NUMERALS.

I have compared the Dakotan numerals with all others accessible to me, including some of the forms of more than five hundred dialects. I can find less than half a dozen American or Turanian sets that resemble any Dakotan set as much as the English numerals resemble the Hebrew. The similarity of the Dak to the I E numerals can therefore be accounted for only as the result of special relationship or of accident. Except as noted below all changes are in accordance with well sustained laws.

1, A S an, Lith (w)ena; Dak (w)-an, ind. article wanzhi one, wancha one, once.

2, I E dwa; Min d(o)pa; Iowa n(o)wa; Dak n(o)m pa cf A S ta two; Dak ta a pair.

3, I E traya; Iowa tanye; Dak ya -mni [or ya (m) ni?]

4, I E k-atwar; Iowa towa; Dak S topa; Y tom; T tol.

5, I E kankan, kwankwan; Mand kikhun; Dak zaptan?

6, I E kswakswa; Win hakwa; Iowa shagwa; Dak shakpe.

7, A S seowon; Dak shakowin.

8, I E aktu, Gk hokto; Dak Y sh-akdo-ghan; Sant sh-ahdo-ghan.

9, I E nawan; Dak na (pchi) wan-ka.

10, I E dwakan; Lat decem; Dak wikchem-na.

5, I E k = Dak z otherwise sustained but not proved. Kw = kp = tp = pt, t and k being interchangeable before labials in Dak.

7, Neither A S seowon nor Dak shakowin are legitimately deducible from saptan. Perhaps sakan, sakwan was the true base.

8, Either Gk h or Dak sh may equal I E s. Dak d for I E t is rare but S. hd, Y. kd is a favorite combination.

9, I cannot explain inserted pchi.

10, In Dak m and n are interchangeable before labials, but m for I E n is here unsupported.[I] D cannot stand before w in Dak.