Now, did the writer of our incoherent sentence mean that the start and the scream were co-ordinate, - were independent actions, for the purpose in hand of exactly equal value? Or did he mean that, as a matter of fact, one of these actions was a part of the other, was subordinate? If so, he should have employed a subordinating connective, such as with. "I started up and screamed," means that there were two independent actions, one as significant as the other; "I started up with a scream," means that the two actions really formed one, the former addressing itself to the eye, and in case of physical contact to the sense of touch as well; the latter, a slightly secondary one, addressing itself solely to the ear. This nicety of distinction, in so simple a case apparently unimpor tant, is among the most subtile secrets of effective style; no confusion of thought is commoner than that which confuses subordinate matters and co-ordinate. Nor is there any more direct path to precision of thought than that which leaves subordinate matter on one side and co-ordinate on the other.

To appreciate the full value of skilfully used connectives, we cannot do better than glance back at the passage from Sir Thomas Browne already before us. Its notable coherence, which in total effect quite atones for the weakness of its massing, is due wholly to the connectives. The second word-since-subordinates the opening clauses; the five ands are strictly coordinate in meaning; the such binds the main clause firmly to the subordinate ones that precede; and the

or, slightly loosening the alternative clause with which the sentence ends, goes far to relieve the impression of tension sure to be produced by too sustained a period. Of these connectives the most subtile is such, whose connective meaning does not instantly appear. It is the most subtile because it is placed, not at the beginning of the clause which it binds to the preceding, but in the body of it. To use a figure of speech, it dove-tails style instead of merely gluing it; and this dove-tailing of style is a thing worth attending to. In producing a firmly coherent effect, connectives in the body of a clause or sentence are surprisingly more efficient than initial connectives. Also and too, for example, are decidedly firmer than and; so, in that preceding clause, the connective for example more firmly knits this sentence to the preceding than this clause, with an initial so, is knit to the clause before it. And so I have touched on the two chief guides to precision in the use of connectives: distinguish between subordinate and co-ordinate matter; and prefer connectives in the body of a clause to initial ones.

So much for the principle of Coherence in detail. The test of coherence appears in my very statement of the principle: Is there any chance of mistaking the relation of a word to its neighbors? So far as this chance exists, and it cannot always be avoided, — a sentence is incoherent.

The historical growth of coherence in English style is too large a subject to discuss here. I shall venture,

then, in very few words, to state what I now think about it. In brief, I think and perhaps a study of

my four typical examples will bear me out - that coherence in the order of words has tended on the whole to strengthen; that coherence in construction is far more rare than it used to be; and that coherence in the use of connectives has on the whole tended to grow firmer and more subtile as thought has gained in freedom and precision. In the conflict between good use and the principle of Coherence, then, we find the principle farther advanced than the principle of Mass, but by no means as far as the principle of Unity. And the point where it is now weakest is constructions; few writers nowadays practically remember that phrases similar in thought may to advantage be similar in form.

Toward the end of the last chapter I called your attention to a matter to which we must now revert. Having considered the dangers of offence against good use in our choice of words, you will remember, and having pointed out what notable differences of effect we might secure within the limits of good use by judiciously varying our choice of words, I proceeded to inquire how a careful writer should proceed in his search for the kinds of words that should produce the effects he has in mind. In our discussion of sentences we have now reached this same point: we have discussed the dangers of offence against good use in composition; we have seen how within the limits of good use different kinds of sentences can

produce very varied effects; and we have seen how judicious application of the principles of composition to sentences of any kind -long or short, periodic or loose, balanced or unbalanced- may help us to vary and to define the effects we have in mind. It is our business now to inquire concerning sentences, just as we inquired concerning words, in what these effects consist.

There is no need of repeating in detail what I said then. I pointed out, you will remember, the inevitable discrepancy between the limited number of words in our possession and the virtually infinite number of thoughts in the mind of every living man; and I showed how in fact every word we use or hear not only names an idea, but suggests along with it a considerable number of others: the idea it names it denotes; the ideas it suggests it connotes.

What we then found true of words by themselves must obviously be true, in a vastly greater and more complicated degree, of words in composition. Composition combines every phase of the words it brings together; in the organism of the sentence denotation. and connotation fuse. Take the simplest of examples, -two words: I speak. As I utter these words in combination, the pronoun calls up certain individualities of face and form and manner and dress, and what not. If any one else should utter the same words, the whole connotation would alter. The changed denotation of the pronoun, of course, would be the chief feature of the alteration; but this change

would be more than enough completely to alter the connotation of the verb. Or take a somewhat longer example, but just as simple, where there is no change in denotation at all. Some years ago a gentleman died hereabouts, whose literary style was much admired by the friend who wrote an obituary notice of him: "His English," ran the sentence, which I have remembered for years, "was purified by constant study of the best models: the English Bible, Shakspere, Addison, and Fisher Ames." I confess that this sentence, which has often made me laugh, is what has chiefly kept alive in my mind the memory of our deceased fellow-citizen. But if his admirer had turned the phrase the other way, without altering his denotation a bit, he would have secured a connotation if not more favorable to the immortality of his subject, at least more consonant with its dignity: "His English was purified by constant study of the best models: Fisher Ames, Addison, Shakspere, and the English Bible." Of compositions, then, we may say just what we said of words: in the first place, they so name ideas that we may identify them; in the second place, they inevitably suggest at the same time a very subtile and complicated set of associated ideas and emotions. In short, compositions, like words, inevitably possess both denotation and connotation; and whoever would intelligently compose sentences must know, in deciding what effect he would produce, both what he would denote and what he would connote.