Canterbury Tales (London, 1909), 50 and CR45.1_01Meyer_Lee.indd 8

9 robert j. meyer-lee 9 and Ln, we are left with three possible dispositions for MerSqL supported by the following counts of witnesses: nineteen manuscripts with MerSqL joining the Merchant s and Squire s Tales, fourteen lacking MerSqL altogether, and four with MerSqL joining the Merchant s and Franklin s Tales. Of course, as Helen Cooper has remarked, textual correctness is not a matter of democracy, and a head count [of manuscripts] will reveal more about lines of transmission than it does of the quality of the text itself. 12 Moreover, while an analysis of lines of transmission would serve to develop an argument about the relations among the three groups of manuscripts I have delineated and hence a scenario for how the corresponding dispositions of MerSqL evolved, this analysis is a large project in and of itself, has been performed toward more general ends by many others, and in any event cannot, by the nature of the surviving evidence, be definitive. 13 In lieu of such analysis and cued by the elitist approach to textual correctness indicated by Cooper, I therefore turn to smaller, more editorially favored groups of manuscripts, which confirm the overall counts for the full set of fifty. Among the ten base manuscripts of the Variorum Chaucer, in five, MerSqL joins the Merchant and Squire: Ad 3, Dd, El, Ha 4, and, presumably, the mutilated Gg, all of which are type a except for the anomalous Ha 4. In three, MerSqL is absent: the type b He and the type c Cp and La. The remaining two manuscripts are the anomalous Hg, in which MerSqL joins Merchant and Franklin, and the type d Pw, in which MerSqL joins Clerk and Franklin. Narrowing the set further and considering only what are taken to be the six earliest manuscripts, we have three in which MerSqL joins Merchant and Squire (Dd, El, Ha 4 ), two that lack MerSqL (Cp and La), and one, Hg, in which MerSqL joins Merchant and Franklin. To be sure, the consistency of these counts with those for the set of fifty manuscripts stands in need of great qualification because of the sheer importance of Hengwrt, without which the balance of evidence would appear 12. Helen Cooper, The Order of the Tales in the Ellesmere Manuscript, in Martin Stevens and Daniel Woodward, eds., The Ellesmere Chaucer: Essays in Interpretation (San Marino, Calif., 1995), , at For Manly and Rickert s discussion of how the relations among CT manuscript families bear on the disposition of MerSqL, see M R, esp. 2:42 43 and 2:489. Throughout this discussion Manly and Rickert assume the priority of the type a or Ellesmere order. For an account that, in contrast, assumes the priority of the Hengwrt order, see the introduction to N. F. Blake, ed., The Canterbury Tales by Geoffrey Chaucer: Edited from the Hengwrt Manuscript (London, 1980), 8 12, which contains a succinct version of an argument that Blake has frequently made. Variorum waffles on the question of the priority of the two manuscripts in respect to order, tending toward Manly and Rickert s conclusions but influenced by Blake s; see esp and Baker s comments in the last paragraph (119). Owen s The Manuscripts is entirely devoted to a study of lines of transmission and how they may have determined the features of the manuscript groups. CR45.1_01Meyer_Lee.indd 9

10 10 The Chaucer Review to tip in favor of the two dispositions of either MerSqL joining Merchant and Squire or being absent. But before reducing the question of MerSqL to a facet of the long-standing debate about the extent of Hengwrt s authority, we will benefit by expanding the consideration of MerSqL to its textual relationships with the other links constituting Fragments IV and V that is, as represented in the Riverside, MerPro linking the Clerk s Tale to the Merchant s Tale (IV ), and the Squire-Franklin Link (SqFranL ) that the Riverside prints as the end of the Squire s Tale (V ). As the above review of the dispositions of MerSqL has suggested, there are significant interdependencies between these dispositions and those of the two other links. In the full set of fifty manuscripts, in every instance in which MerSqL joins Merchant and Squire (if we account for the presumed contents of missing leaves), MerPro also appears, and, in all but two of these, MerPro joins Clerk and Merchant. The exceptions are Ry 1 whose odd use of type a links in a type d order (in which the Merchant s Tale follows the Squire s Tale ) I noted above and the disordered and late anomalous manuscript Ch. In all cases for the other four dispositions of MerSqL, MerPro is always absent except in four type b manuscripts in which MerSqL is absent but MerPro follows the Squire s Tale, in which position the dramatic relevance of its first line ( Wepyng and waylyng... ) is obviously lost. Turning to SqFranL, we find (again accounting for the presumed contents of missing leaves) that when MerSqL joins Merchant and Squire, SqFranL joins Squire and Franklin in every instance but four. In these SqFranL is absent, and once more the exceptions include Ry 1 and Ch. The other two are the anomalous To, in which the Physician s Tale intrudes between the Squire s and Franklin s Tales, and the thoroughly contaminated Se ( M R, 1:496), in which the Franklin s Tale occurs near the end of the work, between the Pardoner s and Manciple s Tales. In the majority of the fourteen instances in which MerSqL is absent, SqFranL is also absent; in six type b or d manuscripts (Gl, Ln, Ph 3, Py, Ra 3, Tc 1 ), however, the Squire s Tale follows the Man of Law s Endlink, and SqFranL subsequently joins the Squire s and Merchant s Tales In the type d Ph 3, SqFranL seems originally to have appeared twice. As Manly and Rickert note, the loss of a folio probably carried with it the end of the SqT (F ), the adapted link (F ) [i.e., SqFranL joining SqT and MerT ], and the beginning of the MeT (E ) (1:428). But another copy of SqFranL then precedes FranT, which follows Mel near the end of the manuscript. Since this tale and link derive from a different source ( a highly edited and quite unknown MS [ M R, 1:429]) than that of the earlier portion of CT, Manly and Rickert believe that the exemplar for the earlier portion had lost FranT. For this reason, for the purposes of my survey I consider Ph 3 to take the same form as the closely related Gl. CR45.1_01Meyer_Lee.indd 10

11 robert j. meyer-lee 11 When MerSqL joins the Merchant s and Franklin s Tales or the Clerk s and Franklin s Tales, in the long or short form, SqFranL always joins the Squire s and Merchant s Tales. The interdependencies pertaining to the three relevant dispositions of MerSqL thus tabulate as follows: in fifteen of the nineteen instances in which MerSqL joins Merchant and Squire, both MerPro and SqFranL also appear in their type a positions (i.e., the positions printed in the Riverside ), and the four exceptions have no authority. In the fourteen instances in which MerSqL is absent, MerPro and SqFranL never appear in their type a positions, at least one of the two is always absent, and in four cases (the three type c manuscripts and Hk) both are absent. And in the four instances in which MerSqL joins Merchant and Franklin, MerPro is always absent and SqFranL always joins Squire and Merchant. A consideration of the smaller, better groups of manuscripts further confirms and clarifies these findings. If we account for the likely contents of lost leaves, in the nine Variorum manuscripts (ignoring Pw), all five in which MerSqL joins Merchant and Squire place MerPro and SqFranL also in their type a positions; two of the three manuscripts in which MerSqL is absent also lack both MerPro and SqFranL, and the other (He) contains just MerPro. And, in the one manuscript in which MerSqL joins Merchant and Franklin (Hg), MerPro is absent and SqFranL joins Squire and Merchant. Among the six earliest manuscripts, three contain all three links in their type a positions (Dd, El, Ha 4 ), two lack all three links (Cp, La), and the final one is Hg. Th e ready conclusion that one may derive from these counts is that MerSqL, MerPro, and SqFranL are of a piece: they form a single, if variant, textual intervention into the Canterbury Tales. Hence, although the four tales making up Fragments IV and V are notoriously mobile, the evidence for these links resolves into only three defensible editorial options: give all three links in their type a positions (with MerSqL as an unbroken passage), drop all three links, or supply just MerSqL and SqFranL in their Hengwrt forms (i.e., joining, respectively, Merchant to Franklin and Squire to Merchant). At this point, we must dip briefly into the murky waters of the debate about the extent of Hengwrt s authority as the earliest and, for most recent editors, best manuscript of the Tales, since we must assess how much of this authority to extend to its disposition of the links, which is unique among the small groups of manuscripts. As is well known, paleographical and codicological evidence strongly indicates that the two links that appear CR45.1_01Meyer_Lee.indd 11

12 12 The Chaucer Review in Hengwrt were inserted quite awkwardly only after the tales that they join were copied. 15 Moreover, most readers find the content of the Hengwrt links ill-suited to their use. In particular, in respect to MerSqL, readers have noted several difficulties with the first three lines of the SqH portion in comparison with their type a counterparts (here quoted from Ellesmere): Hg: Sire Frankeleyn / com neer / if it your wille be And sey vs a tale / for certes ye Konnen ther on / as muche as any man El: Squier come neer / if it youre wille be And sey somwhat of loue / for certes ye Konnen ther on / as muche as any man (V 1 3) In the Hengwrt version, the first line cannot be scanned into ten syllables, the second line s And sey vs a tale does not sound like Chaucer s idiom, and the third line s Konnen ther on does not make near as much sense with the antecedent a tale as it does with of loue. When combined with the manuscript evidence, internal reasons such as these have led most scholars who have considered the question to argue that the Hengwrt links represent adaptations of the type a versions, cobbled into the spaces left between the already copied and misordered tales. 16 Yet, while this scenario seems to me easily the most likely, it is by no means certain. In particular, the clumsy insertion of the Hengwrt links, when considered along with the absence of any of the links in the early manuscripts Cp and La, suggests the possibility that the Hengwrt links may not be Chaucer s, whether or not they are adaptations of their type a versions. This is more or less the position N. F. Blake takes in his Hengwrt-based edition of the Tales, by which logic he relegates both links to an appendix. 17 (Blake assumes the non-chaucerian original of the links to be the Hengwrt versions, 15. For an especially cogent account of this evidence, see chapter 9 in Ralph Hanna, Pursuing History: Middle English Manuscripts and Their Texts (Stanford, 1996), For this argument, see, for example, Larry D. Benson, The Order of The Canterbury Tales, Studies in the Age of Chaucer 3 (1981): , esp ; Cooper, The Order, esp ; and Jill Mann, Chaucer s Meter and the Myth of the Ellesmere Editor of The Canterbury Tales, Studies in the Age of Chaucer 23 (2001): , esp See the more extended discussion in N. F. Blake, The Ellesmere Text in Light of the Hengwrt Manuscript, in Stevens and Woodward, eds., The Ellesmere Chaucer, CR45.1_01Meyer_Lee.indd 12

13 robert j. meyer-lee 13 which, for the internal reasons mentioned above, seems to me unlikely; but Blake s logic still holds if the author of the links originally wrote the type a versions, and some sort of miscommunication occurred between this author and Adam Pinkhurst regarding their disposition.) Although Blake s argument has won few converts, we cannot eliminate the possibility, since the manuscript evidence remains equivocal enough to require a subjective appeal to what feels like Chaucer s verse to decide the matter either way. In this regard, David Lawton s clever and provocative speculation that Thomas Hoccleve wrote these links serves as a helpful reminder that, in short passages, the distinction between Chaucer s verse and that of his most accomplished disciple putting forth his best effort at imitation may be difficult if not impossible to discern. 18 Considerably less likely, from both manuscript and internal evidence, is the argument that the Hengwrt links are Chaucer s original and that the type a versions, as well as MerPro, are scribal adaptations to fit the different tale order. As I have indicated, even Blake, with his famous commitment to the authority of Hengwrt, does not accept this possibility, although recently working from the new supposition that both Hengwrt and Ellesmere may have been products of Chaucer s lifetime he has argued that both versions may be Chaucer s, with Hengwrt the witness to a first draft of the complete work. 19 This contention, too, seems to me unlikely from the internal evidence, but in any event supports the authority of the type a links as Chaucer s supposedly final version. Hence, although there will always be some very small room for doubt, we may resolve the three editorial options for these links into only two truly defensible ones: either print MerPro, an unbroken MerSqL, and SqFranL in their type a positions, or somewhat less plausibly drop all three links, with whatever annotation one deems appropriate to document the variation. 20 To decide firmly between these options, one ultimately has to appeal, as I have said, to one s sense of what seems authentically Chaucerian. If the passages are Chaucer s (and they certainly seem so to me), then Fragments IV and V should be printed as a continuous sequence of four dramatically linked tales. 18. David Lawton, Chaucer s Narrators (Cambridge, U.K, 1985), N. F. Blake, The Links in the Canterbury Tales, in Susan Powell and Jeremy J. Smith, eds., New Perspectives on Middle English Texts (Cambridge, U.K., 2000), This argument recalls Skeat s in The Evolution of the Canterbury Tales, where, as already mentioned, he asserts that all versions of the links (except the short form of MerSqL ) are authorial in origin. 20. Owen s argument about the evolution of the six earliest manuscripts and, in particular, that Ha 4 and Cp represent alternative responses to the disorder of Hengwrt dovetails with this present argument about the two options for the Fragment IV V links ( The Manuscripts, 7 22). CR45.1_01Meyer_Lee.indd 13

14 14 The Chaucer Review If they are not his, then Fragments IV and V should be broken into four separate, unlinked fragments. 21 But in neither case should Fragments IV and V, as such, exist. E d i t i o n s Given the commitment to the type a Ellesmere manuscript that the Riverside inherited from the editions of F. N. Robinson, who in turn inherited it from W. W. Skeat, one may find it odd that all these editors responded to the manuscript evidence for MerSqL by including the passage but dividing it into two. As I suggested at the outset of this article, this decision muddies the waters of what the term fragment supposedly signifies. More egregiously, it divides a passage that Chaucer or whoever wrote the link left as continuous, and, as a consequence, it splits two tales when, in all its variants, the passage was plainly designed to solder two of them together. Yet far more editions than the Riverside and its predecessors are guilty of this decision: for over one hundred years, almost every edition of the complete (or mostly complete) Tales has somehow divided MerSqL between MerE and SqH. The only exceptions of which I am aware are Blake s edition and the student-oriented ones of E. T. Donaldson, Lesley A. Coote, and Robert A. Pratt which do not mark any of the fragment breaks. 22 In addition, the recent (and also student-oriented) Broadview Canterbury Tales by Robert Boenig and Andrew Taylor advertises itself as an edition of specifically Ellesmere and accordingly prints MerSqL as an unbroken passage; however, bowing to tradition, it retains the separation of the Merchant s and Squire s Tales into Fragments IV and V (by means of an entire folio, as with the other fragment breaks) and starts line numbering anew at the point in MerSqL where SqH begins. Explaining the latter decision, the editors remark, Though Ellesmere runs the Host s reaction to the Merchant s tale and his 21. Or, as I have suggested in Manuscript Studies, 33, the tales and links might still be printed continuously, but along with some sort of acknowledgment of the likely co-authorship of the work in this regard. 22. E. T. Donaldson, ed., Chaucer s Poetry: An Anthology for the Modern Reader, 2nd edn. (New York, 1975), which follows Ellesmere in order, includes no paratextual indications of fragments, even in line numbering, which begins anew with each tale. Lesley A. Coote, ed., The Canterbury Tales (Ware, Hertfordshire, 2002) which, as an edition of Ha 4, is a pre-chaucer Society throwback to the days when that was the most favored manuscript also provides no paratextual indications of fragments, and does not even include line numbers. Robert A. Pratt, ed., The Tales of Canterbury: Complete (Boston, 1974), which follows a modified version of the Chaucer Society order (about which, see below), indicates fragments only, and hence obscurely, in its running heads (e.g., IV(E) /V(F)1 11 [372]) and implicitly in the line numbering. CR45.1_01Meyer_Lee.indd 14

15 robert j. meyer-lee 15 words to the Squire together, they are separated in other manuscripts and also in modern editions, apparently unaware that only modern editions and not other manuscripts make this separation. 23 All other editions from the days of the Chaucer Society to the present divide MerSqL, and, as the examples of Benson s and Hanna s comments in the Riverside suggest, editors have also typically registered some sort of recognition of the problematic nature of this decision. Almost without fail, they include some explicit comments in this regard, which range from intentional obfuscation, to perhaps genuine confusion, to blunt if thereby self-contradictory pronouncements of the actual unity of MerSqL. The most noteworthy or influential examples should suffice to illustrate this range of commentary. We may start with Jill Mann s very recent and widely available Penguin Classics edition of the Tales, in which we witness an especially informed and lucid example of self-contradiction. Mann has made plain her views on the topic of Fragments IV V in a 2001 article, which more generally argues against the common supposition that Ellesmere has been scribally improved in comparison with Hengwrt. In the process of countering Blake s contentions about the priority of Hengwrt, she forcefully asserts the authenticity of the Ellesmere MerSqL and SqFranL, and hence also of the continuity of the former link. 24 In her edition of the Tales, however, she splits MerSqL at the traditional juncture, subtitling MerE The Merchant s Epilogue and SqH The Squire s Prologue. Then, in her note to these lines, she clearly (if tersely) indicates the manuscript evidence to the contrary of this editorial decision ( they [the lines of MerSqL ] form a continuous unit in all the manuscripts in which they occur ), pleading conformity with the current practice of line-numbering as her rationale for nonetheless dividing the passage. 25 A similar sort of contradiction between editorial belief and decisionmaking, but expressed more tentatively and less clearly, is evident in both Robinson s first and second editions, which represent the break between 23. Robert Boenig and Andrew Taylor, eds., The Canterbury Tales (Peterborough, Ont., 2008), 225n2. An early twentieth-century edition for a German audience John Koch, ed., Geoffrey Chaucer s Canterbury Tales (Heidelberg, 1915) should also be mentioned in this regard; like Boenig and Taylor s text, Koch s is a presentation of Ellesmere (with variants from several other manuscripts and printed editions) and so provides an unbroken MerSqL as SqPro ; but also like Boenig and Taylor, Koch separates MerSqL from MerT with a marker of the new Fragment V and Chaucer Society Group F. 24. E.g., Whether the MerT-FranT and SqT-MerT links in Hg were (a) authorial, but received late, or (b) specifically composed to fill gaps in the manuscript, it is difficult to see why the scribe... should have wanted to alter them at a later stage so as to use them to link different tales in El; why not leave them as they were? unless, of course, he knew that the El links were the correct versions, because he himself had altered them to fit their place in Hg (Mann, Chaucer s Meter, 88 89). 25. Jill Mann, ed., The Canterbury Tales (London, 2005), 940. CR45.1_01Meyer_Lee.indd 15

16 16 The Chaucer Review Fragments IV and V in the form the Riverside later inherits, i.e., MerE appears as Epilogue to the Merchant s Tale, SqH appears as Introduction to the Squire s Tale, and a page break and the heading Fragment V (Group F) separate the two. In his explanatory note to Fragment IV, Robinson remarks, In fact, IV and V are really connected, as they stand, and might be regarded as one group, leaving uncertain why such really connected fragments only might form a single group of tales. 26 Later, in his note to SqH, Robinson is more expansive, if no more definitive: Fragment V, comprising the Squire s Prologue and Tale and the Franklin s Prologue and Tale, regularly follows the Merchant s Epilogue in the best MSS. Although the Squire s Prologue does not contain any reference to the preceding piece, it makes a satisfactory transition therefrom. Consequently it has been argued by several scholars that Fragments IV and V should properly be considered as forming a single consecutive group. In fact in MS. El the whole passage from IV, 2419, through V, 8, is written continuously and headed The Prologe of the Squieres Tale. 27 Here the first two sentences, although they in effect argue for the unity of MerSqL, verbally presuppose the existence of SqH as an independent textual unit, which does not contain any reference to the preceding piece (which indeed it would not, if it did exist independently). This presupposition then weakens the appeal to manuscript evidence, which Robinson like Boenig and Taylor much later limits to Ellesmere. 28 Benson s corresponding explanatory notes in the Riverside observe the same fact about Ellesmere but, in their evident confidence in the Ellesmere tale order, are far less provisional; and Hanna s textual note to SqH, quoted earlier, is much more definitive than Robinson is about the manuscript evidence, as is Benson s revised explanatory 26. These comments are identical in Robinson s two editions, for which, see F. N. Robinson, ed., The Poetical Works of Chaucer (Boston, 1933), 813, and The Works of Geoffrey Chaucer, 2nd edn. (Boston, 1957), Again these comments are identical in Robinson s first and second editions; see, respectively, The Poetical Works, 821, and The Works, Albert C. Baugh, perhaps influenced by this presupposition, rather misleadingly remarks in his note to SqH (which he represents similarly to Robinson), These uninspired lines are frequently written in the MSS as the concluding lines of The Merchant s Epilogue (Albert C. Baugh, ed., Chaucer s Major Poetry (Englewood Cliffs, N.J., 1963), 459. While providing some information about the actual unity of MerSqL (though frequently here should read always ), Baugh also implies with the adjective uninspired and passive tense are frequently written the possibility that SqH is a scribal appendage to MerE invented for the situation in which SqT follows MerT. As we have seen, this supposition has no manuscript basis whatsoever. CR45.1_01Meyer_Lee.indd 16

17 robert j. meyer-lee 17 note to Fragment V in his 2000 Riverside -derived edition of the Tales. 29 Th is increased confidence, of course, also exacerbates the contradiction between this editorial commentary and these editions exact replicas of Robinson s representation of the fragment break. One of the starkest divisions of MerSqL occurs in John Fisher s edition. Since each fragment appears here as a part prefaced by introductory material, in the case of Fragments IV V, SqH follows MerE after three pages and several headings. In his footnote to SqH, Fisher comments, Lines found before SqT only in E&c. They go with iv in Hg&c, Squier being changed to Marchaunt or Frankeleyn in other MSS as the context requires.... Although not strictly erroneous, the comment is rather misleading in its implications i.e., that a unified MerSqL is unique to Hg&c and those manuscripts that do not link Merchant and Squire, and, conversely, that in E&c SqH does not go with iv The corresponding footnote in Mark Allen s recent revision of Fisher s edition remains misleading: Lines generally found following in the MSS, although they do not often introduce SqT. 31 As we have seen, generally here should read always ; and, while Allen s remark about the lines not often introducing the Squire s Tale may refer to the frequency with which the Man of Law s Endlink instead introduces the tale, it may easily be misconstrued as meaning that the SqH more often introduces another tale, which is not the case. But the most obfuscating comment about MerSqL is probably Skeat s, whose edition has to varying degrees influenced all those heretofore mentioned. Skeat splits MerE and SqH between Groups E and F, subtitling MerE Epilogue to the Marchantes Tale and placing the rubric The Squire s Prologue in square brackets above SqH but after the intervening group and tale headings. Then, in his note to SqH, Skeat remarks, There is nothing to link this tale with the preceding one; hence it begins a new Group. In many MSS. (including E[llesmere]) it follows the preceding Epilogue without any break. 32 Here the assertiveness of the first statement (echoed in Robinson s less assertive does not contain any reference to the preceding piece ) is belied, without explanation, by the simple fact provided by the second, which 29. For Benson s comments, see Riverside 879, 890, and Larry D. Benson, ed., The Canterbury Tales: Complete (Boston, 2000), See John H. Fisher, ed., The Complete Poetry and Prose of Geoffrey Chaucer, 2nd edn. (Fort Worth, 1989), John H. Fisher and Mark Allen, eds., The Complete Canterbury Tales of Geoffrey Chaucer (Boston, 2006), W. W. Skeat, ed., The Complete Works of Geoffrey Chaucer: Edited, from Numerous Manuscripts, 7 vols. (Oxford, ), 5.370; vol. 4 contains the text of CT. CR45.1_01Meyer_Lee.indd 17