University of Virginia Library

The preliminary work on this problem was done some years ago in collaboration with Dr. James W. Alexander, to whom I am greatly indebted throughout this paper.

Chaucer, p. 1039.

H. F. Heath in the Globe Chaucer, p. li; F. N. Robinson, loc. cit.; C. F. Bühler in "A New Lydgate-Chaucer MS.," MLN, LII(1937), 5-9.

For convenience I shall not distinguish between the MSS and Caxton's printed version.

Heath: (1) F; (2) A Cx Ff H H2 P. Robinson: (1) A2 CC2 F Ff H H2; (2) A Cx P. Bühler: (1) A CC2 F Ff H2 P: (2) A2 H M. (See below for the meanings of the sigils.) Although the alphabetical arrangement of the sigils shows most imperfectly the many differences between these classifications, one will note the flexibility with which the MSS are shifted from group to group. Obviously at least two of these classifications must be largely incorrect.

Dr. Bühler's classification is based mainly on the variants in lines 5, 11, 16, and 20; Robinson's (a tentative classification) and Heath's are published almost without explanation.

Note that Robinson omits M and CC1 (both unavailable to him); Bühler, Cx and CC1 (the latter unavailable to him); and Heath (whose classification was published in 1898), A2 CC1 CC2 M (some of the errors in Heath's classification probably arise from the small number of MSS which he used, but cf. Flügel, "Kritische Bemerkungen zur Globe-edition von Chaucers Werken," ESt, XXVII [1900], 60). CC1 (a transcription of which is printed in my "Four Unpublished Chaucer MSS" [forthcoming in MLN]) and the Black Letter texts had not been classified until my study.

Most of them have been published by the Chaucer Society (for A F Ff H H2 P, see Parallel Texts, pp. 448-49; for A2, More Odd Texts, p. 41; for Cx, Supplementary Parallel Texts, p. 155), and I have relied on their transcriptions. In addition, I have used the transcription of M published by Dr. Bühler (op. cit., 6-7) and the transcription of CC2 published by H. N. MacCracken ("An Odd Text of Chaucer's Purse," MLN, XXVII [1912], 228-29). The MSS are all from the Fifteenth Century (Cx was printed ca. 1477); A and F are from the first half. Except for CC1, the dating of which is necessarily my own, I have relied on generally established dates. (Although questions of date are not important in the classification of the Purse MSS, I might point out that the MS. tree which I propose preserves chronological probability.)

"Two British Museum MSS," Anglia, XXVIII (1905), 1-28. The evidence which Miss Hammond used, however, is completely outside the Purse MSS.

Henceforth called "shared unique readings." These readings are in A2, of course, but at the moment that fact is irrelevant (cf. fn. 10 below).

As a general thing, the phrase "the reading of the rest of the MSS" means exactly what it says. To avoid an unnecessary complication, however, I exclude from consideration the readings of MSS already considered provided they are in the same line of descent as the MS. being discussed. For instance, it is at present of no significance whether A2 agrees or disagrees with the rest of the MSS. (The variant readings of A2—as well as those of the other MSS—may be found in the Table of Variants below.)

This useful abbreviation, borrowed from the Chicago Chaucer, means, of course, "the exclusive common ancestor [of H and M]."

On the basis of this reading Dr. Bühler groups CC2 with A P (he does not consider Cx). The omission, however, is of the sort that any scribe might well be tempted to make, since the line is metrically too long and the yf is not necessary to the sense (see below for a further discussion of this variant). In view of the quite strong evidence in lines 5 and 11 for grouping CC2 with a, one is practically forced to consider the omission of yf as accidental coincidence.

CC1, in the same book as CC2, contains the last stanza and the envoy. Although it is of a different provenance from CC2 (see section VI), and apparently in a different hand, there may be some connection in both texts' reading must instead of mote.

CC1 reads as in this worde downe here. The agreement in word-order with H2 seems nonsignificant (see fn. 17 below).

Since the Thynne print is clearly copied from Cx (see section V below), which reads Syn, this reading may have little significance.

Fortunately, as they seem never to have been investigated carefully (even Thynne was unclassified before this study, although the classification is simple [cf. Robinson, op. cit., p. 1039]).

The evidence for this line is as follows:

As we shall see, we must assume that F Ff form a group independently descended from x; hence we must assume the reading of Cx P A to be that of x (note also the reading of A2 H). CC1's reading, however, leaves us with two additional problems: the agreement in word-order with H2 and the agreement in the variant worde with the Stow print. For the first of these I can offer no explanation except to remark on the general badness of CC1's text, a badness which suggests memorial transcription. I am forced to regard the agreement in word-order with H2 as accidental in view of the rather strong evidence for grouping CC1 with Cx P in l. 24. The second agreement is even more remarkable when considered with the next line, for there CC1 reads tresour with Stow. Yet we know that Stow's reading derives from the 1545(?) Thynne. It is chronologically impossible, of course, that CC1 be descended from Thynne or Stow, and there seems to be no reason for thinking that Thynne or Stow utilized CC1, since their texts are otherwise in close agreement with their printed predecessors. The explanation must lie in the character of the variants. Both tresour and worde make a sort of obvious sense, and hence both may be sophistications; or they may be misprints in Thynne and Stow, graphic errors in CC1 (note that CC1's scribe drops a letter in shave, l. 19, reading shae). (Professor Hench has called my attention, since I wrote this note, to the fact that word is commonly found for world in Middle English texts (cf. NED). Hence the form in CC1 is quite probably only a variant spelling.) Such striking agreements as the tresourworde pair show how necessary it is to consider all variant readings in constructing a tree rather than simply those that seem significant.

Apparently meaningless (the writing is clear).

The omission of yf must be considered non-significant for grouping CC2 with A P in view of the evidence set forth in section III.

CC1's reading perhaps derives from a spelling all har mys, since in the non-book hands of the period har and my do not look unlike.

F For I am shave as nye as is a ffrere; Ff For I am shaue as ys any frere.

To be considered as significant only with the reservation expressed in fn. 15 above.

There are many differences in spelling, of course; but since no edition can pretend to reproduce Chaucer's spelling, the fact that the three texts are spelt somewhat differently is not important. My procedure has been to follow F's spelling rather rigidly, departing from it only when it is unique (but making no distinction between y and i and u and v: hence sovne instead of soune in l. 9). Skeat's departures from F's spelling, incidentally, seem to follow no consistent plan: for instance, he adopts the spelling of Cx P, purs, in l. 15, although the spelling of F, purse, is found also in A A2 [CC1] H [M]; but in l. 1 he retains F's purse, although Cx Ff P spell the word purs. Skeat's through for thurgh, l. 17, is found in no MS.

My reading, metrically less regular than Skeat's and Robinson's, perhaps needs a word of comment. Although yf is omitted by some MSS (A CC2 Cx P), it is found in two of the three branches of the tree and hence must be assumed to have been in the original. Furthermore, it would seem more probable that a scribe would emend to make the line more regular than emend to make it less regular. (It may be well to remind the reader that the x of the tree is only the archetype of the eleven Purse MSS; that it is not necessarily identical with Chaucer's own copy of his poem [indeed, most probably is not exactly identical]; and hence that the x of the tree may itself have contained errors.)

One can never know, of course, what the final text will be until the work is done.

Op. cit., pp. 8-9. Since Dr. Bühler is arguing from a unique reading—and the assumption is usually made that unique readings are per force to be regarded as of scribal origin—, the reader should know that in Dr. Bühler's tree the Morgan MS. is placed at one remove from the original. Even so, there are logical difficulties in assuming "the earlier version." Instead of one earlier version, at least two earlier versions must be assumed, one composed of the Morgan MS. alone (solely on the basis of its unique variant in l. 17).

CC2 ends at line 14.