The castra of Frontinus

Andrew Wilson The castra of Frontinus Frontinus’ treatise on Rome’s aqueducts contains a detailed breakdown of water distribution for each aqueduct line, listing quantities of water distributed in the emperor’s name (in nomine Caesaris) and for private and public uses. The figures for public use are further broken down into distribution to castra, public buildings (opera publica), display fountains (munera), and public fountains (lacus). This is potentially one of the most important parts of his work for understanding how Rome’s distribution network functioned, yet it is also one of the most frustrating, as several of the key terms are unclear. First, the figures given for individual lines do not match the totals, as there has evidently been corruption of some of the figures in the manuscript tradition. Secondly, it remains uncertain whether the unit of measurement employed, the quinaria, has any real validity as a unit to measure the flow of water1. Frontinus seems to have understood it as a unit of cross-sectional area; but this is inadaquate for measuring water distribution as one needs to know the rate of flow through the aperture of a given section – which in the case of pipes is determined by the head of pressure, which is unlikely to have been uniform at all offtakes. Thirdly, it is not clear on what principles water was distributed in nomine Caesaris – does this refer to distribution to the emperor’s properties (and those of the imperial family), or to other public or military destinations, or also to private individuals to whom the emperor had granted privileges? Fourthly, what does Frontinus mean by the 18 castra which are listed as a category of distribution under opera publica? The most recent commentator on Frontinus, Rodgers, is unsure: ‘It is not clear what F. means by castra. The numbers alone indicate more than the castra praetoria, which was in F.’s day the barracks for the cohortes urbanae (RE Suppl. 10, 1026) as well as for the praetorian guard, and might in any case have received its water nomine Caesaris2.’ Following Evans3, he suggests that castra could have been used for the stationes of the cohortes vigilum – but does not remark (a) that there is no evidence that these were ever called castra rather than stationes, and (b) that there were only 7 of them (one for every two regions of the city), which would leave us 10 castra short of Frontinus’s total. There were also 14 excubitoria for the vigiles but if we wanted to include these (minor) buildings under the heading of castra we would now have too many entities (21 as against Frontinus’ 18). The Castra Misenatium, Castra Peregrina and Castra Ravennatium are only attested, epigraphically or archaeologically, well after Frontinus4. Rodgers continues: ‘Similar terminology might have been in use for collegia who provided public services (cf. TLL 3, 561.11); castra is thus used in the Notitia regionum for 1 See Bruun Appendix C in Rodgers 2004, 342-346 for the most recent and balanced discussion. Rodgers 2004, 245-246. 3 Evans 1994, 10. 4 LTUR I (1993), ad voces. 2 440 Andrew Wilson lecticarii, silicarii and the like, but their ‘military’ organisation (cf. RE 12, 1093) may have postdated F5.’ Indeed, the use of castra for the seats of these guilds is not attested before the fourth century AD, and moreover, one would expect such guilds to pay for their water use6. One could try to proceed in another manner. Frontinus, and other writers, such as Vitruvius, use the term castellum, which originally meant a small fort, in the sense of a water distribution tank (sometimes castellum divisorium). Castellum is of course the diminutive of castrum or castra, and one could argue that Frontinus should be using castra to mean some larger entity than a distribution castellum – for which the obvious candidate would be a large reservoir or storage cistern (as distinct from a settling tank, which is a piscina limaria). Bruun’s not dissimilar suggestion was that castra here meant ‘some sort of fountain, cistern or castellum … it cannot be regarded as implausible that another category of “public water posts” or of “local cisterns” was intended by castra7’. Evans rejected this argument on the grounds that castra in such a sense should be downstream of a castellum and would necessary be smaller than castella, and therefore the diminutive-master relationship should operate the other way around: ‘Even more problematic is the use of the term to indicate a “local cistern,” if castra were supplied through castella: we would expect the diminutive form to indicate the smaller reservoir or tank, not the larger one within the distribution net8.’ However, this misunderstands the physical realities of Rome’s enormously complex distribution system. Castella were not necessarily large structures; they need to be little more than junction boxes, or tanks from which a number of pipes branched off. The stone distribution tank at Porta Viminalis is very small9, while outside Rome, the two best known distribution castella, at Pompeii and Nîmes each held only a few cubic metres of water10. Reservoir cisterns would be substantially larger, with capacities of several hundred or thousand cubic metres. Secondly, a reservoir cistern might well be upstream of several castella, which acted as local distribution nodes at points along the branch governed by the reservoir cistern. If we understand castra as reservoir cisterns, there is no problem with the size relationship between them and distribution structures implied in the castra/castellum linguistic relationship. Indeed, the use of the military term castellum as a water distribution tank is otherwise unmotivated unless it is the diminutive of a larger water-related structure which was called castra. There is no other readily understandable reason to call a distribution tank a castellum; it does not look or act in any way like a fort. This begs the question, though, of why castra was used to mean a water reservoir? It might come either from the massiveness of the structure, or possibly from some loose sense of a parallel between storing water and housing troops – perhaps, especially, the use of castra for an overnight marching camp suggested a metaphorical transference to the overnight storage function of a reservoir. If castra were used in this way, the formation of a diminutive, castellum, for a smaller water-related structure would then be unsurprising. By the time Frontinus wrote, in the reign of Trajan, the technology and use of storage reservoirs was well developed. Their function was probably to accumulate reserves of water overnight, when demand was generally less, that could be distributed the following day. In 5 Rodgers 2004, 245-246, again following Evans 1994, 10. Bruun 1991 254-256. 7 Bruun 1991, 253. 8 Evans 1994, 10-11 n. 44. 9 Lanciani 1880, Tav. VI.2; Ashby 1935, 151. 10 Cf. Ohlig 2001 for the Pompeii castellum. 6 The Castra of Frontius 441 this way the demand placed by installations such as public baths could be mitigated11. The largest reservoirs might also accommodate some seasonal fluctuation in supply. The large terminal reservoir known as the Piscina Mirabilis on the Bay of Naples is Augustan, and the vast cisterns at La Malga in Carthage are now thought to be either Augustan or Claudian12. At Rome there is a considerable amount of evidence for the use of storage reservoirs on branches of the network, associated especially with large public baths and imperial palaces, and while the bulk of the evidence post-dates Frontinus, some belongs to the first century AD. The first-century examples include a Caligulan cistern below the so-called Domus Tiberiana on the NW corner of the Palatine13, and the Neronian cistern on the Palatine beneath the audience chamber of Domitian’s Domus Flavia. Lanciani considered the cistern on the Palatine under S. Bonaventura (one chamber of which served as refectory for the monks) to be probably Flavian. He calls it a piscina pensile, presumably meaning that the water-reservoirs were supported on vaulted substructures, and refers to a a large lead pipe on which was found a chiave di metallo corintio di peso fine a libbre novanta14 – a bronze stopcock weighing 90 pounds – suggesting that the cistern acted as a storage or regulation reservoir whose outlet could be controlled and implying an outlet pipe of correspondingly large size. A parallel might be drawn with the stopcock found in the tap chamber of the Bordj Djedid cisterns at Carthage15. More detail is available for a series of reservoir cisterns which postdate Frontinus, and while they cannot be used as direct evidence for the widespread use of reservoir cisterns in Rome at the time Frontinus wrote, they suggest that by Bruun’s “local cisterns” we should be thinking in terms not of small structures, but of sizeable reservoirs on local branches of the network. The Arcus Caelimontani branch of the Aqua Claudia supplied two reservoir cisterns on the Caelian at the place called Aquaeductium, near the substructures of the Temple of Divus Claudius. The recent study by Tucci demonstrates a Severan date for one of these, and shows that the other is pre-Severan, perhaps Trajanic or first-century AD16. Large cisterns in the Via S. Nicola da Tolentino, with a capacity of c. 1,965 m3 were originally considered by Lugli to be in Hadrianic brickwork, although he includes them in a list of Trajanic buildings dated AD 110-112; no detailed evidence is given for either supposition17. In 1873-1875 three aqueduct channels were seen heading towards the site of the Ministry of Finance, and Lanciani connected these, plausibly, with three large reservoir cisterns found during construction works for the Ministry. These he therefore identified as the terminal cisterns of the Marcia, Tepula and Julia, but considered them to be of the second century A.D. as the whole water supply arrangements of that area appeared to have undergone modification since the original construction of the lines18. These cisterns in fact governed only some branches of the aqueducts, as pipes also diverged upsteam of this point near the Porta Viminalis, for instance from the small circular distribution castellum in that area. Lanciani also records that large cisterns on the south side of Via Principe Eugenio, between Piazza Vittorio Emmanuele and Porta Maggiore were destroyed in the summer 11 Wilson 1998, 89-91; 2001. Hans Vanderleest, pers. comm. 13 Van Deman 1924. 14 Lanciani 1880, 160. 15 Wilson, 1998, 83-84. 16 Tucci 2006. 17 Lugli 1940, III, 33-37; 1957, 603. Capacity is estimated from dimensions of 4 chambers 38.55 m long x 4.32 m wide and 2.95 m high to the spring of the vault. The structure was on two levels, but the lower one served only as substructures, not filled with water. 18 Lanciani 1880, 95 and tav. VI fig. 1a. 12 442 Andrew Wilson of 1879; their brickstamps indicate a mid second-century date19. Approximate dimensions from the plan in FUR f. 31 suggest dimensions of 18.4 by 15.3 m, giving a footprint of c. 280 m3, which might equate to a capacity of c. 840-980 m3 given a water depth of 3-3.5 m. Lanciani in FUR f. 24 shows a piscina a. 1597 a Sisto V destructo near the E. end of Piazza Vittorio Emanuele. Dimensions from the plan appear to be c. 24 by 8 m, giving a footprint of 192 m2, which might imply a storage capacity of 400-700 m3, depending on water depth. Enormous reservoir cisterns were associated with the imperial thermae of the second and third centuries: The Sette Sale, feeding the Baths of Trajan, had a capacity of c. 7,000 m3, and the capacity of the cisterns for Baths of Caracalla is estimated at c. 10,000-11,500 m3 20. Pirro Ligorio’s measurements for the Botte dei Termini21, which supplied the Baths of Diocletian, suggest perhaps 6600 m3 – while Ligorio is notoriously unreliable, these figures are not out of line with those for the other reservoirs which fed imperial thermae. Frontinus’ delivery figures for different structures on different aqueduct lines may give some insight into the character of the various categories he lists (Table 1)22. The major discrepancy between Frontinus’ figures for totals and those obtained by adding up the figures for individual aqueduct lines occurs with castra – 37 quinariae (Frontinus’ total is too low). The figures for lacus are fairly consistent; they get from 2.04 to 2.45 quinaria on average each. Munera (ornamental display fountains) get between 2.0 and 22.33 quinaria each – most received c. 4 times the average lacus, though some were considerably more lavish. The greatest range comes with opera publica and the castra. Distribution to opera publica varied from the low figures of 2.33 for the Tepula and 2.73 for the Marcia, to the exceptionally high figure of 86.25 for the Virgo. This latter figure is undoubtedly influenced by the Virgo’s supply of a few large public works – the Baths of Agrippa, the Euripus and probably the Baths of Nero23; the category of opera publica probably included imperial thermae (but not privately operated smaller neighbourhood baths). Interestingly, the Virgo supplied no castra. Average figures for the distribution to castra on different lines vary from 4 to 50 quinaria, and this considerable variation may also provide a subsidiary argument against seeing them as supplying the stationes and excubitoria of the vigiles. Such variation might, however, be consistent with supplying reservoir cisterns of different sizes. In sum, while the manner in which Frontinus and his staff calculated their water distribution figures may never become fully clear to us, Bruun’s suggestion that by castra Frontinus meant some kind of specialised feature on the water distribution network remains attractive, and the best candidate for such a feature would be reservoir cisterns on local branches of the network. In the present state of our knowledge, a specialised use of the word castra to mean reservoir cisterns makes at least as much sense as trying to find enough separate barracks or “camps” for military or other associations to make up Frontinus’ total of 18. 19 Lanciani 1880, 174; tav. VIII fig 6a. Manderscheid - Garbrecht 1994 (12,500 m3); Lombardi - Corazza 1995 (10,000 m3). 21 Quoted in Lanciani 1880, 96-97. 22 Frontinus, De Aquis 78-86; usefully summarised in Bruun 1991, 102-103. 23 Evans 1994, 109. 20 The Castra of Frontius 443 Bibliography Bruun 1991: C. Bruun, The water supply of ancient Rome: A study of Roman Imperial administration (1991). Evans 1994: H. B. Evans, Water distribution in ancient Rome. The evidence of Frontinus (1994). Lanciani 1880: R. Lanciani, Topografia di Roma antica. I commentarii di Frontino intorno le acque e gli acquedotti. Silloge epigrafica aquaria (also printed in Reprinted in Memorie della Reale Accademia dei Lincei, Serie 3 no. 4 (1881), 215-616, which was reprinted as Le acque e gli acquedotti di Roma antica [1975]) (1880). Lombardi - Corazza 1995: L. Lombardi - A. Corazza, Le Terme di Caracalla (1995). Lugli 1940: G. Lugli, I monumenti antichi di Roma e Suburbio. Supplemento. Un decennio di scoperte archeologiche (1940). Manderscheid - Garbrecht 1994: H. Manderscheid - G. Garbrecht, Die Wasserbewirtschaftung römischer Thermen. Archäologische und hydrotechnische Untersuchungen A Forschungsbericht (1994). Ohlig 2001: C. P. J. Ohlig, De aquis Pompeiorum. Das Castellum Aquae in Pompeji: Herkunft, Zuleitung und Verteilung des Wassers With CD of Photographs (2001). Rodgers 2004: R. H. Rodgers, Frontinus: De aquaeductu urbis Romae (2004). Tucci 2006: P. L. Tucci, ‘Ideology and technology in Rome’s water supply: castella, the toponym AQVEDVCTIVM, and supply to the Palatine and Caelian hills’, JRA 19 (2006), 94-120. Van Deman 1924: E. B. Van Deman, ‘The House of Caligula’, AJA 28.4 (1924), 368-398. Wilson 1998: A. I. Wilson, ‘Water supply in ancient Carthage’, in Carthage papers: The early colony’s economy, water supply, a private bath, and the mobilization of state olive oil (JRA Supplement 28) (1998), 65-102. Wilson 2001: A. I. Wilson, ‘Urban water storage, distribution and usage in Roman North Africa’, in A. O. Koloski-Ostrow (a cura di), Water use and hydraulics in the Roman city (Archaeological Institute of America Colloquia and Conference Papers, New series 3) (2001), 83-96. Appia Anio Vetus Marcia Tepula Iulia Virgo Claudia and Anio Novus Calculated Total Frontinus’ Total in Aq. 78 Number of castella Delivery to castra (quinariae) Number of castra Average to Delivery to Delivery to Average to Delivery to Number Average to Average to Number Number lacus lacus munera opera publica opera publica of opera castra munera of lacus of munera (quinariae) (quinariae) (quinariae) (quinariae) (quinariae) publica (quinariae) 4.00 14 123 8.79 1 2 2.00 92 226 2.46 50.00 19 196 10.32 9 88 9.78 94 218 2.32 10.63 15 41 2.73 12 104 8.67 113 256 2.27 12.00 3 7 2.33 0 0 0.00 13 32 2.46 34.50 11 181 16.45 3 67 22.33 28 65 2.32 0.00 16 1380 86.25 2 26 13.00 25 51 2.04 20 35 51 14 17 18 1 1 4 1 2 0 4 50 42.5 12 69 0 92 9 149 16.56 18 374 20.78 247 18 326.5 18.14 96 2302 23.98 247 200 (!) 279 95 2301 12 39 39 107 8.92 226 485 2.15 394 10.1 591 1333 2.26 591 1335 386 Andrew Wilson Table 1: Numbers of water elements and deliveries to them in quinaria, by aqueduct line, according to Frontinus’ figures (de Aquis 78-86). The Alsietina is omitted because all its water was distributed extra urbem. 444 Aqueduct