any points of interest seem to be in cases where we find either a contrast aimed at in the passage which forms the link, or a number of repetitions succeeding one another, with differences in the passages connecting them. These two constitute the two great branches through which this primitive idea diverged into thousands of Arias, Lieder, Nocturnes, Romances, Scherzos, and other lyrical pieces on the one hand, and the movement which still retains its name of Rondo on the other. As an early example of the first we may take the song 'Roland courez aux armes' from Lully's opera 'Roland,' which is too long for insertion here, but will be found in the 136th chapter of Hawkins's 'History of Music.' In this there are 12 bars of melody in C, concluding in that key; followed by 12 more bars, in which there is modulation first to the relative minor A, and then to the dominant key G major, in which key this portion concludes; after which the first twelve bars are resumed precisely as at first, and so the whole concludes. Here the employment of modulation in the connecting passage is a strong element of contrast, and indicates a considerable advance in musical ideas on the obscure tonality of the preceding example. On the other hand, almost contemporary with Lully, there are, in the works of Couperin, numerous specimens of the Rondo, consisting of a number of repetitions, with differences in the connecting passages. In these the passage with which the movement commences is repeated over and over again bodily and without disguise, and separate short passages, of similar length but varying character, are put in between. Couperin was particularly fond of the Rondo-form, and examples may be found in profusion in his works. The one which is perhaps best known and most available for reference is the 'Chaconne en Rondeau,' published in the sixth number of Pauer's 'Alte Claviermusik.' A point specially observable in them is the rigidity and absence of any attempt at sophistication in the process. The sections are like crude squares and circles fitted together into a design, and no attempt, or very little at best, is made to soften off the outlines by making the sections pass into one another. The chief subject is distinct and the episodes are distinct, and the number of repetitions seems to depend solely on the capacity of the composer to put something in between. Still it is clear that the virtue of contrasts both of style and of key is appreciated, though the range of modulation is extremely limited. It is noticeable moreover, as illustrating the point of view from which Form at that time was regarded, when recognised as such, that the divisions of the Rondo are marked with extra emphasis by a Fermata or pause. From this to such a Rondo as we find in the Partita in C minor of Bach is a great step. Here there are no strongly marked divisions to stiffen the movement into formality, but it flows on almost interruptedly from first to last. The episodes modulate more freely, and there is not such rigid regularity in the reappearance of the main subject. It appears once outside of the principal key, and (which is yet more important) is brought in at the end in an extremely happy variation; which is prophetic of Beethoven's favourite practice of putting identical ideas in different lights. The next stage of development of this form—and that probably rather a change than an improvement on the above beautiful little specimen of Bach—is the Rondo of Haydn and Mozart. Their treatment of it is practically the same as Couperin's, but in many cases is strongly modified by the more important and elaborate 'First-movement-form,' which by their time had grown into clearness of system and definition. The Rondo-form pure and simple has remained till now much as it was in Couperin's time, gaining more in expansion than in change of outline. Even the great Rondo of Beethoven's 'Waldstein' Sonata (op. 53) consists of the repetition of a subject of some length interspersed with episodes; with modifications in the length of the episodes and the repetition of one of them, and a great Coda founded on the principal subject to conclude with. The further consideration of the Rondo as affected by the 'first movement' form must be postponed till after the examination of the latter.
By the side of the primitive Rondo above quoted a form more complex in principle is found. In this form the relations of harmonic roots come largely into play, but its most striking and singular feature is the manner of the repetition by which it is characterised. And in this case examples drawn from various early sources which agree in the peculiar manner of the repetition will be of value, as above indicated. In this form the movement is divided into two halves, and these again into two sections. The first half, or complete period, comprises a sort of rough balance between the amount which tends to the Tonic and the amount which tends to the Dominant, thereby indicating the division into two sections; and the second half begins with passages which have more freedom in the distribution of their roots, which constitutes its first section, and ends with a quotation of the last bars or figures of the first half, which constitutes its second section. This will be best understood from an example. The following is a very early specimen of the dance tune called a 'Branle' or 'Brawl,' from the 'Orchesographie' of Thoinot Arbeau (Langres, 1545):—
In this it will be observed that the first half of the little tune is divided at (a) by the strong emphasis on the Dominant, from which point it returns to the Tonic, and so closes the first