opportunity for the half-break at the end of the first tercet, corresponding to the same midway in the octave and to the full break at the latter's close. It would be a mistake, however, to dogmatise where the difference is so slight, and the poet will probably instinctively use the tercets in just correspondence with his emotional impulse. The Italian masters recognised as the best that division of the sestet into two distinct tercets (which they termed volte, or turnings), which, while not interfering with what Mr. Watts calls the ebb-movement of the sestet, are fully capable of throwing out two separate lights in one gleam-like the azure hollow and yellow flame in burning gas. The sestet of the true sonnet then may be expressed by the following formula :— a b-c-a-b-c This is the normal type, but the following are among entirely permissible and more or less appropriate variations : a a 4 5 6 7 8 9 10 11 12 13 14 15 16 17 b b bba a а bb а a bb ΔΙΧΧΟ b bb 'A b RDCOR с с a а b a iv. a a a a хээ с с с с а xxviiic. APATK The figures in the third division of this Table denote examples among the sonnets in this book of the variation in question. Of these, it seems to me that the two most musical-the least disturbant to the melodic wave -are the first and third, The occurrence of a rhymed couplet at the close of the sonnet is rare indeed in Italian literature: I cannot recall a single example of it among the classic masters of the sonnet, and even in later times I fancy it would be difficult to find a single good Italian example worthy the name with this termination. But it does not necessarily follow that a closing couplet is equally unpleasant to the ear in English, for in the latter practically all sonnets are what the Italians call mute, that is, the rhyming terminals are in one syllable, while in the language of Petrarca and Dante they are trisyllabic and dissyllabic—a circumstance materially affecting our consideration of this much debated point. Not only are there few good English sonnets with dissyllabic terminals (I remember none with trisyllabic throughout, and do not suppose there is an example thereof to be found) but there are few of any quality. Among the best are two to be found in Mrs. Alice Meynell's delightful Preludes, and one of these is only partially so constructed; both will be found in this volume, 'Spring among the Alban Hills,' as No. CXxxviii, and 'A Day to Come' in its due place in the appendix. But notwithstanding the differences in terminal structure, it is open to question if the rhymed couplet-ending be not almost as disagree able to the English as to the Italian ear. One of the chief pleasures of the sonnet is the expectancy of the closing portion, and when the ear has become attuned to the sustained flow of the normal octave and also of the opening lines of the sestet, the couplet comes upon one with an unexpected jar, as if some one had opened and banged-to a door while the musician was letting the last harmonious chords thrill under his touch. There has been a good deal written on this point, and Mr. Hall Caine and others have succinctly pointed out their reasons for strongly objecting to it. It is, moreover, perhaps the last point on which sonneteers themselves will agree-the last redoubt to be held before the victorious assault of the critical enemy. Writing some three years ago on this subject, I stated that "if the arrangement of lines suits the emotion, I am not offended by a concluding rhymed couplet, or by the quatrains used to so such purpose by Shakespeare, Drayton, and TennysonTurner" but then, undoubtedly, only one side of the question was clear to me. Continuous study of the sonnet has convinced me that while many English poems in this form, even by good writers, are markedly weakened by rhymed couplet-endings, there is one sonnet-form where the closure in question is not only not objectionable but is absolutely as much the right thing as the Petrarcan type is for the octave. Most writers on the sonnet either state generally that they object or that they do not object to the rhymed couplets at the close: thus one anonymous critic writes that he fails 'to see wherein a couplet-ending is not as musical as any other arrangement, that indeed it is demonstratably so by the citation of some of the most striking sonnets in one language'-while on the other hand Mr. Caine refers to the closure in question as being as offensive to his ear as the couplets at the ends of scenes and acts in some Shakespearian plays. It seems to me now that there are broadly speaking but two normal types of sonnet-structures the Petrarcan and the Shakespearian: whenever a motive is cast in the mould of the former a rhymed couplet-ending is quite out of place and is distinctly unpleasant to the ear; whenever it is embodied in the latter the couplet-closure is eminently satisfactory. Thus no sonnet commencing with the two-rhymed solid octave should ever have a sestet with a rhymed couplet-ending, and no sonnet composed in three separate quatrains should close in any other way. Before, however, considering the five chief types (primarily, two), I may finish my general remarks on the early history of the sonnet. That by the fourteenth century the mature sonnet was fully understood and recognised is evident from the facts (set forth by Mr. Tomlinson) that of the forty examples attributed (one or two of them somewhat doubtfully) to Dante, thirty-three belong to the strict Guittonian type: of the three hundred and seventeen produced throughout a long period by Petrarca, not a single one has more than two rhymes in the octave, and only fifteen have any variations from the normal type (eleven in alternate rhymes, and four with the first, third, sixth, and eighth lines harmonising); while two hundred and ninety agree in having nothing more than a double rhyme both in the major and in the minor system one hundred and sixteen belonging to the pure Guittonian type, one hundred and seven with the tercets in two alternate rhymes (Type I. in foregoing table), and sixty-seven with three rhymesounds, arranged as in Type VII. in foregoing table. Again, of the eighty sonnets of Michael Angelo, seven-eighths are in the normal type. It is thus evident that, at a period when the Italian ear was specially keen to all harmonious effects, the verdict of the masters in this species of poetic composition was given in favour of two sonnet formations-the Guittonian structure as to the octave, and the co-relative arrangement of the sestet a-b-ca—b—c, or a—b-a-b-a-b, with a preference for the former. Another variation susceptible of very beautiful effect is that of Type IX. (ante), but though it can most appropriately be used when exceptional tenderness, sweetness, or special impressiveness was sought after, it has not been much availed of. An example of this type may be found in the following pages (No. clxxii.), but I may also quote here in exemplification of it one of the most beautiful of all Italian sonnets. It is one of Dante's, and is filled with the breath of music as a pine-tree with the cadences of the wind-the close being supremely exquisite: while it will also afford to those who are unacquainted with Italian an idea of the essential difference between the trisyllabic and dissyllabic terminals of the southern and the one-syllable or 'mute' endings of the English sonnet, and at the same time serve to emphasise what has been |