trifling variations in the two parts.  The twelve three-part sonatas were issued, as has been said, in 1683.  They are pure, self-sustaining music, detached from words and scenic arrangements; nothing approaching them had been written by an Englishman, nor anything so fine by an Italian.  Indeed, in their own particular way they are matched only by the composer’s own four-part sonatas published after his death.  We must not look for anything like form in the sense that word conveys nowadays; there is no unalterable scheme of movements such as there is in the Haydn symphony, and within each movement there is no first subject, second subject, development and recapitulation.  All that had to be worked out nearly a century later.  The set forms of Purcell’s day were the dances.  The principle of Purcell’s sonata form is alternate fast and slow movements.  Nothing more can be perceived; there is nothing more to perceive.  Sometimes he commences with a quick piece; then we have an adagio or some slow dance; then another quick piece.  In other cases the order is reversed:  a slow movement may be followed by a slower movement.  He makes great use of fugue, more or less free, and of imitation, and, of course, he employs ground-basses.  The masculine strength and energy, the harsh clashing discords, are not less remarkable than the constant sweetness; and if there is rollicking spring jollity, there are also moments of deepest pathos.  There is scarcely such a thing as a dry page.  It is true that Purcell avowed that he copied the best Italian masters, but the most the copying amounts to is taking suggestions for the external scheme of his sonatas and for the manner of writing for strings.  He poured copiously his streams of fresh and strong melody into forms which, in the hands of those he professed to imitate, were barren, lifeless things.  Many of these sonatas might almost be called rhapsodies; certainly a great many movements are rhapsodical.  In set forms one has learnt from experience what to expect.  In the dance measures and fugues, after a few bars, one has a premonition (begotten of oft-repeated and sometimes wearisome experience) of what is coming, of the kind of thing that is coming; just as in a Haydn or Mozart sonata one knows so well what to expect that one often expects a surprise, and may be surprised if there is nothing to surprise one.  But in many of Purcell’s largos, for example, the music flows out from him shaped and directed by no precedent, no rule; it flows and wanders on, but is never aimlessly errant; there is a quality in it that holds passage to passage, gives the whole coherence and a satisfying order.  Emerson speaks of Swedenborg’s faculties working with astronomic punctuality, and this would apply to Purcell’s musical faculties.  Take a scrappy composer, a short-breathed one such as Grieg:  he wrote within concise and very definite forms; yet the order of many passages might be reversed, and no one—­not knowing the original—­would be a penny the wiser or the worse.  There is no development.  With Purcell there is always development, though the laws of it lie too deep for us.  Hence his rhapsodies, whether choral or instrumental, are satisfying, knit together by some inner force of cohesion.