THE SAKAI AND SEMANG DIALECTS.- 55 
given in the above table are well attested, and it is noticeable 
how little agreement there is between the Sakai on the one hand 
and the Sémang and Bésisi respectively on the other. It is true 
that one appears to be the same in groups I and II and possibly 
this is due to the fact that [Lis a mixed group of Sakai with a tinge 
of Sémang in it, as is evidenced by other words common to these 
two groups. But the author’s attempt to derive the forms of - 
eroups I, II and III from the purely Mon-Annam forms of group 
IV is more or less conjectural, and even if it is correct it leaves 
one with the curious result that the pure Sakai is (as regards 
the numerals) further removed from the regular Mon-Annam 
type than the mixed Bésisi and its neighbours. This group IV 
consists of a string of outlying dialects scattered along the 
border line between the pure Sakai and the Jakun, in a tract of 
country which extends from Ulu émbéling and Kuantan 
(Pahang) to the Jasin district of Malacca and from Kuala 
Langat (Sélangor) to Ulu Indau (Johor). In this group 
alone * do the numerals extend beyond four, and that fact as 
well as their singularly good state of preservation (in these very 
mixed dialects) seems to me to indicate that these Mon-Annam 
numerals were not native to the aboriginal dialects of the 
Peninsula but were imposed from without, and that they either 
have nothing whatever to do with the Sakai numerals (from 
which they certainly cannot be derived) or that they have 
filtered through into Sakai in degenerate forms. It seems very 
unlikely that the pure Sakai first imposed its numerals (in a 
primitive form) on the Jakuns who speak Bésisi etc., and then 
proceeded to corrupt them while the Bésisi preserved them 
unchanged. 
So far as this evidence goes, it appears to me to tell against 
the conclusion which the author ultimately arrives at, viz: that 
all the aboriginal dialects of the Peninsula are branches of the 
~ Mon-Annam stock. 
gee It will be seen, too, that it is a mistake to regard the various 
dialects as corruptions, in different degrees, of one single type 
w. Some rather dubious lists of numerals beyond ‘‘four” are 
given by two or three authorities, but all differ inter se and are suspect - 
ed on that ground. 
R. A. Soc., No. 39, 1902. 
