LANGUAGES 895 



])eeii the case. At the same time we find the most arcliaic Bantu dialect 

 in existence at the })resent day on the western slopes of Blount Elgon. 

 The next most arcliaic dialect perhaps is Lukonjo of 8outliern Kuw«Mizori. 

 but Lukonjo is run rather hard for this post of secondary pre-eminence bv 

 Luganda and Runyoro and by the Kiemba of Southern Tanganyika, It 

 would be easy now to fix on [Mount Elgon as having been tlie hub of the 

 Bantu universe but for one detail, with which I am afraid I must weary 

 the two readers who may be still remaining in my audience. There is a 

 very marked feature in the bulk of the Bantu languages in the presence of 

 the syllable "Pa-" as a place prefix. In the majority of the archaic Bantu 

 languages the " Pa-" prefix is always associated with locality. The oldest 

 Bantu word for " place " was "apantu," which was analogous to " umuntu," 

 a man, "ikintu," a thing, etc. Now the consonant "p" is a very unstable 

 letter. It so easily degenerates between the human lips into •' v," '• f," "• w," 

 and " h." But in my own small researches into phonology I have never 

 known the ''h " to develop into a "p." Xow throughout the Bantu languages 

 of the I'ganda Protectorate the locati\e prefix is never •• Pa-." With the 

 exception only of the Luganda language it is invariably "Ha-." In 

 Luganda, it is true, the " Pa-" prefix has become " Wa-," which is, no 

 doubt, a less marked deterioration. It is, however, an almost omnipresent 

 feature in all the Bantu tongues round the shores of the Victoria Xyanza, 

 on Kuwenzori, at the north end and west coast of Tanganyika, and even 

 through British East Africa close to the shores of the Indian Ocean,* that 

 the place prefix should never be " Pa-" but nearly always " Ha-." 

 Therefore all the Bantu languages to the south and west of this large 

 area which retain •• Pa-" or '• Va-" as the locative prefix are in this 

 respect in a more archaic condition than the Bantu languages of the 

 Uganda Protectorate.! 



Formerly the obstacle to my theories of locating the original home 

 of the Bantu race between the Victoria Nyanza and the White Nile lay 

 in the non-existence of the tenth (•' Iti- " or " Izi-") prefix ; but since I 

 have discovered this to exist in the languages of Kavirondo and of West 

 Elgon, and also in the Lukonjo of Ruwenzori, there only remains the 

 problem of the - Pa- " prefix to be solved ; and if this solution is not adverse 

 to the derivation of the Bantu tongues from the region now dwelt in by 

 the Baganda, Banyoro, and Kavirondo Xegi-oes, we may be obliged to 

 change our views as to the ultimate source of the Bantu people and 

 language from West Africa to North-East Africa. Dr. Shrubsall has 

 shown the Karamojo Negroes by their physical conformation to be of 



* Except, interestingly enough, in Kikamba of Ukamba, where it is "Pa." 

 t The " Ha-" disease, if I may so style it, spreads westwards down the west 

 coast of Tanganyika and right across to the T'pper Congo. 



