362. The particles κεν and ἄν, as we have seen, are used to mark a predication as conditional, or made with reference to a particular or limited state of things, whereas τε shows that the meaning is general. Hence with the subjunctive and optative κεν or ἄν indicates that an event holds a definite place in the expected course of things. In other words, κεν or ἄν points to an actual occurrence in the future.1
κεν is commoner in Homer than ἄν. In the existing text κεν occurs about 630 times in the Iliad, and 520 times in the Odyssey; while ἄν (including ἢν and ἐπήν) occurs 192 times in the Iliad and 157 times in the Odyssey. Thus the proportion is more than 3:1, and is not materially different in the twο poems.
It is part οf Fick's well known theory that ἄν was unknown in the original Homeric dialect; a systematic attempt to restore the exclusive use of κεν in Homer has been made by a Dutch scholar, J. van Leeuwen2, who has proposed more or less satisfactory emendations of all the places in which ἄν now appears. It is impossible to deny the soundness of the principles on which he bases his enquiry. When the poems were chiefly known through oral recitation there must have been a constant tendency to modernize the language. With Attic and Ionic reciters that tendency must have led to ἄν creeping into the text, sometimes in place of κεν, sometimes where the pure subjunctive or optative was required by Homeric usage. Evidence of this kind of corruption has been preserved, as Van Leeuwen points out, in the variae lectiones of the ancient critics. Thus in Il. 1.168 ἐπεί κε κάμω is now read on the authority of Aristarchus; but ἐπὴν κεκάμω and ἐπήν κε κάμω were also ancient readings, and ἐπήν is found in all οur MSS. Similarly, in Il. 7.5 Aristarchus read ἐπεί κε κάμωσιν, and the MSS. are divided between ἐπεί κε and ἐπήν κε (or ἐπὴν κεκ.). There is a similar variation between the forms ἢν and εἴ κε (or αἴ κε) in the phrases αἴ κʼ ἐθέλῃσθα, αἱ κ ἐθέλῃσι, etc. Thus in Il. 4.353 (= 9.359) the MSS. nearly all have
ὄψεαι ἢν ἐθέλῃσθα καὶ αἴ κέν τοι τὰ μεμήλῃ
but αἴ κʼ ἐθέλῃσθα, which gives a better rhetorical elect, is found in
Il. 8.471 ὄψεαι αἰ κʼ ἐθέλησθα
(so all MSS., ἢν ἐθ. as a v. l. in A), also in Il. 13.260, 18.457, Od. 3.92, etc. Similarly, in
Il. 16.453 ἐπεὶ δὴ τόν γε λίπῃ
the v. l. ἐπήν is given by good MSS. (D, G, L, and as a variant in A). And the line
Il. 11.797 Μυρμιδόνων, αἴ κέν τι φόως Δαναοῖσι γένηαι·
is repeated in Il. 16.39 with the variation ἢν ποῶ for αἴ κεν. In such cases we can see the intrusion of ἄν actually in process.
Again, the omission of ἄν may be required by the meter, or by the indefinite character of the sentence (§ 283) e.g. in
Il. 15.209 ὁππότʼ ἂν ἰσόμορον ἐθέλησι
both these reasons point to ὁππότε ϝἰσόμορον κτλ. So in
Il. 2.228 εὖτ’ ἂν πτολίεθρον ἕλωμεν
read εὖτε πτ., and in
Od. 11.17 οὔθʼ ὁπότʼ ἂν στείχῃσι
read οὔθʼ ὁπότε (ὅτε κε, which Van Leeuwen proposes in these two places, is not admissible, since the reference is general).
Several reasons combine to make it probable that the forms ἢν and ἐπὴν are post-Homeric. The contraction oφ εἰ ἄν, ἐπεὶ ἄν is contrary to Homeric analogies (§ 378*), and could hardly have taken place until ἄν became much commoner than it is in Homer. Again, the usage with regard to the order of the particles excludes the combinations ἢν δέ, ἤν περ, ἢν γάρ—for which Homer would have εἰ δʼ ἄν, εἴ περ ἄν, εἰ γὰρ ἄν (§ 365). Again, ἤν cannot properly be used in a general statement or simile, and whenever it is so used the meter allows it to be changed into εἰ.
Il. 1.166 ἀτὰρ ἤν ποτε δασμὸς ἵκηται
Od. 5.120 ἤν τίς τε φίλην ποιήσετʼ ἀκοίτην
(ἥ τίς τε in several MSS.)
Od. 11.159 ἢν μή τις ἔχῃ εὐεργέα νῆα
Od. 12.288 ἤν πως ἐξαπίνης ἔλθῃ
Il. 20.172 ἤν τινα πέφνῃ
(in a simile)
Similar arguments apply with even greater force to ἐπήν. Of the 48 instances there are 18 in general sentences, and several others (Il. 4.239, 16.95; Od. 3.45, 4.412, 5.348, 11.119, 15.36, 21.159) in which the reference to the future is so indefinite that ἐπεί with a pure subjunctive is admissible. It cannot be accidental that in these places, with one exception (Od. 11.192), ἐπήν is followed by a consonant, so that ἐπεί can be restored vwithout any metrical difficulty. On the other hand, in 13 places in which ἐπήν is followed by a vowel the reference is to a definite future event, and accordingly we may read ἐπεί κʼ. In the combination ἐπὴν δή, which occurs seven times, we should probably read ἐπεὶ δή, or in some places ἐπεί κεν (as in Od. 11.221). The form ἐπειδάν occurs once, in a simile (Il. 13.285); hence we should read ἐπεὶ δή (not ἐπεί κεν, as Bekker and Nauck, or αἴ κεν as Menrad).
The distinction between general statements and those which refer to an actual future occurrence has hardly been sufficiently attended to in the conjectures proposed by Van Leeuwen and others. Thus in
Od. 5.121 ἤν τίς τε φίλον ποιήσετʼ ἀκοίτην
(in a general reflection)
Van Leeuwen would read αἴ κέν τίς τε, and in
Od. 12.288 ἤν πως ἐξαπίνης ἔλθῃ
he proposes αἴ κά που. So in Il. 6.489, Od. 8.553 ἐπὴν τὰ πρῶτα γένηται (of the lot of man) he bids us read ἐπεί κε. If any change is wanted beyond putting ἐπεί for ἐπήν, the most probable would be ἐπεί τε; see § 332. On the other hand he would put ἐπεί for ἐπὴν in such places as
Od. 1.293 αὐτὰρ ἐπὴν δὴ ταῦτα τελεντήσῃς τε καὶ ἔρξῃς (cp. Od. 5.353, 18.269)
where a definite future occasion is implied, and consequently ἐπεί κεν (which he reads in Od. 4.414) would be more Homeric. In Od. 6.262 αὐτὰρ ἐπὴν πόλιος ἐπιβήομεν we should perhaps read ἐπεί κε πόλεος (˘ˉ) : see § 94.2.
In a few places the true reading may be εἰ or ἐπεί with the optative
Od. 8.511 αἶσα γὰρ ἦν ἀπολέσθαι, ἐπὴν πόλις ἀμφικαλύψῃ
(ἐπεὶ . . . ἀμφικαλύαι, as in Il. 19.208 we should read ἐπεὶ τισαίμεθα)
Od. 21.237 (= 383) ἢν δέ τις . . . ἀκούσῃ μή τι θύραζε προβλώσκειν
(εἰ δέ τις . . . ἀκούσαι)
Il. 15.504, 17.245, 22.55 & 487.
The form ὅτʼ ἄν occurs in our text in 29 places, and in 22 of these the meter admits ὅτε κʼ (χʼ), which Van Leeuwen accordingly would restore. The mischief however must lie deeper. Of the 22 places there are 13 in which ὅτʼ ἄν appears in the leading clause of a simile (ὡς δʼ ὅτʼ ἄν . . .), and in three others (Il. 2.397, Od. 11.18, 13.101) the sense is general; so that ὅτε κʼ is admissible in six only (Il. 7.335 & 459, 8.373 & 425; Od. 2.374, 4.477). It cannot be an accident that there are so many cases of ὅτʼ ἄν where Homeric usage requires the pure subjunctive and no similar cases of ὅτε κεν; but for that very reason we cannot correct them by reading ὅτε κʼ. Meanwhile, no better solution has been proposed, and we must be content to note the 16 places as in all probability corrupt or spurious.
It is one thing, however, to find that ἄν has encroached upon κεν in Homer, and another thing to show that there are no uses of ἄν which belong to the primitive Homeric language.
The restoration of κε(ν) is generally regarded as especially easy in the combination οὐκ ἄν, for which οὔ κεν can always be written without affecting either sense or meter. The change, however, is open to objections which have not been sufficiently considered. It will be found that οὐκ ἄν occurs 61 times in the ordinary text of Homer, while οὔ κεν occurs 9 times, and οὔ κε 7 times. Nοw of the forms κεν and κε the first occurs in the Iliad 272 times, the second 222 times. Hence, according to the general laws of probability, οὔ κεν and οὔ κε may be expected to occur in the same proportion; in the ordinary text this is the case (9:7). But if every οὐκ ἄν were changed into οὔ κεν, there would be 70 instances of οὔ κεν against 7 of οὔ κε. This clearly could not be accidental, hence it follows that οὐκ ἄν must be retained in all or nearly all the passages where it now stands3 And if οὐκ ἄν is right, we may infer that the other instances of ἄν with a negative—22 in number—are equally unassailable.
Another group of instances in which ἄν is evidently primitive consists of the dactylic combinations ὅς περ ἄν, ᾗ περ ἄν, εἴ περ ἄν. Van Leeuwen would write ὅς κέ περ, etc.; but in Homer περ usually comes immediately after the relative or εἰ, and before κεν (§ 365). Similarly οὐδὲ γὰρ ἄν (Il. 24.556) and τόφρα γὰρ ἄν (Od. 2.77) cannot be changed into οὐδέ κε γάρ, τόφρα κε γάρ, since the order γάρ κεν is invariable in Homer. In these uses, accordingly, ἄν may be defended by an argument which was inapplicable to οὐκ ἄν, viz. the impossibility of making the change to κεν.
The same may be said of the forms in which ἄν occurs under the ictus of the verse, preceded by a short monosyllable (˘ˉ́).
Il. 1.205 ᾗς ὑπεροπλίῃσι τάχʼ ἄν ποτε θυμὸν ὀλέσσῃ
Od. 2.76 εἴ χʼ ὑμεῖς γε φάγοιτε, τάχʼ ἄν ποτε καὶ τίσις εἴη
Od. 9.77 τίς ἂν τάδε γηθήσειε
So τίς ἄν, Il. 24.367; Od. 8.208, 10.573.
Il. 4.164 ἔσσεται ἦμαρ ὅτʼ ἄν ποτʼ κτλ.
Cp. Il. 1.519, 4.53, 6.448, 9.101.
Il. 8.406 (= 420) ὄφρʼ εἰδῇ γλαυκῶπις ὅτ ἂν ᾧ πατρὶ μάχηται
So καὶ ἄν and τότ ἄν (see the instances, § 363.2.c), σὺ δʼ ἄν (Il. 6.329), ὃς ἄν (Od. 21.294, cp. Od. 4.204, 18.27, Il. 7.231). In this group, as in the last, we have to do with recurring forms, sufficiently numerous to constitute a type, with a fixed rhythm, as well as a certain tone and style.
The combination of ἄν and κεν in the same clause is found in a very few places, and is probably not Homeric. In four places (Il. 11.187 & 202; Od. 5.361, 6.259) we have ὄφρʼ ἂν μέν κεν κτλ., where the place of ἄν is anomalous (§ 365). Fοr οὔτʼ ἄν κεν (Il. 13.127) we should probably read οὔτʼ ἄρ κεν, and so in Od. 9.334 τοὺς ἄρ κε (or rather οὕς ἄρ κε) καὶ ἤθέλον αὐτὸς ἑλέσθαι (cp. Il. 7.182 ὃν ἄρʼ ἤθελον αὐτοί). In Od. 18.318 ἤν περ γάρ κε should be εἴ περ γάρ κε (supra).
363. Uses of κεν and ἄν. It will be convenient, by way of supplement to what has been said in the chapter on the uses of the moods (1) to bring together the chief exceptions to the general rule for the use of κεν or ἄν in subordinate clauses; and (2) to consider whether there are any differences of meaning or usage between the two particles.
- In final clauses which refer to what is still future, the use of κεν or ἄν prevails (§ 282, 285, 288, 293, 354). But with certain conjunctions (especially ὡς, ὅπως, ἵνα, ὄφρα) there are many exceptions: see §§ 285-289, 305-307. When the purpose spoken of is not an actual one, but either past or imaginary, the verb is generally "pure."
In conditional clauses the subjunctive and optative generally take κεν or ἄν when the governing verb is in the future, or in a mood which implies a future occasion (imperative, subjunctive, optative with κεν or ἄν). On the other hand in similes, maxims, and references to frequent or indefinite occasions, the particle is not used. But
a. Sometimes the pure subjunctive is used after a future in order to show that the speaker avoids referring to a particular occasion; cp.
Il. 21.111 ἔσσεται ἢ ἠὼς ἢ δείλη ἢ μέσον ἦμαρ,
ὁππότε . . . ἕληται
b. In our texts of Homer there are many places in which κεν or ἄν is used although the reference is indefinite; but the number is much reduced if we deduct the places in which it is probable that κε (or κʼ) has crept in instead of τὲ (τʼ) : see 283.b. The real exceptions will generally be found where a clause is added to restrict or qualify a general supposition already made.
ll. 3.25 μάλα γάρ τε κατεσθίει, εἴ περ ἂν αὐτὸν
(eνen in the case when, etc.)
Od. 21.293 οἶνός σε τρώει μελιηδής, ὅς τε καὶ ἄλλους
βλάπτει, ὅς ἂν μιν χανδὸν ἕλῃ
(in the case of him who takes it greedily)
So ll. 6.225, 9.501 & 524, 20.166; Od. 15.344, 19.332 (§§ 289, 292, 296). In these places we see the tendency of the language to extend the use of κεν or ἄν beyond its original limits, in other words, to state indefinite cases as if they were definite—a tendency which in later Greek made the use of ἄν universal in such clauses, whether the event intended was definite or not.
The change is analogous to the use of the indicative in a general conditional protasis; when, as Mr, Goodwin expresses it, "the speaker refers to one of the cases in which an event may occur as if it were the only one—that is, he states the general supposition as if it were particular" (Moods and Tenses, § 467). The loss of the Homeric use of τε, and the New Ionic use of ὁ, ἡ, τό as a relative with indefinite as well as definite antecedents, are examples of the same kind.
- Up to this point the particles κεν and ἄν have been treated as practically equivalent. There are however some diferences of usage which remain to be pointed out.
a. In negative clauses there is a marked preference for ἄν. In the ordinary text of the Iliad ἄν is found with a negative 53 times (nearly a third of the whole number of instances), κεν is similarly used 33 times (about one-twentieth). The difference is especially to be noticed in the Homeric use of the subjunctive as a kind of future (§§ 275, 276). In affirmative clauses of this type κεv is frequent, ἄν very rare; in negative clauses ἄν only is found.
b. κεv is often used in two or more successive clauses of a sentence; e. g. in both protasis and apodosis.
ll. 1.324 εἰ δέ κε μὴ δώῃσιν, ἐγὼ δέ κεν αὐτὸς ἕλωμαι κτλ.
In disjunctive sentences.
Il. 18.308 στήσομαι, ἤ κε φέρῃσι μέγα κράτος ἦ κε φεροίμην
Od. 4.692 ἄλλον κʼ ἐχθαίρῃσι βροτῶν, ἄλλον κε φιλοίη
And in parallel and correlative Clauses of all kinds.
II. 3.41 καί κε τὸ βουλοίμην καί κεν πολὺ κέρδιον εἴη
Il. 23.855- ὃς μέν κε βάλῃ . . .
857 ὃς δέ κε μηρίνθοιο τύχῃ, κτλ.
Od. 11.110 τὰς εἰ μέν κʼ ἀσινέας ἐάᾳς νόστου τε μέδηαι,
καί κεν ἔτʼ εἰς Ἰθάκην κακά περ πάσχοντες ἵκοισθε·
εἰ δέ κε σίνηαι κτλ.
ἄν, on the other hand, is especially used in the second of two parallel or connected clauses.
ll. 19.228- ἀλλὰ χρὴ τὸν μὲν καταθάπτειν ὅς κε θάνῃσι . . .
230 ὅσσοι δʼ ἂν πολέμοιο περὶ στυγεροῖο λίπωνται κτλ.
Od. 19.329- ὃς μὲν ἀπηνὴς αὐτὸς ἔῃ καὶ ἀπηνέα εἰδῇ . . .
332 ὅς δʼ ἂν ἀμύμων αὐτὸς ἔῃ κτλ.
So ll. 21.553 εἰ μέν κεν . . . εἰ δʼ ἂν κτλ.; Il. 3.288 ff. εἰ μέν κεν εἰ δέ κε . . . εἰ δʼ ἂν (the last an alternative to the second).
The only instance of ἄν in two parallel clauses is
Od. 11.17 οὔθʼ ὁπότʼ ἂν στείχῃσι πρὸς οὐρανὸν ἀστερόεντα
οὔθʼ ὅτʼ ἂν ἂψ ἐπὶ γαῖαν κτλ.
and there we ought to read ὁπότε στείχῃσι, according to the regular Homeric use of the subjunctive in general statements (§ 289.2.a)
c. There are several indications of the use of ἄν as a more emphatic particle than κεν. Thus the combination ἦ τʼ ἄν surely in that case occurs 7 times in the Iliad, ἦ τέ κεν only twice. Compare the force of καὶ ἄν in
Il. 5.362 ( = 457) ὃς νῦν γε καὶ ἂν Διῒ πατρὶ μάχοιτο
Od. 6.300 ῥεῖα δʼ ἀρίγνωτʼ ἐστί, καὶ ἂν πάϊς ἡγήσαιτο
Il. 14.244 ἄλλον μέν κεν . . .
ῥεῖα κατευνήσαιμι, καὶ ἂν ποταμοῖο ῥέεθρα
I would put any other to sleep, even Oceanus, etc.
Cp. also τότ ἄν (then indeed, then at length), in
Il. 18.397 τότ ἂν πάθον ἄλγεα θυμῶ.
Il. 22.108 ἐμοὶ δὲ τότʼ ἂν πολὺ κέρδιον εἴη κτλ.
Il. 24.213 τότʼ ἂν τιτὰ ἔργα γένοιτο
Od. 9.211 τότʼ ἂν οὔ τοι ἀποσχέσθαι φίλον ἦεν.
And τίς ἂν (quis tandem) in
Il. 9.77 τίς ἂν τάδε γηθήσειεν
Il. 24.367 τίς ἂν δή τοι νόος εἴη
Od. 8.208 τίς ἂν φιλέοντι μάχοιτο
Od. 10.573 τίς ἂν θεὸν οὐκ ἐθέλοντα κτλ.
The general effect of these differences of usage between the two particles seems to be that ἄν is used either in an adversative sense—with a second or opposed alternative—or when greater emphasis has to be expressed.
This account of the matter is in harmony with the predominance of ἄν in negative sentences. When we speak of an event as nοt happening in certain circumstances, we generally do so by way of contrast to the opposite circumstances, those in which it will happen.
οὐκ ὅν τοι χραίσμῃ κίθαρις
the lyre will not avail you
(viz. in battle—whatever it may ado elsewhere).
The accent of the particles must not be overlooked as a confirmation of the view now taken. Evidently ἄν is more likely to convey emphasis than the enclitic κεν. We may find an analogy in the orthotone and adversative δέ, which stands to τε and the correlated τε . . . τε somewhat as we have supposed ἄν to stand to κεν and κεν. . . κεν.
364. Original meaning of ἄν and κεν. The identity of the Greek ἄν with the Latin and Gothic an has been maintained with much force and ingenuity by Prof. Leo Meyer. The following are some of the chief points established by his dissertation.4
- The Latin an is used by the older poets in the second member of a disjunctive question, either direct, as egοne an ille injurie facimus? or indirect, as utrum scapulae plus an cοllus calli habeat nesciο (both from Νaevius). The use in single questions is a derivative one, and properly implies that the question is put as an alternative.
Plaut. Asin. 5.1.10 credam istuc, si te esse hilarum videro.
AR. An tu me tristem putas?
do you then think me (the opposite, viz.) sad?
Amph. 3.3.8 derides qui scis haec dudum me dixisse per jocum.
SO. an illut joculo dixisti? equidem serio ac veto ratus.
In these places5 we see how an comes to mean then on the contrary, then in the other case, etc. So in Naevius, eho an vicimus? what then, haνe we conquered?
- In Gothic, again, an is used in questions of an adversative character.
Luke X. 29 an hvas ist mis nȇhvundja
("he willing to justify himself, said"):
and who is my neighbor ?
John xviii. 37 an nuh thiudans is thu
are you a king then?
- These instances exhibit a close similarity between the Latin and the Gothic an, and suggest the possibility of a disjunctive particle (or, or else) coming to express recourse to a second alternative (if not, then. . .), and so acquiring the uses of the Greek ἄν. This supposition, as Leo Meyer goes on to show, is confirmed by the Gothic aiththau and thau, which are employed (1) as disjunctive particles, or, οr else, and (2) to render the Greek ἄν, chiefly in the use with the past indicative. Thus we have, as examples of aiththau
Matth. v. 36 ni magt ain tagl hveit aiththau svart gataujan
you can not make one hair white or black
Math. ix. 17 aiththau distaurnand balgeis
(neither do men put new wine into old bottles)
else the bottles break
John xiv. 2 niba vêseina, aiththau qvêthjau
If it were not so, I would have told you
[= it is not so, else I wοuld have told you]
Jοhn xiv. 7 ith kunthêdeith mik, aiththauu kunthêdeith etc.
if you had known me, you should have known, etc.
Similarly thau is used (1) to translate in double questions, as in Math. xvii. 17 whom will you that I release unto you, Barabbas or (than) Jesus? and after a comparative (= than); frequently also (2) in a conditional apodosis, especially to translate ἄν with past tenses.
Luke vii. 39 sa ith vêsi praufêtus ufkunthedi thau
this man, if he were a prophet, would have known
Sometimes also with the present (where there is no ἄν in the Greek)—the meaning being that of a solemn or emphatic future.
Mark xi.26 ith jabai jus ai aflêtith, ni thau . . aflêtith
if you dο not forgive neither will . . . forgive
(οὐδὲ . . . ἀφήσει)
Math. v. 20 ni thau qvimith
(except your righteousness shall exceed, etc.)
you shall in no case enter, etc.
(οὐ μὴ εἰσέλθητε)
This use evidently answers to the Homeric κεν or ἄν with the subjunctive and future indicative: ni thau qvimith = οὐκ ἂν ἔλθητε, ni thau aflêtith = οὐδʼ ἂν ἀφήσει.
- If now we suppose that ἄν, like aiththau and thau, had originally two main uses
(1) In the second member of a disjunctive sentence (= else, or else), and
(2) In the conditional apodosis (= in that case rather.
We can explain the Gοthic and Latin an from the former, the Greek ἄν from the latter. The idiomatic "ellipse" in ἦ γὰρ ἂν . . . ὕστατα λωβήσαιο else you would outrage fοτ the last time will represent an intermediate or transitional use. We can then understand why ἄν should often accompany negatives, and why it should be used in the latter clause of a sentence. The main difference of the two uses evidently is that in the first the clauses are coordinate, in the second the clause with ἄν is the apodosis or principal clause. Thus the two uses are related to each other as the two uses of δέ
(1) As an adversative conjunction.
(2) In the apodosis.
- The use of ἄν in final clauses may be illustrated by that of thau in
Mark vi. 56 bêdun ina el thau . . . attaitôkeina
παρεκάλουν αὐτὸν ἵνα κἂν . . ἅψαωνται
that they might touch if it were but, etc.
With ἵνα, ὡς, etc. ἄν may have had originally the same kind of emphasis as κἄv in this passage, "that in any case," "that if no more then at least, etc." The use in a conditional protasis following the principal clause may be compared with
Luke ix. 13 niba thau . . . bugjaima
(we have no more) except we should buy
(= unless indeed we should buy)
The particle κε(v) is found in Aeolic, in the same form as in Homer, and in Doric, in the form κα. It is usually identified with the Sanskrit kam, which when accented means well (wοhl, gut, bene), and as an enclitic appears to be chiefly used with the imperative, but with a force which can hardly be determined (Delbrück, A. S. pp. 150, 503). A parallel may possibly be found in the German wοhl, but in any case the development of the use of κε(v) is specifically Greek.
- 1. "Im Allgemeinen steht das Resultat durchaus fest: κεν beim Conjunctiv und Optativ weist auf das Eintreten der Handlung hin" (Delbrück, Synt. Fοrsch. i. p. 86), This view is contrary to the teaching of most grammarians (see especially Hermann on Soph. O. C. 1446). It wil1 be found stated very clearly in an article in the Philological Museum, vol. i. p. 96 (Cambridge 1832).
- 2. De particularum κέν et ἄν apud Homerum usu (Mnemosyne, XV. p. 75). The statistics given above are taken from this valuable dissertation.
- 3. It will be seen that the argument is of the same kind as that by which it was shown above (§ 283.b) that τε must have been often changed into κε. The decisive fact in that case was the excessive occurrence of κε. Here it is the absence of any such excess which leads us to accept the traditional text.
- 4. ἈΝ im Griechischen, Lateinischen und Gοthischen, Berlin 1880. The parallel between the Greek ἄν and the Gothic thau and aiththau was pointed out by Hartung (Partikeln, ii. p. 227).
- 5. Taken from Draegerʼs Historische Syntax, i. p. 321, where many other examples will be found.