When it is remembered how such theories fill the literary air, it will be realized what great restraint E. E. Cummings imposed on himself in the matter of alliteration and other tricks with letters. He would not, we feel, let such theories run away with him to the extent of forcing his choice of words to depend more on the sense of their sounds than on the sense of their images. His choice of _swarms_, for instance, is primarily determined by the three meanings combined in the word (the crowding sense, the bee-buzzing sense, and another hitherto not noted--the climbing sense associated with _spires_ and the eye looking up to the light); not by the occurrence of _s_ and _z_ or by the presence of _warm_ in _swarms_, though these are accidents of which he takes every advantage. And this is the way such things should happen in poetry, by coincidence. The poet appreciates and confirms rather than elaborately stage-manages. A certain amount of superstitious faith in language is necessary if the poet is going to perform the sort of miracles expected of and natural to poetry. CHAPTER II THE PROBLEM OF FORM AND SUBJECT-MATTER IN MODERNIST POETRY MODERN French poetic theory lays a great deal of emphasis on the phonetic sense of words; and has done so increasingly since the Symbolists. For a long time, indeed, the French have been dissatisfied with the success of poetry as compared with other arts, and have attempted to remedy its supposed deficiencies by bringing it closer to music. To do this they have had to insist on a musical meaning accompanying the word-meaning, on introducing a system of letter-notation similar to musical notation. Three lines from Paul Valéry will illustrate this picture-making in poetry by the help of sounds: Il se fit Celui qui dissipe En conséquences son Principe, En étoiles, son Unité. Now, since we are able to recognise _dissipe_, _conséquences_, _Principe_, and _Unité_ by their English parallels, we must rewrite these lines in some practical phonetic notation which will completely divorce them from any associated meaning, if we would test their direct phonetic value: Eel s’ fee s’ lwee kee deesseep p’ Ahng kohnsaykahng s’ sohng Prangseep p’, Ahng aytwal l’, sonn Ewneetay. This is the best rough phonetic approximation that we can make without the use of a formal phonetic system. We are immediately impressed by the recurrence of the strong _s_ and the narrow _ee_ sound, as we are supposed to be. This might denote a number of things: a man whetting a scythe, a child writing on a slate, or a serpent trying to talk. On the other hand, such sounds might have nothing to do with the subject; as in the couplet: As fleecy sheep we leap Across this grassy sweep;