ple of Coherence is that which applies chiefly to matters of detail. To distinguish it from the principle of Mass, indeed, detail might have been a better name for it.

For this very reason, the principle of Coherence is far more difficult to discuss in a few minutes than either of the others. Examples of the observance and the violation of it take so many and such varied forms that at first sight the whole matter seems almost hopeless. I believe, however, that coherence of sentence is dependent on one of three pretty simple general devices; that all the rules I have found to guide us toward it will fall under one of three broadly general ones. By stating these and briefly discussing each in turn, I can certainly treat the subject with more decision than otherwise.

The general principle, we may remember, is this: in a sentence the relation of each word, and each clause to the context should be unmistakable. Now, the mutual relations of words and clauses, indicated primarily in our uninflected language by order of words, may be made evident in three ways: by the actual order of words in detail, by the grammatical forms into which we throw our clauses, and by the use of connectives. Three subordinate rules or principles have therefore phrased themselves in my mind: concerns coherence in the order of

.

iefly to mat-
principle of

better name

Coherence is minutes than ervance and

aried forms

almost hopeof sentence ple general to guide us dly general sing each in more deci

er, is this:

and each

ble. Now, indicated - order of ys: by the ammatical

and by the es or prin mind: my e order of thought thought

in constructions, is this,-phrases that are similar significance should be similar in form. The thi which concerns coherence in the use of connectiv is this, when the order of words and the form constructions prove insufficient to define the relati of a word or a clause to the context, connecti should denote that relation with precision. Th three subordinate rules of coherence I propose to d cuss in turn. They may be discussed most conv iently by means of broadly typical examples.

The example which first occurs to me of cohere in the order of words is one from my own experien Writing a lecture on a part of our subject, pa graphs, — which will be before us later, I put down following sentence: "A glance at any printed pa will show that the points in paragraphs which m readily catch the eye are even more notably than sentences-the beginning and the end." On revisi I found this sentence unsatisfactory. It had uni it was tolerably massed; so far as the principles composition went, then, the trouble must fall und the head of coherence. Under this head my fi question was whether the trouble lay in the act order of the words. So far as good use permits reminded myself, words connected in thought sho be kept together, words distinct in thought k apart. In this troublesome sentence what words longed together in thought, which were not togeth in fact? At a glance I saw that "in paragraph

kent apart two words "points" and "which"

that in thought belonged together; at another glance I saw that the clause, "even more notably than in sentences," not only separated words "are" and "the beginning"-that in thought belonged together, but that in thought this clause belonged with the which had likeother words, "in paragraphs," wise proved out of place. "In paragraphs even more notably than in sentences," then, formulated itself as a distinct clause which demanded insertion in a sentence that without it ran thus: "A glance at any printed page will show that the points which most readily catch the eye are the beginning and the end." Where did the qualifying clause, without which the meaning was obviously incomplete, belong? Obviously between the main verb-"show" and its object; for in some degree it qualified both verb and object. So the sentence fell into this far more coherent form: "A glance at any printed page will show that in paragraphs, even more notably than in sentences, the points which most readily catch the eye are the beginning and the end."

In this single example, then, we may see how to apply a general principle of coherence commonly stated in a number of apparently independent rules: Qualifying words should be close to words they qualify and carefully separated from words they might qualify, but do not; Parenthesis is undesirable; and related in thought should be

ther glance ly than in "are" and ed together, d with the had likeeven more

ed itself as

in a sence at any

hich most the end." which the ? Obvi- and its verb and

more cohe

will show

n in sen

the eye

e how to ommonly nt rules: ney qualy might le; and

ould be

To turn to coherence in constructions, I think of better example of it than the passage from Ral already before us. What preserves its looseness f incoherence is simply and solely the admirable formity of its constructions. First comes the apos phe; then three perfectly independent clauses constructed exactly alike, each admirably balan and notably antithetical: this identity of construc instantly groups them-where they belong toget in the mind of any reader. Finally comes the 1 clause explanatory of the three preceding: slightly ferent in significance, it demands a slight alteratio construction, that it may stand sufficiently apart; not varying from the others in mood or in gen character, it preserves, like them, careful balance antithesis. This example, of course, is old-fashion it applies the principle in a form rather exagg ated for modern style. But it shows more distin than less exaggerated examples the value, in co rence, of balance and antithesis, and of parallel c structions. A very modern example of incohere a sentence from a college theme may serve show, in very few words, how the principle t Ralegh so carefully observed is nowadays commo violated. An undergraduate dabbler in fiction engaged in telling a story where he assumed the cl acter of a young and beautiful woman assaulted b spider: "I started up," he wrote, " and a scream heard." Now, in the context there was no consi

within hanning to ha atoutled her

scream; and except for the purpose of calling attention to the hearers of the scream there could have been no possible reason for changing the construction to the passive voice, and for shifting the subject. What he meant was not what he wrote: it was one of two other things,"I started up and screamed," or "I started up with a scream." In short, he managed, in eight words, to commit the two most common and needless offences against coherence in constructions. He shifted his subject, and altered the voice of his verb from active to passive.

In considering how to improve this incoherent little sentence, we are brought face to face with the third subordinate principle of coherence. When the order of words and the form of constructions prove insufficient to define the relation of a word or a clause to the context, connectives should denote that relation with precision. At first, I dare say, you were surprised to have me say that he meant one of two dif ferent things: either, "I started up and screamed," or, "I started up with a scream." Off-hand there appears here little if any difference in meaning; but really there is a difference which I believe to be very profound. In the first sentence-"I started up screamed" the two actions, starting and screaming, are co-ordinate: the function of and is to assert that ends it connects are of precisely

-

and