of, appended to an adverb in -15. The adverb in us, by the side of which we frequently find an adverb in -a (cf. μέγα, μόγις, μάλα, μόλις, &c.), seems to be an abbreviated inflexion of adjectives in -v- or -ɛv-. The v is turned into , which probably always followed it in the complete termination -vis or -hvis before the v was vocalized; thus the Sanscrit root seid is represented both by ud-wo and id-oos; hve, the root of υἱός, appears in φύ ω and filius, ἐμοί = ἐμε-φί, appears as ἔμυ in Baotian Greek (Apollon. de Pronom. p. 364), 'Oïhɛús- Filɛús appears as 'Theús or Olleús (Hermann, de Emend. Gr. Gr. 415), &c. At any rate all comparisons in tv and -totos seem to be ultimately referable to positives in -vs, -ɛvs, or to adverbs in -ig. Although these forms in -vs or vis do not exist, they may be presumed as possible from the analogies pointed out by Buttmann (Ausführl. Sprl. § 67, Anm. 6): namely, by the side of μuax-pós, alox-çós, and vex-pós, we have the forms unx-v-vo (in which the root-syllable is guna'd as in the Sanscrit superlatives, and in μήκιστος, μήκος), αἰσχύνη, and νέκυς. Similarly, καλλίων, κάλλιστος, though immediately connected with καλός = καδ 2ós, most probably refer to a by-form xaλ-λús; cf. xah-λú-vo after the analogy of πολύς, πολλός, πραύς, πρᾶος, &c. We do not, however, agree with Buttmann in regarding the termination -vs and -og as originally the same. On the contrary, the terminations in -vg, as will hereafter be shown, contain a distinct pronominal addition to the crude form of the word, while in the ending -og the nominative suffix is immediately attached. It is no slight proof of the durability of these suffixes, that Sanscrit words in us or -u are always represented in Greek by corresponding adjectives in -vg or substantives in -v: thus for the Sanscrit svâdus, prithus, urus, âçus, laghus, mridus and madhu, we have the Greek ἡδύς, πλατύς, ευρύς, ὠκύς, ἐ-λαχύς, βραδύς and Mεdú. The Sanscrit bahus has two Greek representatives, Badus and naxús, just as ahis stands between Exis and opis, or han between gav (φόν-ος,ἐπέφν-ον) and θαν (θάνατος, θνήσκω, Benary, Berl. Jahrb. August, 1834, p. 229). There are cases, in which the superlative termination is -06-TOS - -ως-τος, as in πολλ-οσ-τός, for which we might have had πόλιστος; as may be seen by a comparison of ὀλιγοστός with ὀλίγιστος. For the form πολύς admits of an adverb in -ts, as πολλός does of an adverb in -ws, and perhaps, after all, the difficult word tozódotos in Soph. Antig. 857, may contain a remnant of this other superlative of πολύς. The name' Οδυσσεύς= Ολυσσεύς οι Ολισσεύς points to an original oyεús, which would fully account for ólíɣotos (see Kenrick, Herod. p. 281, and Varron. p. 142). The terminations -Tegos, -Tatos, are sometimes appended to adverbs in -s from by-forms in -ɛús, as in λαλίστερος, πτωχίσ-τερος, ἁρπαγέστατος, &c. We must not confuse this with the fact that adjectives, in -ns or -nts and -¿vτs, and even

some in-wov (-ov-s), -ığ (-ıx-s), regularly form the comparative and superlative in 6-Tegos, -έo-tatos; for in these cases we have remains of the adverb in -ws. Nor does the apparent analogy of magis favour the derivation of μέγιστος from an imaginary comparative μέγις for usyiov; the comparative of mag-nus has suffered all sorts of abbreviations; in major the gi has become j as usual, while magis has lost the u of the ending: the g has become z in the Zend mazista, and has vanished altogether from the French mais. There is no reason whatever why there should not be a form μεγε-ύς, as well as μέγεθος. Οι the contrary, the existence of μério-tog is a reason for presuming the existence of μεγεύς, just as we might presume the existence of "Αρευς, if we did not know it, from the superlative quo-tos, and as a positive vasus has been inferred from the Zend superlative vahista (Benary, Berl. Jahrb. August, 1843, p. 230). In fact, the adverb uós, as we have already mentioned, is itself an argument in favour of the same supposition. We find ἑκάτερος by the side of ἕκαστος, because, as we shall see, exa is an adverb as well as exaç. So also some adjectives in -vs add the terminations -Tɛoos, -tatos to their neuter used adverbially, as well as employing the other adverbial form in -ɩs. Cf. yâvκύτατος with γλύκιο-τος.

CHAPTER III.

THE PREPOSITION.

168 Independent value of prepositions as positional words. 169 General view of the Greek prepositions according to the elements of which they are composed. 170 Eis-v-s and iv. Apparent use of og as a preposition. 171 Пgó, no̟ó-s, προ-τί, ἀντί, ἄντα. Expression of space and time respectively by the same prepositions. 172 'Eni and dupí identified with abhi and ob. 173 Their agreements in meaning with each other. 174 The use of iní in composition to signify mutuality or interchange, also points to its affinity with dugi. 175 The same appears also from its use with the dative to signify combination or coexistence. 176 Etymology of ¿§. Use of Ex&výoxw and Expέow. 177 Από and παρά. Connexion of the latter with περί, πρός, pre, prater, and per. 178 Пsoi and the Sanscrit pari, para. 179 Ynó and vπéo. 180. Stá and δυς-. 181 Σύν and μετά. 182 Κατά, another form of ἄντα. 183 Ανά and natά are properly correlatives, and are most satisfactorily examined together. 183 'Avα, άv-, ά-, and άño- as negative prefixes; xatά as an affirmative. 185 The intensive 2pa. 186"Av and név related to άvá and zará. 187 Position of av and xév in the sentence.

168 PREPOSITIONS are pronouns or positional words in the strictest sense of the term*. They express relations of place, and in their ordinary use are employed to denote the relative positions of visible objects. Grammarians tell us that they govern cases, and it is the prevailing practice to arrange them according to the cases which they are said to govern. But this is palpably erroneous: for in all languages which have inflected nouns, a case may express by itself any relation which the addition of a preposition could give to it and,

* The preposition is fully discussed in the new edition of Pott's Etymological Researches, the first part of which has just appeared (Etymologische Forschungen auf dem Gebiete der Indo-Germanischen Sprachen; Zaveite Auflage in völlig neuer Umarbeitung; erster Theil: Präpositionen. Lemgo und Detmold, 1859). The whole of the volume (859 pages) is devoted to this subject. As usual, Pott exhibits an inexhaustible abundance of lexicographical details and illustrations, but he seems to be still unacquainted with the analysis of these pronominal forms, which was first expounded, twenty years ago, in the present work, and without which, as we conceive, no definite results can be obtained. In general we think he has rendered himself liable to the happy remark, which he made upon Benfey (Jahrb. d. Wiss. Krit. 1840, p. 629), that "it always excites a feeling of dissatisfaction, when the head of the nail is missed, though we see the hammer falling with busy haste on all sides of it."

in languages which, like the Sanscrit and the modern Russian, have a complete assortment of cases, many relations of place are invariably expressed by the cases without any particle prefixed. Such would have been the fact in the Greek language too, but the rules of euphony, convenience, the influence of writing, and a multitude of other causes, have contributed to mutilate the terminations of the nouns as well as of the verbs, and thus prepositions, the force of which was originally included in the case-endings, have come to be prefixed for the sake of greater distinctness, just as the particular noun is placed after the pronoun, called the article, in repetitions, and just as the nominative case is prefixed to the verb.

169 There are eighteen Greek words which are commonly reckoned as prepositions: six monosyllables, sis, ¿v, ¿¿, ño̟ó, πρός, σύν, and twelve dissyllables, ἀμφί, ἀνά, ἀντί, ἀπό, διά, ἐπί, κατά, μετά, παρά, περί, ὑπέρ, ὑπό. We shall consider these according to the relations which they express, and not according to any arbitrary division of former grammarians*. Since the prepositions retain their original meaning, as words indicating positions and directions in space, more characteristically than any other pronominal words, and also present the simplest combinations of the original elements of the pronouns, we will, previously to examining these Greek forms separately and in detail, endeavour to point out their etymological analysis in a summary manner, and to explain the general principles of their composition.

It has been stated before that the primitive pronouns are three in number, expressing respectively the positions here, near to the here, and there, and that different modifications of direction or position may be denoted by combining these original stems with one another or with the particle la or ra. On examination it will appear that all the Greek prepositions, with the exception of dia which is a form of the second numeral, are compounds of at least two of the primary elements, or of one of

* The object of this chapter is to discuss the origin and signification rather than the syntax of the Greek prepositions: as a supplement to the eases they are fully treated in the Greek Grammar, articles 470–488.

them with -ra. We have already adverted to the principles according to which we would arrange and classify all pronominal compounds (§ 130). After a careful dissection of all the pronominal forms with which we are acquainted, we have arrived at the conclusion, that if any one of the elements of position is combined with -ra, it indicates motion and continuation in a direction of which the element in question represents the point nearest to the subject; and that, by subjoining any one of the pronominal elements to any other of them, we denote a motion or continuation from the position denoted by the first element towards that indicated by the second. Thus we have seen, that the second element when prefixed to -ra (as in tva-ra) expresses motion onwards from the position indicated as near, so as, in fact, to coincide with a word indicating the third position (ta); and that the first element subjoined to the third (as in ta-ma) expresses motion or continuation from the third position towards the first, so as to coincide with the second position (cf. fini-timus, &c.). We shall find this method most amply illustrated by the Greek prepositions. Of these, na-oά, nɛ-Qi, л-оó, л-оó-я, are compounded of the first element and -pa. In the first, which is also written лa-qai, we find both elements in the simplest form. In the second, in which traces of a heavier ending still remain, the vowel of the first element has assumed the lighter form ɛ, according to a principle which will be more fully explained hereafter. In л-o̟ó and л-o̟ó-s, which are in fact one and the same word, another element has been subjoined in the 8, indicating motion or transitiveness, and probably a shortened form of the affix -os, -6to, which plays an important part as the sign of the genitive case. In consequence of this addition, the root-vowel has been dropt before the liquid, and a medium weight given to the vowel of the termination. The forms л-00-τí, лno-tí, also used for л-oó-s, are compounds, one of the preposition -ρó, the other of the simple element of the first pronoun, with the element of the second under the form Tis (§ 152); and both, therefore, denote (the former more strongly) motion from the first to the second position. A similar form is uɛ-tά, which is compounded of the first and third elements, and signifies "with" as a connexion between the here and the there, and "after" as denoting an approximation to their union.