Where can we observe the first steps that rise above this simple perception? I say, as I have always said, In language and in language alone. Language is the oldest monument which we possess of man’s mental power, older than stone weapons, than cuneiform inscriptions, than hieroglyphics. The development of language is continuous, for where this continuity is broken, language dies. After every Tasmanian had been killed or had died, the Tasmanian language _ipso facto_ ceased; and even if any literary remains had survived, the language itself would have to be reckoned, like Latin and Greek, with dead languages. Thousands of them may have disappeared from the earth; in its development a language may have changed as much as Sanskrit to Bengali; but it suffers no break, it remains always the same, and in a certain sense we still speak in German the same tongue as was spoken by the Aryans before there was a Sanskrit, a Greek, or a Latin language. Consider what this signifies. Chronologically, we cannot get at this primitive Aryan speech. Let us assume that Sanskrit, Greek, and Latin were spoken as independent national tongues at least fifteen hundred years before our chronology—what an age had elapsed before these three, as well as the remaining Aryan tongues, could have diverged so much as Sanskrit diverges from Greek and Greek from Latin. The numerals are the same in these three languages, and yet _katvāras_ sounds quite differently from τέσσαρες and _quatuor_ and our four. The words for eight, _octo_ in Latin, ὀκτώ in Greek, and _ashtau_ in Sanskrit, are nearly identical; and it is even possible that the lesser deviations in the pronunciation of these words demanded no great interval of time. But now let us consider what lies behind these ten numerals. There is the elaboration of a decimal system from 1 to 10, no, to 100 (ἑκατόν), Sanskrit _satám_, _centum_. There is the formation and fixing of names for these numbers, which must have been originally more or less arbitrary, because numbers only subordinate themselves with difficulty to one of those general ideas which are expressed in the Aryan roots. Besides these words are, even in their oldest attainable forms, already so weather-beaten, that in most cases it is impossible even to guess their etymology and original meaning. We see that the names for two and eight are dual, while those for three and four clearly have plural endings. But why eight in the primitive Aryan was a dual, and what were the two tetrads, which, combined in _asht-au_, _oct-o_, ὀκτ-ώ, expressed the number eight, will probably never be discovered. It is possible that _asht-i_ was a name for the four phases of the moon, or for the four fingers of the hand without the thumb. Analogies occur in other families of language, but certainty is beyond our reach. If we now consider what mental effort is necessary to work out a decimal system, and to secure general recognition and value for the name given to each number, we shall