would be symbolically represented as follows:

–́×̀×|–́×̀ r ||–́ rr |–́×̀×.

According to this system the pause at secg will be twice as long as that at sōhte, whilst at wudu there will be no real pause and the point will merely indicate the end of the measure.

Others reverted to the view of Bartsch and Schubert that there could be hemistichs with only three accents alongside of the hemistichs with the normal number of four. Among these may be mentioned H. Hirt,[28] whose view is that three beats to a hemistich is the normal number, four being less usual, the long line having thus mostly six beats, against the eight of Lachmann’s theory; K. Fuhr,[29] who holds that every hemistich, whether it stands first or second in the verse, has four beats if the last syllable is unaccented (klingend; in that case the final unaccented syllable receives a secondary rhythmical accent, for example, fḗond máncýnnès) and has three beats if it is accented (stumpf, for example, fýrst fórð gewā́t, or múrnénde mṓd, &c.); and B. ten Brink,[30] who calls the hemistichs with four beats full or ‘complete’ (e.g. hȳ́ràn scóldè, but admits hemistichs with three beats only, calling them ‘incomplete’ from the want of a secondary accent (e.g. twélf wíntra tī́d, hā́m gesṓhte, &c.). The four-beat theory was reverted to by M. Kaluza, who endeavours to reconcile it with the results of Sievers and others.[31] A somewhat similar view is taken by R. Kögel.[32] Trautmann[33] takes Amelung’s view that certain words and syllables must be lengthened in order to get the four accented syllables necessary for each hemistich. Thus, according to Trautmann’s scansion,

would run ×́×|×́×|×́×|⏑́× and

would also have the formula ×́×|×́×|×́×|⏑́×,

ond being protracted to two units. Another instance of this lengthening would, on this theory, occur in the final syllable of the word radores in the hemistich únder rádorès rýne, while in a section like gūð-rinc monig, or of fold-grǽfe, the words rinc and of would be extended to two, and gūð and fold would each be extended to four units, in order to fit in with the scansion ×́×|×́×|×́×|⏑́×. Most of the partisans of the four-beat theory for the hemistich agree in making two of these beats primary, and two secondary; Trautmann, however, does not seem to recognize any such difference in the force of the four accents. All the supporters of the four-beat theory have this in common, that the rhythm of the verse is assumed to be based on time (taktierend), but in other respects differ widely from each other; Hirt, for example, in his last discussion of the subject,[34] claiming that his own view is fundamentally different from that of Kaluza, which again he looks on as at variance with those of Möller and Heusler.

§ 17. The two-beat theory, on the other hand, is that each of the two hemistichs of the alliterative line need have only two accented syllables. In England this view was taken by two sixteenth-century writers on verse, George Gascoigne[35] who quotes the line,

giving as the accentual scheme ` ´ ` ` ´ ` ´ ` ` ´; and by King James VI, whose example is—

In 1765, Percy, in his Essay on Pierce Plowman’s Visions, pointed out ‘that the author of this poem will not be found to have invented any new mode of versification, as some have supposed, but only to have retained that of the old Saxon and Gothick poets, which was probably never wholly laid aside, but occasionally used at different intervals’. After quoting[37] two Old Norse, he gives two Old English verses:—

he continues: ‘Now if we examine the versification of Pierce Plowman’s Visions’ (from which he quotes the beginning—

‘we shall find it constructed exactly by these rules’, which are, in his own words, ‘that every distich [i.e. complete long line] should contain at least three words beginning with the same letter or sound; two of these correspondent sounds might be placed either in the first or second line of the distich, and one in the other, but all three were not regularly to be crowded into one line.’ He then goes on to quote further specimens of alliterative verse from Pierce the Ploughman’s Crede, The Sege of Jerusalem, The Chevalere Assigne, Death and Liffe and Scottish Fielde, which latter ends with a rhyming couplet:

Taken as a whole his dissertation on the history of alliterative verse is remarkably correct, and his final remarks are noteworthy:

Now it is clear that this verse is of exactly the same structure as the verses quoted by Gascoigne:

where the scheme of accents is Gascoigne’s own, showing that he read them as verses of four accents in all, two in each hemistich. They show the same rhythmical structure as the ‘tumbling’ or alliterative line given by James VI[39] (1585):

and described by him as having ‘twa [feit, i.e. syllables] short, and ane lang throuch all the lyne’, in other words with four accented syllables in the verse.

Percy detected very acutely that the Middle English alliterative line stood in close connexion with the Old English alliterative line, and suggested as highly probable that the metre of Pierce Plowman would give a key to the rhythm of that older form of verse, which would have to be read with two accented syllables in the hemistich, and therefore four in the whole line.

Had this essay of Percy’s been known to Lachmann’s followers, many of the forced attempts at reconciling the Old English verse with a scheme that involved a fixed number of syllables in the line would not have been made. Lachmann himself, it must be remembered, admitted the two-beat scansion for Old Norse, Old Saxon, and Old English. Meanwhile other investigators were at work on independent lines. In 1844 A. Schmeller, the editor of the Heliand, formulated the law that, in the Teutonic languages, it is the force with which the different syllables are uttered that regulates the rhythm of the verse, and not the number or length of the syllables (which are of minor importance), and established the fact that this alliterative verse was not meant to be sung but to be recited.[40] He does not enter into the details of the rhythm of the verse, except by pointing out the two-beat cadence of each section. Somewhat later, W. Wackernagel[41] declared himself in favour of the two-beat theory for all Teutonic alliterative verse. In every hemistich of the verse there are according to Wackernagel two syllables with a grammatical or logical emphasis, and consequently a strong accent, the number of less strongly accented syllables not being fixed. The two-beat theory was again ably supported by F. Vetter[42] and by K. Hildebrand, who approached the subject by a study of the Old Norse alliterative verse,[43] and by M. Rieger in his instructive essay on Old Saxon and Old English versification.[44] In this essay Rieger pointed out the rules prevailing in the poetry of those two closely related Teutonic nations, dealt with the distribution and quality of the alliteration, the relation of the alliteration to the noun, adjective, and verb, and to the order of words, with the caesura and the close of the verse, and, finally, with the question of the accented syllables and the limits of the use of unaccented syllables.[45] Other scholars, as Horn, Ries, and Sievers, contributed further elucidations of the details of this metre on the basis of Rieger’s researches.[46]

Next to Rieger’s short essay the most important contribution made to the accurate and scientific study of alliterative verse was that made by Sievers in his article on the rhythm of the Germanic alliterative verse.[47] In this he shows, to use his own words, ‘that a statistical classification of groups of words with their natural accentuation in both sections of the alliterative line makes it clear that this metre, in spite of its variety, is not so irregular as to the unaccented syllables at the beginning or in the middle of the verse as has been commonly thought, but that it has a range of a limited number of definite forms which may be all reduced to five primary types.’ These five types or chief variations in the relative position of the accented and unaccented syllables are, as Sievers points out, of such a nature and so arbitrarily combined in the verse, that they cannot possibly be regarded as symmetrical feet of a line evenly measured and counted by the number of syllables. ‘The fundamental principle, therefore, of the structure of the alliterative line, as we find it in historical times, is that of a free change of rhythm which can only be understood if the verse was meant to be recited, not to be sung.’[48] Soon after the publication of Sievers’s essay on the rhythm of the Germanic verse, the first part of which contained a complete classification of all the forms of the line occurring in Beowulf, other scholars applied his method and confirmed his results by examining in detail the other important Old English texts; Luick dealt with Judith,[49] Frucht with the poems of Cynewulf,[50] and Cremer with Andreas, &c.[51] Sievers himself, after contributing to the pages of Paul’s Grundriss der germanischen Philologie a concise account of his theories and results, expounded them in greater detail in his work on Old Germanic Metre[52] in which he emphasizes the fact that his five-type theory cannot properly be called a theory at all, but is simply an expression of the rules of the alliterative verse obtained by a statistical method of observation. In spite of the criticisms of his opponents, Möller, Heusler, Hirt, Fuhr, and others, he maintained his former views. In principle these views are in conformity with the manner of reading or scanning the alliterative verse explained by English writers on the subject from the sixteenth century downwards, though their terminology naturally is not the same as Sievers’s. We may, therefore, accept them on the whole as sound.

It would be out of place here to enter into the question of prehistoric forms of Teutonic poetry. It will be enough to say that in Sievers’s opinion a primitive form of this poetry was composed in strophes or stanzas, intended to be sung and not merely to be recited; that at a very early period this sung strophic poetry gave way to a recited stichic form suitable to epic narrations; and that his five-type forms are the result of this development. As all the attempts to show that certain Old English poems were originally composed in strophic form[53] have proved failures, we may confidently assent to Sievers’s conclusion that the alliterative lines (as a rule) followed one upon another in unbroken succession, and that in historic times they were not composed in even and symmetrical measures (taktierend), and were not meant to be sung to fixed tunes.

The impossibility of assuming such symmetrical measures for the Old English poetry is evident from the mere fact that the end of the line does not as a rule coincide with the end of the sentence, as would certainly be the case had the lines been arranged in staves or stanzas meant for singing. The structure of the alliterative line obeys only the requirements of free recitation and is built up of two hemistichs which have a rhythmical likeness to one another resulting from the presence in each of two accented syllables, but which need not have, and as a matter of fact very rarely have, complete identity of rhythm, because the number and situation of the unaccented syllables may vary greatly in the two sections.

§ 18. Accentuation of Old English. As the versification of Old English is based on the natural accentuation of the language, it will be necessary to state the laws of this accentuation before giving an account of the five types to which the structure of the hemistich has been reduced.

In the alliterative line, as a general rule, only syllables with the chief accent carry either the alliterating sounds or the four rhythmical accents of the verse. All other syllables, even those with secondary accent, count ordinarily as the ‘theses’ (Senkungen) of the verse[54]:

In compound words (certain combinations with unaccented prefixes excepted) the first element of the compound (which modifies or determines the meaning of the second element) has the primary accent, the second element having only a secondary accent, e.g. wúldor-cỳning, hḗah-sètl, sṓð-fæ̀st.[55] If therefore the compound has, as is mostly the case, only one alliterative sound, that alliteration must necessarily fall on the first part of the compound:

Sometimes it happens that in hemistichs of no great length the second part of the compound carries one of the two rhythmical accents of the hemistich, e.g.

and in a particular form of alliteration[56] it may even contain one of the alliterating sounds, as in the verse:

The less strongly accented derivational and inflexional suffixes, though they are not allowed to alliterate, may occasionally have the rhythmical accent, on condition that they immediately follow upon a long accented syllable, e.g.

The rhythmical value of syllables with a secondary accent will be considered more fully later on.

These general rules for the accent of compound words formed of noun + noun or adjective + noun require modification for the cases where a prefix (adverb or preposition) stands in close juxtaposition with a verb or noun. The preposition standing before and depending on a noun coalesces so closely with it that the two words express a single notion, the noun having the chief accent, e.g. onwég, āwég (away), ætsómne (together), ofdū́ne (down), toníhte (to-night), onmíddum (amid); examples in verse are:

But while the prepositional prefix thus does not carry the alliteration owing to its want of accent, some of the adverbs used in composition are accented, others are unaccented, and others again may be treated either way. When the adverbial prefix originally stood by itself side by side with the verb, and may in certain cases still be disjoined from it, it has then the primary accent, because it is felt as a modifying element of the compound. When, however, the prefix and the verb have become so intimately united as to express one single notion, the verb takes the accent and the prefix is treated as proclitic, and there is a third class of these compounds which are used indifferently with accent on the prefix or on the verb.

Some of the commonest prefixes used in alliteration are[57]: and, æfter, eft, ed, fore, forð, from, hider, in, hin, mid, mis, niðer, ongēan, or, up, ūt, efne, as in compounds like ándswarian, íngong, ǽfterweard, &c.:

Prefixes which do not take the alliteration are: ā, ge, for, geond, oð, e.g.

The following fluctuate: æt, an, bī (big), bi (be), of, ofer, on, tō, under, þurh, wið, wiðer, ymb. These are generally accented and alliterate, if compounded with substantives or adjectives, but are not accented and do not alliterate if compounded with verbs or other particles,[58] e.g. óferhēah, óferhȳd, but ofercúman, oferbī́dan. The following lines will illustrate this:

(a) prefixes which alliterate:

(b) prefixes which do not alliterate:

When prepositions precede other prepositions or adverbs in composition, the accent rests on that part of the whole compound which is felt to be the most important. Such compounds fall into three classes: (i) if a preposition or adverb is preceded by the prepositions be, on, tō, þurh, wið, these latter are not accented, since they only slightly modify the sense of the following adverb. Compounds of this kind are: beǣ́ftan, befóran, begéondan, behíndan, beínnan, benéoðan, búfan, bútan, onúfan, onúppan, tōfóran, wiðínnan, wiðū́tan, undernéoðan.[60] Only the second part of the compound is allowed to alliterate in these words:

Most of these words do not seem to occur in the poetry.

(ii) In compounds of þǣr + preposition the preposition is accented and takes the alliteration:

(iii) weard, as in æfterweard, foreweard, hindanweard, niðerweard, ufeweard, &c., is not accented:

§ 19. The secondary accent. The secondary or subordinate accent is of as great importance as the chief or primary accent in determining the rhythmical character of the alliterative line. It is found in the following classes of words:

(v) After a long root-syllable of a trisyllabic word a short second syllable (whether its vowel was originally short or long) may bear one of the chief accents of the line, e.g. bōcère, bíscòpe:

or may stand in the thesis and be unaccented, as

This shows that in common speech these syllables had only a slight secondary accent.

(vi) Final syllables (whether long or short) are as a rule not accented even though a long root-syllable precede them.

§ 20. Division and metrical value of syllables. Some other points must be noticed with reference to the division and metrical value of the syllables of some classes of words.

The formative element i in the present stem of the second class of weak verbs always counts as a syllable when it follows a long root-syllable, thus fund-i-an, fund-i-ende not fund-yan, &c. In verbs with a short root-syllable it is metrically indifferent whether this i is treated as forming a syllable by itself or coalescing as a consonant with the following vowel, so that we may divide either ner-i-an, or ner-yan; in verbs of the first and third class the consonantal pronunciation was according to Sievers probably the usual one, hence neryan (nerian), lifyan (lifᵹan), but for verbs of the second class the syllable remained vocalic, thus þolian.[61]

In foreign names like Assyria, Eusebius, the i is generally treated as a vowel, but in longer words possibly as a consonant, as Macedonya (Macedonia). As to the epenthetic vowels developed from a w, the question whether we are to pronounce gearowe or gearwe, bealowes or bealwes cannot be decided by metre. Syllabic l,m, n (l̥, m̥, n̥) following a short root-vowel lose their syllabic character, thus sĕtl, hræ̆gl, swĕfn are monosyllables, but er coming from original r as in wæter, leger may be either consonantal or vocalic. After a long root-syllable vocalic pronunciation is the rule, but occasionally words of this kind, as túngl, bṓsm, tā́cn, are used as monosyllables, and the l, m, and n are consonants. Hiatus is allowed; but in many cases elision of an unaccented syllable takes place, though no fixed rule can be laid down owing to the fluctuating number of unaccented syllables permissible in the hemistich or whole line. In some cases the metre requires us to expunge vowels which have crept into the texts by the carelessness of copyists, e.g. we must write ḗðles instead of ḗðeles, éngles instead of éngeles, dḗofles instead of dḗofeles, and in other cases we must restore the older and fuller forms such as ṓðerra for ṓðrā, eṓwere for ḗowre.[62] The resolution of long syllables with the chief accent in the arsis, and of long syllables with the secondary accent in the thesis, affects very greatly the number of syllables in the line. Instead of the one long syllable which as a rule bears one of the four chief accents of the verse, we not unfrequently meet with a short accented syllable plus an unaccented syllable either long or short (⏑́×́). This is what is termed the resolution of an accented syllable. A word accordingly like fároðe with one short accented syllable and two unaccented syllables has the same rhythmical value as fṓron with one long accented and one unaccented syllable, or a combination like se þe wæs is rhythmically equivalent to sécg wæs.

§ 21. We now come to the structure of the whole alliterative line. The regular alliterative line or verse is made up of two hemistichs or sections. These two sections are separated from each other by a pause or break, but united by means of alliteration so that they form a rhythmical unity. Each hemistich must have two syllables which predominate over the rest in virtue of their logical and syntactical importance and have on this account a stronger stress. These stressed syllables, four in number for the whole line, count as the rhythmical accents of the verse. The force given to these accented syllables is more marked when they carry at the same time the alliteration, which happens at least once in each hemistich, frequently twice in the first and once in the second hemistich, and in a number of instances twice in both hemistichs. The effect of the emphasis given to these four words or syllables by the syntax, etymology, rhythm, and sometimes alliteration, is that the other words and syllables may for metrical purposes be looked upon as in comparison unaccented, even though they may have a main or secondary word-accent.

In certain cases, in consequence of the particular structure of the hemistich, there is found a rhythmical secondary accent, generally coinciding with an etymological secondary accent, or with a monosyllable, or with the root-syllable of a disyllabic word. Sievers looks on these syllables as having in the rhythm of the verse the nature of a minor arsis (Nebenhebung); they rather belong to the class of syllables standing in thesis but with a slight degree of accent (tieftonige Senkung).

The two sections of the alliterative line rarely exhibit a strict symmetry as to the number of the unaccented syllables and their position with regard to the accented syllables. In the great majority of cases their similarity consists merely in their having each two accented syllables, their divergence in other respects being very considerable. It is to be noted that certain combinations of accented and unaccented syllables occur with more frequency in one hemistich than in the other, or are even limited to one of the two hemistichs only.

Besides the ordinary or normal alliterative line with four accents, there exists in Old English and in other West-Germanic poetry a variety of the alliterative line called the lengthened line (Schwellvers or Streckvers). In this line each hemistich has three accented syllables, the unaccented syllables standing in the same relation to the accented ones as they do in the normal two-beat hemistich.

§ 22. The structure of the hemistich in the normal alliterative line. The normal hemistich consists of four, seldom of five members[63] (Glieder), two of which are strongly accented (arses), the others unaccented or less strongly accented (theses). Each arsis is formed, as a rule, of a long accented syllable (–́), but the second part of a compound, and (less frequently) the second syllable with a secondary accent of a trisyllabic or disyllabic word, is allowed to stand as an arsis. By resolution a long accented syllable may be replaced by two short syllables, the first of which is accented. This is denoted by the symbol ⏑́×. The less strongly accented members of the hemistich fall into two classes according as they are unaccented or have the secondary accent. This division depends ultimately on the logical or etymological importance of the syllables. Unaccented syllables (marked in Sievers’s notation by ×) whether long or short by etymology, are mostly inflexional endings, formative elements, or proclitic and enclitic words.

Secondarily accented verse-members, mostly monosyllabic and long (denoted by ×̀, and occasionally, when short, by ⏑̀), are root-syllables in the second part of compounds, long second syllables of trisyllabic words whose root-syllable is long, and other syllables where in ordinary speech the presence of a secondary accent is unmistakable. The rhythmical value of these syllables with secondary accent is not always the same. When they stand in a foot or measure of two members and are preceded by an accented syllable they count as simply unaccented, and the foot is practically identical with the normal type represented by the notation –́× (as in the hemistich wī́sra wórda), but these half-accented syllables may be called heavy theses, and the feet which contain them may be denoted by the formula –́×̀, as in wí̄sfæ̀st wórdum (–́×̀|–́×). A hemistich like the last is called by Sievers strengthened (gesteigert), or if it has two heavy unaccented syllables in both feet, doubly strengthened, as in the section gū́ðrìnc góldwlànc (–́×̀|–́×̀). In these examples the occurrence of a heavy unaccented syllable is permissible but not necessary; but in feet or measures of three members they are obligatory, being required as an intermediate degree between the arsis and thesis, or strongly accented and unaccented member, as in þégn Hrṓðgā́res (–́|–́××̀), or fýrst fórð gewà̄t (–́|–́××̀), or hḗalǣ̀rna mǣ́st (–́×̀×|–́). In these cases Sievers gives the verse-member with this secondary accent the character of a subordinate arsis, or beat (Nebenhebung). But it is better, in view of the strongly marked two-beat swing of the hemistich, to look on such members with a secondary accent as having only the rhythmical value of unaccented syllables, and to call them theses with a slight accent. The two-beat rhythm of the hemistich is its main characteristic, for though the two beats are not always of exactly equal force[64] they are always prominently distinguished from the unaccented members of the hemistich, the rhythm of which would be marred by the introduction of an additional beat however slightly marked.

Cases in which the two chief beats of the hemistich are not of exactly the same force occur when two accented syllables, either both with chief accent or one with chief and the other with secondary accent, stand in immediate juxtaposition, not separated by an unaccented syllable. The second of these two accented syllables may be a short syllable with chief accent, instead of a long syllable as is the rule. But in either case, whether long or short, this second beat following at once on the first beat is usually uttered with somewhat less force than the first, as can be seen from examples like gebū́n hǽfdon, Beow. 117; tō hā́m fáran, 121; mid ǣ́rdǽge, 126. The second beat rarely predominates over the first. The cause of this variation in the force of the two beats is to be sought in the laws of the syntactical accent.

In other respects verse-members with a secondary accent obey the same laws as those with a primary accent. They usually consist of one long syllable, but if a member which has the arsis immediately precedes, a short syllable with a secondary accent may be substituted. Resolution of such verse members is rare, which shows that they are more closely related to the thesis than to the arsis of the hemistich. One unaccented syllable is sufficient to form the thesis (×), but the thesis may also have two or more unaccented syllables (××,×××..), their number increasing in proportion to their shortness and the ease with which they can be pronounced, provided always that no secondary accent intervenes. All of these unaccented syllables are reckoned together as one thesis, as against the accented syllable or arsis. The single components of such a longer thesis may exhibit a certain gradation of force when compared with one another, but this degree of force must never equal the force with which the arsis is pronounced, though we sometimes find that, owing to the varying character of the syntactical or sentence accent, a monosyllable which in one case stands in the thesis, may in another connexion bear the secondary or even the primary accent.

§ 23. The number of the unaccented syllables of the thesis was formerly believed to depend entirely on the choice of the individual poet.[65] Sievers first put this matter in its right light by the statistics of the metre.[66] He showed that the hemistich of the Old English alliterative line is similar to the Old Norse four-syllable verse, and is as a rule of a trochaic rhythm (–́×–́×). The proof of this is that in Beowulf, for instance, there are 592 hemistichs of the type –́×|–́× (as hȳ́ran scólde, 10), and that in the same text there are 238 of the type –́××|–́× (as gṓde gewýrcean, 20; hḗold þenden lī́fde, 57), making 830 hemistichs with trochaic or dactylic rhythm, as against eleven hemistichs of similar structure but with an unaccented syllable at the beginning, ×|–́×(×)|–́×, and even four or five of these eleven are of doubtful correctness. From these figures it seems almost beyond doubt that in the type –́×(×)|–́× the licence of letting the hemistich begin with an unaccented syllable before the first accented syllable was, generally speaking, avoided. On the other hand, when the first accented syllable is short with only one unaccented syllable as thesis (⏑́×), we find this initial unaccented syllable to be the rule, as genúmen hǽfdon Beow. 3167 (×|⏑́×|–́×), of which form there are 130 examples, while, as Rieger noticed, ⏑́×|–́× is rare, as in cýning mǣ́nan Beow. 3173. It is perhaps still more remarkable that while the form –́××|–́× occurs some 238 times, a verse of the form ×|⏑́××|–́× is never found at all. The numerical proportion of the form –́×|–́× (592 cases) to –́××|–́× (238 cases) is roughly 5 to 2, and that of ×|⏑́×|–́× (130 cases) to ×|⏑́××|–́× (no cases) is 130 to nothing. The quantity of the second arsis is, as bearing on the prefixing of unaccented syllables to the hemistich, much less important than the quantity of the first arsis. Hemistichs of the type –́×|⏑́× occur 34 times, and in 29 cases the last unaccented syllable is a full word, either a monosyllable or a part of a compound. The same type, with an initial unaccented syllable ×|–́×|⏑́× also occurs 34 times, but then the last syllable is quite unaccented. The proportion of the form –́×|–́× to the form ×|–́×|–́× is 592 to 11, and that of the form –́×|⏑́× to ×–́×|⏑́× is 34 to 34, a noticeable difference.

Further, it was formerly supposed that the number of unaccented syllables following the accented syllable was indifferent. This is not the case. The form –́××|–́× is found 238 times, and the form –́×|–́×× only 22 times. Many of the examples of the latter form are doubtful, but even counting all these the proportion of the two forms is 11 to 1.

If the two accented syllables are not separated by an unaccented syllable, that is to say, if the two beats are in immediate juxtaposition, then either two unaccented syllables must stand after the second arsis, thus –́|–́×× (a form that occurs 120 times in Beowulf), or an unaccented syllable must precede the first arsis and one unaccented syllable must follow the second arsis, thus ×–́|–́× (127 times in Beowulf), or with the second arsis short ×–́|⏑́× (257 times); the form –́|–́× does not occur.

From these statistics it results that hemistichs of the form –́×|–́× are met with about 17 times to one occurrence of the form –́×|⏑́×, and that on the other hand, the form ×–́|⏑́× is about twice as frequent as ×–́|–́×.

§ 24. The order of the verse-members in the hemistich. Every hemistich consists of two feet or measures, each containing an accented syllable. Usually these two feet or measures together contain four verse members, seldom five. In the hemistich of four members, which first falls to be considered, the measures may consist of two members each (2+2), or one may contain one member and the other three (1+3 or 3+1). A measure of one member has a single accented syllable only (–́); a measure of two members has an accented and an unaccented syllable, which may stand either in the order –́× or ×–́; a measure of three members has one accented and two unaccented syllables, one of which has a secondary accent, and the order may be either –́×̀× or –́××̀. Measures of two members may be grouped in three different ways so as to form a hemistich: i. –́×|–́× (descending rhythm); ii. ×–́|×–́ (ascending); iii. ×–́|–́× (ascending-descending)[67]; i. and ii. are symmetrical, iii. is unsymmetrical, but as the number of members in the feet of these three types (2+2 members) is the same, we may call them, as Sievers does, types with equal feet (gleichfüssige Typen), while the others (1+3 members or 3+1 members) may be called types with unequal feet, or measures.

The normal hemistich, then, which consists of four verse-members, will fall, according to the relative position of these measures or feet, into the following five chief types:

a. Types with equal feet (2+2 members)

b. Types with unequal feet

Theoretically type E might be looked on as a type with equal feet, if divided thus, –́×|×–́, but by far the greatest number of instances of this type show at the beginning of the hemistich one trisyllabic word which forbids such a division of feet, as wéorðmỳndum þā́h, Beow. 8.[68] Types like ××–́– and ×̀×–́–́, which we might expect to find, do not occur in Old English poetry. In addition to these ordinary four-membered hemistichs there are others lengthened by the addition of one syllable, which may be unaccented, or have the secondary accent. These extended forms (erweiterte Formen)[69] may be composed either of 2+3 members or of 3+2 members. These extended hemistichs must be carefully distinguished from the hemistichs which have one or more unaccented syllables before the first accented syllable, in types A, D, and E; such a prefix of one or more syllables is called an anacrusis (Auftakt).[70]

The simple five types of the hemistich admit of variation: i. by extension (as above); ii. by resolution (⏑́× for –́) and shortening of the long accented syllable (⏑́); iii. by strengthening of thesis by means of a secondary accent (Steigerung); iv. by increase in the number of unaccented syllables forming the thesis; also (less frequently) v. by variation in the position of the alliteration, and vi. by the admission of anacruses; the varieties produced by the last-mentioned means are not sub-types but parallel forms to those without anacruses.

In describing and analysing the different combinations which arise out of these means of variation, and especially the peculiar forms of the sub-types, the arrangement and nomenclature of Sievers will be followed.[71]

Analysis of the verse types.

I. Hemistichs of four members.

§ 25. Type A has three sub-types, A1, A2, A3.

The sub-type A1 (–́×|–́×) is the normal form with alliteration of the first arsis in each hemistich, or with alliteration of both arses in the first hemistich and one in the second, and with syllables in the thesis which are unaccented according to the usual practice of the language; examples are, þḗodnes þêgnas An. 3, hȳ́ran scólde Beow. 106, gómban gýldan Beow. 11. This is the commonest of all the types; it occurs in Beowulf, according to Sievers, 471 times in the first and 575 times in the second hemistich, and with the like frequency in the other poems.

The simplest modification of this type arises from the resolution of one or two long accented syllables. Examples of resolution of the first arsis are very numerous, cýninga wúldor El. 5, scéaðena þrḗatum Beow. 4, séofon niht swúncon Beow. 517,[72] níðer gewī́teð Beow. 1361. Examples of the resolution of the second arsis are less numerous, as wúldor cýninge El. 291, éllen frémedon Beow. 3, Scýldes éaferan Beow. 19, óft gefrémede Beow. 165; resolution of both in the same hemistich is rare, but is found, as gúmena géogoðe An. 1617, mǽgenes Déniga Beow. 155, gúmum ætgǽdere Beow. 1321.

The chief type is further modified by making the thesis after the first arsis disyllabic (rarely trisyllabic); the formula is then –́××|–×. This modification is frequent, as ríhta gehwýlces El. 910, gṓde gewýrcean Beow. 20, swéordum āswébban An. 72, súnnan ond mṓnan Beow. 94, fṓlce tō frṓfre Beow. 14, wḗox under wólcnum Beow. 8.

Resolution of the arsis may be combined with this disyllabic thesis, as (in the first arsis) wérum on þām wónge An. 22, éotenas ond ýlfe Beow. 112, or (in the second arsis) hā́lig of héofenum An. 89, hélpe gefrémede Beow. 551, or (in both) dúguðe ond géoguðe Beow. 160, hǽleð under héofenum Beow. 52.

The first thesis rarely exceeds two syllables; a thesis of three syllables is occasionally found, as sǽgde se þe cū́ðe Beow. 90, hwī́lum hie gehḗton Beow. 175, and this can be combined with resolution of the first arsis, as swéotulra ond gesȳ́nra An. 565, bítere ond gebólgne Beow. 1431; or with resolution of the second arsis, as ū́tan ymbe ǽðelne An. 873, wī́ge under wǽtere Beow. 1657; or with resolution of both, as réceda under róderum Beow. 310. Examples of thesis of four syllables are (in the first thesis) séalde þām þe hē wólde Beow. 3056, sécge ic þē tō sṓðe Beow. 591. A thesis with five syllables is still less common, as lǣ́ddon hine þā of lýfte Gū. 398, stṓpon þā tō þǣre stṓwe El. 716.

The cases in which the second thesis has two syllables are rare and to some extent doubtful, as wúndor scḗawian Beow. 841 and 3033.[73]

The anacrusis before the type –́×(×)|–́× is also of rare occurrence: examples are swā sǣ́ bebū́geð Beow. 1224, or, with resolution of the first arsis, swā wǽter bebū́geð Beow. 93. Most of the instances occur in the first hemistich; in this position the anacrusis may be polysyllabic (extending sometimes to four syllables), sometimes with resolution of the arsis, or with polysyllabic thesis. Examples: forcṓm æt cámpe An. 1327, gewāt æt wī́ge Beow. 2630; with resolution, ābóden in búrgum An. 78; genéred wið nī́ðe Beow. 828; disyllabic anacrusis ic wæs éndesǣ́ta Beow. 241; with resolution, þǣr wæs hǽleða hléahtor Beow. 612; trisyllabic anacrusis, oððe him Óngenþḗowes Beow. 2475; four-syllable anacrusis, þæt we him þā gū́ðgeatwa Beow. 2637; monosyllabic anacrusis with disyllabic thesis, as in mǣ́gðe gehwǣ́re Beow. 25, āblénded in búrgum An. 78; disyllabic anacrusis with disyllabic thesis, ge æt hā́m ge on hérge Beow. 1249; trisyllabic anacrusis with disyllabic thesis, þū scealt þā fṓre gefḗran An. 216; monosyllabic anacrusis with trisyllabic thesis, gemúnde þā sē gṓda Beow. 759; monosyllabic anacrusis with resolution of first arsis and trisyllabic thesis, ne mágon hie ond ne mṓton An. 1217; with resolution of second arsis, gewā́t him þā tō wároðe Beow. 234; disyllabic anacrusis, ne geféah he þǣre fǣ́hðe Beow. 109; combined with thesis of four syllables, ofslṓh þā æt þǣre sǽcce Beow. 1666.

The sub-type A2 is type A with strengthened thesis (i.e. a thesis with secondary accent) and with alliteration on the first arsis only. This sub-type has several varieties:

(i) A2a, with the first thesis strengthened (–́×̀|–́×); frequent in the second hemistich. The second arsis may be either long or short (–́×̀|–́×, or –́×̀|⏑́×). We denote –́×̀|–́× by A2al and –́×̀|⏑́× by A2ash, or, for brevity, A2l, A2sh. Examples of A2l are, gódspèl ǣ́rest An. 12, wī́sfæ̀st wórdum Beow. 626, hríngnèt bǣ́ron Beow. 1890; with resolution of the first arsis, médusèld bū́an Beow. 3066; with resolution of the second arsis, gā́rsècg hlýnede An. 238, hórdbùrh hǽleða Beow. 467; with resolution of both, fréoðobùrh fǽgere Beow. 522; with resolution of the strengthened thesis, súndwùdu sṓhte Beow. 208; resolution of the first arsis and thesis, mǽgenwùdu múndum Beow. 236; resolution of the first thesis and the second arsis, gū́ðsèaro gúmena Beow. 328.

Examples of A2sh are numerous, as wǣ́rfæ̀st cýning An. 416, gū́ðrìnc mónig Beow. 839, þrḗanȳ̀d þólað Beow. 284; it is exceptional to find the second arsis short when the thesis which precedes has no secondary accent, as Hrḗðel cýning Beow. 2436, Hrúnting náma Beow. 1458, ǽðeling bóren Beow. 2431; with resolution of the first arsis, séaronèt séowað An. 64, snótor cèorl mónig Beow. 909, sígerṑf cyning Beow. 619, mágodrìht micel Beow. 67, &c. Most of the hemistichs which fall under this head have double alliteration.

(ii) A2b, with the second thesis strengthened (–́×|–́×̀). Most of the cases of this type occur in the first hemistich; when they occur in the second hemistich the measure –́×̀ is usually a proper name, not a real compound. Examples: Gréndles gū́ðcræ̀ft Beow. 127, lḗofa Bḗowùlf Beow. 855; with resolution of the first arsis, gámol ond gūðrḕow Beow. 58; with resolution of the second arsis, béorna béaducræ̀ft An. 219; with resolution of both, séfa swā séarogrìm Beow. 595; with resolution of the strengthened thesis, lónd ond lḗodbỳrig Beow. 2472; with resolution of both the second arsis and thesis mǣ́g ond mágoþègn Beow. 408.[74]

This type may still further be varied by a first thesis of two or more syllables, ū́t on þæt ī́glànd An. 15, fólc oððe frḗobùrh Beow. 694, réste hine þā rū́mhèort Beow. 1800; by resolution of the first arsis, glídon ofer gārsècg Beow. 515, of the second, lā́d ofer lágustrḕam An. 423, sýmbel on sélefùl Beow. 620; by resolution of the thesis with secondary accent, éahtodon éorlscìpe Beow. 3173; the anacrusis is rarely found, as gesā́won séledrḕam Beow. 2253, and double alliteration (in the first hemistich) is the rule in this form of type A.

(iii) A2ab, with both theses strengthened –́×̀|–́×̀, bā́nhū̀s blṓdfā̀g An. 1407, gū́ðrìnc góldwlànc Beow. 1882, ǣ́nlī̀c ánsȳ̀n Beow. 251; with resolution of first arsis, wlítesḕon wrǽtlī̀c Beow. 1651, and of the second arsis, glḗawmṑd góde lḕof An. 1581, gū̀ðswèord géatolī̀c Beow. 2155, and of both first and second arsis, héorowèarh hételī̀c Beow. 1268; with resolution of the first (strengthened) thesis, nȳ́dwràcu nī́ðgrìm Beow. 193; with resolution of both the first arsis and the first thesis, býrelàde brȳ́d gèong Gū. 842; with resolution of the second strengthened thesis, égeslī̀c éorðdràca Beow. 2826; with resolution of the first and second thesis, fýrdsèaru fūslìcu Beow. 232. This form of the type has also as a rule double alliteration.

The sub-type A3 is type A with alliteration on the second arsis only and is limited almost entirely to the first hemistich. A strengthened thesis occurs only after the second arsis; this sub-type might therefore be designated A3b.

Verses falling under this head, with their alliteration always on the last syllable but one, or (in the case of resolution) on the last syllable but two, are distinguished by the frequent occurrence of polysyllabic theses extending to five syllables, in marked contrast to types A1 and A2 where theses of one or two syllables are the rule, longer theses the exception. In A3, however, shorter theses are met with along with the usual resolutions: a monosyllabic thesis in hwǣ́r se þḗoden El. 563, ḗow hēt sécgan Beow. 391; with resolution of first arsis, wúton nū éfstan Beow. 3102; with resolution of the second arsis, þús me fǽder mīn El. 528, íc þæt hógode Beow. 633; with disyllabic thesis, hḗht þā on úhtan El. 105, hǽfde se gṓda Beow. 205; with resolution of the first arsis, þánon he gesṓhte Beow. 463; with resolution of the second arsis, wéarð him on Héorote Beow. 1331; with strengthened second thesis, éart þū sē Bḗowùlf Beow. 506; with trisyllabic thesis, gíf þē þæt gelímpe El. 441, fúndon þā on sánde Beow. 3034; with resolution of the first arsis, hwǽðere mē gesǣ́lde Beow. 574, of the second arsis, sýððan ic for dúgeðum Beow. 2502; with strengthened second thesis, nṓ hē þone gífstṑl Beow. 168; with thesis of four syllables, swýlce hī mē gebléndon Cri. 1438, hábbað wē tō þǣm mǣ́ran Beow. 270; with resolution of the first arsis, útan ūs tō þǣre hȳ́ðe Cri. 865; with resolution of the first and second arsis, þóne þe him on swéofote Beow. 2296; with strengthened second thesis, nṓ þȳ ǣr þone héaðorìnc Beow. 2466; with thesis of five syllables, sýððan hē hine to gū̀ðe Beow. 1473; with thesis of six syllables, hȳ́rde ic þæt hē þone héalsbḕah Beow. 2173. These forms are also varied by monosyllabic anacrusis combined with monosyllabic thesis, þe ḗow of wérgðe El. 295, þæt híne on ýlde Beow. 22; with strengthened second thesis, þæt híne sēo brímwỳlf Beow. 1600; with disyllabic thesis, ne þéarft þū swā swī́ðe El. 940, gesprǽc þā sē gṓda Beow. 676; the same with resolution of the first arsis, gewítan him þā góngan Cri. 533; disyllabic anacrusis and disyllabic thesis, ne gefrǽgn ic þā mǣ́gðe Beow. 1012; with resolution of the second arsis, geséah hē in récede Beow. 728; with strengthened second thesis, ge swýlce sēo hérepà̄d Beow. 2259; monosyllabic anacrusis with trisyllabic thesis, on hwýlcum þāra bḗama El. 851; with four-syllable thesis, gewī́teð þonne on sealman Beow. 2461; with resolution of the first arsis, ne mā́gon hī þonne gehȳ́nan Cri. 1525; with resolution of the second arsis, gesā́won þā æfter wǽtere Beow. 1426. The last measure may be shortened exceptionally to ⏑́×, as wǽs mīn fǽder Beow. 262.

On the whole type A seems to occur more frequently in the first than in the second hemistich; in Beowulf out of the 6366 hemistichs of which the poem consists, 2819 fall under this type, and of these 1701 are first and 1118 second hemistichs.[75]

§ 26. The chief type B, ×–́|×–́, has apart from resolutions only one form. But as the second thesis may consist of either one or two syllables, we may distinguish between two sub-types, B1 (with monosyllabic second thesis) and B2 (with disyllabic second thesis). The commonest variation of the type occurs in the first thesis, which may be polysyllabic.