Lecture II, 30 - Oxford Scholarship Jump to ContentJump to Main Navigation
Jacob Wackernagel, Lectures on SyntaxWith Special Reference to Greek, Latin, and Germanic$

David Langslow

Print publication date: 2009

Print ISBN-13: 9780198153023

Published to Oxford Scholarship Online: September 2009

DOI: 10.1093/acprof:oso/9780198153023.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2015. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 27 August 2016

Lecture II, 30

Lecture II, 30

(p.745) Lecture II, 30
Jacob Wackernagel, Lectures on Syntax

David Langslow (Contributor Webpage)

Oxford University Press

Abstract and Keywords

This final chapter first reviews the available means of expression of negation — both inherited and secondary, including (Lecture 27) prohibitive particles — and (Lecture 28) the placement of negative markers, and their use with nouns. There follow (Lecture 29) remarks on quantitative (as opposed to qualitative) negatives — expressions for ‘no one’, ‘nothing’, ‘never’ — and the negation of general statements,and (Lecture 30) discussion of some special uses of Gk μή and Lat. ne. The privative prefix, and the use generally of negatives in compounds, both nominal (possessive and determining) and verbal, are the subjects of Lecture 31; the meaning of privative compounds, and forms that compete with them, are considered at the start of Lecture 32. The chapter concludes (Lectures 32–3) with discussion of the accumulation of negatives (negative + privative compound, qualitative + quantitative negative, prohibitive + declarative negative, pleonastic negation), and finally of the coordination of negated clauses and phrases.

Keywords:   compound, coordination, declarative, negative, pleonastic, possessive compound, privative, prefix, prohibitive, qualitative

With regard to the prohibitive particles Gk μή and Lat. (II, 258–9 above), there are some further special points to report, in addition to what they have in common with the simple negative particles (which we have already discussed). We considered earlier (I, 213–16) how the use of the mood of the verb with these particles has shifted in main‐clause prohibitions compared with the situation in Indo‐European. But in both Greek and Latin these particles serve also to introduce subordinate clauses, a function favoured by the frequent placement of the negative in clause‐initial position (II, 259–60 above).

In the first place, μή‐ and ‐ clauses can serve, just like positive clauses of wishing, to justify an appeal or a declaration of will in the preceding clause. A sentence such as Il. 10. 65 (Agamemnon to Menelaus) αὐ̑θι μένϵιν, μή πως ἀβροτάξομϵν ἀλλήλοιϊν (‘stay there, lest by chance we two miss each other’) has exactly the same structure as e.g. Il. 23. 71 (Patroclus' ghost to Achilles) θάπτϵ μϵ ὅττι τάχιστα πύλας Ἀίδαο πϵρήσω (‘bury me as quickly as possible let me pass inside the gates of Hades’). In both cases, the subjunctive clause can be rendered either with an independent clause—‘stay here: (for) I don't want us to miss each other’; ‘bury me: (for) I wish to pass through the gates of Hades as soon as possible’—or with a subordinate clause introduced by ‘to prevent’ or ‘in order that’.1 It is noteworthy that Nicanor (I, 22 above) | in cases of this sort vacillated between strong and weak punctuation; see FRIEDLÄNDER (1857: 29–30). We can recognize the close connection between the subjunctive clause and the preceding clause in the asyndeton between them; in speech, the link would have been marked also by intonation.

In the above example, the μή‐clause indicates something refused by the speaker himself at the time of speaking. To that extent it still has the character of an independent clause. An attached μή‐clause of this type can also, however, convey a refusal that justifies the action of another participant, e.g. at Il. 5. 844–5 Ἀθήνη | δυ̑ν᾽ Ἄϊδος κυνέην, μὴ μιν ἴδοι ὄβριμος Ἄρης (‘Athene put on the cap of Hades so that mighty Ares should not see her’). Here, the μή‐clause would make no sense on its own. It can be understood only through its connection with the preceding (p.746) clause, and the refusal is thought of not from the narrator's viewpoint, but from that of the agent reported in the immediately preceding clause. The μή‐clause has thus acquired unequivocally the value of a subordinate clause. In the above example, this value emerges clearly from the optative in the μή‐clause, which is conditioned by the preterite in the preceding clause (cf. I, 26, 238–9 above). The same applies to clauses in which someone tells of his or her own earlier actions, e.g. Od. 9. 100–2 (Odysseus to the Phaeacians) κϵλόμην…ἑταίρους…νηω̑ν ἐπιβαίνϵμϵν… | μή πώς τις λωτοι̑ο ϕαγὼν νόστοιο λάθηται (‘I commanded my companions to board the ships, for fear that by chance one of them should eat some lotus and forget his homecoming’), or, with shift of mood, 376–7 πάντας ἑταίρους | θάρσυνον, μή τίς μοι ὑποδϵίσας ἀναδύη (optative; ‘I tried to hearten all my companions to prevent anyone faltering out of fear’).—The Latin negative purpose clauses with developed in an exactly analogous way.

Already in Homer, negative purpose clauses are introduced not only with bare μή but also with ἵνα μή, ὡς μή, ὄϕρα μή, and later also with ὁπως μή (‘so that not’). Since positive purpose clauses usually began with ἵνα, ὄϕρα, ὡς, etc., speakers liked to introduce negative ones too in the same way, and carried on adding those particles before the μή. The alternation of introductory particles may be seen in the selfsame phrase, e.g. at Il. 11. 704–5 τὰ δ᾽ ἄλλ᾽ ἐς δη̑μον ἔδωκϵ | δαιτρϵύϵιν, μή τίς οἱ ἀτϵμβόμϵνος κίοι ἴσης (‘the remainder he [Neleus] gave to the people to divide in such a way that as far as was in his power no one should go defrauded of their proper share’) compared with Od. 9. 41–2 κτήματα…δασσάμϵθ᾽, ὡς μή τίς μοι ἀτϵμβόμϵνος κίοι ἴσης (‘we divided the spoil, so that as far as was in my power no one should go defrauded of an equal share’). Indeed, even the hypothetical particle, which was completely foreign to the independent μή‐clause, is admitted alongside μή, though only three times in the Odyssey, in ὡς ἂν μή (2. 376=4. 749, and 16. 84).—In the face of this tendency, bare μή gradually fell into disuse in clauses of this type. It survives only rarely in Herodotus and Attic, and not at all in certain orators, including Lysias and Hyperides; for details, see Ph. WEBER (1884–5). Even so, μή meaning ‘lest, to prevent’ is still occasionally to be found even in the Koine and into the imperial period (see RADERMACHER 1911: 158 [=1925: 195])—and even in modern Greek (THUMB 1910: §280). |

In Latin, by contrast, simple remained standard, although, especially in the earlier language, ut ne is also not uncommon (or ut…ne: II, 260 above). This is comparable not only with Gk ὡς μή in purpose clauses, but also with utinam ne in main clauses of wishing, e.g. at Enn. Trag. 208–9 Jocelyn (the start of the Medea) utinam ne in nemore Pelio securibus caesa accidisset…trabes (‘would that the timber…had not fallen hewn by axes in a Pelian grove’), where from the classical period on utinam non is more normal (cf. II, 259 above).—The same applies to quī (‘how’) ne in Terence, Andria 335 ego id agam, mihi qui ne detur ‘I shall work to prevent her being given to me’—where Donatus comments: ‘ueteres frequenter (p.747) ne pro non dicebant’ (‘in early Latin ne was frequently used for non’)—and possibly also to quo ne in Horace, Satires 2. 1. 37; HAND (1829–45: IV, 36) documents quo ne also for one or two late writers.2 By contrast, in Cicero, Letters to his Friends 7. 2. 1 (no. 52 Shackleton Bailey) praefinisti quo ne pluris emerem ‘you fixed in advance the price above which I may not make the purchase’, a prohibition is embedded in a relative clause (cf. Livy 34. 6. 14, etc.). Different again is the rhetorician Rutilius Lupus (early 1st c.) 1. 9 (p. 7, 11 Halm), quaeritis maximis sumptibus faciendis, quo modo ne tributa conferatis (‘you seek at very great expense to avoid paying tribute’), a passage translated from the Athenian orator Stratocles (4th–3rd c.), where quo modo ne can only be understood as a translation of Gk ὅπως μή.

Greek and Latin agree in developing from prohibitions words for ‘let alone’, i.e. where the speaker refuses to talk about something: Gk μήτι (γϵ), μὴ ὅτι, etc., Lat. nēdum (also , nē ut). I cannot go into detail on this.3

Greek and Latin also have in common the use of the prohibitive particle to introduce clauses of fearing. This particle can also introduce an independent clause of fearing (and this is the starting point for the subordinate construction). For example, at Il. 16. 128, Achilles, on noticing the fire among the ships, says, μὴ δὴ νη̑ας ἕλωσι καὶ οὐκέτι ϕυκτὰ πέλωνται ‘(I fear) they may now capture the ships, and there may no longer be a means of escape’. In this type of expression, the speaker in a sense fends off from himself what is feared. But a μή‐clause of this type can be attached to a verb of fearing, e.g. at Od. 5. 473 δϵίδω, μὴ θήρϵσσιν ἕλωρ καὶ κύρμα γένωμαι (‘I fear I may become a prey and spoil for wild beasts’). The μή‐clause then indicates what is in the mind of the person afraid, and enters clearly into a relation of dependence on what precedes. The next step is the same as in the case of purpose clauses: fear felt by other people and fear felt in the past by oneself can be indicated in the same way, sometimes with a shift of mood. Even modern Greek has retained this pattern, though πω̑ς (roughly, ‘that’) and similar words are also used.4—The origin of the construction in the notion of fending off is completely forgotten when the verb stands in the indicative. Even this more advanced stage of development | has already been reached in Homer, e.g. Od. 5. 300 δϵίδω μὴ δὴ πάντα θϵā` νημϵρτέα ϵἰ̑πϵν ‘I am afraid that everything the goddess said is true’.—Latin clauses of fearing with are exactly like those in Greek with μή, except that in Latin the indicative is not admitted.

This negative form of clauses of fearing, strange to German eyes, is not confined to Greek and Latin. It is best known to us in French, as e.g. in je crains que tu ne sois malade (‘I fear that you are [lit., not] ill’), but it is also there in (p.748) English lest, formally comparable with Lat. quominus in that it contains ‘less’ (cf. II, 255 above), which is still used after verbs of fearing no less than in purpose clauses. Even in German there are examples: St Paul's words at 2 Cor. 11: 3 ϕοβου̑μαι μή πως…ϕθαρῃ̑ τὰ νοήματα ὑμω̑ν (‘I fear lest by any means your minds should be corrupted’) are rendered by Luther with ich fürchte, dass nicht…eure Sinne verrückt werden.—Latin is more consistent than Greek in that it uses alongside uereor ne ‘I fear that…’ the construction uereor ut ‘I fear that…not…’. This ut is the same as that used in positive prayers, such as ut te di perduint ‘may the gods destroy you’ (cf. also utinam). So, a sentence like Hor. Sat. 2. 1. 60 ut sis uitalis, metuo was originally intended in the sense, ‘I am afraid: (yet) may you have long life!’ On the synonymous Gk μὴ οὐ, Lat. nē nōn, see below, and note οὐκέτι (‘no longer’) in Achilles' words at Il. 16. 128, quoted above.

From clauses of purpose and fearing Gk μή‐clauses came to be used also after verbs of ensuring, taking care, and the like, as e.g. in Plato, Theaet. 145b ὅρα μὴ παίζων ἔλϵγϵν (‘make sure he was not speaking in jest’), where we might translate μή with ‘whether…not’ (German ob…nicht), or indeed Theocr. 12. 36–7 Λυδίῃ…πέτρῃ…,χρυσὸν ὁποίῃ πϵύθονται, μὴ ϕαυ̑λος, ἐτήτυμον ἀργυραμοιβοί (‘the Lydian touchstone by which money‐changers try true gold to see it be not false’), where BÜCHELER (1875: 47) wanted to change what he saw as a ‘structura incondita’ (‘crude construction’): VAHLEN (1891) saw the truth.5 Furthermore, in Greek of the Roman Empire, μή is used after ordinary verbs of asking (cf. KG II, 394, §553.3 n. 3). More or less the same is found in Latin, where e.g. circumspicere nē means at first ‘look around to prevent’ (so perhaps at Plaut. Mil. 955?), but then in Varro, Rust. 2. 10. 1 cum circumspiceret, ne quid praeterisset (‘as he glanced around to see if he had overlooked anything’) it must mean ‘check whether’, i.e. with an indirect question. In later Latin, is well established in the sense ‘whether’. In the Latin Bible, Luke 3: 15 διαλογιζομένων πάντων…πϵρὶ του̑ Ἰωάννου, μή ποτϵ αὐτὸς ϵἴη ὁ Χριστός (‘as they all mused concerning John, whether he were the Christ’) can be translated literally as cogitantibus omnibus… de Ioanne, ne forte ipse esset Christus—Wulfila does the same: þagkjandam allaim …bi Iohannein, niu6 aufto sa wesi Xristus; but contrast Luther: und dachten alle von Johannes, ob er vielleicht Christus wäre.—This provides the basis for constructions in Vulgar Latin such as in the ?third‐century veterinary text Mulomedicina Chironis, 10 aquam ostendis ne bibere uelit ‘show him (the horse) water to see whether he wishes to drink’,7 which lead in turn to Late Latin meaning (p.749) ‘in case, if’ | —on this usage, see LÖFSTEDT (1911: 268) following AHLQVIST (1909: 103–6).8

Apart from clauses of purpose, clauses of fearing, and clauses related to purpose after verbs of wishing, requesting, warning, and preventing (where Greek generally prefers the infinitive), we see in Latin a start made to introduce (with long ē attested at e.g. Plaut. Curc. 36) even in conditional clauses. This involves conditionals containing a prohibition the observation of which determines the validity of the main clause, and they are introduced with dum. We have this several times in the famous Senatus consultum de Bacchanalibus of 186 BC,9 e.g. 8–9 (=16–18) nisei…is…de senatuos sententiad, dum ne minus senatoribus centum adesent, quom ea res cosoleretur, iousisent ‘unless he (the praetor urbanus) has commanded it on the basis of a decision of the Senate, provided that no fewer than a hundred senators were present when the matter was debated’. I would note here that ne with the comparative (as above) is common in the language of public affairs and business in other types of directive clauses with a striking attraction of the negative particle to the comparative, e.g. in the same senatus consultum, 22–3 haice utei…exdeicatis ne minus trinum noundinum (‘[and the Senate decreed that you] are to announce these provisions [in your assembly] over a period of not less than three weeks’),10 or at Livy 30. 37. 6, etc. It is found even without a subjunctive verb, e.g. at Cicero, Laws 2. 66 noluit quid statui nisi columellam tribus cubitis ne altiorem ‘he (Demetrius) did not wish anything to be erected (over a grave) except a small column no more than three cubits high’.—In some cases dum and ne are separated, e.g. Plaut. Capt. 338 quiduis, dum ab re ne quid ores, faciam ‘I'll do whatever you wish, provided that you ask for nothing against my own interests’. In Classical Latin dum ne survives mainly—though not exclusively—in the official language, while the fuller dummodo ne (attested already in pre‐classical Latin) survives somewhat longer. Other synonyms in subjunctive clauses are modo ne and tantum ne ‘only provided that…not’. Note also the use of simple ne at Cicero, Laws 2. 36 tu uero istam Romae legem rogato: nobis nostras ne ademeris ‘you, however, may introduce this law at Rome: just don't take ours away from us’.—Like Gk ϵἰ μή, ὅτι μή, so Lat. dum ne, dummodo ne are also found without a verb, at least in letters, where they may be translated ‘just not, except’: e.g. Cicero, To Atticus 12. 44. 4 (no. 285 Shackleton Bailey) scribas te nihil habuisse, quod scriberes, dummodo ne his uerbis ‘write that you had nothing to say, just not in so many words’ (similarly 6. 1. 4, no. 115 Shackleton Bailey dum ne negotiatori ‘just not for businessmen’).

(p.750) Except in dum ne and related phrases, Latin is not used in conditional clauses. We must not confuse with ‘if not’, as is sometimes done, and the negative with si is either or nōn (cf. II, 251 above). In Greek, however, from Homer on, μή is usual in conditional clauses. Homer does not apply the rule consistently, but in Attic οὐ is generally found in | conditionals only if it belongs very closely with the verb (cf. II, 261 above) or another constituent, or if the ‘if’‐clause contains something that is actually true. This use of μή is in conflict with the prehistory of the particle, and cannot be compared with Lat. dum ne, etc., as Greek ϵἰ μή‐clauses, unlike Latin dum ne‐clauses, have no sense of conveying a demand. No, we have here a striking innovation of Greek, which to the best of my knowledge has yet to be explained. One might be inclined to connect it with the use of ϵἰ and related forms in main‐clause wishes and appeals. If these are negative, μή is required, and one might suppose that it was extended from here to subordinate, i.e. conditional, clauses.11 The problem is that these independent ϵἰ‐clauses are nearly always positive. I know only one counterexample in Homer, at Il. 16. 97–9 (Achilles' prayer to sack Troy with Patroclus alone) αἰ γὰρ…μήτϵ…μήτϵ…, where μή does not even immediately follow αἰ.

One might consider another explanation. It is well known that optative and jussive expressions can also be used with concessive force: this is the reason for e.g. Latin concessive clauses introduced by quamuis. From the meaning ‘conceded that’ it is possible for ‘given that’ and ‘if’ to develop. KRÜGER (1873–91: I, §54.4 n. 2) documents this nicely with imperatival clauses,12 e.g. Antiphon the Sophist (5th c.), B49 (no. 87 DIELS & KRANZ) ϕέρϵ δὴ προϵλθέτω ὁ βίος,…αὕτη ἡ ἡμέρα…καινου̑ δαίμονος ἄρχϵι ‘come now! If life is prolonged (lit. let life go forward, scil. so that one marries),…this day is the beginning of a new destiny’. In Latin, the use of ut+subjv. in such clauses when they are positive—as in the famous line of Ovid, ut desint uires, tamen est laudanda uoluntas (‘even though I lack the strength, yet the will is praiseworthy’, Letters from Pontus 3. 4. 79)—is matched by if something negative is conceded or granted, e.g. Cic. Tusc. 2. 14 ne sit sane summum malum dolor, malum certe est ‘granted that pain is certainly not the greatest evil, still it is certainly an evil’. Was it then once possible to use μή in clauses of this type, meaning ‘given that…not’, and was this μή then reinforced by the regular conditional particle ϵἰ? The striking thing is that neither μή nor, say, ὅπως μή is attested in this sense—Il. 7. 353 ἵνα μὴ ῥέξομϵν ὡ̑δϵ (‘where we do not act as I suggest’) cannot be used as an example.13 Furthermore, this sort of expression seems in general to be more at home in an intellectual age.

(p.751) Whatever view is taken of the origin of this use of μή in conditional clauses, it is very ancient and common to all Greek dialects: suffice it to note αἰ μά(=ϵἰ μή) in early Elean documents.14 The usage has deep roots in Homer, where it already occurs also, albeit not very frequently, in relative and temporal clauses with hypothetical colouring, e.g. Od. 11. 490 (Achilles in the Underworld) ἀνδρὶ παρ᾽ ἀκλήρῳ, ᾡ̑ μὴ βίοτος πολὺς ϵἴη (‘for a man without land of his own, who has not much to live on’). Later, the use of μή in subordinate clauses is extended | further still. This is already the case in Attic, where the construction occurs inter alia in relative clauses with causal overtones containing something factual, e.g. at Thuc. 8. 76. 6 οἵ γϵ μήτϵ ἀργύριον ϵἰ̑χον ἔτι πέμπϵιν…μήτϵ βούλϵυμα χρηστόν ‘since they (the Athenians) had neither money to send them (the army in Samos)…nor good advice (to offer)’.

Post‐classical Greek goes even further, especially in the imperial period (see GREEN 1902). Here we find also ἐπϵὶ μή, ἐπϵιδὴ μή ‘since…not’, ὅτι μή ‘because…not’ or ‘that…not’ (emended away at Antiphon, speech 5. 2115), διότι μή ‘because…not’, and even μέχρι μὲν μηδϵὶς ἀϕι̑κτο παρ᾽ ἡμω̑ν (‘as long as no one appeared from us [Athenians]’) in the sophist Aelius Aristides (2nd c. AD).16 In a sense, then, μή has assumed the character of a negative serving simply for any kind of subordinate clause. The first to give the correct assessment of this phenomenon was the well‐known American Hellenist GILDERSLEEVE (I, 32–3 above). His account (1880) is supplemented by the rich material contained in W. SCHMID's learned and useful work on Atticism in the works of its chief exponents (1887–97: V, 162, Index, s.v. ‘μή'). From imperial prose SCHMID can document even the use of μή instead of οὐ in a simple main clause, e.g. at Aelian, On the Characteristics of Animals 2. 49 (οἱ κόρακϵς) τοὺς γϵιναμένους σϕα̑ς μὴ τρέϕουσιν ‘crows do not feed their parents’ (cf. II, 283–4 below).—This departure from the classical language was noticed already by the ancient linguists: a first‐century AD grammarian referred to it as the Ἀλαβανδιακὸς σολοικισμός (‘the Alabandian solecism’; STEPHANUS, s.v. ‘Ἀλάβανδα'), which must be based on observations of the members of the school of rhetoric which flourished in Carian Alabanda around 100 BC. Much in this most recent use of μή is certainly literary affectation. It is unknown in the modern colloquial language, except that the (p.752) negatives οὔτϵ, μήτϵ and οὐδέ, μηδέ, which also serve as conjunctions, are used interchangeably (see THUMB 1910: §285).17

Parallel to the above development is the increasing use of μή with participles and infinitives. In both cases, the use of the prohibitive particle is a Greek innovation, albeit one that is found already in Homer. Admittedly, Homer uses μή with the participle only in clauses expressing the will or a wish of the speaker, e.g. Od. 11. 613 μὴ τϵχνησάμϵνος μηδ᾽ ἄλλό τι τϵχνήσαιτο (‘may the craftsman who wrought it [Heracles' baldric] never make another!’; cf. Od. 4. 684, Il. 13. 48). The same applies to Hesiod, Works 591 (ϵἴη) βοὸς ὑλοϕάγοιο κρέας μή πω τϵτοκυίης (‘let the meat be of a cow put out to graze who has not yet had a calf’), and to the Hymn to Hermes 92 καί τϵ ἰδὼν μὴ ἰδὼν ϵἰ̑ναι (‘and be unseeing of what you have seen’). GILDERSLEEVE (1897a: 244) rightly characterizes as a step beyond epic usage the instance in Pindar, Nem. 4. 30–1 ἀπϵιρομάχας ἐών κϵ ϕανϵίη λόγον ὁ μὴ ξυνιϵίς (‘a man who does not know the proverb [“it is right that the doer suffers also”] would obviously have no experience of battle’). This secondary use of μή with the participle originates in negative conditionals: ὁ μὴ ξυνιϵίς is roughly equivalent to ἐάν τις μὴ ξυνιῃ̑ (‘if someone does not know’). The normal Attic use of the negatives is best illustrated by a passage of Xenophon, Anab. 4. 4. 15 ἐδόκϵι…πολλὰ ἤδη ἀληθϵυ̑σαι | τὰ ὄντά τϵ ὡς ὄντα καὶ τὰ μὴ ὄντα ὡς οὐκ ὄντα (‘he had the reputation for bringing in accurate information: when he said something was there, it was there, and when he said it wasn't, it wasn't’), where the final words correspond to ἃ ἂν μὴ ᾐ̑, λέγϵι ὡς οὐκ ἔστι (‘whatever is not there, he says is not there’). In later Greek, the use of μή with the participle expands considerably until it becomes the standard participial negative. In certain parts of the New Testament, οὐ is hardly found at all any more with the participle, and in modern Greek μή alone is regular.18

With the infinitive, μή is most naturally used when it functions as an imperative (I, 266–7 above), as e.g. at Il. 16. 839 (Hector quoting Achilles to the dying Patroclus) μή μοι πρὶν ἰέναι ‘do not come back to me before…’; in Vedic, even here the plain negative, not the prohibitive, is used, while Young Avestan appears to agree with Greek.19 It is equally natural in prayers, as at Il. 2. 413 (Agamemnon's prayer) μὴ πρὶν ἐπ᾽ ἠέλιον δυ̑ναι καὶ ἐπὶ κνέϕας ἐλθϵι̑ν ‘let not the sun set and the darkness descend before then’, and after δός ‘grant’ (cf. I, 267–8 above), e.g. Od. 9. 530 (Polyphemus' prayer) δὸς μὴ Ὀδυσση̑α πτολιπόρθιον οἴκαδ᾽ ἱκέσθαι (‘grant that Odysseus, sacker of cities, may not reach his home’). It also makes sense, once it is admitted with the infinitive at all, in oaths, e.g. Od. 2. 373 (Telemachus to Eurycleia) ἀλλ᾽ ὄμοσον μὴ μητρὶ ϕίλῃ τάδϵ μυθήσασθαι (‘but swear that you will (p.753) not tell this to my dear mother’; cf. 4. 746–7). But already in Homer μή is used in denials of fact, not only after verbs of swearing, as at Il. 23. 585 (Menelaus to Antilochus) ὄμνυθι μή τι ἑκὼν τὸ ἐμὸν δόλῳ ἅρμα πέδη̑σαι ‘swear that you did not deliberately baffle my chariot by trickery’, but elsewhere, too, e.g. Il. 18. 500 (in the lawsuit on the Shield of Achilles) ὁ δ᾽ ἀναίνϵτο μηδὲν ἑλέσθαι ‘the other party denied that he had taken anything’—exactly like Ar. Knights 572–3 ϵἰ δέ που πέσοιϵν…ἠρνου̑ντο μὴ πϵπτωκέναι ‘if they ever chanced to fall, they denied that they had fallen’. (For a parallel in main clauses, see II, 283 below.)—In general, here, too, we can see a steady growth in the use of μή. In Attic, two types of instances deserve to be highlighted. Since ὥστϵ+infinitive can also express a desired consequence, which hence called for μή, ὥστϵ μή became generally common in Attic, even with purely factual consequences. The negative with the substantival infinitive is always μή, without exception. The starting point here were instances such as Aesch. Agam. 206–7 βαρϵι̑α μὲν κὴρ τὸ μὴ πιθέσθαι, βαρϵι̑α δ᾽ ϵἰ τέκνον δαΐξω (‘not to obey is a heavy doom, but heavy too if I slay my child’), where, as the parallel ϵἰ‐clause shows, the infinitive has hypothetical force and hence calls for μή. On the strength of cases like this, μή became the standard negative with this use of the infinitive. Later in the same play, we have 568–9 παροίχϵται δὲ τοι̑σι μὲν τϵθνηκόσιν τὸ μή ποτ᾽ αὐ̑θις μηδ᾽ ἀναστη̑ναι μέλϵιν (‘[toil] is over for the dead, so that they will never care even to rise again’). Here the infinitive reports a fact: ‘for the dead the thought even of rising again is gone’. We must not be unsettled by the unnecessary use of μή here (quite apart from its repetition), just as at Thuc. 2. 49. 6 ἡ ἀπορία του̑ μὴ ἡσυχάζϵιν…ἐπέκϵιτο ‘the desperate feeling of being unable to keep still…was oppressive’, and very frequently elsewhere (cf. below).—As with the | participle, so with the infinitive, the New Testament has nearly always μή. On the general steady expansion of μή+infinitive in later Greek, see especially the works of GILDERSLEEVE and SCHMID (II, 281 above). Modern Greek cannot furnish a comparison here, as it has very largely lost the old infinitive (cf. I, 275–6 above).

The distribution of οὐ and μή in combination with nouns and adjectives can be left to one side, since in general the same applies to them as to infinitives and participles. For example, the μή‐phrase (in bold) in Thuc. 4. 87. 4 οὐδὲ ὀϕϵίλομϵν …μὴ κοινου̑ τινὸς ἀγαθου̑ αἰτίᾳ, τοὺς μὴ βουλομένους ἐλϵυθϵρου̑ν (‘we ought not to free those who do not wish it except by cause of some common good’) is equivalent to ϵἰ μὴ κοινου̑ τινὸς ἀγαθου̑ αἰτία ὑπάρχϵι (‘unless the cause of some common good obtains’); and similarly in Theocr. 6. 19 τὰ μὴ καλὰ καλὰ πέϕανται (‘foul has seemed fair’), τὰ μὴ καλὰ = ἃ ἂν μὴ καλὰ ᾐ̑ (‘whatever is not fair’; for examples of a close link between μή and nouns, see II, 264–5 above). A peculiar combination, common especially in tragedy, is ὁ / ἡ μηδέν ‘the worthless one’, and τὸ μηδέν ‘nothing’ (also without the article), which is much more frequent than οὐδέν (cf. BRUHN 1899: 150–1).

(p.754) By way of conclusion to this section, I must come back to independent clauses. Here, too, the use of Gk μή, unlike that of Lat. , expanded considerably, and several types of main clause have μή without prohibitive meaning and hence with the verb in the indicative. First, there are statements containing μή as a straightforward negative following an oath formula, as at Il. 15. 36–42 (Hera to Zeus) (ἴστω νυ̑ν τόδϵ γαι̑α …) μὴ δι᾽ ἐμὴν ἰότητα Ποσϵιδάων ἐνοσίχθων | πημαίνϵι Τρω̑ας (‘[let Earth be my witness…] it is not by my will that Poseidon the Earth‐shaker is afflicting the Trojans’), or Arist. Birds 194–5 (the Hoopoe) μὰ γη̑ν μὰ παγίδας μὰ νϵϕέλας μὰ δίκτυα, | μὴ ᾽γὼ νόημα κομψότϵρον ἤκουσά πω (‘by Earth, by traps, by bird‐nets, by snares, I never heard a more ingenious scheme’). Obviously, the pattern normal in promises on oath has been extended to statements on oath; compare the use of μή with the infinitive after verbs of swearing (II, 282 above), and also the shift in the meaning of ἀρνϵι̑σθαι (II, 258 above).—Secondly, there are cases like Od. 9. 405–6, where the other Cyclopes ask the moaning Polyphemus, ἠ̑ μή τίς σϵυ μη̑λα βροτω̑ν ἀέκοντος ἐλαύνϵι; | ἠ̑ μή τίς σ᾽ αὐτὸν κτϵίνϵι δόλῳ ἠὲ βίηϕιν; Here we can translate, ‘hopefully, no one is driving off your flocks against your will, or trying to kill you by trickery or force?’ (cf. Od. 6. 200). In unreal questions of this kind, what is asked is also wished for, and hence μή is used. Accordingly, throughout post‐Homeric Greek until the late period, μή is regular in questions which hope for or expect a negative answer (this is more accurate than what was said earlier, I, 235 above). Apart from simple μή, we find also μή τι, ἀρα μή (<ἠ̑ ἄρα μή, i.e. Homeric ἠ̑ μή with the addition of ἄρα), μω̑ν (< μὴ οὐ̑ν),20 occasionally with pleonastic μή or οὐ or οὐ̑ν. |—No less easy to understand than the two cases above is μή with indicative ὤϕϵλ(λ) ον+infinitive of an unfulfilled wish, e.g. Il. 9. 698 (Diomedes to Agamemnon) μηδ᾽ ὄϕϵλϵς λίσσϵσθαι…Πηλϵίωνα ‘you should not have appealed to…the son of Peleus’, relatively common in Homer, and also in later writers. The form would have required negative οὐ, the sense led inescapably to μή; cf. Lat. utinam ne (II, 277 above).—On the completely unmotivated use of μή for οὐ in main clauses in Imperial Greek prose, see II, 281 above.


(1) Richardson (1993) on Il. 23. 71 compares 22. 418 λίσσωμαι ‘let me supplicate’, which ‘is virtually a final clause’.

(2) These include Boethius (5th–6th c.), Consolation of Philosophy 2. 7.

(3) See further, on Greek, KG II, 130, Denniston (1954: 143–4), Schwyzer & Debrunner 707; on Latin, Hofmann & Szantyr §331.

(4) In modern Greek, a verb of fearing can be followed either by the simple complementizer ὅτι / πω̑ς or by μή(ν), μήπως, (νὰ μή); see Thumb (1910: §268.3), Holton et al. (1997: 451–2).

(5) Vahlen took ἐτήτυμον as an adverb qualifying φαυ̑λος, ‘truly false’; I have followed the text, punctuation, and translation of Gow (1950), q.v., ad loc.

(6) That is, negative ni+interrogative particle ‐u.

(7) On this anonymous text, see now Fischer (1989), and Adams (1995: 6 n. 29, and Index, s.v.). The instruction quoted by W. is rendered into ‘good’ Latin by the later veterinary writer Vegetius, who used the Mulomedicina, as aquam etiam offeres ut, si uoluerit, bibat ‘you should also offer him water so that, if he wishes, he may drink’.

(8) On in Late Latin, see further Hofmann & Szantyr 542.

(9) CIL I2. 581: cf. n. 38, p. 525 above for further references.

(10) On the problematic phrase trinum noundinum, see Primavesi (1993).

(11) It seems that this is still the standard view: see Schwyzer & Debrunner 322, 594–5, and note also Gildersleeve (1902: 132–6).

(12) Cf. Cooper (1998: I, §54.4.1).

(13) Does W. mean that Il. 7. 353 cannot be used because the line is spurious, and that Aristarchus was right to athetize it? Or does he mean that here ἵνα (or, with Aristarchus, ἵν᾽ ἂν) means the same as ἐάν ‘if’, with ‘a relatively easy extension [in the meaning of ἵνα] from local to circumstantial, or concrete to abstract, application’—so Kirk (1990: ad loc.), with reference to Chantraine (1953: 268)? In the latter case, μή would not be unsupported. (This passage is not indexed in Wackernagel (1916), nor in his Kl. Schr.)

(14) Cf. e.g. Buck no. 62 (Olympia, 6th c.) αἰ δὲ μὰ συνέαν ‘but if they do not act together’, with μά for μή, the very open ē of Elean being written with α (Buck §15); cf. however, Buck no. 61 (Olympia, before 580) αἰ ζὲ μέ̄.

(15) On the Murder of Herodes 21 ὅτι <οὐ τῃ̑ ἐ>μῃ̑ προνοίᾳ μα̑λλον ἐγίγνετο ἢ τύχῃ ‘that these things happened not by design <on my part> but rather by chance’, with Jebb's conjecture, for ὅτι μὴ προνοίᾳ of the manuscripts.

(16) This is from speech 6, On Sending Reinforcements to those in Sicily: the opposite argument, §33 Lenz & Behr (p. 582 Dindorf). In §40 of the same speech, the use of μή for οὐ averts hiatus, but that is not the case in this passage; see Pernot (1981: 108, 126), and KG II, 447 n. 1.

(17) Holton et al. (1997: 423) mention and illustrate only οὔτε … οὔτε.

(18) Furthermore, unlike usual μή in modern Greek, this μή will usually not appear as μή ν, even if the next word begins with a vowel or voiceless consonant; cf. Holton et al. (1997: 423).

(19) For some Vedic examples, see Whitney (1889: §982cd); on Avestan, Reichelt (1909: §§684, 698).

(20) See further Schwyzer & Debrunner 589, and Chantraine, s.v. ‘οὐ̑ν’.