Sanskrit is tara, the Greek teros. This might, at first sight, be taken for a demonstrative element, but it is in reality the root tar, which means to go beyond, which we have likewise in the Latin trans. This trans in its French form très is prefixed to adjectives in order to express a higher or transcendent degree, and the same root was well adapted to form the comparative in the ancient Aryan tongues. This root must likewise be admitted in one of the terminations of the locative which is tra in Sanskrit; for instance from ta, a demonstrative root, we form ta-tra, there, originally this way; we form anyatra, in another way; the same as in Latin we say ali-ter, from aliud; compounds no more surprising than the French autrement (see p. 55) and the English otherwise.

Most of the terminations of declension and conjugation are demonstrative roots, and the s, for instance, of the third person singular, he loves, can be proved to have been originally the demonstrative pronoun of the third person. It was originally not 8 but t. This will require some explanation. The termination of the third person singular of the present is ti in Sanskrit. Thus da, to give, becomes dadâti, he gives; dha, to place, dadhati, he places.

In Greek this ti is changed into si; just as the Sanskrit tvam, the Latin tu, thou, appears in Greek as sy. Thus Greek didōsi corresponds to Sanskrit dadâti; tithesi to dadhati. In the course of time, however, every Greek 8 between two vowels, in a termination, was elided. Thus genos does not form the genitive genesos, like the Latin genus, genesis or generis, but geneos = genous. The dative is not genesi (the Latin generi), but geneïgenei. In the same manner all the

Thus

regular verbs have ei for the termination of the third person singular. But this ei stands for esi. typtei stands for typtesi, and this for typteti.

The Latin drops the final i, and instead of ti has t. Thus we get amat, dicit.

Now there is a law to which I alluded before, which is called Grimm's Law. According to it every tenuis in Latin is in Gothic represented by its corresponding aspirate. Hence, instead of t, we should expect in Gothic th; and so we find indeed in Gothic habaip, instead of Latin habet. This aspirate likewise appears in Anglo-Saxon, where he loves is lufað. It is preserved in the Biblical he loveth, and it is only in modern English that it gradually sank to 8. In the 8 of he loves, therefore, we have a demonstrative root, added to the predicative root love, and this 8 is originally the same as the Sanskrit ti. This ti again must be traced back to the demonstrative root ta, this or there; which exists in the Sanskrit demonstrative pronoun tad, the Greek to, the Gothic thata, the English that; and which in Latin we can trace in talis, tantus, tunc, tam, and even in tamen, an old locative in men. We have thus seen that what we call the third person singular of the present is in reality a simple compound of a predicative root with a demonstrative root. It is a compound like any other, only that the second part is not predicative, but simply demonstrative. As in pay-master we predicate pay of master, meaning a person whose office it is to pay, so in dada-ti, give-he, the ancient framers of language simply predicated giving of some third person, and this synthetic proposition, give-he, is the same as what we now call the third person singular in the

Sanskrit is tara, the Greek teros. This might, at first sight, be taken for a demonstrative element, but it is in reality the root tar, which means to go beyond, which we have likewise in the Latin trans. This trans in its French form très is prefixed to adjectives in order to express a higher or transcendent degree, and the same root was well adapted to form the comparative in the ancient Aryan tongues. This root must likewise be admitted in one of the terminations of the locative which is tra in Sanskrit; for instance from ta, a demonstrative root, we form ta-tra, there, originally this way; we form anyatra, in another way; the same as in Latin we say ali-ter, from aliud; compounds no more surprising than the French autrement (see p. 55) and the English otherwise.

Most of the terminations of declension and conjugation are demonstrative roots, and the s, for instance, of the third person singular, he loves, can be proved to have been originally the demonstrative pronoun of the third person. It was originally not s but t. This will require some explanation. The termination of the third person singular of the present is ti in Sanskrit. Thus dâ, to give, becomes dadâti, he gives; dha, to place, dadhati, he places.

In Greek this ti is changed into si; just as the Sanskrit tvam, the Latin tu, thou, appears in Greek as sy. Thus Greek didōsi corresponds to Sanskrit dadâti; tithesi to dadhati. In the course of time, however, every Greek 8 between two vowels, in a termination, was elided. Thus genos does not form the genitive genesos, like the Latin genus, genesis or generis, but geneos = genous. The dative is not genesi (the Latin generi), but geneïgenei. In the same manner all the

regular verbs have ei for the termination of the third person singular. But this ei stands for esi. Thus typtei stands for typtesi, and this for typteti.

The Latin drops the final i, and instead of ti has t. Thus we get amat, dicit.

Now there is a law to which I alluded before, which is called Grimm's Law. According to it every tenuis in Latin is in Gothic represented by its corresponding aspirate. Hence, instead of t, we should expect in Gothic th; and so we find indeed in Gothic habaip, instead of Latin habet. This aspirate likewise appears in Anglo-Saxon, where he loves is lufað. It is preserved in the Biblical he loveth, and it is only in modern English that it gradually sank to 8. In the s of he loves, therefore, we have a demonstrative root, added to the predicative root love, and this 8 is originally the same as the Sanskrit ti. This ti again must be traced back to the demonstrative root ta, this or there; which exists. in the Sanskrit demonstrative pronoun tad, the Greek to, the Gothic thata, the English that; and which in Latin we can trace in talis, tantus, tunc, tam, and even in tamen, an old locative in men. We have thus seen that what we call the third person singular of the present is in reality a simple compound of a predicative root with a demonstrative root. It is a compound like any other, only that the second part is not predicative, but simply demonstrative. As in pay-master we predicate pay of master, meaning a person whose office it is to pay, so in dada-ti, give-he, the ancient framers of language simply predicated giving of some third person, and this synthetic proposition, give-he, is the same as what we now call the third person singular in the

indicative mood, of the present tense, in the active voice.1

We have necessarily confined ourselves in our analysis of language to that family of languages to which our own tongue, and those with which we are best acquainted, belong; but what applies to Sanskrit and the Aryan family applies to the whole realm of human speech. Every language, without a single exception, that has as yet been cast into the crucible of comparative grammar, has been found to contain these two substantial elements, predicative and demonstrative roots. In the Semitic family these two constituent elements are even more palpable than in Sanskrit and Greek. Even before the discovery of Sanskrit, and the rise of comparative philology, Semitic scholars had successfully traced back the whole dictionary of Hebrew and Arabic to a small number of roots, and as every root in these languages consists of three consonants, the Semitic languages have sometimes been called by the name of triliteral.

To a still higher degree the constituent elements are, as it were, on the very surface in the Turanian family of speech. It is one of the characteristic features of that family, that, whatever the number of prefixes and suffixes, the root must always stand out in full relief, and must never be allowed to suffer by its contact with derivative elements.

There is one language, the Chinese, in which no analysis of any kind is required for the discovery of its component parts. It is a language in which no coales

1 Each verb in Greek, if conjugated through all its voices, tenses, moods, and persons, yields, together with its participles, about 1300 forms.