In 3 follows the construction of cava aedium, with the five kinds of roof construction. In 4 he continues:

Atriorum vero latitudines et longitudines tribus generibus formantur. Et primum genus distribuitur uti longitudo cum in quinque partes divisa fuerit tres partes latitudini dentur, alterum cum in tres partes dividatur duae partes latitudini tribuantur, tertium uti latitudo in quadrato paribus lateribus describatur inque eo quadrato diagonios linea ducatur et quantum spatium habuerit ea linea diagonios tanta longitudo atrio detur. Altitudo eorum quanta latitudo fuerit quarta dempta sub trabes extollatur reliquo lacunariorum et arcae supra trabes ratio habeatur. Alis dextra ac sinistra latitudo cum sit atrii longitudo ab triginta pedibus ad pedes quadraginta ex tertia parte eius constituatur. Ab quadraginta ad pedes quinquaginta longitudo dividatur in partis tres semis, ex his una pars alis detur. Cum autem erit longitudo ab quinquaginta pedibus ad sexaginta quarta pars longitudinis alis tribuatur. Ab pedibus sexaginta ad octoginta longitudo dividatur in partis quatuor et dimidiam, ex his una pars fiat alarum latitudo. Ab pedibus octoginta ad pedes centum in quinque partis divisa longitudo iustam constituerit latitudinem alarum. Trabes earum liminares ita altae ponantur, ut altitudines latitudinibus sint aequales. Tablino si latitudo atrii erit pedum viginti dempta tertia eius spatio reliquum tribuatur. Si erit ab pedibus triginta ad quadraginta ex atrii latitudine tablino dimidium tribuatur. Cum autem ab quadraginta ad sexaginta latitudo dividatur in partis quinque ex his duae tablino constituantur. Non enim atria minora cum maioribus easdem possunt habere symmetriarum rationes. Altitudo tablini ad trabem adiecta latitudinis octava constituatur. Lacunaria eius tertia latitudinis ad altitudinem adiecta extollantur. Fauces minoribus atriis e tablini latitudine dempta tertia, maioribus dimidia constituantur. Imagines item alte cum suis ornamentis ad latitudinem alarum sint constitutae.

Latitudines forum ad altitudinem si Doricae erunt uti Doricae, si Ionicae erunt uti Ionicae perficiantur, etc.

"The length and breadth of atria are constructed in four styles : first, length to breadth, as 5:3; second, 3:2; third, √2: 1, or the diagonal to the side of a square. The height should be the breadth less (quarta dempta) up to the beams in the ceiling or the architrave, where there are columns. . . . For the alae on the right and

left the breadth should be determined when the length of the atrium is 30 to 40 feet at part of it; from 40 to 50 feet the length should be divided into 3 parts, and one of these be given to the alae (ie. 4). When the length is from 50 to 60 feet, part of the length should be assigned to the alae. From 60 to 80 feet the length should be divided into 4 parts, and so on. The beams of the alae should be placed so high that the height shall be equal to the breadth."

"The tablinum, if the breadth of the atrium is 20 feet, taking off of this distance, should have the rest assigned to it. If it is 30 to 40 feet, the breadth of the atrium should be assigned to the tablinum; if from 40 to 60 feet, should be given."

The reason is then given for this sliding scale. Then he continues: "The height of the tablinum to the beam should be made with added to the breadth. Its ceiling should be raised with of the breadth added to the height. The fauces should be made in smaller atria, with part taken from the breadth (i.e. of the tablinum), in larger with. The imagines should be placed with their ornaments as high as the breadth of the alae. The width of the doors, in proportion to their height, should be as Doric, if they are Doric, etc. The opening of the compluvium should be left not less than nor more than of the width of the atrium, its length in proportion to the length of the atrium." Then peristyles, then various closed rooms.

The first question is as to the meaning of fauces. I have looked up the use of the word in all the passages I could find, and I think the history of the word throws some light on the present investigation. In the figurative use the notion in a Roman mind was evidently that of an entrance, just as we use mouth, a hole leading into (or out of) something, like os and ostium. Cato uses it of the mouth of a limekiln; Lucretius says it is the word for the Greek κpaτýp as applied to a volcano, fauces perhibemus et ora (VI. 702).

Ennius, in a passage quoted by Cic. de Div. I. 48, says of the carceres of the circus "pictis e faucibus." Cicero uses patefactis terrae faucibus (N. D. ii. 37) of a supposed opening of the earth, an idea evidently derived from the bursting out of a volcano. In two other places he uses it of positions that command an entrance, as of Corinth, posita in angustiis atque faucibus Graeciae ut claustra locorum teneret, and of Manlius' force in faucibus Etruriae. As Manlius was at Fæsulæ and Arretium, the idea can only be that he holds the key to Etruria, a

use which Sallust follows in Cato's speech on the same subject, but more loosely: Catilina cum exercitu faucibus urget, alii (the conspirators who remained) intra moenia, a use like "thundering at the gates," though perhaps there may be here a feeling also of catching by the throat, as in prehendere faucibus or fauces used figuratively. (Cf. Livy XXI. 9.) Two other passages in which Cicero uses the word will be referred to hereafter. Cæsar uses it twice along with portus for the mouth of a harbor; and this becomes a standing expression in other writers, alternating with ostium and os. The author of the Bellum Alexandrinum uses it once in the sense of a pass; but the passage plainly shows that it means entrance. Quibus ex castris cum locus angustus atque impeditus esset transeundus Pharnaces in insidiis pedites, etc., disposuit, magnam autem multitudinem pecoris intra eas fauces dissipari iussit paganosque et oppidanos in his locis obversari, ut sive amicus Domitius eas angustias transiret nihil de insidiis suspicaretur, cum in agris et pecua et homines animadverteret versari tanquam amicorum adventu, sive ut in hostium fines veniret praeda diripienda milites dissiparentur (B. A. 36). It is to be noticed that this is an eastern pass. Virgil uses the word some ten times in all, five times of the mouth of the infernal world, and in two of these it is coupled with words of entrance. Æn. VI. 273, vestibulum1 ante ipsum primisque in faucibus Orci; and Geo. IV. 467, Taenarias fauces alta ostia Ditis. In two of them again he has a hint at the literal meaning: Æn. VII. 569, ingens vorago aperit fauces; and VI. 241, sese halitus atris faucibus effundens. In Geo. I. 207 he uses fauces Abydi of the Dardanelles (Hellespont), which is like fauces portus, and became the regular use for such places, as in Lucan IV. 594, Threicias fauces; and Plin. N. H. III. 1, faucibus oceani (of Gibraltar).

In Geo. IV. 427 siccis faucibus is used of the fountains of a river, with the common Virgilian play, perhaps, on the literal meaning. There remain two passages where the use seems to approach the later meaning of a pass; but in one of them, Æn. XI. 525, the words are angustae fauces aditusque maligni, which clearly have the notion of an entrance into a larger place, not a pass through some place. In the other (same book, 516) Virgil uses biviae fauces alone, but of

1 Probably vestibulum here means the atrium; but this makes no difference.

the same place; and this passage by itself would hardly seem to mean anything different from the other. In Ciris, 463, the words are deserit angustis inclusam faucibus Isthmum, which may be compared with Cicero's use, where the idea of entrance, as we saw, was clear.

Nonius quotes from Accius (fr. X. Ribbeck): in salti faucibus (cf. examples from Livy below).

The use of the word by Livy is peculiar. In what remains of the earlier books the word does not occur in this sense of pass, even in the description of the Caudine Forks, where we should expect it. In fact, nowhere is a pass in Italy so called, if we exclude the cases in Virgil, above mentioned. In IX. 2, describing the Caudine Forks, Livy says: sed ita natus locus est, saltus duo alti, angusti silvosique sunt, montibus circa perpetuis inter se iuncti. Iacet inter eos satis patens clausus in medio campus, per quem medium iter est. Sed antequam venias ad eum intrandae primae angustiae sunt et aut eadem qua te insinuaveris retro via repetenda aut per alium saltum, etc. Then afterwards, ad alias angustias. So in VII. 39, saltu angusto. In X. 1 he uses fauces of the mouth of a cave.

In Book XXI. 23 occurs the first use of the word for a pass: fauces quae Hispanias Galliis iungunt; and again, XXI. 30, fauces pervias esse, of the Alps; and XXI. 38, fauces saltus. It does not occur again for a long time in this sense, i.e. alone; but in XXII. 4 we have ipsas fauces saltus, and similar phrases often afterwards, and in XXVI. 25 fauces Thessaliae, used of Thermopyla. After this we have it used constantly of Greek and Eastern passes, but not of Italian. From the XXII. book on, Livy uses saltus in this sense 64 times, fauces 29 times, augustiae 22 times.

In only a small proportion of the 29 cases is fauces used alone, distinctly, as a pass. In 11 places fauces saltus is used, or some other phrase indicating that fauces is the entrance, saltus the pass itself; such as, XXII. 15, saltus qui in artas coactus fauces imminet mari (at Terracina); XXXXII. 54, fauces... qua Tempe adeunt; XXXXII. 67, ante ipsa Tempe in faucibus situm Macedoniae. In six places it is used of an isthmus, apparently a regular use. (Cf. Cicero's use, above cited.) In five other places it is used like faucibus Etruriae in Cicero. Thus there are only seven places where it seems to be used like saltus, and some of these approach the narrower meaning.

It would seem as if this use of the word were derived from the Greek múla, which is used of precisely these same places. Especially does this appear from XXXII. 5, quae ad Antigoneam fauces sunt (stena vocant Graeci). Unless he had some such notion in his mind, Livy would naturally use angustiae, as he does in the earlier books, and of all Italian passes except where he uses saltus. Cf. XXXVI., intra portas loci eius (of Thermopyla).

Q. Curtius uses the word of precisely these passes in Cilicia, which are called múa; as in III. 8, 19, Alexander ad fauces quibus Syria aditur, et Dareus ad eum locum quem Amanicas pylas vocant. Cf. Xen. Anab. I. 4. 4, πύλαι τῆς Κιλικίας καὶ τῆς Συρίας.

Pliny the Elder uses it about 20 times, in all the senses, but not of any Italian place. So, also, Lucan; and the use becomes general. But I have not been able to find any use of it for Italian places, except in fauces saltus and similar phrases, above referred to.

The two passages from Cicero which I have left untouched are those in which he speaks of faucibus macelli. We do not know the construction of this market, but we know that it was a building; and, for such a purpose, it must have had a large open court in the centre, and was, no doubt, surrounded by shops opening out (or by shops opening in), like the so-called Augusteum, at Pompeii, which has been by many supposed to have been a market, though the internal arrangements make that doubtful. This building has two passages into it, somewhat like the entrance to the Music Hall in Boston, though not so long. This is doubtless what is meant by Cicero. Thus we have two passages in Virgil which clearly refer to an entrance to a house, and two in Cicero which probably do, as well as one in Gellius (see later), and then the general notion of the word, which was clearly that of an entrance, and continued to be so, even after the word had come to be used for a pass. As in Vitruvius' description, the word must be used to denote the ordinary, though apparently not the literary, name of the place referred to (on account of the nature of his book), it seems almost certain that he uses it with the old meaning of an entrance into one place, rather than a passage between two.

The passage in Gell. XVI. 5 is directly on the point, being a comment on Virg. VI. 273, but is, unfortunately, somewhat obscure, on account of his uncertainty as to the meaning of vestibulum :