have led to like results in independent instances. But it is not only the simple primary roots which run through all Asiatic speech: the secondary and sometimes the tertiary roots are also common. There are not less than 170 roots in all which can be so traced, and many derived words are also common to languages which are very widely separated by grammatical construction; as has been observed over and over again by all students of comparative philology. These coincidences are too numerous to be passed over as merely accidental, and especially important when we consider the method of formation of the secondary roots.

The secondary roots are of two classes, namely, first those which are reduplications, and secondly those which are formed by combination of two primary roots.

The course of time caused these combinations to be so elided and shortened into monosyllables that their origin was thereby obscured, but in many cases it is very clearly traceable yet. The reduplication gave the impression of continuous action of one kind, whereas the secondary category enabled man to distinguish sounds and actions which were akin, but not identical; and it was on this basis that language rose to become what it already was five thousand years ago.

As an instance of reduplication may be noticed the root kak, common to Aryan and Finnic speech, with the meaning to 'cackle,' and clearly a reduplication of ka, 'to call.' In the nineteenth century was invented the nursery word puff-puff for an engine or train, derived from its puffing continuously; while the word puff itself is only a reduplication of the older pu, to blow or pant. The category of roots which combine. two distinct sounds is best understood by the consideration of what are called 'clam-shell' words in Chinese, which are common in the spoken rather than in the written language, and due to the disintegration of the original dialects, which has rendered so many words indistinguishable from each other. Thus to 'spy' is expressed by kwei-kien, 'to peep-look'; and ta-shui, to strike-sleep,' means to fall asleep. Chinese also gives us instances of reduplication, such as kien, 'to look,' and kien-kien, 'to look earnestly' (that is to repeat the look), but

even with these aids great confusion would exist without the use of tones; for fu according to its tone signifies to call,' 'to support,' a treasury,' 'a wife,' 'riches,' or 'father.'* Chinese, indeed, is not a type of primitive, but rather of decayed language, and nothing is more unsafe than to trust to its vocabulary, although the lessons to be drawn from its grammar are of the highest interest.

The secondary roots seem, then, to have arisen from the primary, just as the clam-shell words have been devised in Chinese, with the object of clearly distinguishing varieties of meaning; and, as in the Chinese, they (and the tertiary roots also) are sometimes due to the prefixing of some word, such as as, ta, etc., signifying action. The prefixed 8 in Hebrew, in s Assyrian, in Egyptian, and in Aryan speech, has a causative meaning, and represents the old verb As, 'to be' or 'become'; while the prefixed t is the remains of the old verb ta, 'to beat,' 'to compel,' and so 'to cause.' As instances of the clam-shell arrangement, we may recall BAR, ' to shine,' found in all Asiatic languages, and probably a combination of the two primary roots already mentioned, BHA, 'to shine,' and 'AR, ‘to burn'; or in another class RAG, 'to go,' from RA, 'go,' and GA, 'go.' The Semitic languages are so rich in these secondary roots, and have so few left of the primary, as to lead to their grammatical arrangement in dictionaries under the headings of triliteral or bisyllabic roots; though only in about a third of the cases is the root truly triliteral, being often formed either by reduplication, or by a vowel, and presenting in the imperative its oldest form as a monosyllable.

The Asiatic languages are divided into four great groups, which again are connected in two pairs. The first pair includes the Egyptian which is agglutinative (or in other words, has not as yet softened down the joints of speech by the wear and tear of the words), and the Semitic languages which are inflexional (that is to say, which have been so melted down by time as often not to show the mode of building up the word); but these two groups are connected not only by a large

* E. J. Eitel, Dictionary of the Cantonese Dialect, p. 19.

vocabulary, including names of colours and numbers, particles, and common verbs, but also by the use of the pronouns ANK, 'I,' and ANT,' thou,' and by the grammatical structure which places the genitive after its nominative, and which distinguishes the feminine by the addition of T. To this great group belong also the Berber languages of North Africa. The second pair includes the Mongol or Turanian languages, which are agglutinative, and the Aryan, which are inflexional; and these, though perhaps less intimately connected than the ceding, are distinguished by placing the genitive before the nominative, and by the use of the pronouns M or V for 'I,' and T or S for thou.'

pre

It would seem that the separation of these groups must have occurred very early, and before the use of pronouns and of a regular syntax had arisen, but, on the other hand, there are demonstratives, and a plural form, which serve to tie together all four families of speech, just as they are also connected by the secondary roots. Thus the demonstratives ma, 'this' (from MA, 'to be'), sa, 'that' (from As, 'to exist'), and KA, 'who' (from GA, 'to be'), occur throughout all Asiatic speech, however much the various families may differ in the specialising of other pronouns. The oldest method of forming the plural by reduplicating the word—as in Bushman language, in Akkadian, and in Chinese-seems to have given way at an early period to the suffixed N, to mark multitude.' Thus na is the Mongol plural, found also in Akkadian and Hittite. In Semitic languages N is the common masculine plural, as also in Egyptian, and it is one of many plurals used by the Aryans. The connection of Egyptian with Semitic, and of Mongolic with Aryan speech, has been indicated by many well known scholars. The ultimate connection of all four families is shown by the identity of roots and demonstratives, and of many simple words.

6

Such then briefly sketched appears to be the natural basis of speech, as indicated not only by the most ancient languages, but also by the exclamations and the imitative words of our own time, and by the customs of savages and of primitive peasants. There is nothing arbitrary or conventional in

the origin of human words; but there is a true distinction between intelligent language and animal cries; and already five thousand years ago the great families of Asiatic speech had become widely distinguished, by structure and grammar even more than by vocabulary, and had (with exception of the Aryan) been committed to syllabic writing in at least three different scripts, which only resemble one another in those signs which refer to the simplest ideas of movement, grasping, beating and pouring, or in emblems which represent animals, and the early weapons and manufactured articles (pots, vases, boats, thrones, houses, altars, and the like) which man was able to construct.

may now turn to the simple names of animals, to words denoting climate, and to those which are connected with arts and civilization; and may thus endeavour to learn something of the common home and early manners of the original Asiatic stock. The names of weapons (even including the bow), of colours, of numbers, and of deities, do not run through the whole circle of the various families, as do the simpler ideas already noticed. They are formed on the same principles, but form very various roots. Thus the names of weapons denote thrusting, cutting, striking, and hacking; and defensive armour is named from roots meaning to cover, to hide, to strengthen, or to protect. In the same way the name of the bow usually comes from one of several roots meaning to bend, but the root is not the same one in every case. The abstract idea of colour is very late, but the four chief colours are very early named. The colour red is connected with words for fire and for blood; blue is connected with words for the sky; white with words for light and shining; and black with words for burning. Yellow and green are little distinguished in early languages, and are connected with words for the sun and for light. Purple and other secondary colours were only distinguished in later times. The Semitic and Egyptian languages agree in the words for red, white and black, and connect them with blood, light and burning; but no such

comparison is possible in other cases. The names of numbers are connected, as has often been shewn, with the fingers

and the hand, but these are more variable and distinctive than perhaps any other class of words. The names of metals come from roots meaning to be yellow, white, red, or dark, to be heavy or strong; but the discovery of the use of metals was not made till long after the separation of the great stocks, in the various cradles in which they grew up as distinct races. Of all the names of deities only those which come from the root as, 'to breathe,' seem to suggest any connection; but it is from this root that the Sanskrit asura, the Zend ahura, the Norse Esyr, the Akkadian es, the Turkic es, and the Egyptian aas, appear all to have been formed, meaning in every case the breathing' or 'living' one.

The animals named by early man were those most conspicuous, most terrible, or most useful, and appear to include the lion, and perhaps the dog, the ox and perhaps the sheep, the ass and not impossibly the camel, but not the horse. Animals distinguishable by the cries they uttered appear to have been named from those cries-especially in the case of birds; but animals which had no distinctive cry were named from some marked peculiarity-the hump, the horns, the speed, or the savageness, which chiefly distinguished them.

Names for the lion were formed from the root RA or LA, 'to roar,' which appears to be clearly imitative. Thus we have the Akkadian ur, 'lion,' and the Turkic ars-lan, where the suffix signifies 'beast.' In Hebrew we find ari, 'lion,' and from the cognate roots la, 'to roar,' come the Hebrew labi, the Aryan leo, and perhaps (if not a loan word) the Egyptian labu. It has been supposed that these names are loan words from Semitic speech, but since the root is common to all the various languages, it appears possible at least that the lion was named from the very earliest times.

Among the many names for cattle none are commoner than those which come from the roots MU and BU, 'to bellow.' Hence we have the Akkadian am, bull,' the Turkic en-ek, 'cow,' the Egyptian am, 'cattle,' and the Mongol buku, ‘bull;' but a more distinctive word is TOR, which seems to mean probably 'horned,' and which appears not only as Taurus, but as the Semitic Thor, and the Mongol Shor. The bull, whether tamed