The grinding stones of the Greek colony of Olbia
(Hyères, Var, France)

Introduction

The grinding tools from Protohistory to Late Antiquity along France’s Mediterranean coastline have not, to date, been the object of a systematic survey. This renders the current study of nearly 200 grinding stones unearthed during different excavations of the Greek colony of Olbia de Provence even more relevant. It is one of the area’s largest assemblages and sheds new light on the evolution of grinding tools and techniques in the Western Mediterranean spanning a period of almost a millennium, a timeframe that is particularly compelling as it saw the transition from mills driven with a to and fro movement (hand querns and hopper rubbers) to circular models driven with a rotary movement.

Olbia de Provence was founded around 350 BC on the Mediterranean coast at the Gulf of Giens about 120 km east of the Rhone Valley. According to ancient sources, it served as a mooring point for vessels sailing between Massalia (Marseille) and the Italian Peninsula (Fig. 1). The site eventually passed under Roman influence in 49 BC after the capture of Marseille by Caesar, and remained under Roman dominance until the 6th century AD.

The site was first associated with the antique city of Olbia in the 19th century due to the discovery of an inscription. Decades later, Jacques Coupry, professor of Greek history at the University of Bordeaux, carried out a series of excavations from 1947 to 1951 and then from 1956 to 1972. His extensive explorations brought to light a square urban area surrounded by a rampart with corner towers and a fortified East gate.

The features of the city, on a sort of island linked to the mainland by a tombolo, comprised two perpendicular roads dividing the urban centre into four quarters. Each of the quarters comprised 10 insulae with dwellings of identical dimensions (34.5 x 11 m). One specific insula (IV) was subsequently totally excavated by Michel Bats between 1982 and 1989, and later between 2002 to 2008. This insula originally consisted of three modules of equal size corresponding to a dwelling and possibly a workshop (blacksmith, …). Its stratigraphic levels reveal a sequence of occupation stretching from 325 BC, the year the city was founded, to the site’s abandonment in AD 525.

A first study of Olbia’s milling tools was carried out two decades ago by Jean-Louis Reille (2001) based on the 110 stones unearthed during the earlier excavations by J. Coupry. The findings of Reille’s study highlight the predominance of certain types of raw materials, notably volcanic imports from the quarry of Embonne at Agde on the Mediterranean coast about 200 km to the west, as well as long-distance exports from the quarry of Embonne to other Greek colonies.

The current state of knowledge on milling tools from France’s
Mediterranean coastline

The history of research on Protohistoric and Antique grinding tools along the Mediterranean coast of France was initiated by the pioneering analyses by Michel Py (1992) and Stéphanie Raux (1999) of the mills from the ancient port of Lattara (Lattes, Montpellier). The Lattara assemblage casts light on the evolution of milling from the 5th century BC to the end of the 2nd century AD. The study by Raux focused on the finds from the nearby indigenous settlement of La Cougourlude (several hundred metres from Lattara) excavated by Isabelle Daveau. The Cougourlude assemblage dates from 700 to 475 BC. Hence the combination of data from the two neighbouring sites offers an overview of grinding stones and milling over a period of nearly 900 years (Daveau & Py 2005).

The information on the “saddle” querns (henceforth hand querns) and hopper rubbers of these two sites was recently reviewed out by the authors of this paper in the framework of a Collective Research Programme focusing on the Lez Valley coordinated by I. Daveau (Jaccottey et al., this volume). The different analyses highlight the introduction of the rectangular hopper rubber, also known as the Olynthus mill, at the moment of the foundation of Lattara. The earliest examples of this mill imported from the Italian Peninsula were unearthed in horizons dating to about 450 BC (Stratigraphical Units, SU 52935 and 27871). This mill type endured several centuries as it appears in later in 1st century BC contexts. The first circular rotary querns, in turn, only emerged at the end of the 5th century BC, preceding the demise of hand querns in the first half of the 4th century BC (Jaccottey et al., this volume).

Another significant survey of ancient mills in the French Mediterranean area was conducted by Chausserie-Laprée (1998) on stones from eight different sites in the Municipality of Martigues (about halfway between Olbia and Lattara) ranging from the Late Bronze Age to the 1st century AD. This survey highlights the critical role of the different types of hand querns over time, with the introduction of rectangular “tables” in about 450 BC that could have served as either the lower stones of hand querns or hopper rubbers. Hopper rubber upper stones, in turn, appear around 350 BC, a century after their introduction at Lattara, and rotary querns only make their appearance towards 200 BC, two centuries years later than at Lattara.

The notion of a chronological gap between the introduction of rotary mills to the west and the east of the Rhone River is not new (Jaccottey et al. 2013). Alonso (1997) places the earliest rotary mills farther to the southwest in Catalonia during the Early Iberian period (550 to 400 BC) at sites such as Turó de Ca N’Olivé, Alorda Park (Barcelona) and Els Vilars (Lleida) (Alonso 1997; 1999; Team Alorda Park 2002; Portillo 2006). The oldest in France in this western zone correspond to two upper stones from between 450 and 400 BC at the oppidum of Pech-Maho in the Municipality of Sigean (Gailledrat-Solier 2004, 416-17). Eight other rotary mills from this area are from levels dating presumably to between 550 and 450 BC (Portillo 2006, 445-56), a timeframe that is nonetheless questionable (Jaccottey et al. 2013; Longepierre 2014). Other rotary mills preceding the 3rd century are known at Ruscino (Longepierre 2014; Gailledrat et al. 2014), La Ramasse (Garcia 1993, 223-25; Reille 1995, 202) and Lattara (cf. above), all to the west of the Rhone. Other sites to the west of the Rhone in the 3rd century BC yielding rotary mills include the Oppidum of Moulin Peyriac-de-Mer (Solier & Fabre 1969), Roque Fabrègues (Larderet 1957: 37), Castels in Nages-et-Solorgues (Py 2007; Reille 2002) and Marduel (Reille 2002). Although there are two cases to the east of the Rhone at the Oppidum of Teste-Nègre at Pennes-Mirabeau from the 3rd century BC, these mills only become common to the east of the Rhone in the 2nd century BC. Examples in this later timeframe comprise those brought to light at Martigues (Chausserie-Laprée 1998), Entremont and Bouc-Bel-Air (Rothé & Tréziny 2005, 785).

The situation of hopper rubbers is quite the contrary. They are in fact more common along the French Mediterranean coastline to the east of the Rhone (J. Chausserie-Laprée presentation at the Saint-Julien-sur-Garonne colloquium, 2009), in particular at Martigues (Chausserie-Laprée 1998), Bouc-Bel-Air (Rothé & Tréziny 2005,785), Pennes-Mirabeau, La Courtine at Ollioule and Evenos (Longepierre 2012: 101-02). Other hopper rubbers, at times older models such as those of Lattara (cf. above), were nonetheless unearthed at sites west of the Rhone such as the Oppidum of Nages (Py 2007; Reille 2002) and Nîmes (Longepierre 2012, 345).

Olbia’s grinding stone corpus

The 192 grinding tools from Olbia were collected, as noted above, during different excavation campaigns and thus under different conditions of observation. To this lot can be added 15 other undetermined fragments. The earlier Coupry excavations yielded a total of 142 (139 milling tools), whereas the excavations of Bats at Insula VI unearthed 65 (53 mills) (Fig. 2).

Type	Coupry excavations	Bats excavations
Possible grinding stone fragments	3	12
Undetermined grinding stones	2	15
Querns and hopper rubbers (upper and lower stones)	64	15
Rotary querns	73	23
Total	142	65

The comparison of the types of mills unearthed during each of the two series of excavations is significant as it reflects a major development of archaeological fieldwork techniques and a new awareness of the value of grinding tools. Although the comparison of the number of hand querns and hopper rubbers to that of rotary querns from the first excavator to the second is roughly equivalent (Coupry 46.7%; Bats 39.5%), the number of smaller fragments and undetermined cases is much greater (41.5% to 3.5%) among Bat’s later excavations of Insula VI. This great difference is explained by a systematic approach adopted during the more recent fieldwork in identifying and sampling exogenous volcanic rocks (even very small fragments).

The work at Insula VI identified 65 cases ranging from the earliest Greek horizons (end of the 4th to the beginning of the 3rd century BC) to Roman levels (end of the 5th and beginning from the 6th century AD) (Fig. 3). The present study has only taken into account those from the pre-Roman period (Phases 7 to 9), as well as the earlier Greek stones reused in levels from about 40 BC (Phase 6/7). The corpus of the current analysis therefore comprises 33 stones, all in fills or reused as construction material for floors or walls.

Chronology	Phase	Number
325 – 200 BC	9	2
250 – 200 BC	9	1
200 – 175 BC	8	16
200 – 125 BC	8	3
125 – 40 BC	7	4
40 BC	6/7	7
40 BC – AD 60	6 A et B	5
AD 30 – 50	6 C	4
AD 60 – 80	5	17
AD 475 – 525	2	2

Methodology

As noted above, Olbia’s assemblage includes different types of mills driven with either a to and fro or rotary movement. The first comprises two types: hand querns and hopper rubbers.

The earlier “classic” hand or “saddle” quern consists of a stationary lower stone and a mobile upper stone. This model comprises a very specific type of lower stone or “table”, widespread throughout southern France during the Iron Age, characterised by a regular quadrangular form with vertical or oblique edges, completely regularised extremities and a flat back. It is labelled Type A2 and A3 in the typology of lower stones by Py (1992, 185) at Lattara and Type 6b in the Martigues classification (Chausserie-Laprée 1998, 221). It is noteworthy that these rectangular tables could nonetheless be paired with either upper stones of hand querns or hopper rubbers characterised by a central slit. The term “table” is therefore reserved for this very specific type of lower stone (Jaccottey et al., this volume) serving alternately for two different types of mills which appears to be characteristic of southern France (and possibly elsewhere) during the Iron Age.

The mobile upper stone of the earlier hand querns (molette in French) driven with a to and fro motion is at times in the literature known as a “rubber” or “mano”. Another type of upper stone, very different from the first, is rectangular and bears a slit through its upper surface, a sort of hopper serving for a continuous feed of grains. It formed part of a mill that Childe (1943) labelled a “hopper rubber”. Moritz (1958, 46), in turn, in his seminal study, preferred to call it “Olynthus mill” due to the detailed research of this mill type carried out by Robinson and Graham (1938) at the site of Olynthus (Greece). This term was retained by Frankel who was the first to offer a typological classification of the mill (Frankel 2003). Yet “Olynthus mill”, in our opinion, is linked to a very specific type restricted to Greece and the eastern Mediterranean. Since a number of variants of the type have in fact been discovered throughout the Mediterranean Basin, even as far as Central Europe (Frankel 2003, 2), we prefer Childe’s more generic term “hopper rubber” (molette à trémie) over that of “Olynthus mill” and likewise discard the term molette à fente used at times in the south of France (Py 1992).

The detailed study of the hopper rubbers of Martigues (Chausserie-Laprée 1998, 227-230) yielded data leading to a hypothetical reconstruction of its system of traction which differs from the traditional view (Fig. 4). The upper stone, equivalent in width to that of the lower table, is driven in the sense of the table’s axis (as opposed to laterally with a long bar or rod). This is evidenced by the slight convexity of the longitudinal profile of its grinding surface. The upper stone’s slit serving to feed the grains is thus perpendicular to the axis of the driving motion allowing a wider extension of the grains. Moreover, the presence of horizontal flutes carved into the edge or extremity of certain upper stones facilitated attaching a wooden frame serving, with other elements of the rig, to drive the stone with a to and fro motion (Chausserie-Laprée 1998, 229, fig. 17). This means of operating the mill, bolstered by experimentation, yielded a patina or polish on the edge of its rim originating from firmly stabilising the stone with one hand while driving the rig with the other.

Rotary querns comprise two circular stones of roughly equal diameter, namely the meta (lower stationary stone) and the catillus (mobile upper stone). The grains are inserted through the eye of the catillus which was driven by a lateral handle.

The differing pairings of lower/upper hand querns, table/hopper rubber, and meta/catillus correspond to different types of mills (Jodry 2011) that can then be further classified into subtypes.

This study advances descriptions of the different features of each stone which form part of a database, including their mobile or stationary elements, their fittings, their discovery context and, when possible, their type. The descriptions of hand querns focus on general shape, longitudinal and transversal sections, grinding surfaces (curvature, use-wear, …), edges, extremities, back, and traces of fashioning (when visible). The descriptions of hopper rubbers in the database resemble those of hand querns except for other specific criteria (rim, hopper, slit, dressing …). Likewise, those of rotary querns, besides their metric values, focus on specific typological elements such as the shape of the eye of the catillus (Robin & Boyer 2011) and meta (Chaussat 2011) and other features linked to the system of traction (Jodry et al. 2011). These descriptions, at times complex, are completed by drawings that follow standardised guidelines designed specifically for the representation of ancient mills (Hamon et al. 2011; Jaccottey & Farget 2011). The drawings offer several perspectives including, at least, a bird’s eye view, a cross section and a longitudinal section or side view. Shades of grey and hatched or dashed lines serve to highlight use-wear and polish, features that help interpret their function.

The petrography of the stones of Olbia from the first series of excavations was determined by Jean-Louis Reille (2001). His seminal study reveals that most (90%) of the tables, hopper rubbers and rotary querns were hewn from volcanic basalts. The remaining can be equally divided into rhyolites (5%) and sandstones (5%). Furthermore, three quarters of the volcanic material is vesicular basalt characteristic of the outcrops of the Lower Hérault Valley. Their provenance was, for the most part, Agde (94%) and to a lesser extent Saint-Thibéry and Bessan (Fig. 1). Besides these relatively distant sources (270 km), certain stones hail from other volcanic rock outcrops near Toulon (30 km) and from the rhyolite outcrops around Esterel. The Olbia assemblage is complemented by local sandstones and conglomerates. Reille’s analyses of the Olbia corpus also highlights a preference for volcanic stones from quarries under Greek influence, notably the distant colony of Embonne (Agde), rather than those from non-Greek spheres, notably near Toulon.

The current study therefore takes into account the initial petrological determinations by Jean-Louis Reille of the finds from the Coupry excavations. Identifications of the stones from the later Bats excavations, in turn, are based on macroscopic observations, when possible, following the criteria advanced by Reille. Most of the later finds (32 of 33) are grey vesicular basalt (with the exception of a local conglomeratic sandstone). This method nonetheless has its limitations as volcanic rocks cannot necessarily be provenanced only based on macroscopic criteria. In any case, the predominance of basalt, especially in the case of the six presumably from Agde, appears to confirm Reille’s observations. Certification of the origin of three volcanic samples, nonetheless, would require further geochemical analyses.

In order to facilitate inserting Olbia’s older finds into the current study, the stones were first presented by chronological phase before advancing, in the form of a conclusion, a broad overview of their evolution. The chronological aspects of this study therefore stem for the most part from data gleaned from the more recent excavations by Bats of Insula VI as the stones from the earlier Coupry excavations are, unfortunately, devoid of stratigraphic context. The findings were then compared with those of the assemblages of Lattara (Py 1992) and Martigues (Chausserie-Laprée 1998), as well as with examples from elsewhere in the south of France published by S. Longepierre (2012).

Phase 9 (325-200 BC):
Olbia’s earliest mills

This earliest phase yielded three basalt fragments. The oldest (76) is from a floor level (SU 6521) dated between 325 and 200 BC. A more recent fragment (85) is from a layer (SU 63170) ranging from 275 to 200 BC. The most recent (87) was re-used in a wall (MR 61488) from between 250 and 200 BC. Two of the three (85 and 87) are grey, highly vesicular volcanic basalts that resemble products scored at the quarries of Agde (Reille 2001). The origin of the third (76), a light grey basalt with small vesicles and some dark green olivine crystals, clearly differs.

The three, due to their high fragmentation (76: 3 x 0.5 cm; 85: 4.5 x 2 cm), defy typological classification. Number 85 reveals no feature (except its rock type) that can link it to a grinding stone while, 76, in turn, bears a vertical edge suggesting it to be a hand quern. But the classification stops there. 87, by contrast, corresponds to the edge of a rotary quern measuring about 30 cm in diameter reused in a wall from 250 to 200 BC. It is, to date, Olbia’s oldest rotary quern. Rotary querns prior to 200 BC to the east of the Rhone River are only known at Pennes-Mirabeau at the Oppidum of Teste-Nègre (Jaccottey et al. 2013). Those of Martigues (Chausserie-Laprée 1998) and other sites of Provence are likewise subsequent to the 2nd century BC. The Olbia case is therefore the only rotary quern in this geographical area that can be securely placed in the second half of the 3rd century BC.

Phase 8 (200-125 BC)

Although this phase offers a greater number of stones, they are highly fragmented and difficult to qualify. Fifteen fragments (69-74, 77-84 and 86) date from circa 200 BC (SU 61239, 61286, 61340 and 61491). Another (68) is slightly more recent (SU 61238), about 200-175 BC. Finally, three large fragments (100, 102 and 126) date from between 200 and 125 BC. They served in the paving of a road along the southern edge of Insula VI (SU 62059).

These stones, as in the case of those of the older levels, are primarily highly vesicular grey basalts occasionally bearing white “inclusions” that appear to come from Agde. Fragment 70 contains black crystals (pyroxenes) suggesting a provenance from Coiron, an outcrop in the south of the Massif Central (Reille 2002). Two other fragments, 68 and 102, are respectively characterised by small or few vesicles. The lot also comprises a rotary quern fragment (86) unearthed by Coupry hewn from a coarse beige conglomerate containing centimetric quartz and feldspar clasts in a sandy matrix of unknown provenance (Reille 2001: 209).

The small basalt fragments from SUs 61286 (70, 72-74) and 61340 (77 and 79-84) offer little typological elements. These two units also contained two fragments of what are either hand querns or hopper rubbers. The first (71, SU 61286) has a flattish grinding surface, whereas the second (78, SU 61340) has a vertical edge. Two other fragments (68-69), also from levels of fill (SU 61238 and 61239) and with flat grinding surfaces, correspond to either tables or hopper rubbers.

The most complete fragments, tables 100 and 102, and an upper stone (126) of a hopper rubber, were recovered in secondary position in the road bordering the Insula. The tables bear vertical to slightly convex edges, respectively 4.5 cm (102) and 7 cm (100) thick, and totally regularised surfaces. Their preserved grinding surfaces are plano-concave longitudinally and plane transversally. A polish or patina is visible along their rims and absent or hardly visible along their central area. Their lower face is flat and in one case (100) reveals scars of the removal of flakes from the edge. A second (102) has a completely regularised surface (Fig. 5). These fragments are comparable with a series of more complete examples reused in later Roman masonry (89, 90, 104, 139 and 142) and especially with the tables from the Coupry excavations (150-151, 153-154, 157-160, 162, 167-168, 172, 178, 180-181, 183, 185-186 and 195). Although no quadrangular table is complete, it is possible to ascertain the width of three, notably 181 (31 cm), 162 (33 cm) and 183 (34.5 cm) that concord with the widths (from 22.5 to 39 cm) of tables from Lattara in contexts ranging from 500 to 150 BC (Py 1992; Jaccottey et al., this volume, p. 87). Type 6b tables of Martigues from between 450 and 190 BC (Chausserie-Laprée 1998: 220-222) are also comparable with those of Olbia as their widths range from 24 to 43 cm. One can be compared to the table of Olbia that measures 33 to 34 cm in width and 53 cm in length. Moreover, one of the complete tables at Lattara (SU 981, 80 3) is 36 cm wide and 53 cm long. The two fragments reveal regularised convex edges and an irregular base bearing, in one case (100), flaking scars originating from the edge and, in the second case (102), a flat and regularised surface. Comparisons in the first case can be drawn with a piece from the Coupry excavations whose lower face’s edge was fashioned by percussion carried out with a wide-bladed metal tool (Fig. 5: 162), a step of the operational sequence yielding a convex and irregular lower surface. Coarse regularisation visible on a series (150-151, 154, 157-160, 181, 185-186) eliminated the scars without modifying the back’s transversal profile. The grinding surfaces of two cases (Fig. 5: 153, 183) are entirely plane with fully regularised backs by either a pick or a pointed metal chisel. Their active surfaces, plano-concave or longitudinally plane and transversely plane, imply that they were paired with lower stones of identical width.

Despite the greater total number of hopper rubbers, Olbia has no complete case. No. 126 is best preserved (three-quarters; 45.2 cm wide; 10 cm thick) (Fig. 6). Three other fragments (196-198), one still in the brickwork (198), also offer reliable data as to their width (33 to 36.3 cm). These were paired with tables of similar width as described above (162, 181 and 183). The length of 196, preserved up to its central slit, is estimated to be 39 cm. It is likewise possible to calculate the length of other finds unearthed by Coupry (101, 161, 163, 166, 170-171) at between 37 and 43 cm. The hopper rubber upper stone of Insula VI is wider than the previous examples and was paired with a table measuring between 37 and 43 cm in width and a length surpassing 60 cm and comparable with the largest cases identified at Martigues (Chausserie-Laprée 1998: 221). A distinction between two rectangular upper stones of hopper rubbers ranging from 30 to 35 cm in width, on the one hand, and square models measuring 40 to 45 cm on the other hand, confirms the observations of Martigues (Chausserie-Laprée 1998, 227). The widths of the hopper rubbers of Lattara, in turn, are between 32 and 36 cm with lengths between 38.5 to 40 cm. The typological features of those from Lattes and Martigues are also identical, namely square or rectangular with rims surrounding an elongated slit stretching across almost all of the stone’s entire width. This case also bears a horizontal flute carved along the centre of its two extremities. Among the Coupry assemblage, flutes of this type were carved in one case on each of its extremities and edges (Fig. 6: 164), whereas a second case bears one on its edge but not along its extremity (Fig. 6: 166).

The assemblage is completed by the half of a catillus of a rotary quern measuring 39 cm diameter (86) (Fig. 7) with a circular eye (Type 1A, Robin & Boyer 2011) and a complex lateral horizontal handle hole (Type 1, Jodry 2011). The extremity of the roughly triangular sectioned lateral horizontal perforation attains the upper face through a vertical shaft. The space served to insert a horizontal wooden fitting (acting as a base for the vertical handle) which was secured by a vertical pin. This type of elbow-shaped handle system is not common among the Protohistoric rotary querns of southeastern France. Among the examples there are three of sandstone from Ruscino, two bearing similar vertical handle holes (Longepierre 2012, 201, 203, 210 and 21), three of either calcareous sandstone or conglomerate from the Oppidum of Chastelard aux Lardiers (Longepierre 2012, 207, 211; Longepierre 2014, 306), one of basalt from the Oppidum of Labaume with a pin hole (Longepierre 2012, 212), one of basalt from Lattes with a square handle perforation joining the upper surface (982-73-3; Py 1992, 205 and fig. 17), one of sandstone from Marduel (Phase IIIa, 300-250 BC) with two circular lateral perforations, the oldest connected to the upper face by a pinhole (Py & Lebeaupin 1989), and finally one of andesite from Aix-en-Provence at the Terrain Coq (between 125 and 75 BC) with a circular blind perforation (Longepierre 2014, 305). It is noteworthy, nonetheless, that none features a lateral perforation in the form of an inverted triangle like that of no. 86.

Phase 7 (125-40 BC)

This last group (11 cases) linked to Olbia’s Greek phase comprises grey vesicular basalts presumably quarried at Agde. The lot consists of a hopper rubber fragment (101) from a fill (SU 6379) (Fig. 8) dated to around 125 BC. The group also includes rotary querns (53, 63-67, and 105) dated to between 125 and 40 BC that were either discarded in a fill (SU 9007) or reused in structures (SU 6365 and 6264). Those from 40 BC were unearthed exclusively in layers of fill (SU 6062, 6385, 50021, 61197, 61337 and 61514). Unlike those of previous phases, these stones are less fragmented and only three, either hand querns or hopper rubbers (43, 88 and 90) and undetermined (75), could be drawn.

The three unillustrated fragments are either elements of hand querns or hopper rubbers with flat grinding surfaces and vertical edges (43, 88 and 90). The nature of a fourth with a vertical edge is unclear (75). The corner of a hopper rubber upper stone (101) with a partly preserved receptacle and central slit is estimated to measure approximately 43 cm in length. It also bears an elongate horizontal flute and traces of a patina or polish on its rim and on its edge stemming from its means of traction (Chausserie-Laprée 1998, 229).

All of the basalt rotary quern metae of Olbia bear blind circular eyes (Fig. 9), a feature common to Martigues and Lattes. The majority of volcanic metae at Olbia and Lattes (Py 1992) reveal a diagonal edge and a flat to slightly convex base.

All of the rotary quern catilli range in diameter from 34 to 39 cm. The group is characterised by a rectilinear diagonal edge (Fig. 10: 64-66) (except 53 with a straight vertical edge), circular eyes (Fig. 10: 65-67) (Type 1A, Robin & Boyer 2011) and especially large quadrangular cuttings on the upper surface stretching radially from the edge towards the eye intended to lodge the base of a handle (Type 6, Jodry et al. 2011).

The study of the rotary models from the Coupry excavations indicates that this is Olbia’s most common type of catillus (Fig. 11). Straight and diagonal edges are common to basalt models. This type of edge is present among the catilli of the southeast of France since 450-200 BC, notably at Pech-Maho and Ruscino for the period 300-200 BC (Longepierre 2014, 295-96). This edge type for the period 200 to 1 BC is characteristic of the Agde productions, as opposed to those from the area of Toulon with more convex edges, and Coirons with inwardly inclined edges (Longepierre 2014, 303-04). The type is known at Lattes (Pys 1992) from 425 BC (Jaccottey et al., this volume) and at Martigues between 200 and 100 BC. At Entremont it is recorded between 190 and 90 BC (Longepierre 2014), at Terrain-Coq (Aix-en-Provence) between 125 and 75 BC and at Pennes-Mirabeau between 100 and 50 BC (Longepierre 2014, 305). Catilli with rectilinear vertical edges in later historical phases are still observed at Olbia in levels from the first half of the 1st century AD (92, MR 61018), as well elsewhere in southeastern France in the AD 1 to 200 sequence (Longepierre 2012, 253-56).

Seven of the rotary querns (07-08, 21, 38, 133, 149 and 190) have circular eyes that at times are flanked by notches or cuttings on the upper surface (133, 149 and 190) (Fig. 11) (Type 3A, Robin & Boyer 2011) intended to lodge a wooden rynd. Traces of oxide suggest iron served to attach the wooden fitting. This element, in tandem with the spindle fixed in the eye of the meta, served to assemble the mill and tenter the catillus during grinding (Fig. 12). Circular eyes are common among the rotary querns of Lattes although this particular system with opposite rynd cuttings on the upper surface is an isolated case (Py 1992, 209; fig. 18: 90). The two rotary querns bearing this type of rynd fitting at Martigues date from 150 to 100 BC (Chausserie-Laprée 1998, 122-123). Fifteen other catilli of this type bear large quadrangular lateral cuttings on their upper surfaces serving to lodge handle bases (07-08, 15, 18, 20-22, 29, 33, 36-37, 110, 149, 190-191). This handle fitting type is common at Lattes since about 250 BC (Py 1992) and at Martigues since between 200 and 100 BC (Chausserie-Laprée 1998, 121-123 and 125). Elsewhere in the French Mediterranean it is common in the 2nd and 1st centuries BC (Longepierre 2012) and persists into the first two centuries AD.

Conclusions

In spite of the relatively limited number of mills stemming from clear stratigraphical contexts, the assemblage from Olbia de Provence sheds new light on the question of mills and milling throughout the Protohistory of southeastern France. Noteworthy is the absence of “traditional” hand quern upper stones (saddle querns) among the finds of Insula VI. Moreover, the sole hand quern fragment among the Coupry finds (108) (Fig. 13) suggests that this type played a marginal role when compared to hopper mills and rotary querns. The practical absence of quern upper stones is particularly noteworthy as between the founding of the colony and the middle of the 3rd century BC (a period known for the introduction of the earliest rotary querns) the only upper stones identified among the assemblage are hopper rubbers. The single hand quern fragment is of grey vesicular basalt with a semi-circular transversal section and fully pecked edges and back. Its grinding surface is longitudinally plane and transversely plano-convex. It corresponds to the “overlapping” type, that is, upper stones whose length surpass the width of their lower couples. This particular type designated Type C at Martigues was unearthed in the settlement’s earliest phase between 375 and 360 BC (Chausserie-Laprée 1998, 222) and is therefore one of the oldest grinding stones at Olbia.

The other grinding stones driven with a to and fro motion, that is, the hopper rubbers, dominate the Olbia corpus until the end of the 2nd century BC (towards 100 BC) (Fig. 14-15). The few isolated cases of hand quern and hopper rubbers from later levels were reused as building material. Hence, although the two oldest rotary querns date to between 250 and 200 BC, the rotary type is most common to the levels extending from the end of the 2nd century BC (starting at 125 BC) to the 1st century BC, and then to the phases subsequent to the Roman conquest.

A compelling question that emerges from this survey is that of the persistence of the use of mills driven with a to and fro motion in spite of the introduction of rotary querns. The great number of hopper rubbers in layers of fill preceding the end of the 2nd century BC implies that they saw a prologued use. This appears to be confirmed by their great quantity (79 compared to 96 rotary querns). The hopper mill, introduced in Greece or Anatolia towards the outset of the 5th century BC, is characteristic of the Greek world (Alonso 2015, 30) and only later spread throughout the Mediterranean Basin. In the south of France it appeared at Lattes towards the middle of the 5th century BC (Jaccottey et al., this volume). The rotary mill, in turn, first appeared in the northeast of the Iberian Peninsula. The oldest examples are from the Early Iberian period (550-400 BC) at settlements such as Turó de Ca n’Olivé, Alorda Park, Els Vilars, Penya del Moro, Tossal del Moro of Pinyeres, Puig de la Nau, Escudilla and Los Villares (Alonso 1995; 1997; 1999; Team Alorda Park 2002; Portillo 2006).

Other types of mills driven with a rotary motion elsewhere in the Mediterranean world, notably the Morgantina type, were recovered in the shipwreck of El Sec off the coast of Majorca among a Phoenician cargo dating to between 375 and 350 BC (Arribas et al. 1987, 7). Mills driven with a rotary motion are also reported in Sicily in the Punic city of Motya at the end of the 5th century BC (Moritz 1958, 55; Wefers 2011) and at the colony of Morgantina in the 4th and 3rd centuries BC (White 1963, 205). Beyond Sicily, rotary mills were not introduced into the Italian Peninsula (Moritz 1958, 62, 121; Amouretti 1968, 144; Py 1992, 195, 197) and North Africa until much later, in the 2nd century BC. The arrival of the mola hispaniensis to Italy (Caton, De Agricultura, 10.4 and 11.4) could have been subsequent to the Roman conquest of the Iberian Peninsula (Py 1992, 197). Moreover, rotary mills do not appear in Greece until in the 1st century BC (Runnels 1988; 1990; Frankel 2003, 18) and in the Middle East are only known in the 1st century AD (Alonso & Frankel 2017, 468). Furthermore, the spread of rotary mills to the Eastern Roman Empire appears to have taken place only in a later phase under the influence of Roman legions (Alonso 2015, 30).

The oldest certified examples of rotary querns in France are two catilli (Fig. 16: 1) dating from between 450 and 400 BC at Pech-Maho in the Municipality of Sigean (Gailledrat-Solier 2004, 416-17). It must be noted that the 550-450 BC range of a second group of eight stones from Pech-Maho (Portillo 2006, 445-56) is questionable (Jaccottey et al. 2013; Longepierre 2014). The oldest rotary querns at Lattes are from a level (SU 27648) from 425-400 BC (Jaccottey et al., this volume), a period largely dominated by hand querns. Rotary mills are present in a limited manner throughout the 4th century BC (Raux 1999, 471, 473) and only begin to play a key role at Lattes at the outset of the 3rd century BC (Fig. 16: 2) (Py 1992, 190). The recent excavation (I. Daveau, Inrap) of the Oppidum of Castelnau-du-Lez near Montpellier yielded a rotary meta (Fig. 16: 3) from a level dated between 475 and 400 BC (currently under study by L. Jaccottey and S. Cousseran-Néré). Other rotary models prior to the 3rd century are recorded at Ruscino (Fig. 16: 4) (Longepierre 2014; Gailledrat et al. 2014) and La Ramasse (Fig. 16: 5) (Garcia 1993, 223-25; Reille 1995, 202).

The systematic inventory of stones from certified Late Iron Age contexts farther north in France, far beyond the study area (notably in south central Auvergne; Mennessier-Jouanet & Deberge, dir. 2017), suggests a northward rotary quern spread during the first half of the 4th century BC throughout what are today the Municipalities of Gerzat and Artonne (Fig. 16: 6-7) into the Plain of Limagne (Mennessier-Jouanet 2013, 382). This geographical distribution further bolsters the notion that rotary querns were limited to sites to the west of the Rhone Valley.

Other settlements in the south of France to the west of the Rhone River from the 3rd century BC also yielded rotary querns, notably the oppida of Moulin Peyriac-de-Mer (Solier & Fabre 1969), Roque at Fabrègues (Larderet 1957, 37), Castels in Nages-et-Solorgues (Py 2007; Reille 2002) and Marduel (Reille 2002). Yet this mill type only became predominant in this area during the second half of the 3rd century BC.

Two rotary querns are nonetheless identified to the east of the Rhone Valley in southern France at the Oppidum of Teste-Nègre Pennes-Mirabeau in levels dated to the 3rd century BC. Furthermore, two others dated to between 250 and 200 BC were unearthed at Olbia de Province in a context where the mill types are dominated by hopper rubbers.

The situation farther north in France throughout the 4th and 3rd centuries BC is difficult to diagnose (Nouvel et al. 2009). Systematic analyses conducted in Auvergne (Fig. 16: 10-12) (Mennessier-Jouanet & Deberge, dir. 2017) and Champagne-Ardenne (Fig. 16: 14-18) (Buchsenchutz et al. 2017; Jaccottey et al. 2017), bolstered by a series of regional studies, have identified concentrations in the Limagne Plain and the Marne Valley that shed light on the question of the adoption of the different types of Protohistoric mills. Yet the vague chronological ranges of most of these northern France settlements, often surpassing a century, do not allow their querns to be referenced for this study. It is nonetheless safe to state that rotary querns from certain settlements of the Massif Central and surrounding areas appeared prior to 250 BC (La Tène B2), whereas more to the north they only emerged in the second half of the 3rd century BC (La Tene B2 and C1). This spread therefore suggests a progressive south-north distribution of rotary querns throughout the whole of Gaul.

The use of rotary mills surged as of the 2nd century BC in the southeast of France as evidenced by the assemblages of Martigues (Chausserie-Laprée 1998), Entremont, and Bouc-Bel-Air (Rothé & Tréziny 2005, 785). Their adoption therefore took place more swiftly among the Celts than among the Graeco-Romans (Buchsenschutz & Pommepuy 2002, 177). Moreover, this study clearly demonstrates that hopper rubbers were only supplanted by rotary models at Olbia at the outset of the 1st century BC, a few decades prior to the Roman conquest of Marseille in 49 BC. This “impermeability” or resistance on the part of the Greek world to adopting an Iberian innovation (the rotary mill) could, as noted by Alonso, be explained from a cultural standpoint, more specifically, along the lines of cultural identity (Alonso 2015, 33). The geographical position of Olbia along the western line of distribution of mills driven with a rotary motion fully illustrates the cultural “rivalries” and technical choices (hopper mill/rotary mill) over time. A preference for hopper rubbers over an extended timeframe – in spite of access to rotary mills – underscores the predominance of the Greek influence in Gaul’s eastern Narbonnaise. The acceptance of this technology elsewhere in Gaul hence paved the way for the introduction of rotary mills several decades earlier.

---

Bibliography

• Alonso, N. (1997): “Origen y expansión del molino rotativo bajo en el Mediterráneo Occidental”, in: Garcia, D. and Meeks (eds), Techniques et économie antiques et médiévales: le temps de l’innovation: colloque international (CNRS.) Aix-en-Provence, 21-23 mai 1996, 15-19.
• Alonso, N. (1999): De la llavor a la farina. Els processos agrícoles protohistòrics a la Catalunya Occidental, Monographies d’Archéologie Méditerranéenne, 4. Lattes: Publications du CNRS.
• Alonso, N. (2015): “Moliendo en ibero, moliendo en griego: aculturación y resistencia tecnológica en el Mediterráneo occidental durante la Edad del Hierro”, in: Dossier Sociedades de frontera. Colonización, aculturación e impacto económico / Border societies. Colonization, acculturation and economic impact, Vegueta, Anuario de la Facultad de Geografía e Historia, 15, 23-36.
• Alonso, N. and Frankel, R. (2017): “A survey of ancient grain milling systems in the Mediterranean”, in: Buchsenschutz, O., Lepareux-Couturier, S. and Fronteau, G. (eds), Les meules du Néolithique à l’époque médiévale: technique, culture, diffusion, Actes du 2^e colloque du Groupe Meule, Reims, du 15 au 17 mai 2014, 461-478.
• Amouretti, M.-C. (1986): Le pain et l’huile dans la Grèce antique, Evolution des techniques agraires d’Hésiode à Théophraste, Centre de Recherche et d’Histoire Ancienne de Besançon, vol. 67, 1986
• Beaune de, S.A. (2000): Pour une archéologie du geste. Broyer, moudre, piler, des premiers chasseurs aux premiers agriculteurs, CNRS éditions.
• Buchsenschutz, O., Jaccottey, L., Jodry, F. and Blanchard, J.L., dir. (2011): Évolution typologique et technique des meules du Néolithique à l’an mille sur le territoire français. Table ronde de Saint-Julien-sur-Garonne (F) du 2 au 4 octobre 2009, Aquitania Suppl. 23.
• Chaussat, A.-G. (2011): “La liaison des meules rotatives manuelles protohistoriques et gallo-romaines : œil et anille”, in: Buchsenschutz et al., dir.,  359-366.
• Chausserie-Laprée, J. (1998): “Les meules des habitats protohistoriques de Martigues, in Entremont et les Salyens”, Documents d’Archéologie Méridionale, 21, 211-235.
• Childe, V. G. (1943): “Rotary Querns on the Continent and in the Mediterranean Basin”, Antiquity, 17, 19-26.
• Daveau, I. and Py, M. (2015): “Grecs et Étrusques à Lattes : nouvelles données à partir des fouilles de la Cougourlude”, in: Roure, R. (ed.), Contacts et acculturations en Méditerranée occidentale : Hommages à Michel Bats. Actes du colloque d’Hyères (15-18 septembre 2011), Paris, Aix-en-Provence: Errance, Centre Camille-Jullian, 31-42.
• Équipe d’Alorda Park (2002): “Les meules rotatives du site ibérique d’Alorda Park (Calafell, Baix Penedés, Tarragone)”, in: Procopiou & Treuil, 2002, 155-175.
• Frankel, R. (1999): Wine and Oil Production in Antiquity in Israel and Other Mediterranean Countries, Sheffield Academic Press, Sheffield.
• Frankel, R. (2003): “The Olynthus Mill, Its Origin, and Diffusion: Typology and Distribution”, American Journal of Archaeology, 107, No. 1 (Jan., 2003), 1-21.
• Gailledrat, E. and Solier, Y. (2004): L’établissement côtier de Pech Maho (Sigean, Aude) aux VI^e-V^e s. av. J.-C. (fouilles 1959-1979). Monographies d’Archéologie Méditerranéenne, 19. Lattes: CNRS.
• Gailledrat, E., Longepierre, S. and Cure, A.-M. (2014): “Les moulins de Gaule méridionale (450-1 av. J.-C.): types, origines et fonctionnement”, in: Alonso, N. (dir.), Reunió Molins i mòlta al Mediterrani occidental durant l’edat del ferro, Revista d’Arqueologia de Ponent, n°24, 331-346.
• Garcia, D. (1993): Entre Ibères et Ligures. Lodévois et moyenne vallée de l’Hérault protohistoriques, Revue Archéologique de Narbonnaise Suppl. 26, CNRS Éditions.
• Georges, V., Jaccottey, L. and Robin B. (2017): “Les molettes à rainure: nouveaux indices sur le perfectionnement du matériel de mouture dans le bassin de la Loire à l’âge du Fer”, in: Buchsenschutz, O., Lepareux-Couturier, S. and Fronteau, G. (eds.), Les meules du Néolithique à l’époque médiévale: technique, culture, diffusion, Actes du 2^e colloque du Groupe Meule, Reims, du 15 au 17 mai 2014, 377-382.
• Hamon, C., Farget, V., Jaccottey, L., Milleville, A. and Monchablon, C. (2011): “Quelques propositions de normes de dessin et d’une grille d’analyse pour l’étude du matériel de mouture va-et-vient du Néolithique à l’âge du Fer”, in: Buchsenschutz et al., dir. 2011, 39-50.
• Jaccottey, L. and Farget, V. (2011): “Les normes de dessin des meules rotatives”, in: Buchsenschutz et al., dir. 2011, 51-68.
• Jaccottey, L., Alonso, N., Deffreessigne, S., Hamon, C., Lepareux-Couturier, S., Brisotto, V., Galland-Cretty, S., Jodry, F., Lagadec, J.-P., Lepaumier, H., Longepierre, S., Milleville, A., Robin and B., Zaour, N. (2013): “Le passage des meules va-et-vient aux meules rotatives en France”, in: Krausz, S., Colin, A., Gruel, K., Ralston, I. and Dechezlepretre, T. (dir.), L’âge du Fer en Europe. Mélanges offerts à Olivier Buchsenschutz, Ausonius Collection Mémoires 23, 405-419.
• Jaccottey L., Fronteau, G., Buchsenschutz, O., Jodry, F. et Saurel, M. avec la collab. de Cousseran-Nere, S., Galland-Crety, S., Garcia, C., Gisclon J.-L., Lepareaux-Couturier, S., Monchablon, C., Picavet, P., Robert, B. and Robin, B. (2017): “Les meules de l’âge du Fer en Champagne-Ardenne”, in: Buchsenschutz O., Lepareaux-Couturier, S. and Fronteau, G. (dir.), Les meules du Néolithique à l’époque médiévale: technique, culture, diffusion. Actes du 2^e colloque du Groupe Meule, Reims, du 15 au 17 mai 2014, 87-95.
• Jodry, F. (2011): “Le Lexique” in: Buchsenschutz et al. (dir.) 2011, 19-33.
• Jodry, F., Jaccottey, L., Robin, B., Picavet, P. and Chaussat, A.-G. (2011): “Les manches et l’entraînement du moulin rotatif manuel”,in: Buchsenschutz et al., dir., 299-317.
• Larderet, P. (1957): “L’oppidum préromain de la Roque, commune de Fabrègues (Hérault)”, Gallia, 1957, 15-1, 1-39.
• Longepierre, S. (2012): Meules, moulins et meulières en Gaule méridionale du II^e s. av. J.-C. au VII^e s. ap. J.-C., Monographies Instrumentum, 41, éditions Monique Mergoil.
• Longepierre, S. (2014): “Les moulins de Gaule méridionale (450-1 av. J.-C.): types, origines et fonctionnement”, in: Alonso, N. (dir.), Reunió Molins i mòlta al Mediterrani occidental durant l’edat del ferro, Revista d’Arqueologia de Ponent, n°24, 289-309.
• Mennessier-Jouanet, C. and Deberge, Y., dir., (2017): Chronologie du mobilier du second âge du Fer en Auvergne, RACF Suppl. 65.
• Mocci, F. and Nin, N. (2006): Aix-en-Provence, Pays d’Aix, Val de Durance, CAG 13-4.
• Moritz, J.-A. (1958): Grain-Mills and Flour in Classical Antiquity. Oxford, Clarendon Press: Oxford University Press.
• Portillo, A. (2006): La mòlta i triturat d’aliments vegetals durant la Protohistòria a la Catalunya Oriental. Doctoral dissertation. Departament de Prehistòria, Història Antiga i Arqueologia, Universitat de Barcelona, Barcelona.
• Py, M. (2007): “La zone H de l’oppidum des Castels (Nages-et-Solorgues, Gard)”, Documents d’Archéologie Méridionale, 29-30, 277-344.
• Py, M. (1992): “Meules d’époque protohistorique et romaine provenant de Lattes”. Lattara, 5, 183-232.
• Py, M., Lebeaupin, D. and the collaboration of Duday, H., Fabre, V. and Tillier A.-M. (1989): “Stratigraphie du Marduel (Saint-Bonnet-du-Gard), IV. Les niveaux des IV^e et III^e s. av. n. è. sur le chantier central”, Document d’Archéologie Méridionale, 12.
• Raux, S. (1999): “Les objets de la vie quotidienne à Lattes au IV^e siècle avant notre ère”, Lattara, 12, 446-514.
• Reille, J.L. (1995): “La diffusion des meules dans la vallée de l’Hérault à l’époque protohistorique et l’identification microtexturale des basaltes”, Documents d’Archéologie Méridionale, 18, 197-205.
• Reille, J.L. (2001): “L’importation des meules domestiques dans la forteresse grecque d’Olbia (Hyères, Var) entre le II^e s. av. n. è. et le Haut Empire”, Documents d’archéologie méridionale, 24, 207-211.
• Reille, J.L. (2002): “Meules à grains de provenance septentrionale (Coirons, Massif Central) sur deux sites protohistoriques du Languedoc oriental”, Documents d’archéologie méridionale, 25, 225-232.
• Ribeaux, P. (1986): Cortaillod-Est, un village du Bronze final 3, L’homme et la pierre, Archéologie Neuchâteloise 3, 140.
• Robin, B., Boyer, F., and the participation of Chaussat A.-G. (2011): “La liaison des meules rotatives manuelles protohistoriques et gallo-romaines : œil et anille”, in: Buchsenschutz et al. (dir.) 2011, 351-358.
• Robinson, D.-M. and Graham, J-W. (1938): Excavations at Olynthus. Vol. 8, The Hellenic House. Baltimore.
• Rothé, M.-P. and Tréziny, H. (2005): Marseille et ses alentours, CAG 13/3.
• Roux, V. (1985): Le matériel de broyage. Etude éthnoarchéologique à Tichitt (R.I. Mauritanie). Edition Recherche sur les Civilisations, mémoire n°58.
• Runnels, C. (1988): “The rotary querns”, Institute of Nautical Archaeology Newsletter, 15, 3, 1988, 30-31.
• Runnels, C. (1990): “Rotary querns in Greece”, Journal of Roman Archaeology, 3, 1990, 147-154.
• Solier, Y. and Fabre, H. (1969): “L’oppidum du Moulin à Peyriac-de-Mer (Aude). Fouille 1966, 1967 et 1968”. Bulletin de la Société d’Études Scientifiques de l’Aude, 69, 69-106.
• Wefers, S. (2011): “Still using your saddle quern? A compilation of the oldest known rotary querns in western Europe, in: Williams, D. and Peacock, D. (dir.), Bread for the people: the archaeology of mills and milling, Proceedings of a Colloquium Held in the British School at Rome 4th-7th November 2009, University of Southampton, Series In Archaeology n°3, Archaeopress, 67-76.
• White, D. (1963): “A survey of millstones from Morgantina”, Amercian Journal of Archaeology, 67, 199-206.