'hand' to which it is to be referred will be found below,

with its African parallels :

AMERICA.

AFRICA.

Ehnek tirahho, 'five'.1 Mandingo tere, 'inner

The Niger, the Gambia, and the coast of Guinea enclose the area of these African words, and of others like them in the Polyglotta Africana. In the following comparisons

1 The Ehnek is a Californian language.

2 The Kadiak, Kuskutshewak, and Labrador are Esquimaux dialects.

between Australia and Africa, the African words are all from Senegambia and Sierra Leone :

Australia (Port Essington) rujut, 'fingers'.

The next coincidences seem of no great consequence. For, when any form contains only one consonant, it is not

unlikely that resemblances may occur by chance, as in such cases the choice would be limited to some eight sounds, b, g, d, l, m, n, r, s. And even this number might be reduced; as may be seen, for instance, from Malay and Polynesian 'twos', zua, dua, lua, rua, hua, ua, or from the Greek and Latin negative particles, μn and ne. The coincidences now to be noticed may receive, however, some importance, as far as they relate to Southern Asia and Southern Europe, from their agreement with likenesses already observed (pp. 35-37) :

California (Sekumne) ma, 'hand'.

Texas (Comanche) mowa, 'hand, arm'.

Still less stress can be laid on the following African resemblances to our own one', which belong chiefly to Guinea :—eni, ine, inya, inye, onya, onyi, unye, unyi, weni, wono, wunyi, na.

Of all decades, the Japanese is perhaps the most isolated, as it is the most simple. It has the two following forms:

The first series of forms would originally be the same as the second, from which it differs by the addition of -ts as a suffix. The whole decade seems derived from six names for finger', ts, f-t, m, n, y, k; which six names could be reduced to five, if yo and ko were identified together, as both might be with a Japanese and Chinese (Amoy) form for five', go. An additional Japanese 'two', ni, which is also Chinese, Tibetan, and Nepalese, would supply the n finger involved in nanats and kokonots. If yo and ko are virtually the same finger', then the number of 'fingers employed in the Japanese decade would be five, thus giving one name for each finger of the hand. If, on the contrary, yo and ko are not the same 'finger', then ts would be the name for 'finger' in general, and fito, mi, yo, no, and ko would be the peculiar designations of the five fingers respectively, as we ourselves have thumb, forefinger, middle finger, ring-finger, and little finger, where there are six terms in all, finger, thumb, fore, middle, ring, and little.

Out of these elements the Japanese decade would apparently be thus formed:-From ts, the generic name for 'finger', and perhaps the proper name of one of the fingers as well, would be derived the suffix -ts for all the members

H

of the decade; and also, by reduplication, itsuts, 'five', = hand = finger-finger, and tsuds, 'ten', hands. Compare the Gafat (Abyssinia) edzhedzhe, 'hand', a reduplication of edzhe, 'one', i. e. finger'. Removing the suffix, -ts, we shall next find the following Japanese names for I, III, IV, and their doubles, II, VI, VIII: —

I.

fito

f'ta

II.

Though four' is the double of 'two', yet it could not be obtained by modifying f'ta, 'two': for f'ta is itself only a modification of its half, fito, 'one'. A new word is thus required for four'.

From these six forms we may see that plurality, or duality, is implied in Japanese numerals by a change of vowel; in two cases out of three, by changing o into a. There is something like this in Koriak, where, as Pott has noticed, three' seems converted into four' by the same change of o into a; as in ngroka, ngraka-rohgau, ragauniyokh, niyakh-n'roch, n'rach, three, four'. Here the change might be an indication of doubling, if'three' were 2+1, and four', 2 + 2 x 1; i. e., if ni, n', ng be 'two', as ni, nyi, gni, are in Nepal, and if yokh, roch, roka, be 'one', as yek, ri, rik, are in Nepal, and as roka is in the North Australian of Croker Island, a numeral allied perhaps to the rujut, 'fingers', of Port Essington.1 A like peculiarity exists in Kamtshatkan 'threes' and 'fours':tshok,tshak--tshúk, tshaak-tsúk, tshaak-tzogeltsh, tzageltsh tshúk, tshaaka. Here tza, tsha, may be 'two', while tzo, tsho, tsu, tshu, are 'one'.

As o thus seems singular, and a dual, in Japanese, it becomes probable that no in koko-no, nine', is one', and 1 Compare the Peruvian (Quichua) rukkana, finger'.