48

THE IONICVS A MINORE OF HORACE

regards as significant are such as might be got by writing' operosae que déae méns' for 'operosaeque Mineruae,' or male morálta fúgit Bellerophontem,' if it is conceivable that in any circumstances Horace would have perpetrated such a thing. Now it is to be observed that Horace was writing a poem, not arranging an exhibition of all possible variations in an ionic foot. What he admits we know; but for laying down what he excludes, the material we have, forty feet in all, is far too scanty. Verse endings such as 'déae méns' violate the canon of coincidence of ictus and word accent which the school of Professor Goodell discovers in the last foot of the hexameter; and if all of Vergil had been lost except the last four-fifths of Aeneid I., fifteen times forty cases could have been adduced to show this ending was un-Vergilian.

It is impossible, then, to prove that Horace avoided this form of the ionic foot. But that he did avoid it is possible. And this, though a mere hypothesis, shall presently be considered. The suggestion that the coincidence of a principal wordaccent with the metrical ictus of the third syllable of the ionic foot makes that syllable more prominent, and may thus give to the reader, if he needs it, 'a help to the perception' of an 'unusual rhythm,' is not in itself absurd. But what help is it to him to know that the accent must not fall elsewhere than on one of the three beats of the ionic foot? In agitáto' (I mark accented syllables by italics) the accent may be helpful; but dare lúdum' is distracting, and in 'patruáe uer-,' 'meliór Bel-,' is it aught but a hindrance? If Horace designed a coincidence of accent and ictus which should be a guide to the reader, why does he now on the one hand leave the syllable upon which the ictus falls without a word-accent, and now on the other hand allow accents on two at least of the syllables upon which it does not fall? Surely there is but one conclusion. In this ionic metre at any rate the word-accent was immaterial.

I have already hinted that to understand Horace's handling of this metre we should in the first instance look not to the choric odes of the Greek drama but to those of his avowed models, the lyrists, of which Hephaestion Enchiridion 71, 72, has given us specimens from Alcman

and from Alcaeus,

Εκατον μὲν Διὸς υἱὸν τάδε Μῶσαι κροκόπεπλοι,

Ἔμε δειλὰν ἔμε πασᾶν κακοτάτων πεδέχοισαν.

Studying these examples, and disembarrassed of the accentual fantasy, we realize the probability that in Horace's treatment of this metre there were definite limits to the disintegration of the ionic foot by caesuras. Everyone of the forty feet in Horace, and of the eight feet in the Greek examples quoted above, satisfies the condition that the ionic foot shall be either (A) unbroken, or, if broken, then (B) broken at not more than a single point. The admitted varieties are (a) uu--, ' miserarumst'; (b) --, neque dulci'; (c) uu-l-, ' patruae uer-'; and (d) u|u--, of which there are two examples, que Mineruae' and '-uit in undis.' For since a preposition was pronounced with its following case, 'per apertum' belongs to the first type, and in 'nequ(e) amoris' the elision ('coniunctio uocalium ') binds the words together. In ́ lauer(e) aut ex-' there is nothing exceptional; for there is no breach at the elision and the unaccented 'aut' is pronounced with the following word. But (operosae)que deae mens' breaks the foot in two places. That it would disintegrate its unity we can both see and hear; and if Horace did avoid it, this avoidance is now perfectly comprehensible. It is but a consequence of the rule of the metre, which we may formulate anew as follows. In the Horatian ionicus a minore metre each foot must be contained in a single word or divided at one point only, which, however, may be after any syllable in the foot.

J. P. POSTGATE.

GEOGRAPHICAL LORE IN THE LIBER GLOSSARVM.

In the encyclopaedia portion of the Liber Glossarum the compiler introduced numerous historical and geographical excerpts of varying length. The writers from whose works the geographical extracts are primarily taken are Isidore, Orosius, and Eutropius; but though the compiler has in many cases appended the labels ESIDORI, PAVLI HOROSI, or simply OROSI, and EVTROPI to the entries, this is by no means always the case. A few of the excerpts are of great length; thus, the longest of all, Hispania (HI 233), which is labelled PAVLI HOROSI, SVLINI, OROSI, fills a whole column of the Paris MS. (P.).1 Other long geographical entries are IT 12 (Italia),2 GA 52-4 (Gallia Belgica, Gallia Lugdunensis, Gallia),3 and RO 105-8 (Roma). The length of HI 233, coupled with the fact that the next entry (HI 234) fills half a column of P., and contains Isidore's remarks on Spain (Etym. 14, 4, 28-30), has been used as an argument by those scholars who maintain that the Liber Glossarum is the work of a Spanish compiler.5 To this the obvious retort would seem to be that the presence of three passages about Gaul might with equal justice be used to support the view that the compilation was made in France. Other evidence of a 'Spanish hand' in certain geographical items will be given below, but it is of such a character that it adds nothing to the evidence for the vexed question of provenance.

The short geographical glosses are mainly taken from two sources: either they are quotations from the above-mentioned authors and some others (e.g. Jerome), the lemmata being sometimes labelled, sometimes not, or else they are notes on placenames occurring in Vergil. Of these some are labelled VIRGILI, and are, in fact, marginalia from a Vergil MS. used by Ansileubus'; some are marked DE GLS-that is to say, they are Vergil glosses incorporated in the Abstrusa glossary as it has come down to us, or in the earlier and larger version of that work which was used by the compiler of Lib. Gloss., some few again are unlabelled, but can be identified with reasonable certainty as marginalia of a Vergil MS."

There remains a small residue of geographical lemmata which call for a short investigation, and which are the subject of this paper.

Among the textbooks on geography which were favoured in the earlier part of the Middle Ages was the Cosmographia of Julius Honorius, last published by Riese in his Geographi Latini Minores. Of this little work two versions have been preserved :

1 Contains Oros. 1, 2, 69-72; Solin. 23, 1-3; Oros. 1, 2, 73-4; id. 5, 23, 16.

2 Labelled ESIDORI; it contains Etym. 14, 4, 18-9.

3 GA 52 (OROSI) contains Oros. 1, 2, 63 ; 53 (unlabelled)=id. 1, 2, 64-5; 54 (ESIDORI) contains Etym. 14, 4, 25-7.

• RO 105, labelled ESIDORI, but not in his works as they have come down to us; 106 Oros. 2, 4, 1-3; 107. A variant version of Isid. Etym. 15, I, I, and Eutr. 1, 1; 9, 15, though the item is only labelled ESIDORI. 108, Isid. Etym. (with variations) 15, 1, 55; 9, 4, 42.

5 In primis by Goetz; but from his latest work, C.G.L.I., published in 1922, it would seem that

the earlier, called A by Riese, has come down to us in only one MS., but of the slightly later and fuller version (B in Riese) there are several early MSS. The still later adaptation of Aethicus does not concern us here. The Liber Glossarum contains six lemmata, which are clearly taken from a version of the Cosmographia. On the whole they agree most nearly with B, but there are one or two important differences. All these excerpts are explanations of river names, and read as follows:

1

BE 178 B<a>etis fluuius: nascitur in campis Hispaniae, occidit in Oceano occidentali, currit milia trecenta decem.

Cf. Riese, p. 36, B1: 'milia CCCCX B. A omits this entry, and L.G. the sentence, found in B, about the river Singillius.

DV 169 Durius fluuius: nascitur in campis Hispaniae, infundit in Oceano occidentali inter duos oppidos (!). Post hoc currit milia quingenta octuaginta.

Cf. Riese, p. 36, 4, and B 4: 'milia CCCCXCV B. L.G. differs from both A and B, but is nearer to A. Is the barbarous inter duos oppidos' an unintelligent abbreviation of' diuidens Galliciam et Lusitaniam' in B?

HI 39 Hiberus fluuius: nascitur sub Astericis (-ur-) montis Pirinei uicinus inlustrans Hispanias. Infundit se mari iuxta Dertosa <m>. Currit milia ducenta

tria.

Cf. Riese, p. 37, 4, and B 4: 'Astyribus . . . in mari . . . milia CCIIII B.' This gloss is labelled ESIDORI, but the label clearly refers to the next lemma in L.G., which is from Isid. Etym. 13, 21, 31. montis has apparently been corrected to montes in L.G. In one important particular L.G. differs from both A and B; both the latter read Tarracona.

ME 3 M<a>candros: fluuius qui nascitur in campis Mesiaticis (As-), bicornius currit quasi sint duo redigentes se ad unum. Influit in mare Cicladum (Cy-). Currit milia octoginta tria.

Cf. Riese, p. 45, 1, and B 1: 'milia DCCCXCVII B,' otherwise B and L.G. are identical.

TA 81 Tagus fluuius: nascitur in montibus Hispaniae, occidit in Oceano occidentali. Currit milia sescenta decem.

Cf. Riese, p. 36, 2, and B 2: campis Hisp. A, campo B, milia CCCCII B' (but there is some discrepancy in the MSS.). In montibus' in L.G. is a great improvement on the other versions. The phrase in campis (or -o)' recurs regularly in the Cosmographia, a fact from which Kubitschek would deduce that the maps available did not distinguish clearly between high- and low-lying territory. The correction of L.G. is the more noteworthy, because the Tagus rises at an altitude of nearly 6,000 feet, so that 'in campis' is singularly inappropriate.

TI 13 Tiberis: Italiae fluuius nascitur ex monte Appen<n>ino; occidit Tyrr<h>eno mari iuxta Romam. Currit milia XDCCCCLXXXVIII.

Cf. Riese, p. 39, B 8: 'in mari . . . milia CCCC B.' A omits the entry. The addition of Italiae and iuxta Romam in L.G. shows that our compiler was proud of his Italian geography. The distance in both versions is of course fantastically wrong, since the real length of the Tiber is 245 English miles.

1 The two versions of the Cosmographia will in the sequel be referred to as A and B, the Liber Glossarum version as L.G. The italics indicate

verbal agreement between L.G. and A or B, or both.

Taken as a whole, the version in L.G. shows more knowledge of Spain than A and B. The reference to Dertosa in HI 39 is particularly striking. This town is situated on the Ebro, and from here to the river mouth, a distance of twenty-two miles, the river is navigable. Tarracona, on the other hand, is over forty miles from the estuary. With regard to the estimates of length, the honours are even between the two versions. L.G.'s estimate for the Tagus, 610 miles (circa 575 English miles), is astonishingly near the true figure, which is 566 miles. But in the case of the Durius and the Baetis, whose real length is 485 and 375 English miles respectively, B is somewhat more accurate. The real length of the Ebro is about 450 English miles, so that both versions are hopelessly wrong; the same is true of the estimates for the Maeander and the Tiber.

The question how this version of the Cosmographia came in the first instance to be inserted in the Liber Glossarum cannot be answered with complete certainty. That Julius' work was popular is known, and one of the main reasons for this was doubtless, as Kubitschek suggests, due to Cassiodorus' advice to monastic pupils to improve their minds by reading it. Moreover, it is certain that the MS. of Isidore used by the compiler of the Liber Glossarum belonged to the 'interpolated' or Spanish family. Hence it may be surmised that, just as the chapters on Rhetoric and on gems in this Isidore MS. seem to have contained additions from Julian of Toledo and Pliny, so the geographical portions had some additions from Julius Honorius. The accurate knowledge displayed about Dertosa and the mouth of the Ebro-though not its length-would be quite in keeping with this explanation, and does not therefore form an additional argument for the Spanish origin of Lib. Gloss.

There are thirteen further geographical glosses whose source cannot be determined with certainty. It will be convenient to consider these in three groups for reasons that will become obvious in the sequel:

I.

AU 172 Aufidus: amnis Italiae qui in Adriaticum mare orion (?) funditur (fundatur PF).

This may well be a garbled version of a scholium on Verg. Aen. 11, 405. The last word but one was evidently corrupt in the archetype of L.G., since all the MSS. concur.

CA 646 Carcharidus (Gangaridum): amnis Armeniae.

The item is labelled DE GLS, and a comparison with GA 94-5 and NI 50 shows that the interpretation really belongs to Niphates (cf. Thomson, op. cit., p. 89). The confusion must be put down to the Abstrusa compiler, since the mistake would arise when the marginalia on Verg. Geo. 3, 27 and 30, were transferred to the glossary. In his zeal for double entry and cross-reference the compiler combined the lemma word of one entry-misreading it in the process-with the interpretation of another.

MI 85 Minci<u>s: amnis Galliae Mantuam praeterfluens.

The corrector of A(mbrosianus) makes the quaint addition Ergo Mantua est Galia' (cf. Verg. Ecl. 7, 12, or Geo. 3, 15). This seems to be a mere variant of the next lemma (MI 86)- Mincius: amnis Mantuam praeterfluens, ingens, flexuosus'— which Thomson (op. cit., p. 70) claims for the larger Abstrusa glossary.

NO 77 Nola: ciuitas Campaniae.

Perhaps a marginal comment on 'uicina Vesaeuo ora iugo' (Verg. Geo. 2, 224) in the Vergil MS. used by Ansileubus.'

1 Cass. Instit. diu. script. 25. The passage is quoted in full by Kubitschek (op. cit., coll. 617-8).

PA 58 Padus: amnis qui inter Italiam Galliamque oritur; profluit ad

Orientem uersus prope Rabennam.

Possibly from a Vergil scholium on Aen. 9, 680; this is more likely than that the gloss is a shortened form of Isidore, Etym. 13, 21, 26.

IS 9

II.

Isara fluuius Galliae influens R<h>odano.

RU 244 Rutupina: urbs Brit[t]an<n>iae.

VA 179 Varus: fluuius qui ex Ligurtinis (-ust-) Alpibus fluit in mare Tyrrhenum.

The last of these three glosses is labelled DE GLS, the other two are unlabelled. They have been grouped together because they may all go back to the same source. The form of the second lemma-word is curious, for instead of Rutupiae the derivative adjective in the feminine singular or neuter plural is given. But if this is a note on Lucan (6, 67), Rutupinaque litora feruent,' the mistake is intelligible; the scholiast treated the adjective as a noun. Pliny refers to the Varus several times, and is followed by Mela, but there is no similarity between the gloss under consideration and these passages; nor can Florus be regarded as the source. Where Pliny is the source of a Lib. Gloss. item, directly as in the case of certain lemmata dealing with gems, or indirectly via Isidore, the verbal correspondence between the original and the derived gloss is always sufficiently near to leave no room for doubt. Vibius Sequester in his list of river names includes the Varus (Riese, op. cit., p. 152, 9). He is not the source of Lib. Gloss., but, as is well known, his lists of river and mountain names, etc., are taken from sundry classical poets, amongst them Lucan. Further, in Vibius is the entry Isara Galliae currit in Rhodanum' (ib. 149, 11). These two river names come from the same passage of Lucan (1, 404 and 399), and the suggestion may thus be hazarded that these two lemmata in Lib, Gloss. are like RU 244 marginalia from a Lucan MS. The further question how such Lucan marginalia found their way into Lib. Gloss. is one of great difficulty, for one reason, because one cannot be sure that the label against the third item is not misplaced. That Ansileubus' used a Lucan MS. is not a feasible assumption, for then we should expect to find a far greater number of Lucan glosses. It is of course well known that many glosses in Lib. Gloss, contain quotations from Lucan, but these probably come from a Vergil MS. containing parallel passages in the margin. the three glosses under consideration got into the Abstrusa glossary in this way, or, alternatively, direct into the Lib. Gloss. from a Vergil MS.-in which case the label is wrong-it is hard to see what Vergilian passages could have been illustrated by these Lucan extracts, which then would have provided the material for separate glosses. If the label against VA 179 is misplaced, ' Ansileubus' may have got his information at second hand-e.g. from the interpolated version of Isidore. But for the present the answer to this question must remain in doubt.

III.

If

CI 67 Cydnus: fluuius Siciliae (Cilic-) e Tuuro (Tau-) monte means (et muro P). The item is labelled VIRGILI, but is misplaced, for the next lemma contains a Vergil quotation. The Cosmographia (Riese, p. 44) mentions this stream, but there

1 That the lemma is a gloss on Ausonius, Parent. 7, 2 ('tellus Rutupina '), is quite unlikely ; the case for Lucan is strengthened by taking the three glosses together; and, besides, Ausonius is not an author found in the 'quotation' glosses

as Lucan is.

2 E.g. N.H. 3, 35 ('amnis Varus ex Alpium monte Caenia profusus'); Mela, 2, 72; Florus, I, 19, 4.

* On this question cf. Thomson, op. cit., p. 54.