(no by-forms with e are noted in NED.), 1485, Rutland Papers (Diehl 141), pris the gelon1) 'gallon', 1492, Records of the Church of St. Mary at Hill, London (Zopf 20), Mexymelyanys (= Maximilian's), 1483, Cely Papers, no. 118, excercary, exercarys ('accessary'), 1451, Paston Letters no. 175 (vol. II, ed. 1904)2).

Considering these safe instances of e for u, I am now inclined to explain several of the previous examples (especially those in the first and second group) as phonetic spelling indicating that had passed into (a) in the pronunciation of some speakers in the 15th cent. An indirect proof of the existence of (a) in the early 16th cent. is afforded by the word sack French (vin) sec. NED. explains it as a hyperliterary form on the analogy of Standard English sack (< *sakk-iz) by the side of seck. It is not very likely that one single word would have called forth a hyperliterary form at this early date when a Standard pronunciation was far from fixed in all its details. The distribution of the early forms with a and e point to a different explanation. To judge from the material in NED. early forms with a occur only in Southern English (London etc.) 3), whereas e is kept in the Northern forms. Hence a for e is probably due to the substitution of southern English (æ) for French (ɛ), which is a sound between English (e) and (a). From the middle of the 16th cent. date such unambiguous phonetic spellings as Cremer (= 'Cranmer'), at stren (= 'in the Strand'), Machyn (noted by Diehl 141). Of a somewhat later date are bedly, peck, (after 1600) in Suffolk Records (Binzel 13). Earlier forms with a for e (nat, kattels, Binzel p. 13) may reflect the dialectal sound development (e)

1) NED. gives no variants with e. On the other hand sellary, celary ‘salary'), ibid., which Zopf likewise explains as a phonetic spelling, may be a phonetic doublet (see NED.), and heving, ibid., may contain a long or be weakly stressed.

2) In these examples (a personal name and a learned Latin word) e cannot very well be due to the interchange between French e and a in pretonic syllables seen in revish ravish etc. In exercary the stress must have been on the first syllable, and er in the second syllable denotes the vocal murmur (ə), which consequently must have occurred both for the unstressed e and the unstressed er at this time.

3) The earliest Northern reference is seake, 1536, the earliest Southern reference is sakkes, 1531.

>(); cf. Engl. Vowels, p. 59. The palatal pronunciation of ǎ is not noted by the Grammarians until the next century (Mauger 1652, and Ben Jonson (?) 1640, Cooper 1685). Cf. Engl. Vowels 121, 188.

It may be said that not much importance can be attached to any of my phonetic spellings as they are all comparatively isolated or sporadic. This argument is, however, anything but convincing. It can be answered with the old fable of the sticks and faggots. A single stick can be broken, but not so the entire bundle. I trust that the whole body of accumulated evidence will resist the attempts of the most captious critics to invalidate it. Moreover the conclusiveness of the spellings is much increased by the fact that in spite of their sporadic occurrence they are all the result of one and the same tendency. When the Continental pronunciation was becoming obsolete and the vowels had already commenced to assume their specifically English sound-values, sporadic attempts were made to denote these new sounds by means of the Continental symbols (with their old sound-values kept). Spellings of this kind were chiefly used by persons who were unaccustomed to wield the pen and therefore followed a natural inclination to write words phonetically, i. e. as they pronounced them. Naturally this tendency is particularly noticeable in the spelling of learned and unfamiliar words.

The following is a selected list of phonetic spellings indicating the different phases of the great vowel-shift. If not otherwise stated, they are all taken from original MSS. of the 15th cent.

(See Selectet List p. 319.)

Ekwall also considers my conclusions from the statements of the French Grammarians too far-reaching. Nevertheless he makes only a few objections to my interpretation of the evidence, and out of these only one seems to be to the point. Ekwall is probably right in assuming that Erondelle's (1606) comparison of the French e feminine with the English phrase is he come refers to e in he (vulgarly pronounced 'a') not to o in come. There is, however, no reason for assuming that u in murderer was ever pronounced with (ü) (as in Swedish 'hus'). E. has not tried to confute the arguments I have

Selected List of Phonetic Spellings (15th cent.) showing the Vowel-Shift.

1) It is doubtful whether any conclusions can be drawn from this spelling which I have found only in two instances from the 15th and three (Saveraigne, Chamley, farniture, English Vowels 81) from the 16th cent. An additional instance is however Samersett, Machyn, p. 182. It is worth noticing, however, that a modern vulgar variant of (^) is sometimes rendered with a, as in jadge (= judge'), nathing (= 'nothing') etc. (Shaw, Captain Brassbound's Conversion, p. 217).

2) This spelling, which is not discussed in my English Vowels, is noted without comments by Sussbier from the Cely Papers, no. 28. Similar forms indicating the change of qu> iu are 6—7 lude, 6 due (= 'dew3), 6 butyful, NED. Hence rhymes such as few: true, beauty: duty in Shak. and Spenser need not be incorrect (thus Horn Hist. Gramm. p. 106). After (s) ğu seems to have passed into in or at a still earlier date, to judge by such phonetic spellings of 'show', 'showed' as 3-4 schued, 5 shued, 6 shuyd, 6 shue (= show sb.), NED., shuyd Paston Letters no. 465 (wrongly explained as a scribal error in E. Vs. 84).

adduced in favour of my assumption that ME. й was pronounced in early NE. with a sound approximating to the modern (^). Cf. Engl. Vowels, pp. 72, 81, 106 (Festeau), 133 f., Anglia 1914, p. 423 f., Anglia Beiblatt 1917, p. 78, and below.

That the identification of the English (in 'come' etc.) with the French and German by the Grammarians should not be interpreted literally and that consequently it does not indicate a pronunciation (o) seems probable on the following grounds.

(1) The acoustic resemblance between the English (^) (in 'come') and the French (2) (in 'comme') which even now makes (^) sound as an o to French ears (see Sweet, HES., 798) and which is occasionally hinted at by authors of early grammars. Smith (1795) states that the English sound in 'come' etc. has no French or German correspondent. The Parisian pronunciation of 'sotte' and the German pronunciation of 'hott' (cry to horses) are not unlike 'but still not near enough the u' (Stichel 29). In the modern comic literature Frenchmen are sometimes represented as pronouncing (2) for (A)').

(2) The German and Scandinavian grammars that identify the English in 'come' with the German and the Scandinavian Ŏ, have evidently copied this rule from French sources. This is the case with König (1706) (see Driedger, 20), the first German who gives a comparatively independent account of the English pronunciation. (Tellæus' Grammatica Anglicana 1665, is a mere translation of Alphabet Anglois 1625, see Engl. Vowels 96.) On the early Swedish Grammars see Gabrielson in 'Studier i Modern Språkvetenskap' VI, Uppsala 1917.

(3) Several Grammarians identify the English with the French or German but write in their transcriptions not only o for English but also eu and ö, which points to (^). Such Grammarians are Bellot 1580, Mason 16222), Arnold 1718 (W. Müller, 46), Pell 1735, Berry 1766 (Stichel, 29) and Wagner 1789 (Löwisch, 74).

In specimens of English-German (mixed language) we still

1) In Anstey's Lyre and Lancet (1906) a French chef says moch (p. 122) for 'much', toch (p. 134) for 'touch', and cot for 'cut' (ibid.).

2) Bellot and Mason write eu for u only before r (in 'church' and ‘shirt'), where it may be due to combinative sound-change.

find and written for English (A), e. g. sopper, Botterflei, Kostimersh (= 'customer'), Diffikölti, gestödied etc. (see Dunger, Engländerei, 77 f.). Lastly it is noteworty that Flint (1740) while identifying o in 'come' etc. with French o, nevertheless adds that it is sounded as French eu:

“L'u bref Anglois approche beaucoup de l'o bref François, mais il a un son plus obscur, dit le même auteur que je viens de citer (Wallis), c'est le son d'eu bref dans le mot Serviteur prononcé négligemment. Vous aurez le son d'u bref Anglois, si vous prononcez l'o François extremement bref & serré, & vous verrez qu'en le prononçant ainsi, vous donnez presque le son de l'e feminin François c'est à dire, d'eu bref."

Flint, Prononciation de la Langue Angloise (1740), p. 7.

(4) Comparisons between the English й with Ŏ short (or long) occur at a date when there can be no doubt that й had its present pronunciation:

"u in tub, cup

eu in French neuf, dem o ähnlich in ob oder der interjektion topp", Vollständige englische sprachlehre von J. G. Flügel, 1824.

"kann durch ein völlig entsprechendes deutsches schriftzeichen nicht wiedergegeben werden. Der laut liegt zwischen dem und dem deutschen ö in Köpfe, Lucas, Engl.-deutsches wörterbuch, 1854.

Walker (Principles § 165) says that 'the long sound which seems the nearest relation to u (in but) is o in note, tone.

In his Historische neuengl. laut- und formenlehre Ekwall points out that Wilkins (1668) and Cooper (1685) look upon u in cut and the unstressed e in better as identical from a qualitative point of view. This identification need not be exact, but it certainly points to (a) rather than to (0). Ekwall omits to mention Wallis' identification of u in cut with (c) in serviteur, which likewise speaks in favour of (^) 1).

In support of the theory that early NE. й was pronounced (0) Ekwall, ibid., p. 54, suggests that (^) in 'none' and 'nothing' may have originated by the shortening of (0:) to (0), which afterwards passed into (^). This is a mere conjecture which is not supported by the evidence.

When and qu were shortened the result was of course (5) as in 'knowledge', 'blossom', 'fodder', 'nonce' etc. In point of fact (5) is evidenced in 'nothing' by Hart (1569), Gill (1621), Peyton (1756) (Stichel, 43) and Siret Parquet (1769)

1) To anyone who has heard the sound (o) in 'come' etc. pronounced by provincial speakers it is difficult to believe that this sound could have been compared either with the vocal murmur (ə) or French (œ) (œ:).

J. Hoops, Englische Studien. 52. 3.

21