[1] There can be no harm in stating the simple fact, that a long and intimate acquaintance, extending over many years, with the habits and methods of the scribes of the fourteenth century, has made me almost as familiar with the usual spelling of that period as I am with that of modern English.

It is little more trouble to me to write a passage of Chaucer from dictation than one from Tennyson. It takes me just a little longer, and that is all. In Fragments B and C of the Romaunt, many fifteenth-century spellings have been retained.

[2] See my paper on this subject, printed for the Chaucer Society. Prof. Herford has drawn attention to an unlucky misprint in vol. i. p. 80, where I speak of the pp. of the verb to see as y-seen. Of course I meant y-seyn; see the Glossarial Index. He further remarked, quite correctly, that Chaucer never employs the form seen or y-seen, nor ever rimes it with words in -een. Yet this very form, unknown to Chaucer, occurs thrice in Fragment B, viz. in ll. 3066, 4461, 5571; and in each case it rimes with been. This is a strong hint to those who can appreciate it. A highly characteristic word in Fragment B is dool, in the sense of 'grief'; so also is grete, to weep. But I have no space here to continue the argument. The form sloo, to slay, and other peculiarities suggest that the original dialect of Fragment B was not pure Northumbrian, but Lincolnshire or North East Midland.

[3] For example, l. 4690 of the Romaunt is called l. 4693 in Morris's edition; whilst Book IV of Troilus begins, in the same edition, in the wrong place.

[4] This is the real reason why it was necessary to retain the unauthorised order of the Groups introduced by Dr. Furnivall (see vol. iii. p. 434). To initiate yet another mode of reference would have caused much inconvenience.

[5] The following are some of the more remarkable blunders in the old text. 196. myscoueiting. 274. wo omitted; no sense. 379. er omitted; no sense. 442. ay (for shal). 444. grace (!); for face. 567. Two syllables short. 773. hem omitted. 1007. And for As was; no sense. 1018. wyntred; no such word. 1058. prile for prikke; there is no such word. 1089. durst; for thurte. 1187. sarlynysh (!). 1201. gousfaucoun (!). 1281. And she (!); for Youthe; corrected by Ten Brink. 1313. loreyes; no such word. 1334. Mere nonsense. 1369. Parys (!); for paradise. 1399. it omitted. 1447. garden (!); for yerde in. 1453. goodmesse (!); for good mes (see 3462). 1591. entrees (!); for estres. 1608. laughyng (!); for loving. 2285. Farce; for Fard. 2294. knowith (!); for lauhwith or laughith. 2301. pleyneth; for pleyeth. 2236. londes (!); for Loues. 2650. whider (!); for weder. 3337. cherisaunce; for chevisaunce. 3693-8. Though for Thought; rennyng for rewing; come for to me; the merest nonsense. 4322. wente aboute (!); for wende ha bought; (corrected by Kaluza). 4358. in omitted; no sense. 4366. charge; for change. 4372. MS. yone wole; Th. you wol; for yon wal. 4478. Imperfect. Many more errors, of less consequence, might be added to the list.

[6] Roundel 1 has sustene, kene, grene, quene, sene. In sustēne, the long e is close (Ten Brink, Chaucers Sprache, p. 48); the A.S. words are cēne, grēne, cwēn(e), gesēne, all with close e. Roundel 2 has lēne, bēne, mēne, clēne, all with A.S. ǣ or ēa. Also mēne, of French origin, with open ē; Ten Brink, p. 49.

[7] There is no such word as mena. Critics seem to think that In mena means 'in the middle'; but nothing can be more absurd than to decline a French adjective like a Latin one.

[8] The 'slips' on which the glossaries to these works were written were preserved, and have all been incorporated into the Glossarial Index in the present volume.

[9] Treatise on the Astrolabe; Prologue, l. 43 (vol. iii. p. 176).

[10] I have been courteously provided with proof-sheets from time to time; but my text of Troilus had already been prepared before I was able to make any real use of them.

[11] Chiefly prepared by Miss Gunning and Miss Wilkinson; with liberal additions by Mr. J. H. Hessels, who assisted me in the revision.

[12] The Glossaries to William of Palerne and Havelok were originally prepared by Sir F. Madden, and very well done. We also owe to the same editor a full and satisfactory glossary to Layamon.

[13] In A. 4172, thair occurs, in avowed imitation of the Northern dialect; yet in the line above we find hem instead of them.

[14] For references, see the Glossary.

[15] We even find the double form knittinge, knettinge in Boethius, where there are no rimes to influence the word-form.

[16] Cf. dint of thonder, HF. 534; but, as dint is not a riming word, it may be put for dent.

[17] Hence, in D 51, we should read senne (the Kentish form), to rime with brenne.

[18] Here the standard English thrust is really Southern. We also find thraste, C 260; but this is from A.S. thrǣstan.

[19] I also frequently employ (èè) for open long e; and (éé) for close long e, especially in the Glossary. It is also often usual to employ ę for the open e, and ǫ for the open o. Thus (ae) = (èè) = (ęę); and (ee) = (éé).

[20] It is well known that the mod. E. delight is falsely spelt. The M.E. is delyt (O.F. delit). It rimes with parfyt, appetyt, whyt (see Glossary); never with right or bright.

[21] When the Anglo-French scribes discarded the A.S. symbol æ, they had no certain symbol for the sound (æ) left. Hence, probably, the occasional use of the form thet, to denote the A.S. þæt.

[22] Dr. Sweet gives the sound (ai), as in G. mein. But he adds: 'The distinction between ai and ei, as in day and wey, was probably still kept up in Chaucer's pronunciation, but the two diphthongs were beginning to be confused, probably through the a of ai being modified nearly to the sound of our vowel in man.' However, the rimes prove that Chaucer never distinguishes between them at all; and I believe these diphthongs had been confused much earlier. The Anglo-French scribes could have known but little difference; since ai had already become F. open e in the later text of the Chanson de Roland. Again, Norse only exhibits ei, not ai, so that our raise was M.E. reise, also written raise (Icel. reisa). Very significant is Chaucer's rime of eyse with reyse, D 2101. Nearly everywhere else, the mod. E. 'ease' is spelt ese, eese; and the pronunciation was unquestionably (èè·zə) = (ae·zə), as it rimes with please and appease, words in which even the mod. E. spelling with ea shews that the long e was once open. It follows that reyse was (rei·zə) or even (rèè·zə); certainly not (rai·zə). So again, I should say that the statement that the a of ai was 'modified nearly to the sound of our vowel in man' might have been much more strongly asserted. In such a word as day, from A.S. dæg, the a was already (æ) at the first, and needed no modification at all. It was already spelt dei before A.D. 1200; see Specimens of O. English, ed. Morris, Pt. i. p. 20, l. 79.

[23] In sonne, the n is double; but not in sone.

[24] I use italic y for the consonantal sound of y in ye; because I use (y) for the vowel u in Iuge (jy·gə).

[25] I do not here distinguish between primary and secondary accents. For this distinction, see below (§ 98).

[26] Mod. E. to soar, O.F. essorer, Low Lat. *exaurare; so that the long open o is due to Lat. au.

[27] Store has the o from Lat. au; cf. instaurare. And radevore is from F. ras de Vaur, with o from au; correctly.

[28] I omit dore, door, riming with underspore; perhaps the o was here (u); cf. A.S. duru.

[29] Similarly, in Fragment A of the Romaunt, we find róte riming with swóte, 1025, 1661; and, on the other hand, thròtes riming with harlòtes, nòtes, 191, 507. By way of a glaring contrast, note the rime abood (abòòd) with wood (wóód) in Fragment B of the Romaunt, l. 3159.

[30] Theoretical forms are denoted, in philology, by a prefixed asterisk.

[31] An apparent exception occurs in A.S. cēace, Anglian cēce, M.E. chēke, mod. E. cheek; with unstable ē. Its ēa is unusual, and due to the preceding c. The Du. form kaak shews that its vowel really answers to Germanic ǣ, Goth. ē.

[32] As already noted above; p. xxiv.

[33] Spēre, with close long e, means 'sphere.' It makes all the difference to the sense as well as to the rime.

[34] Whatever test be applied to Fragment B of the Romaunt, the result is always the same, viz. always against its genuineness. Thus it has the rime clène, gréne, 2127; and actually séén, clèèn(!), 2921; clèn-e being always dissyllabic in Chaucer.

[35] Nede once occurs as need, riming with hèèd, head, B. Duch. 1253.

[36] For clear examples of a contrary practice, cf. the rimes gréne, clène, Compl. of the Blk. Knight, 125; Flower and the Leaf, 289; Rom. Rose (B), 2127; wéne, lène, Rom. Rose, 2683.

[37] There once rimes with dere, adj., Legend, 1870. See note 39 below.

[38] Were twice rimes with dere, adj., B. Duch. 773, Clk. Ta., E 882. See note 39 below.

[39] Dere usually has close e (A.S. dēore); but it also rimes with there, were; see notes 37, 38 above, and cf. A.S. dȳre.

[40] Or we may read Mercuri', mury', mari'd, tari'd, beri'd, to-scater'd, contráry', and so on.

[41] MSS. E. Hn. Ln. have Dauit, but it is a childish alteration; of course David is meant. Hl. Cp. Pt. have Dauid.

[42] Better written ones only three lines below; nothing is gained by making words rime to the eye.

[43] The frequent use of o for short u (cf. A.S. duru) by Anglo-French scribes is a source of some trouble to the student.

[44] See vol. i. p. 93, French text, ll. 1-4; p. 94, ll. 19, 33; p. 95, l. 44; &c.

[45] I only cite the pages; all in vol. ii.

[46] All of the alleged exceptions are easily explained by remembering that Gower habitually used Kentish forms. Thus the Kentish for minde is mende; it therefore rimes with ende, wende. The Kentish for pit is pet (still in use), which rimes with let, set. The Kentish for hilles is helles, which rimes with elles. Hid is Kent. hed, riming with fled. Sin is Kent. senne, riming with kenne. Lesseth (Gow. iii. 12) should be lisseth, gives relief; cf. iii. 82, l. 19. It does not appear that Gower is wrong in a single instance.

[47] Correctly printed hedde in Chalmers' British Poets, ii. 67. Pauli's edition is a sad snare.

[48] When writing in French, Gower rimes loisir with obeir; in Balade XXXIV (quoted by Warton).

[49] For is wente read his wente, i.e. his path. This is all that is needed to restore the sense. Wente is a sb., not a pp.

[50] It occurs in no MS. but F.; and the writing in F. (in this passage) is quite late, and of no authority.

[51] Quite 180, in my opinion, if not more; about 4 in every 100 lines. Surely a large percentage.

[52] Chatterton added two lines to Chaucer's stanza, one of the usual length, and the second an Alexandrine. This ten-line stanza occurs in his Battle of Hastings.

[53] Every student of Old French poetry of the fourteenth century must be aware that none of Machault's Balades (in Tarbé's edition) have envoys; and that a large number were written, without envoys, by Froissart, Eustache Deschamps, and Christine de Pisan. Besides, Chaucer introduces a Balade into his Legend of Good Women, which could not have had an Envoy, from the nature of the case; there was no one to address it to.

[57] And yet again, but with repeated rimes, in his Womanly Noblesse; see vol. iv. p. xxv.

[58] The word virelai was taken to mean a lay with a veer or turn in it, owing to a false etymology. The original word was, however, vireli, and the true formula for it was very different. See P. Toynbee, Spec. of Old French, pp. lix. 301. Cf. Ballades, Rondeaus, &c., edited by Gleeson White, London, 1887; p. lxxvi.

[59] The references are, generally, to the Canterbury Tales; A 50 = Group A l. 50.

[60] The forms within parenthesis express the pronunciation, according to the symbols explained above. Cf. Ten Brink, Chaucers Sprache, § 256.

[61] The Glossary has purposely been made very full in order to save references here and elsewhere. Thus ende occurs, finally, in A 15; in the middle of B 481; also in A 197, where the final vowel is slight, but should just be sounded.

[62] Sometimes written -is.

[63] But never peyn for peyne, as in Rom. Rose, 2912, 3184, 3574, 3772, 4323, 4444, 4930; Flower and Leaf, 62.

[64] The prefix y- is not counted as a syllable in this case; y-shette is the same as shette.

[65] The Ellesmere MS. has hise as the plural form; but it is monosyllabic.

[66] In speaking to one person, thou and ye are frequently confounded. Hence in the imperative, the singular and plural forms are frequently confounded also.

[67] See the long list of 183 strong verbs, with an alphabetical index, in Morris's Specimens of English, Part I; Introduction, p. lxix.

[68] But amb, and, ang become omb, ond, ong; hence clomb, &c.

[69] Note the infin. answer-y, short for answer-i-en.

[70] The Glossary (s.v. Ben) gives 'Be, 1 pr. s. am, 3. 588.' This is an oversight; be is here the infinitive = 'to be.'

[71] 'The air that is supplied for the production of the voice-vibrations is capable of being used only in volumes or jets; or, if we attend to the force used in producing them, in pressures.... The law of monopressures, as it may be termed, is a law that operates, and must operate, in the process of articulation. Speech is possible only in monopressures.... One inhalation may suffice for several monopressures. One full breath may suffice, for one who is an expert in husbanding the vocal current, for 30, 60, or even 80 monopressures. Each of these, however, is a vocalised jet of air, condensed and made vocal by a separate effort of the will, just as each note, in a tune rapidly played on the pianoforte, is produced by a special touch, however slight.'—From Accent and Rhythm, explained by the law of Monopressures. Part I. Edinburgh, 1888; an anonymous work, which deserves to be better known.

[72] These symbols are somewhat varied from those employed by the author of 'Accent and Rhythm,' whom I have quoted in the last note (p. lxxxiv.). I owe to him the idea of using them.

[73] See, on this subject, the essay by M. Freudenberger, Ueber das Fehlen des Auftakts in Chaucers heroischem Verse; Erlangen and Leipzig, 1889. I may claim to have been the first to notice this peculiarity, viz. in the Aldine edition of Chaucer, by Dr. Morris, 1866; i. 174.

[74] On the other hand, Lydgate did not shrink from these unmelodious forms. We find form 13 in: 'Up he roos | maugre all' his foon'; Storie of Thebes, 1149; in Spec. of Engl. pt. III. ed. Skeat.

[75] More strictly, as marked in the Ellesmere MS., the caesura really falls earlier, so that 'Ginglen' stands alone; see below, § 107.

[76] Ten Brink quotes many instances of elision, where there is no need for it; thus the -e in wonne (A 59) comes at the caesura, and should be kept.

[77] The e is very light; cf. mod. E. so stately.

[78] Cf. J'rúsalèm, A 463, D 495. Not Jérwsalèm, with w as a consonant, as Ten Brink suggests; such a pronunciation is practically impossible.

[79] The e in párishe is suppressed, by the position of the accent on the a (§ 111); it is not really elided.

[80] It is worth while to place the two stanzas in juxta-position. I accordingly quote them here.

[81] Seneca is often quoted as the author of maxims or proverbial sayings, really found in Publilius Syrus and Caecilius Balbus.

[82] St. Augustine's story found its way into the Gesta Romanorum.