tenable: e.g. "Mwici contains the root Mu (for Mbu=the sea)." The mu in mwisi (Bentley: mwixi=smoke, haze, etc.), is the prefix, and the word is identical with the Zulu umusi, Swahili moshi, and Herero omwise. In Nyanja the prefix is atrophied, and the word appears as utsi, and in other languages (as in Konde ilyosi and Kamba jioki), we find a different prefix. Neither is it at all probable that the prefix mu has anything to do with mbu=the sea, if only because the m in mb is a nasalising of the labial, and not likely to be found without the latter.
Again, the derivation of ka zila, applied to prohibitions, from nzila (njila)=a road, appears doubtful when we remember the Zulu verb zila, meaning "to abstain from, as from certain words or actions, as from certain kinds of food" (Colenso's Dictionary). In Ronga, yila (evidently the same word), and in Ila (the language of the people usually called Mashukulumbwe in N.W. Rhodesia) ku zhila, mean "to be forbidden, tabooed." It may also be pointed out that in all Bantu languages we find many identical words (or rather, one should say, groups of sounds) of different meanings, and, probably, different etymologies. The identity may have been produced by phonetic decay, of which we see the ultimate result in the hundreds of identical monosyllables in Chinese; in any case, it is highly improbable that the Zulu gula="to be ill" has any connection with the Nyanja word gula, which means "to buy," or the Kongo nika="to grind," with Zulu nika="to give." Thus, for instance, it is possible (though one would be sorry to dogmatise on the subject) that the different meanings of kanga (see p. 114) have nothing to do with one another; and, if so, all interpretations based on the contrary hypothesis naturally fall to the ground. How Nyambi (p. 116) means "the spirit or personality of the four," it is difficult to see. Ya is "four"—but Mr. Dennett gives "ia=to be" as one of the constituents of the word—and he cannot have it both ways.
The matter contained in Mr. Dennett's book is so abundant, and of so varied a character, that a full discussion within the limits here imposed would be impossible. All that can be done here is to add a few comparative notes.
The name Nyambi seems, from its occurrence alternately with Nzambi, among tribes both north and south of the Bavili,
