head, and so meet his fate. A confused vituperative clamor this, each speech independent of every other, but all combining in a single exasperating effect. To separate these speeches into independent paragraphs would be wholly to destroy the effect. They should be written in a single paragraph.

So much for what unity of paragraph means. We may understand it still better by inquiring how to test it. While not scientifically exact, I have found the test I shall propose to you very instructive. A paragraph has unity when you can state its substance in a single sentence; otherwise it is very apt to lack it.

This subject is physically too large to be conveniently illustrated here. I must ask such of you as wish to prove it by observation, then, to make observations for yourselves. One or two examples from my own experience, however, may be suggestive aids. At Harvard College, some years ago, I had occasion to consider in detail Burke's speech on Conciliation with America. I have never read a more astonishingly lucid presentation of a very complicated subject. How is this lucidity secured? was one of my first questions. Pencil in hand, I analyzed the whole speech; and from beginning to end I found not a single paragraph whose substance could not be summed up in a single sentence. Again, there is in this country a newspaper whose style is always notable for certainty of effect: I mean "The Nation." I often dislike what it says, but I have rarely found

in it a leading article that at least rhetorically I have not admired. On analysis I have shown myself again and again that whoever write these leading articles in "The Nation"-I refer to the political articles, not to the scholarly letters, and so on, which are often disfigured with all the most lifeless pedantries of modern Germany - rarely write a paragraph whose substance cannot be summed up in a single sentence. Of this masterly making of paragraphs in "The Nation," I shall have more to say when I come to speak of their mass.

Perhaps it may be worth our while here to glance at this whole matter of unity from another point of view. The mere physical bulk of paragraphs makes this method far simpler here than in the case of sentences. What, we may now ask ourselves with some hope of a simple answer, are the chief dangers of offence against unity of composition? Obviously they are two: first, we may break up discourse into needlessly small fragments, thereby, in this case, confusing the function of the paragraph with that of the sentence; in the second place, we may crowd into a single unit of composition incongruous matters, thereby, in this case, confusing the function of the paragraph with that of the whole composition.

From this consideration follows directly a practical suggestion. Excessive length of paragraph, resulting in heterogeneity, and excessive brevity of paragraph, resulting in isolated fragments of style, are alike unfavorable to unity. Proverbial wisdom is wisest after

all here we are face to face with a special case of what we all know, in medio tutissimus ibis. I may add that a study of the historical development of English tends on the whole to show that unity of paragraph is constantly, though slowly, improving. As in the case of unity of sentence, then, principle and usage tend to agree.

To revert for a moment to the matter with which I began this chapter, you will remember that the way to use what we get into our heads about paragraphs is precisely opposite to the way to use what we know about sentences. In that case, we apply our knowledge in revision; in this case, we apply it in prevision,-in the deliberate planning of our work. It follows, then, that whoever wishes his work to produce the effect secured by intelligent unity of paragraph may wisely set about the task of writing as deliberately as this on a sheet of paper he may prudently write down a scheme of the work he wishes to execute, phrased in as many independent sentences as he would ultimately have paragraphs in his composition; and in filling out this scheme he may wisely confine each of his paragraphs to one of the aspects of his subject which he has provisionally phrased in a single sentence. Unless inspiration override all canons of art, it sometimes does with all of us, I know of no rule of literary conduct more fruitful of good than this. So we come to the principle which governs the external form of paragraphs, the principle of Mass: that the chief parts of each composition should be so

placed as readily to catch the eye. More than the other principles of composition this applies to written discourse, for only written discourse appeals directly to the eye. To be sure, written discourse is closely related to spoken. The principle of Mass will be found by no means useless to a mere talker. But, at least for our purposes, it is primarily a matter not of spoken style, but of written. Now, paragraphs are essentially elements of written discourse. It follows directly that the principle of Mass that the chief parts of a composition should be so placed as readily to catch the eye—is above all applicable to the composition of paragraphs.

In paragraphs, too, the oral usage which we saw in terfere with the principle in the composition of sentences has no existence at all. The principle, then, is not only theoretically applicable to paragraphs, but to a great degree actually so applicable in practice. How conspicuous the chief places in any paragraph are, a glance at any printed page will show. Trained or untrained, the human eye cannot help dwelling instinctively a little longer on the beginnings and the ends of paragraphs than on any other points in the discourse. Let any one of you take up a book or an article, hitherto strange, and try in a few minutes to get some notion of what it is about. Whoever has tried to do even very little reviewing for the newspapers; whoever has tried to collect authorities for a legal brief, — knows the experience disagreeably well. First, you instinctively look at the beginning of the

article or book, then at the end; then, turning over the pages, you skim them, in other words, you glance at the beginning and at the end of each paragraph, to see whether it is a thing you wish to read more carefully. And if the paragraphs in question be well massed, you are made aware of it by the fact that the process of intelligent skimming is mechanically easy: that you can, apparently by instinct, arrest your attention on those parts which serve your purpose. If, on the other hand, as is more frequently the case, the paragraphs in question be ill massed, you find difficulty in discovering what you want. All this is quite independent of sentence-structure, and of unity, and of coherence. It is a simple question of visible, external outline; and it means, in other words, that the beginning and the end of a paragraph are beyond doubt the fittest places for its chief ideas, and so for its chief words.

A definite question now presents itself to us: Is there any test by which we may decide what the chief ideas and the chief words in any paragraph ought to be? We have already seen that a paragraph should possess unity; we have already seen that the test of unity in a paragraph is whether we can sum up its substance in a single sentence. Now, clearly the chief words in a typical sentence are the subject and the predicate. Clearly, then, in general, the chief ideas in a paragraph are those which are summarized in the subject and the predicate of the sentence which summarizes the whole. Our question, then, proves