C. The nominative of the gerund is used with the verb est, and with a substantive in the dative case to express the agent; and though both est and dative are often omitted, the full expression seems to be the original, and best expresses the meaning. Now as est mihi domus means 'there is a house for me' or 'I have a house,' so est mihi agendum means 'there is acting for me' or 'I have an acting.' And since egi or sometimes est mihi (or a me) actum means "I have acted,' 'action has been taken by me,' and ago, or sometimes a me agitur, means 'I am acting,' or 'action is being taken by me,' est mihi agendum can be used without risk of confusion for 'I have action to take,' 'I am to act.' This notion of an action existing, but unperformed, is in fact the notion of an action requiring to be done. The action exists in the imagination, as a spur to the will. The phrase thus acquires or contains the notion of obligation1, but the notion does not lie in the word itself, etymologically considered, but in the predication; and predication may exist in this case as in others, though the word est be not present to express it.

That this notion of obligation is not in the form, but in the predication of an abstract substantive, denoting action, is seen from the similar meaning of the verbal substantives in -ion (cf. § 1400): e.g.

Mihi cautiost, ne nucifrangibula excussit ex malis meis (Pl. Bac. 598), 'I must take care' = cavendumst. So Ter. Andr. 400. Me sinas curare ancillas, quæ meast curatio (Pl. Cas. 2. 3. 43)=quod mihi est curandum.

The use of opus est, usus est is analogous also. Both these expressions have the notion absorbed into the words themselves, so that opus and usus have come to mean not 'work' and 'employment' but 'need.' Usus est filio viginti minis is not what it should literally mean 'some one is employing 20 minæ for my son,' but 'my son requires 20 minæ.’

1 The equivalence of the gerund and gerundive to an infinitive with oportet is clearly seen in such sentences as In cubilibus quom parturient, acus substernendum, quom pepererunt, tollere substramen et recens aliud subicere (Varr. B. B. 3. 9. 8), where tollere and subicere are used as if substernere oportet had preceded. So Quod si dies notandus fuit, eumne potius, quo natus, an eum quo sapiens factus est? (C. Fin. 2. 31), where eum (object) shews that notare oportuit is supposed to have preceded. See other passages in Madvig's note ad C. Fin. 1. c.

Other phrases somewhat similar have obtained a different shade of meaning, that of possibility, e.g. quod versu dicere non est (Hor. S. 1. 5. 87); neque est te fallere quicquam (Verg. G. 4. 447); necnon et Tityon terra omniparentis alumnum cernere erat (id. A. 6. 595). So also regressus inde in tuto non erat (L. 38. 4), 'there was no retracting''they could not retract.' Erat nulla omnino recusatio (C. Rab. P. 10), 'there was no refusing'=it could not be refused; ibi occultatio nulla est (C. Att. 9. 13 § 5), "there no concealment is possible' (Nägelsbach, § 58). How much in these matters depended on the context and on association with particular phrases or turns of language, is seen from comparing est dicere (above) with haæc fere dicere habui (§ 1345), ‘I had to say' = hæc mihi erant dicenda.

There is nothing in the origin of this usage of the gerund which should restrict it either to transitive or to intransitive verbs, and accordingly we find instances of both. But while the gerund from intransitive verbs is at all times frequent, with and without an indirect object, the gerund from transitive verbs is usual, only where no object is specified. Thus serviundum populo est, utendum est populo, eundum est, agendum est, are all normal expressions; but agendum est hanc rem is used by two writers only in more than isolated passages. These writers are Lucretius and Varro, both comparatively early writers, and both lovers of older language than their time. The following list contains all the instances (except in Varro) that I have met with:

PLAUTUS Trin. 869 (quoted § 1398).

AFRANIUS. Optandum uxorem quæ non vereatur viri (ap. Non. p.

496=99 ed. Ribbeck). As however this is a single line, we can scarcely be sure that optandam (as part of an infinitive sentence) was not what Afranius wrote:

LUCRETIUS I. III (quoted § 1398); 138; 381; 2. 492; 1129;

3. 391 (e conj.); 4. 777; 5. 43; 6. 917 (3. 626 and 926 are not properly instances, as the accusative is subject to an oblique predicate and is not a mere object).

VARRO L. L. 7. 4; 9. 48; ib. 111; R. R. I. 6 § 1; II § 2; 12 § 1; 17 § 5; 20 § 1 bis &c.; 2. 2 § 13; § 18; 7 § 11; § 12 &c.; 3. 9 § 6; § 8; § 13 ter &c. (Instances of transitive

gerunds without a direct object expressed, and of the gerundive

used predicatively, are also common.)

CICERO Scaur. 7; Sen. 2 (quoted § 1398).

CATULL. 39. 9 (quoted § 1398).

VERGIL A. 11. 230 (on authority of Donatus and Servius).

SILIUS 11. 562; 15. 105.

QUINTIL. 4. 5 § 17 (Halm reads audenda, against the Ambr. MS. which has audendum).

PAPINIAN ap. Dig. 15. 1. 50 § 3 (admittendum esse distinctionem

&c.).

TRYPHONINUS ap. Dig. 49. 15. 12 § 2.

PAULUS ap. Dig. 2. 11. 12 § 1; 3. 5. 19 § 1 (reddendum rationem intellegi); 46. 2. 19 fin.

If we may suppose, as seems most natural, that these instances of the transitive gerund used with a direct object are but relics of a use much more common once, they form a significant fact in the history of these forms. The eventual disappearance of this use altogether and its rarity in most writers find a ready solution in the rise of the gerundive and its application as a predicate.

D. This predicative use of the gerundive was inevitable, when its use for the oblique gerund had become established, and may very probably have been developed along with that. For virtus colenda est stands in the same relation to virtutem est colendum that ad virtutem colendam does to ad virtutem colendum. Just as the object of the gerund is in some sort frequently an object of that on which the gerund depends, so the object of colendum est would also be a possible subject to est. That is to say, virtutem mihi colendum est and virtus mihi est might together produce virtus mihi est colenda. As soon as this last usage had grown up, and the conception of a passive signification became attached to an adjectival gerund, (as the gerundive would at first appear to be,) the active use (e.g. virtutem est colendum) becomes an awkward-looking anomaly, and is retained only where it is justified by the analogy of other passive forms, i.e. as a so-called impersonal in the third person and neuter gender. And as they said virtus colitur, virtus culta est, and virtuti invidetur,

virtuti invisum est, so while they could say virtus colenda est, they might say virtuti invidendum est. But to say virtutem colendum est would have been as irregular as to say virtus invidenda est; and if virtutem colendum est was to be allowed, there would be risk of people saying virtutem cultum est and virtutem colitur. But such expressions as actum est and agitur were used absolutely, though from a transitive verb; and accordingly agendum est was similarly retained.

There are two points which, though certainly not conclusive, seem to afford some confirmation of this view, that the passive sense of the gerundive is really due to an attraction and not to any original passive meaning.

(a) The agent with passive verbs is regularly expressed by the ablative with ab; the agent with the gerundive is regularly expressed by the dative (§ 1152). There are, it is true, instances of the use of a dative with the passive participle, very few of its use with other parts of the passive verb: and there are instances of the ablative with ab being used with the gerundive; but they are by no means numerous; and moreover it is not absolute, but relative, frequency which has most bearing on this point. Of the relative frequency the best test is the examination of a continuous passage.

In the first book of Cicero de officiis, if I have counted rightly, there are

(a) 22 instances of ab with agent after past participle passive; after finite passive verb; dative of agent after gerundive.

(b) 15 (c) II

In the second Philippic the numbers are respectively (a) 31, (b) 12 and (c) 8.

In neither of these writings is there a single instance of the agent being expressed either by the dative after a passive, or by the ablative with ab after a gerundive.

The speech pro lege Manilia happens to contain two instances of ab with the agent after the gerundive. But it contains (in Madvig's edit.) 24 instances of ab with the agent after passive verbs [viz. 17 instances of (a), and 7 of (b)]; 10 instances of (c); and no instance of the dative of the agent after a passive. (I do not reckon instances of videor, for they belong to a different class, § 1148.)

So much for the relative frequency of the usage. As to its absolute frequency, all the instances of the gerundive used with the agent expressed by the ablative with ab, that I have been able to find referred to in grammars and annotations, are the following:—

CORNIFICIUS 1. 3 § 5.

CICERO Or. 2. 20 § 85; 3. 36 § 147; Part. Or.

30 § 103; Verr.

3. 24 § 60; Font. 19 § 42; Cacin. 12 § 33; Man. 2 § 6; 12 § 34; Agr. 2. 35 init.; C. Rabir. 2 init.; Mur. 26 § 54; Sull. 8 § 23; post red. in sen. 12 § 31; Har. Resp. 3 § 5; Sest. 18 § 41; Balb. 3 § 7; Planc. 3 § 8; 32 § 78; Scaur. 22 § 44; Mil. 38 § 104; Phil. 3. 8 § 21; 13. II § 24; 14. 4 § 11; Att. 6. 6 § 4; 10. 4 § 6; 11. 3 § 3; 13. 30 § 2; Fam. 3. II § 3; 13. 16 S2; 15. 4 § II; Brut. ap. C. Fam. II. 20.

LIVIUS 9. 40 § 17 (initium fari ornandi ab ædilibus).

OVID Met. 8. 710,

JUVENAL 12. 14.

In most cases the reason of divergence from the ordinary construction is evident (see § 1147). It will be observed that no instance is produced from early writers1, and that Cicero has most of these instances, as he has of those on p. lxviii.

(b) The other point is that deponent verbs have the gerundive in full use just as much as other verbs. Now deponents are verbs which have received passive (or reflexive) inflexions in order to adapt their original stem to the meaning they are to bear, and cannot consequently make a further use of these inflexions in order to have a passive meaning as well as an active one. And besides the forms with (ordinarily) passive inflexions, which with them have an active meaning, they also take the present and future participles belonging to the active voice, and attach to them active meanings. If therefore the gerundive is really passive, it would be the single passive form to which they gave a passive sense, i.e. the passive of that which appears to us, who translate the deponent by an active verb, to be their ordinary active sense.

1 Ter. Andr. 156 is not an instance.