(a father), geros not gerōn (an old man). This is a very curious instance of retention of an old principle which had seemed to be quite obscured. The old Greek forms ended in pre-historic times with s; and this s in pater-s, geront-s having been dropped, the vowel was lengthened by compensation (§ 38). The modern Greek has replaced the s. In our own language there is a noticeable tendency to form new verbs in ise, e.g., modernise, rationalise, &c. ; this suffix corresponds to the Greek suffix -izō, and came into English through the French -iser in a comparatively small number of verbs; but the list is yearly on the increase. Very parallel is the German verb-suffix -iren when a German wants to naturalise a foreign word this suffix is repeatedly employed, e.g. constru iren; nay, even though ise may be there before, as central-is-iren. 41. This principle of analogy naturally acts, as in the examples which we have been considering, over large classes of words. But there are also changes produced by it in single words, or in but one or two. Thus peas ends in s, because the original final vowel e has been dropped. Hence it came to be regarded as a plural, and a singular pea was made for it. But pease or pese is the old singular form, and one may hear peasen from country-folk still. It is well known that the genitive its is a late form which does not occur in the Bible, his being used instead. The old English pronoun of the third person was he (masc.), heo (fem.), hit (neut.); hit was also the neuter accusative; so t was only the mark of the neuter in these two cases, and had no place whatever in the genitive case. When the initial h fell off, the history of it became obscure; its connection with he was lost; and as genitives were regularly formed by adding s, it was added here too. Both these instances, and many others which might be given, show the mistaken application of a rule to cases for which it was not made; exceptional forms are made to follow the usual analogy. 42. The influence of analogy is often seen in the way in which we make our compound words. In English mis was prefixed to words to express something bad; it occurs as a noun in our older writers, e.g. in the story of William the Werwolf (man-wolf) where we have the line (532): "And to mende my misse I make my avowe." i.e. I make my vow to mend my fault. We still trace the noun in the adverb amiss; also in compounds such as misdeed, mistake; and this was the regular English form for the purpose. Something of the same sort was expressed by dis in Latin, and in the Norman part of our language, as in disturb, discord, &c. ; parting in two seems to have been the primary notion of the word. Now when the English and the Norman vocabularies coalesced, it was natural that Norman suffixes should sometimes get prefixed to English words, and vice versa; and so instead of the English mis-like, there sprang up the mongrel dis-like, half Norman, half English; and by degrees it came to be the rule that all compounds of this sort required dis, on the analogy of those already existing. There is a well known instance in which one English prefix has driven another out. We had in old English fore before, as in fore-tell; and also for, equivalent to German ver (ver-bieten for bid), and Latin per; the idea through or across has brought in by implication the further idea of harm or evil; thus for-swear has the same sense as periuro in Latin; and for-shapen could be used in the same sense as mis-shapen. But the history of this word was forgotten; and compounds with fore increased, till by degrees for was wrongly spelt fore in several words, whose etymology is thereby darkened. We talk of fore-closing in law, and to fore-go a thing, and in each case the false spelling suggests a false = = derivation; fore-fend does not mean 'strike before,' but represents for-fend, 'strike across,' or 'out of the way,' 'prevent.'. Note this last word; the English prefix is combined with a Latin root; which is seen in defend, &c. 43. These instances are enough to show how great an effect this cleaving to a rule, through right or through wrong, may have on a language. I have not time to point out how much of the same effect of analogy upon the mind is to be found in syntax; but Greek scholars may find good traces of it in the history of the genitive with the verb. Uniformity in accentuation is also produced in this way; in English we habitually throw back the stress as far from the end of the word as we can; and when we adopt foreign words, we accentuate them at last after some struggles in the same way (Cp. Ch. VIII., 36). This uniformity is not found in older English, as is obvious to anyone who will look at early rimed poetry, e.g. the metrical Northumberland Psalter. There in the translation of the Eighth Psalm, the verse 'Out of the mouth of babes and sucklings hast thou ordained strength,' appears as "Of mouth of childer and soukánd, where 'soukánd' corresponds to 'lánd'. It was only by degrees that the analogy was established. 44. I shall mention but one more result of analogy. This is the change not merely in the suffix or prefix of a word, but in the whole word which is often caused by the attempt to find some meaning in that which seems to have none. This is strikingly exemplified in names of places. names of places. These commonly contain the name of some person; and if that proper name go out of common use, it is almost certain that the name of the place will be altered so as to represent some known object. Thus the Cumberland lake, Buttermere, was the mere of Buthar, presumably one of the many Norwegians of that name who made themselves homes in the country at Butterhill, Buttergill, &c. Clearly there is no sense in the change; no meaning whatever is gained by it; but butter' was a familiar word, the proper name was unfamiliar ; hence the change. Just in the same way, and in the same country, Bôt-haug, i.e. Bôt's hill, became Boathill, Geit's-garth became Gate-scarth, Sólvar's-seat became Silverside. The Norwegians became Englishmen, as much as the other invaders of England; they were absorbed into the greater body, and their descendants bore English names: and the old proper names were forgotten. Similarly Lizard Point is said to be a corruption of Lazar-point, i.e. an out-of-theway place for lepers. Other corruptions of the same sort are well known; how Dun-y-coed, the Keltic of hill the wood,' has become Dunagoat; how the French Chartreux' has become the Charterhouse; and even the fairly intelligible 'Burgh Walter' has become Bridgwater. 45. Scientific terms naturally suffer severely by this method of handling. Gardeners make strange havoc of the names of plants. I knew one who always called China asters, Chinese oysters; and the power of finding an analogy must have been strained to the uttermost in the man who called chrysanthemums— Christy anthems! Names of diseases are pulled about in the like manner in country talk. In Sussex bronchitis is called the 'brown crisis,' and typhus : sometimes passes into 'titus fever.' We saw above how local etymology acts on the names of animals ($ 19). 46. I have thus shown you the different kinds of change which are found in the form of words apart from their meaning. I have pointed out the general heads to which these changes may be referred, and tried to convince you that underlying the ceaseless variation of spoken languages there are some perma nent principles of general application. We have seen. incidentally that all people are not affected alike by these principles, but that in one language there is more substitution, in another more assimilation; in one language the consonants will be affected, in another the vowels, and so on. But in all that we have yet done we have been seeing how languages change from some previously existing type. We have begun with the phenomena of language which are before our eyes, and tried to work back to some older form. Can we now see how that form was itself developed; how language grew up to a certain point, not how it has been decomposed therefrom? CHAPTER II. SOME OF THE WAYS IN WHICH LANGUAGES HAVE BEEN FORMED. 1. BEFORE a boy has got very far in his Latin grammar, he finds that he must say erit-one word only-when in English he would say 'he shall be.' He will learn that erit can be traced back to an older form es-sya-ti (see Ch. V., 14), and that the parts of that word carried respectively the meanings be-shall-he.' But there was never a time in the history of the Latin language, nor indeed centuries before Rome was founded, when those parts could be used separately. Similarly he will find that erat suffices instead of his own two words 'he was'; sit represents 'he may be ;' fuerit is equivalent to 'he may have been.' From these he will infer that it is the custom of the language to express by one word modes of action which we express by several distinct words. Turning to the nouns he will find saxi when we should say 'of a |