the relative is omitted but in which the relative, if expressed, would be the object of a preposition, the only idiomatic position for the preposition which the language permits is at the end of the sentence. And in complex sentences containing the double relative what as the object of a preposition, the more natural word order calls for the preposition at the end of the sentence. So we have two such perfectly idiomatic forms of expression as This is the very chair he sat in, and I know what you are thinking about. These sentences may seem to have a slightly colloquial flavor, but if so, this is because the literary style is so much inclined to favor artificial and theoretical refinement of language. But to demand that these two sentences be changed to This is the very chair in which he sat, and I know about what you are thinking would be putting the language into a bed of Procrustes and stretching it until the life had been stretched out of it.

Many other locutions in the language with all the prestige of long and respectable use back of them are nevertheless often condemned as incorrect by the theorist who rests all correctness upon logic. A case in point is represented by those words which supposedly name an absolute form or condition, words like round, square, perpendicular, horizontal, and other geometrical terms, or words like full, perfect, complete, finished, and many similar words. The objection is that since these words express ideas which when rightly conceived are fixed and final, it is logically impossible, therefore linguistically indefensible, to use the words comparatively. By this rule it would be incorrect to speak of something as being more round or most round, more square or most square. For anything which is round or square or perpendicular or horizontal is that

and nothing else, and therefore there can be no question of variability about it. So the idea of perfect does not permit such modifications as more perfect or most perfect. But obviously no degree of logical condemnation of such constructions can make them incorrect, or can diminish their use in the practical life of the language. And in this instance it must be said that the logician's argument is unusually feeble. For when these words of absolute meaning are thus used comparatively, they plainly do not have their absolute meanings. When one of two objects is described as being rounder than the other, the word "rounder" is not a comparative degree of the geometrical idea of roundness, but all the statement means is that one of the two objects approximates more nearly than the other the state of roundness. And when the word "perfect" is used comparatively, it is not thought of as expressing an absolute state, but the word becomes in this use nothing more than a vague intensive of approval.

Another construction in general use is that illustrated by such a sentence as I don't go to the theatre oftener than I can help. But the logician reconstructs this. He tells us to say what we mean and that what we mean is, I don't go to the theatre oftener than I can't help. But though this reconstructed sentence corresponds in detail more exactly to the logic of the situation, the plain fact is that however correct logically it may be, idiomatically it is incorrect. Idioms are often defined as phrases the sense of which cannot be deduced from the words considered separately, like to wade in his own blood, to put up with, bids fair, in this connection, to have a falling out, and hosts of similar expressions. If the logical test were applied to these phrases in detail, then all of them would be incor

rect. And undoubtedly persons obsessed with the notion of logic often do refrain from using simple, homely idiomatic English. They prefer a They prefer a more mathematically exact English, and they are not troubled if it is also a cold and lifeless English.

The double negative is no longer permissible in cultivated English, and the reason ordinarily given for its disappearance from approved speech is that it violates logic. Two negatives, the common statement runs, make an affirmative. But two negatives do not make an affirmative in the intention of the person who uses the double negative, nor does the cultivated speaker when he hears the illiterate speaker deliver himself of a sentence containing two negatives take these negatives as an affirmative. The two negatives are an affirmative to him only when he sits down and in the pride of intellect reflects that two negatives ought to make an affirmative. But for hundreds of years throughout the whole of the AngloSaxon and the Middle English periods, the cultivated English language employed two negatives to express negative meaning. The second negative merely strengthened the meaning of the first negative. And if a third was added, as in Chaucer's He nevere yet no vileynye ne sayde, the expression still did not become an affirmative but only a negative one shade stronger.

From the foregoing discussion it becomes apparent that a little logic is a dangerous thing. It is a dangerous and futile practice to quibble over the details of expression the intent and meaning of which are not in doubt. What counts in the end is not how the language might express or ought to express meaning, but how it does express meaning. If the meaning is expressed, the purpose of

language is satisfied. For all language is logical in the sense that it conveys a meaning when the language communication is adequate. But the language communication may be adequate and the form of it may still not be analyzable into logically clear and clean-cut elements. The part of wisdom in such instances is to take language as we find it.

XII

AUTHORITY

ANALOGY, logic, and etymology as principles of correctness in English are at least based upon reasonable judgments which are reached by the observation of some of the genuine processes of language. Their insufficiency as final guides to the use of good English arises from the fact that the living language of a people is not immediately subject to reason. Language is a possession of society, and social activities do not respond as quickly and regularly to rational direction as perhaps they ought to do. In this chapter, however, a different method of determining correctness will be examined, a method in which rational principles are subordinated and an appeal is made to an authority which removes from the individual the responsibility of making judgments of his own.

Authority is of various kinds. Upon every person rests the necessity of submitting to a power beyond individual control in carrying on the activities of the language with due respect to its proprieties. But there is a great difference between a blind submission and an intelligent、 submission to authority. It is the former of these two kinds of submission to authority that will be considered in the paragraphs of this chapter.

One kind of blind submissiveness in language is that which goes in general with temperamental conservatism. By nature many persons seem inclined to like only that which is customary and established, and to look with