assumed by Saran and Sievers. See also Martin, Der Versbau des Heliand p. 61 ff. and Schipper, Grundriss p. 60.

§ 37. Criticism of Sievers' System (cont.).

Sievers could not at first by means of his theory explain why the final thesis of his types is always monosyllabic, the inner thesis monosyllabic or disyllabic and the initial thesis generally of two or three syllables. It was not till later that he recognized the real reason by going back to an original verse of the form ××××××××, viz. one 'more' (x) only of the original verse corresponds to the monosyllabic final thesis of his types, two 'moren' (xx) to the inner thesis of one or two syllables, and three 'moren' (xxx) to the initial thesis of two or three syllables.

Sievers' 'senkungen',however, differ not only in the number of their syllables, but also in their stress. Sometimes they are unstressed ('unbetont'), sometimes they have a subsidiary stress ('tieftonig oder nebentonig') [see § 28], and the same subsidiary stresses ('Nebentöne') [root-syllables of the second part of compounds or long middle syllables of three-syllable words, in D2 (D4) even independent monosyllabic strongly stressed words] are at one time (D1, E-D2) counted as mere 'senkungen', and at another time (C-B) as 'hebungen' of full value. And yet for this different treatment of precisely the same stresses no other ground can

be advanced except the rigid holding to two beats only.

The surplus thesis of the 'expanded' D-verses Sievers has rightly derived from his scheme of the original verse, but again he cannot give any reason why this surplus thesis is suddenly able to act as a special 'member' ('Glied') of the verse; why the constant limit of four members can be exceeded just in these verses, which are comparatively rare. His verses of five members ('fünfgliedrige Verse'), therefore, fall out of the frame of his scheme, which is built up on four members.

$ 38. Criticism of Sievers' System (cont.).

But even if, with Sievers, we are willing to look on only those two arses, which are made prominent by the alliteration, as hebungen of full value, and to regard the two others as 'senkungen', yet even with this assumption his combination of the four variously stressed members to form feet is not free from objection, for it partly comes into conflict with the natural division of speech into speechgroups (Sprechtakte), and "it is one of the chief features of German verse that the groups, into which it falls, fall in with the groups of natural speech and begin with the most strongly stressed syllable" (Paul's Grundriss2 II, 2 Metrik p. 50). With this in types A, D and E Sievers' division. into feet agrees, e.g.: lange | hwile, geong in | geardum, land ge-sawon, wis-fæst | wordum

să-līdende, féond | mancynnes, sīde | sā-næssas murnende | mod, weord-myndum | pāh, worold-are for-geaf etc. In types B and C, on the other hand, a simple word is often divided by the commencement of a new foot: purh mine hand. under Heorotes hrof — him sẽ yl|desta, hū bà æde lingas etc. Even where this does not happen, as e.g. him þā Scyld | gewāt in gear-dagum, ofer lagu-\strate etc. the new foot begins with a more weakly stressed syllable.

In Type D2 (D4) again two independent words are unnecessarily pressed into one foot and the second is in stress subordinated to the first, although it is grammatically quite independent and strongly stressed, e.g. blad | wide sprang, grette | Geata leod, atol | ýđa geswing | etc.

Sievers' division of types B, C and D2 (D1), therefore, cannot be looked on as correct. We must divide these types, too, according to the speech-groups of ordinary speech, as is the case in types A, D1, E. Thus type B must be not ×2×2, but ×× þurh | mine | hand, under | Heoroteshrōf, him pa | Scyld gewät, he pas | frōfre gebād etc. Type C not ×× or ×±lú× but: xlz:× Or xl,'x: him sẽ | yldesta |, hū bà | ædelingas, ofer | hron-rāde |, in | gear-dagum etc. Type D2 (D4) not, but 12×12 blæd | wide | sprang, grette | Geata | leod, atol | ýta ge-| swing etc.

In this way Sievers' rising foot of two members

x, which is not in keeping with the rule of Paul quoted above, is settled; and we see that the present middle foot of Types B and D2 (D4) is likewise a falling foot of two members, like the first or second foot of A. We see, too, that the present foot of three members in C is exactly like the foot of three members of D1. Further the frequent 'shortening' in type C becomes easier to understand, for the second hebung in C is in this grouping a 'nebenhebung', which directly follows a full haupthebung. A weaker hebung can be filled by a long syllable or by a short one.

§ 39. Criticism of Sievers' System (conclusion). Lastly we now see that the types D1 and D2 (D4) do not belong so closely together as in Sievers' system. Type D2 is a variety of B, differing from B only in the more strongly stressed beginning of the verse; in the same way D1 is a variety of C. This close relationship between B and D2 on the one hand, and C and D1 on the other, to which Hirt (p. 47) and Fuhr (p. 20 f.) called attention, is best seen when we put side by side verses from Beowulf, which are quite, or almost quite, alike except in the commencement:

In B and C more weakly stressed words stand at the beginning of the verse. These cannot or need not alliterate. In D2 and D1 the verses begin with more strongly stressed words, which must, or at least can, alliterate. Except for this B and D2, C and D1 are quite alike, and the relation of D" to B, and of D1 to C is precisely like that of A1 to A3.

§ 40. The correct Division of Sievers' Types.

We must, therefore, bring Sievers' B and D2 on the one hand, and C and D1 on the other hand into the same scheme, and arrange Sievers' five types, or six, if D1 and D2 (D4) are to count as two, as follows:

I. (2+2) Type A: 2×2× lánge | hwile.

|

a) Type B: ×× þurh | mine | hand. II. (1+2+1){ b) Type D2: 212x12 blæd|wide| spráng. wéord-myndum | páh. a) Type C: xlex him sẽ | ý dèsta,

III. (3+1) Type E:

IV. (1+3)

or: ×× in | geár-dàgum.