An Imperial Flour Mill on the Janiculum
p. 73-89
Résumés
Nel 1990e ancora nel 1991 scavi di emergenza sul Gianicolo hanno portato alla luce i resti di un mulino ad acqua, precedentemente conosciuto soltanto tramite uno schizzo eseguito nel 1886 dal Lanciani. 11 mulino consiste in due canali paralleli, a fianco dello speco dell’Aqua Traiano. Nei canali erano sistemate ruote "undershot" mosse dall’acqua proveniente dall’acquedotto. La larghezza delle ruote è stata accertata (ca. 1. 70 m). La costruzione di un mulino pubblico nella prima metà del III sec. può essere col legata con la contemporanea fusione delle distributiones e dell’ufficio del curator aquarum.
In 1990 and 1991 salvage excavations on the Janiculum brought to light the remains of a water-mill, previously known to us only through a sketch made by Lanciani in 1886. The mill consists of two parallel canals, flanking the specus of the Aqua Traiano. In the canals were installed undershot wheels, turned by the water coming from the aqueduct. The width of the wheels was ascertained to be c. 1. 70 m. The construction of a public mill in the first half of the third century may be linked to the contemporaneous fusion of the distributiones and of the office of the curator aquarum.
Texte intégral
1During the era of urban expansion in late nineteenth-century Rome, Rodolfo Lanciani saw near the Porta San Pancrazio an ancient structure that he identified as a water-mill1. The monument disappeared almost immediately beneath the Via Giacomo Medici, one of the streets of the new residential quarter, and the only record of the discovery remains Lanciani’s pencil plan, apparently drawn during the construction of the street (fig. 1)2. Although the plan has always been of great interest for the study of both the topography of the Janiculum and ancient mills, it did not prove easy to recognize in its subject a water-mill, and the interpretation of the monument as a group of reservoirs, in some way associated with a mill, has generally been followed (Van Buren and Stevens 1917; Wikander 1979, 18-20, n. 39; Wikander 1985, 158-159).
2In the spring of 1990 and again in the fall of 1991 the laying of new electric and gas lines in the Via Giacomo Medici offered an opportunity to examine the structure seen by Lanciani more than a century earlier. With the authorization of the Soprintendenza Archeologica di Roma, three trenches were excavated, all within the confines of the street (fig. 2). It became apparent almost immediately that Lanciani’s complex had come back to light. Although limited in scope, the new probes produced considerable evidence for the identification, dating, and use of the building; and this included conclusive proof that the complex seen by Lanciani was indeed a water-powered flour mill.
1. Lanciani’s plan of 1886 (fig·1)
3There was first a question concerning the correct location of Lanciani’s complex. His plan describes a building through which pass three water-courses, located very close to the intersection of the Via Giacomo Medici and the Via Angelo Masina. During the construction of the American Academy in Rome in 1912-13, the specus of an aqueduct was discovered on the site of the new building, with the same alignment as that of the channels in the complex seen by Lanciani. The aqueduct was identified as the Aqua Traiano by Van Buren and Stevens, who considered it to be separate from Lanciani’s complex (Van Buren and Stevens 1917). More recently Lucos Cozza has demonstrated that the location of the mill, as shown in the pencil plan and in the FUR, is incorrect, at least as regards the present location of the intersection of the two streets (Cozza 1986-87). Lanciani was misled by a change in the actual layout of the new street system on the Janiculum, which shifted northwards the intersection, while maintaining unchanged the axis and orientation of the Via Medici itself. Retaining the former relation in the FUR, Lanciani’s draftsman drew the structure too far to the north. It should rather be located some twenty meters south of the crossing of the two streets. A first result of the excavation of 1990 was to prove Prof. Cozza correct: the aqueduct passes through the center of Lanciani’s complex, which is thus located some considerable distance from the intersection3.
4After initial uncertainty, owed in large part to the numerous and destructive modern trenches for gas and electric lines, it proved possible to establish the location on Lanciani’s plan of the walls visible in the excavation (fig. 2. trench 1). Of assistance was the north channel, which at a certain point turns at an angle of 60 degrees towards the aqueduct; the angle lay within the excavation and also appears in the pencil plan (figs. 1,3). It was thus a question of the central and right hand (or north) channels of the mill as shown on the plan. We quickly found that Lanciani’s recorded measurements were always precise, but that the architectural features were not correctly proportioned. It is then a question of a sketch, drawn no doubt during the work in progress along the new street4.
5It became evident that the Via Medici mill was very seriously damaged not long after its initial discovery. Various architectural features in the pencil plan did not appear in the excavation, a situation that created confusion until we realized that the structure had been cut down to a lower level in order to create a consistent grade for the new street (figs. 3, 4). The concrete walls excavated in 1990 and 1991 lie consistently, in fact, ca. 0,30 m below the present paving of the street5. We have then Lanciani’s sketch plan of the mill complex at an elevation higher than that of the preserved remains. The difference in level cannot be precisely measured, given the absence of elevations on the drawing; but in any case it is not likely to be much more than ca. 0,50 m6.
2. Ground plan and water channels
6In 1991 we acquired new evidence for the plan of the building (fig. 2). Trench 3 revealed the location and alignment of its eastern edge, which is more or less perpendicular to the aqueduct (fig. 8). A somewhat less well-preserved edge was found at the north in trench l7. From this it can be seen that the building was probably rectangular in plan and consisted of a concrete platform thrown over the pre-existing specus of the Trajanic aqueduct. The overall width is at least 13 m, the length perhaps 23 m8.
7As indicated in Lanciani’s drawing, three parallel channels cut through the complex (fig. 2). In the center is the specus of the aqueduct, continuing the alignment of the segment beneath the American Academy (fig. 5)9. The reticulatum walls are coated with hydraulic plaster, but the vault, noted by Lanciani as still standing, was evidently destroyed by the street-builders. The elevation of the bottom of the aqueduct below the curb along the west side of the Via Medici showed a difference of 0,204 m from the elevation taken beneath the American Academy (73,71 m a. s. l.). Van Buren and Stevens record the fifty-meter long section beneath the Academy as being almost level, with an inclination of 0,03 m in adistance of 26 m; the gradient to the east has markedly increased, as the specus begins its descent towards its presumed destination at the Villa Spada10.
8Of interest are three cuttings in the hydraulic cement of the specus, placed vertically on either side of the channel (fig. 5). These evidently served for the insertion of the frame of a hinged or sliding wooden door that could be used to block the flow of the aqueduct, diverting the water into one of the lateral channels.
9Flanking the aqueduct and parallel to it are the two channels, each 1,70 m wide, to be identified as mill-races (figs. 3, 4, 7, 8). These are paved in bipedales, their walls coated with hydraulic plaster. At the east the south channel was vaulted (fig. 2; the springing of the vault can be seen at the top of fig. 7)11. The barrel vault did not, however, continue the full length of the channel, and it seems certain that the south channel was unvaulted for most of its length12. The north channel was probably similar. The eastwards gradient of the north channel can be roughly calculated at 2% (0,04 m in 2,0 m), thus considerably greater than the almost level aqueduct beneath the Academy.
10Inserted in the north channel at a later period was a low dam with a triangular drainage hole (h. 0,72 m; figs. 3, 4, 10). The upper surface of the dam preserved a longitudinal groove apparently intended for a wooden beam, perhaps to support a sluice-gate above. To the east of the dam, lime incrustations on both walls of the north channel provided considerable evidence for water levels (visible in figs. 4, 10). At its greatest volume the water ran ca. 0,50 m deep. Heavy lime deposits ca. 0,15-0,25 m above the floor were produced by a diminished quantity of water flowing over the dam. Construction of the dam may have taken place at a time when water flow in the aqueduct had decreased; the height of the dam may correspond more or less to the original water level, which would then have declined by at least half during the period when the dam was in use13. Cut into the brick floor upstream from the dam is a rough channel leading to the triangular drain (visible in figs. 3, 4); this late and crudely cut device was evidently intended to eliminate any residual water in the north channel, at moments when the main flow had been returned to the aqueduct14.
3. The gear-pits
11The area between the two channels or millraces consists of a concrete platform covering the aqueduct. The original surface is entirely lost, but two important features, sunk beneath the original surface level, are partially preserved (figs. 2,6). These rectangular spaces can be recognized as gear-pits, and it is they that prove that the complex in Via Medici was indeed a water-powered flour mill15. In them meshed the cogwheels that transfered the power generated by the horizontal mill-wheel to a vertical axle, at the top of which was attached the upper millstone (fig. 14). The surviving north gear-pit 1 (fig. 6) was separated from the adjacent channel by a wall (w. 0,36 m). Through a window or port in this wall (the sill of which is 0,70 m above the floor of the channel) passed the axle of the mill-wheel. Set in the gear-pit wall opposite the axle-port was a reused marble bearing block of remarkable form (fig. 6), intended to sustain one end of the axle; and a similar block on the opposite side of the channel will have supported the other (see below). Only a part of south gear-pit 1, along the south millrace, was preserved; it was apparently similar to the better preserved north gear-pit 1, except for the fact that the floor level is ca. one meter higher (cf. fig. 14; on this difference, see below).
12Such gearpits are found in other mills and were intended to provide a dry space for the meshing of the cogwheels (Athens: Parsons 1936, 70-74, 83-85, fig. 10; Spain 1987, 344-348; Baths of Caracalla: Wikander and Schiøler 1983, 52, fig. 10). The gearing was relatively simple: a vertical cogwheel attached to the axle of the mill-wheel engaged with a lantern pinion mounted at the bottom of the vertical axle, at the top of which was attached the upper millstone. An approximate reconstruction appears in fig. 1416. According to the text of Vitruvius, the horizontal cogwheel was larger than the vertical one (resulting in a slowly turning millstone), but this is not likely to have been the case in the Via Medici mill (Landels 1978, 23-24)17. Lanciani recorded stone blocks flanking the axle-ports between the gear-pits and the north channel, and he also drew them at a larger scale (fig. 1 ; see also Van Buren 1917, pl. 15,4). These features, which were lost in 1886 along with the upper part of the structure, are best understood as the stone mounting blocks for the ends of the wooden bridge that supported the vertical axle; their proximity to the adjacent wall is so close that only a small lantern pinion can have been mounted on the vertical axle, as shown in fig. 14. With a much higher gear ratio, the millstones of the Via Medici mill thus turned at a faster rate than did those of the Vitruvian type.
4. The mill-wheels
13The notable width of the millrace (1,70 m) as well as the lack of an elevated water source are factors arguing for the use here of undershot mill-wheels, powered by the current, instead of the more common overshot wheels, turned by centrifugal force generated by water falling on them from above (on the distinction Landels 1978, 19-21). The wheels in the mills in the Athenian agora or the Baths of Caracalla are of the second type, which is more efficient. Vitruvius describes an undershot mill powered by a river, but in the imperial period aqueducts were often used as a reliable source of power, especially in cities. The Via Medici mill appears to be the first known example of a mill with undershot wheels driven by the water of an aqueduct18.
14A chance discovery, caused when a backhoe inadvertently cut a gas line to the American Academy, produced evidence for a second mill installation to the west of the first19. Here appeared the bearing block for the opposite end of the mill-wheel axle, inserted into the north wall of the north millrace (fig. 9); the marble block was scarred by the movement both of the axle journal and its metal housing. The axle of this second mill-station is 2,60 m distant from the axle of the first; with a diameter of ca. 2,30 m (calculated from the height of the bearing block above the millrace floor), the mill-wheels were evidently spaced very closely. Lanciani’s plan lets us restore at least four mill-stations in the north millrace, but if the estimate of the building’s length of ca. 23 m is correct, space would be available for at least two more, or six in all (fig. 13). There was presumably an equal number in the south millrace. The Via Medici mill, with its two parallel millraces and multiple mill-stations, thus strikingly resembles the well-known mill at Barbegal in Provence — at least in plan. At Barbegal there were eight stations in either millrace, and the source of water was again an aqueduct (Benoit 1940; Sellin 1983)20. Yet there are also differences: the Gallic mill was constructed on a steeper slope and must have required quite a different roofing system; and the mill-wheels were probably of the overshot type21.
15The Via Medici mill provides new evidence for the dimensions and functioning of undershot wheels in the Roman period. The width of ca. 1,65 m is much greater than that of any known examples of the overshot type22. The surface of the pinnae or paddles exposed to the current thus measured ca. 1,65 by 0,50 m, or 0,825 sq. m. The axles had a maximum length of 3,40 m., and, as already noted, the diameter of the wheel was ca. 2,30 m. No evidence is preserved to indicate whether the wheels were equipped with the external rims necessary in overshot models23.
5. The water supply
16It seems certain that the north millrace was fed by water diverted from the Aqua Traiana, by means of the mechanism described above24. The section shown in fig. 14 reveals that there is a difference in level of ca. one meter between the floors of the gear-pits of the north and south millraces. If, as seems probable, the wheels in the two millraces were of the same dimensions, it follows that the floor of the south millrace must have been about a meter higher than that of the north25. If this assumption is correct, then it is unlikely that the Aqua Troiana can have served as the source of water in the south millrace. Given the very slight gradient of the specus, it would have been necessary to tap the aqueduct at some very distant point, in order to bring water into such an elevated millrace.
17The south millrace may then have been fed by a different source. In 1927 Van Buren and Stevens noted the existence of a second aqueduct channel in the garden of the Villa Richardson, some 75 m SE of the Via Medici mill (fig. 15, b; Van Buren 1927). The specus was identified as that of the Augustan Aqua Alsietina. Not far away was a second channel, parallel to the first but wider. Van Buren and Stevens consider the masonry of the second channel to be «of about the third century A. D. », but say little else about it. The combination of specus and parallel wider channel strongly suggests the presence of another mill, outside the circuit of the walls of Aurelian. Could the water from this second aqueduct (presumably the Alsietina) have been rerouted to the Via Medici complex, to drive the wheels in the south millrace?26 Unfortunately there is no indi cation of the elevation of this second specus at the crest of the hill where the water could have been diverted27.
18In this hypothesis water from both the Traiano and the Alsietina powered the Via Medici mill and, after driving the mill-wheels in the two millraces, flowed together in the specus of the Traiano. Although such mixing of waters of different quality may seem unlikely, it may not have been considered a problem if the primary function of both aqueducts had become the production of power for state-run flour mills. In any case, water of whatever quality that had passed through the flour-dusted millraces would probably not have been considered potable down in the Transtiberim.
19The difference in level of the two millraces may also have been expressed in a two-tiered working floor for the mill above, but for practical reasons it seems more likely that a single level was maintained there (fig. 14)28. The vertical axles of the northern row of mills will in that case have been approximately one meter longer than those to the south. In the restored section, it is assumed that the floor of the mill consisted of planks laid over millraces and gearpits, supported on the various walls, now lost, that appear in Lanciani’s plan.
6. Evidence for dating
20Both gear-pitspits contained deposits of considerable interest. Numerous tiles indicate that the mill was roofed, as we would have supposed. Two lamps with burned nozzles suggest that the gear-pits were dark and required illumination (fig. 12)29. There were many fragments of well-used millstones, both upper catillae and lower metae (fig. 11)30. Most were of grey basalt containing creamy quartz-like inclusions, a stone possibly to be identified as leucitophyre, commonly used for millstones and quarried near Orvieto31. The diameter of the millstones is ca. 0,75 m; it seems possible that two similar but perfectly preserved, and apparently little used, metae in the collection of the American Academy come from the Via Medici mill32. Organic material included several kernels of wheat33. Finally we can note the presence of three bronze coins from the floor of north gearpit 1. More than seven-hundred small coins were found in the mill in the Athenian agora, thirty in the gear-pit there, and similar coins have been found in other mills; such coins probably fell through cracks in the plank flooring and may document small purchases of flour by private citizens (Parsons 1936, 88)34.
21The only evidence for dating the mill itself came from the 1991 excavation. Although outside the mill structure nothing seems to be preserved of the ancient ground level, the lower part of a foundation trench does survive along the NE side of the building, cut into the clay of the hillside to permit the laying of the concrete substructure (trench 3, fig. 8). The latest material from the foundation trench includes the rim of an African red-slip plate of the first half of the third century35. The mill cannot thus be apart of the original Trajanic project completed in A. D. 109. The same conclusion is also clearly indicated by the return of the north millrace to the main channel of the aqueduct: here the builders of the mill had to cut through the pre-existing aqueduct wall, and in doing so they left the Trajanic aggregate exposed, plastering only the millrace walls up to their juncture with the aqueduct wall (fig. 7, top). We would not expect such rough work if the mill had been part of the original plan. The available evidence thus argues that the Via Medici mill was constructed during the third century.
22As for its abandonment, there are two sorts of evidence. The gear-pit deposits contained nothing later than ca. A. D. 400. The pottery is local, typical of the fourth century in its open, carinated forms and micaceous fabric; and the lamps are of Bailey’s "fat-globule" type, dated again to the fourth century (fig. 12)36. Of the three small bronzes from the floor of the north gear-pit n° 1, the two that can be identified are quarti of the later fourth century37. Both mill-stations appear to have gone out of use ca. A. D. 400.
23It is nonetheless certain that water continued to flow through the north millrace at least as late as the second half of the fifth century. Three nummi were found in a small depression in the floor of the millrace, one of which (the only legible coin) is a Roman issue of Libius Severus with the monogram of Ricimer, of ca. 461-46738. It thus seems possible that the mill was abandoned gradually. In any case our somewhat limited evidence does not indicate that the mill in the Via Medici was still in operation at the time of the siege of 537, when – the aqueducts cut by the Goths – Belisarius transfered vertical mill-wheels from the Janiculum to the Tiber in order to continue providing bread to the besieged population. As Wikander has noted, the narrative of Procopius indicates that those wheels had a width of two feet, and so they must have been of the overshot type; the much wider undershot wheels of the Via Medici mill (which, with a width close to six feet, would have been better adapted for use in the river) had evidently ceased to turn in the fifth century and so were not available at the time of the Gothic emergency (Wikander 1980, 31-32).
7. Water-mills and the distributiones
24Construction of the Via Medici mill in the third century implies a new primary function for the water of the Aqua Troiana, a major acqueduct39. The decision to take such a remarkable step must have been made at a very high level, and it indicates clearly that the mills were state-owned and -operated. The production of flour had previously been accomplished privately, by bakers or individuals (Rickman 1980, 197; Chastagnol 1960, 56-59; Coarelli 1987, 452-456). The entry of the state into the process of bread production implies the existence of new administrative structures.
25Wikander has suggested that the mills of the Janiculum were constructed by Aurelian, whose city walls in the Transtiberim were specifically intended to defend the mills (Procopius, 5, 19, 9), and who is generally credited with having begun distributions of bread to the urban population, in place of the traditional grain (Wikander 1979, 23-24)40. As has been seen, the archaeological evidence for the Via Medici mill would permit a construction date in the time of Aurelian (270-275). However, the ancient sources in fact state only that Aurelian increased the quantity of the bread allotment, and the change from grain to bread could have been introduced by a predecessor41. Coarelli has argued that this was done by Severus Alexander (Coarelli 1987, 448-456)42. As he has rightly pointed out, the shift to bread distributions must have required numerous and profound changes in the administration of the frumentationes, including the assumption of new responsibilities by the state in the local transport of grain and flour, milling, storage, baking, and a new method for the actual distributions. Decentralization must have been the general aim. Unlike grain, which could be handed out monthly from a central point (the Porticus Minucia), bread had to be distributed daily at many different places throughout the city (Coarelli 1987, 453-454; Rickman 1980, 197). The change from grain to bread distributions cannot have been made without careful planning, and it may not have taken place all at once43. Assumption of the task of milling flour may then have been one of the necessary changes, and the construction of the Via Medici mills could be associated with the decision by the administration of the frumentationes to begin distributing bread in place of grain. The third-century mill in the substructure of the Baths of Caracalla could be yet another element in a new system; as Wikander and Schiøler have noted, the two-station mill there may well belong to the original construction of the baths, which were dedicated in A. D. 216 (Wikander 1983, 61-62; Coarelli 1987, 450)44. Use of both the location and the water-supply would in any case have required state authorization45.
26Indeed decisions to use aqueduct water to grind grain would everywhere have required the consent and cooperation of the curator aquarian. It is therefore tempting to associate the construction of the new mills on the Janiculum and in the Baths of Caracalla with the merging of the administrations of the curator aquarian and the frumentationes (Rickman 1980, 97, 253-256; Pavis 1976, 35-36; Chastagnol 1960, 56-57)46. The new imperial post of curator aquarum et Miniciae is first known under Septimius, under whom the merging of the two offices may have taken place47.
27No specific reason has been adduced for the decision to combine the previously separate administrations (cf. Rickman 1980, 196). The potential of the aqueducts as a convenient power source for grinding flour may have been a major incentive. The background for this change will then have been the desire to shift to a system of bread distributions. Technological considerations (use of aqueducts to power mills with multiple stations, improvements in mill efficiency) must have played a part. Consequences of the changeover would include the construction of mills, presumably throughout the city; the extension of state control over the pistores; and the merging of the previously distinct administrations. Construction of the new mills necessarily drew on the practical experience of the personnel of both frumentationes and the office of the curator aquarum. The survival of the senatorial position of praefectus frumenti danài (now evidently subordinate to the curator aquarum et Miniciae), at least until the time of Gordian III (238-244), or Philip the Arab (244-249), may indicate that in certain parts of the city, or for some citizens, grain distributions continued, along side the new distributions of bread48.
8. Industrial activity on the Janiculum
28The Via Medici mill was not the only industrial plant on the Janiculum for which there is archaeological evidence. A probable second mill in the garden of the Villa Richardson, powered by the Aqua Alsietina, has already been noted (fig. 16, b). In 1933 Van Buren and Stevens documented further evidence. The Aqua Traiana follows the same orientation east of the Via Medici mill, continuing in the direction of the seventeenth-century Villa Spada, which has long been thought to mark the site of the monumental fountain at the urban end of the aqueduct, and which now serves as the residence of the Irish ambassador to the Holy See (fig. 16, e)49. In 1928, during the construction of the Villino Monami, which lies between the Via Medici and the Villa Spada, Van Buren and Stevens noted the presence of both a secondary aqueduct veering to the SE and a Watergate of travertine, which had evidently been installed immediately downstream in the specus of the Traiana (fig. 16, c; Van Buren 1933, 71-72)50. This secondary channel is oriented toward a building of opus caementicium on the site of the present Norwegian Institute (fig. 16, d; Van Buren 1933, 73)51. The travertine Watergate must have controlled the flow from the specus of the aqueduct into the secondary channel. There is nothing to reveal what purpose this arrangement may have served. The slope is greater here and a water-mill with overshot wheels is a possibility52.
29The excavations of 1990 and 1991 have thus exposed small parts of what must have been a more extensive industrial establishment on the Janiculum, made possible by the source of power provided by the two aqueducts. In an earlier time the hilltop was presumably given over to villas and gardens, as it is in part today, but in the third century of our era it was totally transformed as a consequence of the adoption of a new plan to feed the urban populace and the decision to exploit the aqueducts as a source of power. The importance of the complex on the Janiculum is indicated by the decision of Aurelian to defend it with a long and expensive salient of the new city wall. Other components of this industrial complex are likely to survive within the third-century walls, and future excavation and study here would probably cast further light on questions of historical, topographical, and technological interest.
Discussion
30F. ZEVI: Volevo soltanto chiedere a Malcolm Bell se tra l’Aqua Traiana, che immagino sia di Traiano,e la legislazione traianea a proposito degli incentivi che si davano ai pistores lui vede un rapporto.
31M. BELL: È un’idea molto interessante. Però Frontino non può parlare dell’Aqua Traiana, e non parla nem-meno della possibilità di mulini attivati dagli acquedotti. Sarebbe una novità connessa al compimento dell’Aqua Traiana. ma purtroppo mancano le fonti scritte. Invece per quanto riguarda i nostri scavi non abbiamo trovato la minima indicazione della datazione dell’edificio (scavi 1990). Nel futuro speriamo di allargare lo scavo verso l’Accademia Americana dove c’è la speranza di trovare l’altra metà dell’edificio. Ma non vorrei azzardare, a questo punto, una data nel II secolo per il complesso. [Per gli scavi del 1991, posteriori al dibattito, che hanno rivelato informazioni cronologiche, v. il testo sopra].
Bibliographie
Des DOI sont automatiquement ajoutés aux références bibliographiques par Bilbo, l’outil d’annotation bibliographique d’OpenEdition. Ces références bibliographiques peuvent être téléchargées dans les formats APA, Chicago et MLA.
Format
- APA
- Chicago
- MLA
Bibliographical references
Acqua 1986: Il trion fo dell’acqua: acquee acquedotti a Roma, iv sec. a. C. -xx sec. Rome, 1986.
Ashby 1935: ASHBY (T. R.), The Aqueducts of Ancient Rome (ed. LA. Richmond). Oxford, 1935.
Bailey 1975: BAILEY (D. M.), A Catalogue of the Lamps in the British Museum. I. London, 1975.
Bell 1992: BELL (M.), Mulini ad acqua sul Gianicolo. Archeologia Laziale, XI, 1992,67-74.
Benoit 1940: BENOIT (F.), L’usine de meunerie hydraulique de Barbegal (Arles). RA. 15, 1940, 19- 80.
Celano 1974: CELANO (C.), Notizie del hello, dell’anticoe del curioso della città di Napoli. Napoli, 1974 (reprint).
Chastagnol 1960: CHASTAGNOL (Α.), La préfecture urbaine à Rome sous le Bas-Empire. Paris, 1960 (Publ. de la Fac. de Lettres et Sciences Humaines d’Alger, 34).
Coarelli 1981 : COARELLI (F.), L’area sacra di Largo Argentina. I, 1. Topografia e storia. Roma, 1981 (St. e mater, dei Museie Monum. comun. di Roma, X).
Coarelli 1987: COARELLI (F.), La situazione edilizia di Roma sotto Severo Alessandro. In: L’Urhs. Espace urbain et histoire (ier siècle av. J. -C. -iiie siècle ap. J. -C.). Rome, 1987 (ColI. EFR. 98), 448-456.
Cozza 1987: COZZA (L.), Mura aureliane, 2. Trastevere. Il braccio meridionale: dal Tevere a Porta Aurelia-S. Pancrazio. BCAR. 92, 1987-88, 162-169.
Grassi 1988: GRASSKM. T.), Rinvenimenti monetali da Angera, scavi 1980-1984. BNum, 11. 1988.
10.1017/S0068246200011193 :Hayes 1970: HAYES (J. W.), Late Roman Pottery. London, 1970.
Hill 1960: HILL (P. V.), KENT (J. P. C.), CARSON (R. A. G.), Late Roman Bronze Coinage. London. 1960.
Homo 1904: HOMO (L.), Essai sur le Règne de l’Empereur Aurélien (270-275). Paris, 1904 (BEFAR. 89).
Jacono 1938: JACONO (L.). Tecnologia romana, la ruota idraulica di Venafro. In: L’Ingegnere, 15 dicembre 1938, 850-853.
Johannowsky 1990: JOHANNOWSKY (W), Appunti su alcune infrastrutture dell’Annona romana tra Neronee Adriano. Boll. Archeol., 4, 1990, 1-13.
Landels 1978: LANDELS (J. G.), Engineering in the Ancient World. London, 1981.
Parsons 1936: PARSONS (A. W.), A Roman Water-mill in the Athenian Agora. Hesperia, 5, 1936,70-90.
Pavis 1976: PAVIS D’ESCURAC (H.), La préfecture de l’annone, service administratif impérial d’Auguste à Constantin. Rome, 1976 (BEFAR, 226).
Peacock 1980: PEACOCK (D P S.), The Roman Millstone Trade, a Petrological Sketch. World Archaeology, 12, 1980. 43-53.
Pflaum 1963: PFLAUM (H. G.), Du nouveau sur les agri decumates à la lumière d’un fragment de Capoue, CIL, X, 3872. BJ, 163, 1963, 233-237.
10.56021/9780801825545 :Reynolds 1983: REYNOLDS (T. S.), Stronger than a Hundred Men, a History of the Vertical Water Wheel. Baltimore - London, 1983.
Rickman 1980: RICKMAN (G.), The Corn Supply of Ancient Rome. Oxford, 1980.
Roos 1986: ROOS (P.), For the Fiftieth Anniversary of the Excavation of the Water-mill at Barbegal: a Correction of a Long-lived Mistake. RA, 1986, 327-333.
Sagui 1948: SAGUI (C. L.), La meunerie de Barbegal (France) et les roues hydrauliques chez les anciens et au Moyen-Age. Isis, 38, 1948, 225-231.
Sellin 1983: SELLIN (R. J.), The Large Roman Water-mill at Barbegal (France). HTechn, 8, 1983, 81-109.
Spain 1984: SPAIN (R. J.), The Second-Century Romano-British Water-mill at Ickham, Kent. HTechn, 9, 1984, 143-180.
Spain 1987: SPAIN (R. J.), The Roman Water-mill in the Athenian Agora, a New View of the Evidence. Hesperia, 56, 1987, 335-353.
Tengström 1974: TENGSTRÔM (E.), Bread for the People. Studies of the Corn Supply of Rome during the Late Empire. Stockholm, 1974.
10.2307/3601953 :Thompson 1954: THOMPSON (M.), Coins from the Roman through the Venetian Period. Princeton, 1954 (The Athenian Agora, II).
10.2307/4238499 :Van Buren 1917: VAN BUREN (A. W.) and STEVENS (G. P.), The Aqua Traiana and the Mills on the Janiculum. MAAR. 1, 1917, 59-61.
Van Buren 1927: VAN BUREN (A. W.) and STEVENS (G. P.). The Aqua Alsietina on the Janiculum. MAAR, 6, 1927, 137-146.
10.2307/4238575 :Van Buren 1933: VAN BUREN (A. W.) and STEVENS (G. P.), Antiquities of the Janiculum. MAAR, 11,1933,69-79.
Wikander 1979: WIKANDER (Ö.). Water-mills in Ancient Rome. OpRom, 12, 1979, 13-36.
Wikander 1983: WIKANDER (Ö.) and SCHI0LER (T.), A Roman Water-mill in the Baths of Caracalla. OpRom, 14, 1983, 47-64.
Wikander 1985: WIKANDER (Ö.), Archaeological Evidence for Early Water-mills-an Interim Report. HTechn, 10, 1985, 163-165.
Wikander 1991: WIKANDER (Ö), Water-mills and Aqueducts. In: Future Currents in Aqueduct Studies (ed. A. Trevor Hodge). Leeds, 1991, 141-148.
Notes de bas de page
1 I am grateful to Dr. Emmanuele Gatti of the Soprintendenza Archeologica di Roma for having entrusted the excavation of the mill seen by Lanciani to the American Academy. Dr. Franco Astolfi provided valued counsel in 1990, Dr. Lanfranco Cordischi in 1991. Prof. Lucas Cozza followed the progress of the excavation and offered useful assistance. In 1990 we received drafting assistance from the British School in Rome through the kindness of its director, Dr. Richard Hodges; I am especially grateful to Cathy Couts and Andrew Hanesh. Paul Roberts of the British School is preparing a report on the late Roman pottery. Prof. Kathryn Gleason (University of Pennsylvania) reported on the organic remains. The work was carried out in both seasons by volunteers from the American Academy. Special thanks are owed to Karin Einaudi, Elfriede Knauer, Elizabeth Fentress, Fred Biggs, Brandt Junceau, and David Wright; Dora Yeoh contributed to the drawing that appears here in fig. 15. The written sources for water-mills in Rome have been conveniently collected by Örjan Wikander (1979, 13-35). A brief report on the Via Medici mill appears in Bell 1992, 67-74.
2 Biblioteca Vaticana, Cod. Vat. Lat. 13043, 121 r. The plan was copied by Lanciani in Forma Urbis Romae, Milan, 1893-1900, folio 27; and in Van Buren 1917, 59, n. 2; 60. fig. 1; and again in 1933, 70-71, fig. I. The plan itself was first published in Cozza 1987, 162, fig. 46. Although it bears the clearly written date of 2 March 1880, Cozza has presented strong reasons for believing that the year should be 1886 and not 1880, when the development of the Janiculum had not yet begun. It seems at least possible that the Vatican drawing is a later copy of an earlier plan, made at the time of the discovery of the mill in March, 1886. The inscribed date of 1880 might have resulted from a misreading of the date of the original drawing. The preparation of the plates of the FUR in the late 1890’s might be a possible occasion for the execution of such a copy.
3 Lanciani’s error resulted in a contradiction regarding the line of the city wall of Aurelian: on the pencil drawing (fig. 1) a section of wall can be seen only two meters south of the complex, identified as "mura aureliane"; but in FUR that identification had to be abandoned, as the known line of the wall was some twenty meters to the south.
4 It will be noted that the parts of the mill accessible to Lanciani do not extend much beyond the edges of the new street into the adjacent lots. It is therefore likely that he could draw only what was visible on the surface, no doubt in the difficult conditions of an open construction project.
5 The 1991 excavation indicated all too clearly that little if anything remains of the original ground levels adjacent to the structure. A long trench cut by Italgas in the street to the south of the complex did not, in fact, encounter any remains of Aurelian’s wall, which Lanciani saw here in 1886. It was probably entirely destroyed in building the street. Uphill to the south, ten meters outside the line of the wall, the Italgas trench cut through what appeared to be a ditch parallel to the city wall, a feature that may have been a part of the fortifications on the crest of the hill. The ditch was filled with domestic debris of the fifth century A. D.
6 See the text below at n. 16.
7 The south side of the building indicated in figs. 2 and 13 is hypothetical, the north side serving as a model in its relation to the north millrace. Lanciani’s drawing, difficult to interpret, may indicate a different form for the south side.
8 At the west the building must end at the point where the aqueduct passes beneath the city wall. The length cannot thus be greater than ca. 23 m.
9 Dimensions: w. 0,95 m; h. to shoulder of vault: 1,38 m. The description offered in Van Buren 1917, 59-61, applies as well here.
10 On the Villa Spada, see text at n. 49.
11 As in the case of the aqueduct, the vault of the south channel was destroyed in 1886; Lanciani’s drawing notes its existence.
12 For the method of covering the two millraces, see text at n. 28.
13 The specific purpose of the dam in the functioning of the millrace remains uncertain.
14 The channel is flanked by low mortar “shoulders” resting on the floor. These would have interfered with the action of the mill-wheel in north station 1, which must therefore have been out of use when the channel was cut, presumably in the fifth century.
15 North gear-pit 1: 1,52 by 0,77 m.
16 This replaces the initial attempt at a reconstruction published in Bell 1992, fig. 8.
17 Vitruvius, De Arch., X, 5, 2:... in uno capite axis tympanum dentatum est inclusum. Id autem ad perpendiculum conlocatum in cultrum versatur cum rota pariter. Secundum id tympanum maius item dentatum planum est conlocatum, quo continetur. Ita denies tympani eius, quod est in axe inclusum, inpellendo denies tympani plani cogunt fieri molarum circinationem.
18 Wikander 1991, 141-148. The seventeenth century Carmignano aqueduct at Naples was intended to power numerous mills in the city: see Celano 1974,461 -466. I thank Carlo Knight for this reference.
19 The term “mill station” will be used to refer to the individual installations, each equipped with a millstones. wheel, gears, and set of millstones.
20 See also Roos 1986, 327-333. The Barbegal mill has recently been studied by Ph. Leveau, whom I thank for kindly communicating to me the preliminary results of his work.
21 For a suggestion that Barbegal had undershot wheels. Reynolds 1983,40; against such a view are the following points: 1) the considerable slope at Barbegal, more suitable for overshot wheels; 2) the narrow wheels actually used there (w. 0,70 m); and 3) the like lihood that overshot wheels would be used wherever possible, given their greater efficiency.
22 Baths of Caracal la: w. ca. 0,50 m, d. ca. 1,95-2,10 m (Wikander 1983,53 and fig. 7). Barbegal: w. ca. 0,70 m., d. ca. 2,2 m (Benoit 1940). Athens: w. 0,54 m, d. 3,24 m (Parsons 1936, 80, fig. 10). Venafro: w. 0,297, d. 1. 85 m (Jacono 1938, 850-853).
23 Rims do not seem necessary in undershot wheels. The amount of energy produced by an ancient undershot wheel has not yet been calculated. That it was less than that of an overshot wheel is clear (Landels 1978, 21 -22). For an estimate of the power and productivity of the Barbegal mill, Sagui 1948,225-231 ; for a more recent and realistic estimate, Sellin 1983, 98-101.
24 On the two aqueducts that crossed the Janiculum, the Traiano and the Alsietina, Ashby 1935, 183-179 (Alsietina), 295-307 (Traiana); Acqua 1986,72-79 (Alsietina, A. Liberti Silverio). 113-19 (Traiana, P. Virgili).
25 The floor of the south millrace was excavated in trench 2, at the point where the channel changes direction to return to the aqueduct (fig. 8). Here the level is approximately the same as that of the aqueduct (and therefore of the north millrace). If the assumptions made in the text are correct, there must have been a one-meter change in level in the south millrace, between south gear-pit I and the return to the acqueduct.
26 If the putative mill in the garden of the Villa Richardson was abandoned because it lay outside the wall of Aurelian, and if the Alsietina was only then rerouted to the Via Medici mill, then the latter establishment would necessarily belong to the period of Aurelian. There is no evidence to suggest that the southern millrace was added to the Via Medici mill, that the mill was, in effect, constructed in two phases.
27 Van Buren and Stevens record the elevation of the channels in the lower garden of the Villa Richardson as 71 m above sea level.
28 A two-tiered main floor was suggested in the reconstruction published in Bell 1992,74, fig. 8.
29 For the lamps, see n. 36. Lamps were also found in the gear-pit of the mill in the Athenian agora: Parsons 1936, 88.
30 A list of known millstones used in water-mills ("powered millstones") is provided in Wikander 1985, 163-165.
31 The quality of stone is evidently that described by Peacock 1980,43-53, esp. 44-46.
32 Lanciani records millstones NE of the mill; see fig. 1. I have not been able to establish when these two metae, which are of a different quality of basalt, arrived at the American Academy.
33 Report of Prof. Kathryn Gleason.
34 Most of the coins from the mill in Athens come from the area that was presumably beneath the mill floor (Parsons 1936, 88); the comparable space in the Via Medici mill is not preserved. For coins from a rural mill in England, Spain 1984, 170.
35 Hayes 1970,52-55, form 31, n° 4: «early-mid third century».
36 Bailey 1975, type R, 377-381, in particular the example Q.
37 The two legible coins bear the legends Securitas Reipublicae (ca. A. D. 364-378) and Concordia Augustorum (ca. A. D. 375-383); Hill 1960, nn. 710-738,743-762. It is tempting to associate these two abandonment deposits with Alaric’s sack of the city in A. D. 410.
38 Thompson 1954, 101, n° 1677, p. 2; Grassi 1988, 133, n° 284. An illegible nummus was found on the floor of the south millrace, which must also have continued in use into the fifth century. The picture offered by the excavation of continued operation at least into the second half of the fifth century is confirmed by CIL, VI, 1711, the Edict of the urban prefect Dynamius concerning the short-changing of customers by the molendinarii, usually dated to the 480’s. The inscription was seen by the Anonymus Einsiedlensis on the Janiculum.
39 As has been suggested above, water from the Alsietina was probably also used in the new mill.
40 Homo 1904, 177-183.
41 S. H. A., Aurei., 47, 1 ; Aur. Vict., 35, 7; Zosimos, 1,61.
42 Severus Alexander constructed horrea publica in omnibus regionibus (S. H. A., Alex., 39, 3)-a decentralization of grain storage that might be called for in a new system of bread distributions; and he also set up opera mechanica plurima (S. Η. Α., Alex., 22,4), which Coarelli interprets as mills.
43 A law of A. D. 364 indicates that milling might still take place in the bakery; Cod. Theod14, 3, 7.
44 The substructures of other imperial baths could also have provided the necessary space and power for flour mills, as a fifth-century writer recommends: Palladius, Opus agriculturae, I, 41 (42): si aquae copia est, fusuras balnea rutti debent pistrina suscipere, ut ibi formatis aquariis molis sine animalium vel hominum laborefrumentafrangantur.
45 For what would seem a multiple-wheel mill near Beneventum, beneath a bridge on the Via Appia and presumably public, Johannowsky 1990, 10.
46 On the statio aquarum, and the interesting proposal that it was located in and around the northernmost of the temples in the Largo Argentina, adjacent to the Porticus Minucia Frumentaria, Coarelli 1981, 43-45.
47 CIL, V, 7783; CIL, XIV, 3900. On the grounds of the existence of a praefectus Miniciae in ca. A. D. 188 (CIL, VIII, 12442), H. G. Pflaum has dated the merging of the two administrations to the time of Commodus (Pflaum 1963, 233-237, with lists of officials). The praefectus Miniciae can be understood to be either the same position as the praefectus frumenti dandi, or as a separate one. In this latter view, propounded by Pflaum, the praefectus Miniciae serves under the curator aquarum et Miniciae, and the two positions must the have been created no later than ca. A. D. 188. In the other view, the curator came into being at the time of the earliest known inscriptions in which the position is named, i. e. under Septimius. This seems correct to me.
48 Grain must now have been made available to the pistores, a possible new responsibility of the praefectus frumenti dandi. On the distribution (erogatio) of grain to the bakers in the fourth century, supervised by a custos ac mensor, Tengström 1974, 71 -72.
49 I am grateful to Dr. Gearoid O’Broin, the present ambassador, for his kindness in permitting members of the American Academy to examine the substructure of the Villa Spada.
50 Τwo fragments of the Watergate are preserved in the garden of the Villa Monami, as was pointed out to me by Prof. Cozza. The width of the secondary channel was recorded as 0,73 m.
51 Part of the core of a concrete wall can still be seen in the furnace room of the Norwegian Institute.
52 Van Buren also published an early imperial funerary inscription of P. Clodius Diaphanes, a glutinarius, found in the construction of the building that houses the Norwegian Institute (Van Buren 1933, 73-76, pl. 3, fig. 3). A tomb might be possible here before the construction of the third-century walls, and one also wonders whether a glue factory would have required water-power.
Auteur
University of Virginia, Charlottesville
Le texte seul est utilisable sous licence Licence OpenEdition Books. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.
Les bois sacrés
Actes du Colloque International (Naples 1989)
Olivier de Cazanove et John Scheid (dir.)
1993
Énergie hydraulique et machines élévatrices d'eau dans l'Antiquité
Jean-Pierre Brun et Jean-Luc Fiches (dir.)
2007
Euboica
L'Eubea e la presenza euboica in Calcidica e in Occidente
Bruno D'Agostino et Michel Bats (dir.)
1998
La vannerie dans l'Antiquité romaine
Les ateliers de vanniers et les vanneries de Pompéi, Herculanum et Oplontis
Magali Cullin-Mingaud
2010
Le ravitaillement en blé de Rome et des centres urbains des début de la République jusqu'au Haut Empire
Centre Jean Bérard (dir.)
1994
Sanctuaires et sources
Les sources documentaires et leurs limites dans la description des lieux de culte
Olivier de Cazanove et John Scheid (dir.)
2003
Héra. Images, espaces, cultes
Actes du Colloque International du Centre de Recherches Archéologiques de l’Université de Lille III et de l’Association P.R.A.C. Lille, 29-30 novembre 1993
Juliette de La Genière (dir.)
1997
Colloque « Velia et les Phocéens en Occident ». La céramique exposée
Ginette Di Vita Évrard (dir.)
1971