character than any others which exist. Their most fundamental characteristic is the triliterality of their roots. With rare and insignificant exceptions, every Semitic verbal root—the pronominal roots are not subject to the same law—contains just three consonants, no more and no less. Moreover, it is composed of consonants alone. That is to say: whereas, in the Indo-European and other tongues, the radical vowel is as essential a part of the root as any other, even though more liable than the consonants to phonetic alteration, in the Semitic, on the other hand, the vocalization of the radical consonants is almost solely a means of grammatical flexion. Only the consonants of the root are radical or significant elements; the vowels are formative or relational. Thus, for example, the three consonants q-t-l form a root (Arabic) which conveys the idea of 'killing;' then qatala means 'he killed;' qutila, 'he was killed;' qutilū, 'they were killed;' uqtul, 'kill;' qātil, 'killing;' iqtāl, 'causing to kill;' qatl, 'murder;' qitl, 'enemy;' qutl, 'murderous;' and so on. Along with this internal flection is found the use of external formative elements, both suffixes and prefixes, and also, to a limited extent, infixes, or inserted letters or syllables; yet they are but little relied on, and play only a subordinate part, as compared with their analogues in the languages of other races; the main portion of the needed inflection is provided for by means of the varying vocalization of the root, and what remains for affixes to do is comparatively trifling. The aggregation of affix upon affix, the formation of derivative from derivative, so usual with us (it was illustrated in a former lecture by such examples as inapplicabilities and untruthfully), is a thing almost unknown in the domain of Semitic speech. This truly Procrustean uniformity of the Semitic roots, and this capacity of significant internal change, separate the languages to which they belong by a wide and almost impassable gulf from all others spoken by the human race. So far as we can discover, the varying vocalization of the roots in these languages is an ultimate fact, and directly and organically indicative of a variation of meaning: it is not, like the occasional phenomena of a somewhat similar char-
