soon become fixed; there are few if any vestiges of the intermediate links, and so the better half of the evidence of the change is wanting. (3) Among the incumbrances or illusions of language may be reckoned many of the rules and traditions of grammar, whether ancient grammar or the corrections of it which modern philology has introduced. Grammar, like law, delights in definition: human speech, like human action, though very far from being a mere chaos, is indefinite, admits of degrees, and is always in a state of change or transition. Grammar gives an erroneous conception of language: for it reduces to a system that which is not a system. Its figures of speech, pleonasms, ellipses, anacolutha, pros to semainomenon, and the like have no reality; they do not either make conscious expressions more intelligible or show the way in which they have arisen; they are chiefly designed to bring an earlier use of language into conformity with the later. Often they seem intended only to remind us that great poets like Aeschylus or Sophocles or Pindar or a great prose writer like Thucydides are guilty of taking unwarrantable liberties with grammatical rules; it appears never to have occurred to the inventors of them that these real “conditores linguae Graecae” lived in an age before grammar, when “Greece also was living Greece.” It is the anatomy, not the physiology of language, which grammar seeks to describe: into the idiom and higher life of words it does not enter. The ordinary Greek grammar gives a complete paradigm of the verb, without suggesting that the double or treble forms of Perfects, Aorists, etc. are hardly ever contemporaneous. It distinguishes Moods and Tenses, without observing how much of the nature of one passes into the other. It makes three Voices, Active, Passive, and Middle, but takes no notice of the precarious existence and uncertain character of the last of the three. Language is a thing of degrees and relations and associations and exceptions: grammar ties it up in fixed rules. Language has many varieties of usage: grammar tries to reduce them to a single one. Grammar divides verbs into regular and irregular: it does not recognize that the irregular, equally with the regular, are subject to law, and that a language which had no exceptions would not be a natural growth: for it could not have been subjected to the influences by which language is ordinarily affected. It is always wanting to describe ancient languages in the terms of a modern one. It has a favourite fiction that one word is put in the place of another; the truth is that no word is ever put for another. It has another fiction, that a word has been omitted: words are omitted because they are no longer needed; and the omission has ceased to be observed. The common explanation of kata or some other preposition “being understood” in a Greek sentence is another fiction of the same kind, which tends to disguise the fact that under cases