Now the more specific the name the less extensive its application, and the less extensive its application the smaller the probability of its appearing in more languages than one. No one would expect the word _brother_ to occur in the Gaelic (_brathair_), and in the Latin (_frater_), if Gaels, Englishmen, and Romans, without any name for brother in general, had merely known an _elder brother_ by one separate single name, and a _younger one_ by another, as is really the fact in America. What we should look for in such a case would be the equivalents to words like _cadet_, and these might differ in languages otherwise allied. Names, then, for common objects are often of so specific a kind in the American languages, that they differ in cases where, if more general, they would agree. _The numerals._--Another class of words, which in many languages agree, differs in the American, viz., that of the numerals. In the Indo-European tongues these agree even where other words differ.[140] The converse, however, takes place with the tongues in question. Languages, alike in other points, shall count differently. Can this be explained? I submit the following doctrine, based upon the difference between absolute numerals like _two_ and _three_ (words which mean two units, and three units exclusively and irrespectively), and concrete numerals like _brace_ and _leash_. Between these two classes of words there is the following difference. Absolute numerals give no choice, concrete numerals do. Out of two tribes, wherein the intelligence of each is so little capable of generalization as not to have evolved abstract and absolute numerals like those of the Indo-European nations (_one_, _two_, &c.), the only way of counting is by the adoption of some material object in which the _number_ of its parts is a striking characteristic; in which case there is so much room for arbitrary selection that allied languages may take up different words. It is not to be supposed that unless the English, Greeks, Gaels, Slavonians, and the members of the Indo-European stock in general, had broken off from the common stem at a period subsequent to the evolution of absolute numerals that their names for the first ten units would be so like as they are. On the contrary, there would most certainly have been a difference; _two_ being expressed in one quarter by a word like _brace_, in another by such a term as _couple_, in a third by _pair_, and so on. Now this latitude exists and bears fruit with the American languages. One takes the name for (say) _two_ from one natural dualism, another from another--one calls it by the name for _a pair of hands_, another by that of _a pair of feet_, a third by that of _a pair of shoes_, &c.