The Canterbury Tales Glosses and the Manuscript Groups

85 The Canterbury Tales Glosses and the Manuscript Groups Stephen Partridge The several kinds of marginal and interlinear glosses in the manuscripts of the Canterbury Tales are a neglected body of evidence. One reason for that neglect may be that the most complete published account of the glosses, that in the Manly and Rickert edition of the Canterbury Tales (3: 483-527), suffers from confusing presentation and a lack of any firm conclusions. In this paper I will review the accuracy and organization of the Manly and Rickert appendix on the glosses, and then go on to suggest a basic classification of the manuscripts according to which glosses they include. Finally, I will discuss in detail two instances where the glosses can draw our attention to relatively little known manuscripts. Any remarks on the accuracy of the Manly and Rickert appendix devoted to the glosses must be to some degree unfair to the editors. Clearly their energies were focused on establishing the text of the Canterbury Tales and recording variants, and in fact Manly attributes the gathering of material for the chapter on the glosses, and for several other chapters, to a graduate assistant, Robert Campbell (1: xiii). Nevertheless, it is necessary to point out the limitations of this appendix, since scholars continue to rely on it as the most comprehensive and accessible published record of this material.1 A review of some of the most widespread glosses, those in the MLT, will give a fair picture of how far the Manly and Rickert appendix is reliable. The list of where these nine glosses appear (3: 492) can be corrected as follows: •Ad 3 does not have all of the glosses, but only those at 771, 925, 1132, 1135; the lines where the others usually occur are absent from Ad3 due to loss of leaves. • In addition to those at 161, 421 (partial), and 771, Bo2 has the gloss at 197. • In addition to the gloss at 421, absent from both Cn and Ma, Cn lacks those at 1132 and 1135; these lines are absent from Cn due to the loss of a leaf. •Ha 5 lacks the gloss at 161, in addition to that at 1135. •Ry 2 does not lack the gloss at 161, and so should be added to the list of manuscripts that contain all nine. Glosses in Dd at II. 309, 421, 771, and 1132 are described as ‘defective’; these do run close to the inner edges of the leaves, and so are reproduced imperfectly in the facsimiles used by Manly and Rickert, but they are entirely legible in the manuscript itself. Here in the MLT, and throughout the Manly and Rickert appendix, many of the errors that do occur are confined to the record of later and less important manuscripts, and of shorter glosses. Moreover, none of the errors I have listed affects our understanding of the manuscript relations, and the same is generally true of the many singleA New Catalogue of the Manuscripts THE C ANTERBURY T ALES PROJECT OCCASIONAL PAPERS I, 1993, 85-94 86 Canterbury Tales Project Occasional Papers I manuscript variants purposely omitted by Manly and Rickert, and of their occasional misrecording of variants.2 The transcripts from the earliest manuscripts are highly accurate, and as a result, there are very few places where one could take issue with the Manly and Rickert ‘editing’ of the glosses. In the MLT, there are only three editorial decisions I would question. At II. 197, the second word should probably be spelled ‘Phoronei’, as in Hg, Dd, and other manuscripts, rather than ‘Phorenei’, as in El, Ch, and others. Similarly, at 295, ‘Protholomeus’ (in Ch, Dd, and Hg) is better than El’s ‘Ptholomeus’. At 771, ‘cui mens’ (in Ha5 and Hg) is perhaps preferable to ‘cuius mens’ (in Ch and El). More serious than any errors in transcription are the difficulties posed by the the manner in which the glosses are presented. Manly and Rickert state that they ‘shall attempt to group them in such ways as to aid in disclosing their significance’ (3: 483), but two examples will illustrate the inconsistencies and overcomplication that obscure their presentation. In the Knight’s Tale, the list of glosses is divided into at least ten distinct groups: Latin ‘long glosses’ which are widespread in the manuscripts3 ; explanatory glosses (and comments) shared by El, Hg, and others; interlinear explanatory glosses shared by El, Hg, and others; explanatory glosses in Hg but not in El; headings and comments in El and others but not in Hg; ‘major glosses’, usually headings and comments, in other manuscripts (not El or Hg), including a few from Ha3 and Se; lines where manuscripts (almost always other than El or Hg) repeat names mentioned in the text; remaining glosses in Ha3 , divided into longer headings and shorter glosses (including explanatory ones); remaining glosses (headings and comments) in Se; and finally, the ‘numerous instances’ of ‘Nota’ and ‘Nota bene’ (3: 484-90). Overlapping and conflicting here are several ways of classifying the glosses: by kind (for example, explanatory, comment, or heading) or by length; by which manuscripts contain them (the editors are concerned to distinguish any shared by El and Hg from those limited to other manuscripts, but also to highlight those that distinguish El and Hg from each other); and even by how the glosses are placed in the manuscripts (between the lines of the text or in the margin). In the account of the Wife of Bath’s Prologue, similar categories are invoked, but in rather different ways, so that the glosses are listed thus: El glosses, usually source glosses, with notes on other manuscripts which have the same or similar glosses at the same lines4; source glosses in Ad1-En3 (but not in El), with notes on where other manuscripts resemble these; headings shared by Ad1-En3 and Cn-Ma; source glosses, comments, and explanations in other single manuscripts or small groups; English headings in El, sometimes shared by one or two other manuscripts; headings in other manuscripts; and finally, all remaining glosses, including headings, names, comments, and explanations from El, Hg, and other manuscripts (3: 496502). The presentations for other parts of the Canterbury Tales likewise vary according to the character of the evidence, but some general patterns are apparent. Manly and Rickert usually leave for the end of their account of each tale those glosses occurring only in single manuscripts and small groups, especially those not in El or Hg, since these are of little or no value 87 for classification. Usually listed first are glosses in El and/or Hg, particularly source and explanatory glosses. This arrangement reflects an assumption which seems plausible enough=that source and explanatory glosses are the ones least likely to have been added independently, and thus are of most value for assessing manuscript relationships.5 Since some other kinds of glosses, such as headings and names, are more easily derived from the text, it is more difficult to be confident that those shared by several manuscripts had their origins in a common ‘ancestor’ manuscript. Probably one reason that Manly and Rickert give priority to source and explanatory glosses in El and/or Hg is that any glosses of early and widespread occurrence, and not easily derived from the text, are those most likely to have their origins early in the manuscript tradition. Clearly the origins of the glosses preoccupy Manly and Rickert, though here, as so often, their remarks on each tale do not seem to reflect a consistent theory of the origins of each of the manuscript groups. In their summaries for several tales, the editors comment on which glosses are likely to derive from a common source, but it is unclear whether they are always thinking of the same source, because they give it various names: ‘the archetype’ (3: 484, 490, 512); ‘the same fair copy of the original’ (3: 510); ‘O1’ (3: 511, 515); or, for less widespread glosses, ‘the ancestor of one line’ (3: 519), ‘the ancestor of cd’ (3: 504).6 Then, in their brief conclusion (3: 525-27), Manly and Rickert conjecture, rather tentatively, on the question of whether some of the glosses were Chaucer’s, and the possible models and motives for the glossing.7 They are sceptical of the glosses in WBP, MerT, and FkT, but say others ‘a priori seem possibly to have been due to Chaucer himself’=the KnT epigraph, and those in ClT, SNT, and MLT. The phrase ‘a priori’ evades, of course, essential questions of what their reasons are for accepting certain glosses as Chaucer’s and not others. They seem to be impressed by the fact that the MLT glosses reflect unusual learning and are widespread in the manuscripts; but they reject the El glosses on WBP, which include learned astronomical glosses like those on MLT, and they make no comment on the PardT glosses, which are almost as widespread as those in MLT. Nor do Manly and Rickert offer any explanation of why the ClT glosses, which they are inclined to accept as Chaucer’s, appear in almost exactly the same manuscripts as the MerT glosses, which they reject. The impression given here, as so often in this edition, is that Manly and Rickert devoted so much time and care to gathering their evidence that they were unable to interpret it fully; and that their conclusions, therefore, should be regarded as a tentative and inconsistent draft. What is especially surprising, in light of their obvious concern with how the glosses might help with classification of the manuscripts, is that Manly and Rickert never summarize their information on which manuscripts contain which glosses. Out of the welter of detail in their account, it is in fact possible to distinguish three groups of manuscripts. A clearly defined set of source glosses had entered the tradition by the time the earliest surviving manuscripts were copied, and the surest basis for dividing the manuscripts into these three groups is according to which of these early glosses they contain. This classification, however, is also valid for other 88 Canterbury Tales Project Occasional Papers I kinds of glosses, such as explanatory glosses and headings. In these features, too, the manuscripts I group together resemble one another more consistently than they do manuscripts in the other two groups.8 Here I list the manuscripts in each, together with the tales where source glosses enable one to distinguish the groups. For the sake of simplicity, I list only the verse tales where source glosses appear in El and/or Hg:9 1. Manly and Rickert groups c and d, and Ha4 : KnT, MLT, SqT, and PardT. 2. Manly and Rickert group a, Ad1-En3, Bo2, Ch, Ha5, Hg, and Ht-Ra2: KnT, MLT, SqT, and PardT; MilT, ClT, MerT, FkT, PhyT, and MkT; and, in fewer manuscripts, PrT, SNT, and MancT. 3.Ad 3, El, Ra1, and Tc2: KnT, MLT, SqT, and PardT; MilT, ClT, MerT, FkT, PhyT, and MkT; WBP, WBT, SumT, and additional glosses in FkT. How do these groups relate to the Manly and Rickert textual groups? The evidence of my first and third groups can be summarized briefly; the situation in the second group demands more sustained attention, and so I will discuss it last. For my first group, the evidence of the glosses is consistent with Manly and Rickert’s conclusions about which manuscripts are most similar in text. The variants in the glosses suggest that scribes copying these manuscripts generally took glosses from the exemplars they used for their text; where the variants in the glosses enable one to distinguish smaller groups, these groups coincide with those established by the textual variants=for example, Pw and Ha2-Lc. The transmission of the glosses in these manuscripts seems to have been rather straightforward, despite the fact that many were probably produced in the same milieu of London craftsmen, and there must have been many opportunities to compare and combine the material in multiple exemplars.10 The readings in the glosses also argue that while some of these manuscripts, such as Cp and Lc, are careful copies of their immediate exemplars, the c and d manuscripts are ultimately derived from a single scribal copy of the Canterbury Tales which contained some mistaken readings in its relatively few source glosses.11 My third group is quite different, for here glosses are shared by manuscripts not closely related in text. What this group demonstrates, therefore, is the variety of ways that scribes could gather material. The scribe of Ra1 apparently copied the text and glosses of Mc, and drew on El or a similar manuscript for additional glosses.12 Ad3 is a slightly different case, for here the scribe seems to have drawn on El or a similar manuscript for the text of a few tales, but for the glosses in many. My second group of manuscripts demands more detailed examination. For some manuscripts in this group, the evidence of the glosses confirms the conclusions of Manly and Rickert about the text. It supports the two pairs, Cn-Ma and Ad1 -En3 , and the relationship of these pairs to each other and to the earliest extant manuscript of the a group, Dd. On the other hand, the mid-century pair Ht-Ra2 seem to derive their glosses from Hg. Thus they may exemplify scribal habits like those reflected in Ad3 and Ra1, since Ht and Ra2 are both loosely related to the d group in text. Elsewhere in my second group it is not obvious what model of scribal behavior explains the similarities in glossing, for the manuscripts=Bo2, Ch, Dd, Hg=were not 89 classified together in the Manly and Rickert textual groups. I would like to consider more carefully what the glosses suggest about the relationships among these manuscripts. Bo2 dates from the second quarter of the fifteenth century, and its many unique variants make its ‘text’ of the glosses less accurate than those of Ch, Dd, El, and Hg. There is no reason to believe, however, that Bo2 is copying from any of these or from any other surviving manuscript. In MerT, there are eleven glosses at IV. 1311-85 from Albertanus of Brescia; all are in Bo2 , as well as in Ad3 , Ch, Dd, and El. Hg lacks one of these glosses, at IV.1380, and so none of these other manuscripts is derived from Hg. Bo2 is not copied from El either, because at IV. 1362, 1369, and 1376 it shares with Hg (and other manuscripts) readings which are superior to variants in Ad3 El Tc2 .13 That Bo2 is not a copy of Dd is made clear by the situation in ClT, where Bo2, like Ch, El, and Hg, contains at IV. 40 a gloss from Petrarch which is absent from Dd. There are a few variants in Bo2 which link it with other manuscripts in my second group, and these are scattered throughout the Canterbury Tales in a way which suggests steady association with this group. Some of these set it together with a manuscripts and Ad1-En3 and apart from Ch Hg: the wording of the headings in GP; the omission of the widespread gloss at I. 1164; the name of ‘Percius’ inserted in the gloss at V. 207. Yet other readings make Bo2 more like Ch, Dd, and Hg than Cn-Ma and Ad1-En3. Most obviously, Bo2 includes more of the early glosses than the latter four manuscripts. In ClT, Bo2 lacks four of the glosses from Petrarch, and part of a fifth, but Ad1-En3 and Cn-Ma agree in omitting ten; and in PhyT, Bo2 is one of only a few manuscripts to contain all three glosses at VI. 14, 16, and 240 (the others are Ad3 Ch Dd El Hg Ht), while Ad1-En3 and Cn-Ma have none of them. In sum, the glosses in Bo2 apparently are derived from a lost early copy that was also a source for the glosses in several other manuscripts in my second group; and the immediate exemplars of Bo2 were more complete and accurate than those used by Cn-Ma and Ad1-En3. This situation suggests, in turn, that the text of Bo2 deserves greater attention, despite the relatively late date of the manuscript. Although there is no space here to analyse the readings in the text of Bo2, we might at least recall the remarks of Manly and Rickert on this manuscript (1: 67): ‘Since Bo2 is ultimately derived from manuscripts of the best traditions, its readings are always worth consideration as supporting evidence, with due allowance for the great proportion of individual errors.’ The third-quarter manuscript Ch is even later than Bo2, but its ‘text’ of the glosses is both more complete and more accurate. Ch is particularly similar to Dd and Hg in the glosses it contains and in its readings, so it is best to consider all three together. These three manuscripts are linked most clearly by their inclusion of several glosses in PrT. In addition, they share with other manuscripts several readings, particularly in Fragment IV, which are superior to variants in Ad3 El Ra1 Tc2; thus the glosses of Ch are not derived from El. Nor are they derived from either Dd or Hg, as is shown most clearly by the situation in ClT. Of the 28 El glosses from Petrarch’s Epistola, Ch, Dd, and Hg each lack one; but each manuscript omits a different The Glosses and the Manuscript Groups 90 Canterbury Tales Project Occasional Papers I gloss=Dd that at 40, Hg 197, Ch 295. At V. 721, Ch gives the best reading, ‘Nec fonte’, whereas El has simply ‘Fonte’ and Hg, which is damaged, most likely read ‘Nec fronte’; that is, Ch, along with En3 and Ps, gives a correct reading where both El and Hg err. The evidence consistently suggests that the glosses in Ch, Dd, and Hg were derived from a lost copy with a highly accurate ‘text’ of the glosses, and that none of these shared immediate exemplars with El. This conclusion argues that Ch should play a greater role in efforts to establish the text of the Canterbury Tales.14 There are, I think, further lessons to be drawn from the evidence of Ch. First, the quality of this manuscript demonstrates what is too often forgotten, that while the earliest surviving manuscripts of this (or any) text may in general be more accurate, and useful for establishing a text, than later copies, there can of course be crucial exceptions. As a witness to a highly accurate lost exemplar, Ch seems more useful than others from the first half of the century which continue to receive more attention, such as La and Pw. Moreover, Manly and Rickert’s remarks on the textual relationships of this manuscript point to questions about their methods for classifying manuscripts=questions which already have been raised elsewhere.15 Manly and Rickert consider Ch in agreement with different manuscripts in different tales=most often they link it with Ad3 , a, El, Gg, and Hg=and ‘independent’ in several (1: 87-8). The impression created is that their method of classifying manuscripts by common error has failed them here, because the manuscript is so accurate that they have insufficient data for any firm conclusions. Rather than admit this, however, they assert that the texts in this manuscript ‘were picked up from many different sources’ (1: 87). But other kinds of evidence point to very different conclusions. The consistent similarities in glosses through much of Ch, Dd, and Hg argue strongly that all are indebted to the same complete (lost) copy of the Canterbury Tales. To the similarities remarked above may be added others that, by themselves, might carry less weight but can help to confirm the general pattern. Ch, Dd, and Hg agree in the unusual infrequency of their glosses in the prose tales; moreover, there are no page heads in Ch or Dd, and Hg, where many of the page heads are added by a second hand, gives weak evidence for the presence of such page heads in its exemplars. The order of the Canterbury Tales is a feature so subject to accidental and intentional change that it is best regarded as weak evidence for classification. But the similarities of Ch and Hg in this respect are distinctive and striking. I have highlighted them by adding slashes to the following outlines: Hg: I III / II SqT MerT FkT SNT / ClT VI VII IX X (corrected for misplacement of quires containing part of VII and IX). Ch: I Gam III / ClT VI VII IX / FkT II MerT SqT PlTale VIII / X. In Ch, Gamelyn was added in a separate stage of work, and so was probably not in the scribe’s first set of exemplars, and PlTale was added by a second, later hand. Taking account of this makes the similarities between the two manuscripts even more apparent; they are, in fact, the only two witnesses, among all surviving manuscripts, for the sequence I-III, and are likewise the only two for the sequence ClT-VI-VII-IX. It appears that the scribes of Ch and Hg drew on the same immediate exemplar, though each also must have 91 supplemented this by referring to other exemplars as well. This probability highlights even further the potential value of Ch as a witness to the text. Moreover, it has implications for the transmission of the text, because it argues against any hypothesis that the exemplars of Hg were either dispersed or edited beyond recognition shortly after Hg was produced.16 While I have been able to argue on the basis of only a limited body of evidence, the glosses draw our attention to Bo2 and Ch, and to their relationships with the much earlier and better known manuscripts Dd and Hg. The glosses suggest that these four manuscripts are linked by their connection with one or more lost but early, accurate, and complete manuscripts of the Canterbury Tales. Clearly there is a need for reexaminaion of the text in these manuscripts, and it seems to me likely that the Sheffield and Oxford Canterbury Tales Project can make this work both less taxing and more comprehensive. Notes 1The Variorum Chaucer volume devoted to the Squire’s Tale, for example, simply reproduces the Manly and Rickert list of glosses in that tale (Baker 1990, 82-3); but this list fails to record five of the manuscripts where the gloss at V.609 appears, including two manuscripts collated for the Variorum edition. In addition to those manuscripts listed at 3:511 in Manly and Rickert (and in Baker 1990, 83), this gloss is in Cp Ha5 Ph3 Pw Ry2. 2 There is not enough space here to provide an apparatus of variant readings, even for the MLT. Corrections and additions to the Manly and Rickert appendix would add further information about individual manuscripts but relatively little about their relationships. 3 In fact the ‘gloss’ at I. 859 is more properly regarded as part of the heading to the tale, and the gloss at I. 1164 is only four words long. 4The Manly and Rickert account of the WBP glosses is misleading because it overstates the number of manuscripts that derive their glosses from a common exemplar or ‘ancestor.’ The El glosses are shared only by Ad3 , Ra1 , and Tc2 ; other manuscripts, including Ad1 -En3 , preserve independent series of glosses which sometimes closely resemble the El glosses in their wording. These series, provided by glossators who were, in all likelihood, unaware of one another’s work, are preserved in manuscript groups c and d; Ad1-En3; Ne; and Mc-Ra1 . 5The consistency in wording in the manuscripts of most of the source glosses, and the rarity of the explanatory glosses, demonstrate that they were not added coincidentally but derive from no more than a few ‘ancestor’ manuscripts. 6 There is additional evidence that Manly and Rickert gave special attention to glosses that may derive from O1: ‘The Petrarch glosses were certainly inherited by Hg Bo2 El’ (3: 508); the MLT glosses appear in ‘representative manuscripts of all groups’ (3: 492); the gloss at VII.694 is ‘the one occurring in the most manuscripts’ in Sir Thopas (3: 519). 7Charles A. Owen Jr. (1982) argues that even the earliest glosses are scribal in origin; but I do not believe that his arguments outweigh the evidence in The Glosses and the Manuscript Groups 92 Canterbury Tales Project Occasional Papers I favour of Chaucer’s authorship offered by Germaine Dempster (1943), Robert A. Pratt (1966), and Robert E. Lewis, in his edition of Lotario dei Segni’s De Miseria Condicionis Humane (1978, 32-9). For some hypotheses about the motives for the glosses, see the articles by Graham Caie (1975 and 1984), Susan Schibanoff (1988), and Thomas J. Farrell (1989). 8An apparent exception to this generalization is a set of explanatory glosses generally limited to Ch Dd El Hg. But such glosses, which were usually written between the lines of text, seem to have been the kind most readily omitted by scribes as they copied an exemplar. These four manuscripts, therefore, amount to reasonably good evidence for the presence of these glosses in my second and third groups; these explanatory glosses should not be taken as evidence for a contradictory scheme of classification that would distinguish Ch Dd El Hg from all other manuscripts. 9 In other words, I have omitted glosses limited to single textual groups, small groups, or single manuscripts=unless that single manuscript is El or Hg. In the prose tales, headings, comments, and names, which are of less value for classification, far outnumber source and explanatory glosses. The limited value of this complex body of evidence, therefore, does not justify the effort required to present even a general picture of the evidence in all manuscripts. Some special cases, where manuscripts are strikingly similar in their glosses (or lack thereof) for the prose tales, are discussed later in this paper. 10 Doyle and Parkes (1978) identify Cp and Ha4 as by the same prolific London scribe. There is additional codicological evidence for contacts among those producing c and d manuscripts; the scribe of En2 modelled his work on that of the Cp-Ha4 scribe with unusual closeness, and Lc seems to have been produced by some of the same people as Pw. Pace Hanna (1988, 38), to judge by surviving manuscripts, the d text seems to have exerted most influence between late in the first quarter of the fifteenth century and the end of the second quarter. Ha2 is one d manuscript which was obviously corrected and supplemented by reference to several exemplars. 11 The clearest evidence in the glosses for the close relationship of c and d is a series of shared scribal variants in the MLT glosses. Additional evidence can be found in the several scribal glosses in WBP which are limited to these two textual groups; these are easily derived from the text, and thus have no claim to Chaucer’s authority. The close relationship of c and d is reflected in many of Manly and Rickert’s statements in their edition, and is argued in detail by Dempster (1948), who hypothesizes that the d text was produced by a conflation of the c ancestor and Hg. 12 See Manly and Rickert 1: 451 for the relationship of Mc and Ra1 . 13 Those readings are: IV. 1362, ‘Jacob enim’ for ‘Jacob’; 1369, ‘Et Abigail’ for ‘Abigail’; and 1376, ‘superius benigna’ for ‘benigna’, and ‘est crudelius’ for ‘crudelius est’. 14 Manly and Rickert were clearly impressed by the text of Ch (1: 87-8); Hanna includes readings from Ch in the Textual Notes of The Riverside Chaucer, and remarks on its value in his article of 1991. 15 Blake (1983) criticizes Manly and Rickert for grouping manuscripts according to variants within the line. 93 16 Owen (1991, 7) conjectures that the Hg exemplars were dispersed; Blake (1985) argues throughout that the Hg exemplars were Chaucer’s original papers, and that they were modified in successive stages by scribes experimenting with text, links, and order. Hanna (1989) and Dane (1991) offer objections to Blake‘s theory. Bibliography Baker, D. C. (Ed.) 1990. The Squire’s Tale. A Variorum Edition of The Works of Geoffrey Chaucer. Volume II, Part Twelve. Norman and London: University of Oklahoma Press. Benson, L. D. (Ed.) 1987. The Riverside Chaucer. 3rd edition. Boston: Houghton Mifflin. Blake, N. F. 1983. ‘The Editorial Assumptions in the Manly-Rickert Edition of The Canterbury Tales.’ English Studies 64: 385-400. =. 1985. The Textual Tradition of the Canterbury Tales. London: Arnold. Caie, G. D. 1975. ‘The Significance of the Early Chaucer Manuscript Glosses (with Special Reference to the Wife of Bath’s Prologue).’ Chaucer Review 10: 350-60. =. 1984. ‘The Significance of Marginal Glosses in the Earliest Manuscripts of The Canterbury Tales.’ In Chaucer and Scriptural Tradition, ed. David Lyle Jeffrey, 75-88. Ottawa: University of Ottawa Press. Dane, J. A. 1991. ‘Copy-Text and its Variants in Some Recent Chaucer Editions.’ Studies in Bibliography 44: 164-83. Dempster, G. 1943. ‘Chaucer’s Manuscript of Petrarch’s Version of the Griselda Story.’ Modern Philology 41: 6-16. =. 1948. ‘A Chapter of the Manuscript History of the Canterbury Tales’. PMLA 63: 456-84. Doyle, A. I., and Parkes, M. B. 1978. ‘The Production of Copies of the Canterbury Tales.’ In Medieval Scribes, Manuscripts and Libraries: Essays Presented to N. R. Ker, ed. M. B. Parkes and A. G. Watson, 163-210. London: Scolar Press. Farrell, T. J. 1989. ‘The Style of the Clerk’s Tale and the Function of its Glosses.’ Studies in Philology 86: 286-309. Hanna, R. III. 1988. ‘Authorial Versions, Rolling Revision, Scribal Error? Or, The Truth about Truth.’ Studies in the Age of Chaucer 10: 23-40. =. 1989. ‘The Hengwrt Manuscript and the Canon of The Canterbury Tales.’ English Manuscript Studies 1100 - 1700 1: 64-84. =. 1991. ‘Presenting Chaucer as Author.’ In Medieval Literature: Texts and Interpretation, ed. T. W. Machan, 17-39. Medieval and Renaissance Texts and Studies, volume 79. Binghamton, N. Y.: Center for Medieval and Early Renaissance Studies. Lotario dei Segni (Pope Innocent III). 1978. De Miseria Condicionis Humane. Ed. Robert E. Lewis. The Chaucer Library. Athens, Ga.: University of Georgia Press. Manly, J. M. and Rickert, E. 1940. The Text of the Canterbury Tales. 8 vols. Chicago: University of Chicago Press. The Glosses and the Manuscript Groups