¹ The only tale orders not covered by this list are haphazard ones (Ch, Hk, etc.), obvious distortions of known orders (Ps, To, Nl, etc.), and the order of √Ad³ which will be treated en passant. For the order in any extant MS of CT see John M. Manly and Edith Rickert, The Text of the “Canterbury Tales” (Chicago, 1940), vol. i; the chapter “The Order of the Tales” is in vol. II, 475–494; helpful charts are printed immediately after that chapter.

² On the “constant groups” a, c, d, and 6, see Manly and Rickert, ii, 51–70, and 480–486, or pp. 395–401 of the writer's “Manly's Conception of the Early History of the Canterbury Tales”, PMLA, LXI (1946), 379–415. The chief point in connection with the order of the tales is that the textually related MSS which constitute a “constant group” have the same order (accidental deviations disregarded) and undoubtedly inherited it from their common ancestor.

³ Manly and Rickert, i,270–275; ii, 477–479.

⁴ “A Chapter of the Manuscript History of the Canterbury Tales; The Ancestor of Group d; the Origin of its Texts, Tale order, and Spurious Links”, PMLA, LXIII (1948), 456–484.

⁵ “The Canterbury Tales in 1400”, PMLA, L (1935), 100–139.

⁶ His few remarks on the formation of the √c order (i, 95–96, and ii, 42; ii, 482 is merely descriptive) are not very helpful. The statement that the √c editor “moved” the SqT to follow the ML Endlink (ii, 42) would seem to reflect a belief in some pre-√c arrangement (other than Hg's, where SqT follows MLT), but no such possibility seems to be in Manly's mind in I, 95; yet cf. ii, 482, where Block G is said to be “brought forward.” The further statement that, as far as internal evidence goes, G's position before C is as good as the one generally adopted (ii, 96) is hardly correct, for C is always before B², thus before the reference to Rochester.

⁷ For the evidence see pp. 467–469 of the writer's article on √d (ref. above, n. 4).

⁸ The order of El is shared by group a, Gg, En³, and others; the √d order was adopted by the editor of √b

⁹ Manly and Rickert, i, 220,223.

¹⁰ Ibid., i, 95–96, 220–222, 567 (plate); iv, 499.

¹¹ Ibid., i, 270; II, 477–479.

¹² They are absent only in disarranged or incomplete MSS; see the charts in Manly and Rickert, n, after p. 494.

¹³ Taking the pieces in their √c order, we find b and c closely related not in Gen. Pro. but in KnT, not in MiT but in RvT, not in CkT but in MLT and SqT, not in D nor C1T but in MeT, not in FkT nor G but in C, not in the first four tales of B² but in MkT, not in NPT but in H-I.

¹⁴ The same could of course be said of all the other MSS which, on and off, appear in the “Large Group” higher than √c.

¹⁵ For instance the tale orders of Bw, Ry¹, Mc, Ra ³, whether or not through √b, derive from √d; En³ owe their order to an a MS.; Bo² and Ps have distortions of the a-El order; Bo¹ combines d and a-El.

¹⁶ Manly had ample ground for postulating the existence of such a MS for almost all but the very short pieces. Some of the significant variants are listed in the discussions in vol. n; as a rule there are many more in the corpus of variants.

¹⁷ We cannot expect such continuity to be very striking, for, by the time most of the extant MSS were written, many shifts of affiliation had occurred, especially at the end of pieces, where scribes would often stop their day's work; but in comparison with the conditions in E-F e.g., where no resemblance at all can be found between the Stammbatime of any two tales, such continuity as one observes through C or D does seem significant.

¹⁸ Nothing more natural than that a copy consisting of portions (perhaps neat quires of regular size) bearing figures, catchwords, or the like, and intended for use as original, should be prepared on Chaucer's directions for any series of tales which he considered finished. But no such copy could be made of E-F, e.g., as long as the SqT remained unfinished, nor of Block G if more work was to be done on CYPT. This inference as to the condition of the Originals at Chaucer's death can of course be drawn from the surviving MSS only on the assumption that most of those MSS derive their texts of most CT pieces from copies made from the originals after Chaucer's death; on the opposite view and on Manly's grounds for holding it, see pp. 384–390 of the writer's article on Manly and the history of CT (ref. above, n. 2); for a discussion of some facts bearing on these alternatives, see “On the Significance of Hengwrt's Change of Ink in the Merchant's Tale”, MLN, LXIII (1948), 325–330.

¹⁹ My main reservation is about the ML Endlink, where the variant Squire in B 1179 may be due to misreading and be older than any attempt to arrange the tales; see below, n.31.

²⁰ Smooth transmission of the three links in their correct forms from the three Originals to Ha⁴ would of course be a sufficient explanation of the correctness of the MS as to sequences and links throughout E-F. But at least the Me-Sq Link came to Ha⁴ via Hg's adapted form of Me-Fk Link (Manly and Rickert, ii, 284–285, 286; vi, 507–508). It was restored to its proper functions hardly through a lucky guess (if so, why Sir Squyer com neer rather than Doctor or Shipman?) but in response to correct information. Whether Ha⁴'s Sq-Fk Link was also derived from Hg we shall never know, for Ha⁴ has lost the quire that covered F 617–1223 (and undoubtedly contained the link).

²¹ For the evidence of contamination in GP, KnT and MeT see Manly and Rickert, ii, 91,103–104,280; of contamination or genetic relation in MiP, ii, 138; of genetic relation in SNT, ii, 433. For KnT and MiP, the man who borrowed readings from El or √E1 would be the scribe of the ancestor of Ha⁴ and Ii; for GP and MeT there is no indication that it was not Ha⁴ himself.

²² A good case could be made for any of the following alternatives : 1) The only authentic reading is Manciple; the word first written and then erased by the Hg scribe was intended to tie PsT to a piece (FkT? PdT?) of which the text, unlike that of the McT, had been obtained before PsP was transcribed. 2) The erased reading is the only authentic one. It referred either to a pilgrim for whom no tale was ever written or to one whose tale was later placed elsewhere by Chaucer. That authentic reading was erased by the Hg scribe and replaced with Manciple when, having obtained the McT, he found that it was the only piece which he could conveniently put immediately before PsPT, at least in that MS. The relative adequacy of the sequence (allusions to time disregarded) plus the dependence of later editors on the tale order of Hg would account for the total disappearance of the authentic reading. 3) The erased word and Manciple are both editorial, the first corresponding to the plan of the Hg scribe before he had obtained McPT, the latter to the plan which finally he carried out. On this hypothesis the reference in the Original might have been to any pilgrim whose tale, written or merely planned, Chaucer could at any time have thought of placing just before PsT; or the Original might have had a blank awaiting Chaucer's decision; cf. E 1305–1306. The absence of comment on the story just told would fit well with this third possibility. There is another alternative, that the erased word reflected no intention of Chaucer or any editor, but was due to a moment of inattention on the part of the Hg scribe, who, on that hypothesis, would not have noticed his mistake at once, at any rate not before line 2 was written, for the/and the h of fro the have been retouched, undoubtedly because damaged by the eraser.—I wish to thank Sir William LI. Davies, Librarian of the National Library of Wales, for his kindness in answering my questions on these and several other points. On the ink of Manciple in H 1 Sir William writes : “ ‘Manciple’ and the tops of the ‘f ‘ and ‘h’ are in the same ink; it appears slightly lighter than the rest of the page. But it was applied lighter on the pen over the erasure probably to avoid running; the outlines of the points show distinctly. It is definitely not the yellow ink” (i.e. not the ink of the opposite and preceding pages).

²³ I leave out of this discussion the references to the time of day, less striking than the place names and very confusing; it seems safe to assume that no early editor paid much attention to them.

²⁴ The PhT and ShT start respectively on the verso of the 7th and the recto of the 4th of the 8 folios making up quires 24 and 26.

²⁵ When he obtained the Me-Sq and Sq-Fk Links he left unchanged his Sq-Me-Fk-SN-Cl order.

²⁶ If, when he copied the Mel-Mk Link, the Hg scribe (as seems highly probable) had neither Block D nor CYPT, the question of the order of the towns could not at the time have presented itself to him; the only awkward point likely to strike him would be the disproportion between the tales before and the tales after the mention of the half-way town of Rochester. The indications that Block D was copied later than the bulk of Hg (the portions in black ink) are in my opinion very clear; Manly seems overcautious on this point (i, 273–274).

²⁷ Not even any such thing as an accidental B²-H arrangement of the copies from which Hg made his transcriptions, thus nothing that could have led to the appearance of the same sequence in other MSS.

²⁸ Even supposing that the √c editor, while working out his own arrangement, paid absolutely no attention to the allusions in the text (yet placed A and I correctly), and further, that he considered both B¹-SqT and D-Cl-Me as blocks (which would reduce to seven the number of loose units to be placed between the terminals A and I)—even so, the chance of his placing B² immediately before H would be only one in seven.

²⁹ For the details see the charts in Manly and Rickert, II, after p. 494.

³⁰ Manly describes it as type a disarranged (i, 42). Much of the disarrangement is obviously due to delay in obtaining copies of CkT and CYPT. Yet why was SNT placed immediately before CITî This suggests acquaintance with the Hg order, as does of course the B²-H sequence. It may be worth recalling that Ad³ derives its Sq-Fk Link from Hg's adapted form, and in SqT shows contamination from Hg.

³¹ This alternative is favored first by the fact that Shipman, Summoner, and Squire all begin with s, which makes misreading a plausible explanation; secondly by the blank which the Hg scribe left between MLT and SqT, as if expecting a link of just the length of the end-link. The endlink, to all appearance, barely escaped destruction; at any rate there is no indication that more than one copy was made directly from the Original. This may be due to Chaucer's transference of the Mel from the Man-of-Law to himself. If written on the same page as the end of Mel, the endlink, at the time of that transference, would probably be crossed off, and Chaucer, even if definitely intending to make use of it, might see no reason why it should at once be recopied. The scribe who copied the endlink from the Original would, on this hypothesis, have put it after his transcription of the legend of Constance on the suggestion of an old explicit to Mel as tale of the Man-of-Law or of a title to the end-link.

³² On the Fk-SN sequence see pp. 474, 480, of my article on √d- As to C-B² there is, with regard to the main lines of derivation from the Originals, enough continuity through Block C and the first three tales of Block B² to suggest rather definitely that the sequence was favored by the arrangement of Chaucer's MSS, if not at his death at least by the time trans-criptions were made from them. Yet the internal structure of the “Large Group” is quite different in PdPT and ShT, as though the √Large Group copies of these two tales had not been parts of one set. In that case the sequence C-B² in √c, since it cannot have been suggested by anything in the text, would very probably be due to some outside influence.

³³ Hg had at first put it immediately after Block A; on his insertion of Block D see above n. 26.

³⁴ On Ht see Manly and Rickert, 1,252–253; II, 271–272,279,285,298,307,386,462; the derivation in most of these cases is more probably from Hg than from √Hg; it is certainly so with the E-F links. The Sq-Me Link of Py, Ra³, and others, and the Sq-Fk Link of Ad³, Ps, etc., show derivation from the adapted form initiated in Hg (Manly and Rickert, II, 298–299).

³⁵ See above, n. 7.

³⁶ Manly and Rickert, II, 284–285; vi, 507–508; note the variants in F 1,2,6, and 7.

³⁷ On the order of the tales that constitute Blocks A, C, D, B², see above, pp. 1127.

³⁸ Throughout Blocks B² and H and Block I down to 680, Ha⁴ and the much later MS Ld¹ (1430–1450) are very close together, probably copied directly from the same pages. It is hard to imagine that those pages were not part of one MS., which then may very well have had the Hg order B²-H-I, and may have covered much more of CT than just these three blocks. I am inclined to think that Ha⁴ copied several other pieces from it. A comparison of the extensive editing peculiar to Ha⁴ and Ld¹ in Blocks B², H, and I to 680 with the editing common to Ha⁴ and its temporary associates in other parts of CT might support that possibility and would be a good, well-defined subject for a student. On editing possibly by the scribe of Ha⁴, see PMLA, LXI, 400, n. 117.

³⁹ Manly and Rickert, i, 220, 223–224.

⁴⁰ See above, n. 20.

⁴¹ Manly and Rickert, II, 481.

⁴² In that alternative, √c's reversion to the B'-Sq sequence would be due either to Hg or to the fact that the √c editor did not have the genuine link to introduce the SqT but had the ML Endlink, possibly with the variant “Squire” in B 1179.

⁴³ See above, n. 38.

⁴⁴ Manly and Rickert, II, 475.

⁴⁵ Whether or not this is the case, the El arrangement may have been initiated on El or on an earlier MS. The absence of signs of hesitation in El does not definitely suggest that there was an antecedent; the Hg scribe would have no hesitation; another might make a table before starting.

⁴⁶ I make the suggestion with utmost caution, for I am very far from sharing Manly's conviction that the H-I sequence cannot have been intended by Chaucer; see above, n. 22.

⁴⁷ For instance, it is possible that Chaucer's MS. of B¹ was found, or was reported to have been found, next to his MS. of Block A, and likewise his MS. of ShT next to PdT; there may have been a feeling that the Man-of-Law, an important person, was the one to whom the Host would turn immediately after Block A; notice may have been taken of the references to Bobbe-up-and-doun and the Blean Forest (H 2–3) which of course require that the MeT be placed very near the end.

⁴⁸ The ML Endlink, had the man responsible for the a-El order obtained it in what may quite possibly have been its authentic and only pre-1400 form, viz. as ML-Sh Link, would have given him sufficient ground for breaking the C-B² sequence. But I incline to think that he never had it in any form. The explanation which Tatlock offers for its absence in El— that both endlinks, the Man-of-Law's and the Nun's Priest's, were rejected by the editor as loose ends drawing attention to the fragmentary state of the work {op. cit., pp. 112, 114, 118–119)—does no longer seem necessary, as Manly's classifications for MLT and NPT so definitely confine each of the endlinks to one genetic family, in neither case that of El.

⁴⁹ Op. cit., p. 129.1 am not sure that the idea is not already reflected in the Hg order. Except for SNT, whose insertion before C1T was an afterthought, Hg has, ahead of C and B², no headless tale which he does not seem to have expected to link to the preceding one, witness the spaces left before SqT and MeT and the precautions at the start of FkT (Manly and Rickert, i, 271–272).

⁵⁰ See above, n. 48.

⁵¹ On PsP see n. 22.

⁵² In his second edition, Caxton, having obtained the Sq-Fk Link, put the FkT between SqT and Block D; Wynkyn de Worde moved Caxton's Me-Sq-Fk series to their proper place immediately after C1T and thus, in a round-about way, came to duplicate the order of Ha⁴.