are two accented words in the first and one in the second half of the verse. Only occasionally do we meet with verses containing merely one alliterative word in each half-verse, or with so-called crossed alliteration" (cf. Lawrence, Chapters on Allit. Verse, p. 77).

Metrically the Morte Arthure is less correct than the Troy-Book, whereas the deviations from the identity of stressed and alliterative words are most frequent in Piers the Plowman. These facts are generally known, and will be confirmed and illustrated by the following investigation, which will, therefore, at the same time present a contribution, although a modest one, to the knowledge of the metrical composition of Middle English alliterative verse.

Chapter I.

We possess various means for ascertaining the word-stress. in Middle English: 1. The language of the poets: rhythm, alliteration, rhyme. 2. Certain changes in the language: weakenings of sounds, syncope etc. 3. Conclusions drawn from Modern English, both from the accentuation of the present day, and from direct evidence of an earlier date, e. g. from the Manipulus Vocabulorum belonging to the sixteenth century (cf. Morsbach, Me. Gramm. § 18).

Among those different tests, we shall principally use that of alliteration, for, as Schipper (Grundriss II p. 1038) puts it: The supreme law for the connection between word-stress and metrical stress requires, in all verse based upon the principle of accent, that the latter should be in agreement with the former. This applies in an equal measure to the alliterative line and to „, equipedal“ verse (,, gleichtaktige Versarten“)“. Where in our texts this agreement is not found to exist, we shall have to decide by means of one or other of the criteria mentioned above, whether the accentuation in question can be justified or not.

We divide the material collected in the following pages into two principal groups, a Germanic or English and a Romance group. In the further subdivision of the former we follow the one adopted by Morsbach in his Middle English Grammar.

For reasons already referred to in the Introduction we base our conclusions in the cases that will come under discussion, in the first place on the Troy-Book.

A. The Germanic or English Element.')

I. Original Nominal Compounds and their Analogues.

árowsmythis, 1588.

belmakers, 1589.

belt stid, 5940.

bladsmythis, 1592.

a) In the Troy-Book:

bódword, bódeword, 6262, 8315.

burgh-men, 8570.

éuensangtyme, 8919.

góldsmythes, 1584.

hérne-pon, 8775.

horse fete, hórsfet, 5834, 6560.

léfs-ales, léfe-sals, 337, 1167. (Cf. Chaucer, Reves T.: levesel). nightwácche, 7352:

Nightwacche for to wake, waites to blow,

But also skóute-wacche, skówte wacche, 1089, 6042. sópertyme, 3398.

forward (= agreement), 548, 602, 636, 651, 704, 2440, 2727, 3123, 7985, 9312; (= vanguard) 1148, 5860.

The first part of the word has the chief stress, quite regularly. We find this accentuation already in OE. and likewise still in the Modern Dutch voorwaarde.

fórwise, 2539, 3950.

afterwarde, 121.

éftsones, 2478, 7245, 11518. (Cf. p. 16.)

áuerthwert, óuerthwert, 7532, 8348.

These also have the regular stress.

wanspede has the stress on the first syllable in v. 9327; in v. 7945 on the second:

My wonsped to aspie in dispite ay.

If the rime-letters are here placed correctly in the first half verse, we must assume that the accentuation of this word was

1) In this division we also place words of Romance origin, when they are provided with a Germanic prefix, and formations like sopertyme.

a shifting one. In OE. the syllable wan- (won-) was stressed, as it is still in Modern Dutch, e. g. wanhoop, despair

mishap, 2069, 13133.

misrewle, mýsrewle, 6128, 7952.

In OE, in such compounds, the prefix mis- was regularly stressed. In ME. the accentuation varies (cf. Morsbach, § 242, and Anm. c.). In our texts we have discovered only one example of unstressed mis-: myserúle, Rich. the Redeles, Pass. 4,3. In the Manip. Vocab. we find misdéede 52, 33, but mishappe 27, 27, and misrewle 95, 44. In Mod. E. this prefix is unstressed, as in mistake, mishap, or has a weak stress, as in misdeed. Sweet (A New Engl. Gram. § 919) says: "Some prefixes which have a very definite meaning and are phonetically capable of being detached from the body of a word have in consequence come to be felt as independent words, the prefix and the body of the word being balanced against one another, as it were, by each receiving equal stress", and quotes among his examples the word mis conduct in which the dots indicate his accentuation. To me it seems more than doubtful that both parts of the word receive equal stress", although certainly the prefix is not altogether unstressed.

Compounds with the negative particle un-, which in OE. had still mostly the chief stress on the first part, shift their accent in ME. In Mod. E. this prefix is usually unaccented, or has a secondary stress, as in unbelief (cf. Sweet, NE. Gram. § 919, and Morsbach § 242 and Anm. a).

vnbest (monster), 7766.

vnkýndness, 144, 1923.

vnpóssible, 258.

unstithe, 117.

vncléne, 1639, 1845.

unfaithful, 714.

vntrúly, 723.

b) in the Morte Arthure.

bule-fyre, 1048.

blód-hondes, 3640.

cáremane, 957. (Cf. Oxf. Dict.: carman.)

cópe-borde, 206. (Cf. Oxf. Dict.: cupboard.)

crosse-dayes, 3212.

dede-thráwe, 1150:

The theeffe at the dede-thrawe so throly tyme thryngez Cf. Gamelyn 24:

On his deep bed to a-bide Goddes wille.

Here, perhaps, a shifting of accent took place (cf. Morsb. §§ 26, 27). In OE. the first part of the word had of course the chief stress (cf. Beowulf 2901)

dule-cotes, 4336.

éuensange, éuesange, 894, 900.

eye-liddes, 3953:

Lokes one his eye-liddis, that lowkkide ware faire.

Probably here also a shifting of accent may be assumed.

fáa-mene, 303.

fóte-mene, 1989.

hannde-brede, 2229.

hánsemane, hánsemene, 2662, 2743. (In the Gloss. Index explained as henchman, page. Cf. Skeat, Etym. Dict.: henchman.) hérne-pane, 2229.

kételle-hattes, kéttille-hatte, 2993, 3516, 3995.

mórne-while, 2001, 3223.

neke-bóne, 2771:

And brustene his neke-bone, that alle his breste stoppede! Here also, perhaps, a shifting of accent. Cf. Burns's Tum o'Shanter:

Whare drunken Charlie brak's neck-bane. schaft-monde, 2546, 3843, 4232.

schippemene, 1212.

schirreues, 725.

schýnbawde, 3846.

tóppe-castelles, 3616.

watyre-mene, 741.

wolfe-heuede, 1093.

fórchipe, 3678.

fóretoppe, 1078.

fórheuede, 1080.

forestayne, 742.

fórtethe, 1089, but