^{page 105 note 1} Since Capua joined Rome in 340–338 B.C. and since the colony of Cales was founded in 334, we might even assume that the Via Latina was the first military highroad to Capua. Cf. Papers of British School at Rome, iv. 4Google Scholarseq. The Via Appia is clearly out of the question unless Appius made use of an old track) as it was not constructed until 312 B.C.

^{page 106 note 1} Nissen, , Ital. Land. ii. pp. 818, 819Google Scholar. He identifies this road with the Via Aurelia Aeclanensis, but beyond the republican milestone (C.I.L. ix. 6073) existing at S. Maria della Manna, between Grottaminarda and Ariano di Puglia, which he wrongly refers to this road and not with Mommsen, (C.I.L. ix. p. 602Google Scholar) to the Via Appia, he has no evidence at all for the existence of this road.

^{page 107 note 1} Livy, xxii. 13, 1. Hannibal ex Hirpinis in Samnium transit, Beneventanum depopulatur agrum, Telesiam urbem capit. ‘Hirpinis’ is Weissenborn's reading. Grasso, (Studi, vol. iii. pp. 1–18Google Scholar) wishes to read ‘ex Arpinis.’

Dr. Reid feels a difficulty about the reading ‘ex Arpinis’ upon the score of Latinity. He does not know of any parallel to this use of ex with the name of a town population, not a people. He points out also that one would gather from the reading ex Arpinis that Hannibal went straight to Beneventum. But evidently both Livy and Polybius make him do destruction to the south of it before he came near it. This leads naturally to the supposition that he passed by Aquilonia, which suits the reading ex Hirpinis.

^{page 107 note 2} The difficult question of the Via Minucia is quite unsolved. From Cicero, , ad Att. ix. 6Google Scholar (cohortesque sex guae Albae fuissent ad Curium via Minucia transisse) and from the indications supplied by Caesar, B.C. i. 16 and 24, it is clear that Alba Fucens lay upon it, and from Horace, , Ep. i. 18, 20Google Scholar (Brundisium Minuci melius via ducat an Appi), that it was an alternative route bo the Via Appia between Rome and Brundisium. It is unfortunate that we do not know where Curius, Caesar's partisan, was when he was joined by the six defaulting cohorts. It could not have been another name for the Via Valeria, which, constructed in 154 B.C., led in Strabo's time from Tibur to Alba Fucens and Corfinium. In 48–49 A.D. it was prolonged to the ostia Aterni (C.I.L. ix. 5973) by Claudius as the Via Claudia Valeria. Bunbury, (Dict. Geog. ii. 1282Google Scholar) thinks that the Via Minucia may have been the road described by Strabo (vi. 3, 7) between Brundisium and Beneventum, but it is impossible to reconcile this with Cicero, , ad Att. ix. 6Google Scholar. It may be suggested that the road running from Corfinium on the Via Valeria to Beneventum on the Via Appia through Sulmo, Aufidena, Aesernia, Bovianum Undecimanorum and Saepinum may have been partially or wholly the Via Minucia. Brundisium would then be reached from Beneventum by Strabo's road. Such a road is indicated by the Itineraries (C.I.L. ix. p. 203). It is tempting to imagine that a cross road may have run between Corfinium and Strabo's road, reaching the latter at Aequum Tuticum, but the character of the country no less than the entire absence of evidence, militates against this supposition. Mommsen, says (C.I.L. ix. p. 589Google Scholar) vasta regio quae interiacet inter Valeriam et Latinam Traianamque antiqua aetate viis publicis populi Romani fere caruisse videtur. Cf. also C.I.L. vol. ix. Tab. iii.

^{page 108 note 1} Horace, , Sat. i, 5Google Scholar.

^{page 108 note 2} Strabo, vi. 3, 7, c. 282, 283.

^{page 108 note 3} We must notice that Strabo says ‘ἁμαξήλατος μᾶλλον’ of the Via Appia and must assume that carriage traffic was not out of the question on the other road, which he designates as ‘ἡμιονική’—i.e. more convenient for mules than carriages.

^{page 109 note 1} Satires, i. 5, 77–92.

^{page 110 note 2} Mommsen, (C.I.L. ix. p. 62Google Scholar) and Nissen, (Ital. Land. ii. p. 845Google Scholar) both favour Ausculum. Grasso (Studi di storia antica, 1893, p. 57–75) likewise suggests Ausculum, but later in the appendix to the same edition (pp. 146, 147) he is inclined to place his faith in Herdoniae. This idea he develops in the introduction to Studi, vol. ii. 1896, with the conclusion that Herdoniae must have been the oppidulum, since the correct form (Herdoniae) is clearly unmetrical, much more so than Ausculum. Besides, Ausculum is on a lofty hill three or four miles north of the direct line from the Calaggio valley to Cerignola or Canosa and no ancient road would have gone in and out of Ausculum. On the score of actual distance neither Ausculum nor Herdoniae entirely satisfies the equation. If Horace's ‘vicina Trivici villa’ be located at the summit level on the watershed between the valleys of the Fiumarella and the Calaggio, then Ausculum (29 kilometres or about 20 Roman miles distant) is too near and Herdoniae (43 kilometres or some 28 Roman miles) is too distant. The question is complicated by our uncertain knowledge of the time spent by Horace between Beneventum and Trivicum. If he spent only one day, perhaps we must place his villa somewhere in the Vallone dei Franchi before the summit level is reached, and so Ausculum is more satisfactory than Herdoniae, because by taking the villa farther back the distance to Ausculum more nearly approaches 24 miles. That the ‘vicina Trivici villa’ is a varying factor naturally makes the oppidulum correspondingly variable. It is only a theory of despair to suppose that Horace was mistaken in the distance, and as neither Ausculum nor Herdoniae is convincingly satisfactory, perhaps Horace passed the night at some village whose memorials have perished with it.

Dr. J. S. Reid thinks that one point against Ausculum is that Horace would not have hesitated to contract the name to Ausclum, if he wanted to get it into the verse (so Silius Italicus, viii. 440). As the u between c and l is not represented on the Oscan coins, one may conjecture that the local pronunciation of the name was probably still Ausclum. As to the 24 miles, some astounding errors are on record as made by Roman travellers. See Hunter, L. W., ‘Cicero's Journey to his Province in 51 B.C.’ in J.R.S. vol. iii. Pt. i. 1913, p. 73–97Google Scholar.

^{page 111 note 1} An inscription of Aeclanum, (C.I.L. ix. 1156Google Scholar) speaks of work “in via ducente Herdonias.’

^{page 112 note 1} The coast route between Butunti and Gnatia is 49, the inland route is 45 miles long.

^{page 113 note 1} De la Berge, Essai sur le règne de Trajan, p. 108, thinks, on the other hand, that the operations began in 109 A.D. and finished three or four years later.

^{page 113 note 2} C.I.L. ix. 5998–6055. The original stones of Trajan constitute the bulk of the extant milestones of this road.

^{page 113 note 3} Cohen,² 647–652.

^{page 114 note 1} Cohen,² 667.

^{page 114 note 2} For the earlier sculptures incorporated in the Arch of Constantine see E. Strong, Roman Sculpture, p. 291. Prof. Petersen (Röm. Mitth. 1890, 73 ff) has shown that all the eight panels on the Arch of Constantine belong to a monument erected in 176 A.D. to commemorate the double triumph of Marcus Aurelius over the Germans and Sarmatians which took place in the closing months of the year. Cf. Jones, H. Stuart in Papers of the British School at Rome, vol. iii. p. 251Google Scholarseq.

^{page 115 note 1} Op. cit. p. 108.

^{page 115 note 2} Grasso, , Studi, vol. ii. p. 26Google Scholar.

^{page 115 note 3} Pratilli, La Via Appia, p. 432. Cf. C.I.L. ix. p. 593. Mommsen, says (C.I.L. ix. p. 27Google Scholar) PRATILLIVS in Via Appia (1745) quae de Apulia profert, ea quatenus vera sunt, proficiscuntur fere ab Johanne Baptista DELLO IACONO Bitontino, ….

^{page 117 note 1} The chief literature upon the arch is as follows:—

E. Petersen, Römische Mitteilungen, 1892, p. 241 et seq.

von Domaszewski, A., Jahreshefte des Österreich. archäologischen Instituts ii. 1899, 173Google Scholar, Die politische Bedeutung des Traiansbogen im Benevent.

Rossini, Archi trionfali, tav. 38–43.

Meomartini, Monumenti di Benevento, 1889, pp. 9–218 and Tav. 1–29.

^{page 117 note 2} In the construction, however, of the long bridges over the Cervaro and the Carapelle (see pages 142, 146 infra) the engineers would encounter very considerable difficulties in finding a firm foundation and would probably have to dig quite deep through the subsoil. It was not possible from an examination of the remains of these bridges appearing above ground to form an idea of the nature of the foundations, but probably excavations would yield interesting results.

^{page 118 note 1} We found no traces of these embankments on the probable course of the road below Buonalbergo; whether the later brickwork at Ponte S. Spirito (see p. 135 infra) could be connected with these repairs is quite uncertain. And, indeed, as far as we could judge on the spot, it would be hard to say where these embankments would be employed. The road did not there appear to coast along the slopes of the earth hills (where cuttings would hardly be permanent and embankments or supporting walls a desideratum), but climbed straight up to each summit level and then continued along the top until the next valley. Moreover, traces of Roman reconstruction are extremely rare all along the course of the road.

^{page 120 note 1} C.I.L. iii. 1456.

^{page 120 note 2} The names of the curators of the Roman roads under the Empire have all been collected by Cantarelli, Bull. Arch. Comm. 1891, p. 90 et seq.

^{page 120 note 3} Cf. Hirschfeld, Verwaltungsbeamten, p. 208, note 1. After the time of Claudius equestrian curators are met with only rarely and they are exclusively concerned with viae minores; their position corresponded to that of the procuratores but it was exceptional for them to be designated as procuratores. Cf. C.I.L. vi. 1610. The title, however, of procurator is probably genuine in the inscription mentioning Aelianus, Q. AxiusC.I.L. iii. 1456Google Scholar. CVRATOR AD POPVL(VM) VI[ar(um)] TRAIANAE ET AVRELIAE (et) AECLANENSIS. It is certain, Hirschfeld thinks, that here we are not to think of an equestrian curator of the Via Traiana, but that the words ‘ad populum’ are to be referred to the roads near the large military highways.

^{page 120 note 4} The Italian Staff Maps (published by the Instituto Geografico Militare) on the scale of I: 50,000 were used as detail maps and the Touring Club Italiano Maps as general maps. All names mentioned in the topographical description of the Via Traiana are referred in footnotes to the particular sheet of the Staff Map on which they are to be found. Of the five maps accompanying the article one (p. 119) is a general map (scale 1: 1,500,000) embracing the whole of the Via Traiana and the Via Appia (from Beneventum onwards). The remaining four are detail maps showing all the topographical names mentioned in the text; they divide the road into four sections, Beneventum to the Mutatio Aquilonis, the Mutatio Aquilonis to Herdoniae, Herdoniae to Barium and Barium to Brundisium.

^{page 120 note 5} C.I.L. ix. pp. 26, 592–598.

^{page 123 note 1} Although this road, described by Strabo (loc. cit.), is designated by modern writers as the Via Egnazia, there is no classical authority whatever to justify the use of this word.

^{page 123 note 2} Staff Map, 1: 50,000; Benevento, 173, 2.

^{page 123 note 3} See map attached to Meomartini, Del cammino della via Appia, 1907.

^{page 123 note 4} Cf. Staff Map, 1: 50,000; Benevento, 173, 2.

^{page 123 note 5} Cf. Staff Map, 1: 50,000; Benevento, 173, 2.

^{page 124 note 1} Cf. Staff Map, 1: 50,000; Benevento, 173, 2.

^{page 124 note 2} Cf. Staff Map, 1: 50,000; Benevento, 173, 2.

^{pahe 124 note 3} Meomartini, Monumenti di Benevento, p. 257, wrongly calls the Ponte Valentino mediaeval and sees no Roman remains in the bridge.

^{page 124 note 4} Meomartini (op. cit. and loc. cit. supra) quotes ‘alia via (clearly the Via Traiana) que vadit ad S. Valentinum’ nel diploma del principe Arechi di riconferma delle concessioni precedenti al Monastero di Santa Sofia (di Benevento).

^{page 125 note 1} Cf. Staff Map, 1: 50,000; S. Giorgio la Molara, 175, 1.

^{page 126 note 1} Cf. Staff Map, 1: 50,000. S. Giorgio la Molara, 173, 1, 1909.

^{page 126 note 2} Pratilli, La via Appia, Naples, 1745, p. 506, says that between Paduli, Buonalbergo, Castelfranco and Crevacuore traces are seen of an ancient paved road leading to Troia. This is the Via Traiana. But the author erroneously regards it as another road and makes the Via Traiana diverge from the Via Appia at Aeclanum.

Mommsen, (C.I.L. ix. p. 122Google Scholar) and Nissen (op. cit. p. 816) both place Forum Novum at Monte Male. Kiepert's, map (C.I.L. ix. Tab. iiGoogle Scholar.) wrongly represents the Via Traiana (printed Via Appia) as passing south of Paduli, whereas it really goes north.

^{page 127 note 1} Op cit. p. 258.

^{page 127 note 2} Notizie degli Scavi, 1899, p. 149. The dedication is to one M. Rutilius Macedo from his wife Licinia Marcella. He had been curator at Puteoli and ii. vir at Beneventum. The inscription is probably later than the construction of the Via Traiana. Bones, fragments of bricks and pottery were found in the vicinity.

^{page 127 note 3} At the Masseria to the north-west is a finely sculptured cornice block with interesting decoration (Fig. 8) and a tombstone of the common ‘baulo’ type (in shape like a portmanteau, with a semicircular top) bearing the sepulchral inscription C.I.L. ix. 1440. At a house further north-west is the notable inscription C.I.L. ix. 6005 (Fig. 1) which alone records the construction of the bridges along the Via Traiana. It is built into the wall of the farm-house.

^{page 130 note 1} The Ponte S. Marco dir(uto) is marked on the Staff Map (1: 50,000; Montecalvo Irpino, 174, 4), but the position given to it is quite a kilometre too far downstream.

^{page 132 note 1} Cf. Meomartini, I monumenti di Benevento, p. 303.

^{page 132 note 2} Here was found C.I.L. ix. 6010 (misprinted 6011) referring to extensive repairs of the Via Traiana carried out by Septimius Severus and Caracalla in 210 A.D. On a tile of the Ponte delle Chianche there was seen the inscription PONTV·TRA (C.I.L. ix. 6011, 6078, 2).

One of the tiles still in situ bears the stamp . The tiles are 0.045 to 0.05

metre thick. A tegula mammata (with the mammae knocked off) is seen on one place. The discovery of the inscription and of the tile are due to Dressel. Mommsen conjectures that the meaning is pont(es) v(iae) Tra(ianae). Cf. C.I.L. xv. p. 6, under vii.

^{page 134 note 1} Cf. Staff Map, 1: 50,000; Montecalvo Irpino, 174, 4.

^{page 134 note 2} I Monumenti di Benevento, p. 303.

^{page 134 note 3} The courses are well over two Roman feet high (0·62, 0·68, 0·71 m. respectively), a fact which also points to the period of Trajan. No traces indeed of earlier construction have been found by us along the course of the road.

^{page 136 note 1} Mommsen, , C.I.L. ix. p. 122Google Scholar; Hülsen, in Pauly-Wissowa ‘R. E.’ i. 605Google Scholar, and Supplement and Nissen., op. cit. ii. p. 816Google Scholar give all available information concerning it.

^{page 136 note 2} C.I.L. ix. 1418, 1419.

^{page 136 note 3} Mommsen, says (C.I.L. ix. p. 592Google Scholar), ‘licet omni tempore vici exigui condicionem non egressum rei viariae Italiae inferioris tanquam cardo fuit, quadrivium scilicet viarum primariarum, alterius hinc Roman ducentis per Appiam, inde Brundisium per Traianam, alterius autem hinc pergentis Luceriam et ad mare superum, inde Venusiam et ad fretum Siculum.’

^{page 136 note 4} Grasso, , Siudi, vol. i. p. 39–57Google Scholar. It rather joined Aeclanum and Herdoniae.

^{page 136 note 5} Pratilli, op. cit. p. 515. Cf. Staff Map, 1: 50,000; Bovino, 174, 1. N.W. corner.

^{page 136 note 6} Analyse géografique de l'Italie. Paris 1744, p. 218Google Scholar.

^{page 136 note 7} Storia della regia citta di Ariano. Rome, 1794, p. 5Google Scholar.

^{page 136 note 8} Grasso, , Studi, vol. i. pp. 79–147Google Scholar.

^{page 138 note 1} Meomartini (I monumenti e le opere d'arte di Benevento, 1889, p. 261) takes the road too much to the west after leaving Aequum Tuticum and so loses what is certainly the exact course. It may have passed, according to him, through Vescellium to the west of Castelfranco in Miscano, between this and Rosetto, where there is now the Bosco Vetrosello. He thinks that the Itin. Anton. confused Vescellium with the Mutatio Aquilonis. Nissen, (Ital. Land. ii. p. 843Google Scholar) takes the road from Aequum Tuticum to Mutatio Aquilonis (near Cappella S. Vito) without fixing any intermediate points.

^{page 138 note 2} The Masseria S. Vito is 971 metres above the sea, the highest point of the Via Traiana. Just before reaching it a narrow defile is passed from which magnificent and comprehensive views are obtained both ways. To the N.E. the mountains slope down to the Apulian plain, with Garganus looming on the sky line, to the S.W. and S.E. a fine panorama of mountain and valley is seen from M. Vulture to M. Taburno.

^{page 138 note 3} de Petra, G., Rendiconti del R. Accademia di Napoli, xii. (1898) p. IIIGoogle Scholarseq.; cf. Grasso, vol. iii. p. 9, note 3. Hülsen in Pauly-Wissowa, ‘R. E.’ Supplementum to article ‘Aquilonis mutatio,’ p. 114. An inscription of 1504 over the fountain speaks of it as the Fons Aquilonensis, so that the inscription was then already known.

^{page 139 note 1} C.I.L. ix. p. 87, mutatio Aquilonis …. diversa ab Aquilonia hodie Lacedogna sub Vibino fuerit necesse est.

^{page 139 note 2} Cluver does not mention the mutatio Aquilonis at all. Pratilli (op. cit. p. 503) jumps from Samnium to the Hirpini and from the Hirpini to Apulia. He entirely confuses the Aquilonia in Samnium, the Aquilonia in Hirpinis, and the mutatio Aquilonis. Corcia, (Storia delle due Sicilie, Napoli, vol. ii. p. 531Google Scholar) would place the mutatio Aquilonis at the Buccolo di Troia and regard Aquilonis as coming from aquilo, the north wind. He writes, ove a forza di scalpello si vede aperto il monte per tracciarvi la strada e cosi gagliardi vi spirano i venti nella stagione invernale, che bene spesso atterrati vi rimanevano iviandanti coi carichi e le vetture, circostanza la quale ci spiega la ragione onde fu cosi detta dagli antichi.’ Wesseling, too (Ancient Itineraries, p. 610), says, ‘diceres ab Aquilone vento mutationi nomen haerere, nisi Apuliae proprii nominis ventus Atabulus infestior fuisset.’ No ancient author speaks of the Aquilo, but there is a mediaeval document, published first by Ughelli, (It. sacra, vol. iGoogle Scholar. epis. Troia) and then by Vitale, (Storia di Ariano, Rome, 1794Google Scholar, doc. in Appendix) in which mention is made of this Aquilo in connexion with the boundaries of the territory of Troia. ‘… et rediens ad sinistram usque ad fluvium Aquilonis descendit usque ad transitum Colonnelli.’

^{page 140 note 1} Horace, Epodes, iii. 16Google Scholar, siticulosae Apuliae.

^{page 140 note 2} Cf. Staff Map, 1: 50,000; Ariano di Puglia, 174, 3.

^{page 140 note 3} Nissen, , Ital. Land. ii. p. 843Google Scholar. The road crosses the Buccolo di Troia (905 metres) gradually by cuts in the rocks for three miles. On the Apulian side it drops 300 metres sharply in three turns.

(Storia delle due Sicilie, ii. p. 531.)

^{page 141 note 1} For its history see Hülsen, , Pauly-Wissowa, , ‘R.E.’ vol. i. 443Google Scholar. Mommsen, , C.I.L. ix. p. 85Google Scholar. Excavations at Tro a are described in Not. d. Scavi, 1903, p. 349. Cluver (Ital. Antiq. p. 1202) wrongly identified it with Accadia, a mountain village just to the north of S. Agata di Puglia, west of Candela. Holste (ad Cluver., p. 271) corrects him by reference to Cuniferius, a monk of M. Casino, in his life of S. Secundinus, as follows: ‘Haec vero civitas, si nominis significationem advertimus, Ecana enim dicta est, antiquissima fuit, cum monumentorum marmoratio, scenarum columnatio, eminentia culminum id designent. Huic vero in reconciliatione Troia nomen imponitur, ut egregii titulus nominis auctoramentum faciat novitati. Putamus ista fidem posse quaerentibus facere, quos sub Troiae nomine appellatio noverit Urbis Ecanae.’

Pratilli (op. cit. p. 515 et seq.), who describes the course of the Via Traiana in these parts with surprising accuracy—though he does not call it by the name of the Via Traiana—rightly identifies Aecae with Troia. According to him it was only after the eleventh century that Troia was built upon the present site.

^{page 141 note 2} C.I.L. ix. p. 26. The Tab. Peut. gives Aecas-Pretorium Lauerianum Mucerie Apulie (appingitur domus)-viiii-Arpos-xxi-Siponto.

^{page 142 note 1} Staff Map, 1: 50,000; Troia, 163, 2.

^{page 143 note 1} Staff Map, 1: 50,000; Foggia, 164, 3. Pratilli, op. cit. p. 517, is singularly correct in his description of the road here.

^{page 145 note 1} Cf. Staff Map, 1: 50,000; Ascoli Satriano, 175, 4.

^{page 147 note 1} C.I.L. ix. p. 64. He regards Furfane as being in the neighbourhood of the modern Cerignola., C.I.L. ix. p. 26Google Scholar. He thinks that the turres duae represent Canusium (oppidum, ad quod pinguntur turres nomine omisso, Canusium esse recte intellexerunt viri docti (Mannert, ix. 2, 74); sed num recte ita ordinaverim ut supra factum est fecitque similiter Desiardinius, p. 212, parum constat).

^{page 147 note 2} Studi, vol. ii, pp. 37–41; 49–56.

^{page 148 note 1} Its exact orthography is very doubtful. Herdōnǐae is the generally recognised form, as can be gathered from the various citations of the word which we possess. The singular form may have been used at the end of the republic (as we find in Livy and Strabo), but the plural form was certainly preferred later. Herdōnia, however, is found in Sil. It. viii. 567, as the metre requires.

^{page 149 note 1} Ital. Ant., p. 1202.

^{page 149 note 2} Ad. Cluv. pp. 271, 272.

^{page 149 note 3} Op. cit. p. 429.

^{page 150 note 1} Staff Map, 1: 50,000; Ascoli Satriano, 175, 4.

^{page 150 note 2} On the uncertain interpretation of the Tab. Peut. see pages 147, 148 supra.

^{page 151 note 1} Cf. Staff Map, 1: 50,000; Ascoli Satriano, 175, 4.

^{page 152 note 1} Giovio, and Alberti, , Memorie storiche di Cerignola, Napoli, 1785Google Scholar, reprint, Faenza, 1883.

^{page 152 note 2} Livy, xxii. 18: Ex Paelignis Poenus flexit iter retroque Apuliam repetens Gereonium pervenit, urbem metu quia conlapsa ruinis pars moenium erat, ab suis desertam; dictator in Larinate agro castra communiit. Ibid. 24, dein castra ipsa propius hostem movit duo ferme a Gereonio milia in tumulum hosti conspectum (Hannibal).

Polybius, iii. 100, 3, ἀϕικόμενος δὲ ςρὸς τὸ Гερούνιον, ὅ τῆς Δουκαρίας ἀπέχει διακόσια στάδια.

^{page 152 note 3} Cf. pages 147, 148 supra.

^{page 152 note 4} Antequam autem ad Canusium pervenirent, apud mansionem hodie Cerignola, cui Itin. Hieros. tribuit nomen ad xi miliarium a Canusio Horatius et comites rursus in Viam Appiam (sic !) ingressi sunt.’

Conte, Memorie filologiche sull' antichita della Chiesa di Cerignola, Napoli, 1887, p. 12, alters the length of the ancient mile and proposes Cerignola as the oppidulum of Horace.

^{page 153 note 1} If this be correct, we may consider the original form of Cerignola as being Certniola; hence it is dactylic and cannot be a candidate for Horace's oppidulum. But Horace, supposing that Ausculum was the oppidulum, could have come by a very easy route to join the line of the later Via Traiana at Cerignola itself or at the Masseria Monte Gentile, halfway to Canosa. Cf. Staff Map, 1: 50,000; Cerignola, 175, 1. Pratilli (op. cit. p. 511–514) describes such a road from Equotutico (which he places between Trivicum and Canosa, different from the Equumtuticum near Ariano) to Canosa via the Calaggio valley.

Only two other inscriptions (besides the milliarium) are attributed to Cerignola. Of these one (C.I.L. ix. 684) is interesting as being a dedication to the ‘Bonn Dea’ by one Sexlilia.

^{page 154 note 1} Pratilli, op. cit. p. 518, describes the course of the road from Herdoniae to Canusium. At Ad Sextum, which he wrongly reads instead of the mut. undecimum of the Itin. Hieros., he says there are remains of ancient buildings to the left of the road. Cerignola he places two miles to the north of the road upon a small hill. He mentions the milestone (lxxxi.) to be seen there.

^{page 154 note 2} Pratilli, op. cit. p. 519. Nissen, , op. cit. vol. ii. p. 854Google Scholar.

^{page 155 note 1} General references to Canusium, : C.I.L. ix. p. 34, 35Google Scholar; P.W. ‘R.E.’ vol. iii. 1501; Nissen, It. Land., ii. 853–856; Rotnanelli, , Topogr. storica, ii. 262Google Scholar; N. Jacobone, Canosa antica.

^{page 155 note 2} Garrucci, Monete d'Italia, 94.

^{page 156 note 1} Horace, , Sat. i. 5. 92Google Scholar: nam Canusi (panis) lapidosus, aquae non ditior urna | qui locus a forti Diomede est conditus olim; Sat. i. 10, 30: Canusini more bilinguis.

^{page 156 note 2} Strabo, vi. 3, 9, p. 283, mentions its degeneracy: .

^{page 156 note 3} Its harbour, to which ships came up from the Adriatic, was perhaps identical with Cannae. Strabo, vi. 3, 9, p. 283: .

^{page 156 note 4} Before the construction of the Via Traiana it was on one of the regular routes between Beneventum and Brundisium. Strabo, vi. 3, 7, p. 283, considered supra p. 108. Frequent mention is made of it in accounts of journeys.

^{page 156 note 5} Philostratus, , vit. soph., 2, 1, 5, p. 551Google Scholar: ᾣκισε δὲ (Herodes Atticus) … τὸ ἐν τῇ Ίταλίᾳ, Κανύσιο ṍἡμερώσας ὕδατι μάλα τούτου δεόμενον. Not. d. Scavi, 1894, 408, records an inscription on a water pipe, R(ei) P(ublicae) C(anusinorum) cur(ante) P. GRAEc(idio) FIRMO.

^{page 157 note 1} For the characteristics of mediaeval fortifications constructed from Roman materials, see Gsell, , Les monuments antiques de l'Algérie, vol. ii. p. 344Google Scholarseq.

^{page 158 note 1} Staff Map, 1: 50,000; Barletta, 176, 1.

^{page 158 note 2} The milestones C.I.L. ix. 6040, 6042, 6043, 6044, were discovered at and in the vicinity of Corato.

^{page 158 note 3} See Angelo Mosso, Le origini della civiltà Mediterranea, 1910, p. 167 seq.

^{page 158 note 4} See Gervasio, Michele, I dolmen e la civiltà del bronzo nelle Puglie. Bari, 1913, pp. 1–69Google Scholar.

^{page 159 note 1} Op. cit. p. 338 sqq., Figs. 107, 108.

^{page 160 note 1} ‘Inde Rubos fessi pervenimus, utpote longumCarpentes iter et factum corruptius imbri.’ (Sat. 1, 5, 94).

In Pliny (iii. 11, 105) we read of the Rubustini: and perhaps in § 102 where we read ‘Poediculorum oppida Rudiae (ruriae Lugd.) Egnatia Barium, we ought to read Rubi Egnatia, unless we are to suppose that Pliny erroneously assigned Ennius’ birthplace to the Poediculi. The auctor libri coloniarum interpolati (p. 262) gives Rubustinus ager. We possess a considerable number of coins from Rubi, silver and bronze, bearing the inscriptions ،Рυψ and ،Ρβαστείνων. Sambon, Monnaies de la presqu'île Italique, p. 214.

^{page 160 note 2} C.I.L. ix. p. 33. Inter Rubos et Canusium quae videtur in itinerari s interponi mansio Rudae sive Budae, incidens fere in oppidum quod nunc est Andria, titulis paene caret. Non recte ad earn referri locum Plinii, iii, 11, 102, modo monui, p. 6.

^{page 160 note 3} Cf. Staff Map, 1: 50,000; Barletta, 176, 1.

^{page 161 note 1} iv. 55, 29 ‘Haec tam rustica malo quam Butontos.’

^{page 161 note 2} To the stretch between Ruvo and Bitonto belong the milestones C.I.L. ix. 6040–6051. For excavations at Bitonto see Not. d. Scavi, 1882: 242, 1886: 239, 1887: 204, 1897: 433.

^{page 161 note 3} Mommsen notes: statio adiungitur, errore ut videtur, extremae viae VI. sub Lubatia: numerus xx. utrum pertineat ad viam Norva—Ezetium numero carentem an ad aliam viam parum constat.

^{page 162 note 1} Strabo, vi. 3, 7, p. 282.

^{page 162 note 2} The form Καιλία is better attested than that of Καιλία. Strabo (loc. cit.) gives ‘Καιλία’; Ptolemy, iii. 1, 73, says ῞Απουλῶν Πευκετίων μεσόγειοι … Κελία. In the Tab. Peut. and the Geog. Raven. we read Celia. The liber coloniarum interpolatus (p. 262) gives ‘Caelinus ager’ and in an inscription from Rome we read (C.I.L. vi. 2382, 6, c. 33: C· VALERIVS C·F· CLA · MABCVLIN· CAEL)· From this inscription it is clear that Caelia was registered in the Tribus Claudia. From Caelia come bronze and silver coins inscribed Καιλίνων. The inscriptions from Caelia are mostly funerary and none of them cast any light upon the local constitution. Kaibel, , I.G. xiv. 686Google Scholar, records a Greek fragment.

^{page 162 note 3} Mommsen, , C.I.L. ix. p. 30Google Scholar, says: ‘Mediterranea quoque via Gnathia Butontum ducens stationes habet ad Veneris, Norve, Ezetium et ipsas auctoribus quos habemus ignotas; temere enim Netium oppidum, quod Strabo vi. 3, 7, p. 282, inter Caeliam et Canusium interponit, ad Ezetium illud rettulerunt alio omnino loco situm. Lapides scriptos hae quoque partes adhuc non magis dederunt.

^{page 163 note 1} Cf. Staff Map, 1: 50,000; Bari delle Puglie, 177, 2.

^{page 163 note 2} Ibid.

^{page 163 note 3} Ibid.

^{page 163 note 4} All available information concerning Barium is given by Mommsen, , C.I.L. ix. p. 30, 31Google Scholar; Hülsen, in Pauly-Wissovva, , ‘R.E.’ iii. 19Google Scholar; and Nissen, , Ital. Land. ii. 358Google Scholar. According to Horace, Sat. 1, 5, 92, it was famous for its fish, while Pliny, , N.H. xiv. 69Google Scholar, is probably to be referred to its wines. Tacitus' mode of expression (Ann. xvi. 9 Silanus tamquam Naxum deveheretur Ostiam amotus post municipio Apuliae, cui nomen Barium, clauditur) would lead one to think it unimportant.

^{page 163 note 5} Livy, xi. 18, inter duumviros ita divisa tuenda denis navibus maritima ora, ut promunturium iis Minervae velut cardo in medio esset; alter inde dextram partem usque ad Massiliam, laevam alter usque ad Barium tueretur.

^{page 164 note 1} C.I.L. ix. pp. 25, 26.

^{page 164 note 2} It is curious that the Itin. Hieros. should be more exact here than the Itin. Anton. The total distance between Beneventum and Brundisium, as measured along the probable course of the road, is almost 205 miles. The Itin. Anton. gives a total of 206, the Itin. Hieros. of 199 miles.

^{page 166 note 1} That the Diriam of the Geogr. Raven, is to be preferred to the Dertum of the Tab. Peut. may be suggested from Pliny, , N.H. iii. 11, 105Google Scholar, who speaks of Dirini.

^{page 166 note 2} Nissen, op. cit. p. 860, places Turres Caesaris at Polignano, Diria at Monopoli. Pratilli (op. cit. pp. 534–543) describes the Via Traiana from Barium to Gnathia. At Torre Ripagnola (Staff Map, 1: 50,000; Mola di Bari, 178, 3, where ‘Ruderi’ are marked) he places Turres Caesaris. He is certainly wrong in identifying Turres Iulianas with ‘Turres Aurelianas,’ ‘Turribus,’ and ‘Turris Caesaris.’ The three latter are clearly the same, but ‘Turres Iulianas’ is nearer Barium. He says that pavement is visible for three miles near S. Vito (south of Torre Ripagnola) at a distance of fifty paces from the sea. Here he would place the station of Arnesto, but it is almost certainly to be identified with Turres Caesaris, as Pratilli himself later suggests. The station of Vertum or Diria he locates at the Torre Orto just north of Monopoli. Between Torre Orto and Monopoli he says that pavement is to be seen. South of Monopoli he states that the road passed quite near the ‘fortino di S. Stefano’ and the ‘Torre di Centola’ (both of these are given on the Staff Map, 1: 50,000; Monopoli, 190, 1, as S. Stefano and Torre Cindola). The Torre di Palasciano farther south may perhaps be represented by the modern Lo Sciale on the same map. Much pavement, he says, is to be seen from there to Gnathia.

^{page 166 note 3} Sambon, op. cit. p. 213; Mommsen, , C.I.L. ix. p. 30Google Scholar, says: 'eos enim tradunt in Polignanensi territorio potissimum eruderari.

^{page 166 note 4} The name of the city is given in various forms. A bronze caduceus from the neighbouring Fasano, (I.G. xiv. 685Google Scholar) reads ГΝΑΘΙΝΟΝ; two tiles, Ibid. 2401¹ and 2402² give respectively Г]ΝΑΘΙΣ and Г]ΝΑΘΙΣ. The form Gnatia is used by Horace, , Sat. i, 5, 97Google Scholar (dein Gnatia lymphis | iratis exstructa) Mela ii, 66, and the Geogr. Rav. iv. 31. The locative form (Gnatiae) is read in the Itin. Anton. 313. Gnatia comes from the Tab. Peut. and also from the Geog. Rav. Egnatia and the corresponding Greek iorm ГΝΑΘΙΣΈγνατία are given respectively by Pliny, , N.H. ii. 240Google Scholar, iii. 102; Ptolemy, iii. 1, 13, and Strabo, vi. 3, 8, p. 282. The locative form of this (Egnatiae) we discover in the Itin. Anton. 117; Ignatiae comes from the Geog. Rav. v. i. Ignatinus [ager] is used in the Lib. coloniarum, p. 262; Leonatiae by the Itin. Hieros. 609 and Augnatium by Guido, 27, 71. See Pratilli, op. cit. pp. 544, 545; Pepe, L., Notizie storiche ed archeologiche dell' antica Gnathia, Ostuni, 1882Google Scholar.

^{page 167 note 1} Strabo, vi. 3, 8, p. 283. . Pliny, iii, 102. Ptolemy, iii, 1, 13.

^{page 167 note 2} M. Mayer, op. cit. and loc. cit.

^{page 167 note 3} Nissen, , op. cit. vol. ii. p. 860Google Scholar, note 10, says: Gnatia lymphis iratis exstructa kann angesichts der reichen und tremichen Quellen des Ortes nicht auf Wasserarmut gedeutet werden.

^{page 168 note 1} Cf. Staff Map, 1: 50,000; Ostuni, 191, 3.

^{page 168 note 2} Cf. Carta d' Italia del Touring Club Italiano, Lecce.

^{page 168 note 3} In Italie pittoresque (Tableau historique et descriptif de l'Italie, du Piémont, de la Sardaigne, de Malte, de la Sicile et de la Corse, Paris, 1835), p. 49, we read: On distingue encore ça et la quelques dalles intactes de la voie antique et je découvris moi-même au milieu de la plaine un vaste fragment de construction réticulaire. Etait-ce un temple? une villa? un tombeau? C'est ce que je ne saurais dire. Tout ce que ce je puis affirmer, c'est que c'est un débris romain. We are unable to state the nature or position of this construction in opus reticulatum.

^{page 169 note 1} Camassa erroneously says that the viaduct was used for the Via Appia. He means the Via Traiana.

^{page 169 note 2} Compare the structure of the buttresses in the bridges across the Cervaro and Carapelle near Ordona.

^{page 170 note 1} Cf. Camassa, op. cit. pp. 27, 28.

^{page 170 note 2} Camassa, op. cit. p. 21. The survivor of two twin columns (the other fell in 1528 and was transported to Lecce) is to be seen near the harbour at Brindisi. It is composite in style; the pillar is made of cipollino, the pedestal and capital of white marble. The capital is adorned with twelve figures (head and shoulders alone), four of which represent Jupiter, Neptune, Minerva and Mars; the remaining eight are Tritons. The whole is nineteen metres high and bears on its base the following incomplete inscription:—

ILLVSTRIS PIVS ACTIB · ATQ: REFVLGE

PTOSPATHA LVPVS VRBEM HANC STRUXIT AD

QVAM IMPERATORES MAGNIFICIQ: BENIGNI

The two columns are probably to be referred to the time of Sulla, who, in order to recompense Brundisium for its co-operation in the Social War, granted it ἀτέλεια, the nature of which is in dispute. Cf. Appian, , B.C. i. 79Google Scholar. . The question has been examined in detail by Henderson, B. W. (Classical Review, xi. p. 251Google Scholarseq.) Holding that it is impossible to interpret ἀτέλεια in any other sense than that of exemption from portoria, he suggests that these dues which were universally abolished by the Lex Caecilia of 60 B.C. were reinstated probably by Nero. Trajan, however, in order to develop the commercial prosperity of the east coast of Italy, revived the gift to Brundisium of exemption from portoria which Sulla had bestowed upon it. As Strachan Davidson suggests (Appian, B.C. i. p. 82Google Scholar, note 6) it is hard to see why such a gap in the Italian customs cordon should have been allowed, but no more satisfactory explanation can be offered.