It will be seen that the remotest archetype here deduced does not differ, in the Tale-order A-F, from the two other main MStypes represented by the Ellesmere and by Harley 7334. This seems proven from the numberings in Corpus and Mm and from the sequence appearing when the misplaced Squire of this archetype and of the Harley 7333 archetype is returned to his proper place. Certainly, until evidence is forthcoming for another theory, we may argue that it was an order like that of Harley 7334, but without the E and F links there seen, which lay behind the Corpus-7333 group, and that it was the misreading of the deleted word Sompnour, still seen in Harley 7334's Man of Law endlink, which set on foot all the displacements characteristic of the Corpus-7333 line of descent. We arrive therefore at the conclusion that the archetype of the Corpus-7333 group represents the most primitive known form of the Canterbury Tales, a form which went into circulation before Chaucer had written the E and F links, and which received those additions during the early stages of copying. This conclusion also brings with it several corollaries. First, if the E and F links were indeed later written, the fusion of these four Tales into a whole was, as ten Brink affirmed, made by Chaucer some time after the completion of the Tales themselves, or of most of them. That it was later than the writing of the Clerk's envoy may be argued from the fact that in many Corpus-Mm manuscripts the Clerk's envoy, when existing, does not end ready for the Merchant; and when it does so end in other groups, I would suggest that its rearrangement had been made to meet the (therefore later written) Merchant headlink. It follows that the conditions of the MSS can thus, in minor points at least, give us some light on the relative dates of the Links, perhaps of the Tales, and on the workings of Chaucer's mind. For is it not a plausible inference that in these later E and F links Chaucer was endeavoring to make more appropriate the telling of two of the Tales (Merchant and Franklin) by their narrators? The vividly autobiographical tone of these two introductions is of the same character as are the Wife of Bath's prologue and the Pardoner's exhortation, pieces of the framework which every student would assign to Chaucer's ripest maturity; and yet the Tales which they prelude are not in themselves so appropriate to their narrators as are those, e. g., of Clerk and Squire. This leads readily to the assertion that there are in the Canterbury Tales three sorts of Tales; those which Chaucer had previously written, assigned to a pilgrim whom he created later, after the idea of the pilgrimage had occurred to him; those written with the pilgrim in mind; and those written after the poem was in progress and forced upon a pilgrim. Cp. ten Brink, Hist. Eng. Lit. II: 149. The evidence which we can collect for this or any chron ological division comes partly from our personal judgment as to plot and treatment in a Tale, partly from Chaucer's own allusions, lastly, from the manuscripts. The first of these sources has hitherto received the most attention. Ten Brink did not deal minutely with the Canterbury Tales; but Skeat and Koch base their remarks sometimes upon their personal judgment regarding Tale-structure, sometimes upon a parallelization of the known facts of Chaucer's life which has led them to extremes. Thus, the one critic suggests that the coarser Tales were probably composed after Chaucer's loss of his wife; the other considers that the ponderous theological portions of the Tales were the expression of Chaucer's period of poverty and anxiety. As actual evidence, such opinions have no solid value. Facts which do serve as basis for a chronological theory are to be found in the Chaucerian text. When in the Legend of Good Women prologue (of 1386?) Chaucer alludes to the stories of Palamon and Arcite and of St. Cecilia as in existence, this is a bit of definite evidence, already well used. When in the Gg text of the Legend of Good Women prologue and in the Tales of Wife, Merchant, and Franklin, Chaucer speaks of and uses a book unmentioned by him elsewhere, we may, with Professor Lowes,1 argue that this group of work was of similar date, and subsequent to Chaucer's introduction to the material which he applies so lavishly and with such enjoyment. Or when we find Chaucer, in the Monk's Tale, translating the story of Ugolino from the 33d canto of Dante's Inferno, while in the tragedy of Caesar (in the same Tale) he amalgamates Brutus and Cassius into one person, shall we not argue that the Caesar was written before he knew Dante? for in the 34th canto of the Inferno Brutus and Cassius are unmistakably separate persons. Compare the two translations of Virgil's "pernicibus alis", and the probability that the wrong one antedated the right, noted in Professor Lowes' paper, page 857. Skeat's note on this amalgamation is entirely pointless; what the writer of the Serpent of Division blundered in throws no light upon Chaucer, from whom the later author (Lydgate?) probably obtained the error; and no remark is made upon the curious fact that the same mistake appears in King Aelfred's translation of Boethius. Add here also that a line in the tragedy of Nero (487) is from Dante, as noted by Cary, and we query whether we cannot safely date the four "modern instances" and Nero (at least) after Chaucer's acquaintance with Dante, the Caesar before that time. Should not the Croesus, moreover, be dated later than most of the tragedies, as the earlier form of the endlink (see above) makes no citation from it? 1 Publ. Mod. Lang. Ass'n 20: 749 ff. See Tatlock in Mod. Phil. 3: 367. Again, does not the miswriting alto stilo for Petrarch's alio stilo, translated by Chaucer as high style (Clerk's prol. 41, Tale 1092), make it probable that Chaucer executed that translation before he punned on "high style" in the Squire's Tale lines 105-6?1 Similarly, might not the agreement in idea and wording between the "Verba Hospitis" and the Melibeus endlink lead us to believe that the former, perhaps the original endlink to the Clerk's Tale, was written earlier than the connective between Melibeus and the Tales of Monk and Nun's Priest? In which case we would surmise that the latter Tales were added to the B2 fragment before the present Clerk-envoy was written, which again, as I argue, antedated the rearranging of that Envoy and the writing of the Merchant headlink which meets it. Further, do not the lines of the Man of Law's headlink, But Chaucer, though he can but lewedly On metres and on ryming craftely with their strong resemblance to the Host's comment upon the Rime of Sir Thopas, suggest to the student that that part of the headlink, at least, was once intended to come later than Chaucer's public failure as a doggerel romancer? For where is the point of the Man of Law's apology for Chaucer unless this apology followed Chaucer's public disgrace? Moreover, is it not possible that this headlink, with its conclusion promising a prose story, once introduced the Melibeus, which still follows Sir Thopas as Chaucer's second narrative? This latter has been suggested by Furnivall, and the only other conjecture regarding the original prose narrative has been that of Professor Lowes, see his paper already cited. It is especially interesting in this connection to find that Lounsbury, Studies I: 418, would date the Man of Law's headlink earlier than the Medea story of the Legend of Good Women; with this consideration and that above advanced, I would suggest for the ML headlink a date parallel not with the second, but with the first, period of work upon the Legend. When discussing such a repetition as that just mentioned between the Man of Law's headlink and the Host's words interrupting Sir Thopas, it is necessary to decide first what view we shall take of Chaucer's tendency to repeat himself. Sometimes these echoes and agreements are of a character which indicate nearly contemporary composition; for instance, the description of the tournament encounter, lines 1745 ff. of the Knight's Tale, and of the seafight in the Cleopatra of the Legend, lines 56 ff.; compare also the refusal of the Knight to describe the wedding feast, lines 25-30, with lines 37-44 of the Cleopatra. Or we might parallel the 1 See Professor G. L. Hendrickson's paper in Mod. Phil. 4: 190; to which I may add that the two MSS of Petrarch's Latin in the library of Corpus Christi College, Cambridge, show in one case alio stilo, in the other case alto stilo. Professor Hendrickson's conjecture is thus established as fact. remarks of the onlooking common people in Knight's Tale 1655 ff. with the Squire's Tale 202 ff.; see moreover the notes by Lowes, loc. cit. page 791. But the case is not so clear in verbal echoes like the following: Trouthe is the hyeste thing that man may kepe Frankl. Tale 751 Trouthe is a thing that I wol ever kepe Allone withouten any companye That it is lyk an heven for to here That ech hir lovede that loked on hir face what eyleth thee to wryte The draf of stories and forgo the corn? Eek Plato seith, whoso that can him rede, For pitee renneth sone in gentil herte Can. Yeo. Tale 491 This line, according to Paget Toynbee, is from Dante, Inferno 5100. Note also that Dante took it from Guido Guinicelli, see the Volgari Eloquio. Our interpretation of these echoes can hardly be uniform. The last-mentioned group of lines must be due to conscious repetition on Chaucer's part; and the two parallels preceding it may well have seemed immaterial to him. But the first four cases have more the appearance of a sort of formula, repeated half-mechanically if the keyword came into Chaucer's memory. The second case of the list is particularly interesting from the fact that there occurs at the |