but thow wer fals to twoo'; giving six feet at least to the line, and a syllable over.

a.-Addit. 28617. A fair MS., but only a fragment, as already noted (p. xlvii). It confirms many of my readings; as, e.g., in Il. 1995, 2019, 2020, 2199, &c. It varies in 1. 1999, but gives there an excellent reading :-That is nat derk, and ther is roum

and space.

B.-Camb.

B. Camb. Univ. Library, Ff. 1. 6. Contains the Legend of Thisbe only. A late and poor MS., of small account.

y.-Rawl. C. 86 (Bodleian Library). Contains the Legend of Dido only. A poor text, with many errors. Yet it seems to be of the first class, and preserves ll. 960-1. It confirms my readings of 11. 1048, 1074, 1079, 1139, 1144, 1159 1174, 1195, 1196, 1215, 1366.

F.-Fairfax 16 (Bodleian Library). This is the valuable MS. which contains so many of the Minor Poems. It is described in my Introd. to the Minor Poems; vol. i. p. 51. I have taken it as the basis of the edition, though it was necessary to correct it in all the places where the MSS. of the first class have better readings. It is the best MS. of the second class, and Bell's edition does little more than follow it, almost too faithfully, though the editor professes to have collated with it the MS. A. described above. The same text, in the main, reappears in the editions by Thynne, Morris, Corson, Gilman. The scribe is careless, and frequently leaves out essential words; he also omits ll. 249, 487, 846, 960, 961, 1490, 1643, 1693, 1998, part of 2150, 2151, 2152, part of 21532, 2193, 2338 (in place of which a spurious line is inserted in a wrong place), and 2475. Besides this, the scribe often ruins the scansion of a line by omitting an essential word in it, as has already been mentioned. Thus in l. 614, he drops the word for, which occurs in all the other MSS. The scribe often wrongly adds or omits a final e, and is too fond of substituting y for i in such words as him, king. When these variations are allowed for, the spelling of the MS. is, for the most part, clear and satisfactory, and a fair guide to the right pronunciation. Rejected spellings are given in footnotes as far as 1. 924; after which I have made such alterations as are purely trivial without giving notice. Even in l. 1-924 I have changed hym into him, and kyng into king; 1 Not 1491, as Bell says; he has mistaken the line.

"From gelen to gayler; Dr. Furnivall has not got this quite right.

and, conversely, strif into stryf, (where the y denotes that the vowel is long), without hesitation and without recording the change. My text is, in fact, spelt phonetically; and, after all, the test of a text of Chaucer is to read it with the Middle-English pronunciation as given by Dr. Sweet in his Second Middle-English Primer, and to observe whether the result is perfectly in accord with the flowing melody so manifest in the Canterbury Tales.

B.-Bodley 638. Closely related to MS. F., and almost a duplicate of it, both being derived from a common source. B. is sometimes right where F. is wrong; thus in l. 1196 it has houyn, where F. has heuen, See Introd. to the Minor Poems, vol. i. p. 53. Of course this MS. belongs, like F., to the second class. It preserves 1. 1693 (missing in F.); otherwise it omits all the lines that are omitted in F., as well as ll. 157, 262, 623, 1345, 1866; all of which F. retains. Like F., it has a spurious line in place of 1. 2338.

Tn.-Tanner 346 (Bodley). This is a MS. of the second class, strongly resembling F.; see Introd. to the Minor Poems, vol. i. P. 54. It preserves ll. 1693, 2193, 2475; otherwise it omits all the lines omitted in F., as well as the latter half of 1. 1378 and the former half of 1. 1379. It has a spurious line in place of 1. 2338. It is clear that F., B., and Tn. are all from a common source, which was an older MS. not now known.

§ 13. DESCRIPTION OF THE PRINTED EDITIONS. Th.-Thynne's edition; A.D. 1532. This follows, mainly, the MSS. of the second class; its alliance with F., B., and Tn. is shewn by its containing the spurious form of 1. 2338. But it gives the genuine form also, so that in this place three lines rime together. It is more complete than any of those MSS., preserving the lines which they omit (excepting 11. 960, 961), save that it omits ll. 1326, 1327 (doubtless by oversight), which are found in these three MSS., and indeed in all the copies. Probably Thynne used more than one MS., as he sometimes agrees with the MSS. of the first class. Thus, in 1. 1163, he reads vpreysed had, as in C., T., A., P., instead of up-reyseth hath, as in F., Tn., B. He might, however, have corrected this by the light of nature. In ll. 1902, 1923, Thynne alone gives the right reading Alcathoe; unfortunately, both these lines are missing in MS. C. The chief faults of Thynne's edition are its omission of ll. 960, 961, 1326, 1327, and its spurious 1. 2338. Thynne was also unfortunate in following, in general, the authority of a MS. of the second class.

Some later editions.-Later editions appeared in the collected editions of Chaucer's Works, viz. in 1542, (about) 1550, 1561, 1598, 1602, 1687; after which came Urry's useless edition of 1721. Excepting the last, I suppose the editions are all mere reprints; each being worse than its predecessor, as is almost always the case. At any rate, the edition of 1561 is a close reprint of Thynne, with a few later spellings, such as guide in place of Thynne's gyde in l. 969. This edition of course omits 11. 960, 961, 1326, 1327; and gives the spurious 1. 2338.

According to Lowndes, other later editions of Chaucer's Works are the following:-Edinburgh, 1777; 18mo. 12 vols.-Edinburgh, 1782; 12mo. 14 vols.-In Anderson's British Poets, Edinburgh, 1793-1807; royal 8vo. 13 vols.-In Cooke's British Poets, London, 1798, &c., 18mo. 80 parts.-In Chalmers' English Poets, London, 1810; royal 8vo. 21 vols. I suppose that all of these are mere reprints; such is certainly the case with the edition by Chalmers, which merely reproduces Tyrwhitt's edition of the Canterbury Tales, and follows 'the black-letter editions' throughout the other poems. The same remark applies to the edition printed by Moxon in 1855, and attributed to Tyrwhitt as editor. Other editions are those by S. W. Singer, London, 1822, fcp. 8vo. 5 vols.; by Sir H. Nicolas (in the Aldine edition of English Poets), London, 1845, post 8vo. 6 vols. ; and by Robert Bell, London, 1855, 12mo. 8 vols. The last was really edited by Mr. Jephson.

Bell's (so-called) edition was conveniently reprinted in four volumes, in Bohn's Standard Library; a revised edition of this was published in 1878, with a Preliminary Essay by myself. Of the Legend of Good Women, the editor (Mr. Jephson) remarks that 'the text of the present edition is founded upon a careful collation of the MS. Fairfax 16, in the Bodleian Library, and MS. Arch. Seld. B. 24'; i.e. upon a collation of F. with A. It gives us the text of MS. F., with the missing lines supplied from Thynne or from MS. A. It omits ll. 960, 961, and inserts ll. 1326, 1327 in the wrong place, viz. after 1. 1329. At 1. 2338, it gives both the correct and the spurious forms of the line; so that here (as in Thynne) three lines rime together. In 1. 2150-3, the same confusion occurs as is noticed below, in the account of Morris's edition. The chief gain in this edition is that it has a few explanatory notes. Of these I have freely availed myself, marking them with the word 'Bell' whenever I quote them exactly; though

they were really written, as I am told, by Mr. Jephson, whose name nowhere appears, except at p. 12 of my Essay, as prefixed to the revised edition.

The Aldine edition was reprinted in 1866, on which occasion it was edited by Dr. Morris. With respect to the Legend of Good Women, Dr. Morris says that it is copied from MS. F., collated with MSS. A., C. (privately printed at Cambridge by Mr. H. Bradshaw, 1864), and MSS. Addit. 9832 and 12524. In this edition, variations from the MS. (F.) are denoted by italic letters, but such variations are very few. Practically, we here find a correct print of MS. F., with most of the missing lines supplied by collation, and with very few corrections. Lines 960, 961 are, however, still omitted, though found in MS. C.; but ll. 1326, 1327 (also omitted by Thynne) are duly given, being found, in fact, in MS. F. At 1. 2338, the correct line is given, but the spurious line is also retained; so that (as in Thynne) three lines here rime together. In the former part of 1. 2153, a part of 1. 2150 is repeated, giving us by instead of eek; the fact is that the scribe slipped from gayler in 1. 2150 to gayler in l. 2153, omitting all that came between these words. Nothing is said about the interesting form of the Prologue as existing in MS. C. There are no explanatory notes.

Besides the English editions, two editions of the Legend of Good Women have appeared in America, which demand some notice.

Of these, the former is a very handy edition of the Legend of Good Women, published separately for the first time, and edited by Professor Hiram Corson. The text is that of Bell's edition; but the explanatory notes are fuller and better, and I have carefully consulted them. At the end is an Index of all the words explained, which really serves the purpose of a glossary. This is certainly the best edition I have met with.

The other edition is that of Chaucer's Works, edited by Arthur Gilman, and published at Boston in 1879, in three volumes. The Legend of Good Women occurs in vol. iii. pp. 79-183. The harder words are explained in footnotes, and there are just a few notes on the subject-matter. The chief point in this edition is that the editor quotes some of the more remarkable variations in the Prologue from MS. C., which he says is 'evidently an earlier one than the one followed in the text, Fairfax 16, in

the Bodleian Library, Oxford.' Yet his text is a mere reprint from that of Morris; it omits ll. 960, 961, and gives 1. 2338 both in its correct and in its spurious form. Consequently, it contains 2722 lines instead of 2723. The true number of lines is odd, because of the Balade of 21 lines at l. 249.

The net result is this; that none of the editions are complete, and they are all much the same. After twenty editions, we are left almost where we started at first. Thynne's edition was founded on a MS. very closely resembling F., but more complete ; still it omits four lines, and gives 1. 2338 twice over, in different forms. The same is true of all the numerous reprints from it. Bell's edition restores ll. 1326, 1327, but in the wrong place; whilst Morris's edition restores them in the right place. These lines actually occur in MS. F. (in the right place), and could hardly have been unnoticed in collating the proofs with the MS. These editions are both supposed to be collated with MS. A. at least, but the results of such collation are practically nil, as that MS. was merely consulted to supply missing lines. The editors practically ignore the readings of that MS., except where F. is imperfect. Hence they did not discover that MS. A. belongs to a different class of MSS., and that it frequently gives earlier and better readings. But even A. omits ll. 960, 961, though it also rightly suppresses the spurious form of 1. 2338.

§ 14. SOME IMPROVEMENTS IN MY EDITION OF 1889. No real advance towards a better text was made till Dr. Furnivall brought out, for the Chaucer Society, his valuable and exact prints of the manuscripts themselves. This splendid and important work gives the texts in extenso of all the MSS. above mentioned, viz. MSS. C., F., Tn., T., A., and Th. (Thynne's ed.) in the 'Parallel-Text edition of Chaucer's Minor Poems,' Part III; MSS. B., Addit. 9832, P., and Addit. 12524, in the 'Supplementary Parallel-Texts,' Part II; and MSS. a, ß, y, in 'Odd Texts,' 1880. But for the invaluable help thus rendered, the edition of 1889 would never have been undertaken, and I should never have attained to so clear an understanding of the text. I have already said that Dr. Furnivall was the first person who succeeded in numbering the lines of the poem correctly; indeed, most editions have no numbering at all.

I have not thought it necessary to encumber the pages with wholly inferior readings that are of no value, but I have carefully