three very simple heads, each of which might be phrased as a simple proposition. Various as they are, all these directions concern either what may be included in a given composition (a sentence, a paragraph, or a whole); or what I may call the outline, or perhaps better, the mass of the composition,-in other words, where the chief parts may most conveniently be placed; or finally, the internal arrangement of the composition in detail. In brief, I may phrase these three principles of composition as follows: (1) Every composition should group itself about one central idea; (2) The chief parts of every composi tion should be so placed as readily to catch the eye; (3) Finally, the relation of each part of a composition to its neighbors should be unmistakable. The first of these principles may conveniently be named the principle of Unity; the second, the principle of Mass; the third, the principle of Coherence. They are important enough to deserve examination in detail.

I have said that all compositions should have unity, -in other words, that every composition should group itself about one central idea. The very terms in which I have phrased this principle suggest at once the chief fact that I have tried to keep before you in the earlier part of this chapter, that words are after all nothing but arbitrary symbols standing for ideas. So really, when we come to consider the substance of any composition, we may better concern ourselves rather with what the words stand for than with the visible symbols themselves. If we once know what

ideas we wish to group together, the task of finding words for them is immensely simplified; on the other hand, if in the act of composition — an act which is generally rather hasty - we have grouped together a number of words, the question of whether we shall leave them together, or strike out some, or add some, is generally to be settled by considering not what visible forms our composition has associated, but what ideas. Now, the principles on which we may properly group ideas together are as various as anything well can be. In the first place, as we have just seen, there are various kinds of compositions, sentences, paragraphs, and those larger kinds which for convenience I have grouped under the single head of wholes. Obviously there is in good style some reason why the unity of the sentence should be more limited than that of the paragraph, and the unity of the paragraph than that of the whole. Yet, as our purposes in composing vary, we may perfectly well devote to a single subject George Eliot, for example a book, a chapter, a paragraph, or a sentence. Any decently written life of George Eliot - Mr. Cross's, let us say - has unity, in that it groups itself about one central idea; namely, the notable writer in question. Any history of English fiction in the nineteenth century—to be sure, I do not at this moment recall one worth mentioning would probably contain a chapter about George Eliot which would possess unity for precisely the same reason. So, in a general account of contemporary English literature, we should be rather

surprised not to find at least a paragraph devoted to George Eliot, and this paragraph would have unity for precisely the same reason that caused us to recognize it in the imaginary chapter, or in Mr. Cross's book. And a very short article a leader in a newspaper, for example - which should deal with modern novels in general would be more than apt to contain at least a sentence about George Eliot, of which the unity would be demonstrable in exactly the same way. In other words, the question of scale in many aspects important - has very little to do with the question of unity. The question of unity is whether for our purposes the ideas we have grouped together may rationally be so grouped; if we can show that they may, we are safe. Analogies are often helpful: we may liken the grouping of ideas in compositions to the grouping of facts in statistics. A group of statistics, such as the director of the Harvard gymnasium calls anthropometic, may concern a single individual; again, a genealogy concerns, as the case may be, a family, or a group of families related by blood or marriage; a local history, such as we have hundreds of in New England, properly concerns a considerable number of families who have lived at different times under the same political conditions; a State or a national census concerns the entire population of State or nation, and groups it too in any number of different ways. But each of these things has a unity of its own; and to a certain degree each larger group contains each smaller one.

Here, I think, is the chief thing to keep in mind: just as the sentence is a group of words, the paragraph is a group of sentences, and the whole a group of paragraphs. We should take care that each group has, for our purpose, a unity of its own; and that the unity of each larger group is of a kind that may properly be resolved into the smaller unities of which it is composed.

In considering the question of unity, then, we consider rather what the words stand for than the visible words themselves. In considering the second principle of composition, the principle of Mass, I conceive the case to be different. Style, you will remember, I defined as the expression of thought and emotion in written words. Written words we saw to be visible material symbols of that immaterial reality, thought and emotion, which makes up our conscious lives. What distinguishes written words from spoken, literature from the colloquial language that precedes it, is that written words address themselves to the eye and spoken words to the ear. Though this fundamental physical fact has been neglected by the makers of textbooks, I know few more important. The principle of Mass, you will remember, the principle which governs the outward form of every composition, — is that the chief parts of every composition should be so placed as readily to catch the eye. Now, what catches the eye is obviously not the immaterial idea a word stands for, but the material symbol of the idea, - the actual black marks to which good use has in course

cance.

of time come to attach such subtile and varied signifiIn these groups of visible marks that compose style certain parts are more conspicuous than others. Broadly speaking, the most readily visible parts of a given composition are the beginning and the end. Run your eye over a printed page; you will find it arrested by every period, more still by every one of those breaks which mark the division of paragraphs. Compare a book not broken into chapters - Defoe's "Plague" for example — with a book in which the chapters are carefully distinguished; and you will feel, on a conveniently large scale, the extreme mechanical inconvenience of the former arrangement. On the other hand, compare the ordinary version of the Bible-broken into verses whose separation is based chiefly on the fact that each by itself will make a tolerable text—with the Revised Version, in most respects so deplorably inferior as literature: in the former case, it is mechanically hard, unless somebody is reading aloud to you, to make out which break is important, which not; in the latter case, the task is mechanically easy. Or again, remark a fact that is becoming in my literary studies comically general: familiar quotations from celebrated books are almost always to be found at the beginning or the end. "Music hath charms" are the opening words of Congreve's "Mourning Bride." Don Quixote fights with the windmill very early in the first volume; he dies with the remark that there are no birds in last year's nests near the end of the last. Until I read