tonic syllables of the words; should the accent fall upon an atonic syllable, the duration of the note to which the tonic syllable is sung may be increased, to avoid the apparent discordance between the musical accent and the tonic syllable. The musical accent must fall upon the rime and the rhythm adopted at the outset will persist throughout the poem.

Hence a study of the words will give the key to the interpretation of the tune. If, for instance, the poem shows accented followed by unaccented syllables or trochees as the prevalent foot, the first "mode" is indicated as providing the principle to be followed in transposing the Gregorian to modern notation. When these conditions are reversed the iambic foot will prevail and the melody will be in the second mode. It is not possible here to treat this complicated question in full detail for which reference must be made to the works of J. Beck. But it is clear that the system above outlined is an improvement upon that proposed by such earlier students of the subject as Riemann, who assumed that each syllable was sung to a note or group of notes of equal time value. There is no evidence that such a rhythm was ever employed in the middle ages, and the fact that words and music were inseparable in Provençal lyrics shows that to infer the nature of the musical rhythm from the rhythm of the words is a perfectly legitimate method of inquiry.

A further question arises: how far do the tunes correspond with the structure of the stanza as given by Dante? In some cases both tune and stanza correspond in symmetrical form; but in others we find stanzas which may be divided according to rule conjoined with tunes which present no melodic repetition of any kind; similarly, tunes which may be divided into pedes and coda are written upon stanzas which have no relation to that form. On the whole, it seems that the number of tunes known to us are too few, in comparison with the large body of lyric poetry existing, to permit any generalisation upon the question. The singer accompanied himself upon a stringed instrument (viula) or was accompanied by other performers; various forms of wind instruments were also in use. Apparently the accompaniment was in unison with the singer; part writing or contrapuntal music was unknown at the troubadour period.

As has been said, the stanza (cobla) might vary in length. No poetical literature has made more use of rime than Provençal lyric poetry. There were three typical methods of rime disposition: firstly, the rimes might all find their answer within the stanza, which was thus a self-contained whole; secondly, the rimes might find their answer within the stanza and be again repeated in the same order in all following stanzas; and thirdly, the rimes might find no answer within the stanza, but be repeated in following stanzas. In this case the rimes were known as dissolutas, and the stanza as a cobla estrampa. This last arrangement tended to make the poem a more organic whole than was possible in the first two cases; in these, stanzas might be omitted without necessarily impairing the general effect, but, when coblas estrampas were employed, the ear of the auditor, attentive for the answering rimes, would not be satisfied before the conclusion of the second stanza. A further step towards the provision of closer unity between the separate stanzas was the chanso redonda, which was composed of coblas estrampas, the rime order of the second stanza being an inversion of the rime order of the first; the tendency reaches its highest point in the sestina, which retained the characteristic of the chanso redonda, namely, that the last rime of one stanza should correspond with the first rime of the following stanza, but with the additional improvement that every rime started a stanza in turn, whereas, in the chanso redonda the same rime continually recurred at the beginning of every other stanza.

Reference has already been made to the chanso. A poetical form of much importance was the sirventes, which outwardly was indistinguishable from the chanso. The meaning of the term is unknown; some say that it originally implied a poem composed by "servants," poets in the service of an overlord; others, that it was a poem composed to the tune of a chanso which it thus imitated in a "servile" manner. From the chanso the sirventes is distinguished by its subject matter; it was the vehicle for satire, moral reproof or political lampooning. The troubadours were often keenly interested in the political events of their time; they filled, to some extent, the place of the modern journalist and were naturally the partisans of the overlord in whose service or pay they happened to be. They were ready to foment a war, to lampoon a stingy patron, to ridicule one another, to abuse the morality of the age as circumstances might dictate. The crusade sirventes¹⁴ are important in this connection, and there were often eloquent exhortations to the leaders of Christianity to come to the rescue of Palestine and the Holy Sepulchre. Under this heading also falls the planh, a funeral song lamenting the death of a patron, and here again, beneath the mask of conventionality, real emotion is often apparent, as in the famous lament upon Richard Coeur de Lion composed by Gaucelm Faidit.

Reference has been already made to the tenso, one of the most characteristic of Provençal lyric forms. The name (Lat. tentionem) implies a contention or strife, which was conducted in the form of a dialogue and possibly owed its origin to the custom in early vogue among many different peoples of holding poetical tournaments, in which one poet challenged another by uttering a poetical phrase to which the opponent replied in similar metrical form. Such, at any rate, is the form of the tenso; a poet propounds a theme in the first stanza and his interlocutor replies in a stanza of identical metrical form; the dispute usually continues for some half dozen stanzas. One class of tenso was obviously fictitious, as the dialogue is carried on with animals or even lifeless objects, such as a lady's cloak, and it is possible that some at least of the discussions ostensibly conducted between two poets may have emanated from the brain of one sole author. Sometimes three or four interlocutors take part; the subject of discussion was then known as a joc partit, a divided game, or partimen, a title eventually transferred to the poem itself. The most varied questions were discussed in the tenso, but casuistical problems concerning love are the most frequent: Is the death or the treachery of a loved one easier to bear? Is a lover's feeling for his lady stronger before she has accepted him or afterwards? Is a bad noble or a poor but upright man more worthy to find favour? The discussion of such questions provided an opportunity of displaying both poetical dexterity and also dialectical acumen. But rarely did either of the disputants declare himself convinced or vanquished by his opponents' arguments the question was left undecided or was referred by agreement to an arbitrator.

A poetical form which preserves some trace of its popular origin is the pastorela¹⁵ or pastoral which takes its name from the fact that the heroine of the piece was always a