The application of animal and water power: From the Mediterranean to Britain
The earliest milling tools were worked by human effort, using specially selected stones or wooden artefacts to pound, crush and grind various raw materials, including cereal grains. The use of animal power for turning millstones marked a significant step forward in the mechanisation of the milling process. It enabled the use of larger millstones with a corresponding increase in efficiency and output.
The earliest and best known form of animal-powered corn mill is the hourglass type, sometimes referred to as the Pompeiian mill.
Usually made of volcanic stone, the Pompeiian mill comprises two main components, the meta and the catillus. The meta is a rounded cone set in a masonry base, the podium, usually with a skirting around it on which the ground product was collected. The catillus takes the form of two joining hollow cones with a narrow waist between them. The lower cone fits over the meta and the upper acts as a hopper into which grain is put. The catillus was turned by a mule or donkey attached to the stone by a timber yoke, the animal walking in a tight circle around it. Some smaller versions were probably turned by slaves.
The Pompeiian mill appears to have been invented in the central Mediterranean area between the 6th and 4th century BC. The earliest reliably dated examples are those from the 4th century BC El Sec wreck off the coast of Majorca, which chemical analysis indicates were made on Sardinia. It has been suggested that this type of mill was originally developed for use in the mining industry, for crushing rock. If so it was soon adapted for grinding grain and by the 3rd century BC was in use in bakeries around the central Mediterranean, replacing the earlier lever-operated Olynthus mills. It also spread further north in Spain and France; Cato (234-149BC), in his book on agriculture written c.160BC, refers to saddle querns, Spanish mills ‒ that is, rotary querns ‒ and also donkey mills for grinding grain (On Agriculture 10.4).
A small number of examples have also been found in Britain. Part of a catillus, now reconstructed in the Museum of London, was found below Princes Street, London, in the 1920s. The stone from which it was made has been analysed as being lava from the Volvic area of central France.
A possible bakery with six roughly circular stone bases which may have been the podia of animal-powered Pompeiian-type mills, was found during excavations at the Roman town of Silchester, Hampshire. The building which housed these features appears to date from an early phase of occupation of the site, but no associated oven was recorded.
Circular buildings, such as those found at Roman villa sites at Langton, Malton, North Yorkshire, first excavated in the 1920s and more recently at Ingleby Barwick, Stockton on Tees, may also have housed animal-powered mills. At both sites other buildings for grain storage and processing, such as threshing floors and grain drying kilns, have also been identified.
The earliest water-powered devices are thought to date from about the middle of the 3rd century BC, at about the same time as the Pompeiian type animal mill was becoming more widespread around the central Mediterranean.
There are basically two forms of waterwheel, defined by the plane in which they turn. The horizontal waterwheel is thought to have originated in Byzantium and the vertical waterwheel in Alexandria, both at about the same time.
The earliest classical references to waterwheels are somewhat later. Strabo, writing in the first century BC, mentions a watermill at the palace of Mithridates at Pontus, in modern day Turkey, which appears to have been in existence in about 71BC although the type of waterwheel cannot be determined (Geography 12.3.30). The often-quoted epigram by Antipater of Thessalonica written c.20BC-AD10, which describes a watermill in poetic terms, has prompted much discussion as to whether the waterwheel was horizontal or vertical. However, his reference to gearing and that the nymphs leap down on the wheel suggest it was vertical. Interestingly, both Antipater and Strabo refer to millstones from Nisyros, a small Greek island to the north-west of Rhodes (Geography 10.5.16).
It seems feasible, however, that the first water-powered grain mill was the result of connecting a water-lifting wheel to animal-powered millstones, using a pair of right-angle gears to change the plane of rotation from vertical to horizontal. This is what is described, in simple terms, in the earliest written description of a water-powered corn mill in On Architecture, ten books on architecture, astrology and engineering compiled by the Augustan engineer Marcus Vitruvius Pollio, which dates from about 25BC.
Vitruvius describes a vertical waterwheel driven by a current of water, so working on the same principle as a water-lifting wheel, which drives a millstone through a pair of gears. Beyond this, the description is somewhat ambiguous, although in his introduction to machines and implements, Vitruvius does state that it was unnecessary for him to describe commonplace machinery, which included mills.
As well as water-driven millstones, Pliny the Elder (AD23-79) implies that water-powered pestles were used in Italy in the 1st century AD for hulling and pounding grain (Natural History 18.23).
Other literary references to milling in classical sources indicate a growing familiarity with such technology, the millstones being powered by human muscles, animals or water. Diocletian’s Edict of AD301 set down the prices of a watermill, horse mill, donkey mill and hand mill and later in the 4th century the agriculturalist Palladius considered that watermills made both animal and human effort unnecessary and suggested using the outflow of water from baths to drive waterwheels (On Agriculture I.41(42)).
Writing in the early 6th century, Cassiodorus noted ‘The vast numbers of the Roman people in old time are evidenced by the extensive Provinces from which their food supply was drawn, as well as by the wide circuit of their walls, the massive structure of their amphitheatre, the marvellous bigness of their public baths, and the enormous multitude of mills, which could only have been made for use, not for ornament’ (Variae 11.39.1).
Evidence of a complex of watermills has been found through archaeological excavations on the Janiculum Hill in Rome, water being supplied from an aqueduct, the Aqua Triana. It has been suggested that the mills may have been built in the late 2nd or early 3rd century AD, as part of the reorganisation of the annona, when handouts of grain to the poorer citizens of Rome were replaced by handouts of baked bread. Evidence for a pair of mills driven by overshot waterwheels has been found in the basement of the Baths of Caracalla, also in Rome, which it is thought were in operation between c.AD275 and AD537.
A number of other vertical-wheeled mill remains have been found by archaeological excavation throughout Europe and from evidence gathered it appears that Roman waterwheels were relatively low powered, perhaps no more than 1½ to 2 horsepower (>1.5kW) on average, capable of driving a single pair of millstones.
One of the earliest mill sites is that at Avenches, in Switzerland, which dates from the 1st century AD and timber components from three Roman waterwheels were found at Hagendorn, also in Switzerland which have now been reconstructed and dated to the late 2nd century.
In France, the building date of the remarkable milling installation at Barbegal, near Arles in Provence has recently been re-assessed, placing it in the 2nd century. Here two rows of eight mills were built down the steep slope of a limestone ridge. Water was brought to the top of the site by aqueducts and fed onto a series of overshot waterwheels, each taking water in turn from the one above. It has been suggested that Barbegal may represent a municipal rather than an imperial monument, built to supply flour to the citizens of Arles.
A similarly impressive group of eight watermills fed from an aqueduct has also been identified at Ephesos, in present day Turkey. Deposits on the walls of the wheel chambers suggest both breast and overshot vertical wheels were used. The mills here were active between the 4th and early 7th centuries AD and a large number of millstone fragments indicate their use for grain milling. The remains of a water-powered stone saw were also found in one of the mill rooms.
In the second volume of History of Corn Milling, Richard Bennett and John Elton offered no evidence to support the theory that the Romans had brought the watermill into Britain, but suggested it was a later introduction which was widely used by the Anglo-Saxons, based on the evidence of Domesday Book. Within ten years of this statement, the archaeologist F. Gerald Simpson excavated a Roman building at Haltwhistle Burn Head, to the south of Hadrian’s Wall in Northumberland, which he interpreted as a watermill dating from the 3rd century AD. Among the finds were the remains of a timber trough in which an undershot waterwheel would have turned and fragments of millstones.
Two further sites along the Wall, at Chesters and Willowford Bridge, were subsequently suggested as the likely locations of watermills, based on evidence in the form of water channels and millstones, dated to the 3rd century AD. However, at neither site can the evidence be considered conclusive.
In the south of the country, two sites of Roman watermills, at Ickham in east Kent and Fullerton, Hampshire, have been excavated. The Ickham site produced evidence for two phases of timber-built mills, the earlier dated to the mid 3rd century and the later sometime before the end of the 4th century. A complex arrangement of timber stakes and piles and millstone fragments were found, with water supplied from the Little River Stour. At Fullerton the mill site was first investigated in the 1960s and re-visited in the first decade of the 21st century. Watercourses and the remains of at least two mills were found, the earlier being dated to the 3rd century. A later mill, located upstream of the first site, was active around 360-380. As at Ickham, a number of millstone fragments were found.
The most recent discovery of a Roman watermill was made in the aftermath of the severe flooding suffered in the area around Cockermouth, Cumbria, in 2009, on the banks of the river Derwent. Stone foundations and fragments of a timber structure that supported a waterwheel were found, broadly dated from coin and pottery evidence to the 1st or 2nd century AD. The mill is thought to have been connected with the nearby Roman settlement at Papcastle.
Although only a handful of Roman watermills have been identified by archaeology, millstones, that is, stones too large to have been worked by hand, have been found on a wide range of Roman sites and thus point to a much wider distribution of mills. In many cases the stones have been found re-used in later, secondary contexts such as floors or oven bases and it is possible that the fragments found on sites where there is little evidence of or potential for water-power use were from animal-powered mills.
Like contemporary quernstones, Roman millstones were made from both locally available stone, such as sandstone conglomerates, and also lava imported from Germany.
The earliest Romano-British millstones had sloping grinding faces, the upper, runner stones being concave and the lower stones convex, sometimes with furrow dressing laid out in a manner similar to that found on more modern stones.
Later Roman millstones were larger in diameter with flatter grinding surfaces, a development also seen in quernstones. In addition, two iron millstone spindles have been found in hoards of Roman ironwork, at Great Chesterford, Essex and the Roman town of Silchester, in north Hampshire.
An iron artefact unearthed close to a large villa site at Mansfield Woodhouse in Nottinghamshire in the 1930s has also recently been identified as a feed cone, a device which controlled the feed of grain into the eye of an upper millstone.
There is only very limited evidence for the use of horizontal waterwheels in the Roman empire. One of the earliest illustrations, a mosaic from the Great Palace of Byzantium, shows a vertical wheel apparently attached to a building, while a second mosaic, and one from Utica in Tunisia, both show buildings with water flowing out from arches at the lower level, which may or may not represent horizontal-wheeled mills.
The earliest known horizontal-wheeled mill is thought to be that at Chemtou, the former Roman site of Simitthus in western Tunisia. Here a masonry structure containing three horizontal waterwheels was built into the remains of a collapsed Roman bridge across the Medjerda river. The mills are thought to have been built soon after the collapse of the bridge in the reign of Constantine, in the late 3rd or early 4th century AD, and to have gone out of use after a working life of 20 to 30 years. No information concerning the ownership of the mills – whether they were built to serve the army or the civilian settlement attached to the nearby quarries, or perhaps the imperial estates in the Mejerda valley – is available, but the multiple installation suggests an intended output in excess of purely local requirements.
The waterwheels at Chemtou have been described as a ‘helix-turbine’ type, using both pressure and the kinetic energy of the water. Some doubts, however, have been cast on the Roman date attributed to the mills as the sophisticated water supply arrangements, with the waterwheels rotating in close-fitting stone-lined shafts, has no parallels until the later Middle Ages. The only closely comparable installation, currently undated, was discovered at Testour (Tichilla), further down the Oued Medjerda towards Tunis, in 1993.
Archaeological finds from Europe indicate the dominance of vertical-wheeled mills and in Britain similar evidence of Roman watermills suggests that they had undershot waterwheels between 2 and 3.6m in diameter. By way of contrast horizontal-wheeled mills became more common in the succeeding Anglo-Saxon period.
Select bibliography
- Bennett, P, Riddler, I and Sparey-Green, C. 2010: The Roman Watermills and Settlement at Ickham, Kent, Archaeology of Canterbury NS, 5, Canterbury Archaeological Trust
- Lewis, M.J.T. 1997: Millstone and Hammer. The Origins of Water Power, University of Hull
- Morgan, M.H. 1960: Vitruvius. The Ten Books on Architecture, New York, Dover
- Peacock, D. 2013: The Stone of Life, Southampton Monographs in Archaeology, New Series 1, Southampton, Highfield Press
- Simpson, F.G. 1976: Watermills and Military Works on Hadrian’s Wall, Kendal, Wilson
- Spain, R. 2008: The Power and Performance of Roman Water-Mills, BAR International Series 1786
- Watts, M. 2011: A newly identified milling artefact from Roman Britain. In Williams, D. and Peacock, D. (eds), Bread for the People: The Archaeology of Mills and Milling, 2011, 93-6
- Wefers, S. 2015: Die Mühlenkaskade von Ephesos, Mainz, RGZM
- Wilson, A. 1995: Water-power in North Africa and the development of the horizontal water-wheel, Journal of Roman Archaeology, 8, 499-510
- Wilson, A. 2002: Machines, Power and the Ancient Economy. Journal of Roman Studies, 92, 1-29