r is always strongly trilled; never reduced to a vocal murmur, as frequently in modern English.

s (s); as in sit (sit). But voiced to z (z) between two vowels, and finally, as in ryse (rii·zə), to rise, shoures (shuu·rez).

sh (sh), as in modern English, ssh (shsh); as in fresshe (fresh·shə).

u short; (y). The French sound, as in Iuge (jy·jə). Rarely (u), as in cut (kut), ful (ful); which are not French words.

u long; (yy). Not common; and only French. Ex. vertu (vertyy·); nature (natyy·rə).

v (v), as in modern English. But the MSS. very rarely use this symbol. The sound of v was awkwardly denoted by the use of u, followed by a vowel; as in loue (luv·ə), love. In the present edition, v is used throughout to denote the consonant.

we final; (wə), but often merely (u). Ex. arwes (ar·wez); bowe (bò·wə, bòu·ə); morwe (mor·u). So also blew (blee·u); newe (nee·wə).

wh (wh), as in the North of England; not a mere w, as in the South.

For the sound of th, modern English may be taken as the guide; and the same remark applies to the distinction between f and v, and to the variable sound of s. Moreover, every letter should be distinctly sounded; the k in knee (knéé) and the w in wryte (wrii·tə) were still in use in the time of Chaucer, though now only preserved in the written forms.

§ 23. It will readily be understood that the M.E. vowel-sounds were intermediate between those of Anglo-Saxon and of modern English. They can best be understood by consulting the table at p. 42 of my Primer of English Etymology; and, for French words, that at p. 126 of my Principles of English Etymology, Second Series. The pronunciation of M.E. and of Anglo-French vowels did not materially differ. Instead of here reproducing these tables, I give the approximate pronunciation of the first eighteen lines of the Canterbury Tales. But we must remember, that the pronunciation of words in a sentence is not always the same as when they are taken singly, owing to the accent (or want of accent) due to their position. The word his (hiz) may have its initial h aspirated, when standing alone; but in the phrase his shoures, it is taken along with shoures, loses its accent and its initial h, and becomes (iz). Words are much affected by the manner in which they are thus grouped together. I denote this grouping by the use of a hyphen, and mark the accented syllables by a sloping stroke over every accented vowel; as is usual^[25]. The elided final e is denoted by ('). There is no elision at the medial pause; see below (§ 116). The medial pause is here denoted by a sloping stroke, as in the Ellesmere MS.

§ 24. The above example also shews the mode of scanning the lines, as will be more particularly explained hereafter. It will be seen that the normal number of accents in the line is five, though the fifth line, quite exceptionally, has six, with an additional accent at the cæsural pause. It may also be noted here, by the way, that accents are by no means of equal strength. The accents on with in lines 1 and 5, on to in line 2, and on is in l. 4, are but slight; whilst those on the former syllables of straunge and strondes in line 13 are of unusual force.

§ 25. Rimes illustrating the pronunciation of long O and long E.

It has been said that the values of the M.E. vowels are intermediate between those of the Anglo-Saxon and the modern vowels. The best and surest guide to them is afforded by the A.S. sounds, and it is worth while to illustrate this by special instances.

Let us consider the case of the open and close o. These are distinguished by their origin. Thus open long o (ao) arises (1) from A.S. ā; or (2) from the lengthening of A.S. short o at the end of an open syllable. I have observed that Chaucer frequently makes a difference between the open o that arises from these two sources.

The M.E. (ao) from A.S. ā was doubtless wholly long. Examples occur in lore (lao·rə), lore, from A.S. lār; and in more (mao·re), more, from A.S. māra.

But the M.E. (ao) from the lengthening of A.S. short o was probably somewhat less full, or only half-long, or perhaps, as Dr. Sweet suggests, was somewhat closer. At any rate, Chaucer usually makes a difference between this sound and the former. To keep up the distinction, I shall now write (òò) for the former open o, and (ò) for the latter; so that lore and more will be denoted by (lòò·rə), (mòò·rə). Examples of the other (ao) occur in forlore (forlò·rə), from A.S. forloren, forlorn; to-fore (tóó-fò·rə), from A.S. tō-foran; and in the curious word more (mò·rə), a root, from the A.S. mora. In the fourth stanza of Troilus, Book V, Chaucer distinguishes between (òò) and (ò) in a very marked manner, since the riming formula of the stanza is ababbcc, i.e. the first line rimes with the third, and the second with the fourth and fifth. Observe, that Chaucer emphasizes this variation by making a similar distinction between open and close e in the preceding stanza. I here give the pronunciation of the whole stanza; and, in order not to confuse the marks over the (o) with those of accentuation, the accent is here denoted by (·) placed after the accented vowel or syllable.

The same distinction is preserved throughout the whole of the poem of Troilus, as may be seen by the following references, where the numbers refer, not to the lines, but to the stanzas.

lore, more; I. 93. sore, more, sore; I. 96; where the former sore is from A.S. sāre, adv., and the latter sore is of French origin^[26], sore, more, lore; I. 108, 156; II. 81, 192; III. 35. more, sore; III. 139, 151; IV. 19, 129, 161; V. 97, 106, 171. rore (A.S. rārian), sore, more; IV. 54. yore (A.S. geāra), more; IV. 214; V. 8. yore, more, lore, V. 47. evermore, more; V. 117. more, sore, evermore, V. 194. more, evermore, yore, V. 248. Also: more, Antenore; IV. 95; where Antenore, being a proper name, may be treated much as the author pleases. And further: more, restore, IV. 193; V. 239; where the o in restore is due to Lat. au. And lastly, pore, rore, V. 7: where the o in pore is of variable quality, from O.F. povre (Lat. pauperem).

On the other hand, we find another set of words in Troilus, in which the open o was originally short. Examples are: tofore, wherfore, bore, i.e. born; II. 202: from A.S. tōforan; from A.S. hwǣr combined with fore; and A.S. boren. y-shore, bifore, therfore; IV. 143; where y-shore, shorn, is from A.S. gescoren. therfore, bifore; IV. 149. forlore, mŏre, heretofore, V. 4; already noticed above.

In all the above examples, the open o occurs before r; the only other examples of open o from original short o are seen in Book I. stanzas 13 and 30. In both these stanzas we find the riming words spoken, wroken, broken, which obviously belong to the same set. Broken is from A.S. brŏcen; but spoken and wroken are new forms, altered from the A.S. sprecen and wrecen by analogy with the very word broken here used. Chaucer never rimes these words with tōken, from A.S. tācen.

§ 26. An analysis of the rimes in the Minor Poems reveals an exceptional use of but one word ending in -ore, viz. the word more. On account, probably, of its frequency and utility, we find it used to rime with heretofore and heerbefore; both examples occurring in the Book of the Duchesse, 189, 1127. This shews that the rime was permissible, and the difference extremely slight. Nevertheless we find, with the exception of these two instances only, that the Minor Poems again present two distinct sets of rimes: (1) from A.S. ā, the words evermore, namore, more, sore, lore, rore, yore, together with tresore (of F. origin, from Lat. thesaurum); and (2) from A.S. o, the words before, bore, wherfore, lore (A.S. loren), herebefore, tofore.

§ 27. In the Legend of Good Women, the result is just the same. The exceptional rimes are shewn by mōre riming with before, 540, 1516; with y-swore, 1284; and with therfore, 443. But with these exceptions, we find, as before: (1) the set of words more, yore, sore, with the French words store and radevore^[27]; and (2) the set bore, forswore, swore (all past participles), and therfore.

§ 28. In the Canterbury Tales, we find from Mr. Cromie's Rime-Index, pp. 185, 189, that the word mōre is again used exceptionally, riming once with the pp. bore, A 1542, and frequently with before; but we find, further, that before is also used exceptionally, riming once with more and lore, E 789; once with sore, D 631; once with more and yore, E 65; and once with gore, A 3237, from A.S. gār. Similarly, therfore rimes with yore, E 1140. But, with these exceptions, we again find the two sets kept distinct, viz. (1) evermore, namore, more, lore, hore (from A.S. hār), gore, ore (from A.S. ār), rore, sore; together with the French restore; and (2) before, bore, y-bore, forlore, swore, therfore, wherfore^[28].

In spite of all the exceptional uses of the two words more and before, we cannot but see, in the above examples, a most remarkable tendency to keep asunder two vowel-sounds which it must have required a delicate ear to distinguish. This is interesting, as proving exceptional care on the part of the author.

We find, accordingly, that later writers did not take the same pains. Thus, in Lydgate's Complaint of the Black Knight, 218, we find sore (from A.S. sār) riming with tore, pp. (from A.S. toren). In Fragment B of the Romaunt of the Rose, it is startling to find more actually altered to mar or mare (the Northern form) in order to rime with thar (for there), 1854; with fare, 2710; and with ar, 2215.

§ 29. Open and close ō. After making the above investigation, we shall naturally expect to find that Chaucer takes care to distinguish between the open ō and the close one; and such is really the case.

The chief source of long close o is the A.S. and Icel. ō. Ex. bóók, forsóók, dóm, bóne (a boon); from A.S. bōc, forsōc, dōm, and Icel. bōn. The distinction between the two kinds of o is perfectly easy to follow, because the sounds are still kept apart in modern English, in which the old open long o is now a close ō, whilst the old close ō is lowered to the sound of ū (uu).

Easy examples occur in A.S. bān, M.E. boon (baon, bòòn), mod. E. bone; as contrasted with Icel. bōn, M.E. boon (boon, bóón), mod. E. boon (buun). In other words, the mod. E. bone was pronounced in M.E. so as to rime with lawn; whilst the mod. E. boon was then pronounced so as to rime with lone.

A few exceptions occur, shewing occasional relaxations of the general rule. They are doubtless due, as Ten Brink suggests, to a paucity of rimes in some particular ending. Thus, when the long o is absolutely final, as in go (gao), do (doo), Chaucer considers these as permissible rimes, and pairs them together freely; and owing to such usage, we even find agoon (agaon) riming with doon (doon) in Troilus, ii. l. 410. But this is the only instance in Troilus of this character; in all other places, the ending -oon relates to the open o; the riming words being alloon, anoon, atoon, boon (bone), foon (foes, A.S. fan), goon, noon, stoon; to which add roon, it rained, woon, quantity. In the Cant. Tales, B 3127, we find the rime dōm, doom, hōm, home; but words in -ōm are, of course, extremely scarce, so that there was little else to be done. For a like reason, sooth (sooth) sometimes rimes with wrooth (wraoth), Bk. of the Duchesse, 513, 519, 1189; and sothe (soo·dhə) with bothe (bao·dhə), Sec. Nonnes Tale, G 167; Troil. iv. 1035.

With these few exceptions, the rule of distinguishing the two qualities of o is rigorously observed. Thus we find in Troilus, rimes in -òòt, viz. hoot, noot, woot, wroot, A.S. hāt, nāt, wāt, wrāt, ii. 890, 1196, iv. 1261. And we find, on the other hand, rimes in -óót, viz. foot, moot, soot, A.S. fōt, mōt, sōt, iii. 1192. Once more, we find, in the same poem, rimes in -òte, viz. hote, note, grote; cf. A.S. hāte, adv., A.F. note (Lat. nŏta), O. Friesic grāta; iv. 583. And yet again, there are rimes in -óte, viz. bote, fote, rote, sote, from A.S. bōt, fōt, Icel. rōt, A.S. swōte, adv.; ii. 345, 1378, v. 671, 1245. Every one knows the first rime in the Cant. Tales, that of sote, rote, (pronounced as mod. E. soata, roata)^[29].

§ 30. Open and close ē. In like manner, Chaucer distinguishes to some extent, and with certain rather more numerous exceptions, between the open and close long e. This is a somewhat more intricate matter, so that it is best to give the results succinctly. It is also a little more difficult to follow, because modern English has confused the sounds; though they are frequently distinguished by a different mode of spelling, the old open e being represented by ea, and the old close e by ee. A good example occurs in the case of the words sea and see. The former, in Chaucer, is (sae) or (sèè), with long open e; whilst the latter is (séé), with long close e. Both were written see in M.E.; with the result, that the words were spelt alike at that time, though pronounced differently; but are spelt differently now, though pronounced alike. The difference in spelling is due to an Elizabethan habit, when the two sounds were purposely distinguished; and it may be remarked that such words as are spelt with ea are precisely those which still have a peculiar pronunciation in Ireland. Some writers try to denote this by using such spellings as say, tay, baste, mate, and the like, instead of the standard English sea, tea, beast, meat.

§ 31. Stable and unstable ē. The two kinds of ē are best understood by observing their sources.

Before we can shew these clearly, it is necessary to observe that the A.S. ǣ has two values, which must be carefully distinguished. The first, which I shall call 'stable ǣ,' because it regularly produces an open ē in M.E., answers to Germanic and Gothic ai, and is generally due to mutation. Thus hǣlan, to heal, answers to Goth. hailjan, and is mutated from hāl, whole, Goth. hails. This produced M.E. hēlen (haelən), with open ē. Again, M.E. sprēde, to spread (note ea in the modern form), answers to a Gothic *spraidjan^[30]; for, although no such Gothic form actually occurs, we can infer it from comparison with the G. spreiten; cf. G. heilen with Goth. hailjan above.

The second kind of ǣ, which I shall call the A.S. 'unstable ǣ,' because it occurs in forms which are treated both ways in Chaucer, answers to an original Germanic ǣ, Goth, ē, and does not arise from mutation, though it may arise from gradation. Thus the M.E. dēde, deed, A.S. dǣd, answers to Goth. gadēds, a deed, G. That; and the contrast between the vowel in G. That and that in G. heilen, to heal, is very clearly marked. It is from words of this class that some trouble arises.

§ 32. If we inquire further, why there should have been any difference of development in such cases, and how the same form could, apparently, yield both an open ē and a close one, I believe that a clear answer can be given. For it is precisely in such cases that we find different forms in the Old Mercian (or Midland) dialect and in the A.S. (or Southern). Thus, whilst the A.S. (Southern) form of 'deed' was dǣd, the Mercian form was dēd. In fact, the mod. E. deed is clearly Mercian, and that is why it is not spelt with ea in Elizabethan English. Hence Chaucer had, ready to his use, two forms of this word. One was the Southern dèèd, with open ē, from A.S. dǣd; the other was the Midland dééd, with close ē; and, as the Midland dialect was then rapidly gaining the ascendency, he could hardly go wrong if he sometimes used the more popular form. Chaucer knew nothing of etymology, but he knew how words were pronounced by his cotemporaries; a fact which sufficiently explains his habits.

In order to complete this part of the case, it is necessary to add that the M.E. ē which results from A.S. ēa is ALWAYS open^[31].

§ 33. A similar ambiguity occurs in the case of a long e which we should expect to be close. Here again we must distinguish between two kinds. The A.S. ēo yields an M.E. ē which is ALWAYS close; as in dēop, deep, M.E. déép. Again, there is an A.S. ē which results from mutation, as in A.S. blēdan, to bleed, from blōd, blood; and the resulting M.E. ē is ALWAYS close, as in blēden (bléédən), to bleed.

But there is also the UNSTABLE vowel in the M.E. y-sēne, visible. Of this word the A.S. forms are various; we find gesīene, gesȳne, gesēne, all three. Of these, gesīene is the earlier spelling of gesȳne, and may be neglected; but gesȳne and gesēne still remain. Gesȳne is the usual A.S. (Southern) form, whilst gesēne is Midland and Northern. From the Midland gesēne came M.E. ysēne (iséénə), with close e, regularly; and this is the form which Chaucer usually adopts. The A.S. gesȳne would have developed regularly into M.E. ysȳne (isiinə), just as the A.S. mȳs answers to M.E. mȳs, mod. E. mice. But the y-sound was difficult of treatment, as the true sound (yy) was lost; and Ten Brink has observed a corresponding variation in the development of A.S. short y, which became sometimes short i and sometimes short open e in M.E. In the same way, I should suppose that this A.S. long y corresponded to a Kentish long open e; thus producing M.E. ysēne (isèènə), in which the e was open. There is a remarkable example of such a variety in the development of the A.S. fȳr, fire. This usually became M.E. fyr (fiir), with long i; but in Troilus, i. 229^[32], we have the remarkable form afere (afèèrə), on fire, riming quite regularly with were (wèèrə), were (from A.S. wǣron), and with stere, to stir (from A.S. styrian). Indeed stere, to stir, is really another example of the like development, since the e in it is merely lengthened from an A.S. short y.

§ 34. Summary. As this investigation has run to some length, I here give a summary of all the above results.

Open and close ō. 1. The M.E. open and close ō have resulted in mod. E. sounds which are still kept apart; cf. M.E. stòòn and M.E. dóóm with the mod. E. stone and doom.

2. A.S. ā produced M.E. open ō. A.S. o, when lengthened, also produced M.E. open ō. But the two M.E. sounds somewhat differed, and Chaucer avoids riming them together. The few exceptions are noted above; the commonest of these being due to the variable treatment of the words mōre and before.

3. A.S. and Icel. ō produced M.E. close ō. Chaucer avoids riming the close ō with the open one; the chief exceptions being when the vowel-sound is final, and in other cases where rimes are scarce.

4. The different spellings of the mod. E. sea and see, now pronounced alike, answer to the different sounds of the M.E. form see. If the ee was open, it meant the sea; if it was close, it was part of the verb to see.

5. The A.S. ēa produced M.E. open ē.

6. The A.S. ǣ, if answering to Gothic ai, produced M.E. open ē. But if answering to Goth. ē, the M.E. ē was close in the Midland dialect, but was allowed to rime with open ē in Southern; giving Chaucer a choice of forms.

7. The A.S. ēo and ē (if arising from mutation of ō) produced M.E. close ē.

8. In words such as A.S. gesȳne, Mercian gesēne, visible, the M.E. y-sēne had an ē which rimed with open ē in Kentish, and a close ē in Midland, giving Chaucer a choice of forms.

§ 35. It will be now easily understood, that Chaucer's general rule, of avoiding the riming of close ē with open ē, admits of a considerable number of exceptions, in which the ē is really of a doubtful or unstable character.

It is clear that, in considering Chaucer's forms, we must set aside, as UNSTABLE, all words in which long e corresponds either to a Germanic ǣ (Gothic ē, German ā), or otherwise to A.S. unstable ȳ (Mercian ē). I proceed to enumerate the chief of these, as occurring, first of all, in Troilus.

Words ending in -eche. The verb ēche, to eke, answers to A.S. ȳcan. Leche, a leech, is allied to Goth. lēkeis, a physician. Speche, speech, is from the stem seen in sprǣc-on, they spoke, with the same vowel, originally, as in Goth. brēkun, they broke. All these words have unstable e.

-ede. Dede, deed; A.S. dǣd, Goth. gadēds. Drede, to dread, A.S. on-drǣdan, O.H.G. trātan. From V. 1654-7, it is difficult to draw any clear inference; brede should have open ē (cf. A.S. brād, Goth. braids); hede, heed, goes with A.S. hȳdan, and its vowel is unstable; and Diomede, though the e should be close, is at proper name, and needs no exact treatment.

-eke. Besides the correct form èèk (A.S. ēac), Chaucer has a form eke, with unoriginal final e; he probably connected it with the verb eche, to eke, in which the e is unstable, as it arose from mutation.

Cheke answers to A.S. cēace, Anglian cēce, mod. E. cheek; but here the ēa is not the usual A.S. ēa, being merely due to the initial c, and the West-Germanic type is *kākā (New E. Dict.), answering to Germanic *kǣkā; whence the A.S. original form *cǣce; so that the e is unstable, by the rule above given.

-ele; -ene. Rimes in -ēle and -ēne are all regular. So also in -eme, -emeth. The rimes in ēmen are imperfect.

-epe. Slepe has unstable e; cf. Goth. slēpan.

-ere. Unstable e occurs in fere, fire, as explained above; also in here, to hear, A.S. hȳran, hēran; and again, in dere, dear, A.S. dȳre (as well as dēore). Also in yere, year, because the ēa in A.S. gēar is not the usual diphthong ēa, but due to the preceding g; the Goth. form is jēr, so that the M.E. is unstable, by the rule. Bere, a bier, is from the verbal stem bǣr-on, corresponding to Goth, bērun; hence the e is unstable.

But a real exception occurs in the riming of lere, to teach, with here, here (T. ii. 97, iv. 440). Lere, A.S. lǣran, Goth. laisjan, should have the open e; but it here rimes with a word in which the e is close. This is one of the exceptional words noted by Ten Brink (Chaucers Sprache, § 25). No explanation is offered, and I know of none, unless it be that it was confused with lére, cheek, from A.S. hlēor. But we must note the fact.

-ete. The exceptional words are bihete, mete (to dream), strete, street. Bihete is really a false form for bihote (A.S. bihātan); the e is due to confusion with the pt. t. bihēt, where hēt is for A.S. hēht, the result of contraction; hence the e is doubtful and unstable. Mete, to dream, is from A.S. mǣtan, of unknown origin; hence we may regard the e as doubtful. Strete, a street, answers to A.S. strǣt, Mercian strēt, mod. E. street; hence the e is unstable, as explained above.

-eve. Ten Brink (Ch. Studien, §§ 25, 23) thinks that leve, sb., leave, was treated as if with close e by confusion with bilēven, to believe, which, he says, has close e. Whatever be the right explanation, we must set aside leve, leave, as an exceptional word. So also eve, eve, A.S. ǣfen, Mercian ēfen, has a variable vowel; see Sweet, O.E. Texts, p. 602.

§ 36. Having now considered the doubtful cases, which may be altogether set aside, it remains to draw up the list of words in which the quality of the long e, at least in Troilus, admits of no doubt. The result gives us a valuable set of test-rimes, by which the genuineness of a poem attributed to Chaucer may be investigated. Of course, a few divergences may admit of explanation; but the presence of a large number of them should make us extremely suspicious.

The list is as follows.

(A) The following words (in Troilus) have open e only. (I omit some doubtful cases, in addition to those discussed above; and only give those which ought certainly to have the open vowel.)

teche, to teach.

dede, dead; lede, lead (the metal); rede, red. Also lede, to lead; sprede, to spread. Other words in -ede are doubtful.

breke, to break, speke, to speak, wreke, to wreak, have open e; but it was originally short, and these words are kept apart from others.

bene, bean; clene, clean; lene, lean; mene, to mean.

hepe, heap; lepe, to leap.

there, there; were, were; where, where. Also ere, ear; gere, gear; tere, a tear. (Fere, fear, has unstable e; cf. G. Gefahr.)

bere, to bear, dere, to harm, swere, to swear, tere, to tear, besides bere, a bear, spere, a spear^[33], were, a weir, here, her, stere, to stir, likewise have open e; but the e was originally short, and these words are kept apart from those in the preceding set.

bete, to beat; grete, great; hete, heat; spete, to spit; swete, to sweat; threte, to threat. Also ĕte, to eat, foryĕte, to forget. (I omit doubtful cases.)

reve, to reave; greve, a grove. (But leve, to leave, is doubtful.)

(B) The following (in Troilus) have close long e only.

seche, to seek; biseche, to beseech.

forbede, to forbid; nede, need; yede, went. Also bede, to offer, blede, to bleed; brede, to breed; fede, to feed; glede, a glowing coal; spede, to speed; stede, a steed.

meke, meek; seke, to seek.

bitwene, between; grene, green; kene, keen; quene, queen; tene, vexation; wene, to ween.

kepe, to keep; wepe, to weep; also depe, deep.

fere, companion; yfere, together; here, here.

bete, flete, grete, mete, to mend, float, greet, meet; swete, sweet.

leve, dear.

§ 37. Of course, the rime-tests consist in this, that not one of the words in class A can possibly rime with one of those in class B, either in Troilus or in any genuine work of Chaucer.

To test this, we must first refer to Cromie's Rime-Index to the Canterbury Tales, under the headings, -eche, -ede (-eede), -eke, -ene, -epe, -ere, -ete, -eve.

The only apparent exceptions that I can find are two; and they are worth notice.

Under -eepe, we are told that leepe, 3 s. perf., rimes with keepe, n. obj. The reference is to Group A, 2688. When we look, we find that the Ellesmere MS. has wrong spellings; the words should be leep, keep. Or rather, we find that the final e is not real, but only represents a meaningless flourish in the MS. Now it is a neat point of grammar that, although lepen, to leap (A.S. hlēapan), has an open e, its past tense (A.S. hlēop) has a close e; so that the rime is quite correct. In both words, the e is close.

The other case (A 1422) is worth citing. Mr. Cromie says, at p. 108, that here, adv., rimes with the inf. bere, to bear; which is, in my view, impossible.

The lines run thus:—

This is a case where the sound decides the sense. The e in bere is properly short; hence the same is true of here. Accordingly, here is not an adverb, nor does it mean 'here'; it is the personal pronoun, A.S. hire, and it means 'her'; precisely as it does in Troilus, ii. 1662.

§ 38. In the Minor Poems, the following passages are the only ones that I can find that present any difficulty.

In the Death of Blaunche, 1253, we find need riming with heed (head); so that need has here, apparently, an open e. Ten Brink has noted this exception (at p. 20), and explains it by remarking that there is a double form of the word in A.S., viz. nēad as well as nēod. At any rate, we see that the word nede cannot be relied on as a test-word, and must be struck out; though there is only this one example of its use with open e.

In the Death of Blaunche, 773, we find dere (dear) riming with were, were. And once more, viz. in Clk. Ta., E 882, we find were riming with dere; but, after all, dere (see § 35) has unstable e. The Death of Blaunche presents many difficulties, and the text of it is far more uncertain and unsatisfactory than that of any other genuine poem.

In the House of Fame, 1885, we find the rime here (here), lere (to teach). This only shews that lere is here once more used with the close e; I have already said (§ 35) that it is no sure test-word.

I just note the rime of here (here) with were (perplexity); H. Fame, 980. Were is of F. origin; and several such words have the close e; see Ten Brink, p. 48.

In the Legend of Good Women, 1870, we have the unusual rime there (there) with dere (dear). Ten Brink has noted this (p. 20). He remarks that it is the only example in which there seems to have close e; but it is rather one of three cases in which dere has open e (from A.S. dȳre).

These are all the difficulties which I could find, after a search through the Index to the Minor Poems. The only modifications they suggest are these: the word need is once found riming with heed (head); and the word dere (though it usually has a close e) really has unstable e (A.S. dēore, dȳre).

It is interesting to apply the results to other Poems.

The beautiful Roundels entitled Merciless Beauty answer the test surprisingly (§ 4). In the first stanza, the author uses the rimes sustene, kene, grene, quene, sene, where all the vowels are close, if we include sene, which has the variable e (close in Midland). In the second stanza, the rimes are pleyne, cheyne, feyne, atteyne, pleyne, all of French origin, in which the sound is slightly varied to that of the nearest diphthong. And in the third stanza, we find lene, bene, mene, v., clene, mene, s., in which the e is now open.

In the poem called A Compleint to his Lady, the final stanza of which, with Chaucer's name appended, was discovered by Dr. Furnivall after I had claimed it for Chaucer, every rime is entirely perfect, and many of them are highly characteristic of him, being used elsewhere very freely.

The poem which I have called An Amorous Complaint has every rime perfect, except in l. 16, where the author rimes do (with close o) with wo, go (with open o). It has already been shown that Chaucer frequently does this very thing (§ 29).

§ 39. This shews one side of the argument. It is instructive to turn to a piece like The Complaint of the Black Knight, which we now know to be Lydgate's, as printed in the Aldine Chaucer, vi. 235. In the very first stanza we find white riming with brighte and nighte, which, to the student of Chaucer, is sufficiently astonishing. Other non-Chaucerian rimes are seen in pitously, malady (st. 20), where the form should be maladye, and the same error occurs in st. 27; in ageyn, tweyn, peyn (34), where the latter forms should be tweyne, peyne; in forjuged, excused (40), which is not a true rime at all; in ywreke, clepe (41), a mere assonance; in feithfully, cry (65), where I cry should rather be I cry-e; in wrecche, with short e, riming with leche, seche (68); seyn, peyn (for peyn-e, 82); went (for went-e), pt. t., shent, pp. (93); peyn (for peyn-e), ayeyn (93); quen-e, dissyllabic, seen (miswritten sene), monosyllabic, (97). Here are twelve difficulties in the course of ninety-seven stanzas; but there are more behind. For the test-words already given above would alone suffice. The riming of sōre with tore (A.S. toren) has already been noticed, in § 28. In st. 4, we find swéte, sweet, paired off with hète, heat; in st. 18, we find gréne paired off with clène; and in st. 86, we have rède, red, paired off with spéde, to speed. That is, we have here four exceptions in the course of 97 stanzas, being more than can be found in the whole of Chaucer's genuine works put together. In fact, the indiscriminate riming of close and open e is a capital test for Lydgate and for work of the fifteenth century. Using this test alone, we should see cause to suspect The Flower and the Leaf, which has three false rimes of this class, viz. ète, to eat, swéte, sweet (st. 13); bète, pp. beaten, actually riming with the pp. set (31); and gréne riming with clène (42); not to mention that the author makes the dissyllabic words wene, grene, rime with the pp. seen (36); and again, grene, tene rime with the pp. been (56); and yet again, grene rime with the pp. seen (57), and with been (77). On this point alone, the author differs from Chaucer SEVEN times^[34]!

The Court of Love differs from Chaucer in instances too many to enumerate; but, as to this particular point, I only observe the riming of gréne with clène, l. 816; and of dére with require, l. 851; but we may alter require to the Chaucerian form requere. At l. 79, we find the dissyllabic grene; it rimes with the monosyllable been.

§ 40. Similar tests apply to open and close o. We might arrange these, similarly, into two classes, viz. (A) with the open sound, and (B) with the close sound; and we should find that they do not rime together; i.e., if we first eliminate those words which are observed to be of a variable character. For a few exceptions, see § 29. I give the list below.

It is also curious to observe that, in Troilus, the words wolde, nolde, sholde, usually rime together. Wolde rimes with biholde once only, iii. 115; but sholde never rimes with any words but wolde and nolde. In the Cant. Tales, wolde rimes with several words, but sholde only with wolde and nolde. The only exception is in the Book of the Duchess, 1200, where sholde rimes with tolde. It would greatly improve the sense as well as the metre to substitute wolde for sholde in this passage.

§ 41. Now that I have exemplified the mode of using these test-words, I give fuller lists, slightly augmented by help of Mr. Cromie's Rime-Index, and adding a third class (C) of words which have a variable vowel, and are therefore not available as test-words; for it is useful to know the character of these also.

The following is THE KEY to the meaning of the lists.

1. (A) contains words with open long e and open long o. The chief sources of open long e are (1) A.S. ēa and (2) the stable A.S. ǣ answering to Goth. ai (O.H.G. ei) and usually due to mutation of A.S. ā. We may include words with A.S. short e, though these often keep the vowel somewhat short; perhaps it was only half-long.

The sources of open long o are (1) A.S. ā and (2) a lengthening of A.S. short o; perhaps the latter was only half-long.

2. (B) contains words with close long e and close long o. The chief sources of close long e are (1) A.S. ēo and (2) A.S. ē (from mutation of ō). The chief source of close long o is A.S. ō.

3. (C) contains words with variable long e and variable long o. The chief source of variable long e is the unstable A.S. ǣ answering to Gothic ē (Germanic ā); this ǣ occurs in sprǣc-on, third stem of the strong verb sprecan, and in its derivative sprǣce, whence M.E. speche, speech. It also appears to arise from sounds corresponding to A.S. īe, ȳ, mutation of ēa, ēo.

Chaucer's use. Words in (A) rime with each other, but never rime with words in (B). Words in (B) rime with each other, but never with words in (A). Words in (C) rime with words both in (A) and (B).

-eche. (A) tèche, bitèche. (B) séche, biséche. (C) eche, to eke, leche, speche.

-ede. (A) dede, dead, hede, head, lede, lead (metal), rede, red, sprede, to spread. (B) bede, to offer, blede, v., brede, v., crede, fede, forbede, glede, nede^[35], spede, v., stede, a steed. (C) dede, deed, drede, s. and v., hede, to heed, rede, to advise. Words in -hede almost always shew open e, but a few exceptions occur.

-eke. (A) brĕke, v., spĕke, v., wrĕke, v., awrĕke, ywrĕke, with (original) short e; leke, leek. (B) meke, seke, v., seke, sick, biseke.

-ene. (A) bene, bean, clene, lene, adj., mene, to mean, unclene. (B) bitwene, grene, kene, quene, tene, vexation, wene, v. (C) sene, adj., visible, y-sene (the same), shene, bright^[36].

-epe. (A) chepe, to buy, hepe, lepe, v., stepe, bright. (B) crepe, v., depe, kepe, wepe. (C) slepe.

-ere. (A) bĕre, a bear, bĕre, to bear, dĕre, to harm, ĕre, to plough, hĕre, her, spĕre, spear, stĕre, to stir, swĕre, to swear, tĕre, to tear, wĕre, a weir, wĕre, to defend; all with (original) short e. Also ere, ear, gere, gear, tere, tear; and there^[37], were^[38], where. (B) fere, companion, here, here, yfere, together. (Here belong the F. words, chere, clere, manere, matere, spere, sphere.) (C) bere, bier, dere, dear^[39], fere, fear, here, to hear, lere, to teach, yere, year.

-ete. (A) bete, to beat, grete, great, hete, heat, spete, to spit, swete, to sweat, threte, v., wete, wet, ybete, beaten. Also ĕte, to eat, foryĕte, to forget, mĕte, meat (originally with short e). (B) bete, to mend, flete, to float, grete, to greet, swete, sweet. (C) bihete, to promise, forlete, to let go, lete, to let, mete, to dream, shete, sheet, strete, street.

-eve. (A) bireve, deve, pl., deaf, greve, grove, reve, to reave. (B) leve, dear, reve, a reeve. (C) eve, eve, leve, to believe, bileve, belief, leve, to permit. Note that yeve, to give, usually rimes with live, to live, as in mod. English.

-o. All words in -o are allowed to rime together; of these, to, therto, unto, do, fordo should have the close sound.

-olde. Nolde, sholde, wolde, usually rime together. Occasionally wolde rimes with other words. In only one case does sholde rime with tolde (B. Duch. 1200), where wolde would make better sense.

-one. (A) alone, echone, bone, bone, grone, to groan, lone, loan, mone, to moan, one, one. (B) bone, boon, eftsone, mone, moon, sone, soon. (C) done, to do. [Note that sŏne, son, wŏne, to dwell, are really written for sune, wune, and only rime with each other.]

-onge. [Note that songe, pp., spronge, pp., tonge, yonge, are really written for sunge, sprunge, tunge, yunge. They rime together, but are quite distinct from fonge, honge, longe, stronge, wronge; just as in mod. English.]

-ook. (A) ook, strook. (B) awook, book, cook, forsook, hook, look, quook, shook, took, wook.

-oot. (A) boot, he bit, goot, goat, hoot, hot, noot, know not, smoot, smote, woot, know, wroot, wrote. (B) foot, moot, must, soot.

-ooth. (A) clooth, gooth, looth, ooth, wrooth. (B) dooth, sooth, tooth.

-ore. Bifore, bore, pp., born, forlore, pp., more, a root, shore, pp., swore, pp., therfore, wherfore, originally had a short o, and usually rime together. Hore, pl., hoary, lore, more, rore, sore, yore, have open long o, and usually rime together. In a few cases, bifore and more rime with words in the other set.

-ote. (A) grote, groat, hote, hot, throte, throat (from A.S. þrotu). (B) bote, satisfaction, fote, rote, root, swote, sweet.

The above lists are offered for what they are worth. I believe them to be fairly correct; but they may not be quite exhaustive. Nevertheless, they record ascertained facts; and the facts remain true and useful, even if the theories be wrong.

§ 42. Some peculiarities of rime.

The subject of Chaucer's rimes is fully discussed by Ten Brink; Studien, p. 190. As the critical reader will necessarily consult this work, it is only necessary to give here a few of the chief results.

Chaucer's rimes are usually either (1) masculine, or (2) feminine. Masculine rimes are those in which the rime is confined to a single final syllable, as 'licour,' 'flour'; Prol. l. 3. Feminine rimes are those in which the rime extends through two syllables, as 'sote,' 'rote'; Prol. l. 1. It is necessary to remember that every unaccented final e at the end of a line is to be sounded, and constitutes a syllable.

Sometimes the rime extends, apparently, over more than two syllables; but it will be found that, in such a case, the penultimate or antepenultimate syllable can either be suppressed, or consists of the shortest possible sound. Ex. swévenis, swéven is, really swév'nis, swév'n is; B 4111. Béryis, méry is; B 4155. Victórië, glórië; A 2405. Mercúrië, múryë; A 1385. Máriëd, táriëd; B 3461. Bériëd, a-blákebériëd; C 405. To-scát'red, y-flát'red; D 1969. Contrárië, Ianuárië; E 2319; &c.^[40] Note that feminine rimes are extremely numerous, and are sometimes kept up through whole stanzas in such a poem as Troilus. Thus, in Troilus, iii. 407-434, we find four consecutive stanzas, or twenty-eight consecutive lines, in which every rime is feminine; and this is by no means an extreme case. Feminine rimes are extremely old in English, and are found even in Anglo-Saxon.

§ 43. The most striking examples are those in which the feminine rime is composed of two distinct words, as these prove at once the reality of the final -e. Thus Ro-me rimes with tó me; A 671. You-the rimes with allow thee (aluu·dhe); F 675. Ty-me, with by me; G 1204. Similarly, the final -es of the plural substantive constitutes a syllable, as shewn by such a rime as werk-ës, derk is; G 64. In such a case, some scribes write werkis for werkes, to make the rime more complete, but it is quite needless, as there is no necessity for an absolute coincidence of vowel-sound in a mere unaccented syllable. In Lenvoy a Scogan, 15, it would be quite absurd to alter goddes to goddis(!), merely because it rimes with forbod'is; the really weak part of the rime is in the linking of the short o in goddes, with the longer o in forbode. For the same reason, the rime of lyte is with dytees (HF. 621) is good enough; indeed, we cannot write dytis (as Ten Brink proposes to do) because the word meant is the plural of ditee. Unusual rimes of this sort are still in common use, especially where a slightly humorous effect is intended; and this may very well excuse the above examples, as well as such rimes as Davit^[41] (for David), eructavit, D 1933; saveth, significavit, A 661; wounded, wounde hid, B 102; agon is, onis^[42], D 9; and the like.

§ 44. There are several cases in which the rimes are rather to be considered as permissible than exact. The frequent riming of go (gao) with do (doo) has already been noted. Similarly, owing to the paucity of words ending in open ē, the word sèè, sea, is allowed to rime with close ē. The proper M.E. form of 'beast' is bèèst, which rimes, exactly, with èèst, east, and with almèèst, almost; but, inexactly, with forèst, in which the e is short. Yet, in Sir Thopas, B 1944-8, we find the words forest, best, est, almest, all reduced by the scribe to the same apparent form. In G 1324, we find bréést (A.S. brēost), breast, riming with préést, priest, exactly; but elsewhere bréést is treated as if the e were short, so that it rimes with lest (Kentish form of lust), A 2983; E 617. The mod. E. form suggests that the vowel was beginning to be shortened. In the rime upŏn, gōn, G 562, the o in the former word is short, but in the latter is long; both are open, and the rime is admissible. A similar variation in vowel-length is seen in the riming of hădde, had, with blāde, blade, A 617, and with spāde, spade, A 553; and here again, some scribes try to better the matter by using the form hade. The rime is really (had·də), (spaa·də); and the right lesson to be learnt from it is, that the a in spade was still (aa), and thus very different in sound from the a in mod. E. spade (speid). Long and short u are rimed in hous (huus), Caucasus (kau·kasus·) D 1139; and elsewhere. Note neyghebores, dores^[43], i.e. (nei·həbuu·rez), (du·rez); in HF. 649. One of the most licentious rimes is in Troil. ii. 933, viz. riden, abiden, yeden, properly (rid·n), (abid·n), (yééd·n); which suggests that yeden is here (yĕd·n); and we are reminded of the M.E. form of the verb 'to give,' which hovers between yĕven and yiven, and rimes in Chaucer with liven, to live, though frequently written yeven. The singular form yēde rimes with nede (néé·də) in G 1280, and with dede (déé·də) in G 1140.

Chaucer certainly sometimes uses two forms of the same word; the most noticeable are heer and here for 'here'; theer and there for 'there'; eek and eke for 'eek.' These can be explained by the tendency to add a final -e in adverbial forms. Of course the double form was highly convenient. Remarkable double forms are chivachyë, A 85, and chevauchee, Mars, 144; perryë, A 2936, and perree, B 3550.

§ 45. Repetitions. Such rimes as aff-ecciouns, prot-ecciouns, F 55, wherein the penultimate and antepenultimate syllables are repeated, are disliked by later writers. Chaucer had found many such in Le Roman de la Rose^[44]. In discussing such repeated rimes as seke, to seek, seke, sick, A 17, we must remember that they are common in Old French poetry, though it was usual for the poet to take care that the repeated forms should be used in different senses. This rule Chaucer usually observes; cf. séé, see, sèè, sea, A 3615; here, here, here, to hear, A 4339; style, style, style, a stile, F 105; fern, fern, fern, long ago, F 255; &c. But he also allowed himself such repetitions as nones, noon is, A 523; clerkes, clerk is, B 4425; places, place is, D 1767; &c. We now avoid such rimes as acordes, cordes, HF. 695; acorde, recorde, Parl. Foules, 608; and still more, such rimes (all too easy) as goodnesse, soothfastnesse, E 793; soothfastnesse, wrecchednesse, I 34; more, evermore, Anelida, 240.

§ 46. Mistakes as to Chaucer's uses. Some of the facts concerning Chaucer's rimes have been misunderstood, even by so good a scholar as Prof. Lounsbury, in his Studies of Chaucer, vol. ii^[45]. It is therefore desirable to point out some of these errors.

He calls attention, among others, to the following false rimes:—

Desyre, manere, T. iv. 817 (p. 54). But the right reading is martyre, which alone makes sense. For the actual use of the false rime here censured, see Rom. Rose, 2779.

Kinde, binde, wende, T. iii. 1437 (p. 54). Read winde, that thou mayst wind. 'Gower will furnish a number of similar illustrations' (p. 54)^[46].

Prof. Lounsbury is extremely anxious to prove that assonances (i.e. such imperfect rimes as we see in kepe, eke, with a mere correspondence in the vowel-sound only) occur in Chaucer; and endeavours to strengthen his position by considering various difficult rimes. At p. 60, he says: 'All difficulty with crown and person (R.R. 3201) disappears the moment they receive the forms coroun and persoun (as in Gower, iii. 112, 141, 227, 234).' But Gower has no such forms; he has coróne, persóne in every instance, emphasised by the use of coróned, enviróned (iii. 112), and by such lines as, 'If it in his persón-e be'; ii. 202. Chaucer rimes persone with allone, D 1162; and with done, T. ii. 701, 1485, iv. 83; and he uses the forms córone or córoune and coróne. But R.R. 3201 has, 'And on hir heed she hadde a crown'; and, only two lines below, has the dissyllabic crownet.

'Gower,' we are told, 'rymes the preterite had with bed, leiser with desire, and dore, a door, with the verb dare, in the form dore'; p. 64. Gower does none of these things; he rimes the correct preterite hedde^[47], which means 'hid,' and which Pauli (regardless of sense) turns into hadde, with the form a-bedde (i. 256). Further, he rimes desir with leiser, according to Pauli (ii. 95); but there is no reason why Gower may not have meant to use the form leisir, since that is the true A.F. form corresponding to O.F. loisir (still in use)^[48]. Lastly, Gower rimes dore (durə), a door, with dore (durə), the 1st p. pr. subj. of the verb durren, to dare, corresponding to A.S. durre (ii. 96). The fact that the pres. indicative is dar, with a different vowel, has nothing to do with the passage in question. It is the critic, not Gower, who is here at fault; even Gower must have known that dar is monosyllabic, and could not possibly rime with the dissyllabic sb. dore.

Chaucer uses 'the pp. smitted for smitten'; T. v. 1545; p. 65. Not so; smitted and smitten are totally different words.

Chaucer uses 'the form houn for hound'; T. iv. 210; p. 65. What howne means, I do not know; but, as it is dissyllabic, it cannot mean hound; nor has it any connection therewith.

'In HF. 959, the infin. demeine is found riming with seyen'; p. 71. Not so; it rimes with the dative of the infinitive, to seyne (A.S. tō secganne); precisely as to seyne rimes with reyne in F 313. In the face of this quotation, the next remark loses all its point, viz. that 'the suggestive fact about this peculiarity of ryme is that it is not found in the Canterbury Tales'; the answer being, that it is found there. So again, we find to seyne, peyne, Parl. Foules, 78.

Next we read—'if it be contended that the usage is based upon the derivation of one of the forms from the A.S. gerundial ending -anne, it is enough to reply that its occurrence in these cases is not borne out by the poet's practice elsewhere'; p. 71. Of course, it is not enough; for we cannot divorce Chaucer's language from the general usage of Middle-English, in which very few forms of this character had survived. Even if it were enough, the assertion that there is no other such case happens not to be true; for we often find to done; as in A 3543, 3778, B 770, D 2194, F 334, G 932, I 62.

And again, we find to sene, riming with grene, A 1035. And yet again, to bene, Rom. Rose, 1265. It is impossible to respect arguments which derive all their apparent force from the principle of heaping one mistake upon another.

§ 47. It is tedious to reply to special pleading of this kind. Thus, at p. 72, I am quoted, correctly, as objecting to the false rime in R. Rose, 1981, where the acc. pl. feet is made to pair with the infinitive lete. And we are told that 'the force of this example is altogether impaired by the fact that in the Man of Lawes Tale (B 1104) the same plural rimes with the infin. mete.' So far from impairing my argument, the 'fact' strengthens it immensely; for, in that passage, we have no longer to do with the acc. feet, but with the dative plural in the phrase to fet-e, answering to the A.S. phrase to fōtum, which just makes all the difference. Correctly, it should be to fōte; but the ē was, by this time, so strongly associated with the plural use, that to fēte took its place.

We see that the e was sounded, because there is a third riming word, in the phrase in the strete. Stratmann's Dictionary duly notes this very passage. It is, however, true that Chaucer is not always consistent about this; he has under fete, riming with swete, Book of the Duchess, 399; in a strete, riming with on my fete, HF. 1049; but in the Cant. Tales, we find at his feet, A 2047; al about hir feet, A 2075; unto his beddes feet, A 4213. The one thing which he does not do is to use fete in the accusative, which is precisely what the author of Fragment B of the Romaunt does; unless, as is more likely, he drops the -e of the infin. lete, which Chaucer invariably keeps (at any rate when final). We can easily understand the suppression of a final e; but it is difficult to understand why a writer should invent one.

Once more, when I argue that the rime of entente with the adj. present in R. Rose, 5869, does not accord with Chaucer's usage, the reply is made (p. 72) that entent rimes with the pp. shent in the Man of Lawes Tale (B 930). But it is clear that Chaucer here has entente as usual, and rimes it with the form shent-e, which is the pp. treated as a plural adjective; as in several other places.

Next (on p. 72), Gower is rated for riming the prep. for with the pp. forlore; Gower, C.A. ii. 239. But Gower's phrase is 'that thou art comen fore'; and I suspect that he knew the language of his own time. The fore may answer to the A.S. fore, on account of (Grein, i. 320); or, more probably, that ... fore was taken as the equivalent of therfore, which constantly takes the final e, as in Chaucer, E 1141.

On p. 72, again, it is said that, in F 1273, Chaucer rimes the pt. t. broght-e with nought, i.e. he uses the incorrect form broght. This charge, for once, is quite true, and it is as well to say at once, that Chaucer's rimes are not quite immaculate; but his sins of this description are not, after all, very numerous, and not by any means so numerous as Prof. Lounsbury, for the purpose of his argument, would have us believe. The only right method is to make out a fair list, without straining to make it much worse than it should be.

§ 48. In his Studies, vol. i. pp. 402-5, Prof. Lounsbury makes another attack upon the unfortunate poet's rimes. Many of his instances are wrong; so much so, that four of Chaucer's supposed errors and two of Gower's are admitted to be no errors in vol. iii. 453. It would have been well if all the rest of the charges had been withdrawn at the same time. I here draw attention to them accordingly.

'In Parl. Foules, 121, the preterite broughte rymes with the pp. wrought.' Answer; the rimes are: broght-e, y-wroght-e, thoght-e; the form y-wroghte occurs in the phrase 'with lettres large y-wroghte,' where y-wroghte is treated as a plural adjective; and there is no error at all.

'In Troilus, i. 463, the pp. fled rymes with the preterite bredde.' As before, the phrase is: 'Alle othere dredes weren from him fledde.' Here fledde is treated as a plural adjective, and there is no error at all. One would have thought that Chaucer knew something of the language of his time.

'In Troilus, ii. 1079, the pp. excused [rimes] with the preterite accusede.' But the preterite of accusen was accused; the addition of the full suffix -ede is rare, and chiefly confined to monosyllabic roots.

'In Troil. iv. 1422, the pp. sprad [rimes] with the preterite hadde' The line ends, 'with herte and eres spradde'; where spradde is treated as a plural adjective. No error.

'In Troil. v. 1758, the preterite mette [rimes] with the pp. whet.' It is the same story; the phrase is 'hir speres weren whette.' No error.

'In the Legend, 786, the preterite heryede rymes with the pp. beryed.' As the usual preterite was heryed (hér-y-ed-e being too cumbrous and almost unpronounceable), there is no error.

'In the Legend, 2384, the pp. served [rimes] with the preterite deservede' But the preterite was deserved. The full ending -ede was seldom added to roots of more than one syllable, least of all when the verb happened to be of French origin. By ignoring the habits of the language of Chaucer's time, such objections might have been largely multiplied; it is surprising to find that so few have been noted.

'In the Knightes Tale, A 2343, the preterite signifyede rymes with the pp. cried.' However, the preterite was signifyed.

'In the Man of Lawes Tale, B 559, the preterite mette rymes with the pp. yshet; [in B 435] the pp. converted with the preterite astertede; [in B 547] the pp. exiled with the preterite bigilede; and [in B 1115] the pp. ymet with the infin. lette and the preterite sette.' All the charges against Chaucer break down. The pp. yshet is properly yshette, plural. The preterite of asterten is asterted. The preterite of bigilen is bigiled. And the pp. ymet should be ymette, plural. A critic who imagines that such cumbrous preterites as astertede and bigilede were in common use, should be asked to read Middle-English authors till he meets with a few examples of them.