^{page 388 note 1} ‘The Origin and Development of Latin habeo+Infinitive”, C.Q. lxv (1971), 215–31. I assume that the reader has access to this article, and shall make fewer references to the doxography than would otherwise be appropriate: Mr. Coleman has provided an excellent bibliography (pp. 230 f.).Google Scholar

^{page 388 note 2} All modern discussions are deeply in debt to Thielmann's, P. work (‘Habere mit dem Infinitiv und die Entstehung des romanischen Futurums’, A.L.L. ii [1885], 48–89 and 157–202). He collected more evidence than anyone else had before, and made some important observations which have subsequently been neglected, while presenting the evidence in such a rigidly diachronic manner–this was before de Saussure's time–as to perpetuate in sub sequent discussion and in text-books certain false categories and lines of semantic derivation: see below, p. 396.Google Scholar

^{page 388 note 3} So e.g. Salonius, A. H., Vitae Patrum, 1920, 284 (not well argued) and, more influentially, V. Bulhart in his T.L.L. article on habeo as a sub-section of ‘futurity’ (2457. 67–2458. 12).Google Scholar

^{page 388 note 4} Rose. 100 ‘habeo etiam dicere quern … de ponte … deiecerit’; Acad. 2. 43 ‘quid dicere habeant cur ilia definitio uera sit?’; N.D. 1. 63 ‘de diuis neque ut sint neque ut non sint non habeo dicere’ (translat ing Protagoras' famous remark … ; but that is irrelevant to the accept ability of the Latin locution); Balb. 33 ‘quid habes igitur dicere de Gaditano foedere eiusmodi?’; de or. 10 ‘quid habes igitur de causa dicere?’; ibid. 27 ‘quid habes de orationis praeceptis dicere?’; Div. 2. 136 ‘de nostris somniis quid habemus dicere?’; N.D. 3. 93 ‘haec fere dicere habui de natura deorum.’

^{page 389 note 1} Att. 2. 22.6 ‘de re publica nihil habeo ad te scribere’ ;fam. i. 5a. 3 ‘de Alexandrina re … tantum habeo polliceri me tibi … satis facturum.’

^{page 389 note 2} (a) with dicere: Lucr. 6. 711 (after an explanation) ‘item in multis hoc rebus dicere habemus’; Colum. Arb. 5. 8 ‘sed haec … de toto genere huius arboris habui dicere; nunc …’; Gellius N.A. 14. 1.2 ‘utrum … an … , non habeo dicere’; ibid. 17. 20. 5 ‘habesne nobis dicere … tam apte … compositam orationem ?’ Then in Tertullian and his contemporaries and throughout subsequent Latin, e.g. test. anim. 6. (b) with other ‘inform’ locutions: Hor. Epod. 16. 23 ‘sic placet? an melius quis habet suadere?’ (cf. Fraenke, E.l, Horace, 1957, 44 f.); Ovid Tr. 1. 1. 123 ‘plura quidem mandare tibi, si quaeris, habebam’; Quint. Decl. 7. 12 ‘habeo adhuc ex illo multa referre secreto’; Apul. Fl. 6 ‘qui nihil habet adferre cur prandeat’. This use too continues in and after Tertullian.Google Scholar

^{page 389 note 3} Ovid ex P. 3. 182 ‘nee te si cupiat laedere rumor habet’ (i.e., neque quisquam tibi male dicere habet); [Quint.] Decl. 16. 5 ‘sic habeo queri, quasi relinquar.’

^{page 389 note 4} The latter pair would only be purely potential if they exhibited the tonic forms: ‘what can I do … ?’, ‘What could I do?’.

^{page 390 note 1} Thielmann put great weight on the example from Seneca Rhetor as the earliest example of the ‘must’-habeo, but he made rather heavy weather of it, insisting that the idea of ability was irrelevant to its growth, and treating it as an entirely separate structure. There is this justification for separating the examples discussed from the others at the early period, that the distri bution is different; but it is wrong to treat the locutions as separate in and after Ter- tullian, when there is no longer any distri butional difference between different senses of the structure. It is as if one were to give rigidly separate diachronic accounts of the prehistorically distinct components of the subjunctive or perfect systems of Latin, insisting that the meanings of those separate components remained separate after a single formal system had established itself. I wonder whether the exclusive opposition in German of the sentence-types ‘was kann/ konnte ich tun?’ and ‘was soll/sollte ich tun?’ has not been behind the generally rather unsatisfactory categorization of the material by German scholars.

^{page 390 note 2} D. 48. 5. 16. 3 ‘neque enim laborare habet ut se repraesentet.’

^{page 390 note 3} Strat. 1. 5. 1 ‘Q,. Sertorius in Hispania cum a tergo instante hoste flumen traicere haberet’ (tmeret Hartel, Gundermann) ‘uallum in ripa eius in modum cauae lunae duxit.’ See further below, p. 394.

^{page 390 note 4} G.L.K. 7. 49. 20 ‘ut iam in ambiguita- tem cadat utrum per i quaedam habeant dici an per u’.

^{page 390 note 5} At Statius Th. 6. 159 f. ‘nee uos inces- sere luctu orba habeo’ (abeo pars codd.) the correct reading is aueo (Muller, Ellis), and at Val. Flacc. 1. 671 f. ‘tollique uicissim pontus habet’, Schott's auet should be printed; cf. Hor. C. 4. 11. 7 for the inanimate subject and passive infinitive (cited below). We have already seen enough to know that these two alleged and in any case aberrant habeo- structures have as much right to a place in the grand style of Silver Epic as would a glottal stop in Milton; we shall presently see that they both use aueo elsewhere, and that it fits their manner; we may add that it makes sense, whereas commentators and lexico graphers seem reluctant to explain exactly what habeo is supposed to mean in either context: ‘can’ or ‘must’ will hardly do.

^{page 391 note 1} I take my material from Bickel's T.L.L. article on aueo. It is ungracious to carp at the shortcomings of the less well-organized articles in T.L.L., for it is to bite the hand which feeds us. But it may fairly be said that Bickel's chronological arrangement of the material effectively obscures important facts about the usage of the verb, and that his statement ‘uerbum est poetarum … ab ENNIO adsumpsisse uidetur cic, a posterioribus poetis LIV. ; notandum quod deest apud VERG., ergo rarum est apud inferiores’ in volves two plain and two tasteless errors of interpretation. Kiessling (on Horace C. 4. 11. 8) rightly allowed the verb a place in the Classical sermo familiaris.

^{page 391 note 2} Lucr. 3. 957 ‘semper aues quod abest’, 1082 f. ‘dum abest quod auemus … aliud … auemus’; Hor. S. 1. 1. 94 ‘parto quod auebas’. Such later examples as Statius Th. 7. 11 ‘arma tubasque insatiatus auet (Mars)’, Apul. Fl. p. 96 Oud. ‘hereditatem auebant’ do not correspond strictly to classical norms.

^{page 391 note 3} Cic. Aral. 425 ‘auens epulas ornare’; Lucr. 3. 259 ‘rationem reddere auentem’; . 4. 1204 ‘canes discedere auentes’; Hor. S. 2. 4. 1 ‘non est mihi tempus auenti ponere signa’ (ponere ‘in common’); 2. 6. 99 ‘urbis auentes moenia … subrepere’; Ovid M. 2. 503 ‘accedere auenti’ (Ital.: fugit codd.). Much later in pretentious prose: e.g. Aurel. Vict. 39. 10 ‘imperium auens eripere’ (right pattern), e. 1. ‘Caligula auentibus cunctis deligitur’ (false pattern).

^{page 391 note 4} Varro L.L. 6. 83 ‘aures ab aueo, quod his auemus di<s>cere semper’ (cf. Ciceronian usage, quoted below, p. 392 n. 1) ‘quod Ennius uidetur etymon ostendere uelle in Alexandra’ (tr. 47. 2) ‘cum ait “iamdudum ab ludis animus atque aures auent auide exspectantes nuntium”’; id. Men. 78 ‘quibus instabilis animus ardens mutabiliter auet habere et non habere fastidiliter inconstanti pectore’ (with a jingle? see below); Cat. 46. 7 ‘iam mens praetrepidans auet uagari’; Lucr. 2. 265 ‘non posse tamen prorumpere equorum uim cupidam tam desubito quam mens auet ipsa’; Cic. Phil. 5.13 ‘auet animus apud consilium illo pro reo dicere.’

This last was perhaps the locus classicus; at any rate, everyone had the Philippics drummed into them at school, and in the following examples from the ‘revival’ period, the word-order has become fixed: Tac. A. 4. 71 ‘auebat animus anteire statimque memorare …’; Arnob. Adv. Nat. 5. 26 ‘auet animus scire quibus sitis eloquiis tam periculosa negotia defensuri uel artibus quibus habeatis tam confossis salutem dare personis’ (our two structures combined, with a jingle? See above); Symmachus ep. 6. 67 ‘auet animus edoceri …’. Merob. Pan. poet. 107 ‘Aetium … procerum mens omnis auet’ is a false pattern by Classical standards.

^{page 392 note 1} Varro L.L. 6. 83 ‘auemus di<s>cere’ (quoted more fully above, p. 391 n. 4); Cic. Fin. 2. 46 ‘quid … fiat scire auemus’; 4. 52 ‘auent audire cur malum non sit’; Tusc. 1. 16 ‘aueo tamen audire’; 1. 112 ‘aueo enim audire quidquid est’; Tim. 8 ‘si forte … minus id quod auemus, aliquid uidere, audire, addiscere’; Arat. 341 ‘si solis aues … cognoscere cursus’; N.D. 2. 2 ‘aueo audire … quid sentias.’ While the expression is frequent in Alt., it strangely never occurs in Fam.: Alt. 1. 15. 2 ‘ualde aueo scire quid agas’; 2. 18. 1 ‘intellexi quam suspense animo et sollicito scire aueres quid esset noui’; 4. 3. 1 ‘auere te certo scio cum scire quid hie agatur turn ea a me scire’; 4. 15. 3 ‘auere te scribis accipere aliquid a me litterarum’ (i.e., ‘receive information’; this example suggests Atticus used the expression in the same way); 5. 20. 9 ‘quid actum sit aueo scire’; 6. 1. 19 ‘ea quae sint … aueo scire’; 7. 2. 7 ‘Hortensius quid egerit aueo scire’; 7. 3. 9 ‘nunc aueo scire quid hominis sit’; 13. 35. 2 ‘scire igitur aueo quomodo res se habeat’; 13. 39. 2 ‘set tamen scire aueo qualis ei totius itineris summa fuerit’; 13. 50. 3 ‘quantum acceperit prorsus aueo scire’; 14. 7. 1 ‘nam cum reliqua turn de hoc scire aueo omnia’; 15. 19. 2 ‘hoc enim uero nunc discere aueo.’ The structure is, for Cicero at least, clearly colloquial.

^{page 392 note 2} Cic. Sen. 83 ‘conuenire aueo quos ipse cognoui’ and Lucr. 3. 6 ‘te’ (Epicurus) ‘imitari aueo’ are perhaps arguably tangential to the main type, but Cic. Att. 2. 18. 3 ‘non lubet fugere, aueo pugnare’ and 14. 12. 2 ‘itaque exire aueo “ubi nee Pelopi-darum”, inquit’ fall outside it. This is not surprising in view of the artificiality of the restriction of the use of the infinitive with personal aueo (its other infinitival structures being ‘free’) and in view of the subsequent developments.

^{page 392 note 3} habens+ini., which would be parallel to auens+inf., does not appear before Ter- tullian.

^{page 392 note 4} Hor. S. 1. 4. 87 ‘unus auet quauis aspergere cunctos’; Livy 33. 32. 8 ‘cum … uidere nuntium aueret’, 45. 39. 8 ‘conspicere … ingredientem auent’ (both these close to the older usage); C.L.E. 969 SEIQUIS HAUET NOSTRO (e.g. PROPRIUM) CONFERRE DOLORE, ADS1T NEC PARUEIS FLERE QUEAD LACHRYMIS (from this period on orthographic grounds, and note flere queat ═ flere habet ═ flebit).

^{page 392 note 5} The latest example is the extraordinary ‘ara … auet inmolato spargier’ (cf. 5. 1.4. 87, quoted above) ‘agno’, Hor. Carm. 4. 11. 8, the first case of aueo+pass, inf., and of an inanimate subject, and the only case in Hor. Carm. of the archaic inf. in -ier. The whole passage is a most strange mixture of familiar and archaistic elements and I am at a loss to define the tone precisely; see Kiessling ad loc.

^{page 392 note 6} Lucan 4. 265 ‘inopes … cingere … Caesar auet’; 7. 84 ‘scire senatus auet’; Val. Flacc. 1. 671 f. ‘tollique uicissim pontus auet’ (habet codd., as usual, but cf. Hor. Carm. 4. 11. 8); Silius Ital. 1. 61 (abolere), 5. 533 (reponere), 7. 22 (cognoscere), 14. 183 (euellere), 15. 373 (portare).

^{page 392 note 7} Tacitus A. 12. 36 ‘auebant uisere’ (as Livy: see above); Apul. Plat. 2. 16 ‘haurire auet omnia genera uoluptatis’; Apol. 94 ‘litteras ter et quater aueo … lectitare’; Julius Valerius, Ps.-Callisthenes 3. 32 ‘et ipse sane loqui per litteras aueo cum Candace’; i. 7 ‘aueo enim scire’; Auson. 399. 18 ‘auemus agrum uisere’, 178. 32 ‘cetera quae noscere aues’; Paul. Nol. carm. 24. 429, etc.

^{page 393 note 1} Lucretius keeps within the ‘communication’ field, but does not observe the same demarcation: he, like Cicero, uses habeo dicere only for verbs meaning ‘inform’ (6. 711 ‘item in multis hoc rebus dicere habemus’), but allows aueo+inf. not only with a verb meaning ‘be informed’ (2. 216 ‘illud in his quoque te rebus cognoscere auemus’) but also with a verb meaning ‘inform’ (4. 778 ‘plane si res exponere auemus’). It should be borne in mind in all this that we are describing a restriction which only applies to one of four species of aueo-structure.

^{page 393 note 2} Varro has oneaueo audire expression but no corresponding habeo dicere (see above), this in a pointed scholion; and he is a frag mentary author anyway. He (and Catullus and Ennius) use the animus auet structure; but that is irrelevant here, as it does not rhyme with a habeo-structure.

^{page 393 note 3} In the mid-first century B.C.: cf. Catullus 84, Fordyce's comments ad loc., Nigidius Figulus (first century B.C.) ap. Gell. N.A. 13. 6. 3 ‘rusticus fit sermo si aspires perperam’, and Allen, W. S., Vox Latina, 1965. 44.Google Scholar

^{page 393 note 4} No absolute date can be given for either of the relevant shifts, as they developed, it seems, in different places in different social registers. Some places in Plautus might be taken to suggest that in some dialects they had already taken place (Poen. 313 ēsse et bibere ═ ēsse et uiuere?); and Varro's ‘quibus instabilis animus ardens mutabiliter auet habere et non habere fastidiliter inconstanti pectore’ (Men. 78; see above, p. 391 n. 4) looks like a paradoxical jingle exploiting the shifts. It was well established at Pompeii by the mid-first century A.D., as the frequent interchange of B and V in spelling shows. On the other hand a strong case can be made for supposing that the shifts never happened in British Latin: it would not be relevant to present the evidence here. Again, Consentius (fifth century A.D.) implies that he himself used /w/ in a word like ueni (G.L.K. 5. 395). I cannot agree with Allen's view (Vox Latina, 1965, 42) that there is no evidence at all for /β/ before the first century A.D.

^{page 395 note 1} In my opinion, only a minority of the material from Tertullian and later writers classified by Bulhart as potential, obligatory, and future is rightly presented under these exclusive heads.

^{page 396 note 1} In fact, the atonic can [kn] is often so used in colloquial English, and in late Latin posse was a contender as an exponent of futurity.

^{page 396 note 2} The ‘rules’ for the use of shall and will are artificial and defective in many points, but in none more than the failure of gram marians to distinguish tonic and atonic forms. Even today, Dr. Syntax objects to the use of ‘ll in the first person on the un necessary view that this must represent will: Wood, F. T., Current English Usage, 1962, 264 writes, ‘the abbreviation ’ ll should be used only for will. Shall should be written and spoken in full. Sentences like ‘I'll be twenty-one next August’ are often heard in speech, but they are better avoided, and should not appear in writing.‘ But they have been with us for a long while, cf. e.g. Shakespeare Much Ade m. iv. 8 lie weare this. For examples, and an unsatisfactory statement of the rules, see N.E.D. s.w. shall, will, ‘11, ‘Id.Google Scholar

^{page 397 note 1} The collapse of and -β- as -- will have been less damaging than might at first appear to the distinction of the future and perfect systems. For it will only have affected the first person plural (laudabimus/lauda-uimus), and then only in verbs with potentially identical stems (not therefore monebimus/mormimus). All other persons of the perfect will have had distinct forms by the loss of intervocalic at a much earlier stage: laudai, laudasti, laudát (-áut, -áit); It. lodò, Fr. aima, etc.), laudastis, laudanmt. Forms like laudauimus had never been regularly contracted, precisely because the result would be homophonous with the present (laudamus). The only certain exceptions come in one book of one poet (Prop. 2. 7. 2 flemus, 2. 15. 3 narramus, 2. 15. 9 multamus). There was never any objection to the contraction of forms which would remain distinct (Enn. tr. 160 nomas, Lucr. 1. 60 suemus, etc.). See Sommer, F., Laut- und FormenUhre ², 1914, 564, who does not, however, make clear the permissible circumstances of contraction.Google Scholar

^{page 397 note 2} Tert. adv. Iud. 8 ‘quod esset uenturus et pati haberet‘ and Aug. serm. 40. 1 ‘ali-quando Christiani non erunt et idola rursus coli habent’ are typical.

^{page 398 note 1} My thanks are due to Mr. R. Coleman for his courteous and helpful criticisms of an earlier draft of this paper: responsibility for the general thesis, for particular errors, and for misrepresentation of his views (which I especially hope that I have avoided) will however be all mine, and I end as I began, by referring the reader to his paper (see p. 388 n. 1).