always changed in the same language: sometimes it is retained in a particular group of words-arbitrarily, as it may appear; yet the cause which keeps it there is not the will of any one man or even of many men ; rather it is the general sense that the sound is necessary for the meaning. At any moment, this may cease to be felt; a few people may drop the sound, others may follow them; and after a period of struggle, in which one man pronounces one way and one another, the innocent cause of the war either re-establishes itself or goes the way of its fellows. Thus it is uncertain now whether 'contemporary' will be finally pronounced with the n, or without it: at present even the same person may use both forms. In the same way 'either' varies between eidhur and eedhur (the spelling denotes the actual sounds heard); and it is doubtful whether it will go forward or backward: it will hardly get back to the older form aidhur. The general tendency in English in all such cases is toward the sound ee and the general tendency will probably win in the long run. You may easily find other examples for yourself. These considerations may suffice to show that language is not an abiding work on which man consciously expends his labour: but that it varies according to general principles over which he has no direct control.

38. This brings me to the last point on which I wish to speak. The recognition of these general principles, which govern speech independently of the speaker, has not unnaturally led some philologists to the belief that the science of language should be classed among the physical sciences, rather than among those which deal with the works or the ways of man. In this view languages have been compared to plants, and described as natural organisms, which grow and die out in accordance with fixed laws, independent of the will of man. I cannot enter fully here

into this question: I will only submit one or two points for you to consider.

39. First, the analogy between language and a plant seems incomplete. We may fairly enough speak of the growth and decay of language; meaning thereby the constant development of new forms, to meet the waste caused by the rubbing down of words in daily use or their falling out of use altogether. But the growth is not due to any inherent vitality in languages, as it is in plants it is due to the action of man governed by laws of association—how established we cannot tell— between certain sounds and certain things. Just as we believe that in all history certain consequences necessarily follow certain antecedents; and, if we could know all the antecedents in any one case, we could predict the result with certainty; so in language, there are doubtless causes mental and spiritual, which determine the development of speech, but these also are hidden from our eyes. We must not eliminate the mind of man, as though it were no factor in the production of speech, because we cannot tell with certainty the laws by which it works.

40. Secondly, the death of a language cannot be exactly compared with the death of a plant. A plant dies a natural death when it is no longer capable of receiving from without those elements which are necessary for its growth. But that change in speech, which is so great that one language may be said to have died and a new one to be born, is due indeed to the progressive and never ceasing loss of old elements, but also to the addition of new ones: as when Latin became a 'dead' language, and the Romance languages grew up. When, on the other hand, a language dies out' because all those who speak it have ceased to exist, as the Keltic language in Cornwall, it may die in full vigour and able to perform every function. Such a superseding of one language by another of an entirely different character, is

altogether unlike the ordinary decay of a plant; the language here suffers a violent death. These two considerations seem to me to point to this result: that, while language differs greatly from any ordinary work of human art, it also differs from any natural organism; and the study of language must be classed neither as a historical nor as a physical science, but be placed between the two.

APPENDIX.

(1) Grimm's Law is the name given to a regular interchange of consonants between (i.) Indo-European, with which Sanskrit, Greek, and Latin in the main agree; (ii) the Low German languages; (iii.) Old High German; but this language in its modern form often agrees with the Low German.

The interchange is shown in the following table, where the corresponding sounds are placed horizontally :

By an aspirate is meant a momentary consonant followed by a slight h-sound, not so distinct as in 'backhouse,' 'anthill,' &c., but of the same nature. These sounds, however, are found only in Sanskrit and Greek; in the other languages they are represented by the corresponding continuous consonantsh, ch (German), th, z, f.

The following examples will shew the changes. Greek and Latin forms are given as being well known, instead of Indo-European. English represents Low German :

Note that in Old High German the third change (soft for aspirate) took place only irregularly.

(2) Some of the more important letterchanges in Greek and Latin from the IndoEuropean. Among the vowels we often find that (i.) Indo-European a = Greek and Latin e or o; as Indo-European 'padas' Greek 'podos' (gen. sing.) podes' (nom. plur.) Latin 'pedes' (nom. plur.).

=

=

(ii) In Latin, u often Greek ('ferunt' = 'pheronti'); also i Greek o ('pedis''podos'). (iii.) In Greek a is sometimes weakened to i; thus 'hippos' Indo-European 'akva ;' 'didōmi' = Indo-European 'dadami.'

=

Among the consonants

(i.) In Latin d changes to 7; Ulysses' Greek Odysseus;' rarely to ras 'arbiter''ad-biter' (the 'comer-to').

(ii.) In Greeks at the beginning of a word often