On some Points in Skaldic Metre. § 1. The theory of Old Northern metres proposed by Prof. Sievers has met with so much acceptance from those who have had most opportunity to test its correctness, that it might well seem to be, in its general outlines at least, a completely established fact. Some studies in the stricter forms of skaldic verse have however forced me to the conclusion that in not a few points the system in question, as finally laid down in Altgermanische Metrik, is open to certain strong objections. The discovery of one important point in which skaldic practice seems to be quite opposed to Prof. Sievers' views (see § 2 below) finally supplied the means of testing other assumptions involved in the general theory, and of showing that these are not very well borne out by the facts of the case. What these facts are, and what I believe them to mean, I have stated, as clearly and concisely as possible, in the following pages. Here, however, I wish to say that the time at my disposal being limited, I have not been able to make exhaustive investigations in the whole body of Old Northern verse. Feeling the necessity of having a comparatively small and yet representative field to work in, I selected as the basis for statistics the verses contained in Konrád Gíslason's Udvalg af Oldnordiske Skjaldekvad, as presenting on the whole a safe text and illustrating the practice of many different skalds. The number of lines in this work is about 2000, and in all important cases I have given exact statistics of the types under discussion; a number prefixed to a line indicates the page of Skjaldekvad on which it occurs. But while my statistics are limited to the verses contained in this work, I have gone through the general mass of skaldic verse with sufficient thoroughness to assure myself that it contains nothing materially at variance with the results otherwise obtained, ARKIV FÖR NORDISK FILOLOGI XVI, NY FÖLJD XII. though possibly exhaustive tabulation might give different percentages in some cases. One thing which I have left entirely untouched is the bearing which my views have upon the structure of Old Germanic metre as a whole. The most attractive feature of Prof. Sievers' system is that it brings Old Northern verse into line with Old English and Old Saxon, and supplies them with a common explanation. It must be admitted, however, that skaldic practice is not in favour of this view, nor does there seem to be any imperative reason why there should be this correspondence in detail. We know that dróttkvætt is exclusively Northern; other features may be so as well. To assume that Old Northern metre is found in its primitive form only in some of the Edda poems involves grave difficulties, for it is now well established that these cannot be set down as older than other poems in which the strict skaldic rules are observed. Here, of course, it is possible to reason in several different ways, but incidentally in the following pages I have indicated what seem to me to be strong reasons for reconsidering the whole question as to the standing of these poems. When texts are in the condition in which we have Völuspá and Völundark viđa, to all appearance marked by numerous interpolations and omissions, it is safer not to attach too much importance to their metrical peculiarities. These problems, however, together with many others, such as the structure of ljóðaháttr and málaháttr, do not come within the scope of the present article, which practically resolves itself into an attempt to demonstrate that the skalds observed rules with regard to quantity of syllables and stressing of compounds quite different from what has hitherto been supposed. How far these rules may be ultimately reconcilable with Prof. Sievers' types is a question into which I am not at present prepared to enter. § 2. The clearest evidence that the theory of Old Northern metre presently in favour does not account for certain remarkable features of skaldic verse, is afforded by the lines which Prof. Sievers quotes as undoubted examples of his E-type, that is, in dróttkvætt, xx. These are the following (Altgerm. Metr. p. 104): a Súdvirki lið búðir verdungar, styr gerðu B ógndjarft fyr kné hvarfa ε gunnreifum Aleifi par á hald und Rögnvaldi. I shall not at present insist on the extreme improbability of the metrical structure assumed for some of these lines (especially those of the group 7), though it seems to me almost incredible that a poet who had any ear for verse could have so intended or recited them. There is a much more weighty and material objection than this. It will be observed that in the first four classes given above, the fourth place of the line is occupied by a substantive, and that in seven cases this substantive has a short vowel (lið, styr, skip, etc.), while in the remaining case (kné) it ends in an (originally short) vowel. By Prof. Sievers' theory all these words must be supposed to have the metrical value of; and in fact this is expressly assigned to them in similar positions (see Altg. Metrik p. 58) even when followed by a word beginning with a vowel, e. g. við ský uppi, á aldinn mar orpit, or ok hjörunduð. If we assume, however, that it was the intention of the skald to have, in this place of the line, a substantive metrically equivalent to, it must be admitted that he would most readily have found what he wanted in such words as land, jörð, stein, úlf, etc. Such forms are certainly not less numerous in the language than either of the types, lid, styr or kné, ský. Now, if it proves to be the fact that every dróttkvætt-poet carefully avoided the former types and deliberately selected the latter, it is evident that there must be a good reason for such a choice, and the reason which comes most naturally to hand is that they wanted a short syllable and not a long one. Even if it is maintained that such words as lið, styr etc. are long in this position, the fact remains that the skald was influenced in his choice by considerations for which Prof. Sievers' types supply no explanation. To show that the statement I have made is practically true without exception, I shall deal successively with lines of the four types instanced by Prof. Sievers. It will, however, be necessary at the outset to give up any attempt to separate D- and E-lines; so far as this point is concerned, there is no distinction between them, and it would only lead to needless complications to try to keep them apart throughout the following statistics. The real difference which exists between lines of apparently the same form will be dealt with later on. For the same reason I include in the statistics of the groups 7 and 8 (§§ 5 and 6) lines which would ordinarily be taken as examples of Prof. Sievers' B-type (as par's gollin spjör gullu). As will be seen (§§ 15, 16) this type also seems to require an interpretation altogether different from that hitherto assigned to it. § 3. The a-type, with a trisyllabic word at the beginning of the line, as Súðvirki lið búðir occurs 133 times in the 2000 lines of Skjaldek vad. In 29 cases the word in the fourth place of the line is a substantive, and in every case this has a short vowel, as för, lið, frið, slög, brim, sköp, gram, dag, etc. or (in two cases) ends in a vowel, as skæ, gný. In three cases the -r of the nominative appears (stafr, viðr, sonr) withont affecting the metrical quality of the vowel. In the greater number of the remaining cases the word is a verb, which may have (1) an absolutely short vowel, as gaf, vas, skal, kom etc. also with suffixed 'k = ek as gaf'k, biðk, gat'k, etc.; (2) a short vowel followed by two or more consonants as fell, svall, varð, gekk, etc., also galť'k, felt'k; (3) a final vowel, as brá, dó, frá, so also sák, vá’k; (4) a long vowel followed by one or more consonants, as lætr, hlýtr, bauð, hlaut, also rauð'k, lét'k, etc. It is thus clear that in the case of verbs there was not the same absolute difference in metrical value between short and long syllables as existed in the case of substantives. In the small residue of lines (18) the fourth place is usually occupied by a pronoun, most commonly mik or mér; but sometimes sér, sá, pat, því, etc. are found, and very rarely an adverb as nú, vel. § 4. Lines of the B-type, with dissyllabic word followed by two monosyllables, as greiðendr á skip reiðir, are nearly as numerous as the foregoing class; there are 113 of them in Skjaldekvad. The proportion of substantives is however considerably larger, owing to the fact that this type is, almost without exception, confined to the second line of the couplet (cf. § 14). Here we find 48 nouns, 37 of the form stað, beð, þröm, etc. and 11 ending in a vowel as sæ, gný, bæ, jó, etc. In the remaining lines the proportion of those with a pronoun in the fourth place is very much larger than in the a-class, there being 35 of these. Those which have verbs in the fourth place are thus comparatively fewer, but the types are the same as in the list given in § 3. § 5. In the y-class, the line opens with four monosyllables, as úlfs fót við sker Sóta or the first of these is resolved, as faðir, vas-at, förum, etc. Of this type there are 91 examples in Skjaldekvad, and in 50 of these the |