agast for agast so. In 621, we must read lyte is, not lytel is, if we want a rime to dytees. In 827, I restore the word mansioun; the usual readings are tautological. In 911, I restore toun for token, and adopt the only reading of l. 912 that gives any sense. In 1007, the only possible reading is Atlantes. In 1044, Morris's edition has biten, correctly; though MS. F. has beten, and there is no indication that a correction has been made. In 1114, the right word is site; cf. the Treatise on the Astrolabe (see Note). In 1135, read bilt (i.e. buildeth); bilte gives neither sense nor rhythm. In 1173, supply be. Ll. 1177, 1178 have been set right by Willert. In 1189, the right word is Babewinnes^[8]. In 1208, read Bret (as in B.). In 1233, read famous. In 1236, read Reyes^[9]. In 1303, read hatte, i.e. are named. In 1351, read Fulle, not Fyne. In 1372, adopt the reading of Cx. Th. P., or there is no nominative to streighte; and in 1373, read wonderliche. In 1411, read tharmes (= the armes). In 1425, I supply and hy, to fill out the line. In 1483, I supply dan; if, however, poete is made trisyllabic, then l. 1499 should not contain daun. In 1494, for high the, read highte (as in l. 744). In 1527, for into read in. In 1570, read Up peyne. In 1666, 1701, and 1720, for werkes read werk. In 1702, read clew (see note)^[10]. In 1717, lyen is an error for lyuen, i.e. live. In 1750, read To, not The. In 1775, supply ye; or there is no sense. In 1793, supply they for a like reason. In 1804, 5, supply the, and al; for the scansion. In 1897, read wiste, not wot. In 1940, hattes should be hottes; this emendation has been accepted by several scholars. In 1936, the right word is falwe, not salwe (as in Morris). In 1960, there should be no comma at the end of the line, as in most editions; and in 1961, 2 read werre, reste (not werres, restes). In 1975, mis and governement are distinct words. In 2017, frot^[11] is an error for froyt; it is better to read fruit at once; this correction is due to Koch. In 2021, suppress in after yaf. In 2049, for he read the other (Willert). In 2059, wondermost is all one word. In 2076, I read word; Morris reads mothe, but does not explain it, and it gives no sense. In 2156, I supply nevene.

I mention these as examples of necessary emendations of which the usual editions take no notice.

I also take occasion to draw attention to the careful articles on this poem by Dr. J. Koch, in Anglia, vol. vii. App. 24-30, and Englische Studien, xv. 409-415; and the remarks by Willert in Anglia, vii. App. 203-7. The best general account of the poem is that in Ten Brink's History of English Literature.

In conclusion, I add a few 'last words.'

L. 399. We learn, from Troil. i. 654, that Chaucer actually supposed 'Oënone' to have four syllables. This restores the metre. Read:—And Paris to Oënone.

503. Read 'brighte,' with final e; 'bright' is a misprint.

859. Compare Cant. Tales, F 726.

1119. 'To climbe hit,' i.e. to climb the rock; still a common idiom.

2115. Compare Cant. Tales, A 2078. Perhaps read 'wanie.'

INTRODUCTION TO THE LEGEND OF GOOD WOMEN.

§ 1. Date of the Poem: A.D. 1385. The Legend of Good Women presents several points of peculiar, I might almost say of unique interest. It is the immediate precursor of the Canterbury Tales, and enables us to see how the poet was led on towards the composition of that immortal poem. This is easily seen, upon consideration of the date at which it was composed.

The question of the date has been well investigated by Ten Brink; but it may be observed beforehand that the allusion to the 'queen' in l. 496 has long ago been noticed, and it has been thence inferred, by Tyrwhitt, that the Prologue must have been written after 1382, the year when Richard II. married his first wife, the 'good queen Anne.' But Ten Brink's remarks enable us to look at the question much more closely.

He shows that Chaucer's work can be clearly divided into three chief periods, the chronology of which he presents in the following form^[12].

It is unnecessary for our present purpose to insert the conjectured dates of the Minor Poems not here mentioned.

According to Ten Brink, the poems of the First Period were composed before Chaucer set out on his Italian travels, i.e. before December, 1372, and contain no allusions to writings by Italian authors. In them, the influence of French authors is very strongly marked.

The poems of the Second Period (he tells us) were composed after that date. The Life of Seint Cecile already marks the author's acquaintance with Dante's Divina Commedia; lines 36-51 are, in fact, a free translation from the Paradiso, canto xxxiii. ll. 1-21. See my note to this passage, and the remarks on the 'Second Nun's Tale' in vol. v. The Parlement of Foules contains references to Dante and a long passage translated from Boccaccio's Teseide; see my notes to that poem in vol. i. The original Palamon and Arcite was also taken from the Teseide; for even the revised version of it (now known as the Knightes Tale, and containing, doubtless, much more of Chaucer's own work) is founded upon that poem, and occasionally presents verbal imitations of it. Troilus is similarly dependent upon Boccaccio's Filostrato. The close connexion between Troilus and the translation of Boethius is seen from several considerations, of which it may suffice here to mention two. The former is the association of these two works in Chaucer's lines to Adam—