the Greek euporía. As long, then, as the word experiment expresses this more or less general idea, it has a real existence. But take the mere sound, and change only the accent, and we get experiment, and this is nothing. Change one vowel or one consonant, exporiment or esperiment, and we have mere noises, what Heraclitus would call a mere psóphos, but no words. Character, with the accent on the first syllable, has a meaning in English, but none in German or French; charácter, with the accent on the second syllable, has a meaning in German, but none in English or French; charactère, with the accent on the last, has a meaning in French, but none in English or German. It matters not whether the sound is articulate or not; articulate sound without meaning is even more unreal than inarticulate sound. If, then, these articulate sounds, or what we may call the body of language, exist nowhere, have no independent reality, what follows? I think it follows that this so-called body of language could never have been taken up anywhere by itself, and added to our conceptions from without; from which it would follow again that our conceptions, which are now always clothed in the garment of language, could never have existed in a naked state. This would be perfectly correct reasoning, if applied to anything else; nor do I see that it can be objected to as bearing on thought and language. If we never find skins except as the teguments of animals, we may safely conclude that animals cannot exist without skins. If color cannot exist by itself (åñaν yàp χρῶμα ἐν σώματι), it follows that neither can anything that is colored exist without color. A coloring sub

stance may be added or removed; but color without some substance, however ethereal, is, in rerum naturâ, as impossible as substance without color, or as substance without form or weight.

Granting, however, to the fullest extent, the one and indivisible character of language and thought, agreeing even with the Polynesians, who express thinking by speaking in the stomach, we may yet, I think, for scientific purposes, claim the same liberty which is claimed in so many sciences, namely, the liberty of treating separately what in the nature of things cannot be separated. Though color cannot be separated from some ethereal substance, yet the science of optics treats of light and color as if they existed by themselves. The geometrician reasons on lines without taking cognizance of their breadth, of plains without considering their depth, of bodies without thinking of their weight. It is the same in language, and though I consider the identity of language and reason as one of the fundamental principles of our science, I think it will be most useful to begin, as it were, by dissecting the dead body of language, by anatomizing its phonetic structure, without any reference to its function, and then to proceed to a consideration of language in the fulness of life, and to watch its energies, both in what we call its growth and its decay.

I tried to show in my first course of lectures, that if we analyze language, that is to say, if we trace words back to their most primitive elements, we arrive, not at letters, but at roots. This is a point which has not been sufficiently considered, and it

1 Farrar, p. 125.

may almost be taken as the general opinion that the elements of language are vowels and consonants, but not roots. If, however, we call elements those primitive substances the combination of which is sufficient to account for things as they really are, it is clear that we cannot well call the letters the elements of language; for we might shake the letters together ad infinitum, without ever producing a dictionary, much less a grammar. It was a favorite idea of ancient philosophers to compare the atoms, the concurrence of which was to form all nature, with letters. Epicurus is reported to have said that

"The atoms come together in different order and position, like the letters, which, though they are few, yet, by being placed together in different ways, produce innumerable words." 1

Aristotle, also, in his "Metaphysics," when speaking of Leucippus and Democritus, illustrates the different effects produced by the same elements by a reference to letters. "A," he says, "differs from N by its shape; AN from NA by the order of the letters; Z from N by its position."

It is true, no doubt, that by putting the twentythree or twenty-four letters together in every possible variety, we might produce every word that has ever been used in any language of the world. The number of these words, taking twenty-three letters as the basis, would be 25,852,016,738,884,976,640,000; or, if we take twenty-four letters, 620,448,401,733,

1 Lactantius, Divin. Inst., 1 b. 3. c. 19. Vario, inquit (Epicurus), ordine ac positione conveniunt atomi sicut literae, quae cum sint paucae, varie tamen collocatae innumerabilia verba conficiunt.

2 Metaph., i. 4, 11. Διαφέρει γὰρ τὸ μὲν ἡ τοῦ Ν σχήματι, τὸ δὲ ΑΝ τοῦ ΝΑ τάξει, τὸ δὲ Ζ τοῦ Ν θέσει.

239,439,360,000.1 But even then these trillions, billions, and millions of sounds, would not be words, for they would lack the most important ingredient, that which makes a word to be a word, namely, the different ideas by which they were called into life, and which are expressed differently in different languages.

"Element," Aristotle says, "we call that of which anything consists, as of its first substance, this being as to form indivisible; as, for instance, the elements of language (the letters) of which language is composed, and into which as its last component parts, it can be dissolved; while they, the letters, can no longer be dissolved into sounds different in form; but, if they are dissolved, the parts are homogeneous, as a part of water is water; but not so the parts of a syllable."

If here we take phōné as voice, not as language, there would be nothing to object to in Aristotle's reasoning. The voice, as such, may be dissolved into vowels and consonants, as its primal elements. But not so speech. Speech is preeminently significant sound, and if we look for the elements of speech, we cannot on a sudden drop one of its two characteristic qualities, either its audibility or its significancy. Now letters as such are not significant; a, b, c, d, mean nothing, either by themselves or if put together. The only word that is formed of mere letters is "Alphabet" (å åλpáßŋros), the English ABC; but even here it is not the sounds, but the names of the letters, that form the word.

1 Cf. Leibniz, De Arte combinatoria, Opp. t. ii. pp. 387, 388, ed. Dutens; Pott, Elym. Forsch., ii. p. 9.

One other word has been supposed to have the same merely alphabetical origin, namely, the Latin elementum. As elementa is used in Latin for the ABC, it has been supposed, though I doubt whether in real earnest, that it was formed from the three letters 1, m, n.

The etymological meaning of elementa is by no means clear, nor has the Greek stoicheion, which in Latin is rendered by elementum, as yet been satisfactorily explained. We are told that stoicheion is a diminutive from stoichos, a small upright rod or post, especially the gnomon of the sundial, or the shadow thrown by it; and under stoichos we find the meaning of a row, a line of poles with hunt ing-nets, and are informed that the word is the same as stichos, line, and stóchos, aim. How the radical vowel can change from i to o and oi, is not explained.

The question is, why were the elements, or the component primary parts of things, called stoicheia by the Greeks? It is a word which has had a long history, and has passed from Greece to almost every part of the civilized world, and deserves, therefore, some attention at the hand of the etymolog ical genealogist. Stoichos, from which stoicheion, means a row or file, like stix and stiches in Homer. The suffix eios is the same as the Latin eius, and expresses what belongs to or has the quality of something. Therefore, as stoichos means a row, stoicheion would be what belongs to or constitutes a row. Is it possible to connect these words with stóchos, aim, either in form or meaning? Certainly not. Roots with i are liable to a regular change