109. The suffixes employed in Homer to form adverbs are as follows.
-θι expresses the place where: the chief instances are
From pronouns and prepositions.
τό-θι, ὅ-θι, πό-θι, αὖ-θι, αὐτό-θι, κεῖ-θι (ἐκεῖ-θι only Od. 17.10)
ἑτέρω-θι, ἑκάστο-θι, ἄλλο-θι, ἔκτο-θι, ἔνδο-θι, ἀπό-προ-θι, ὑψό-θι, ἐγγύ-θι
From nouns.
νειό-θι, θύρη-θι (Od. 14.352)
οἴκο-θι, ἡῶ-θι, οὐρανόθι, κηρόθι
Ἰλιόθι, Κορινθό- θι, Ἀβυδό-θι
Note that ἐκεῖ is not fond in Homer.
-θᾶ place
ἔν-θα, ἐνταῦ-θα, ὕπαι-θα (cp. also δηθά, μίνυνθα)
-θε(ν) place, from prepositions
πρόσ-θε(ν), ὄπισ-θε(ν) and ὄπι- θε(ν), ὕπερ-θε(ν), πάροι-θε(ν), ἔνερ-θε(ν)
-θεν place whence, used with nearly the same stems as -θι
ὅ-θεν, πό-θεν, ἔν-θεν, κεῖ-θεν, ἄλλο-θεν, ὑψό-θεν, πάντο-θεν, ἀμφοτέρω-θεν, ἑτέρω-θεν
From nouns.
ἠῶ-θεν, Διό-θεν (Il.), οὐρανό-θεν, ἰππό-θεν, etc.
This suffix is often used with the prepositions ἐξ and ἀπό, as ἐκ Διό-θεν, ἀπʼ οὐρανό-θεν, etc. With the stems ἐμε, σε, ἑ, it forms a genitive.
Il. 1.280 σέθεν δʼ ἐγὼ οὐκ ἀλεγίζω
The form ἕθεν is only found in the Iliad.
-θοῖ, only in ἐνταυ-θοῖ there (Od.).
-τος place
ἐν-τός, ἐκ-τός
Originally, perhaps, it expressed the place whence, as Lat. caeli-tus, divini-tus.
-τις, in αὖ-τις back, again (Attic αὖ-θις).
-σε place whither
πό-σε, ὁππό-σε, κεῖ-σε, ἑτέρω-σε, ἀμφοτέρω-σε, ὁμό-σε
From nouns.
πάντο-σε, κυκλό-σε
-φι(ν), -φις, in νόσ-φι(ν) apart, λικρι-φίς sideways (Il. 14.463). This may be the instrumental ending -φι(ν).
-φᾶ, in μέσ-φα until, lit. meanwhile (Il. 8.508).
-xι, in ᾗ-χι where (lit. which way, Lat. quā).
-xα, with numerals
δί-χα (two ways), τρί-χα, πέντα-χα, ἕπτα-χα.
-χθα, in the same sense, τρι-χθά, τετραχθά.
-κις, -κι; with numerals
δεκά-κις, τετρά-κις, εἰνά-κις, εἰκοσά-κις
and with similar meaning πολλάκις and πολλάκι, ὁσσάκι, τοσσάκι.
The original suffix is -κιs or -κι (not -ᾰκις), but in consequence of its having been used at first with stems ending in -ᾰ (τετρᾰ-, ἑπτᾰ-, δεκᾰ-, εἰνᾰ-), the combination -α-κις came to be felt as the suffix, and was extended to other words by analogy. Α similar explanation applies to the ᾰ of πέντα-χα.
-κας expresses manner; ἀνδρα-κάς = Lat. viritim.
-δε place whither, suffixed to the accusative
οἶκόν-δε, πόλε-μονδε, ἅλαδε. This suffix is peculiar in being an enclitic; in strictness we should write οἶκόν δὲ, πόλεμόν δὲ, etc.
-δίς expresses direction or manner
χαμά-δις, ἄμυ-δις, ἄλλυ-δις, ἐπαμοιβα-δίς (Od. 5.481).
110. Case forms as Adverbs. The suffixes which fοllοw have been explained, with more or less probability, as case endings.
-α manner
ἄρ-α (lit. fittingly), ἄμ-α, μάλ-α, θάμα, τάχ-α, σάφ-α, κάρτ-α, ῥεῖ-α or ῥέ-α, ὦκ-α, ἦκ-α, αἶψ-α, λίγ-α, σῖγα, ῥίμφ-α, πύκ-α, λίπ-α
in Attic κρύφ-α, ρέμ-α.
The adverbs in -ᾰ belong to an early stage of Greek, most of them being confined to Homer. They have generally been taken to be primitive instrumental forms (so Brugmann, M. U. ii. 158, G. G. § 83). It is a question, however, whether the original instrumental ending was -ᾰ or -ε: see Joh. Schmmidt, K. Z. xvi. 292. Those which answer to adjectives in -ύ-s, viz. τάχα, ὦκα, λίγα, κάρτα, θάμα, are explained by Joh. Schmidt as older neuter plural forms (ταχϝ-α, etc.), cp. αἰπά neuter plural of αἰπύ-ς, and πρέσβᾰ (for πρεσβϝ-ᾰ ?) feminine of πρέσβυ-ς. This will not apply to ἄρα, μάλα (since ἀρ-ϝα, μαλ-ϝα would give ᾱρα, μᾱλα). Some may be stems in -n̥, like μέγα: cp. λίγα and λιγαίνα (n̥ι̯ω), λίπα and λιπαίνω, πύκα and πυκν-ός, also the stems κρέα-, γερα- (§ 105.4).
-ῃ or -η way, direction
ᾗ, τῇ, πῇ, ὅπῃ (or πῆ, ὅπη), πάντ-η, λάθρῃ
These forms represent the instrumental of the way by which (Lat., quā, etc.).
It is a question whether they should be written with iota subscript or not. The ancient grammarians prescribed iοtα (Apoll. de Adv. 625.1), and this is confirmed by the forms ᾇ, ὁπᾷ, ἀλλᾷ, παντᾷ on Doric inscriptions (Ahrens, ii. 369). In Homer however the final vowel of πάντῃ (or -η) is frequently shortened before another vowel, which is rarely done in the case of final -ῃ (§ 380).[fn]The question between πάντῃ and πάντη cannot be decided, as Joh. Schmidt supposes (Pluralb. p. 40), by the circumstance that the final vowel is frequently shortened before another vowel in Homer. It is true as was observed by Hoffman (Quaest. Hom. i. p. 58, quoted by Schmidt l. c.) that final η is oftener shortened than final ῃ. In the first four books of the Iliad and Odyssey, as Hartel shows (Hom. Stud. ii. p. 5) -η is shortened 41 times, -ῃ 19 times, and further examination confirms this ratio. But, as Hartel also points out, -η occurs in Homer about three times as often as -ῃ, consequently the shortening of -ῃ is relatively more frequent.[/fn] It is not unlikely therefore that the original instrumental feminine -η took iοta subscript from the analogy of the dative feminine in -ῃ. There were also Doric adverbs of place in -η or -η (πή ποκα, ἑκατερῇ, see Ahrens, ii. 362, Brugmann, M. V. ii. 244), in which η is of course pan-Hellenic; but Ionic πῇ, etc., are connected by the meaning with the Doric forms in -ᾳ. Cp. also λάθρη (-ῃ) with Attic λάθρᾱ (or -ᾳ). The form πάντ-η is an extension of the ending -η to the consonantal declension (as with the adverbs in -ωs).
-ει, -ι time, manner
αὐτο-νυχ-εί (or -ῑ) that very night, Il. 8.197
τρι-στοιχ-ί in three rows
ἀναιμωτ-ί (ῑ) blοοdlessly
ἀμογητ-ί, ἀμαχητί, ἀνουτητί, ἀνιδρωτί, ἀνωϊστί, ἐγρηγορτί
with ῐ
ἕκητι with the will
ἀέκητ-ι without the will
μελεϊστ-ί limb by limb
μεγαλωστί in mighty fashion
Short -ῐ is certain in ἔκητι, ἀέκητι, μελεϊστί, μεγαλωστί, and is not excluded by the meter in ἀμογητί and ἀμαχητί. Where the syllable is long the MSS. are usually divided between -ει and -ι. The evidence of inscriptions is strongly in favour of -ει (H. W. Smyth, The reductiοn of ει to ι in Hοmer, p. 10): but -ῑ can hardly be due to mere itacism, and we have further to explain the forms in -ῐ. The generally accepted view is that -ει is the original locative ending of the ο-declension, which is preserved in the Doric adverbs εἷ, πεῖ, τουτεῖ, τηνεῖ, etc., also in οἴκει (Menander fr. 456). On this view short ῐ must be the corresponding ending of the consonantal declension, and the analogy of forms of that declension must have been extended so as to create a new adverbial ending -τῐ (cp. ἐγερτῐ́ in Soph.). The -ῑ of ἀναιμαωτί, etc., if not a mere error, may be due to contamination between -ει and -ῐ.
αἰεί has been taken to be a locative from the stem αἰϝεσ- (of which the Doric αἰές is the accusative). Mr. H. W. Smyth (l. c.) justly objects to this that the Homeric form would be αἰϝέϊ: and this form, we may add, wοuuld become αἰεῖ, not αἰεί. Hence he derives it from the stem αἰϝο-, Latin aevο-m.
A different account of the adverbs in -ει and -ι is given by Mahlοw (Die langen Vοcale, p. 121). Νοticing that they are mainly compoυnds, especially with ᾰ priv., he compares the numerous Latin adjectives such as ex-animi-s, in-ermi-s, im-belli-s, and shows that change to an I-stem is found in similar words in other European languages. This I-stem in the accusative neuter gives the adverbs in -ῐ, in the locative those in -ει or -ῑ. On this view the doubt between -ει and -ῑ is the same that we meet with in the dative of nouns in -ι-ς (§ 98).
-ως manner; a suffix of which there are comparatively few examples in Homer; the commonest are from stems in -ο.
τῶς, ὥς, πῶς, οὕτ-ως (also οὕτ-ω), ὁμ-ῶς, φίλ-ως, αἰνῶς, καρπαλίμως, ἀσπασίως, ῥηϊδίως, ἐκπάγλως, κρατερῶς, μεγάλως (rare)
From other stems
ἀφραδέ-ως, περιφραδέ-ως
-ω, chiefly from prepositions
εἴσ-ω, ἔξ-ω, πρόσσ-ω, ὀπίσσ-ω, ἄν-ω, κάτ-ω, προτέρ-ω (further on), ἑκαστέρ-ω, ἑκαστάτ-ω (farther, farthest), ἀσσοτέρ-ω (nearer)
Two others are adverbs of manner, ὧ-δε, οὕτ-ω (for which οὕτως is only written when a vowel follows in the same sentence).
The ending -ωs has long been considered to be the Greek form of the original ablatival -ōt (Lat. -ōd) of ο-stems in Greek, however, a final -d would disappear (as in ἄλλο, Lat. aliu-d, etc.) and consequently the theory applies only to the forms without -ς, viz. ὧδε and οὕτω. The difficulty was met by Curtius (Curt. Stud. x. 219) with the suggestion that -τ would pass into -ς before a dental or σ: e. g. οὕτος σοί, οὕτος τίθημι for οὕτωτ σοί, οὕτωτ τίθημι. When two forms οὕτω and οὕτως had thus come into existence as "sentence-doublets" (like οὐ and οὐκ, ἐξ and ἐκ), it wοuld be natural to use οὕτως when it served to prevent hiatus, and the more regular οὕτω in other cases. This explanation was rejected by later scholars (as Brugmann and G. Meyer), and is certainly not quite satisfactory. If Curtius is right we should expect ὧτ δέ to become ὧσδε rather than ὧδε, His view is however defended by Joh. Schmidt (Pluralb. p. 352.)
The ending -ω in ἄν-ω, etc., may be either the ablatival -ōt, or (more probably) an instrumental ending -ō (Mahlow, Die langen Vοcale, p. 86). In Latin, as Mahlow shows, it is probable that the instrumental is represented by the adverbs in -ŏ, as modo, citο, the ablative by archaic -ōd, later -ō. If -ωs and -ω were alternative ablative endings (sentence-doublets) it seems possible that the adoption of -ωs rather than -ω in the adverbs of manner was partly determined by the circumstance that -ω was already familiar in the instrumental use.
The extension of -ωs, -ω to the consonantal declension presents no difficulty. It may be observed, perhaps, that the proper ablative of that declension was unsuited for adverbial use, because it was the same in form as the genitive: e.g. ταχέος was already = of a swift, and accordingly a new wοrd ταχέως swiftly was coined on the model of φίλως, etc.[fn]As adverbs of the genitive absolute form (ταχέος, etc.) must have existed at one time alongside of those in -ωτ from ο-stems, the conjecture may be hazarded that this adverbial -ος was one of the influences which determined the choice of -ως rather than -ω for original -ōt. If so, such a form as πάντ-ως is a sort of contamination of the genitive absolute παντ-ός and the forms in -ω(s).[/fn]
-ου place: ποῦ, ὁμοῦ, ἀγχοῦ, τηλοῦ, ὑψοῦ, αὐτοῦ, -all perispomena. They are the same in meaning as the corresponding adverbs in -ὅθι.
-δὸν, -δην, -δα, forming adverbs of manner, are evidently accusatives from stems in -δο-, -δη- (§ 114).
σχε-δόν nearly, lit. holding-wise
ἀποστα-δόν aloof
ἐμβα-δόν on fοοt
ἀμφα-δόν openly
ἰλα-δόν in crowds
So βοτρυ-δόν, πυργη-δόν, ῥυδόν, συνωχαδόν, etc.; βά-δην (steppingly), τμήδην, κρύβ-δην, κλήδην, ἐπιγράβδην, etc. (all from verbs), also a peculiar group in -ά-δην.
ἐπιστροφά-δην (wheeling about), προτροπά-δην (headlong), ἐπιτροχά-δην, μεταδρομά-δην, ἀμβολά-δην; μίγ-δα, κρύβ-δα, ἀποσταδά, ἀμφα-δά, ἀναφαν-δά, αὐτοσχε-δά.
It is evident that these are much more numerous than the noun stems in -δο, -δη can ever have been. In such cases we have to explain, not the derivation of the individual forms, but the origin of the type.
Other adverbs obtained from accusatives are
ἄκην in silence
ἄντην (ἀντίον, ἐναντίον, etc.) opposite
πάλιν backwards
δηρόν lοng
σχεδίην hand to hand
ἀμφαδίην οpenly
ἀπριάτην withοut purchase
perhaps also ἄγχι (near), ὕψι (aloft), ἰφι (mightily). The form ἶφι is generally taken as the instrumental of ἴ-ς force (§ 104): but this does not explain how it comes to be used as a stem in the adjective ἴφι-α (μῆλα). as well as in compounds, Ἰφι-άνασσα, etc., (Bekker, H. B. i. 160).
Many adverbs are formed with a final -ς, which is liable to be lost before a word beginning with a consonant, as οὕτω(ς) and the adverbs in -κι(ς) already mentioned; other Homeric instances are
ἄχρι(ς) and μέχρι(ς) until
ἰθύ(ς) straight towards
μεσσηγύ(ς) between
ἀτρέμα(ς) quietly
also the preposition ἀμφί, adverb ἀμφίς, and Homeric ἀντικρύ, later ἀντικρύς. Similar adverbs in which -ς is not lost are
ἅλι-ς, μόγι-ς, χωρί-ς
ἀγκάς, ἑκά-ς, πέλα-ς, ἐντυπάς (Il. 24.163)
ἐγγύ-ς; χθέ-ς
and those in -δι-ς, as ἄλλυδις, ἀμοιβηδίς. Note also the group formed by -ς subjoined to a monosyllabic verbal stem.
πύξ with the fist
ἐπί-μιξ in confusion
ἅ-παξ once
μάψ idly
ὀ-δάξ with the teeth (δάκ-νω)
The nature of this -ς is obscure. Brugmann (K. Z. xxiv. 74) connects it with the -ς of the prepositions ἐξ, ἄψ, ἀμφί-ς, holding that it is ablatival. Jοh. Schmidt (Pluralb. 357) supposes a group of neuter stems, like the nouns in -ας, -ες, etc.