There can be no object in following the development of the system of Form further than Beethoven, for it can hardly be said that there is anything further to trace. His works present it in its greatest variety and on the grandest scale; and his successors, great as many of them have been, have not even approached him, far less added to his final culmination. The main tendency observable in later instrumental works is to develop still further the system above discussed of taking one key as central in a group comprising many subsidiary transitions. Schumann's works present remarkable instances of this; Mendelssohn adopts the same practice, but with more moderation; Brahms again is extremely free in the same direction; as may be observed, for instance, in the first section of the first movement of the pianoforte Quartet, op. 25, which is nominally in G minor. This is apparently a recognition of the hypothesis above proposed, that the mind is capable of being more and more educated to recognise the principal key in a chain of transitions which to the audiences of Mozart's day would have been quite unintelligible.
It is now time to return to the consideration of the Rondo-form as found in the works of Haydn and Mozart, in which it was frequently affected by the more important and interesting First-movement-form. It will be obvious that its combination with that form does not offer much difficulty. For that alternation of subject and episode which is the very basis of the Rondo opens the way to the adoption of a second subject in the complementary key as the fittest antithesis to the first statement of the principal subject; and the main point of distinction of the Rondo-form from the First-movement-form pure and simple, is that the first subject reappears after the second in the original key, instead of bringing the first half of the movement to a conclusion in the complementary key. After this deviation the form again follows the system of the first movement; for—as we have already sufficiently pointed out—no fitter place is found to develop the figures and features of the subjects and to modulate freely. In the simpler system of the Rondo this again takes the place of an episode; in both systems the first subject would here recur, and nothing could more fitly follow it than the recapitulation of that subject which occupied the place of the first episode. It is worthy of remark that in the Rondo of the Waldstein Sonata, Beethoven has in this place reproduced the subject which opens the first episode, though the movement is not cast on the system of a first movement. Finally, the subject may reappear yet again in the original key without deviating strongly from that system; so that, as just mentioned, the only marked point of deviation is the return to the principal key after the appearance of the second subject. This complete adaptation is more commonly abbreviated by replacing the 'Development' by a short episode (as in Beethoven's Sonata in E, op. 90); and even further (as in the Finale of Mozart's Quartet in E♭, No. 4), by passing immediately from the second subject to the recapitulation of both subjects in the principal key, and ending with one further final quotation of the real Rondo-subject. This latter in point of fact is to be explained rather as a simple method of establishing the balance of keys by giving an episode in a complementary key, than as based on any preconceived notion of amalgamation with the First-movement-form.
One of the most prominent features in the Rondos of Haydn and Mozart is the frequent rigidity of the subject. It is common to meet with a complete dance-tune divided into two halves, each repeated after the accepted system, and closing formally in the principal key. So that it is in fact a complete piece in itself, and stands out as markedly as Couperin's subjects do with fermatas over the concluding chords. In these cases the tune is not given in extenso at each repetition, but is generally fined and rounded off so as not to affect the continuity of the movement so conspicuously as in its first statement.
The angularity and obviousness of outline which often mark the Rondo form in works prior to Beethoven, were to a certain extent alleviated by the use of ingenious playful treatment of the figures of the chief subject by way of episode; but nevertheless the formality remains, and marks the Rondo of Haydn and Mozart as a thing of the past, and not to be revived in their particular manner in the present day without perpetrating an artistic anachronism. Beethoven's treatment of the Rondo offers great differences, but they are chiefly in point of sentiment, and difficult to define. Prior to his day there had evidently been a persistent tradition that final Rondos were bound to be gay, jaunty, light, or even flippant. With Beethoven such a dogma was impossible; and he therefore took the line of developing the opportunities it offered, either for humorous purposes, in the persistent repetition of a quaint phrase (Sonata in D, op. 10, No. 3), or in the natural and desirable recurrence of a melody of great beauty (Sonata in E, op. 90, and Waldstein). In every case the system is taken out of the domain of mere observance of formula, and its basis vitalised afresh by making it the vehicle of thoughts which can appear in such an order without losing their true significance. In point of fact the Rondo form is elastic enough notwithstanding its simplicity, and if the above sketch has not sufficiently indicated that fact, the study of the movements mentioned, and those in Beethoven's E♭ and G Concertos and B♭ Trio, will lead to the perception of the opportunities it offers to the composer better than any attempt at reducing the various features to a formula.
The Minuet and Trio survive as pure and undeveloped examples of the original source of the larger movements, in immediate contact with their wonderfully transformed descendants. They offer no systematic difference whatever from the dances in the Suites which preceded the perfected