¹ Fabretti, R., De aquis et aquaeductibus veteris Romae dissertationes tres (Rome, 1680; reprint Portland, 1972), 174–5Google Scholar. Elsewhere (49–51) Fabretti quashed a theory that Alexander Severus introduced an ‘Aqua Ianiculensis’ (i.e. the Aqua Traiana) to supply his baths and nymphaeum east of the river.

² Preller, L., Die Regionen der Stadt Rom (Jena, 1846), 201, 246–7Google Scholar, criticized by Jordan, H., Topographie der Stadt Rom im Alterthum, 2 vols in 4 (Berlin, 1878), I.1 476, n. 99; II 294–6Google Scholar; Bloch, H., ‘Aqua Traiana’, American Journal of Archaeology 48 (1944), 337–41CrossRefGoogle Scholar. On Fabretti's evidence, see the section on the Aqua Traiana below.

³ Parker, J.H., The Aqueducts of Ancient Rome, Traced from their Sources to their Mouths (The Archaeology of Rome VIII) (Oxford/London, 1876)Google Scholar; Lanciani, R., ‘Topografia di Roma antica. I commentarii di Frontino intorno le acque e gli acquedotti’, Atti della Reale Accademia dei Lincei. Memoria della Classe di Scienze Morali, Storiche e Filologiche ser. 3, vol. IV (1880), 215–616Google Scholar; Van Deman, E.B., The Building of the Roman Aqueducts (Washington, D.C., 1934)Google Scholar; Ashby, T., The Aqueducts of Ancient Rome (ed. Richmond, I.A.) (Oxford, 1935)Google Scholar; Lloyd, R. B., ‘The Aqua Virgo, Euripus, and Pons Agrippae’, American Journal of Archaeology 83 (1979), 193–204CrossRefGoogle Scholar.

⁴ Evans, H. B., ‘Agrippa's water plan’, American Journal of Archaeology 86 (1982), 193–204CrossRefGoogle Scholar; Evans, H. B., ‘Nero's Arcus Caelimontani’, American Journal of Archaeology 87 (1983), 393–9CrossRefGoogle Scholar; Evans, H. B., Water Distribution in Ancient Rome: The Evidence of Frontinus (Ann Arbor, 1994)CrossRefGoogle Scholar. Richardson, L. JrA New Topographical Dictionary of Ancient Rome (Baltimore/London, 1992)Google Scholar, showed a sporadic awareness not only of the crossings, but of some of their implications; for example, he suggested that the Thermae Etrusci, which were supplied by both the Aqua Virgo and the Aqua Marcia, may have been in the Transtiberim. Oddly, in an otiose entry for the Balneum Claudii Etrusci, which is the same building, he did not mention this possibility.

⁵ Bruun, C., The Water Supply of Ancient Rome: a Study in Roman Imperial Administration (Commentationes Humanarum Litterarum 93) (Helsinki, 1991)Google Scholar; Il trionfo dell'acqua: acque e acquedotti a Roma (Rome, 1986)Google Scholar; Panimolle, G., Gli acquedotti di Roma antica (Rome, 1968)Google Scholar.

⁶ See D. Cattalini, ‘Aqua Marcia’; S. La Pera, ‘Aqua Virgo’; Z. Mari, ‘Anio Novus’, ‘Anio Vetus’ and ‘Aqua Claudia’; A. Mucci, ‘Aqua Appia’; and P. Virgili, ‘Aqua Traiana’, all in Steinby, E. M. (ed.), Lexicon Topographicum Urbis Romae I (A-C) (Rome, 1993)Google Scholar.

⁷ Delbrück, R., Hellenistische Bauten in Latium I (Strasbourg, 1907–1921), 12–22Google Scholar, and Frank, T., Roman Buildings of the Republic: an Attempt to Date them from their Materials (Rome, 1924), 139–41Google Scholar, both offered detailed descriptions of the masonry of the Ponte Rotto but reported no evidence of aqueduct conduits in the Augustan remains of this bridge. The bridge is known to have carried a conduit of the Acqua Felice until the flood of 1599 (Ashby, , The Aqueducts of Ancient Rome (above, n. 3), 306–7 and n. 1)Google Scholar, but its incarnations as a garden in the eighteenth century (Picone, L., Ponti di Roma e vedute sul Tevere di Guiseppe Vasi (Rome, 1986), 70)Google Scholar and the support for an iron bridge in the nineteenth (Delli, S., I ponti di Roma (Rome, 1977), tav. 6Google Scholar; D'Onofrio, C., Il Tevere e Roma (Rome, 1970), 230)Google Scholar leave little hope that any traces of the hydraulics survive.

⁸ Gazzola, L. and Bascià, L., La testata etrusca di Ponte Emilio in Trastevere (Rome, 1992)Google Scholar; Richter, O., Die Befestigung des Janiculum: ein Beitrag zur Topographie der Stadt Rom (Berlin, 1882)Google Scholar.

⁹ Savage, S., ‘The cults of ancient Trastevere’, Memoirs of the American Academy in Rome 17 (1940), 26–56, at p. 29CrossRefGoogle Scholar.

¹⁰ See Champlin, E., ‘The suburbium of Rome’, American Journal of Ancient History 7 (1982), 97–117Google Scholar.

¹¹ Wikander, O., ‘Water mills in ancient Rome’, Opuscula Romana 12 (1979), 13–36, at p. 24, n. 64Google Scholar. Hermansen, G., ‘The population of imperial Rome: the regionaries’, Historia 27 (1) (1978), 129–68, at pp. 160–2Google Scholar, offers a synopsis of the scholarly discussion about these inscriptions. The Capitoline Base may not list all the vici from a given region, only those favoured by Hadrian in some way. See my discussion of the Aqua Traiana below.

¹² Urlichs, L., Die Brücken des Alten Roms (Studien zur Römischen Topographie I) (Munich, 1870), 480–4Google Scholar; Mayerhöfer, A., Die Brücken im Alten Rom (Erlangen, 1883), 47–9Google Scholar. Contra Urlichs, the roofed structure called the Pons Maximus, attested for the year 156 BC (Obseq. 16), is undoubtedly a wooden version of the Aemilius built upon M. Fulvius Nobilior's piers of 179 BC. In 142 the entire superstructure was rebuilt in stone (Livy 40.51.4). For a recent discussion of the stages of the Pons Aemilius, see Coarelli, F., Il Foro Boario dalle origini alla fine della Repubblica (Rome, 1988), 141–7Google Scholar.

¹³ Lanciani, , ‘Topografia di Roma antica’ (above, n. 3), 237–40Google Scholar; Richardson, L. Jr‘The aqueducts of the Transtiberim and the American Academy in Rome’, Classical Society of the American Academy in Rome Newsletter (December 1990), 6–8Google Scholar. The Syrian sanctuary on the Janiculum was supplied by an abundant spring; see Savage, , ‘Cults of ancient Trastevere’ (above, n. 9), 44–9Google Scholar.

¹⁴ Evans, , ‘Agrippa's water plan’ (above, n. 4), 409Google Scholar; Evans, , Water Distribution in Ancient Rome (above, n. 4), 107Google Scholar. In ‘Water distribution in ancient Rome: Quorsum et Cui Bono?’, in Hodge, T. (ed.), Future Currents in Aqueduct Studies (Leeds, 1991), 21–7, at p. 24Google Scholar, repeated in Water Distribution in Ancient Rome (p. 71), Evans ascribed the extension of the Appia to Agrippa rather than to its first restorer, Q. Marcius Rex. Presumably he meant that the Appia was the third aqueduct to be carried on the Pons Aemilius, preceded by two pre-Agrippan lines; yet he was vague about dating the crossing of the Anio Vetus, which he suggested was added ‘much later’ than the construction of the main conduit (Water Distribution in Ancient Rome, 77). He piled yet another aqueduct on to the Aemilius with the arrival of the Aqua Claudia (Water Distribution in Ancient Rome, 121–2).

¹⁵ The Marcia's service of the Aventine was interrupted between Nero's time and Frontinus's own (Frontin. Aq. 76.5, 87.4). Frontinus's claim that the Marcia crossed the river just before his tenure, but concurrently did not serve regiones XI, XII and XIII (81.2), strongly suggests that the Marcia's main (or only) river crossing was nowhere near the Aventine or its river-front.

¹⁶ Dio Cass. 49.42, however, ascribes Agrippa's restoration of the Marcia to 34 BC, the year before his aedileship. The other two systems are not mentioned here. Pliny, , HNGoogle Scholar 36.121 claims that Agrippa ‘lacus DCC fecit, praeterea salientes D’ (or CVI). On the meaning of saliens and lacus, see Bruun, , The Water Supply of Ancient Rome (above, n. 5), 105–6 and n. 43Google Scholar.

¹⁷ Pons can mean any raised part of an aqueduct, but here Frontinus probably meant the river bridges, since their siphon pipes would have needed more regular service than ordinary specus on arches; see Fahlbusch, H., ‘Maintenance problems in ancient aqueducts’, in Hodge, , Future Currents in Aqueduct Studies (above, n. 14), 7–14, at p. 8Google Scholar.

¹⁸ Buzzetti, C., ‘Arcus in Platea Pontis Sanctae Mariae’, Bollettino dell'Unione di Storia ed Arte 77 (1984), 21–5Google Scholar, suggested that the Aqua Marcia crossed at the Pons Aemilius, but this claim has no foundation in evidence.

¹⁹ Evans, , ‘Agrippa's water plan’ (above, n. 4), 409Google Scholar; Lloyd, , ‘The Aqua Virgo (above, n. 3), 195–6Google Scholar; Shipley, F. W., Agrippa's Building Activities in Rome (St Louis, 1933), 33Google Scholar, n. 94.

²⁰ The purpose of the Pons Fabricius was not to initiate a project for a crossing to the Transtiberim but to augment the recently revitalized Aesculapius cult on the island. The 60s BC witnessed a resurgence of the cult's popularity. The main sanctuary on the island was entirely refurbished with tithes from the cult; see Degrassi, D., ‘Interventi edilizi sull'Isola Tiberina nel I sec. A.D.: nota sulle testimonianze letterarie, epigrafiche ed archeologiche’, Athenaeum n.s. 65 (1987), 521–7Google Scholar. A year after the approval of the restored Aesculapium, the Pons Fabricius was built with public funds.

²¹ Piale, S., Degli antichi ponti di Roma al tempo del secolo V (Atti dell'Accademia Romana di Archeologia IV) (Rome, 1834), 13Google Scholar; Richardson, A New Topographical Dictionary of Ancient Rome (above, n. 4), s.v. ‘Pons Cestius’. C. Cestius's brother, Lucius, is also mentioned on the inscription. Some prefer him as the bridge builder; see Besnier, M., L'Île Tiberine dans l'antiquité (Paris, 1902), 107–9Google Scholar.

²² Robinson, O., ‘The water supply of Rome’, Studia et Documenta Historiae et Iuris 46 (1980), 44–86, at pp. 53–5Google Scholar.

²³ Pauly-Wissowa, , Real-Encyclopädie der Klassischen Altertumswissenschaft [hereafter RE], s.v. ‘Cestius’ [7], 2005Google Scholar; Richardson, A New Topographical Dictionary of Ancient Rome (above, n. 4), s.v. ‘Sepulcrum C. Cesti’.

²⁴ Shipley, F.W., ‘Chronology of the building operations in Rome from the death of Caesar to the death of Augustus’, Memoirs of the American Academy in Rome 9 (1937), 7–60CrossRefGoogle Scholar; Strong, D.E., ‘The administration of public building in Rome during the late republic and early empire’, Bulletin of the Institute of Classical Studies 15 (1968), 97–109CrossRefGoogle Scholar.

²⁵ Evans, , ‘Agrippa's water plan’ (above, n. 4), 406Google Scholar, n. 30. If Agrippa introduced the Aqua Iulia in 40 BC, as Dio Cass. 48.32.3 attested, then some of the redistribution of the older systems may have taken place then as well. Frontinus, however, attributed the Iulia to Agrippa's aedileship in 33 (Aq. 9.1).

²⁶ Roddaz, J.-M., Marcus Agrippa (Rome, 1984), 81CrossRefGoogle Scholar.

²⁷ Lanciani, , ‘Topografia di Roma antica’ (above, n. 3), 355Google Scholar.

²⁸ Gatti, G., ‘Di un nuovo cippo terminale delle ripe del Tevere’, Bullettino della Commissione Archeologica Comunale in Roma 15 (1887), 306–13Google Scholar; Degrassi, A. (ed.), Inscriptiones Italiae (Rome, 1931–1986), XIII.1 207, 673Google Scholar.

²⁹ Borsari, L., ‘Regione IX. Di un cippo spettante alla terminazione delle sponde del Tevere’, Notizie degli Scavi di Antichità (1887), 322–7, at p. 326Google Scholar; Borsari, L., ‘Del Pons Agrippae sul Tevere tra le regioni IX e XIIII’, Bullettino della Commissione Archeologica Comunale in Roma 16 (1888), 92–8, at pp. 92–4Google Scholar. It is believed that the remains are still in place in the bed of the Tiber; see Carpano, C. Mocchegiani, ‘Indagini archeologiche nel Tevere’, in Gigli, S. Quilici (ed.), Archeologia Laziale IV (Quarto incontro di studio del Comitato per l'archeologia laziale) (Rome, 1981), 142–55, at p. 144Google Scholar.

³⁰ Lloyd, , ‘The Aqua Virgo (above, n. 3), 197Google Scholar.

³¹ CIL IV 29781Google Scholar. CIL VI 39087Google Scholar, also a much-discussed inscription, seems to delineate avast area in the Campus Martius from some uncertain point ‘ad Tiberim P(edes) M(ille?) […]/ [secundum Tibe]rim ad Euripu[m p(edes)…] / [secundum Euripu]m ad Piscina[m p(edes)…]. This inscription has been understood in various ways, but all recent interpretations have agreed that it refers to some property of Agrippa, perhaps the Horti Agrippae, which extended from some unknown point along the Tiber to the Euripus, and then along the Euripus to a piscina; see Grimal, P., ‘Agrippa et le Champ de Mars’, Revue Archéologique 19 (1942), 24–30Google Scholar; Coarelli, F., ‘Il Campo Marzio occidentale: storia e topografia’, Mélanges de l'École Française de Rome. Antiquité 89 (2) (1977), 807–46, at pp. 815–18CrossRefGoogle Scholar; Lloyd, , ‘The Aqua Virgo’ (above, n. 3), 199–200Google Scholar; Roddaz, , Marcus Agrippa (above, n. 26), 238–41Google Scholar. If one accepts Lloyd's placement of the Euripus, then this area would include the bridgehead of the Pons Agrippae. It is best to interpret the piscina as the Stagnum Agrippae, which was in the vicinity of the northeastern end of the Euripus. Lloyd's contention that the piscina was a settling basin for the Aqua Virgo is unconvincing. We know of no settling basin within the ancient city limits (Richardson, A New Topographical Dictionary of Ancient Rome (above, n. 4), s.vv. ‘piscina’, ‘stagnum’, ‘naumachia’); such an arrangement would in fact be a waste of urban space, as settling tanks were just as effective and far less intrusive outside the city. For recent criticism of Lloyd, see Roddaz, , Marcus Agrippa (above, n. 26), 285–8Google Scholar and the sources cited therein. Bruun, , The Water Supply of Ancient Rome (above, n. 5), 117Google Scholar, n. 5, seems to have misunderstood Lloyd's argument; Lloyd, did not suggest that ‘the Euripus… brought water from the Aqua Virgo to regio XIV’. He repeated the error, on p. 121Google Scholar.

³² On the Pondel, see O'Connor, C., Roman Bridges (Cambridge, 1993), 39, 91–2Google Scholar; Blake, M. E., Roman Construction in Italy from the Prehistoric Period to Augustus (Washington, D.C., 1947), 218Google Scholar. On Stabilio's bridge, see Potter, T. W., Roman Italy (Berkeley/Los Angeles, 1987), 138Google Scholar; Guzzo, P.G., ‘Sacrofano — ponte romano in località Fontana Nuova’, Notizie degli Scavi di Antichità (1970), 330–44Google Scholar.

³³ Le Gall, J., Le Tibre: fleuve de Rome dans l'antiquité (Paris, 1953), 210–11Google Scholar.

³⁴ The former site of the Pons Agrippae may be precisely where Belisarius extended his chain, and where barrier chains were used in later ages. An ancient street identified by Quilici, L., ‘Il Campo Marzio occidentale’, in Città e architettura nella Roma imperiale (Analecta Romana Instituti Danici Supplement 10) (Rome, 1983), 59–85. at pp. 65 and 69Google Scholar, which led to the left-bank bridgehead of the Pons Agrippae (Fig. 10), has the modern name Via or Strada della Catena on the maps of Nolli and of Lanciani, R., Forma urbis Romae (Milan, 1893–1898Google Scholar; reprint Rome, 1990). The street name, which might be explained by the chains around the Palazzo Farnese, predates the palace; see Gnoli, U., Topografia e toponomastica di Roma medioevale e moderna (second edition) (Rome, 1984), 67–8Google Scholar. Leo IV (847–55) extended a similar chain across the Tiber to the south; see Lanciani, R., Storia degli scavi di Roma e notizie intorno le collezioni romane di antichità, 4 vols (Rome, 1902–1912), IV 84–5Google Scholar.

³⁵ See, for example, the map of La Rocca, E., La riva mezzaluna: culti, agoni, monumenti funerari presso il Tevere nel Campo Marzio occidentale (Rome, 1984)Google Scholar. In the light of this reconstruction, it is doubly surprising that La Rocca has refused to site the Pons Agrippae here.

³⁶ Borsari, , ‘Del Pons Agrippae sul Tevere’ (above, n. 29), 94Google Scholar.

³⁷ Lanciani, , Forma urbis RomaeGoogle Scholar (above, n. 34), tav. 14. Nero also built or refurbished the straight avenue leading from the bridge east to his baths in the Campus Martius, often identified as the Via Recta or, more recently, as the Via Tecta; see Richardson, , A New Topographical Dictionary of Ancient Rome (above, n. 4), s.v. ‘Thermae Neronianae’Google Scholar; Palmer, R. E. A., Studies in the Northern Campus Martius in Ancient Rome (Philadelphia, 1990), 58–9Google Scholar.

³⁸ Lloyd, , ‘The Aqua Virgo’ (above, n. 3), 201–2Google Scholar. Could the same emperor's restoration of the Pons Cestius five years later after a flood (Fasti Ostienses, 152 AD (Degrassi, , Inscriptiones Italiae (above, n. 28) XIII.1, 207, 673))Google Scholar have had a similar secondary purpose? If some of the emperor's family brickyards were in the Transtiberim, they may have been fed by an aqueduct carried by the island bridges.

³⁹ Piale, Degli antichi ponti di Roma (above, n. 21); Jordan, , Topographie der Stadt Rom (above, n. 2), I.1, 417–18Google Scholar. The Pons Aurelius is mentioned in the regionary catalogues and Polemius Silvius, but the Pons Antoninus first appeared in the Liber Pontificalis for the year AD 791 and was featured in a number of medieval sources; see Kummer, T., De urbis Romae pontibus antiquis (Schalke, 1889), 8–16Google Scholar. The two names are never found together in a single source.

⁴⁰ Piale, , Degli antichi ponti di Roma (above, n. 21), 19Google Scholar; Jordan, , Topographie der Stadt Rom (above, n. 2), I.1, 417–18 and n. 31Google Scholar. Subsequent scholars have followed their lead. There is no compelling reason why Caracalla merits this distinction; Marcus Aurelius himself certainly qualifies. Elagabalus was also made Caesar under the name M. Aurelius Antoninus, and was an avid builder during his short reign. Even Commodus assumed the purple under the name M. Aurelius Commodus Antoninus, though he later suppressed the Antonine connection. In fact, the Horrea Antoniniana in Ostia have been ascribed to him; see Rickman, G., Roman Granaries and Store Buildings (Cambridge, 1971), 85Google Scholar.

⁴¹ Coarelli, , ‘Il Campo Marzio occidentale’ (above, n. 31), 844–5Google Scholar.

⁴² Quilici, ‘Il Campo Marzio occidentale’ (above, n. 34), passim.

⁴³ See Piale, , Degli antichi ponti di Roma (above, n. 21), 19Google Scholar; Urlichs, , Die Brücken des Alten Roms (above, n. 12), 465–6Google Scholar; Jordan, , Topographie der Stadt Rom (above, n. 2), I.1, 418, n. 31Google Scholar; Kummer, , De urbis Romae pontibus antiquis (above, n. 39), 15–16Google Scholar; Borsari, , ‘Regione IX’ (above, n. 29), 327Google Scholar; Borsari, , ‘Del Pons Agrippae sul Tevere’ (above, n. 29), 95Google Scholar.

⁴⁴ This is confirmed by the reference in the Liber Pontificalis (94) to a flood in AD 791 that entered the city at the Porta Flaminia and exited by breaching the city wall at the Pons Antoninus.

⁴⁵ In AD 102, the Cellae Vinariae Nova[e] et Arruntiana[e] (CIL VI 8826Google Scholar), apparently wine storage and possibly wine manufacturing facilities, and an accompanying schola for the college of vinarii, were dedicated immediately south of the Casa della Farnesina; see Notizie degli Scavi di Antichità (1878), 66Google Scholar; (1879), 15, 40, 68; (1880), 127–9, 140–1, pl. 4. Aurelian laid his wall directly over the main south horreum. At this time or earlier, these buildings were abandoned permanently; see Notizie degli Scavi di Antichità (1880), 127–8Google Scholar.

⁴⁶ Borsari, , ‘Del Pons Agrippae sul Tevere’ (above, n. 29), 96Google Scholar.

⁴⁷ The excavator reported an inner earthwork 3.07 m deep at the wall (Notizie degli Scavi di Antichità (1880), 128)Google Scholar. The Aqua Virgo was already at a rather low level when it approached the river (Lloyd, , ‘The Aqua Virgo’ (above, n. 3), 195–7)Google Scholar. If the segment proceeding from the western bridgehead had remained where it was before, it would either have been hastily buried under several metres of fill or else have penetrated the river wall itself.

⁴⁸ Quilici, , ‘Il Campo Marzio occidentale’ (above, n. 34), 70–3Google Scholar.

⁴⁹ Quilici, , ‘Il Campo Marzio occidentale’ (above, n. 34), 65Google Scholar.

⁵⁰ Clearance zones existed in Roman cities for secondary branches after the initial distribution. Augustus's edict on the aqueduct at Venafrum (CIL X 4842Google Scholar) treated the zones like public roads; see Robinson, , ‘The water supply of Rome’ (above, n. 22), 61Google Scholar.

⁵¹ Could the name Grotte here recall a distribution tank of the aqueduct?

⁵² One literary source attests to the existence of the Pons Antoninus prior to Probus. A life of Hadrias and Hippolytus, martyrs of Valerian's persecutions in the 250s AD, narrates that they were led to the ‘pons Antonini’ and thrown from it, whereupon their bodies washed ashore on the island; see Jordan, H., Novae quaestiones topographicae II (Regimont (Königsberg), 1868), 13Google Scholar; Urlichs, , Die Brücken des Alten Roms (above, n. 12), 463, 473Google Scholar; Kummer, , De urbis Romae pontibus antiquis (above, n. 39), 8Google Scholar. If its chronology is reliable (a questionable assumption), this account must refer to the old Pons Antoninus, the renamed Pons Agrippae.

⁵³ The Pons Fabricius may be another example of a bridge having an alternate name derived from an inscription. In the middle ages it was called Pons Lepidus (or the corrupt ‘Lapideus’) after one of the bridge's Roman restorers, recorded in its inscription. See Gnoli, , Topografia e toponomastica di Roma (above, n. 34), 224Google Scholar.

Fragments of statuary with a head of Valentinian I and vaulting coffers from a bridgehead arch dedicated to Valens and Valentinian (CIL VI 31402Google Scholar) have been found in the area of the Pons Aurelius/Antoninus, which we know was rededicated to Valentinian in the year 366–7; see Lanciani, R., ‘Monumenti rinvenuti nell'alveo del Tevere sotto il ponte Sisto’, Bullettino della Commissione Archeologica Comunale in Roma 6 (1878), 241–8Google Scholar. Except for the head of the emperor, they seem to be spolia from earlier monuments. Dehn, G., ‘Die Bronzefunde bei Ponte Sisto’, Mitteilungen des Deutschen Archäologischen Instituts (Röm. Abt.) 26 (1911), 238–59, at pp. 253–9Google Scholar, identified the coffers and another head found on the site as Severan, and cited this as evidence for the date of the earlier bridge. But the provenance of the head, which is really the only piece of evidence admissible for dating the bridge, is far from certain. Castagnoli, F., ‘Note di topografia romana’, Bullettino della Commissione Archeologica Comunale in Roma 74 (1951–1952), 49–56Google Scholar, and Rocca, La, La riva mezzaluna (above, n. 35), 68Google Scholar, suggested that the bridgehead arch was a restoration of an earlier ‘Arch of Caracalla’, which again assumed that the adjoining bridge was a project of that emperor. There is little evidence to support this hypothesis.

⁵⁴ Richardson, A New Topographical Dictionary of Ancient Rome (above, n. 4), s.vv. ‘Balneae Severi’, ‘Porta Septimiana’ and ‘Thermae Septimianae’.

⁵⁵ Richmond, I. A., The City Wall of Imperial Rome (Oxford, 1930), 223–7Google Scholar, suggested that the wall was built directly over the bath.

⁵⁶ The Marcia, Rome's coldest line (Pliny, , HN 36.121Google Scholar), reached the Transtiberim, but Frontinus had deployed it exclusively for drinking (Aq. 92; however, Caracalla and Diocletian tapped it for their baths; see Evans, , Water Distribution in Ancient Rome (above, n. 4), 92)Google Scholar. Frontinus's improved Anio Novus (Aq. 93), and perhaps the Claudia as well, were cold, although we do not know if the Anio Novus continued to serve the Transtiberim after Frontinus separated it from the Claudia (Aq. 86.1, 91.2–3, 92). However, the winter temperature of the Virgo, whose waters Martial called ‘gelida’ and ‘nivea’ (7.32.1, 11.47.6), has been recorded at 14–15 degrees Celsius, only 2 degrees colder than the source of the Aqua Tepula and 4 degrees warmer than the Iulia; see Lanciani, , ‘Topografia di Roma antica’ (above, n. 3), 342Google Scholar; Pace, P., Gli acquedotti di Roma e il De aquaeductu di Frontino (Rome, 1986), 138–44Google Scholar. At any rate, the large number of balnea and laci recorded in the fourth-century regionary catalogues for the Transtiberim (86 and 180 respectively) leads one to believe that the real problem was not a lack of cold water in the region per se, but a lack of reassignable cold water. There is no word on whether Aurelian's planned baths were ever begun.

⁵⁷ This is confirmed by Blackman, D.R., ‘The volume of water delivered by the four great aqueducts of Rome’, Papers of the British School at Rome 46 (1978), 52–72, at pp. 68–70CrossRefGoogle Scholar.

⁵⁸ Evans, , Water Distribution in Ancient Rome (above, n. 4), 121–2Google Scholar; Mucci, A., ‘Arcus Caelimontani Aquae Claudiae’, in Il trionfo dell'acqua (above, n. 5), 95–8, at p. 98Google Scholar; Platner, S. B. and Ashby, T., A Topographical Dictionary of Ancient Rome (London, 1929), 211Google Scholar; Longhi, G. Marchetti, Bullettino della Commissione Archeologica Comunale in Roma 52 (1924), 229–31Google Scholar. The Fornix Augusti (CIL VI 878Google Scholar), described by a late source as arcus marmoreus in platea pontis S. Marias (= Pons Aemilius), can be read as part of an aqueduct only with the aid of a vivid imagination.

⁵⁹ ‘Immine[ntem ruinam … aquae] / Anienis ṇ[ovae … avertit Val(erius)] / Anthidiu[s v(ir) c(larissimus) a(gens) v(ices) praef(ectorum) praet(orio) et urbi] / insisten[te…] / consula[ri aquarum…] / Nov(…) Syagrio [et Eucherio cons(ulibu)s]’. Although Frontinus separated the Anio Novus and the Aqua Marcia, they may still have crossed on a single bridge. The Marcia's name therefore may have filled the lacuna after the mention of the Anio.

⁶⁰ Bell, M., ‘Mulini ad acqua sul Gianicolo’, in Gigli, S. Quilici (ed.), Archeologia Laziale XI (Undicesimo incontro di studio del Comitato per l'archeologia laziale) (Rome, 1993), 65–72Google Scholar; Savage, , ‘Cults of ancient Trastevere’ (above, n. 9), 28Google Scholar; Wikander, , ‘Water mills in ancient Rome’ (above, n. 11), 13–36Google Scholar; Wikander, O., Exploitation of Water Power or Technological Stagnation? A Reappraisal of the Productive Forces of the Roman Empire (Lund, 1984)Google Scholar; Evans, , Water Distribution in Ancient Rome (above, n. 4), 130Google Scholar.

⁶¹ Wikander, O., ‘Water mills and aqueducts’, in Hodge, , Future Currents in Aqueduct Studies (above, n. 14), 141–8, at p. 143Google Scholar.

⁶² Calza, G., Notizie degli Scavi di Antichità (1932), 188–205Google Scholar; Muñoz, A, Il parco di Traiano (Rome, 1936), 13Google Scholar; Colini, A. M., Bullettino della Commissione Archeologica Comunale in Roma 66 (1938), 244–5Google Scholar.

⁶³ Bloch, , ‘Aqua Traiana’ (above, n. 2), 339–41Google Scholar.

⁶⁴ Evans, , Water Distribution in Ancient Rome (above, n. 4), 131Google Scholar; Bruun, , The Water Supply of Ancient Rome (above, n. 5), 135Google Scholar.

⁶⁵ Evans, , Water Distribution in Ancient Rome (above, n. 4), 107Google Scholar.

⁶⁶ Frutaz, A. (ed.), Le piante di Roma (Rome, 1962), tavv. 189, 208, 221Google Scholar.

⁶⁷ The Aqua Appia's ‘supra terram substructio et arcuatura proximum Portam Capenam passuum sexaginta’ in Fontinus, Aq. 5.5 was further north and much shorter.

⁶⁸ See Frutaz, , Le piante di Roma (above, n. 66), tavv. 228, 238, 245, 261, 278, 333Google Scholar. All of these maps follow Bufalini's orientation, with the praetorian camp at the top of the map.

⁶⁹ Frutaz, , Le piante di Roma (above, n. 66), tav. 228Google Scholar. Even a fragmentary apsidal building by the river nearby is reproduced faithfully.

⁷⁰ Frutaz, , Le piante di Roma (above, n. 66), tavv. 108, 407Google Scholar; Lanciani, , Forma urbis Romae (above, n. 34), tav. 34Google Scholar.

⁷¹ Giovanni Battista Piranesi studied the maps of both Bufalini and Nolli, or the latter's redrawing of the former's map, in reconstructing the aqueducts and walls for his antiquarian maps of Rome; see Frutaz, , Le piante di Roma (above, n. 66), tavv. 69, 70, 420Google Scholar. He seems to have known about the piers southwest of the main Aventine, but reconstructed them as part of the Servian Wall, pierced intermittently by republican gates. He clearly recognized Nolli's ‘antiche ruine’ in the middle of the river, but interpreted them as the piers for a conventional bridge.

⁷² See Evans, , Water Distribution in Ancient Rome (above, n. 4), 130 and n. 8Google Scholar.

⁷³ The ‘Naumachia Traiani’ is cited often as a principal destination for water from the Aqua Traiana. But it is far more likely that Trajan adapted the unpalatable Aqua Alsietina to this purpose, as Domitian had undoubtedly done for his own naumachia. Frontinus reported that none of the Alsietina's water was used inside the city (Aq. 85), but since we know that Domitian's naumachia was built ‘in a certain new place’ (Cass. Dio 67.8.2), this does not necessarily pose a problem.

⁷⁴ Wikander, , ‘Water mills in ancient Rome’ (above, n. 11), 28Google Scholar. More interesting is Wikander's proposal that the Aqua Alsietina, not the Traiana, was the primary power source for the mills (Wikander, , ‘Water mills in ancient Rome’ (above, n. 11), 24–5Google Scholar). Frontinus claimed that in his day the Alsietina was the lowest of all the city's aqueducts and served mostly the low-lying areas (Aq. 18.8). This must have referred only to the elevation of the terminal distribution tank, for the archaeological evidence suggests that the Alsietina entered the city from the heights of the Janiculum quite near the specus of the Traiana. It now seems evident that both aqueducts of the Janiculum supplied the mills. See Evans, , Water Distribution in Ancient Rome (above, n. 4), 112Google Scholar; Bell, M., ‘An imperial flour mill on the Janiculum’, in L'Italie méridionale et le ravitaillement en blé de Rome et des centres urbains des débuts de la République jusqu'au Haut Empire (Actes du colloque international de Naples, 1991; Collection de l'École Française de Rome) (Naples/Rome, 1994), 73–89, at pp. 81–2Google Scholar; Van Buren, A. W. and Stevens, G. P., ‘The Aqua Alsietina on the Janiculum’, Memoirs of the American Academy in Rome 6 (1927), 137–46Google Scholar; Van Buren, A. W. and Stevens, G. P., ‘The Aqua Traiana and the mills on the Janiculum’, Memoirs of the American Academy in Rome 1 (1915–1916), 59–62CrossRefGoogle Scholar.

⁷⁵ Cf. Evans, , Water Distribution in Ancient Rome (above, n. 4), 100, 125Google Scholar.

⁷⁶ The mention in the medieval Mirabilia and Graphia aureae urbis of two bridges south of the Pons Aemilius, the so-called Pons Theodosius and the Pons Valentinianus/-i, is probably a scribal error, the consequence of misinterpreting an earlier manuscript tradition that recorded a single bridge (no doubt the former Pons Probi) dedicated to the co-emperors Theodosius I and Valentinian II.

⁷⁷ Similarly, the Aqua Virgo followed a circuitous route to the north of the city to avoid heavily developed areas; see Quilici, L., ‘Sull'acquedotto Vergine dal monte Pincio alle sorgenti’, Quaderni dell'Istituto di Topografia Antica 5 (1968), 125–60Google Scholar; Evans, , ‘Agrippa's water plan’ (above, n. 4), 408Google Scholar.

⁷⁸ CIL VI 1260 = 31567Google Scholar; see Bloch, , ‘Aqua Traiana’ (above, n. 2), 337Google Scholar. This prescription seems to have applied to all of Rome's aqueducts, even as late as AD 330; see Codex Theodosianus 15.2.1. Augustan Venafrum had clearance zones that were likewise imperial property, but with eight feet of clearance on either side of the conduit (CIL X 4842Google Scholar); see Robinson, , ‘The water supply of Rome’ (above, n. 22), 61Google Scholar.

⁷⁹ Perhaps the arcus divi Traiani mentioned by the regionary catalogues in regio I was part of this segment of the arcade. Trajan may have owned property in the region, since the Privata Traiani (q.v., Richardson, A New Topographical Dictionary of Ancient Rome (above, n. 4)) are thought to have been here.

⁸⁰ Lanciani, , ‘Topografia di Roma antica’ (above, n. 3), 371Google Scholar. Of particular interest in Lanciani's survey of the evidence is a report from 1689 of two aqueduct channels in the slope of the hill below the north face of the temple of Claudius, (Cod. vat. 7849)Google Scholar and Piranesi's description of a channel ‘che girava e portava l'acqua all'intorno’ (Antiq. 1.9, n. 203).

⁸¹ Preller, , Die Regionen der Stadt Rom (above, n. 2), 246–7Google Scholar. Lanciani, , ‘Topografia di Roma antica’ (above, n. 3), 377–8Google Scholar, identified a tract of specus found under Sant'Onofrio on the Janiculum as part of a Hadrianic expansion of the Aqua Traiana. Hermansen, , ‘The population of imperial Rome’ (above, n. 11), 160–1 and n. 90Google Scholar, seems to have been unaware of the evidence for the Traiana's distribution east of the river.

⁸² The only vicus mentioned on the base that is demonstrably outside this geographic distribution is the Vicus Censori of regio XIV on Tiber Island. Perhaps Hadrian, by supplying morewater to the Transtiberim from the Aqua Traiana, was able to divert some water crossing the island by an older aqueduct. The absence of any vici from regio XI is another puzzle.

⁸³ Among these may be counted the Balinea Gratiarum, Antiochiani, Mercurii and Mamertini; and the Thermae Commodianae and Surae. The Balineum Bolani, also in regio I, may have been built by the consul of AD 111 rather than his father (q.u, Richardson, A New Topographical Dictionary of Ancient Rome (above, n. 4)). The later Thermae Severianae and Decianae probably benefited more from the augmentation of the Arcus Caelimontani in AD 201 (CIL VI 1259Google Scholar).

⁸⁴ It is possible, although unproven, that the Aqua Appia, flowing in the opposite direction, hugged the southern slope of the main Aventine on the way to its terminus at the Salinae. In this case, part of the Traiana's swath would have been cleared already for the maintenance of the underground conduit of the Appia. Despite the conclusion of Parker and subsequent topographers that the Appia passed directly through the core of the Aventine on its way to the Salinae (Parker, , The Aqueducts of Ancient Rome (above, n. 3), 10, n. tGoogle Scholar; Van Deman, , The Building of the Roman Aqueducts (above, n. 3), 26Google Scholar; Ashby, , The Aqueducts of Ancient Rome (above, n. 3), 53)Google Scholar, there is absolutely no evidence of this. Such an expedient is highly unlikely, given the simpler alternatives. Frontinus described the final stretch of the Appia as ‘sub Caelio monte et Aventino actus’ (Aq. 22.3). Archaeological evidence suggests that ‘sub Caelio monte’ does not mean that the aqueduct tunnelled through the heart of the hill, but rather that it was buried along the hill's contours (Parker, , The Aqueducts of Ancient Rome (above, n. 3), 6–7Google Scholar; Colini, A. M., Storia e topografia del Celio nell'antichità (Atti della Pontificia Accademia Romana di Archeologia ser. III, vii) (Rome, 1944), 82–7)Google Scholar. We should therefore interpret ‘sub Aventino in the same way.

⁸⁵ Valentini, R. and Zucchetti, G. (eds), Codice topografico della città di Roma, 4 vols (Rome, 1940–1953), I 193–258, and, for the date, esp. pp. 195–6Google Scholar.

⁸⁶ Nordh, A. (ed.), Libellus de regionibus urbis Romae (Skrifter utgivna av Svenska Institutet i Rom ser. II, vol. III) (Lund, 1949), 30, 94Google Scholar. See Jordan, , Topographie der Stadt Rom (above, n. 2), I.1476, n. 99; II 296Google Scholar; Valentini, and Zucchetti, , Codice topografico della città di Roma (above, n. 85), 245, n. 2Google Scholar.

⁸⁷ Panvinio, O., Reipublicae romanae commentariorum libri tres (Venice, 1558), 212Google Scholar; Burns, H., ‘Pirro Ligorio's reconstruction of ancient Rome: the ANTEIQUAE URBIS IMAGO of 1561’, in Gaston, R.W. (ed.), Pirro Ligorio, Artist and Antiquarian (Monograph of the Harvard University Center for Italian Renaissance Studies 10) (Florence, 1988), 19–92, at pp. 24–5Google Scholar.

⁸⁸ La Follette, L., ‘The Baths of Trajan Decius on the Aventine’, in Rome Papers (Journal of Roman Archaeology Supplement 11) (Ann Arbor, 1994), 6–88, at pp. 24–5, 33Google Scholar.

⁸⁹ Fabretti, , De aquis et aquaeductibus veteris Romae (above, n. 1), 174–5Google Scholar.

⁹⁰ Jordan, , Topographie der Stadt Rom (above, n. 2), I.1476, n. 99; II 296Google Scholar.

⁹¹ Lanciani, , ‘Topografia di Roma antica’ (above, n. 3), 511Google Scholar; La Follette, , ‘The Baths of Trajan Decius on the Aventine’ (above, n. 88), 24Google Scholar. On Ligorio's reputation, see Gaston, R.W., ‘Ligorio on rivers and fountains: prolegomena to a Study of Naples XIII. B.9’, in Gaston, Pirro Ligorio, Artist and Antiquarian (above, n. 87), 159–208, esp. pp. 163–72Google Scholar and the sources cited in nn. 21–2.

⁹² Lanciani, , ‘Topografia di Roma antica’ (above, n. 3), 418Google Scholar; Bloch, , ‘Aqua Traiana’ (above, n. 2), 340Google Scholar.

⁹³ Along with the inscription under discussion, Jordan condemned the testimony of a curator of the ‘Thermae Philippi’, a monument that had mysteriously appeared in Leto's edition. Ligorio's text reads as follows: ‘Prope aediculam S. Matthaei in Merulana, in thermis Philippi imp. inventa est haec inscriptio: L. RVBRIVS GETA CVR. P. CCCXXIII / D.N. PHILIPPI AVG. THERM…’ (Panvinio, , Reipublicae romanae commentariorum libri tres (above, n. 87), 168Google Scholar (not 173 as cited in Jordan)). On private conduit owners and their inscriptions on fistulae, see Bruun, , The Water Supply of Ancient Rome (above, n. 5), 63–96Google Scholar; on baths in particular, see pp. 72–6. Bruun showed that the pipe inscriptions follow certain formulaic patterns, none of which resemble Ligorio's. The nearest equivalents in the epigraphic record to a curator of a specific bath building are the curatores of individual temples (RE, s.v. ‘curatores’, 1803) and a Beneventan ‘curator opens thermarum’ (CIL IX 1419Google Scholar). But for an expanded understanding of the use of the term cura, see Bruun, , The Water Supply of Ancient Rome (above, n. 5), 104–6Google Scholar, and Corbier, M., ‘De Volsinii a Sestinum: cura aquae et évergétisme municipal de l'eau en Italie’, Revue des Études Latines 62 (1984), 236–74Google Scholar.

⁹⁴ Lanciani, , ‘Topografia di Roma antica’ (above, n. 3), 511Google Scholar, nos. 85 and 86: ‘aqua traiana bacchuleia 1. bacchuleius vatiens f.p. lxxxix’, found ‘sul colle Aventino verso il tempio di s. Maria’; ‘aquas traian. d. clodius albinus sua perd. cur. ped. cccc.lxxiiii, found ‘in diversi canalette di piombo nell'Aventino’. The other is in regio III (503, no. 15)Google Scholar.

⁹⁵ Burns, , ‘Pirro Ligorio's reconstruction of ancient Rome’ (above, n. 87), 22–3Google Scholar. However, the aqueduct does not appear on Ligorio's antiquarian maps of 1553 or 1561 (Frutaz, , Le piante di Roma (above, n. 66), tavv. 25–32)Google Scholar.

⁹⁶ Lanciani, , ‘Topografia di Roma antica’ (above, n. 3), 379Google Scholar. Piranesi reported a castellum by the purported baths, which he believed belonged to the Claudia. These ruins are probably to be associated with the Privata Traiani.

⁹⁷ Silverio, A. Liberati, ‘Aqua Alsietina’, in Il trionfo dell'acqua (above, n. 5), 72–9, at p. 73Google Scholar. Anderson, J. C. Jr‘The date of the Thermae Traiani and the topography of the Oppius Mons’ American Journal of Archaeology 89 (1985), 499–509, at pp. 505–6CrossRefGoogle Scholar. A large Domitianic pipe, found in situ along the Via di Santa Lucia in Selci, just north of die baths, in 1882 (Lanciani, , Forma urbis Romae (above, n. 34), 23Google Scholar; Bruun, , The Water Supply of Ancient Rome (above, n. 5), 136)Google Scholar may suggest that Domitian had already identified a water supply for the complex.

⁹⁸ See Le Gall, , Le Tibre (above, n. 33), 311Google Scholar. The bridge fell out of use between the building of Aurelian's wall and the compilation of the regionary catalogues during Constantine's reign; see Richardson, A New Topographical Dictionary of Ancient Rome (above, n. 4), s.vv. ‘Pons Neronianus’ and ‘Via Triumphalis (1)’; Coarelli, , ‘Il Campo Marzio occidentale’ (above, n. 31), 819–23Google Scholar; Jordan, , Topographie der Stadt Rom (above, n. 2), I.1417, n. 30Google Scholar. The Chronographus anni 354 (148) records extensive refortification of the city wall in the early fourth century. It is to the emperor Maxentius, as he prepared for invasion in the years or months leading up to the struggle of 312, that we should ascribe this work; and under Maxentius the Pons Neronianus likely met its end. Hadrian's tomb and the Pons Aelius were probably fortified at this time, the Neronianus spoliated, and its bridgehead gate on the left bank narrowed to a postern.