Lava rotary querns of ‘Iron Age type’ in Roman times

In Mayen the production of lava rotary querns of ‘Iron Age type’ continued from the late La Tène period into Julio-Claudian times. The lower quernstone possessed a domed grinding face and the upper stone was double concave in section. While the surfaces of these querns are usually pecked, late examples show a segmented radial grooving on the grinding surfaces. Handle sockets with elbow-shaped (L-shaped) perforation were already an innovation of the late Iron Age. 
Since Augustan times ‘typical Roman’ hand-mills were the main product of the Mayen quarries. They had a meta with a flat conical grinding surface and a catillus with a broad raised rim. The active surfaces were grooved for functional reasons. However, the upper side of the catillus and the sides of upper and lower stones were grooved for decoration, making these rotary querns a characteristic ‘branded’ product. 
Most of the ‘Iron Age type’ quernstones of Early Imperial times are known from the Low Countries where they go under the name of Brillerij-type. A survey of these quernstones reveals several examples found to the southeast of this region. Even after the typical Roman hand mills became the dominant form, some ‘vintage’ Iron Age type querns were still produced for a special clientele. Though, so far, virtually no closely dated specimens are known from contexts after the Batavian revolt.


Introduction
Between Mayen and Mendig, close to the Rhine, Mosel and Lahn, lies one of the largest ancient quarrying areas north of the Alps for the extraction of saddle querns and millstones from "basalt"-like lava (Hörter et al. 1951;Mangartz 2008;Wefers 2012a;Wefers & Gluhak 2010). Already for the saddle querns of the late Hallstatt Period (Joachim 1985: 360) as well as for the Iron Age rotary querns, which were found as far as present-day Hesse (Wefers 2012a: 167), the find distribution suggests that these products were transported on waterways to customers outside the region. Also, in later times fluvial transport remained crucial (Enzmann 2016).
The querns manufactured since the early imperial period in the vicus of Mayen and in villae near the quarries (e.g., Wenzel 2012: 135, fig. 6) usually have a standardized shape and surface modification (Green 2017: 174;Mangartz 2008: 73, 82-83): the runner (catillus) and the lower stone (meta) have a cylindrical basic shape, with a diameter of approximately [35][36][37][38][39][40]. The upper stone has a broad (4-5 cm) and distinctly stepped rim, its underside is adapted to the lower stone with its flat conical grinding surface. The upper stones usually have an 'elbow-shaped' perforation for the handle, "with a broad hole in the outer surface leading through to a slightly smaller hole in the upper surface" (Hill et al. 2011: 351). The underside of the lower stone is hollowed out to reduce the weight. Grooves carved in the grinding surfaces increased the effectiveness of the mills. In addition, grooves were chiselled on top of the runners and on the side surfaces of runners and lower stones for decoration.  10, 56;3. Milne & Wardle 1996: fig. 46, 75;4.) (Photo B. Streubel) The decoration of the quernstones with grooves is considered typical of the products of the Bellerberg volcano at Mayen, which gave them the appearance of a "branded article" (Mangartz 2008: 82). As it is usual with branded products, they were imitated (Rees 2011: 111;Shaffrey 2010: 877;2011: 372). A find in the wreck "De Meern 1" shows that the typical "legionary quern" remained in use at least until the second half of the 2nd century, if not until the first decades of the 3rd century (Jansma et al. 2014: 495;Mangartz 2007;). Blanks for hand mills were rarely exported, which then could get a form deviating from the standard (Baatz 2015: 300, fig. 248;Hörter 1994: 31).
In contrast to the 'typical roman' querns, a quern from a grave in Ettringen near Mayen is of interest, which belongs to the 3rd quarter of the 1st century CE ( Figure 2) (List of findspots with references as supplementary material ("List"), 24). It shows a combination of features known from Iron Age querns with those found at Roman hand mills. Again, both stones roughly have a cylindrical shape. The domed grinding surface of the meta and the biconcave cross-section of the catillus are common features of Iron Age querns, as well as the picked outer surfaces. The sharpening of grinding surfaces with grooves however is considered characteristic of Roman millstones and is rather rare on pre-Roman specimens (Wefers 2012a: 75-78). Obviously, standardization was not taken too seriously here and the production of an old quern type had been continued.

Background
Querns similar to that of Ettringen have been described by Otto Harm Harsema (1967: 141) as "type Brillerij"; he specified the differences to classic Roman handmills which he called "type Westerwijtwerd". He published several examples from the Netherlands (Figure 3: 1-2), including one excavated at Rasquert from a context of the 1st century CE (Harsema 1967: 145, fig. 5:3) and referred to mineralogical analyses, which throughout named Mayen as origin of the raw material (Harsema 1967: 151). Harsema (1967: 150) also noted that the Brillerij type is not present in the article by Fridolin Hörter et al. (1951), who summarized the knowledge at that time of the Mayen quarries and their products. This suggests that quernstones of the Brillerij type were not known in great quantities in their area of origin (see Similar querns with runners of a bi-concave cross section have been called "celtic" querns or "La Tène querns" (Hörter 1994: 22;Mangartz 2012: fig. 6:1). These querns may be regarded as a variant of the Brillerij type querns, though they differ in two details: The hole for the handle in the runners is horizontal, not elbow-shaped, and the underside of the lower stones is not hollowed out. Although querns with runners of bi-concave cross section were particularly common in the Late Iron Age and were produced from various materials (e.g., Wefers 2012a: Pl. 62:362, 70:439, 107:570), there was a great diversity in the design of the lower stones as well as of the edges and eyes of the runners (Wefers 2006: 68, fig. 3-4;Peacock 2013: fig. 4.4;Lepareux-Couturier et al. 2017: fig. 15). To describe the querns which are the subject of this article, the term "type Brillerij" therefore seems most appropriate. Usually also lower stones found separately are assigned to this type (Harsema 1967: 157 no. 8-9, fig. 9:8-9;Hartoch et al. 2015: 278;Van Heeringen 1985: 379 DR 1;). This is also practised here, although lower stones cannot be attributed to the Brillerij type with the same certainty as runners, because in some cases they may have belonged to runners with stepped rim (e.g., Baatz 2015: fig. 248;Engels 1976: Pl. 30:9;Hartoch et al. 2015, Cat. N° 50-51;Stümpel 1991: fig. 5;Wefers 2012a: Pl. 53:278;). Isolated lower stones are therefore mapped with open symbols.
Both, the lower stones as well as the runners of the rotary querns found at Eschweiler-Laurenzberg, have distinctly domed or concave grinding surfaces. In this settlement saddle querns were also still in use (Joachim 1980: 434.438). Bowls with carinated shoulder and profiled or canted lips indicate Germanic inhabitants of the settlement (Lenz 1995: 160), as well as the presence of an unusual cereal spectrum, which "with its primacy of barley has a close connection with the barley-oat-emmer region of the North German lowlands" (Lenz 1995: 161, translated), or with North Holland (Knörzer in Joachim 1980. The husked barley was well storable and may mark the beginning of a 'tradable overproduction' (Kalis & Meurers-Balke 2007: 147), if it was not used as malting barley (Kreuz & Baatz 2003: 22;Shaffrey 2015: 70-72). It is quite possible that the querns were adapted for the processing of this peculiar grain, if not for the crushing of malt.
While the previously mentioned upper stones do not permit any statements to how the handle was attached to them, a fragment of a catillus of the Brillerij type made of lava ( Figure  6:10) from well 12.30 of Oss-Brabantstraat in the Netherlands (List, 55) shows an elbowshaped hole for the handle. The well is dated based on pottery to 185-125 BCE (de Leeuwe 2001: 89) and contained also a runner of a saddle quern made of lava (Knippenberg 2001: 98, fig. 62). Finds from pit 3 in the oppidum "Fosse des Pandours" near Saverne in France (List,65) show that circa 120 BCE two alternatives for the attachment of handles to the upper stones were known (Jodry & Féliu 2009: 70). The pit contained two pairs of quernstones with diameters between 36 and 39 cm, whose material is given as "basalte". One of them corresponds to the type Brillerij ( Figure 5:8). The runner has a horizontal cavity with a round cross-section into which the wooden handle originally was driven. At the salt production site of Bad Nauheim (Germany), "Lattkaute" Brillerij type quernstones of Eifel lava ( Figure 5:9-10) have been found (Wefers 2012a: 111). The site belonged to a spacious area of salines, which was in use from circa 250 BCE until the turn of eras (Hansen 2016: 99). However, few pieces of wood have been found at "Auf der Lattkaute", dating dendrochronologically to Domitianic times (Hansen 2016: 91). In the La Tène period crafts practised in the saltworks area included wood working, ceramic production, forging iron, casting bronze and working amber (Hansen 2016: 94-95); there were also millstone quarries in close proximity (Wefers 2012a: 110.155). The pottery from Bad Nauheim shows Celtic and Germanic influences (Hansen 2016: 99).
In The material of a meta from Creil (Oise, France; Figure 6:9; List, 14) is described as vesicular basalt, with the only known outcrop is at Maar in the district of Daun in the Eifel (now Vulkaneifel district) close to the Rhine (Fémolant 1989: 62). This probably is a misunderstanding: There are several volcanic rock deposits in the environs of Daun (Hörter 1994: 107 no. 45, nos. 49, 53, 59, 88, 93;Mangartz 2008: fig. 46), which were used for quernstones and millstones, but only for local demands, because they are situated far away from the Rhine. If the raw material of this unusual small lower stone originates from the Eifel at all, Mayen would be the more likely source. Creil is situated near the mouth of the river Le Thérain into the Oise, at the crossroads of fluvial and terrestrial routes, near a ford and a harbour (Fémolant 1989: 43-44).
Already during the Gallic War, the Roman army began to use rotary querns of lava from Kottenheim near Mayen. The fragment of a Brillerij type runner was discovered in a military camp near Hermeskeil in Germany (List,32), where quernstones of various shapes and materials document the troop's previous path (Hornung et al. 2015;Hornung 2016: 152, fig. 130:1). To the aftermath of the Gallic War dates a pair of broken Brillerij type quernstones from grave 1/2013 of the cemetery of Bierfeld, "Vor dem Erker" near Nonnweiler, Germany (Gleser & Fritsch 2015: 162, fig. 12.18). (2020)

Quernstones of the Brillerij type from Roman contexts
Several Brillerij type quernstones have been found in Roman military camps of the Augustan period in Germany: in Barkhausen (Figure 7:1;List,6) and Rödgen (Figure 7:2;List,62) fragments of runners; in Hedemünden, a lower stone (Figure 7:4;List,31), which was obviously not used in its original function (Wefers 2012b). Secondarily modified was also a half catillus from the oppidum Martberg near Pommern and Karden (Figure 7:5;List,59), which had been disposed of in a pit of the Oberaden and Haltern horizons (Nickel 2011: 105.255, fig. 45). Its working face has two generations of grooves: very thin and long grooves that look almost fresh, as well as wide and short elongated depressions with rounded edges. On the side face are two horizontal holes for handles. Two holes, placed on the upper surface on both sides of the eye, which was no more centric at the end, belong to a phase of secondary use.
Fragments of Brillerij type querns dating to Augustan times were also found in the context of a supposed watermill at Inden-Altdorf in Germany (List,36). The runner has apparently an elbow-shaped perforation for the handle (Figure 7:3). The pottery from this place, e.g., vessels with cube-shaped profiles of the lips, indicates Germanic operators of the plant (Geilenbrügge & Schürmann 2010: 64, fig. 71).
To the first half of the first century belongs a fragment of a Brillerij type quernstone, which was recovered in the occupation floor of a building in rue Paille-Maille, just outside of Roman Metz, in France (Figure 7:9;List,46). It is made of lava from the Eastern Eifel or the Chaîne des Puys in the Auvergne, and was secondarily converted into a kind of grinding stone or mortar until its base broke through (Asselin et al. 2017). In the settlement area LR42 of the excavation Utrecht, Sportpark Terweide (the Netherlands), both Brillerij type querns ( Figure  7:8-9) and classic Roman hand mills of type Westerwijtwerd have been found. The settlement dates to 20-60 CE, but most of the 'lithic finds' come from a crevasse layer with organic material from 5-9 or 23-127 cal. CE (den Hartog 2009, 22;Mooren 2009: 124, fig. 6:254.255;Taayke et al. 2012: 210 note 369).
In the cemetery "Briedeler Heck" near Briedel,Germany (List,11), spot 4 in tumulus A 9 of the barrow cemetery A, contained the fragment of a Brillerij type upper stone of "basalt"like lava (Figure 8:3), dating to around the middle of the first century or from Claudian to Vespasianic times. In tumulus 3 of the barrow cemetery B, the fragment of a lower stone from the time of Claudius (Figure 8:4) was detected.
Outside the Neronian legionary camp Vetera I near Xanten-Birten (Germany), a fragment of a catillus made of lava (Figure 8:2) was found in the planum of a pit (List,74). It has no broad rim and falls in the variation range of the type Brillerij. Apparently, the pit was associated with the camp of the legio V alaudae and the legio XV Primigenia, which was built around 60 CE and whose suburb was demolished in 69 CE by the Romans due to the threat of the Batavi. After the revolt of the Batavi, the camp was moved about 1 km to Vetera II in 70 CE (Gechter 1987: 620-625).
Outside the Neronian legionary camp Vetera I near Xanten-Birten (Germany), a fragment of a catillus made of lava (Figure 8:2) was found in the planum of a pit (List,74). It has no broad rim and falls in the variation range of the type Brillerij. Apparently, the pit was associated with the camp of the legio V alaudae and the legio XV Primigenia, which was built around 60 CE and whose suburb was demolished in 69 CE by the Romans due to the threat of the Batavi. After the revolt of the Batavi, the camp was moved about 1 km to Vetera II in 70 CE (Gechter 1987: 620-625). (2020)    List, 67) comes from a context intra muros and was found together with a Roman iron trowel and ceramics, including shards of samian ware (Hartoch et al. 2015: 178). From the axial villa of Champion, "Le Rosdia", near Hamois and Emptinne in Belgium fragments of a catillus with biconcave cross section originate (Figure 8:5), as well as a meta with flat radial grooves on the slightly domed grinding surface, both of basalt lava (List, 29). They were found in pit F18.3, which lies south of the main building in an area from which no prehistoric findings are known but numerous pits containing Gallo-Belgic ware are present. Therefore, the querns most likely belong to the Roman phase of the villa.

Journal of Lithic Studies
Two quernstones from northern France are to be listed here under reserve, both from Roman cities. A runner from Bavay (Figure 8:7) is made of "volcanic rock" (Picavet et al. 2011: 221 no. 66, fig. 41:66). Its lateral face tapers slightly upwards, and the grooved grinding surface seems to be rather straight than curved. This runner would only count to type Brillerij if the latter resulted from heavy wear. From a first century context in Meaux (Figure 8:6), a meta is reported, with a high-arched, but only slightly curved grinding surface, whose material is described as "basalte" (List, 45). It differs from all other lower stones discussed in this study: it is thicker near the hole for the spindle than at the margin. Because of its domed grinding surface, a fragment of a meta from Avenches in Switzerland may also be assigned to the "type Brillerij" (List, 3), which is, by far, later than all other pieces. It was excavated from a context of 150 to 200-250 CE. According to the drawing (Figure 8:8), it has as a slight collar around the hole for the spindle, which is an unusual feature of querns of type Brillerij.

Discussion and conclusions
In addition to the 'classic' Brillerij type catilli, which have a concave upper surface and a concave underside, there are also specimens in which the depression of the upside is more straight-lined and those where the brim looks strangely coarse. The runners from "Fosse des Pandours" ( The perforations for the handle sockets show a great diversity, which may be the result of using handles of various materials. Several Brillerij runners have elbow-shaped perforations for the handle (e.g., Figure 3:1-3; 6:10; 7:6, 7; 9:1-4), which would have permitted the attachment of iron clamps as known from Roman hand mills (Figure 1:2), or "a leather strap with wooden peg" as observed at one of the catilli found at Haaksberg veen (van Es & Verlinde 1977: 76) or of a cord or piece of cloth (Alonso et al. 2014: 17, fig. 5). (2020)   Simple horizontal holes for the handle can be found in the runners from "Fosse des Pandours" (Figure 5:8), Martberg (Figure 7:5) and perhaps from the Kastelen plateau ( Figure  8:1). Such a simple form of drilling for the handle can be found even in a catillus with a relatively broad, but slightly rounded rim from a pre-Flavian well in Boxmeer-Sterckwijck in the Netherlands (Melkert 2015: 397,400, fig. 6.27).

Journal of Lithic Studies
The runners of Geldermalsen-Hondsgement (Figure 6:6) and Lierop-Boomen ( Figure  10:1) have an irregularly shaped eye, possibly the result of using wedges to stabilise the rotational movement   fig. 6: a, c). Some upper stones show an eye with polished margin (e.g., List, 5(1), 42), and this may indicate that they were used without a locating tab, like modern querns from Tunisia (Alonso et al. 2014: fig. 5). The same is the case with the eye of the runner from Tongeren which shows small deformations caused by wear due to the spindle (Hartoch et al. 2015: 179). Little used quernstones of Brillerij type show that the working face of the catillus clasped that of the meta, what might have contributed to their centring. The querns of type Brillerij were produced in the area of the Treveri. They were supplied to the territory of the Celtic tribe of the Mediomatrici both in the La Tène period and in Roman times, but they are found also frequently in the area of Germanic tribes ( Figure 11). Several specimens come from the context of Augustan military installations. The distribution range of Brillerij type quernstones resembles that of certain Late Iron Age artefacts (van Heeringen 1985, 383 note 15), such as that of rainbow-cup type staters of the northern group (Schulze-Forster 2010: fig. 5), and that of silver coins of type Scheers 57 with 'dancing man' (Roymans & Aarts 2009: fig. 10). It is also similar to the northwestern circulation area of the glass bracelets of group Haevernick 3b (Deiters 2008: map 1). However, these distribution ranges are not completely congruent. The large number of findspots of Brillerij type quernstones in the Dutch provinces of Groningen and Drenthe has no equivalent in findspots of the aforementioned types of coins and glass armrings.
In the northeastern Netherlands, there are numerous undated querns of Brillerij type, some from Roman contexts, and no securely dated Iron Age examples. There is no evidence that querns of the Brillerij type have been exported to a nearly coinless area during the late La Tène period, but obviously some specimens were traded out of the Roman Empire. In general, the centre of the distribution area shifted northward in Roman times. At the same time the distribution also expanded to the west and to the east. In this process, small rivers could have played a role, and in the west, old routes which were newly restored. For example, Tongeren could have been reached via the Jeker (Hartoch et al. 2017: 219) while the Via Belgica would have provided a shortcut on land (Melkert 2014, 62-63). Salt was negotiated in vast quantities during the Iron Age (Salač 2006: 57). In addition to agricultural products, it may have been of particular importance in exchange for basaltic rock querns (van Heeringen 1985: 383 note 15;Simons 1987: 13) and this could explain the large number of findspots near the coast. Three findspots of Brillerij type quernstones are associated with salt production. In Bad Nauheim, "Lattkaute" (List, 5;Hansen 2016, 94-96) and Monster, "Het Geestje" (List, 47;van Heeringen 1992b, 240), salt production is proven, in Paddepoel (List,58), it is considered possible due to briquetage (Nieuwhof 2008: 289). Already in the Hallstatt period, querns in form of Napoleon's hats arrived at the salines in the Seille Valley, so that even for this early period an exchange of lava products against salt is supposed (Hörter 1914: 288-290;Joachim 1985: 360;Mangartz 2008: 39).
Fragments of lava quernstones as well as lead pans at the Roman salter settlement of Middlewich, Cheshire, are considered to be evidence of military activity (Howard-Davis in Zant 2016, 117-119). In northern Britain, lava quernstones are mainly found on military sites of the first and second centuries (Buckley & Major 1998: 245). In southern Britain, civilian settlements have also been supplied with these goods; notably, those with military presence received particularly many rotary querns made of basalt-like lava (Buckley & Major 2016: 138).
To carve the curved or concave grinding faces of Brillerij type querns was much more difficult and laborious than to carve the flat conical working faces of a Roman hand mill. (2020)  Whether there were any practical reasons for continuing to produce the ancient Celtic form in Roman times is unclear. The use of querns as offerings (van der Sanden 1998;Jodry & Féliu 2009;Trebsche 2013; and in burial rites (Oesterwind 1997: 130-131;Gleser & Fritsch 2015: 168) shows that these machines also had a symbolic meaning. Therefore, it was maybe important for some native groups to use querns in the traditional form. From the time after the second third of the first century there are virtually no Brillerij type quernstones recorded. This corresponds to hand-built pottery in La Tène tradition, which is detectable in the Rhineland until the second third of the first century (Lenz 1995: 159;Lenz & Schuler 1998: 597) and started to disappear rapidly in the last third of the first century (Joachim 1999: 178). The disappearance of the Brillerij type querns and of indigenous wares are indicators of a social change that may have been significantly influenced by the outcome of the revolt of the Batavi.