The Language, Hands, and Interaction of the Two Scribes of the Egerton 2726 Chaucer Manuscript (En1)

40 Canterbury Tales Project Occasional Papers II Figure 2. En1, folio 198v Figure 3. En1, folio 11r For Figure 1, and for the Linguistic Map, see Plate II in the front pages. 41 The Language, Hands, and Interaction of the Two Scribes of the Egerton 2726 Chaucer Manuscript (En1)1 Daniel W. Mosser British Library, Egerton MS 2726 (En1), containing The Canterbury Tales, (defective at the end, from I1084), is classified by Manly-Rickert as ‘the most constant member of the subgroup Dd, the earlier branch of Group a, and is therefore the best representative of the Dd subgroup’ (Manly and Rickert 1940 1:131.)2 En1 has been used by the Variorum as one of the landmark MSS for those texts and parts of texts lacking in Dd (e.g., GenPro 1-252, 505-758; PsT; and several other folios interspersed throughout the text, including substantial portions of KnT.) Thus, although En1 is a much later witness than those manuscripts routinely accorded landmark status, it does serve an authoritative function as a kind of ‘textual surrogate’ for Dd and for the a (or α) text. However, it is not the text of En1 that I wish to focus on here, but rather the remarkable evidence the manuscript provides of the collaboration of two scribes, writing in very similar hands in a strikingly similar and marked form of East Anglian Middle English. The primary purposes of this essay, then, are: 1. to identify the characteristics of hand and dialect that will allow us to distinguish the two scribes (and given the similarity of the two in both categories, this will require some close scrutiny of detailed features); 2. to examine the evidence from which Manly-Rickert inferred that persons other than the scribes proof-read the MS and that the apparent cessation of corrections ‘after f.119….seems to indicate a new policy—perhaps a new master of the shop’ (1:133); 3. to provide an alternative picture of the processes by which this manuscript was produced. Physical description The text is written on parchment, sewn in thirty-three quires, all gatherings of eight except for Quire [7], which is a gathering of six.3 The parchment has been trimmed, as evidenced by cropped flourishes on initials at the top of pages (e.g., 72v, 94v, 100r, 100v.) A page measures 35 x 19.5 cm; the written space is 22.5 x 13.5 cm. In the first part of the MS (through fol. 48), the work of Scribe 1, pages are ruled in reddish crayon, in single columns of 40 lines per page. There is no ruling visible in the work of Scribe 2, whose pages contain between 40-41 lines of text in single columns. The ink used for the text is generally a brownish-russet color, possibly made from an iron-base paste, though the second scribe sometimes uses a darker ink (possibly carbon-based), THE CANTERBURY TALES PROJECT OCCASIONAL PAPERS II, 1997, 41-53 42 Canterbury Tales Project Occasional Papers II especially for corrections.4 As the Video Spectral Comparator at the British Library was effectively able to separate the two inks (i.e., one could be masked out under certain light), they are clearly different, though at times they may have been mixed.5 The collation of En1 (excluding the modern fly leaves) is as follows : QUIRES [1]-[6]8, fols. 1-48 QUIRE [7]6, fols. 49-54 QUIRE [8]8+8 ([8]1+χ8) (i.e., the original 8 parchment leaves plus the inserted Gamelyn), fols. 55-70 QUIRES [9]-[32]8, fols. 71-262 QUIRE [33]8+1 (±[33]9) (i.e., the paper leaf containing the conclusion to the Parson’s Tale and the ‘Retraction,’ replacing what must originally have been a tipped-in singleton parchment leaf), fols. 263-71 The insertion of the eighteenth-century Gamelyn manuscript has produced a somewhat confusing set of two foliations and one partial pagination. The oldest of these, which omits the folios containing Gamelyn, must be the one in blue pencil, which occurs between the other two numberings in the upper right-hand corner of each recto. A pagination in red ink is top-most and begins on fol. 7r with ‘13.’ It is recorded on both recto and verso through p. 106 (where Gamelyn is inserted), following which only rectos are marked; after p. 123, on the recto following the Gamelyn insertion, it stops altogether. The bottom-most numbering is a black pencil foliation that includes Gamelyn. Thus the first recto following Gamelyn is foliated as ‘56’ in blue and ‘64’ in black pencil. My references, except where indicated, are to the most recent (pencil) foliation. Catchwords survive on the last verso of each quire starting with fol. 102. The similarity of the two scribes’ hands—especially in proximity—led Eleanor Hammond to note despairingly that ‘[t]he latter part of the MS is apparently in a different hand, but I cannot say where it changes’ (1908 [1933], p. 174.) Scribe 1 copies folios 1-49r, GenPro through MilT (which ends on the eighth line of 49r), and also copies the first page (fol. 111r) of Quire [14],6 with the exception of the first two lines (Figure 5.)7 Scribe 2 takes over on fol. 49r, copying from RvPro through the end of the MS, with the exception of that portion of fol. 111r copied by Scribe 1. Conceivably, the scribes might have set out to divide up the exemplar for simultaneous copying, with the first ‘chunk’ originally consisting of GenPro through RvT, ending in the MS on 54v, the end of Quire [7], the anomalous gathering. When Scribe 1 finished MilT (at the beginning of Quire [7]), he simply handed over the job of filling out the quire to Scribe 2, who had probably already copied the section of text beginning with CkPro at the top of 55r. The calculations made to do this (similar to a printer’s technique of ‘casting off’) were close enough that they were left with only one extra bifolium in Quire [7] (i.e., they ended up needing only six instead of eight folios for that gathering.) Conceivably, Scribe 2 copied the first two lines of fol. 111r, at the beginning of Quire [14], as a place marker, a cue to remind the first scribe where to take up copying again. Why the first 43 scribe copied only the remainder of the recto of this folio and then disappeared from the project is unclear.8 Both scribes employ a hybrid anglicana script that includes in its repertoire both secretary and anglicana forms for several graphs. Figure 1 reproduces the top portion of fol. 49r, where Scribe 1 finishes MilT and where Scribe 2 begins copying RvPro. A very slight difference in the size of the two hands can be detected here, with Scribe 2’s hand being slightly the larger of the two. As there are many similarities between the two hands, especially in proximity, a more detailed description is required to distinguish their work more clearly. The two scribes differ in the execution of minim strokes, for example in the letters m and n. Scribe 1’s minims (e.g., Figure 3, l. 3: ‘Vn to’ ; ‘Thomas’) are a closer approximation of the textura, or ‘display script,’ style than Scribe 2’s (thus a Bastard Anglicana); that is to say that the strokes are more meticulously formed than those of Scribe 2, though it will be necessary to examine another portion of Scribe 2’s work for this to be apparent, as it would appear that Scribe 2 is imitating the hand of Scribe 1 at this point in his work. In MkPro (Figure 2, l. 3: ‘now ’; ‘comth’ ; ‘remem braunce’ ), for example, the minims have virtually no ‘feet’ at all as compared to those of both scribes on fol. 49r, and of Scribe 1 elsewhere. Another clear and consistent difference can be detected in the execution of the anglicana a graph. Scribe 1 forms two compartments, of about equal size, with the top lobe angling out on the left side and forming an ‘overhang’ above the bottom compartment (Figure 3, l. 1: ‘gadred’ ), similar to the example Petti provides of the anglicana formata letter forms (Petti 1977 p. 14, fig. 12.) Scribe 2, in contrast, forms a single compartment, which is then divided in two with a cross-stroke, creating a top compartment that is smaller and narrower than the bottom one (Figure 2, l. 4: ‘Have’ ), after the fashion of the hybrid anglicana letter form (Petti, p. 15, fig. 14.) Both scribes also employ secretary single-compartment forms of a: Scribe 1 has two versions of the secretary a, one formed with three strokes, creating a very ‘horned,’ angular letter-form (Figure 1, l. 2: ‘man’ ), and a second, consisting of two strokes, which results in a much rounder duct (Figure 4, l. 1, ‘art’ .) Scribe 2 also has two forms of anglicana a: a three-stroke, angular form (Figure 1, l. 2 of RvPro, ‘and’ ), and a more rounded, two-stroke form (Figure 1, l. 4 of RvPro, ‘and’ .) Both also have two basic forms of the letter w, again providing a poor basis for distinguishing one scribe from the other (Scribe 1, Figure 3, l. 2: ‘we’ , and l. 3: ‘wat yng’ ; Scribe 2, Figure 2, l. 3 of MkT: ‘was’ and l. 1 of MkT: ‘bewaile’ .) The second form consists of three strokes: a rounded loop on the left, which sometimes fails to connect to the base of the other part of the graph (for example, five lines up in Figure 2, ‘Beware’ ), followed by a v-shaped stroke that begins with a billowing flourish at the top, and completed by a stroke that forms the top lobe of the B-shaped element on the right. This latter stroke sometimes fails to connect with the body of the graph on the left. This description essentially accounts for Scribe 1’s form as well, except that the proportions of the various elements are not identical to those in Scribe 2’s graph.9 Egerton 2726 44 Canterbury Tales Project Occasional Papers II Figure 4. En1, folio 44v Figure 5. En1, folio 111r 45 Both scribes prefer the looped-ascender anglicana form of d (notable in both hands for the uprightness of the ascender), and both alternate between the anglicana ‘double-diamond’ g and the ‘tailed’ secretary form, though Scribe 1 prefers the anglicana form and Scribe 2 prefers the secretary form. In Figure 3, l. 1, Scribe 1’s characteristic form of the two-compartment g can be seen in ‘gadred’ and ‘to gider’ . Scribe 2’s two-compartment g can be seen in Figure 1, the ninth line of RvPro: ‘gruccœ’ . The economy of Scribe 2’s duct creates a very rounded appearance, as contrasted with the more angular duct of Scribe 1’s graph. There are no examples of Scribe 1’s secretary ‘tailed’ g in the material reproduced here; it is used infrequently, and is very similar in appearance to the graph executed by Scribe 2 in Figure 1, l. 9 of RvPro: ‘light’ . Another variation (also Scribe 2) occurs in the second-tolast line of fol. 49r (not included in the material reproduced in Figure 1), in ‘gledes’ .10 Sigma-shaped s appears in both initial and final position in both scribes’ work, and long s appears in initial and medial positions: Scribe 1, Figure 3, l. 3, ‘seint’ ; l. 4, ‘hors’ ; Scribe 2, Figure 1, l. 1 of RvPro, ‘cas’ ; l. 3 of RvPro, ‘sei∂’. Long s can be seen in Figure 3, l. , ‘lest·’ (Scribe 1), and Figure 1, l. 2 of the RvPro, ‘Absolon’ (Scribe 2.) The head stroke on this graph and on f provides a useful diagnostic for distinguishing the work of the two scribes. For example, in Figure 5, note the ‘droopy’ headstroke in l. 1, in ‘faders’ and l. 2, ‘Grisel∂’ (Scribe 2) as contrasted with the more horizontal headstroke (formed by a broken stroke at the join with the vertical stroke) in l. 3, ‘oft’ , l. 5, ‘fe ´’ , and ‘aspye’ (Scribe 1.) Scribe 1’s emphatic form (two long f graphs in combination) consists of one graph with a horizontal head stroke, ligatured with a second graph that has a more curved, droopy stroke: for example, the first word in Figure 1 and the first word of line 3 in Figure 5 . In contrast, Scribe 2’s emphatic form consists of a pair of ‘droopy-headed’ graphs: (last line of Figure 2.) Both also employ a B-shaped (‘kidney’) s in final position: Figure 3, l. 3, ‘Thoma s’ (Scribe 1); Figure 2, l. 3 of MkT, ‘was’ (Scribe 2.) Note that Scribe 2’s B-shaped s is made with one continuous stroke with a very rounded duct. In the ‘was’ of Figure 2, the first stroke begins as a ligature continuing from the cross-stroke of the two-compartment a. Thorn ( ) and th are used interchangeably. Scribe 2 tends to add more flourish, especially to the tails of y’s and h’s. The overall impression created by Scribe 1’s hand is of an upright orientation, while Scribe 2’s often has a rightward slant, and thus appears more cursive, an impression that is reinforced by the formation of the minim strokes, after the fashion discussed earlier. Further, in part because of a tendency to execute the more modeled, texturastyle graphs, Scribe 1’s duct appears heavier, more angular and ‘chunky,’ in comparison to Scribe 2’s more slender, round, and cursive appearance. Egerton 2726 46 Canterbury Tales Project Occasional Papers II Language The two scribes of En1 exhibit remarkably similar linguistic features. In the Linguistic Atlas of Late Mediaeval English (hereafter ‘LALME,’ McIntosh, Samuels et al. 1986), LP (=‘Linguistic Profile’) 6150 was derived from En1. However, the ‘tranche’ used to construct the profile covered only the PdT extract transcribed by the Chaucer Society in their parallel-text specimens (Zupitza 1890.) The data-base is thus limited to the language of Scribe 2 and covers only a fraction of his work. In the LALME, the MS has been localized to what appears to be Colchester, in Essex.11 My own analysis of both scribes’ work throughout the MS reinforces the LALME findings. The LPs for the scribes can be found online at http://ebbs.english.vt.edu/exper/mosser/catalogue/en1lps.html. Both scribes’ main form for <SUCH>—’soch’—is recorded in the LALME as occurring in the south, in the counties of Sussex, Hampshire, Dorchester, and Somerset. It also occurs as a minor form in Hertfordshire, in the East Midlands (in a text whose main form is ‘swich’) and, of course, is represented as occurring in Essex in the work of En1 Scribe 2. This form, then, might suggest a southern provenance. The two scribes display different primary spellings for <YET>. Scribe 1’s preferred form, ‘yit,’ has a widespread distribution throughout the Midlands. Indeed, many of the dialect forms in En1 enjoy such a wide distribution. However, Scribe 2’s form, ‘yitte,’ is attested only in Middlesex, Northamptonshire, Surrey, and Essex (where Egerton Scribe 2 is the only witness for this form.) Neither of the two forms overlaps in its distribution with the scribes’ forms for <SUCH>, except perhaps in Hampshire, where only the more common of the two forms, ‘yit,’ co-occurs. Thus, on the basis of forms for <SUCH> and <YET>, there is no compelling evidence to localize the manuscript in Essex, or even in the East Midlands. But persuasive evidence for an Essex localization is provided by other forms in the scribes’ repertoire. Two spellings for <SINCE> compete for primary status in each scribe’s work: the main form, ‘seth,’ and a minor variant, ‘sen.’ ‘Sen’ occurs mainly as a northern form, dipping down into Norfolk and Suffolk. It also occurs in the Central Midlands and is represented as the main form in two LPs from Surrey. ‘Seth’ has a more restricted distribution, occurring mainly in the West Midlands and the southwest, and in Essex and south Suffolk. The distribution of ‘yitte’ rules out the West Midlands as a possibility and both the forms for <YET>, as well as the distribution of other forms, serve to rule out the southwest. It should also be noted that there are no instances of u or uy as a reflex of Old English y, a common test for western dialects.12 The evidence of these forms thus suggests the East Midlands, perhaps more specifically East Anglia. The form that occurs in both scribes’ work as a main form, with by far the severest restriction in distribution, is ‘ecch’ (for <EACH>.) The LALME records two occurrences as the main form for texts in northern Essex (in addition to En1), one occurrence in south Norfolk, and one occurrence in Devonshire. The only place where the majority of the forms discussed above co-occur, as well as 47 the numerous other very common forms in En1, is in northwest Essex, in the vicinity of Colchester (see MAP; Plate ii in the front pages.) Several other forms should also be mentioned. The form ‘werld’ occurs in both scribes’ work, as a rare variant in the work of Scribe 2, and as a main form for Scribe 1, occurring 10 times as compared to 11 occurrences of ‘world(ly.)’ The strongholds for ‘werld’ are southern Essex, in the south, and an area in the north, running from Norfolk up the east coast through Lincolnshire, and northward. Colchester is about midway between the borders of these two areas. Scribe 2 also has the unusual spellings for <NOWHERE> and <ANYWHERE> of ‘nougher,’ ‘noughere,’ and ‘ougher.’ Also in the work of Scribe 2 a shift can be detected, beginning with the B2 Fragment (fol. 171v) in the scribe’s preferred spellings for <AGAIN(ST)>. Prior to that section, the form ‘ageyn’ is used predominantly for both <AGAIN> and <AGAINST>. Subsequently, the forms ‘ayein’ (<AGAIN>) and ‘ayeinst’ (<AGAINST>) predominate, especially in PsT. The LALME records the form ‘ayeinst’ only in the profiles of this scribe and in LPs from Hertfordshire (LP 6540) and Rutland (LP 554.) ‘Ayein’ is recorded in the profile for another Essex scribe (LP 6120), and in LPs from Hertfordshire, Kent, London, Shropshire, Suffolk, and Worcestershire. (The scribe of Cambridge Dd.4.24—the MS above En1 in the A or a family—prefers the form ‘a-geyn’ for both words.) A possible explanation for the shift is that the scribe copied a similar form fairly faithfully from his exemplar through PdT, and then began to activate his own favored forms. In general, however, these two scribes appear routinely to ‘translate’ texts into their own dialect in the course of copying. A better explanation is that the set of exemplars used to produce En1 derived from different sources, reflecting different spelling systems. ManlyRickert’s collations determined that En1 and Lincoln Cathedral Library MS 110 (Ln) are affiliated in the texts for the B2 Fragment and PsT (Manly-Rickert 1:132.) It may be that Ln shares only these texts with En1 because they consisted of booklets, or at least gatherings, that were combined with other exemplars to produce the a exemplar and then, sometime subsequently, were dispersed. If this was the case, and if those exemplars shared with Ln contained the ‘ayein(st)’ spellings, then the shift in preferences by Scribe 2 might be more readily explained. There is, unfortunately, nothing in the surviving evidence of provenance to corroborate the localization suggested by the dialect evidence. Manly-Rickert’s account of this evidence (1:134-5), and the narrative they construct to explain it, goes about as far as seems likely to be plausible (perhaps too far in suggesting the MS might actually have been owned by the Cobhams.) The earliest marks of ownership are from the fifteenth century: an owner or reader practicing a salutation to ‘my lord Cobham’ (fol. 158v) suggests an association with the powerful Cobham family of East Chalk, Cowling (or Cooling) Castle, and other locations in northwestern Kent. All of the subsequent marks of ownership also suggest a Kentish provenance. Egerton 2726 48 Canterbury Tales Project Occasional Papers II Evidence of supervision and scribal interaction Manly-Rickert observe that there are numerous corrections by several persons and in different inks, some by the scribe over erasures, others by a contemporary, also over erasures. There are some insertions above the line by a slightly later hand, and a few underdottings of words. (1:131) They also assert that after fol. 119 ‘the corrections almost cease’ (1:133.) There are, however, a number of corrections after this point, e.g. fols. 136v, 4 up, 140v, l. 5, 145v, 9 up, 168v, 18 up, and 172v, 16 up. Those corrections that occur in the portion of the MS copied by Scribe 1, when they are not made by him, are made by Scribe 2 in the role of proof-reader. If we look at two places where whole lines have been inserted in the margins to correct earlier omissions—at folios 11r (Figure 3), the very end of GenPro, and 44v (Figure 4), in MilT—we find the proof of this. In the first example, we can see the spelling of <IF> as ‘yf’ , the spelling preferred by both scribes, and well-attested in northern Essex. Both scribes place a dot over their y’s, though Scribe 2 is far more consistent in this regard.13 An example of ‘yf’ in Scribe 1’s hand can be seen in the line directly above the inserted line. In Figure 1 numerous examples of Scribe 2’s y can be seen; it is formed in just the same way as the y in the correction (e.g., the twelfth line of RvPro, the first word, ‘yf’ .) As detailed previously, in the f as executed by Scribe 1’s hand, the headstroke rarely curls downward, certainly not in the exaggerated fashion of Scribe 2’s form, but is more typically horizontal. Scribe 2’s f always has a downward-curling headstroke (e.g. Figure 1, l. 1 of RvPro, ‘folk’ .) The remaining characteristics of the corrector’s hand also match those of Scribe 2; the flourished v is commonly found on the top line of a page or in an incipit or explicit; the w is unusual for Scribe 2, but similar forms can be found, especially line-initially (e.g., fol. 97v, 6 up, ‘whan’ [not pictured].) Paleographically then, the corrector can be identified tentatively as Scribe 2. Additional support for this assertion appears on fol. 44v (Figure 4.) In addition to the handwriting correspondences, in this correction we have the very unusual spelling ‘koth’ for Chaucer’s <QUOD> (‘kotœ’ is Scribe 2’s usual form) to help us make the identification certain: ‘+ kotœ oo is sely man I am no labbe.’14 (Four lines below the insertion point we can see Scribe 1, the main scribe at this point, writing the more usual brevigraph for the word, ‘q∂.’ In Figure 1, l. 10 of MilPro, the form ‘kotœ’ can also be seen, again in the hand of Scribe 2.) Based on this identification, we can say that Scribe 2 acted as a proof-reader for Scribe 1 and corrected the first 49 folios of the MS in addition to his own work as copyist in the remainder of the MS. Thus, for example, where Manly-Rickert and the Variorum (citing Manly-Rickert) refer to a correcting hand other than that of Scribe 2 (e.g., GenPro 338; Andrew and Ransom 1993, p. 73), I hold that it is the hand of Scribe 2. While one could argue, especially considering the similarity of the two scribes’ hands in the MS, that the handwriting evidence does not compel identification of the corrector with Scribe 2, it would be a very great