¹ Especially in Canterbury Tales by Geoffrey Chaucer (New York, 1928), hereafter cited as “Manly.”—Some statement seems fitting as to the relation of the present study to him and his work. I sent the MS. to him to ascertain if it would be in any way or for any reason impolitic to print. He replied, of course in a kind and liberal spirit, that it would not be. Though in order not to confuse responsibility little or none of the information afforded by him and his co-workers has been embodied here, certain matters or expressions which he criticized have been modified or omitted, and he has saved me from more than one error; certain other opinions have been reinforced with further evidence. My desire has been to confine myself to broader conclusions unlikely to be invalidated by more exhaustive manuscript-detail, and to matters admitting legitimate difference of opinion. In all cases I am grateful to him.

² London and Copenhagen, 1925; cited hereafter as “Brusendorff.”

³ The works cited under these names are respectively Eight Manuscripts of Chaucer's Cant. T. (Ch. Soc., 1913); Evolution of the Cant. T. (Ch. Soc., 1907); Chaucer: a Bibliogr. Manual (New York, 1908). Other works cited hereafter merely under the authors' names are Robinson's edition already mentioned (Boston and New York, 1933), and Sir W. McCormick's The Manuscripts of Chaucer's Cant. T. (Oxford, 1933). For the MSS. the abbreviations of Manly, McCormick, and Robinson are used.

⁴ By the word “editor” (which will recur) I mean someone who took more responsibility for solving problems and getting things into shape than would be expected of an ordinary copyist.

⁵ Of the nine alternatives in MchT 1305–6 none is genuine; Chaucer probably wrote the first six words, hoping vainly to finish the lines later, as Virgil so often did in the Aeneid, and the scribes mended according to their lights, as they so often did. A modern editor I believe should omit all except these six words (as MSS. Hg and Hk, only, do—suggestively as to the history of the text.) After most alterations Chaucer would cancel.

⁶ One which has been doubted (as by Manly, p. 512) is the graunl—haunt couplet about the Friar in the Prolog, 252 a b, in several MSS. In favor of its authenticity I would point out that it is not only in one of the very oldest MSS., the Hengwrt, but that this contains no spurious lines, to say nothing of couplets, whatever; on the contrary it often carefully leaves blanks for lines obviously missing. I do not believe any discoverable MS.-genealogy will prove otherwise likely additions to be spurious; everyone knows there was a vast amount of contamination, both constant and sporadic.

⁷ Lists of cases, not complete and perhaps not without debatable matters, are given by Brusendorff (pp. 116–20); the writer's Harl. MS. 7334, pp. 24–25; and elsewhere. Here I should not include six cases where the word write is used of these orally-delivered tales (K_nT 1201, Mel 2154, MkT 3843, MchT 1739, FklT 1549, SNT 78), which are far too numerous to be “oversights”; Chaucer for the moment thought of himself as the literary man, not as the imaginary reporter, and in all cases but one found the word convenient for a rime. This matter is good to remember when one is tempted to force the realism.

⁸ See, e.g., Skeat's Eight-Text Edition, pp. 2–5; he seems to believe the same. So also Brusendorff (p. 131), who further makes the legitimate, startling, and unacceptable suggestion that Chaucer may not have meant the Sec. Nun's Tale for that lady after all.

⁹ Pp. 82, 127. Mr. Manly so far is undecided (pp. 574–575). These extracts are collected, from the Ch. Soc. prints of MSS., in Skeat's Eight-Text Ed., pp. 9–18. It is noteworthy that they are much commoner in two of the oldest MSS. (El, Hg).

¹⁰ All these and other remarks also are collected by Skeat (pp. 5–9), who thinks some may be due to Chaucer; assuredly not all. These glosses in the Ellesmere MS. are certainly in the same hand as the text, which probably shows that they were copied from its parent. They are commoner by far in El and Hg than in other published MSS.

¹¹ So Manly, Stud. in Philol., xxviii, 616.

¹² Cf. R. K. Root, PMLA, xxviii, 419–420, etc.; Brusendorff, p. 55. Much of Mr. Root's serviceable article, chiefly on Latin works in Italy, may be applied to vernacular works in England.

¹³ See Troilus, v, 1793–98 (also Brusendorff, p. 59); ibid., iii, 726, compared with KnT 2062–64; PardT 584–585 (intended perhaps for Chaucer's scribe as well as the Pardoner's congregation). I have put in the same category the Latin glosses in the earliest MSS. over homonyms and hard words.

¹⁴ Skeat, Evol. C. T., pp. 12, 19, 27; Eight-Text Edit. of C. T. (Ch. Soc., 1909), pp. v, 50; Miss Hammond, pp. 243–244, 250, 262 (she suggests a pirated edition). Professor Carleton Brown, PMLA, xlviii, 1041–42, believes in more than one genuine arrangement and draft; why, unless the earlier was uncorrectable because published? It is a trifle difficult at times to see just what the conception of the history really is.

¹⁵ Miss Hammond, pp. 243–244, and in Mod. Philol., iii, 162 (she seems to overstate in saying that the supposed fact “has long been recognized”); Carleton Brown, PMLA, xxvi, 28–29. The view was well opposed by Koch (p. 420; Anglia Beibl., xxii, 282).

¹⁶ Miss Hammond (p. 260) seems to agree. But there is interest for another reason in observing which the single tales thus selected are (list in McCormick, pp. 535 ff.); one of the commonest is Melibeus. Of course some MSS may be mere débris.

¹⁷ G. H. Putnam, Books and their Makers during the M.A. (New York, 1896), i, 233, 238, 256, 258–259, 267. Conceivably some of the incomplete MSS., without headlinks for example, (when not mere débris) might be derived from such peciae.

¹⁸ I would not say that never can there have been two copies even of a shorter prolog, for whatever reason produced; one must remember the long and short forms of the MkNP link (Mrs. Heseltine, in McCormick, pp. xxix f.), and MS. variations as to extra passages in WBP, MkNP link, and ClT. But all these might have been on extra sheets sometimes overlooked. Such a thing as Mk NP link is specially unlikely to have been published separately.

¹⁹ Mod. Phil., i, 13–18; see Skeat, Oxford Chaucer, vii, 347, lviii.

²⁰ By Dr. Ruth Crosby, Radcliffe College dissertation (unpublished), 1928. Some of the wording in the Tales which seems unfitting on the pilgrimage is, as I have said, due to the fact that Chaucer was thinking of his visible auditors and his readers.

²¹ It is best to keep till later the question whether he left any indication as to the order of the “groups.” Brusendorff (p. 72) once implies some significant arrangement of parts by Chaucer. His point is far from convincing.

²² Opere Volgari, Magheri-Moutier ed. (Florence, 1827–34), xv, 72.

²³ Brusendorff, pp. 72, 130; Manly, Stud. in Philol., xxviii, 616. One or two people have trifled with the possibility that the officious, able, sometimes stupid person, a martinet as to versification, who is responsible for the very early MS. Harl. 7334 was no less a person than Chaucer's friend John Gower, who was all those things, and who did not die till 1408. Chaucer edited by Gower! Stranger things have happened.

²⁴ So far as I know it has been mentioned only in the Chaucer Concordance, p. iv.

²⁵ W. Wattenbach, D. Schriftwesen im M. A. (Leipzig, 1896), especially pp. 554–556, 560–562; Richard de Bury, Philobiblon, chap. 8; F. Kapp, Gesch. d. deutschen Buchhandels (Leipzig, 1886), especially pp. 20–21; Brusendorff, pp. 54–55 (little on vernacular books in England); G. H. Putnam's desultory and uncritical Books and their Makers during the M. A. (New York, 1896) is useful, especially i, 226–274, 302–313. I have not seen most of A. Kirchhoff's work (1853–55). With patience an informing treatise on the subject could be written.

²⁶ Yet in 1898 H. B. Wheatley suggested that the heads of monastic scriptoria would pay Chaucer “something for a new Canterbury Tale,” and competed in publishing the Tales; his ideas are completely misguided (Prices of Bks., p. 2).

²⁷ Kapp, pp. 20–21; Wattenbach, pp. 561–562; Putnam, i, 232, 260, 264, 271, 306 (the reference here seems wrong), 311–313; Ducange, s. vv.; E. G. Duff, A Century of the Engl. Bk. Trade (London, 1905), pp. xi f, xiv, xvi; G. J. Gray, Earlier Camb. Stationers (Oxford, 1904), p. 15; Wheatley, p. 27. There is sometimes mysterious information about prices of books (mostly liturgical), writing materials and copying in the fourteenth and fifteenth centuries in J. E. Thorold Rogers' Hist. of Agric. and Prices in Eng. (Oxford, 1866–1902), vols. i, ii, iii, iv (155, 599 ff., sale at fairs, and prices, sixteenth century); G. G. Coulton, Chaucer and his Eng., pp. 98–100; Wheatley, pp. 51 ff.; Gray, pp. 18–22.

²⁸ The fact that a MS. has the arms of the owner does not necessarily prove that it was not bought, when written to order, or even ready-made, from a bookseller; cf. Putnam, p. 268. I do not discuss how far this is applicable to the CT MSS.; but see, e.g., MSS. Paris and Petworth in McCormick, and p. 121 below. The numerous signs that the CT, as one might expect, appealed to well-to-do and aristocratic people, as shown by Mr. Manly and Miss Rickert, do not of course show that they were not commonly got from regular booksellers. My conclusions are in accord with those of Professor J. W. Thompson in his book in preparation on medieval libraries (my thanks to him).

²⁹ “Prohemye” reprinted by E. Flügel, Neuengl. Lesebuch (Halle, 1895), i, 6–7. The two editions are analyzed by McCormick (pp. 55–76).—Just so Thynne in 1532 had compared various texts of Chaucer, and laments their depravation (Flügel, p. 305).

³⁰ The same in Ne; almost as many in Ha² and a few others.

³¹ I do not wish to overstate this matter. A modern might go through many MSS., perhaps, or through McCormick's book, without noticing these efforts. But once he notices them, he finds a vast amount of evidence that the scribes were constantly trying to deal with the appearances of incompleteness.

³² The texts where these five sets of notations exist are, respectively: En³, Ha³, Hg, Py; Cx², Dd, Nl, Py; Ad¹, Cn, Dd, Ds¹, En¹, En³, Ma; Dd, Ds^l, En¹, Ha⁵, Ma; Dd, Ds.¹ To quite an extent these lists are identical, showing a set policy. These texts are of all dates, Hg and Dd being among the oldest in existence. The fact that a spurious headlink for ShipmT is almost the commonest of all spurious links explains the rarity of its appearance “sine prologo.” None of the above-mentioned notations indicates, as Brusendorff supposed (p. 120), the scribes' belief that Chaucer had finished the poem. It should be added that the scribe of Ha⁵ appears to call attention to the absence of CYP and T though known to exist.

³³ Hk, Ld², Mc; sometimes, but evidently not always, the explanation may be that blanks were left for a rubricator who never arrived.

³⁴ Just so some thirteen MSS. omit the whole of the unfinished couplet, MchT 1305–6 (two others, Hg, Hk, omitting all but the six genuine words). This passage is one of the most suggestive as to the history of the text and the careful comparison and discussion which must have taken place.

³⁵ Bo² and others. I add that Se has something similar at the end of SqT, but wrongly placed.

³⁶ Just so the Northumberland MS., in foisting in the tale of Beryn, actually in a Latin couplet mentions its author, an ecclesiastic of Canterbury (McC., p. 376).

³⁷ Miss Hammond (p. 244) made the explicit assumption that scribes never omitted links which were in their originals—a very large assumption indeed. It vitiates a great part of the argument on the history of the CT produced by this able pioneer.

³⁸ The only other genuine link in stanzas is that between the stanzaic PriT and Thopas; the two other tales in seven-line stanzas are followed by couplet-links. There is not the slightest reason to doubt the genuineness of the Host-Stanza. One might hazard a query as to whether Chaucer once thought of another stanzaic tale as following; later he evidently decided to put all links in couplets.

³⁹ As Manly does fully, Koch and Robinson with brackets. Skeat puts it at the foot of the page, and the Globe editor wholly omits.

⁴⁰ McCormick, p. xix. Mostly or all genuine, but probably cancelled (Skeat); all suspicious (Globe); genuine but cancelled (Koch; Robinson also, but suspecting the final couplet); spurious or genuine, but perhaps cancelled (Manly); genuine, except final couplet, and deliberately omitted by scribes (Brusendorff, p. 89). But cf. Tatlock, MS. Harl. 7334 (Ch. Soc., 1909), pp. 23–24, which says briefly what is said above. It is hard to be sure what idea of the history of the CT is implied by the idea that NPE is genuine but cancelled.

⁴¹ See, e.g., Kittredge, Date of Chaucer's Troilus (Ch. Soc., 1909), p. 32; Tatlock. Devel. and Chronol. of Ch's Works (Ch. Soc., 1907), p. 77. Chaucer's “favorite line” about pity in the gentle heart appears no less than five times. In the two links under discussion there are three pairs of lines with echoing of words and thought (4641–42, 3135–36; 4645, 3131), and there is a general parallel in idea. But so is there in the Prioress' and Second Nun's prologs, and also echoing of words in as many pairs of lines (1668, 54 and 56; 1657, 36; 1664, 50). So is there in the Miller's and Manciple's prologs, and in many other places, a recurrence of situation and idea, and more or less of wording. Mr. Manly notes the parallels between the latter half of MkP and the NPEp, only 1500 lines apart, and infers the cancellation of the latter; he does not note those between its first half (on the Host's wife) and MchP and MchEp (the latter two only 1200 lines apart). Which of these should Chaucer have cancelled? The verbal parallels are less striking than Manly's—also than mine above; but three appearances of a pilgrim's ill-tempered wife are striking enough. Cf. MchP 1223–25 with MkP 3085–90, and 1239 with 3094 ff.; MchP 1222 with Mch-Sq link 2428, 1226–27 with 2432, 1221–22 with 2429 and 2433–34, 1243–44 with 2439–40. Perhaps all this tends to invalidate an argument for cancellation based on likeness in thought or language. Chaucer may have written one of these links long after the other, and without much memory of it, but we cannot affirm that he would ever have changed or even regretted such repetition. I believe the usual omission of NPEp is due to the scribes or “editors,” and has no connection with its resemblance to the earlier link. It is hard to see how it could have any.

⁴² Ry¹ has a CY-Ph link. The MSS. which retain the Host-Stanza refrain still more from spurious links.

⁴³ McCormick, p. xvii. I do not understand why his assistant says that in only 29 “the link introduces the Squire's tale,” though by their own showing it does so in 32 (pp. xvii, 69, 103, 225). No one doubts this link is genuine. Koch and Robinson think it cancelled; Brusendorff has a peculiar and unacceptable theory (p. 72); Manly does not commit himself save by printing it merely in his notes.

⁴⁴ To pass over Brusendorff's fantastic suggestion (p. 72), the only person who has argued at large for the Squire is Dr. C. R. Kase, in the course of an ingenious and painstaking essay in Three Chaucer Studies (New York, 1932), pp. 39–50; but see also Brown in PMLA, xxvi, 27–28 and xlviii, 1054; Skeat, Evol. CT., p. 12. Mr. Kase's arguments are so odd as to be hard to refute briefly, but the following may suffice. The Squire might not have hankered after a Parson's tale, but he was too “courteous, lowly and serviceable,” too well-bred (as shown in his tale and the links before and after it), to volunteer among his elders, his father and all, such high-spirited bluster and such rudeness to a good, learned, and probably elderly priest. Every human being whose soul is sworn by in Chaucer is obviously dead, and there was at hand too fine an assortment of medieval oaths for the Squire to treat his father, riding beside him, so jauntily as to swear by his soul. It is no wonder that, as Mr. Kase says, the dramatic appropriateness of this passage to the Squire “has not received much attention from scholars.” On the other hand, the Shipman, who was in the habit of throwing objectionable persons overboard, would not consider whether or not he had a right to be disrespectful to the Parson; there is rich humor in this pirate heading off a clerical bore by professing fears of heresy; and the breeziness of his “ joly body” and his clinking a merry bell is perfect for him.

Though almost no one has argued for the Sumner, I add that the rollicking speaker here cannot have been that surly fellow; nor would a man who had a suppressed desire to show off his parrot Latin have boasted that there was but little Latin in his maw. (Here I retract a note on p. 218 of Devel. and Chronol.)

In support of his view that the Squire's Tale was written to follow the Man of Law's, yet that it and the Franklin's were meant as a unit, Mr. Kase argues against Kittredge's view, accepted by almost everyone, that group F was written as a part of the “marriage group.” Here again much that he says is odd and irrelevant. After pausing to wonder at his statement that the Franklin's Tale, especially the delightful little verse-essay near the beginning (ll. 761–786) on the Conduct of Life, are not concerned with marriage, I merely point out that there is in this tale “textual linking ” and “reference back to the D and E Groups,” of which it suffices to compare FklT 745 ff. with ClT 351 ff., 751–2 with MchT 1379, 768–70 with WBT 1038–40, 774 with ClT passim (esp. 1044–50), 804–806 with Mch T 1259–60, 818 with WBP 592, 1364 with MchT 2239 ff. (serving as a retort), and with 2280–85 (as illustration). I do not affirm that the one passage was always meant to recall the other; but this suffices to show that the debate on marriage was held in mind in FklT as much as one need expect; Chaucer was writing tales, not polemics. In Mr. Ease's idea that Chaucer's first plan was a Man of Law-Squire sequence, and his final plan (as held by all scholars) the Merchant-Squire, there is nothing a priori unlikely; the question is merely one of evidence, of which Mr. Kase shows none except the commonest reading in l. 1179 and the usual position of SqT, which are better interpreted otherwise.

After this note and article were written, Professor F. Tupper's “Bearings of the Shipm. Prol.,” appeared in JEGP, xxxiii, 352–72. It is pleasing that we almost exactly agree throughout.

⁴⁵ Miss Hammond seems to have been the first to remark on this, in Mod. Phil. iii, 163–164.

⁴⁶ So Skeat continued to believe, though he changed his mind as to other things about it for reasons hard to understand (MLR, v, 430–434). It is not strictly impossible, perhaps, that Chaucer left the spot blank or illegible, that a very early scribe or “editor” (anathema sit) wrote “ Sumnour” on the original MS., and that others used their own judgment. But there is no evidence for such a notion, and my supposition seems far likelier. This is one of the passages that make one believe in lively discussion among early “editors.”

⁴⁷ His omission to indicate the intended unity of B¹–B², as he may have done somehow with other groups, might be due to realizing that the beginning of ShT had to be rewritten. I cannot for a moment believe Brusendorff's idea (pp. 118–119) that the Shipman is meant to be quoting an imaginary female speaker; it requires as much modern punctuation as Skeat's interpretation of MchT 1684–87.

⁴⁸ I repress conjecture here as to the origin and history of this confusion.

⁴⁹ Caxton's two editions, and El and Hg (see pp. 128, 129 below). Caxton for his second, and presumably the El-scribe at times, used a fresh exemplar. The El-scribe probably used more care and intelligence than any other, and the only links he omits are the two which most strongly indicate incompleteness. An intelligent scribe is the riskiest of guides when what we are seeking is the original Chaucer.

⁵⁰ Lists in Miss Hammond, pp. 156–157; Brusendorff, pp. 69–70; Manly, pp. 82–86; the Ch. Soc. Six Text edition; and above all, presumably exhaustively, McCormick, pp. xxv–xxviii, and (with the text) in his account of all the MSS., pp. 1 ff.

⁵¹ Except a half-dozen found respectively only in La and Tc¹ (Mrs. Heseltine, in McCormick, pp. xxv–xxviii).

⁵² Noted by McCormick in his account of each MS.

⁵³ There is backing for this opinion in the Paris MS. It bears both a private coat-of-arms and the name of the “scriptor,” as very few do; it also contains no spurious links and omits most of several tales with a frank expression of distaste. So the least commercial-looking of the MSS. is the least Jesuitical. The mere spurious lines in which it abounds would not be recognized as such, and may have been inherited from its original.

⁵⁴ Furnivall, Temporary Preface (Ch. Soc., 1868), pp. 9–44, esp. 42–43; Skeat, Oxf. Chaucer, iii, 376–379.

⁵⁵ Manly, pp. 77–78; and Stud. in Philol., xxviii, 617; Brusendorff, pp. 125–126; Koch, Pard. T. (Ch. Soc., 1902), p. xx; Robinson, p. 1005.

⁵⁶ Consider the surprising freedom from inconsistencies in the Divina Commedia, and its time-scheme, discoverable only on minute observation. Chaucer was not the perfectionist Dante was; but his love of reality appears in the unobtrusive accuracy shown by Mr. W. C. Curry in many passages, and shown in the Troilus and the Franklin's Tale. On that in Troilus see Root, PMLA, xxxix, 50.

⁵⁷ I venture to draw a parallel with what I have said (Kittredge Anniversary Papers, p. 345) of the FklT astrology; a certain date is pointed to, and on that date the astrological conditions prove to be extraordinarily potent for Chaucer's purpose; how can we believe this accidental?

⁵⁸ Temp. Pref., p. 29. See the present writer in Harl. MS. 7334, pp. 21, 26 (notes). See even Dr. Kase, op. cit., pp. 54, 58.

⁵⁹ The only exception (merely apparent) is in Ln, where the two tales which it is used with are far from each other, and a later hand splits it to introduce each. Even where the pilgrims mentioned are changed, the link remains a unit. Its last part (for the Squire) never has a separate heading (see even Skeat's way of printing); it has no “quod oure hoste,” but continues the preceding speech. These facts may be verified in McCormick's MSS. of CT, and are stated by his assistant Mrs. Heseltine, p. xxii. I am not discussing the complex history of the E-F links, but so far as I know there is no theory even of revision, or anything, to contradict the view that E-F was finally (and so far as known always) intended as a unit.

⁶⁰ To put H and I together will help explain a still worse absurdity, in ParsP itself; “ten of the clokke” (1. 5) was certainly written by Chaucer, though a child could see the time is late afternoon, and several MSS. correct the reading. When he decided to join Manc and Pars, he probably changed form four to ten (as was easy) and postponed changing the rest (not easy). How otherwise can one explain the flagrant impossibility in Pars P2–7?

⁶¹ Manly, p. 655, and Stud. in Philol. (Royster Memorial Studies), xxviii, 613–617; MLN, xliv, 493–496.—Dr. J. A. Work uses some of the arguments used below, but in general he follows these two scholars (JEGP, xxxi, 62–65).

⁵² With horses walking two or three miles an hour there was plenty of distance also for E-F and G since Ospringe. I do not understand why Mr. Manly sees in MancP 4–13 or later any implication of beginning the day's tales; all that seems implied is a momentary lull (l. 5). The situation is a stock one, natural to recur in a series of fragments written as the C.T. were, and especially natural in headlinks—remarks on a fresh start of tales, “time we began,” someone silent (Prol 828 ff., MLP 16 ff., ClP 1 ff., MancP 5 ff.).

⁶³ Cant. T., p. 656; Stud. in Philol., xxviii, 616.

⁶⁴ I could illustrate this from real life in the twentieth century. We might compare Chaucer's own characterization, with conscious humor, of the long and matter-of-fact though highly interesting Melibeus (ll. 2127, 2154) as a “litel thyng in prose,” “this murie tale.” Professor Robinson (p. 16) is with me as to the Pars. T.

⁶⁵ For the passages which supply all this evidence, see Furnivall, Temp. Pref., pp. 42–43; Miss Hammond, pp. 160 ff.

⁶⁶ In D 847 the party has not reached Sittingbourne (sixteen miles from Canterbury); in G 556 they are at Boughton (only six miles).

⁶⁷ P. 505.1 add that he detects the absence of a fixed plan in the geographical contradiction in the MSS. between two links here (pp. 77–78), but (as we saw) not so in another between H and I, and bases a revolutionary theory on it.

⁶⁸ I still believe the evidence distinctly favors the intention of a three-days pilgrimage, with nights at Dartford and Ospringe conceived as coming between A and B¹ and between D and E; see PMLA, xxi, 478–485. E-F must follow D, but not necessarily on the same day. It is likely enough that had the work been completed some of these arrangements would have been changed; we are merely discussing what is indicated by the present evidence, which seems clear if not decisive.

⁶⁹ In MLP 14 the hour is ten in the morning, and the implication is that there have been no tales for some time; the opinion has been expressed above that B¹ should be immediately followed by B².

⁷⁰ Except that in H-I, which cannot be cured by rearrangement.

⁷¹ I therefore take issue with Brusendorff (p. 126), and Manly, pp. 77–78, and Stud. in Philol., xxviii, 617. The debate really comes to the dilemma—which is more likely, a discoverable and fairly thorough plan with rare irrationalities, or no discoverable thorough plan at all? I vote for the former.

⁷² One would like to know, how many of them? See even Professor Carleton Brown, PMLA, xlviii (1933), 1058–9.

⁷³ In five texts (McCormick, pp. 327, 335, 387, 405, 425), mostly very incomplete, and all hopelessly disordered, the undoubted unit B² is broken up, and a part comes earlier than D; the whole of it never. One is almost surprised at this, considering the extraordinary variety of orders in which the work became shaken up. No one could conceivably find MS. precedent for the order B²D here.

⁷⁴ A B¹ D E-F C B² G H-I. The most considerable difficulty here (to ignore the affinity of B¹ for B²) is the mention of Sittingbourne before Rochester. Another of the oldest MSS., Harleian 7334, makes the worse error of putting also Boughton-under-Blee before Rochester.

⁷⁵ Cf. Brusendorff, p. 84; E. Markert, Chaucers Canterbury-Pilger u. ihre Tracht (Würzburg, 1911), p. 4; and Athenaeum (1911), ii, 210–211.

⁷⁶ See the terminal note on the Hengwrt MS. (pp. 133, 134 below).

⁷⁷ Besides this, Sq headlink is adapted to the Franklin and Fkl headlink to the Merchant, distortions correct in El. These differences in content with closely similar text in these two MSS. probably by the same hand and mostly copied from the same original are extremely illuminating as to the sort of thing which happened shortly after Chaucer's death. The suggestion (which has actually been made) that the lack of CYP and T in Hg proves they were a late addition by Chaucer shows an idea of the history of the CT hard to understand, and certainly contradicting what is said of it here. They are lacking also in He, Ra³, Sl², Tc². The sheets containing them may have been temporarily lost, lent, or mislaid from an unbound original, and later returned to it, or were taken by the El-scribe from another original. They were not lost from their right place in Hg during binding or otherwise, but were never there. It is true that it would be hard to prove them never in Hg in a wrong place; it is not quite impossible also that CYP and T were intentionally omitted, as most of them were (with an amusing comment) in the Paris MS.; conceivably there might have been some resentment at the reflections on alchemists or on canons. There may be several ways of explaining their absence from Hg; but far the most likely is absence from the original.

⁷⁸ El used to be called an“edited MS.,” showing such signs of meddling as the“modern instances” in MkT at the end, and the omission of ML and NP endlinks. I cannot for a moment agree with Brusendorff's peculiar view (p. 78) that Chaucer meant the “modern instances” to come last in MkT. Since these peculiarities are also in Hg, they are no argument for or against my position as to the priority of Hg. But all these variations among the oldest MSS. do show how the earlier scribes tried and tried again to get things into a satisfying shape. “Edited MS.” is unmeaning; most of the early MSS. are “edited,” El merely more intelligently. The later MSS., made in wholesale routine, show less use of the wits, often were misled by spurious links, and are more careless as to arrangement; at times there may have been misbinding, and since the original confusion may have become well known, scribes perhaps gave up the problem of order as a bad job.

⁷⁹ WBProl though beginning abruptly carries on the narrative of the pilgrimage, and with a label attributing it to the Wife is satisfactorily connective.

⁸⁰ These number about 24 out of the 57. In the terminal note (pp. 137, 138) I show that Hg also probably had reasons, intelligible though less good, for his arrangement.

⁸¹ For the facts see McCormick and (more conveniently) Manly, pp. 79–81. The nearest other cases are F C, and B²G, in only a fourth or fifth of the MSS.

⁸² By the late and deeply-regretted Samuel Moore, PMLA, xxx, 116–123. He assumes of course that some group-arrangements (which ones?) in the MSS. are authoritative; and speaks of a general agreement (by whom?) that the worst defect of the Chaucer Society arrangement is the position of C. His assumption granted, his argument is sound enough.

¹ Chaucer Soc., Autotype Spec. of Chief Chauc. MSS., facsim. and Forewords; Nat. Libr. of Wales, Charter of Incorp. and Rep. on the Progress of the Libr. (Oswestry, 1909), pp. 50 ff.; respectively.

² Except in those (important in this argument, and with none of the text missing) comprising folios 82–87 (only 6 folios), 152–160 (9 folios, the first separate, with no catchwords, discussed in note 6 below), 225–234 (10 folios, normally arranged, none separate). There are one or two other cases of peculiarities among the 30 quires. F. 203 recto, within a quire, has catchwords, evidently inadvertent. Ff. 161–176 (containing Fkl, SN, Cl) form a quire of 16 ff. These cases are not significant, as the others are.

³ Facsimiles in the National Library's publication just mentioned, of A, ll. 4323–92; and in the Chaucer Soc. Autotype Spec., of B, 1191–1226. These may be convincingly compared with that of El published by the Manchester Univ. Press in 1911. Koch is not aware of the identity of hand.

⁴ Skeat, Evol. of CT, p. 34; Koch, Eight MSS., p. 68 (cf. 128, 132, 142), and Pari. Prol, and T. (Ch. Soc., 1902), pp. xxxvi f.; Spec. of … Unpr. MSS. of CT. (Ch. Soc., 1897) iv, xlvi f.; Athenaeum, 1872, ii, 208 (ten Brink quoted).

⁵ Though the folios have been trimmed. The fact that folios 225–9 (which begin the 10-f. quire) have just under the text the numbering i, ij, iij, iiij, v, with no preceding letter, proves that there were never any other signatures beneath, because these would have made this needless. Further, those familiar with this MS. will probably grant that vestiges of signatures would be likely to remain on some folios.

⁶ It is less confusing to exclude here the very peculiar case between MchT and FklT, rather similar. A quire ends 58 lines before the end of MchT (at E 2360), with the proper catchwords. Then comes a single separate folio beginning with these words and ending the tale (followed by a rubric) at the middle of the verso, the rest of which was left blank, and has no catchwords. Then comes a complete 8-folio quire, the first 1$frac12 pp. of which are as follows. At the top of p. 1 is

“¶Here folwen the Wordes of the Worthy Hoost to the ffrankeleyn,” followed by a $frac14 p. blank, then by the genuine MchSq link (Ey goddes mercy—I wol seye as I kan, E 2419-F 4), which fills the rest of the page without any blank. At the top of p. 2 comes the rest of the link (F 5–8), followed by a $frac12 p. blank containing only

“¶Explicit”and

“¶Here bigynneth the ffrankeleyns tale.” Below the middle of this p. 2 there begins the Fkl “proem” (Thise olde gentil Britons), which with the tale continues normally throughout the quire. There is nothing abnormal except the extra folio, the blanks, and the fact that F1 instead of “Squier” reads “ Sire ffrankeleyn.” Several things are certain. The scribe used the separate folio because he wished to end MchT without committing himself as to what should follow. Why was he so cautious? Since MchSq link was obviously composed to follow MchT, the answer is that it was not then at hand. Why did he at first leave one and a half pages and more blank before Fkl proem? He hoped to find a link. At some time after he began Fkl proem and tale he found the MchSq link. If it had the Fkl in F 1, he inserted it here inevitably. If (as probably) it had the correct reading “Squier,” why did he not insert it before SqT, already copied and with a blank page before it with ample space for MchSq link? Because the wording made it much more necessary that this link should follow MchT than precede SqT, it was too late to change the position of SqT, and it was easy to change “Squier” to “Sire ffrankeleyn” (“Sire” obviously suggested by “Squier”). The absence of catchwords at the end of the separate folio proves that Hg did not know what words would begin the next page; he may or may not have known that MchSq link existed, and have intended FklT to come next. The presumption is, it is true, that this case was originally like the other four where catchwords are wanting, and that he omitted them and left the blank on the mere chance that a link would turn up; but the exposition is clearer through treating this case separately. The whole matter is a good specimen of the early laxity as to safeguarding and altering Chaucer's MS. or the first copies.

⁷ It is notable that in every one of these 4 cases it makes no difference whether or not Chaucer's text indicates what is meant to follow; it gives no indication after Ck or Sumn, but does so after Mel and (as I believe, and in almost all MSS.) after Mane. The one-fourth page blank between Ck and WB is far from unparalleled; not only did our man in writing El leave one of his few blanks here; within quires in Hg he left a page blank between MLT and SqT, a half and a quarter and a half page blank between MchT and FklT (before and after the MchSq link, made over to Fkl, the very peculiar case described above), a half-page blank between FklT and SecNP, a quarter-page blank between PardT and ShipmT. All except part of those in the peculiar case are at the bottom of the page. These blanks are not, as might be fancied, due to a preference for starting a new poem at the top of a page. Hg had none; in the first half or so of the work (from Prol to MancT) he almost always begins a new one on the same page as the last (always, except in a very few cases where by chance the last fills the page). But in the later part of the MS., where are the blank places under discussion, there are about as many cases of beginning on a new page as on the old, and half of these cannot possibly be deemed accidental. The clear reason is that in this part of the MS. the scribe was very doubtful about the completeness and the sequence, and wished to leave space (even thinking of inserting an extra folio if necessary) for rearranging or the supplying of a link; which usually proved to be not forthcoming, either because Chaucer had never written it (as after Fkl, Pard), or because Hg did not find or use it (as after ML, SN, Cl). That this was the main reason for leaving the blanks is almost proved by the peculiar case just mentioned. The blanks of a page or more (21 pp. after SumnT, 1 p. after MLT) betray even more clearly the hope that some connective would turn up. Hg left so long a blank after SumnT because it was so near the end of a quire that he was willing to leave the rest blank; a shorter after MLT because that ends close to the beginning of a quire; at the only other cases where a tale or link ends without a following connective, —NPT, MancT, MelT—the last two end exactly with the quire (allowing the insertion of following folios later if needed), and NPT ends before the middle of the quire, to leave the rest of which blank would have been too much; but the chief reason for no blanks is that early in the MS. the scribe was not so uncertain and suspicious as he was later.

⁸ These cases, where Chaucer's text does not show what tale is to follow, but where Hg continues in the same quire, are NPT-MancP, with no intervening blank space; MLT-SqT, with a page of blank; FklT-SNP, with a half-page of blank; PardT-ShipmT, with a quarter-page of blank; and two cases where (since Hg lacks CYP and ClMch link) Hg found equal uncertainty, SNT-ClP, and ClT with Host's stanza-PhysT, each with an insignificant blank. While in some of these cases leaving some blank space for links possibly to be found, Hg did not trouble here to leave open the question of arrangement.

⁹ None are erased. The very few other irregularities of procedure in Hg mentioned in n. 2 are not nearly enough to justify doubt that the extra sheet, crowded writing and absence of catchwords have the significance I give them.

¹⁰ The / which in other lines marks the cæsural pause was also erased, so the original word was little if at all longer. There seems no possibility of distinguishing anything more of the earlier reading.

¹¹ Except that 4 late MSS. (Ch, Gl, Ra³, Tc¹) changed to yeman, Marchaunt, or Frankelen, trying to adapt to wrongly preceding tales.

¹² Incidentally I add,—even WBP and MkNP link suggest the same original, the passages there in El but lacking in Hg having been perhaps on loose sheets disregarded by Hg. See p. 112 above. I accept with alacrity Manly's theory as explaining things sometimes.

¹³ In most cases marring the verse and style—Marchant in SqFkl link, 675, 696, 699, ffrankeleyn in MchSq link (“F,” 1). The only case where he might conceivably have done this and did not is in the MelMk link; here it would have been a waste of trouble since this is preceded by several blank pages, which he hoped might be filled later.