Author of Varro R.R. Capitula Libri Primi (Goetz, p. 4):

De eis quae seri oporteat et quare legumina appellentur.

appellantur A.

In all of the foregoing passages the natural meaning of ea quae or id quod is "the things which" or "that which" (determinative). To account for the subjunctive by interpreting "things which" or "a thing which" (descriptive1) would be forced. Similarly in the examples of Classes II and III, given below, to interpret the dependent clauses as descriptive is either impossible or unnatural.

CLASS II. THE ANTECEDENT IS A NOUN

A. Antecedent and pronoun are both objects:

Cic. Verr. Act. II, iii. 40. 92:

Audite litteras quas ad Segestanos miserit. Litterae C. Verris. Cic. Leg. Manil. 13. 38:

Itinera quae per hosce annos in Italia per agros atque oppida civium Romanorum nostri imperatores fecerint, recordamini.

B. Antecedent an object; pronoun a subject:

Cic. Leg. Agr. I. 8. 25:

Cum vero scelera consiliorum vestrorum fraudemque legis et insidias, quae ipsi populo Romano a popularibus tribunis plebis fiant, ostendero, pertimescam, credo, ne mihi non liceat contra vos in contione consistere.

C. Antecedent an object; pronoun neither subject nor object:

Ter. Adel. 572:

At nomen nescio

Illius hominis, sed locum novi ubi sit.

The first two examples of Class II have sometimes been understood as indirect questions with an unusual word-order. Thus Richter (ed. 1871) and Deuerling (ed. 1884) explain the order in Leg. Manil. as emphatic. Sure examples of indirect questions with such word-order occur in Plautus: Bacch. 891, "Iam dudum hercle equidem sentio suspicio Quae te sollicitet"; Aul. 778; Curc. 321. There seems to be no sure example in Cicero. Commentators on our

1 For the descriptive clause of fact, with the subjunctive mood, cf. Hale, CumConstructions, pp. 88 ff., German translation, pp. 98 ff.; Hale-Buck, Latin Grammar, § 521, 1.

Leg. Manil. passage refer to Leg. Manil. 2. 6, "Causa quae sit videtis.” However, this passage is not parallel. In "Causa quae sit," causa is the subject, and quae is in the predicate. In the present passages, quae and quas would modify the nouns. Mueller finds the Verres passage difficult, as is shown by his suggested emendation (cf. p. 60).

All the passages of Class II might conceivably be explained as indirect questions with prolepsis or pleonasm.1 In B the subject of the indirect question would be "anticipated" as the object of the main verb. (Cf., as a typical instance of prolepsis, Men. 519, "Uxori rem omnem iam ut sit gesta eloquar.") In A there would be an unusual (and, so far as I know, unrecognized2) sort of prolepsis. Not the subject, but the object, of the indirect question would be anticipated. Leg. Manil. would be interpreted: "Remember the marches, viz., what marches our generals have made." In C Ter. Adel. may conceivably be an indirect question with pleonasm.3 Locum, that is, may pleonastically express the idea of place which is contained in ubi sit. The interpretation would be, "I know the place, viz., where he is." Still, especially for A and C, it seems oversubtle to speak of prolepsis or pleonasm. The difference in feeling between "Itinera, quae . . . . fecerint," for example, and "Itinera quae . . . . fecerunt" must have been, at the most, extremely slight. If there is not, in these examples, actual confusion between indirect question and relative clause, at any rate the indirect question approaches suspiciously near to the function of the relative clause.

ADDITIONAL POSSIBLE EXAMPLES OF CLASS II

In the following examples, more easily than in the preceding ones, the dependent clauses may be understood as indirect questions, with prolepsis or pleonasm. However, these examples, too, may show confusion of the indirect question and the relative clause.

1 For pleonasm and prolepsis, cf. Lindskog, Quaestiones de Parataxi et Hypotaxi apud Priscos Latinos (Lund, 1896), pp. 69 ff. and 75 ff.

? Tincani, in his edition of Leg. Manil. (1889), comments on our passage from this oration: "detto per prolessi." However, as he cites 2. 6, "Causa quae sit," as parallel, he seems not to use "prolessi" in the sense in which "prolepsis" is generally employed. Lindskog, op. cit., p. 83, cites this passage as an example of prolepsis, although on p. 75 he defines prolepsis as the structure "ubi subiectus enuntiati secundarii pro obiecto primarii ponitur" (the italics are mine).

(A).

Varro R.R. iii. 1. 10:

Haec ad te misi, recordatus de ea re sermones, quos de villa perfecta habuissemus.

This passage differs from the preceding passages in that the repetition of de ea re, in the form de villa perfecta, makes it less inevitable to understand the dependent clause as relative. This awkward repetition may well cause the hearer to keep his mind in suspense until he hears the subjunctive habuissemus.

(C).

Cic. Pro Flacco 33. 81:

Habetis causam inimicitiarum, qua causa inflammatus Decianus ad Laelium detulerit hanc opimam accusationem.

Here the presence of qua causa, after inimicitiarum, has much the same effect as de villa perfecta had after de ea re in the preceding example. Inimicitiarum may seem to the hearer to define causas adequately, so that the qua causa clause comes as a surprise and causes him to hold his mind open until he hears the subjunctive. (B).

Ter. Hec. 351:

Omnem rem narrabit, scio, continuo sola soli,

Quae inter vos intervenerit, unde ortumst initium irae.

(D). Antecedent in a prepositional phrase; pronoun a subject: Cic. Ad Quint. Fratr. i. 2. 16:

De singulis tamen rebus, quae cotidie gerantur, faciam te crebro certiorem.

In the last two passages the modifiers omnem and singulis affect the interpretation in much the same way (though to a less extent) as did the modifiers mentioned in the two preceding passages. Their presence makes it somewhat easier than it would otherwise be to regard the dependent clauses as interrogative. If the clauses are so interpreted, there will be an unusual, though not unnatural, sort of prolepsis in Quint. Fratr. The subject of the indirect question will be anticipated, not as the object of the introductory verb (as in the usual kind of prolepsis), but as the principal word of a phrase: "I shall keep you informed of all events: what events occur every day."

(C).

Plaut. Most. 969:

Scio qua me eire oportet et quo uenerim noui locum.

loqui P; "Pro loci ?" Lindsay. This passage differs from Adel. 572, cited above, in that the dependent clause precedes the noun with which it is connected. On this account it seems a little easier to regard the dependent clause in the present passage as interrogative than it was in the case of the Adel. example.

CLASS III. THE ANTECEDENT IS NOT EXPRESSED OR ELSE IS WITHIN THE DEPENDENT CLAUSE

Cic. Inv. ii. 9. 30:

Quae res harum aliquam rem consequantur, faciles cognitu sunt. Edd., including Mueller, emend.

Cic. Lael. 16. 56:

Constituendi autem sunt qui sint in amicitia fines et quasi termini deligendi.

Matius in Cic. Fam. xi. 28. 2:

Nota enim mihi sunt, quae in me post Caesaris mortem contulerint. Caes. B.G. vii. 3. 3:

Nam quae Cenabi oriente sole gesta essent ante primam confectam vigiliam in finibus Arvernorum audita sunt.

Doberenz (ed. 1857) comments: "Der Conjunktiv weil der Gedanke concessiv zu fassen: was doch erst . . oder wiewohl es erst." Later editors generally agree. Bond-Walpole (1887) makes an additional comment: "If quae essent had been the conjunctive of indirect question, auditum est must have stood." This interpretation seems improbable to me. I know of just one edition in which the passage is explained as a confusion of the relative clause and the indirect question-that of Kraner-Dittenberger (1890).

Livy xxi. 21. 1:

Hannibal Sagunto capto Carthaginem novam in hiberna concesserat, ibique auditis quae Romae quaeque Carthagine acta decretaque forent, seque non ducem solum sed etiam causam esse belli, . . . . Hispani generis milites convocat.

It should be noted that in this example auditis is followed by an indirect statement with the infinitive, as well as by the quae-clause.

Livy xliv. 30. 12:

Anicius praetor eo tempore Apolloniae auditis quae in Illyrico gererentur, praemissisque ad Appium litteris, . . . . triduo et ipse in castra venit.

Sargeaunt, on Phormio 845, remarks that Livy often confuses the relative clause with the indicative and the dependent question with the subjunctive. Neither Kühnast nor Riemann appears to have noticed such confusion. The few examples which I cite are taken from the Fügner lexicon, s.v. audio. If this admirable lexicon could have been finished, my collection would be more complete. Sen. Epist. Mor. vi. 1. 4:

Incredibilia sunt quae tulerim, cum me ferre non possim.

Greg. H.F. 5. 36:

Sed quae contra sacerdotes egerit, altius repetenda sunt. In these examples the number and gender of the principal verbs point to the interpretation of the dependent clauses as relative. On the other hand, the mood of the dependent clauses is appropriate to the indirect question and not to the relative clause. The examples appear, therefore, like those of the other two classes, to be mixtures of relative clause and indirect question. Cic. Inv., for example, seems to be a contamination of "Quae res

consequuntur faciles

cognitu sunt" and "Quae res . . . . consequantur, facile cognitu est." It should be noted, however, that there occur in Latin a number (in classical Latin apparently only a small number) of passages in which an impersonal verb in an independent clause is assimilated in number and gender to some word or words in a dependent clause, but in which there is no possibility of confusion of indirect question and relative clause. Cf. Cic. Leg. Agr. ii. 37. 102, "Quae nobis designatis timebatis, ea ne accidere possent, consilio meo ac ratione provisa sunt." It may, then, perhaps be doubted whether the occurrence of the examples of our Class III was a result or a cause of confusion of indirect question and relative clause. Because of the contemporary, or possibly earlier, occurrence of examples of our other two classes, the former supposition seems the more probable one. 1 Cf. Schmalz, Lateinische Syntax, p. 658; Hofmann-Andresen on Cic. Fam. xi. 28. 2.