OF THE CARDINAL NUMBERS. s. 328. In one sense the cardinal numbers form no part of a work on etymology. They are single words, apparently simple, and, as such, appertaining to a dictionary rather than to a grammar. In another sense they are strictly etymological. They are the basis of the ordinals, which are formed from them by derivation. Furthermore, some of them either have, or are supposed to have, certain peculiarities of form which can be accounted for only by considering them derivatives, and that of a very peculiar kind. s. 329. It is an ethnological fact, that the numerals are essentially the same throughout the whole Indo-European class of languages. The English _three_ is the Latin _tres_, the Sanskrit _tri_, &c. In the Indo-European languages the numerals agree, even when many common terms differ. And it is also an ethnological fact, that in a great many other groups of languages the numerals differ, even when many of the common terms agree. This is the case with many of the African and American dialects. Languages alike in the common terms for common objects differ in respect to the numerals. What is the reason for this inconsistency in the similarity or dissimilarity of the numerals as compared with the similarity or dissimilarity of other words? I believe that the following distinction leads the way to it:-- The word _two_=2, absolutely and unequivocally, and in a primary manner. The word _pair_ also=2; but not absolutely, not unequivocally, and only in a secondary manner. {274} Hence the distinction between absolute terms expressive of number, and secondary terms expressive of number. When languages separate from a common stock before the use of certain words is fixed as _absolute_, there is room for considerable latitude in the choice of numerals; _e.g._, whilst with one tribe the word _pair_=_two_, another tribe may use the word _couple_, a third _brace_, and so on. In this case dialects that agree in other respects may differ in respect to their numerals. When, on the other hand, languages separate from a common stock after the meaning of such a word as _two_ has been fixed absolutely, there is no room for latitude; and the numerals agree where the remainder of the language differs. 1. _One_=_unus_, Latin; [Greek: heis] ([Greek: hen]), Greek. 2. _Two_=_duo_, [Greek: duo]. 3. _Three_=_tres_, [Greek: treis]. 4. _Four_=_quatuor_, [Greek: tettara]. This is apparently problematical. Nevertheless, the assumed changes can be verified by the following forms:-- [alpha]. _Fidvor_, Moeso-Gothic. To be compared with _quatuor_. [beta]. [Greek: Pisures], Aeolic. Illustrates the change between [tau]- and [pi]- (allied to _f-_), within the pale of the classical languages. 5. _Five_=_quinque_, [Greek: pente]. Verified by the following forms:-- [alpha]. [Greek: Pempe], Aeolic Greek. [beta]. _Pump_, Welsh. These account for the change from the _n_ + _t_ in [Greek: pente] to _m_ + _p_.