Janiculum Mills Excavations 1998-1999

The mill structure

The west wall of the mill building, with brick relieving arch over the Aqua Traiana.

In the eastern corner of the excavation the external west wall of the mill complex was found, built in brick-faced concrete 0.80 m thick. Up to 19 courses of bricks survived, together with the springing of a relieving arch to carry the wall over the channel of the Aqua Traiana.


During the third century AD, a complex of water-mills was built astride the aqueduct, in brick-faced concrete with a cocciopesto working floor. Two mill races branched obliquely off the Aqua Traiana just inside the west wall of the building, turned to run parallel to the aqueduct for some distance, and then turned to feed back into the aqueduct again. The exact arrangements are uncertain as much of the complex still lies unexcavated under the street and an adjacent house, but from comparing Lancianiís sketch with the excavation plans it seems that there were either 3 or 4 millwheels in the north millrace, and one larger wheel in the south millrace. The wheels were of the undershot type, driven by water striking paddles on the rim as it flowed under them; to provide a sufficiently large surface to strike the wheels had a width of c. 1.65 m. From the position of the axle bearing blocks it can be calculated that the wheels on the north millrace were c. 2.30 m in diameter, and closely spaced, with only 2.60 m between the centres of their axles. This must have reduced efficiency, as each wheel would have operated partly in the turbulence caused by the next. The single wheel on the south millrace must have been larger; although the bearing block is not preserved the gear-pit is a metre higher than that on the north millrace, and the axle can hardly have been less than 1.6 m above the millrace floor, giving a minimum diameter of 3.2 m; perhaps as much as 3.8 m. A larger wheel would have turned more slowly, but could have driven heavier machinery ó either larger millstones, or perhaps two sets of millstones, one at each end of the axle.

Composite plan of the mill complex, using data from the 1998 excavations (to left, with external mill walls in solid red), the 1990/91 excavations (centre, in green), and Lanciani's sketch (roughly scaled redrawing in white). The positions of axles for the mill-wheels on the north mill-race have been reconstructed in yellow, based on the spacing of the two at the east end, whose bearing blocks were discovered by Malcolm Bell's excavations.

On the axle of each waterwheel a cog-wheel would have rotated in a gearpit below the floor of the mill, meshing with a cog-wheel on a vertical spindle, which passed through a hole in the centre of a lower millstone on the floor above the gearpit, and turned an upper millstone supported on an iron cross-bar or rynd.
Gear-pit on the north mill race, and marble bearing block for the end of the axle of one of the mill wheels, excavated by Prof. Malcom Bell in 1990. (Photo: M. Bell.)

Millstone fragments found during the excavation show that the stones were of volcanic basalt. Lancianiís sketch shows 8 millstones lying beside the mill, and two lower stones in the Academyís courtyard, of the same basalt as the fragments from the excavation, may have come from here. These stones are 63 and 71 cm in diameter; Lanciani noted another stone 80 cm in diameter, and fragments found during the excavations belong to stones with diameters of c.75 and 80 cm. Driven by the water-wheels, they would have rotated much faster than animal-powered mills of the sort common at Ostia and Pompeii, and could have ground considerably more flour.

One of two basalt millstones now in the Academyís courtyard, which may have been found during the road construction works in 1886 which first exposed the mill sites.

Water was diverted into the millraces by lowering a sluice gate across the main channel of the aqueduct. A metre or so upstream of the point where the aqueduct entered the mill were found the fittings for an iron grille, socketed into the aqueduct floor and walls and soldered in with lead, to stop debris or foreign objects entering the mill and fouling the wheels.
It is very probable that the mill was connected with the annona or state handout of food to the urban populace. Water to power the mill was taken directly off the Aqua Traiana, and fed back into it, and such use of aqueduct water is inconceivable without state approval and probably state ownership. At some point between the late 2nd century A.D. and the reign of Aurelian, the handout of unground grain in the annona was replaced by a handout of baked loaves, and the state therefore had to assume the functions of milling the grain into flour and baking the bread. Also during this period the offices of the curator aquarum, the head of Romeís aqueduct system, and the procurator Porticus Minuciae, the official in charge of the grain distribution, were merged, and Filippo Coarelli has suggested that this was due to the construction of state water-mills to grind the flour to make the bread required under the new annona arrangements. As the aqueducts were used as the motive power for the mills, it made sense to bring the aqueducts and the annona under the administration of a single official. This is an attractive argument, but must remain hypothetical as we do not know precisely when these changes were made. It is tempting to link the construction of our mills to these changes, but at present the dating evidence allows us only to say that the mills were probably constructed during the third century A.D. This date rests on pottery from the fill of the foundation trench, the latest sherd of which, an African Red Slip sherd of type Hayes 31, was produced during the first half of the third century; the mills cannot therefore be earlier than this.
Concrete floor of opus signinum inside the mill.


The building seems to have gone out of use around A.D. 400. Pottery and lamps found in destruction deposits in the gear-pits date from the late 4th / early 5th century, as does pottery from the layer of silty clay half a metre thick which accumulated over the floor of the mill. A dam was built across the north millrace, perhaps to form the offtake point for a lead pipe found nearby in 1927; clearly, after this the building could no longer have functioned as a mill. Later, the entrances to the north and south millraces were blocked with crude masonry, incorporating re-used material including a millstone fragment. The aqueduct continued to flow, however, and was deliberately made accessible within the former mill by cutting through the opus signinum floor; low walls built of a single course of large re-used tufa blocks flanked the aqueduct at this point. The original filter grille across the aqueduct was replaced by 5 iron bars fixed into a lintel over the aqueduct and socketed into a block on the channel floor below. The most likely explanation for the arrangements in this phase is that they created a place where people could draw water from the aqueduct as though from a well or public fountain. This sort of direct access by the public to the main channel of the aqueduct would have been unthinkable under the High Empire, and seems to suggest that control over access to the water supply network had become considerably less strict by the fifth century A.D.
Late dam across the north millrace, found in 1990. This may have formed the offtake point for a lead pipe found nearby in 1927. (Photo: M. Bell).

Next Previous Home

Go to:

Other Links:

Andrew Wilson

The American Academy in Rome

Dr Andrew Wilson

University Lecturer in Roman Archaeology
Institute of Archaeology
36 Beaumont St
Last updated: 12/03/2001

By: Andrew Wilson ( andrew.wilson@archaeology.ox.ac.uk )