in Homer, αύλαξ, ἁβρός, δαπάνη, δρόσος, κοινός, μελέτη, κωμάζω, OKλnpós. These words are all found in Hesiod and Pindar and, for the most part, are of Indo-European origin.

Hesiod alone has over one thousand words not found in Homer, and he uses both the Homeric dialect and meter. Few indeed of the words used by Hesiod could have come into being in the brief interval between the composition of the Odyssey and the Hesiodic poems.

There can be no doubt that Homer had control of a vocabulary vastly greater than appears in his poems. Nearly one thousand words are saved to Homer by a single appearance in the Iliad; most of these are unquestioned and must have been in good and frequent use at the time of Homer. There are fewer words beginning with in Homer than with any other letter, yet the following åra§ λeyóμeva— ψάμμος, ψεδνός, ψευδάγγελλος, ψευδής, ψευστέω, ψεύστης, ψηλαφάω, ψηφίς, ψιάς, ψύχος, ψύχω, ψωμός-beginning with that letter are as well-attested words as there are in Homer. Over one-half of the words beginning with ψ are ἅπαξ λεγόμενα.

These facts seem to me to take away all force from the reasoning of those scholars who object to the use of deî with infinitive and subject accusative or of ws as a preposition with the accusative, because Homer has them each but once, or to λóyos or to wσTE with the infinitive, because they are each used but twice.

CONCLUSION

Nothing in Homer is to be rejected on the ground that it is used rarely or but once, while even the absence of a word or construction is in itself no proof that the word or construction was unknown or out of favor.

We know, for example, that Emerson was not acquainted with the "Kodak," not because he does not employ the word, but because we have outside evidence that both the word and the thing came into being after his death, but we are not justified in drawing a similar conclusion if the word is not found in the writings of Stephen Phillips or of Alfred Noyes, and thus similarly Homeric silences, unsupported by external evidence, furnish no proof of the poet's knowledge or choice. The use or omission of a word or construction is largely a matter of accident, so that statistics can only show us what the poet

used; they cannot show the range of his potential vocabulary or syntax. We must actually know from some outside source that Homer disapproved of, or was ignorant of, the things he does not mention or of the constructions he does not use before we can draw any important conclusions therefrom.

The word ovos, used but once, is as truly an Homeric word as iOS, which is used over 460 times, while the word pódov, which is not used at all, must have been as familiar to the poet as either.

NORTHWESTERN UNIVERSITY

THE CONFUSION OF THE INDIRECT QUESTION AND

THE RELATIVE CLAUSE IN LATIN

BY A. F. BRÄUNLICH

It has often been observed that the indirect question and the relative clause are sometimes indistinguishable in Latin.1 Examples like Plautus Amph. 133, "Quae illi ad legionem facta sunt memorat pater Meus Alcumenae," are common enough in the Latin of all periods. In these examples it is a matter of indifference whether the dependent clauses be regarded as indirect questions or as determinative relative clauses. It is not so generally recognized that the indirect question and the relative clause were sometimes actually confused. Yet instances of such confusion do occur in the writings that have come down to us. See, for example, Cicero Phil. ii. 21. 50: "Accipite nunc, quaeso, non ea quae ipse in se atque in domesticum decus impure et intemperanter, sed quae in nos fortunasque nostras, id est in universam rem publicam, impie ac nefarie fecerit." "Ea quae. . . . fecerit" is apparently a contamination of "ea quae fecit" (determinative relative clause) and "quae fecerit" (indirect question).

Such examples, though not numerous, are interesting in their bearing upon text-criticism. Because of failure to recognize this kind of confusion, "emendations" have sometimes been suggested or actually adopted. So for the passage just cited Campe proposed a change of fecerit to fecit. "Ich bin der Ansicht," he declares, "dass man sich in diesen Dingen ein Herz fassen. . . . sollte." Of Verres Actio II, iii. 40. 92 (cited p. 63) C. F. W. Mueller remarks: "Miror nemini in mentem venisse sic corrigere: Audite litteras quas misit"; and of Phil. xiv. 3. 9 (cited p. 71) he says that

1 Cf., e.g., Kroll, "Der lateinische Relativsatz," Glotta, III (1910–12), p. 5.

2 The determinative clause is "the clause which points out what person or thing is meant." Cf. Hale, The Cum-Constructions (Ithaca, New York, 1887), p. 85, German translation (Leipzig, 1891), p. 94; Hale-Buck, Latin Grammar (Boston, 1903), p. 260, n. 1, and p. 294, n. 1.

"Zu Cicero," Philologus, X (1885), p. 631. [CLASSICAL PHILOLOGY XIII, January, 1918] 60

Ernesti changes eaque to atque "recte ut videtur." In De Div. i. 38. 82 (cited p. 70) Mueller reads sunt in place of sint. "Nirgends," he says, "kann ein abhängiger Satz, der halb nach einer Frage, halb nach einem Relativsatz aussieht, unzweideutig zum ertsen gemacht werden, indem man ein vorhergehendes einfaches Pron. dem., auf welches sich das Relativum bezöge, streicht. Man sagt nicht id quod oder id quid gestum sit scio, wenn nicht etwa bei dem letzten Beispiel id so viel sein soll wie tantum, also im Plural: id scio, quae gesta sint."2

3

Although the confusion of the indirect question and the relative clause has been noticed by a number of scholars, the examples have never been collected. In the hope that such a collection might have some influence toward establishing a more conservative attitude toward our Latin texts, I have brought together all the instances that I could find. The collection does not pretend to be complete. Doubtless a prolonged search, through manuscripts as well as editions, would reveal more instances. The collection includes clear examples of confusion and also examples in which the use of the subjunctive mood may possibly be otherwise accounted for. The various possibilities of interpretation are mentioned. Manuscript variations that are given in the standard editions are reported.

1 Critical note on Cic. Rosc. Am. 34. 95 (ed. of Cicero's works (1893–98], Part II, Vol. I, p. 62, line 8).

• Review of Baiter-Halm, N.J. für Phil. u. Paed., LXXXIX (1864), p. 629. 'By Hofmann-Andresen, "Ausgewählte Briefe" of Cicero (2d ed., 1885), on Fam. xi. 28. 2; Bonnet, Le Latin de Grégoire de Tours (Paris, 1890), p. 676, n. 3; Kraner-Dittenberger, ed. Caesar, Bellum Gallicum (1890), on vii. 3. 3; Sargeaunt, ed. Phormio (1914), on vs. 845.

Kroll, op. cit., pp. 4 ff., suggests that Trin. 373, "Scin tu illum quo genere gnatus sit," may show confusion of the indirect question and the relative clause. He thinks that the sentence arose by contamination of "Scin tu illum (:) quo genere gnatus est?" and "Scin tu quo ille genere gnatus sit?" and remarks: "Hier mag der erstere Typus frühzeitig als Relativsatz empfunden sein, wenn er es auch ursprünglich nicht war' (p. 5). I see no reason for regarding Trin. 373 as anything else than an indirect question with prolepsis (for prolepsis see below, p. 64, n. 1). Furthermore, in Kroll's imaginary example with the indicative mood, "quo. est" would never, it seems to me, be felt as a relative clause. Quo genere cannot refer to illum as an antecedent; and to supply eo genere would not be natural.

...

Some of the examples are taken from Mueller's review of Baiter-Halm (see above, n. 2) and cross-references in the critical notes of Mueller's edition of Cicero; Hofmann-Andresen on Fam. xi. 28. 2; Fügner, Lexicon Livianum (Leipzig, 1897), 8.0. audio; Schmalz, Lateinische Syntax (4th ed., Munich, 1910), p. 658. The rest have been gathered in the course of my own reading of Latin authors.

CLASS I.

Cic. Phil. ii. 21. 50:

EXAMPLES

THE ANTECEDENT IS A PRONOUN

Accipite nunc, quaeso, non ea quae ipse in se atque in domesticum decus impure et intemperanter, sed quae in nos fortunasque nostras, id est in universam rem publicam, impie ac nefarie fecerit.

Cic. Rep. i. 13. 19:

Ain vero, Phile, iam explorata nobis sunt ea quae ad domos nostras quaeque ad rem publicam pertineant? Siquidem, quid agatur in caelo quaerimus.

pertinent m. I, and most editors (but not Mueller). Skutsch, in Glotta, III (1912), thinks that the rhythm proves that the indicative is right. But see below, p. 73.

Cic. De. Orat. ii. 39. 166:

Et causas rerum vestigabimus, et ea quae ex causa orta sint, et maiora paria minora quaeremus.

sunt L.

A little later, in 40. 171, this passage is picked up in the words: "Ex iis autem quae sunt orta de causis." In the latter passage there is no possibility of confusion with the indirect question; and the indicative, the regular mood of the determinative clause of fact, is employed. Cic. Fam. iii. 10. 11:

Nunc ea quae a me profecta quaeque instituta sint, cognosce. sint M, Mendelssohn, Mueller; sunt G R, Baiter, Wesenbach, Tyrrell-Purser (1890).

Caelius in Cic. Fam. viii. 13. 1:

Non est enim pugnax in vitiis neque hebes ad id quod melius sit intellegendum.

melius sit M H3, Mueller; est Wesenbach, Tyrrell-Purser. The Tyrrell-Purser edition (1890) comments: "The ordinary reading, melius sit, probably arose from meliust; no account can be given of the subjunctive."

Author of Varro R.R. Capitula Libri Primi (Goetz, p. 3):

De eis quae extra fundum commoda fiant aut incommoda.

fiant V B; fiunt A, Victorinus, Goetz.