the normal prose accent, and in two extreme cases he allowed this accent to usurp altogether the function of prosodical length.

§ 24. Early hymns of uncertain date. Here we seem to witness the last stage of transition, before the complete transformation of metre into rhythm. The hymns just considered afford at least prima facie reason for believing that the change was a natural and gradual one, that the rhythmical effect upon the unsophisticated ear was the same in each kind of verse, that the later hymns were not merely syllabic, but accentual, and that the change was begun by the device of making the metrical ictus of quantitative verse coincide with the prose accent of the words, and carried a step further by the substitution, at first only tentative, of accented for long syllables. The hymns of Ambrose and Fortunatus are the only ones in iambic metres which can with certainty be assigned to known authors of this transition period: and for reasons which will appear in the next chapter, the early hymns in quasi-trochaics cannot properly figure in the comparison. The materials are so scanty that our deductions cannot, of course, be regarded as independently conclusive; but it will be seen in the next chapters that their value can be greatly increased by corroborative evidence from other sources.

Many of the so-called Ambrosian hymns, (i. e. hymns in stanzas of four iambic dipodies, either metrical or "rhythmical"), and those attributed to Damasus and Hilary, would be of great value to the student of versification, if only he were in a position to speak confidently of their authorship and date. Those credited to the two authors named, if known to be authentic, would be of especial interest as being the oldest extant specimens of the Latin hymnology; but they are pretty certainly not authentic. To Damasus tradition ascribes the celebrated hymn on the martyrdom of Agatha, in dactylic verse:

21 Jam renitens quasi sponsa polo, etc.

And to Hilary the hymn beginning "Lucis largitor splendide" is ascribed by his biographer Fortunatus, as well as by a fairly consistent tradition. This hymn (written for the most part strictly in quantity) contains two notable lines, namely

and

Ne rapientis perfidi (19)

Haec spes precantis animae (29).

The hymn of Damasus seems to reveal a comparative disregard of quantity in the arsis: for though the blunder in the 5th line might be purely accidental, the false quantity in renitens (or renidens according to other MSS.) cannot have been unnoticed, as the long penult controls the accent of the word. But though both these hymns (like divers others of similar character), could obviously be fitted into our theory very conveniently, the uncertainty of their truc dates makes it unsafe to rely upon them. Both are later than the time of their supposed authors, but we cannot tell how much later except by reasoning in a circle. It is to be hoped that theological, literary and historical scholarship may before long clear up some of these questions, and so make considerable additions to the material now available for our present inquiry.

§ 25. Ambrosian hymns of Adam of St. Victor. Before considering the more complex features of the problem, it will be useful to examine the rhythm of the Ambrosian hymn in its last stage of development. Adam of St. Victor, the skilful (if hardly great) poet of the 12th century, was the most perfect master of Latin "rhythmical" composition, and has left many specimens of his work.

Among these are seven Ambrosian hymns. If these hymns are read accentually, they exhibit a strikingly large number of initial inversions, either single (i. e. of the first foot) as in "Jesu, tuorum militum", or double (i. e. of the first two feet together), as in "Felix ortus infantulae". The frequency of these inversions can be conveniently shown by the following table.

In somewhat more than half of the verses we find initial inversions, either single or double. But in all the 310 lines there are only two possible instances of inversion in the 3rd or 4th foot. These are both in the hymn "Aeterni festi gaudia", viz.:

Quae vellet potest mens pia (32)

and Qua praefulget Augustinus (51)

In the former of these cases the inversion is obviously no inversion at all; the line is to be read with a wrenched accent. This is shown clearly enough by the fact that pia rimes with omnia, scientia and caetera. The latter case is doubtful. The whole stanza is as follows:Datur et torques aurea

Pro doctrina catholica:

Qua praefulget Augustinus

In summi regis curia.

In as much as the line in question does not rime, (contrary to the uniform rule of this hymn), it seems likely that the inversion is real, and that the poet de

(1) The last line of this hymn, "Sancto sit spiritui", is defective, and I have not included it in the table. The same in true of the next hymn.

signedly introduced an entire trochaic line into this stanza: but of course the passage may be corrupt.

The frequency of initial inversions in such hymns as these seems to give color to the theory that the verse does not depend upon accent at all, that it is in theory only syllabic. But the fact that the inversions are practically always initial-(there is no instance of inversion in the 2nd foot alone, unless possibly in 1. 37 of the first hymn in the table, "O mors Christi mirifica)-seems strongly to negative this theory. It is perfectly true, of course, that final inversions would have prevented masculine rime; and the principle of systematic rime is the one principle that is admitted, in addition to that of syllabism, by those who uphold the theory that these verses are not accentual.(1) But this is no sufficient answer, for it is not clear how the principle of rime can be scientifically separated from that of accentual rhythm. If words like Maria and maria had been systematically treated as of equal fitness for the verse-end, we might regard the two principles as quite separate; but they in fact were not. Further confirmation of our view will be found in the next section: but for the present it may merely be remarked that the syllabic theory is that (for the most part) of French and German scholarship, and that our ears are perhaps readier to perceive the true character of these rhythms than those of the French and the Germans, because the latter are accustomed in their own poetry to a much more regular alternation of thesis and arsis than is found in ours, while the former, from the nature of their language, are hardly trained to appreciate an accentual rhythm at all.(2)

(1) See for example Kawczynski p. 117 et seq. It should perhaps be noted that the use of final monosyllables would have enabled the poet to invert the third foot alone without marring the rime, had he cared to do so.

(2) See for example a remarkable comment of Kawczynski's (p. 44). "Les poètes allemands qui imitèrent le vers octosyllabique des romans

It has been assumed in the foregoing paragraphs that these inversions in the poems of Adam are real inversions, not wrenched accents. That this is the case is clear enough from the fact that they occur only at the beginnings of lines. For that in accentual poetry an inverted foot is positively agreeable in the beginning, and in general positively disagreeable at the end of a verse, is clear enough; but no such assertion can be made as to wrenched accents. If a line like

Jesu dulcis memoria,

read with regular stress on the even syllables, was satisfactory to the ear, so would be such lines as

Te nostra vox primum sonet,(1)

Fortunatus to the contrary notwithstanding. And the comments of the grammarians, (see § 36, post), leave no doubt that these verses were read accentually.

§ 26. Trochaics of Adam of St. Victor. The opinion uged in the last section, as to the theoretical presence of the accentual principle in Adam's Ambrosian hymns, receives positive confirmation from his treatment of trochaic rhythms. Kawczynski says(2) (after arguing for the purely syllabic character of Latin "rhythms" in general):-"Il faut toutefois accorder une large exception au rhythme trochaïque. La langue latine ne possédant pas d'oxytons polysyllabiques, et les proparoxytons n'y

français ne parvinrent pas à donner à leurs vers un nombre de syllabes déterminé. Cela les gênait trop probablement. En voici la règle exprimée par un rimeur du quatorzième siècle :

Ouch ich diss getichniss rim

Uef die zal der silben zune,

Sechse, sibene, achte, nune.

Il rime son écrit, dit-il, sur le nombre des syllabes, dix, sept, huit ou neuf. Le seul rhythme qu'on donnait à ces vers consistait donc dans la rime. (!)

(1) From the "Aeterne rerum conditor" of Ambrose.

(2) p. 119.