The Technical Development of the Horizontal Water-Wheel in the First Millennium ad: Some Recent Archaeological Insights from Ireland

Abstract

Recent archaeological discoveries of early medieval horizontal water-wheels in Ireland have shed important new light on the origins and technical development of the horizontal water-wheel in Asia and Europe. As will be argued below, based on this evidence, a number of general patterns are beginning to emerge. In the first of these it is clear that regional variations on basic types had already developed across the island of Ireland from at least the seventh century onwards. Indeed, a number of these also appear to have been designed to accommodate seasonal water flows. The recent Irish evid­ence is also compared with what is currently known from medieval written sources from Europe. Together, these demonstrate clear continuity in the use of certain forms of horizontal water-wheel from the early medieval period up to very recent times.

Keywords:

• early medieval Ireland
• horizontal water-wheels
• medieval technolog­y
• water-power in early medieval Europe

Introduction

To date some 153 sites of early medieval water-mills have come to light in Ireland, which have been dated by dendrochronology, radiocarbon dating and by finds assoc­iation to the period c. ad 612–1124 (Figure 1). Of the latter only six are known to have employed undershot vertical water-wheels. However, while early medieval written sources from Ireland suggest that vertical-wheeled mills were much more common than the archaeological evidence might suggest, the horizontal-wheeled water-mill was undoubtedly the most frequently employed variety in early medieval Ireland. Their water-wheels, as will be shown here, were clearly one of the finest achievements of any early medieval woodworking tradition recorded in any region of medieval Europe as a whole. The continued use of many similar water-wheels in Mediterr­anean Europe, until recent times, indeed, high­lights the durability of the original design. To date, some twelve Irish horizontal-wheeled mill sites have produced water-wheel hubs, while a further example has been recovered from a raised rath (an early medieval settlement site) at Deer Park Farms, County Antrim.¹ Of the latter, only five have thus far been dated. These include Nendrum, County Down (c. AD 619), Kilbegl­y, County Roscommon (late seventh-early eighth century) Deer Park Farms (eighth century), Moycraig (ninth century) and Moneyleck, County Antrim (c. AD 822), and Cloontycarthy, County Cork (c. AD 833). Spoon-shaped paddles have, in addition, been found in association with seven of the recorded wheel hubs, while individual dished paddles have been recorded at Banagher, County Offaly, Tahilla, County Kerry, Drumard, County Derry (c. AD 782), Nendrum, County Down (c. AD 788) and Mullantine, County Kildare (c. AD 804).² However, the wheel hub recovered from the earliest mill at Nendrum is completely different from the others, in that it accommodated flat paddles, angled at around 15°. This is, no less, probably the most significant discovery to have been made in association with an early medieval Irish horizontal-wheeled mill, which has far reaching consequences for our understanding of the early development of such mills in post-Roman Europe. There are, however, a number of long-standing questions which the more recent archaeological discoveries from Ireland can shed important new light upon. These include the antiquity of the flat-vaned (or erroneously named ‘Norse’) form and the development of the scooped, or dished, blade designs. Why indeed, was there such a wide diversity of spoon shaped forms already employed in Ireland in the second half of the first millennium ad? Nonetheless, despite the evident widespread differences in the basic forms, within what is, in relative terms, a geographically small island, the overall dimensions of both wheel hubs and individual paddle forms display close similarities. These latter, and other questions, will be addressed below, along with the implications of the Irish evidence for our understanding of similar developments in Britain and continental Europe.

FIGURE 1. Distributions of water-powered mill sites in early medieval Ireland, c. ad 612–1124.

By reference to Figure 2 it will be seen that the individual profiles of the spoon sections of most early medieval Irish water-wheel paddles are similarly shaped — an effect that is hardly fortuitous. It may be observed that it would have been much easier to omit this curvature when manufacturing the paddles and it is most unlikely that this shape was merely decorative. The likelihood is, then, that this feature was the result of observations made over a long period of time on the efficacy of the basic paddle form, and that the results duly noted became incorporated into the water-wheel design. While no two paddles are exactly the same individual forms do, nonetheless, exhibit close similarities in terms of their basic dimensions (see below) and in the profiles of their spoon sections. The scoop profiles of the Cloontycarthy and Mashanaglass paddles are generally similar to each other, as are those from Deer Park Farms and Nendrum 2. However, while the external curvature on the backs of the paddle scoops is generally well executed, that on one of the Kilbegly examples (blade 27) was cut in four facets (Figure 3). Two further sites, Kilbegly (blade 23) and Moycraig differ from the other recorded examples in that the interior profile of scoop is angular rather than curved (Figure 2). The essential spoon shape of the Deer Park Farms paddle is similar to those from Mashanaglass paddles, although the latter differ markedly from the other early Irish forms in that they were provided with ‘felloe’ pieces on their outer extremities, through which interlocking dowels were inserted to tie them together (Figure 4). This latter technique gave the wheel a rigid outer rim similar to that of a cartwheel. Clearly the most critical part of the manufacture of these wheels was the spacing of the paddle mortises given the limited surface area of the hub. The partitions between individual paddle mortises were, in consequence, extremely thin. The Mashanaglass hub in particular, though slightly smaller than the Moycraig and Cloontycarthy examples, positioned some twenty-three paddles, four more than the former. Owing to the slightly irregular grain of the wood, the Mashanaglass hub mortises were off-centre, a defect which was not, however, as marked on the Cloontycarthy example where these mortises were more skilfully executed (Figure 5). At the centre of the paddle tenons, holes were pierced through which were inserted dowels to secure them in the mortises of the hub. These dowels were carefully interlocked to create a pull into the hub when the individual spoons gave resistance to the force of the water striking them.

FIGURE 2. Horizontal water-wheel paddled forms from early medieval Ireland: A Cloontycarthy, Co. Cork (9^th century); B Moycraig, Co. Antrim (9^th/10^th century); C Mashanaglass, Co. Cork; D Tamworth, Staffordshire (9^th century); E Nendrum, Mill 2, Co. Down (8^th century); F Kilbegly, Co. Roscommon (7^th-8^th century); G Kilbegly Co. Roscommon (7^th-8^th century); H Deer Parks Farm, Co. Antrim (8^th century).

FIGURE 3. The paddles from the Kilbegly mill, Co. Roscommon.

FIGURE 4. The Mashanaglass water-wheel (after Fahy, 1956).

FIGURE 5. The Cloontycarthy water-wheel: A the paddles; B the one-piece wheel hub and axle shaft (Photo by Daragh McGrath, courtesy of Cork Public Museum); C reconstruction of the water-wheel.

The Nendrum 2, Moycraig and Mashanaglass paddles also had dowels which articulated each paddle block, a measure which further reduced the pressure on the paddle tenons (Figure 2). In all cases, the face of the paddle block was slightly skew-cut, which ensured that any given paddle was contiguous with one positioned immediately in front of it on the hub, and also helped to absorb the shock created by the impinging water jet. For the paddle block would have been firmly pressed against the inner face of the paddle immediately in front of it, where the forward movement of the paddles created by the force of the water jet striking them, and facilitated by the axial movement of the water-wheel itself about its bearing, would create a torque which the thin paddle tenons could not withstand. The remains of wooden wedges were investigated on the Kilbegly wheel hub, which would have originally been placed between every second mortise on the hub (Figure 6).³ A similar practice was also in evidence on the Mashanaglass example.⁴

FIGURE 6. The Kilbegly wheel hub: A reconstruction of the water-wheel and its tentering mechanism; B the conserved water-wheel hub.

(Photo by John Sunderland of Valerie Keely and Company)

Although all of the recorded early medieval scooped paddle forms exhibit very close dimensional similarities (see below), island-wide there is a wide diversity of individual forms. To a certain extent, as in the case of basic components of early medieval and later vertical undershot water-wheels, this can be explained by the development of regional millwrighting traditions. However, on the basis of the Kilbegly evidence, where both shallow and deep spoon-blade forms were recovered on the same site other factors relating to seasonal changes in the water supply may have been at play. There are two likely explanations for the Kilbegly find. In the first, it may well have been that two different paddle forms were employed on two separate water-wheels, each operating at different periods during the mill’s use. Ethnographic evidence for the use of traditional horizontal water-wheels in Europe indicates that an individual wheel would commonly be manufactured to suit particular, site-specific conditions. Where this did occur, as in the Valais region of Switzerland, this often led to a bewildering multiplicity of water-wheel types being used contemporaneously, even within a relatively small region.⁵ However, at Kilbegly, the change in paddle form could have been related to changes in the available water supply, where one particular design — deep spooned for good conditions and shallow spooned for when the supply was in some way impaired — was favoured over the other. The second hypothesis posits that these paddle forms were interchangeable and used on the same hub as conditions dictated. Again, there is ethnographic evidence for this, where deep spoon forms were used when rainfall created ideal conditions, but replaced by shallower forms during the summer months when supplies of water ran low. In either case the Kilbegly evidence would seem to indicate that some practices still current in the modern period are in fact older than previously thought.⁶

Two Irish sites have also produced firmly dated evidence for the use water-wheels with inclined flat vanes, of the archetypal ‘Norse’ form. These include the two earliest dated horizontal mills, Nendrum Mill 1 (AD 619) and Ballykilleen, County Offaly (AD 633), although each paddle form is essentially different from the other. Nendrum Mill 1 also produced a wheel hub and driveshaft, fashioned in one piece, the hub portion being just over 0.30 m in diameter, with inclined slots for 18–19 flat vanes (Figure 7).⁷ A further refinement in paddle design, however, was found on the Ballykilleen paddle fragments, where an artificial twist was imparted to the back of the paddles by means of a series of carefully executed sloping cuts. This created what was, in effect, a slightly curved paddle blade. Furthermore, the upper edges of the Ballykilleen paddle fragments exhibited notches which must have positioned staying devices, probably thin iron bands, which either formed part of a concentric band, or ‘shroud’ with struts affixed to the lower rim of the wheel hub.⁸ This practice was recorded by Hubert Knox in County Mayo at the turn of the century and is also widely attested in more recent examples from the Mediterranean.⁹

FIGURE 7. The Nendrum Mill 1 wheelhub (after McErlean, 2007), and a reconstruction.

As has been seen, complete wheel hubs have been recovered from the excavations of Nendrum Mill 1, Kilbegly, Deer Park Farms, Cloontycarthy, Moycraig, Mashanaglass, sites, whilst a radius of the Killinchy wheel hub survives in the Ulster Museum.¹⁰ The Moycraig, Deer Parks Farm, Kilbegly and Cloontycarthy hubs and their driveshafts were fashioned in one piece, a considerable achievement if one considers that these were not lathe-turned and that the whole balance of the axle and the hub must have been equalized by carefully shaving away the exterior faces. The published illustration of the Glenkeen, County Tipperary wheel hub, discovered in 1821, though schematic, also suggests that the hub and mill shaft were cut in one piece.¹¹ The entire length of the Moycraig hub and shaft was 1.76m, the shaft itself thinning in substance in the direction of the hub. The Shanacashel driveshaft is said to have been 6ft (c. 1.80m) long, although this, like the Moycraig example, probably does not take into account the height of the footstep bearing and the power-take-off.¹² At the upper extremity of the Moycraig driveshaft, a vertical groove with two perforations at either side of it had been executed; the spindle being received into this slot and secured there with iron bands, these latter being affixed to the shaft by means of rivets. It is unlikely, however, that both the iron plate at the top of the shaft and the reinforcing band at the base of the wheel hub, which are presently on the Moycraig wheel, are original features. The upper section of the Kilbegly driveshaft was recovered some 2 m to the south of the mill undercroft. Although similar to Moycraig in certain respects, it had an open rynd-bar socket (a slot to receive the spindle and rynd assembly) at its upper extremity.

As early as 1956 Edward Fahy had drawn attention to the striking similarities in the basic dimensions of the Moycraig and Mashanaglass water-wheels, both in terms of their individual paddles and their wheel hubs.¹³ What was more remarkable, however, was that these employed different paddle forms, which were each in use at opposite ends of the island. Fahy’s observations on the likely existence of common dimensional formulae for early medieval Irish horizontal water-wheels has been borne out by more recent discoveries of both paddles and wheel hubs from Kilbegly, Cloontycarthy and Deer Parks Farm and Nendrum Mill 2.¹⁴ We may compare, for example, the total length of individual paddles (Nendrum 2, 46.5 cm, Deer Park Farms, 46 cm, Cloontycarthy, 42 cm and Mashanaglass, 43.9 cm), and blade thicknesses (Nendrum 2, 8.1 cm, Deer Park Farms, 8 cm, Cloontycarthy, 7 cm and Mashanaglass, 7.6 cm). Further correlations are evident in regard both to the height of paddle scoops and the length of paddle tenons. At the two County Cork sites, Cloontycarthy and Mashanaglass, the blade heights were exactly 5 cm, while at Moycraig and Deer Park Farms these were both 7 cm high: at Nendrum 2 and Kilbegly these were, respectively, 8.1 cm and 8 cm.¹⁵ Finally, the length of individual paddle tenons (Nendrum 2, 6.6 cm, Cloontycarthy, 6.6 cm and Mashanaglass 6.7 cm), is also remarkable.¹⁶ Close comparison is also possible between individual wheel hubs and the entire water-wheel as assembled. By reference to Table 1 it will be seen that the diameter of the wheel hubs from Nendrum Mill 1, Moycraig, and Mashanaglass (29.2–30.9 cm) are all very similar, even though flat- and spoon-bladed forms were involved. The individual heights of the recorded examples are even closer. Three of these, Nendrum Mill 1, Mashanaglass and Kilbegly are, respectively, 30 cm, 31.8 cm and 29 cm. The other three only differ from a centimetre to a few millimetres in height. Again, despite the basic differences in design, chronology and geography between the assembled water-wheels two (Kilbegly and Cloontycarthy) are 0.93 m and 0.96 cm in total diameter. In a further three examples — Deer Park Farms, Mashanaglass and Moycraig — the overall diameters differ by only 1–2 cm. The dimensional similarities in early med­ieval Irish horizontal water-wheels, therefore, occur over the entire island and within a chronological span of some three centuries.

TABLE 1. EARLY MEDIEVAL IRISH HORIZONTAL WATER-WHEELS

	Dia. hub	Hub height	No. of paddles	Dia. of wheel when assembled
Nendrum 1, County Down (AD 619)	30.9 cm	30 cm	18/19	Est. 71.9 cm
Kilbegly, County Roscommon (7^th to 8^th centuries)	34 cm	29 cm	24	0.93 m
Deer Parks Farms, County Antrim (mid-8^th century)	26 cm	22 cm	18	1.02 m
Cloontycarthy, County Cork (AD 833)	32 cm	23.3 cm	19	96.4 cm
Moycraig, County Antrim (10^th century)	30.4 cm	22.9 cm	19	1.05 m
Mashanaglass, County Cork (early medieval)	29.2 cm	31.8 cm	23	1.03 m

There are perhaps two possible explanations for this phenomenon, and probably inter-related. The discovery of both a wheel hub and a paddle on an eighth-century settlement site at Deer Park Farms could be an indication that these were intended as templates, either for repairs to an existing mill wheel or to facilitate the manufact­ure of a new one.¹⁷ This would enable a non-specialist craftsman to make repairs to the water-wheel after the millwright had moved on, one supposes, to his next commission, a practice recorded up to very recent times in the Vallais region of Switzerland.¹⁸ However, even if this were the case the dimensional similarities across the recorded examples strongly suggest that ‘setting out’ of each example involved widely recognized and shared units of measurement. In Ireland, as in other areas of Europe during the early medieval period, the written sources are generally vague in regard to units shorter than the Roman foot of 0.296 m. In the Irish sources such measurements are often based on the parts of the human hand. These include a graínne (⅓ in.), an ordlach (c. 1 in.), a mér (1 in. or slightly less) and a bas (4 in., or the width of hand).¹⁹ According to one early medieval source, Bretha Étgid (‘Judgements of inadvertence’) there were three basa (i.e. 4-in. measures) in the old Irish traig, or 1 ft. This can only suggest that the Irish foot consisted of 12 in., which Fergus Kelly believes was close to modern equivalent of 30.38 cm.²⁰ However, its subdivision into twelfths would suggest, perhaps, that it was closer to and, indeed, influenced by the Roman foot of 0.296 m. It is by no means unreasonable to infer that the first millwrights to construct water-mills in Ireland, who were working within a Roman woodworking tradition, brought their own system of measurement with them. Roman carpenters used a regula, a folding bronze ruler, 1 ft in length, and etched with divisions into palmae (fourths of c. 7.4 cm) or unciae (twelfths or inches of c. 2.5 cm).²¹ On present evidence, the small tolerances exhibited in the manufact­ure of the early medieval Irish water-wheel assembly, irrespective of paddle design, geography or chronology, would seem to suggest that more or less standard graduated straight edges were employed by contemporary millwrights. Indeed, the use of precise measurements in the act of setting out of the mortises in the wheel hub was itself explicitly mathematical.²² The dias fidchrann is also listed amongst the carpenters’ tools of the mrugher, or ordinary strong farmer, in Crith Gablach (‘Branched purchase’, c. AD 700), appears likely to have been a pair of wooden dividers.²³

Early medieval Irish horizontal water-wheels and their comparanda in medieval and post-medieval Europe

In many ways the discovery of an inclined flat-vaned water-wheel in the Nendrum 1 mill of ad 619, shed new light on the way in which water-mill forms were adapted in early medieval Europe. It has long been known that mills employing both flat- and scooped-vaned water-wheel forms had co-existed in Iberia, the Mediterranean region and in the parts of the Alps up to very recent times. An extraordinary multiplicity of horizontal water-wheel types, indeed, were recorded in the Vallais region of Switzerland alone. In view of this, the Nendrum discovery, while of international import­ance, should have come as no surprise. We have already seen how the scooped forms recovered in Ireland, while sharing many of the same basic dimensions, exhibited slight yet multiple variations in paddle design. However, based on the early date for flat-vaned water-wheel from Nendrum Mill 1, Thomas McErlean concluded that such wheels developed earlier than spoon or dished forms. He also argued that the seventh-century, concave paddle from Ballykilleen, County Offaly should, in light of the Nendrum evidence, be viewed as intermediate in form between the flat vanes of Nendrum Mill 1 and the fully developed dished paddles of Nendrum Mill 2.²⁴ On this basis McErlean concluded ‘that the relationship between the two forms is shown, on this site at least, to be chronological and not geographical’.²⁵ There are, however, some obvious problems with this assertion. As will be seen below, a two-piece spoon-bladed paddle, dated to the final decade of the seventh century, was recovered from Northfleet, Kent, while two examples of paddles fashioned in one piece, with fully developed scoops from the Kilbegly mill could easily date to the final decades of the seventh century (see above). Based on these findings, rather than succeeding each other, both flat and spoon forms (and the alleged ‘intermediate’ type from Ballykilleen), as in more recent times, are more likely to have operated contempor­aneously in early medieval Ireland. Furthermore, the suggestion that the discovery of flat and spoon forms at Nendrum in the seventh and eighth centuries demonstrates ‘increasing technical improvements in horizontal water-wheel design’ is also questionable.²⁶ The absence of dished paddles in either medieval China or, indeed, in most of the contemporary Islamic societies in south-west Asia, cannot be taken as an indication of a lack of ‘technical improvement’. In both cultural zones considerably more advanced uses of horizontal water-wheels as, for example, in the powering of blast furnaces in Han dynasty China are in evidence.²⁷ Moreover, the traditional arubah penstock mills of south-west Asia, which employed flat-vaned wheels very similar to that discovered at Nendrum Mill 1, were capable of a vastly superior output to any of the early medieval examples recorded in Ireland using water-wheels with dished paddles.²⁸

To date, the only close comparanda for the early medieval Irish water-wheel forms are the dished paddles recovered from the Anglo-Saxon sites at Northfleet, Kent and Tamworth, Staffordshire.²⁹ In both instances, however, the reconstructions of the paddle and water-wheel forms based on these are very much at odds with the archaeological and ethnological evidence. The Tamworth water-wheel, of which a single paddle survived (Figure 8), was reconstructed with only twelve paddles, with a total diameter of 1.22 m.³⁰ Not a single piece of archaeological or ethnological data is adduced for this arrangement which in no small part is derived from the clearly erroneous assumption that the paddle block was believed to be means by which the paddle was affixed to the hub.³¹ Based on the Irish corpus and examples recorded in more recent times, this was clearly not the case. The early medieval Irish paddles were provided with tenons, by which means they were inserted into the wheel hub. In point of fact one of the two paddles recovered from the late seventh-/early eighth-century mill at Kilbegly, E3369:27, was very similar to that from Tamworth in that its tenon was not preserved (see above). But of course paddle tenons were found in the mill debris, while the preserved water-wheel hub made it abundantly clear how the paddles were affixed to it. The reconstruction of the Tamworth water-wheel, therefore, has been misconstrued, and based on the archaeological and ethnological evidence (see below) it would have had a minimum of at least eighteen paddles.³²

FIGURE 8. Water-wheel paddles from Anglo-Saxon watermills: A paddle from Northfleet, Kent (after Watts and Hardy 2011); B suggested reconstruction of Northfleet paddles, as a hubbed wheel, in the Irish fashion and as an annular wheel in the southern European tradition; C Reconstruction of Tamworth paddles.

The interpretation and reconstruction of the Northfleet water-wheel is also highly problematical. However, as we shall now see, this is again based on a series of basic misconceptions involving the design of early medieval horizontal water-wheel paddles. Two paddles were recovered from the Northfleet undercroft in fragmentary forms without ‘clear evidence of how they may have been attached to the hub’.³³ The better preserved paddle was 17 cm long (Figure 8) which, based on the Irish corpus of early medieval examples and those recorded in recent Iberian mills, would repres­ent about 25 per cent of a complete dished paddle. However, in the Northfleet report it was suggested that as ‘the inner end of the better of the two paddles does not appear to be broken, its rounded shape suggests it would be difficult to fix it securely, even with wedges, into hub’. To get around this it was proposed ‘that a circular board, or radial arms, could have been fixed to the hub, and the paddles pegged to them’.³⁴ This is, indeed, how the Northfleet wheel is reconstructed, with what is to all intents and purposes a circular board fixed to base of the wheel hub.³⁵ If such a wheel ever existed the board shown in the suggested reconstruction would add considerably to the weight of the wheel assembly, especially when wet. In addit­ion, it would also defeat the purpose of curving the paddles, which were intended to assist the run-off of water, while at the same time creating extra pressure on the lower bearing. Not surprisingly, there is not a single recorded example of water-wheel with either dished or curved blades employing such a device. Furthermore, even if one is prepared to accept that the Northfleet paddle, as found, was its original size, then surely it would have formed part of an annular wheel? In such an arrangement, which is documented in early medieval China and throughout the Mediterranean and south-west Asia (see below), the paddle would have slotted between two shroud sections. The shroud itself would have been attached to the water-wheel’s axle by clasps arms and thus would not have required a wheel hub.

In Figure 8 a much more plausible reconstruction of the Northfleet paddle is suggested, based on the archaeological and ethnographical evidence. As has been, seen there is no certainty regarding its actual length, except that it appeared ‘not to have been broken’. When compared with the early medieval Irish examples, however, it becomes clear that this is fact a fragment of a paddle, which would originally been much longer. The base of the dished portion would have been held in place by a series of additional pegs positioned along the back of the paddle. Furthermore, the curvature of its back suggests that the base of the dished portion at its tenoned extremity would have been narrower, as on the examples from Deerpark Farms and Mashanaglass. This, in turn, suggests that it is likely to have had a thinner paddle block to enable it to form part of a multi-vaned water-wheel, and not one of twelve paddles as the Northfleet report goes to great lengths to attempt to prove. The discovery of a series of extraordinary geometric scribings on one of the flumes recorded at the site, in the form of ‘daisy wheel’ pattern, believed to be template for setting out twelve equidistant points, is the sole reason for this conclusion. These were believed to be an ‘engineering drawing’ for the setting out of a wheel hub, around 0.30 m in diameter, with twelve mortises for paddles. The diameter of the ‘hub’ is similar to most of the Irish examples (see above), but while it is acknowledged that the Irish wheels had a minimum of least eighteen paddles, and that this was unusual in more recent water-wheels employing dished forms, it was nevertheless concluded ‘that the water wheels at Northfleet were not of comparable design or size to those found at Nendrum and other sites in Ireland’.³⁶ Incredible faith, it is clear, has been placed in the belief that the daisy-wheel pattern is indeed a template for a horizontal water-wheel, even though this flies in the face of the archaeological and ethnological evidence for water-wheels with dished paddles. But how can we be sure that this design was ever intended for setting out the paddle mortises of a horizontal water-wheel? There are, in fact, two important reasons why this is unlikely to have been the case. Let us begin by taking the daisy-wheel design at face value. Based on the Irish evidence the first question we should ask is why would a hub diameter of 0.30 m be required for the positioning of twelve paddles? In the Irish corpus of water-wheels, the close similarity in diameter was unquestionably to provide a surface area in which vertical mortises for eighteen to twenty-four paddles could be accommodated. Surely a wheel hub intended for twelve paddles would have been of a smaller diameter? Indeed, in the suggested reconstruction of the setting out of the hub, the mortises are clearly too far apart for the wheel to have functioned properly.³⁷ The second major objection to the daisy-wheel template hypothesis, is the assumption that ‘the setting out and cutting of the water wheel hub would require precise measurement’, to a degree of accuracy that the geometry of the scribings on the flume would seem to imply.³⁸ How would this have been possible? There is no question that the five recorded Irish wheel hubs, some of which have twice the number of mortises than that proposed for the Northfleet water-wheel, exhibit extraordinary technical skill. However, the manner in which the individual mortises were executed was clearly governed by the grain of the wood. Many of the hub mortises not only vary slightly in size but are also off-centre, a defect which the millwright could largely offset in varying the thickness of the paddle blocks. Even if such accuracy had been desired, the millwright would have been unable to execute it, having largely to contend with vagaries in the grain of the wood. By way of summary, then, the proposed geometrica­l daisy-wheel design on the Northfleet flume, as a template for a horizontal water-wheel, can be discounted. Firstly, because the archaeological evidence for early medieval water-wheels suggests that proposed wheel hub diameter could have accommodated twice the number of paddles. And, secondly, because there are five early medieval examples of horizontal water-wheels, which clearly demonstrate that the millwright worked with and around the grain rather than attempting to impose ‘precise measurement’. Finally, in recent examples of horizontal water-wheels with twelve to fourteen paddles, as for example in the ritrecine wheels (see below) recorded in the Italian regions of Emilia, Romagna, Tuscany, Abruzzi, Lazio, Marche and Calabria, the reduced number of paddles has a clear logic. The individual paddle scoops (cucchiai or catini) have almost three times the surface area of those found on the Iberian peninsula or, indeed, in early medieval Ireland. In fact, their internal profiles are almost equal to a quarter sphere.³⁹

At the time of writing there are no recorded examples of medieval horizontal water-wheel paddles from any of continental Europe, Africa or Asia. The romance languages of southern France, Iberia and Italy do, however, provide some important clues as to predominant paddle design. In Spain the vernacular Castilian term for a horizontal water-wheel with dished paddles, rodezno is attested as early as the twelfth-century in Leon, which had a Calle de Rodezneros, or ‘street of the paddle wheel makers’.⁴⁰ In 1397 when three flour mills, situated on right bank of river Segura in Murcia were valued, it was found that twenty-two álabes (curved or dished paddles) were missing, while documents of 1395–96 from the same region indicate that local rodeznos were made of poplar or elm.⁴¹ An earlier description of a cubo (arubah penstock) mill near Jaén, in the Muslim kingdom of Al-Andalus, in a lease of ad 1114, states that ‘the wheels [dawalib] are made of oak, with iron pintles’.⁴² In the late sixteenth century, the author of the pseudo Turriano codex, tells us the rodezno water-wheel has paddles made of holm oak (evergreen oak), and that the wheel itself can have a diameter of 8 palms (1.64 m dia). In addition, ‘These blades are to be very curved in the part where the water is going to fall, and the deeper they are, the better do they perform their duty’.⁴³The water-wheels illustrated by him also have nineteen to twenty-four curved blades, iron shrouds and could be up to c. 2 m in diameter.⁴⁴Rodezno water-wheels, with twelve to nineteen scooped-sectioned paddles are also illustrated in the manuscript of Francisco Lobato, of Medino del Campo, between c. 1547–85.⁴⁵ Francisco also depicts annular wheels with compass arms of the type found more recently in Spain and throughout the Mediterranean, with dished or spoon (cuchares) paddles.⁴⁶

Elsewhere on the Iberian peninsula the horizontal water-wheel with curved or dished paddles is termed rodet/e (Catalan), rodicio (Galician) and rodizio (Port­uguese).⁴⁷ In the Aveyron region in the southern part of the Massif-Central, France, very similar wheels are known by the Occitan word roudet, as in roudet libre (‘free wheel’), or roudet volant (‘flying wheel’).⁴⁸ The term rodet is also used in fourteenth-century sources to describe the water-wheels of Roussillon, and there is even a reference to ferradures de rodetz at Perpignan in 1345–46.⁴⁹ In modern Tuscany the term for a horizontal water-wheel with dished paddles, ritricene,can be traced back as far as the thirteenth century. But while the horizontal mill is archaeologically attested in Italy from at least the tenth century, the clear-cut documentary evidence first occurs a few centuries later.⁵⁰ A Siennese contract of 12 February 1277, lists retecenis, while the terms reticeno and retricini appear in documents relating to Pistoian mills in the 1280s.⁵¹ The thirteenth- and fourteenth-century retricene mills of medieval Pistoia have, indeed, been well documented by the American historian John Muendel. Spanish rodezno is also etymologically related to ritrecine, although the origins of the latter term and its use to denote a horizontal water-wheel are somewhat obscure.⁵² Nonetheless, the fact that it had a medieval Spanish cognate by at least the twelfth century, suggests that its use in southern Europe was well established by this period. Furthermore, in Dante’s Inferno, completed in first two decades of the fourteenth century there is an allusion to a molin terragno or ‘country mill’, a term which in Italian Renaissance sources designates a horizontal-wheeled mill.⁵³

The Siennese engineer, Mariano Jacobus (‘Il Taccola’, 1382– c. 1458) in a series of notebooks compiled between 1419 and 1433 (Liber tertius de ingeneis ac edifitiis non usitatis), makes a clear distinction between what he terms mulini terraguoli / reticino (‘country mills’ or the horizontal-wheeled variety), and molendinum gallicum (‘French mills’, or the vertical-wheeled type). Of the former he tells us:

[they have] a small wheel near the ground. The millstone grinds faster and better when there are 20 blades [on the wheel] than when there are 16, and still better with 24 than 20, because the water that flows and hits the top of the blades turns it faster because of its greater roundness. Also time will then be saved. Because of the contiguity of the blades the water will not strike between them.⁵⁴

Mariano is clearly describing a horizontal water-wheel and the description of the vertical overshot wheel which follows the above text makes this abundantly clear:

According to the second method such a structure is to be built with wheels of large circumferential area, having cups to receive the water that falls from the aqueduct. This mill is called Gallican, as it is made according to French custom. When the wheel has [a diameter of] eighteen feet, it turns rather well and rapidly by itself, but let it be noted that the mill with the open wheel [retecinam] turns more rapidly than the French or Gallican mill.⁵⁵

But while watermills with a ‘small wheel near the ground’ (ad rucham prope terram)are ‘commonly called a country mill’ many, we are also told, ‘call it an open mill’ (Et molti dicono mulini ad reticino). Indeed, in DeIngeneis (fols 83v and 106v), two horizontal water-wheels with scooped paddles are clearly illustrated.⁵⁶ These are the earliest horizontal water-wheels to be depicted in a European source.⁵⁷ A near contemporary of Mariano, Francesco di Giorgio Martini (1439–1531) in his Trattati completed between 1480–90, illustrates up to thirteen water-powered grain mills and one edge runner mill powered by horizontal water-wheels, and also employs the terms mulino terragnolo, ritecine terragnola and ritrecene.⁵⁸ Most of the wheels illustrated by him are annular and have curved vanes, and at least one has compass arms as in more recent examples, while in another the paddles appear to be mortised into a hub.⁵⁹ In certain instances diameters of 1.67 m (compare with rodezno wheel, of 1.62 m diameter, described in pseudo Turriano codex above) and 1.52 m are provided, while the height of the wheel hub is given as 20 cm. At least one of these was made oak.⁶⁰ A very similar form of retricene wheel is also depicted in a sixteenth-century illustration of a mill at Grancia de Cuna, Sienna.⁶¹

Leonardo da Vinci, who is known to have met Francesco, also owned a manuscript of Francesco’s Trattati and to have actually annotated part of it.⁶² In the Codex Atlanticus collection of Leonardo da Vinci’s sketches (c. 1505–10) at fol. 335r, Leonardo illustrates three water-wheels. The first sketch depicts a vertical overshot water-wheel which actuates what appears to be a stone saw via gears and a crank and rod combination.⁶³ The two other sketches which illustrate alternative methods, presumably in connection with the same machine, depict horizontal water-wheels. One of the latter has dished paddles and the upper end of its driveshaft is connected to a crank and a connecting rod, the second horizontal wheel has flat blades perpendicular to the wheel hub; and in both cases water is discharged onto the wheels by means of a vertical tower and spout similar to those of arubah penstocks.⁶⁴ But Leonardo, unlike Mariano and Francesco, found other applications for the horizontal water-wheel; and in the Codex Atlanticus at fol. 2r–b there are a series of machine drawings for a scheme for drawing cannon staves powered a by a horizontal water-wheel. The wheel in question has twenty-two dished paddles with paddle blocks, which are set into a hub. Similar water-wheels are illustrated by Agostino Ramelli (1531–c. 1610) an Italian campaign engineer brought to France by the Duke of Anjou, later Henry III of France, in his book Le diverse et artificiose machine, published at Paris in 1588.⁶⁵ Many of his designs are concerned with military projects but within the vast array of machines are thirteen devices actuated by horizontal water-wheels and a fourteenth by a combination of a horizontal and a vertical water-wheel. One of the horizontal-wheeled grain mills illustrated by Ramelli seems very likely to have been based on the rodezno/ritrecene types at work in his own day.⁶⁶ Using written sources, therefore, it possible to demonstrate continuity of use of the dished paddled forms recorded in early medieval Ireland, throughout Europe, into the sixteenth century.

The study of the origins and development of flat-vaned horizontal water-wheel, outside of Ireland, is beset with same basic difficulties associated with dished forms. Other than the early seventh-century example from Nendrum, there are currently no recorded medieval discoveries from elsewhere. Once again, our only recourse is to examine contemporary written sources in order to gain some idea of their basic morphology. A largely ignored medieval Chinese painting, known as the ‘Shanghai watermill’, provides a detailed, technical rendition of a large flour-milling complex of the 970s, executed in the jiehua tradition of the Northern Song period. This latter shows two water-wheels, one of about 1 m in diameter and one almost twice that size.⁶⁷ Each wheel is very similar to the annular form, illustrated in Nung Shu, a Chinese technical treatise of ad 1313, where it is shown powering a blast furnace via a complex belt drive.⁶⁸ While technically not the oldest illustration of a horizontal water-wheel (see below), the ‘Shanghai watermill’ certainly provides the first accurate depiction of any water-wheel of this type. It is also the oldest rendition of the distinct­ive annular forms still found in China and Tibet today.⁶⁹ The preponderance of the horizontal-wheeled mill throughout western Asia manifests itself at a relatively early stage in a number of medieval Arabic treatises on automata (hiyal) in which are to be found the earliest known illustrations of horizontal water-wheels. The ninth-century treatise Kitab al-hiyal (‘The Book of Ingenious Devices’) by the Banu Musa bin Shakir contains instructions on how to build one hundred hiyal, eighty-five of which can be classified as novelties or trick vessels.⁷⁰ Some five of these designs (models 70, 90, 92–94) employ miniaturized horizontal water-wheels set in motion by pressurized water jets, and at least two commentators have drawn attention to similar vaned wheels in a later work on hiyal by al Jazari (see below).⁷¹

The horizontal water-wheels illustrated by the Banu Musa, however, employ multiple water jets, but the manner in which these were issued onto the wheels is not elaborated upon within the individual sketches, nor in the accompanying textual descriptions. Nonetheless, that these were in fact miniaturized versions of existing prototypes is largely confirmed by the treatise Kitab fi ma ‘rifat al-hiyal al handasiyya (‘The Book of Knowledge of Ingenious Mechanical Devices’) completed in either ad 1204 or 1206 by Ibn al Razzaz al Jazari.⁷² Three types of water-wheel are described by al Jazari: the horizontal-axled scoop-wheel (dawlab dhu kaffat), the horizontal water-wheel (dawlâb dhu rîshât) and the vertical undershot paddle-wheel (dawlab dhu ajniha).⁷³ The scoop-wheels which he employs in no less than seven of his designs are effectively ‘primitive Pelton wheels’ as his most recent translator, Donald Hill, correctly points out.⁷⁴ However, Hill did not see these as capable of performing anything more than the ‘operation of jack figures’.The operation of a chain of pots, however, as illustrated in Chapter 3 of Category V would have been much more demanding than the operation of automata, and is a fair indication that al Jazari considered such wheels capable of heavy workloads.⁷⁵ Moreover, Hill points to the noria as the model upon which al Jazari based his scoop-wheels, when the most obvious derivation of these wheels are those employed in the traditional rice-hulling mills of Persia.⁷⁶ Three devices actuated by horizontal water-wheels are described by al Jazari, two of which are wine dispensers, the third a water-raising engine. In Chapter 1 of Category II he provides instructions on how to make water-wheels for the wine dispensers out of copper discs, in which ‘the vanes are separated so that they form, as it were, a water-wheel’ (surn al-arha’).⁷⁷ The meaning of surn as used by al Jazari is ‘water-wheel’, while the Arabic raha (pl. arha’)means ‘mill’: the expression surn al-arha’, therefore, means ‘mill-wheel’, which al Jazari uses exclusively for the horizontal variety.⁷⁸ A similar water-wheel is incorporated in a scheme for a musical automaton, in a treatise by Apollonius, translated from Greek into Arabic during the medieval period. The horizontal water-wheel in the latter is described by the author as being of the ‘Byzantine type’, and while the treatise appears to be pre-Islamic, claims that it dates to the third century bc have never been properly verified.⁷⁹

Of the devices powered by horizontal water-wheels, described by al Jazari, the wine dispensing goblets are, of course, simple trick vessels, and the work expected of them is rather lightweight. His hydraulic force pump actuated by a horizontal water-wheel, on the other hand, was intended to raise water ‘about twenty cubits’ from a running source.⁸⁰ This device, he tells us, ‘can be made in two versions’:

The first is to make a vaned wheel, which is the machine’s drive, on a vertical axle. The vanes are turned by the water as millstones [arha’] are turned. It [i.e the water-wheel] is on the lower end of the axle, which rotates in a bearing in the usual way. The upper end rotates in a firm ring. At the extremity of this end is a disc with a flat surface, and on the edge of this disc is a vertical dowel. This device is the activator for the device for raising water.⁸¹

The illustration which accompanies this shows a water-wheel with twenty-two flat and apparently inclined vanes set into a vertical axle, whose lower extremity rests in a bearing. The traditional water-wheels of both eastern and western Asia, therefore, seem likely to be descended from medieval prototypes.

What, then, of the horizontal water-wheels types employed up to recent times: to what extent do they illustrate continuity of use from the early medieval period? Miguel Barceló has observed that

while it is certain that contemporaneous milling traditions are not in themselves indicative of their antiquity it is nevertheless true that taken singly they acquire historical meaning as specific technological adaptations to solve local and regional problems. Thus they do indeed tend to exhibit remarkable technological identity and stability.⁸²

Even in early medieval Ireland the existence of regional millwrighting traditions, in which many of the mill’s components — while continuing to assume recognizable forms and close dimensional similarities — can still exhibit slight variations within a few hundred years and over quite short distances. However, while Barceló’s note of caution is timely, from the archaeological record for early water power in Ireland and Europe a clear pattern of technological continuity is beginning to emerge. In one sense, the surviving sketches of early renaissance engineers provide a vital bridge between the medieval and modern periods. The horizontal water-wheels illustrated by Mariano Jacobus, Francesco di Giorgio Martini and Leonardo da Vinci demonstrate a clear continuity of use from the early medieval period. At the same time, these water-wheels share striking similarities with those that were in use in many parts of southern Europe up to very recent times. In some instances, indeed, as in the case of the survival of the Old Irish terms comla (sluice gate) in Connemara, and sgiathain (horizontal water-wheel paddle) in western Scotland, direct continuity of use, in a clearly defined geographical region may well have been involved.⁸³ Nevertheless, while recent examples can bear an uncanny resemblance to early medieval forms, we can never be certain when and how they were adopted in a particular region. The value of horizontal water-wheel types recorded in recent centuries, therefore, lies in their potential as ethno-archaeological entities, providing insights into the manufacture and use of earlier recorded forms.

From the foregoing, therefore, there can be little doubt that most forms of both flat- and spoon-bladed horizontal water-wheels, in use up to recent times, were based on medieval prototypes. Furthermore, the archaeological evidence from early med­ieval Ireland also demonstrates that the millwrighting traditions operating during this period, also embraced several hybridized forms of horizontal water-wheel. These latter may well have already become regional specialities in early medieval Ireland or, alternately, by reference to more recent practices, the type of water-wheel employed in any given mill could also have been chosen to suit certain site specific conditions. Technological continuity, then, in what are basically unchanged or slightly modified forms, attests to the stability of a technological choice or adoption. It is hardly surprising, therefore, given the widespread use of horizontal-wheeled mills in early medieval Ireland that regional water-wheel types (albeit closely based on a basic design) should have developed at an early stage. However, there is a growing body of evidence to suggest that many of these mills, and their plant, were site-specific adaptations, tailored to meet either a particular type of seasonally variable water supply, or a scale of production appropriate to the community who used it.

Notes on contributor

Dr Colin Rynne’s research interests include medieval agriculture, medieval and post-medieval iron working in Ireland and Irish industrial archaeology. His publications include five books, specialising on the industrial history of Cork city and industrial Ireland. He was also an editor of the multi-disciplinary volume titled The Heritage of Ireland (2000), and has published widely on water power in medieval Ireland and Europe and on the post-medieval and industrial archaeology of Ireland. He is a member of the Editorial Board of this Journal.

Notes

1 C. Rynne, ‘Mills and Milling in Early Medieval Ireland’, in The Mill at Kilbegly. An Archaeological Investigation on the Route of the M6 Ballinasloe to Athlone National Road Scheme, ed. by N. Jackman, C. Moore and C. Rynne (NRA, Dublin, 2013), pp. 115–51, at p. 142.

2 Ibid., pp. 142–43.

3 C. Moore and E. O’Carroll, ‘Worked Wood and Woodland Resources’, in The Mill at Kilbegly, p. 52.

4 E. M. Fahy, ‘A Horizontal Mill at Mashanaglass, County Cork’, Journal of the Cork Historical and Archaeological Society, lxi (1956), 13–57, fig. 6.

5 P. L. Pelet, ‘Turbit et turbine. Les roués hydrauliques horizontals du Valais’, Vallesia, 43 (1988), 25–64.

6 Rynne, ‘Mills and Milling’, pp. 143–44.

7 T. McErlean, ‘The Water-Wheels of Mill 1 and Mill 2, the Water-Wheel Assembly, the Tentering System and the Wheelhouse’, in T. McErlean and N. Crothers, Harnessing the Tides: The Early Medieval Tide Mills at Nendrum Monastery, Strangford Lough (London: Her Majesty’s Stationery Office, 2007), pp. 158–59.

8 A. T. Lucas, ‘A Horizontal Mill at Ballykilleen, County Offaly’, Journal of the Royal Society of Antiquaries of Ireland, lxxxv (1955), 100–13.

9 H. T. Knox, ‘Notes on Gig-Mills and Drying Kilns near Ballyhaunis, County Mayo’, Proc. of the Royal Irish Academy, 26C (1906–07), 263–73.

10 McErlean, p. 158; Moore and O’Carroll, p. 52.

11 T. Lalor Cooke, ‘Description of the Baranaan Cuilawn, and some conjectures upon the original use thereof; together with an account of the superstitious purposes to which it was latterly applied. Also a description of the remains of an ancient mill, which were recently discovered near the ruins of Glankeen church, in the County of Tipperary’, Trans. of the Royal Irish Academy, 14 (1825), 31–45, at p. 30, fig. 2.

12 D. Byrne, ‘Ancient Irish Watermills’, Trans. Kilkenny Archaeological Society, i (1849–51), 154–64.

13 Fahy, p. 22.

14 The Mill at Kilbegly, p. 33; C. Earwood, ‘The Wooden Artefacts’, in C. J. Lynn and J. A. McDowell, Deer Park Farms: The Excavation of a Raised Rath in the Glen Valley, Co. Antrim (London: Her Majesty’s Stationery Office, 2011), pp. 386–416, at p. 398; McErlean, pp. 171–72.

15 McErlean, pp. 172–73, table 8; Moore and O’Carroll, p. 52.

16 McErlean, table 8.

17 Earwood, p. 398.

18 Pelet, p. 54.

19 F. Kelly, Early Irish Farming: A Study Based Mainly on the Law Texts of the 7th and 8th Centuries ad (Dublin Institute of Advanced Studies, Dublin, 1997), pp. 560–61.

20 Ibid., p. 565.

21 R. B. Ulrich, Roman Woodworking (New Haven and London, 2007), p. 54.

22 P. Kidson, ‘A Metrological Investigation’, Jnl of the Warburg and Courtauld Institutes, 53 (1990), 71–97, at p. 74.

23 D. Ó Corrain, ‘Some Cruxes in Críth Gablach’, Peritia, 15 (2001), 311–20. This is also mentioned in Cóir anmann (‘Fitness of Names’) a Middle Irish text which also contains material dating from perhaps the eighth and ninth centuries, right up to the twelfth century; ibid., p. 314; S. Arbuthnot, ed. and trans., Cóir Anmann: A Later Medieval Irish Treatise on Personal Names,Irish Texts Society, 59 (London, 2005).

24 McErlean, pp. 168–69. McErlean also points to the similarities between the Ballykilleen example and that from Dalswinton in Dumfriesshire, Scotland, although neither of the latter are dated and could also be post-medieval.

25 Ibid., p. 169.

26 Ibid., p. 169.

27 J. Needham, Science and Civilisation in China, Vol. 4 part II. Mechanical Engineering (Cambridge University Press, 1965), pp. 374–76. The use of water-power in the early Chinese iron industry, in the Nanyang region, is now known to extend back as far as the third century bc; see D. B. Wagner, Iron and Steel in Ancient China (Leiden: Brill, 1993), p. 263, and D. B. Wagner, Science and Civilization in China. Vol. 5. Chemistry and Chemical Technology: Part II Ferrous Metallurgy (Cambridge University Press, 2008), pp. 241–42.

28 M. Harverson, ‘Chutes and Droptowers Water Delivery to Horizontal Mills’, Transactions of the International Molinological Society,9 (2004) 53–65.

29 Two possible paddles have also been recovered from Nailsworth, Gloucestershire, but are undated. Neither of these has any close parallels in either the archaeological record or with more recent examples in Europe or Asia.

30 P. Rahtz and R. Meesom, An Anglo-Saxon Watermill at Tamworth. Excavations in the Boleridge Street Area of Tamworth, Staffordshire in 1971 and 1978, CBA Research Report 83 (London, 1992), 125.

31 Ibid., fig. 72, p. 104. According to Rahtz and Meeson (ibid., p. 125), ‘The paddle spacing at Tamworth, and the diameter of the wheel-assembly, and that of the main shaft, were agreed by [F. W. B.] Charles on the basis of his calculations and his extensive knowledge of both mills and timber technology’. In the course of this study I was unable to find any published work by Mr Charles on early medieval or later horizontal water-wheels.

32 To their credit, Rahtz and Meeson (ibid., p. 125) do acknowledge this as a possibility, on the basis of the Moycraig wheel but do, nevertheless, go along with Charles’ reconstruction.

33 M. Watts and A. Hardy, ‘The Design and Operation of the Northfleet Mill’, in Settling the Ebbsfleet Valley. High Speed 1. Excavations at Springhead and Northfleet, Kent: the Late Iron Age, Roman, Saxon, and Medieval Landscape. Volume 1: The Sites, ed. by P. Andrews, E. Biddulph, A. Hardy, and R. Brown (Oxford, 2011), p. 327.

34 Ibid., p. 327.

35 Ibid., p. 326, fig. 6.12; p. 334, fig. 6.14.

36 Ibid., p. 327.

37 L. Smith, ‘The Northfleet Mill Pentrough Compass Geometry’, in Settling the Ebbsfleet Valley, fig. 6.24, p. 343.

38 Ibid., p. 342.

39 E. Caruso, Mulini e mugnai in Romagna e nell’Italia de medievo (n.p, 2004), pp. 160–61; G. Rochetta, D’acqua e di farine. Mulini idraulica nel castiglionese (n.p., 2009), pp. 51–53; I. Casoli, ‘La derivazione’, in I mulini ad adqua della Valle dell’Enza. Economia tecnica lessico, ed. by F. Foresti, W. Baricchi, and M. Tozzi Fontana (Casalecchio di Reno, 1984), pp. 72–73; W. Baricchi, ‘Sistema tecnico’, in I mulini ad adqua della Valle dell’Enza, pp. 75–79.

40 T. F. Glick, Islamic and Christian Spain in the Early Middle Ages (Princeton, 1979), p. 233.

41 L. Martínez Carillo and M. Martínez Martínez, Origenes y expansion de los molinos hidráulicos en la cuidad y huerta de Murcia (siglos XII–XV) (Ajuntamiento de Murcia, 2000), p. 99.

42 V. Lagardère, ‘Moulins d’Occident musulman au moyen age (IX au XV siècles), Al-Andalus’, Al Qantara, 12 (1991), 109; S. Selma, ‘De la construcció islàmica al casalici modern: l’evolució molí hidràulic Valencia’, in Els molins hidràulic valencians. Tecnologia, història i context social, ed. by T. F. Glick, E. Guinot, and L. P. Martínez (Valencia, 2000), pp. 103–63; M. Harverson, Mills of the Muslim World (London: SPAB, 2000), p. 13.

43 A. Keller, trans., The Twenty-One Books of Engineering and Machines of Juanelo Turriano,2 vols (Madrid: Fundación Juanel­o Turriano, 1998), vol. 2, p. 322.

44 Ibid., pp. 323, 329, 249.

45 J. A. García-Diego and N. García Tapia, Vida y técnica en el renacimiento. Manuscritó que escribió, en el siglo XVI, Francisco Lobato, vecino de Medina del Campo (Valladolid, 1987), pp. 16–17; N. García Tapia, ‘Los molinos en el manuscrito de Francisco Lobato (siglo XVI)’, in Losmolinos: cultura y tecnología, ed. by L. V. Elias (Madrid, 1988), pp. 151–72.

46 García-Diego and García Tapia, pp. 63–64, fol. 16.

47 F. Sendra Bañuls, Molins d’aigua a la Vall d’ Albaida (Ontinyent, 1998), p. 38; R. Córdoba de la Llave, ‘Some Reflections on the Use of Waterpower in Al-Andalus’, in Economia e energia Sec. XII–XVII, ed. by S. Cavaiciocchi (Florence, 2003), pp. 936–37; E. Viega de Oliviera, F. Galhano, and B. Pereira, Tecnologia tradicional portuguesa: sistemas de maogem, 2 (Lisbon, 1983), p. 112ff.

48 J. P. Azèma, ‘Horizontal Water-Wheels in Aveyron in the Southern Part of the Massif-Central, France’, Transactions of the International Molinological Society, 7 (1994), 85–92, at p. 8.

49 S. Caucanas, Moulin et irrigation en Roussillon du IX^e au XV^e siècle (Paris, 2002), pp. 147, 157, n. 32. However, despite the similarity in the terms for horizontal water-wheel in Catalan and the immediate proximity of Roussillon (where Catalan is still widely spoken) to northern Spain, Sylvie Caucanus maintains that the most widely used variety of water-wheel in the area during the medieval period is not known (ibid., p. 155).

50 P. Brandizzi Vitucci, ‘Circo massimo: contributo di scavo per la topographia medievale’, ArcheologiaLaziale, 9 (1988), 406–16; P. Brandizzi Vitucci, ‘L’emiciclo de Circo Massimo nell’utilizzazione post classica’, Mélanges de l’Ecole Francaise de Rome, Moyen Age, 103 (1991), 7–40.

51 J. Muendel, ‘The Horizontal Mills of Medieval Pistoia’, Technology and Culture, 15 (1974), 200.

52 Ibid., p. 200.

53 Muendel, pp. 194–225.

54 Caruso, p. 165.

55 Muendel, p. 200.

56 Ibid., p. 200.

57 F. D. Prager and G. Scaglia, Mariano Taccola and his book De Ingeneis (Cambridge, Mass.: MIT Press, 1972), p. 107. Ibid., p. 107, 57. G. Scaglia, F. D. Prager, and U. Montag, Mariamo Taccola De Ingeneis. Liber primus leonis, liber secundis draconis, Addenda Books I and II, on Engines and Machines (The Notebook) (Wiesbaden, 1984), pp. 166, 214. It has been suggested by Anthony Bryer and by Philip Rahtz (1981) that a fifth- or sixth-century mosaic in the Great Palace of Constantinople depicts the tail-race outlets of two horizontal-wheeled mills, see A. Bryer, The Empire of Trebizond and the Pontus (London, 1980), p. 410; P. Rahtz, ‘Medieval Milling’, in Medieval Industry, ed. by D. W. Crossley, CBA Research Report 40 (London, 1981), 1–15. Based on what is currently known about medieval and recent water-mills of this type, this seems highly unlikely. In either chute- or arubah-fed mills the tail-race arches would have been much smaller than those depicted, whilst the buildings behind these have little in common with those of horizontal water-mills. There is also an illustration of a late fifteenth-century horizontal mill in Bohemia in the Anonymous of the Hussite Wars Ms (Codex Latinus Monacensis 197, fol. 18v). This has been dated by Hall to 1472–75, and it depicts a shrouded horizontal water-wheel with compass arms and what appear to be flat vanes, see B. S. Hall, The Technological Illustrations of the So-Called ‘Anonymous of the Hussite Wars’: Codex Latinus Monacensis 197. Pt. 1 (Wiesbaden, 1979).

58 C. Maltese, ed., Francesco di Giorgio Martini: Trattati di Architettura Ingegneria e Arte Militare, 2 vols (Milan, 1967), vol. i, 143; vol. ii, fol. 34, Tav 63; fol. 36v, Tav 70; vol. 2, 501, fol. 95v, Tav 326.

59 Ibid., vol. ii, fol. 95v, Tav 326; vol. i, fol. 34, Tav 64.

60 Ibid., vol. ii, 501, fol. 95v, Tav 326; vol. i, 143, fol. 34 Tav 63.

61 Caruso, p. 146, Ms 172, fol. 110v, Fondo Ospedale di Santa Maria della Scale. Archivo di Stao di Sienna.

62 L. Reti, ‘Francesco di Giorgio Martini’s Treatise on Engineering and its Plagiarists’, Technology and Culture, 4.3 (1963), 287–98, at p. 287.

63 S. B. Ek, Väder kvarnar och vatten möllor,Nordiska Museets Handlingar, 58 (Stockholm, 1962), 56; L. Reti, ‘A Postscript to the Filarete Discussion on Horizontal Water-Wheels and Smelter Blowers in the Writings of Leonardo da Vinci and Juanelo Turriano’, Technology and Culture, 6 (1965), 428–41, at p. 437.

64 Ibid., pp. 437–38.

65 M. T. Gnudi and E. S. Ferguson, The Various and Ingenious Machines of Agostino Ramelli. A Classic Sixteenth-Century Illustrated Treatise on Technology (New York, 1976), pp. 13–19.

66 Ibid., pl. 114, p. 293.

67 H. Liu, ‘“The Water-mill” and Northern Song Imperial Patronage of Art, Commerce and Science’, The Art Bulletin,84.4 (2002), 566–95.

68 Needham, pp. 374–76.

69 E. Wehrli and H. Brütsch, Mühlen in Tibet (Opuscula Tibetana. Arbeiten aus dem Tibet-Institut Rikon-Zürich, 1993), pp. 28–30.

70 Banu Musa bin Shakir, The Book of Ingenious Devices (Kitāb al-hiyal), trans. by D. R. Hill (Dordrecht: Reidel, 1979), pp. 19–24.

71 Ibn al Razzaz al Jazari, The Book of Know­ledge of Ingenious Mechanical Devices (Kitab fi ma ‘rifat al-hihal al handasiyya),trans. by D. R. Hill (Dordrecht/Boston, 1974), p. 223.

72 Ibid., p. 223.

73 Ibid., p. 274.

74 Ibid., p. 274.

75 Ibid., p. 182.

76 H. E. Wulff, The Traditional Crafts of Persia (London, 1966), p. 291. In the traditional rice-hulling mills of the Alburz mountains of Iran, described by Hans Wulff, a hollowed-out tree trunk was used to direct a jet of water onto what he characterized as ‘a kind of Pelton wheel’. The latter consisted of some 16 scooped vanes, vertically mounted on a horizontal axle, a mode of construction very similar to the dawlab dhu kaffat illustrated by al Jazari.

77 Ibid., pp. 95–96.

78 Ibid., p. 275.

79 Ibid., p. 275.

80 Ibid., p. 186.

81 Ibid., p. 186.

82 M. Barceló, ‘The Missing Watermill: A Question of Technological Diffusion in the High Middle Ages’, in The Making of Feudal Agricultures?, ed. byM. Barceló and F. Sigaut (Leiden: Brill, 2004), p. 277.

83 C. Rynne, ‘Technological Continuity, Technological “Survival”: The Use of Horizontal Mills in Western Ireland, c. 1632–1940’, Industrial Archaeological Review, 33.2 (2011), 96–105.