makers of dictionaries, very much remains to be done.

Generalisation

The impression left on our minds by a study of Pânini's collection of roots and of the and Specialisa- meanings which he assigns to them would tion of Roots. certainly lead us to suppose that most roots had in the beginning a general meaning. Roots meaning to go, to move, to hurt, to kill, to sound, etc., form a large majority, and this very fact has often been dwelt on as showing the uselessness of Pânini's definitions. Still, the same idea has very much influenced the researches of the students of the science of language, nor can it be denied that during the periods which we have the best opportunity of watching in the growth of language the tendency in the development of the meaning of roots is certainly from the general to the special.

That development is very much influenced by the Influence of use of prepositions, an influence so great Prepositions. that in several, nay in many cases, the same root can be made to convey not only very varying, but sometimes diametrically opposite meanings. This fact is hardly ever taken into consideration by Pânini. He gives to KHAN, for instance, the meaning of tearing or breaking open (avadârana). We know that in real language this breaking open is always confined to the breaking of the soil, and that KHAN means, to all intents and purposes, to dig. The same root, however, with the preposition ni, assumes the meaning of digging in or making firm 1. At a time when houses consisted of a few piles

The two meanings of striking and being firm are combined also in KHAD; see p. 369.

driven into the soil, covered with reeds and leaves, what was dug in naturally assumed the meaning of firm. In Zend the name for house was kata, i. e. dug, possibly hole, cave, and this name still lives in the Persian kadah, house 1. With the preposition pra, however, the root KHAN means to undermine, with ud, to destroy. Thus GRANTH means to tie together, but with ud, to untie. AñK with apa means to drive away, with prati to honour. NÎ, to lead, with vi, means to train, to educate, with pra to desire, with pari to marry. NAM with ud means to raise, with ni, to bow down. MUK is to let loose, but with prati, to fasten.

Now it may happen that one of the meanings depending originally on a certain preposition, came to prevail to such an extent that the root, even without the preposition, retained the same meaning. Thus NAM means originally to bend, but with pra it comes to mean to bend forward, to pay respect, to worship. In the end, however, the simple NAM may be used to convey the same meaning of worshipping, and the substantive namas never means bending, but always veneration.

But although during the time when the growth of language becomes historical and most accessible therefore to our observation Special Meaning original. the tendency certainly is from the general to the special, I cannot resist the conviction that before that time there was a pre-historic period

1 Schrader, Sprachvergleichung, p. 405. The name for bread also, as being baked between layers of hot ashes, is derived from the same root, the Persian nân, Beluchi naghan; Schrader, 1. c. P. 373.

during which language followed an opposite direction. During that period roots, beginning with special meanings, became more and more generalised, and it was only after reaching that stage that they branched off again into special channels.

If we take, for instance, such roots as now convey the most general meanings, such as being and doing, we can in most cases discover, or, at least, guess their former more special purports. AS, to be, meant to breathe, BHU, to be, to become, meant to grow, VAS, in I was, meant to dwell, VAT, in German werden, meant to turn, KHAK, in German geschehen, vorkommen, meant to jut out 1.

To do meant originally to set, like Sk. DH. To do. To work may have been originally the Sk. VÆG, to twist, to strip. The Sanskrit TAKSH and TVAKSH, to make, to prepare, was originally conceived as cutting and shaping wood. From it we have the substantive tvákshas, vigour, and the adjective tvákshîyas, very strong. The Sanskrit tákshan, carpenter, is the Greek TéκTwv. Greek has also Túkos, chisel. Tvash-tar means a carpenter, but also maker and creator. The verb takshati or taksh-noti still retains the meaning of cutting, splitting and carving, but very soon we find it applied to the making of anything, a carriage, a thunderbolt, heaven and earth, also a thought, a word, a hymn, till at last it came to mean simply to make, as Rig-veda iv, 36, 3, pitárâ púnar yúvânâ karáthâya tákshatha, 'you made your parents young again to walk.'

Pânini assigns to TVAKSH and TAKSH the meaning of thinning (tanûkarane). But he gives

1 This process is repeated in modern languages, e.g. to grow bold, to wax angry, to turn pale; see A. Kühn, Wurzelvariation, p. 6.

another root TAKSH (Dhâtup. xvii. 13), which he explains by tvakane, i. e. skinning, and in this sense tvak, skin, lit. what has been cut off, would have been traced back by him to the same root.

At all events the transition from cutting, shaping, trimming to making in general is perfectly clear. Even in Homeric times the TéKTOV represents the stone-mason, the carpenter, the ship-wright, the wheel-wright, the worker in horn, in bone, in ivory, the turner, the joiner, the belt-maker1 and other handicrafts, so that the generalisation of its meaning became almost a matter of course. Téxvn means art in general.

It is most likely that the German schaffen, too, from which Schöpfer, creator, meant originally to scrape, to polish, and then to make. This root appears in σκάπτω, to dig, to hoe, σκέπαρνον, carpenter's adze, Lat. scabere, scabies, scobîna, file, Goth. scaban, scôf, to shave, and skapjan, skôp, to shape; O.N. skapt, shaft; and skip, ship. Even shape and ship, in friendship, would in the end come from the same source.

Another root RAK, to make, seems to me to have been applied at first to the arranging, crossing, and chaining of threads that were to be twined and platted, thus accounting for the Greek páπTew, to twist, to sew, and for the Lithuanian rìnk-ti, to collect2. Even the Sanskrit KR, to make, I do not hesitate to connect with SKR, in the sense

1 See Riedenauer, Handwerk und Handwerker in den homerischen Zeiten, p. 96; Schrader, Sprachvergleichung, p. 397.

2

See H. Möller, K. Z. xxiv. p. 457. P. Schrader, Ling.-histor. Forschungen, p. 175, presupposes Fрán-тw and compares Lith. werpù, to spin.

CC

of cutting, shearing, in fact with the same root that gives us to shear, and in the Greek kepew and σkάe. In Sanskrit we find kri-nâti having the sense of cutting and hurting, and the compounds upaskirati and pratiskirati clearly mean to split, to tear with nails, having nothing to do with kirati, to scatter. The meaning of cutting comes out very curiously in the Prâkrit kalpayati, to cut, and in kalpanî, scissors, etc., Sk. kripânî (p. 372).

These may seem bold combinations, but they are not without ample analogies in the history of language. In Latin materies meant wood, before it came to mean matter. In Greek λŋ meant wood, before it was used in the sense of substance. In Sanskrit dravya, matter, if derived from dru, tree, would show the same transition of meaning, while, if derived from DRU, to run, to cause to melt, it would have had the special meaning of metal, before it took the general sense of matter.

The name for tree or wood, dru, has been derived from the root DAR, to tear off, to decorticate, showing that the tree was conceived and named as the object and product of the act of felling, chipping, peeling, and shaping a tree, just as dép-μa was the skin torn off from an animal and dri-ti in Sanskrit a leather-bag.

Pânini assigns to DARMH the very general meaning of growing (vriddhau), which To be strong. is probably intended for growing strong and firm. In all the passages where this root occurs, whether in nominal or verbal derivatives, firmness and strength are certainly the meanings conveyed by it. Still it seems far more likely that DmH had originally a more special power,