A first approach to women, tools and operational sequences in traditional manual cereal grinding

The milling of cereals, other plants and materials is an activity that is essential to the subsistence of human societies. It is a task carried out for the most part by women. Querns are the most representative archaeological artefacts of this activity. However, querns are only a part of more complex operational sequences that stretch beyond these mechanisms and involve other tools such as mortars and sieves that yield a much broader range of cereal products than those commonly identified. This paper reviews a selection of publications from recent decades address this subject from the point of view of ethnography, ethnoarchaeology and archaeobotany, and takes into account a number of features serving to process the more common types of cereals.

The traditional process of manual cereal grinding, as well as aspects regarding the manufacture, use 21 and socio-economics of querns, has been the object of studies by different authors in many regions of the current article is to present a preliminary overview of the results of previous research with a 25 special focus on the cereal grinding process and women's role in milling.

26
Ethnoarchaeological and ethnohistorical studies of traditional techniques and practices are highly 27 valuable sources in prehistoric and protohistoric research. Although the tools or implements as well 28 as the technical actions can be identical or similar in the different lines of study, the products are 29 difficult to document from the archaeological perspective in spite of the recent growth in 30 archaeobotanical research and experimentation. Yet ethnographical analogies alone cannot be 31 considered as pure evidence of the products and processing techniques of the past, but should serve 32 as a source or model to reflect on these questions. 33 This paper relies almost exclusively on the data available on querns driven with a to-and-fro 34 movement (henceforth referred to simply as querns) and hand-driven rotary querns, and does not 35 take into consideration particulars regarding, for example more complex mechanical mills such as 36 watermills. This research does, nonetheless, include mortars, implements that are essential in the 37 operational sequence of processing certain cereals. It must be stressed that grinding carried out with 38 a quern is based on friction (or shearing, in the case of dressed grinding surfaces) and involves 39 reducing grains into small or fine particles. This differs from the act of pounding in a mortar that 40 results in a crushed, broken or mashed product. Yet the research consulted in this study does not 41 deal equally with the wide variety of variables related to grinding. Certain work focuses on specific 42 questions such as quern and millstone production, whereas other focuses on the role of grinding 43 stones in the operational sequence of processing specific types of plant products. Other studies, by 44 contrast, are of a wider interest as they are centred, for example on the process of manufacture of 45 querns and mortars of stone or wood, or draw up inventories and cartographies of grinding tools in 46 the framework of their domestic units. Another line of research centres on morphological and 47 technical studies of the grinding implements, their function and daily use, the members of the 48 community that operate them, operational sequences of processing more common products and the 49 variety of tasks that go into preparation of food (Hillman 1984;D'Andrea and Haile 2002). Examples 50 of these types of research include the work of Hayden (1987a) and Searcy (2011) in Mexico and 51 Guatemala, Roux in Mauritania (1985), Moreover, there is a clear imbalance of data in favour of querns compared to that of rotary querns. 56 Hand-driven rotary querns, contrary to querns, and in spite of often being cited in general works 57 describing the variety of grinding or processing tools (e.g. Gast 1968;Bruneton 1975 Wollstonecroft 2011). This is not the case of the rotary quern since it was not introduced in Europe 68 until protohistory, toward the middle of the first millennium BC, a time frame to which ethnographic 69 models have rarely been applied (Alonso and Frankel 2017).

70
As noted, manually driven rotary querns in many recent societies are being gradually substituted by 71 either electric or mechanical mills, or by other industrial processes. However, and fortunately for this 72 research, traditional hand milling has not been totally abandoned and in many cases is resorted to 73 attain a specific product that requires a particular type of flour. This type of grinding is also applicable 74 in the case of products associated with rituals or special plates served in festivities of different 75 cultures.

77
Tools and their use in manual cereal grinding 78 79 Ethnoarchaeological studies reveal that certain communities made use of a wide variety of grinding 80 and pounding implements (Alonso 2014 and stone basins, mortars dug into bedrock, mortars dug into large blocks, small mortar 85 of stone and wood, small shallow stone mortars for meat, deep and flat wood mortars and rotary 86 querns. 87 As noted in the "Introduction" section, although this study focuses principally on querns, stone and 88 wood mortars play a vital role in tandem with querns and other tools in food and plant processing. 89 Mortars, in fact, have a more "social" connotation than querns, at least among certain communities 90 such as the Minyanka (Koutiala, Mali), as they could at times be used simultaneously by up to eight 91 individuals (Hamon and Le Gall 2013a, p. 276).

92
Bedrock mortars, that is holes carved into rock outcrops serving to pound products, are of 93 particular interest. Examples outside the villages of the Dogon (Mopti, Mali) served to pound millet 94 and peanuts for oil (Clemente et al. 2002, pp. 88-89). In Sukur (Adamawa, Nigeria), they served to 95 process rice, legumes or fruit (David 1998 types of materials in one area of Central Anatolia can perform identical processes. The author notes 104 that these differences are not necessarily related to a lack of raw material or to different cultural 105 traditions. The choice may simply be due to a particular family or group preference. Similarly, 106 dehusking (discussed below in the "The operational sequence of cereal grinding" section) may vary 107 depending simply on the preference of a group. David (1998, p. 25) records that populations of the 108 Mandara Mountains (Adamawa, Nigeria) rarely use wooden mortars because, unlike the groups in 109 the plains, they prefer products with a higher fibre content and therefore do not remove the husks 110 before grinding.

111
The type of rock chosen to manufacture querns, although variable, is most often local or regional. authors note, however, that no systematic norm can be applied as the choice of petrography of 120 certain groups is determined by the proximity of the rock outcrops. 121 The relation between the choice of raw material, functionality and more general socioeconomic 122 aspects, subjects of more specialised studies, are not clear in the three cases. The link between the 123 functional needs and morphology of querns, for example does not appear to be direct or consistent. 124 Moreover, morphological variation is also not necessarily tantamount to functional variation. 125 However, in certain cases, when there is a need to control the product being ground, the mill's 126 morphology can play a role. An example is when grains are humidified before grinding, a common 127 practice that facilitates grain retention in the mill that will be examined below (Horsfall 1987, p. 352).

128
Querns are known to have very long working lives (sometimes several decades) which leads to a 129 gradual morphological transformation due to wear (Hamon and Le Gall 2013b, p. 119). The factor of 130 the angle of the grinding surface can also be associated with the type of product. In Algeria, 131 Mauritania and Cameroon, for example a flat to slightly concave surface served for cereal grinding. But 132 over time, as the concavity of the grinding surface increased, the stones no longer served for cereals 133 and were converted into condiment grinders (Gast 1968 As noted previously, querns in many localities fall into two large categories depending on whether 160 they serve for cereals or condiments. Although in certain localities they are multifunctional, they 161 never serve to grind oleaginous seeds as these always require pounding in mortars (Teklu 2012, p. 72).

162
Moreover, all of the grinding implements in Nepal are multi-functional with the exception of the 163 watermill which is reserved exclusively to grind grains (Baudais and Lundström-Baudais 2002, p. 175).

164
Multifunctionality can over time be either simultaneous or successive. In fact, a multi-functional use 165 is recorded in certain cultures such as the Maya for querns originally serving for maize but later 166 adapted to grind salt, pigments and calcite (Hayden 1987c, p. 191). Furthermore, distinct populations 167 in identical cultural contexts in a restricted geographical zone can act differently. In Guatemala for 168 example the Q'eqchi' and Poqomam population normally use the same quern to grind maize and a 169 series of other foodstuffs (with the exception, at times, of coffee that was ground in a separate 170 quern). However a similar population in a nearby region, the K'iche', reverted to different querns to 171 grind maize, coffee, achiote and chilli (the last two condiments ground with a smaller model) (Searcy 172 2011, p. 76). 173

Cereals and their processing 175 176
Supply, processing and cooking plant foodstuffs play a fundamental role in the diet and economy of 177 most traditional societies, whether they be farmers or hunter-gatherers. Mills are involved in the 178 following four types of preparations: (1) groats, semolina and flours; (2) beverages and beers; (3) 179 sauces and condiments; and (4) medicines. The register of fruits, nuts, seeds, leaves, roots or tubers 180 ground or pounded for consumption is vast not only in societies using traditional technology, but in 181 kitchens all over the world. Yet since all cannot be treated in this work, we centre on cereals and on 182 the first type of preparations listed above. 183 Many important wild or domesticated plants (particularly legumes and cereals) essential to human 184 diet cannot be digested correctly if consumed raw. For this reason, certain authors (e.g. 185 Wollstonecroft 2011, p. 141) consider that processing and cooking of plant foods plays a vital role in 186 the course of human evolution because it facilitates access to a large quantity of nutrients that 187 humans would not otherwise be able to consume therefore expanding the range of the diet and 188 leading to more secure and stable foodstuffs.

189
The techniques of grinding, pounding and grating have very specific purposes (Katz et al. 1974;190 Stahl 1989, p. 172; Teklu 2012, p. 26). They separate the elements that are suitable for consumption 191 from those that are not (e.g. fibre in both grains and tubers), they change the shape of the product 192 by reducing its size leading to a better digestion, and they help eliminate the toxins contained in 193 some plants. 194 Moreover, in some cases, the techniques lead to the addition of nutrients. In this sense, it is 195 interesting that cereals such as maize, vital to the diet of a great part of the world's population, has 196 nutritional limitations. Maize is, in fact, only of minimum value as a daily source of nutrients if not 197 processed by a specific alkaline technique, and a population depending only on untreated maize 198 would be malnourished (Katz et al. 1974). Evidently, this treatment is only relevant among groups 199 with a high production and consumption of maize (Duffy 2011, pp. 14-15).

200
Cereal is the staple food of most agricultural societies and, from our point of view, a detailed 201 knowledge of its processing is indispensable to understanding the tools, that is the mills and mortars, 202 that take part in its transformation. The data available is unfortunately uneven depending on the 203 type of cereal and the tools engaged to process them. As noted, most ethnographic and 204 ethnoarchaeological mill studies focus on querns driven with a to-and-fro movement. Hence, the 205 data regarding the cereals processed in these types of mills refer basically to maize, sorghum, millet 206 and ragi (finger millet). There is, however, little data regarding the grinding of wheats and barleys by 207 querns driven with a to-and-fro motion (Table 1), the more common cereals throughout the 208 prehistory and protohistory of Europe and the Mediterranean (as observed by Samuel 2010, p. 457).

210
The physical characteristics of cereals 211 212 Cereals present a variety of characteristics that pre-determine not only the crop-processing stages, 213 but their preparation and processing, the focus of the current study (Charles 1984 where the grains and husks do not separate in spite of threshing. These hulled cereals thus 223 require an additional step in the process to separate the grains from the glume and the 224 whole of the spikelet. 225 On the other hand, cereals are sorted depending on whether they are free-threshing (that they 226 easily release the grain during threshing whether it be naked or hulled) or not. For instance, "naked" 227 grains (caryopsis) of certain species such as common wheat (Triticum aestivum) easily release the 228 rachis, the glumes and the palea and lemna. Likewise is the case of hulled barley (Hordeum vulgare) 229 where the threshing separates the "hulled" grain (with the palea and lemna fused with the grain) 230 from the fragments of awns and rachis. Conversely, there are other species, non-free-threshing, that 231 do not release their grains as easily and are reduced to spikelets containing one or two grains. 232 Examples of this type include emmer (Triticum dicoccum), einkorn (Triticum monococcum) and rice 233 (Oriza sativa) (cf. e.g. Hillman 1984, fig. 4; Van der Veen and Jones 2006, p. 218, fig. 1).

234
Cereals are usually stored in the form of grain (either naked or hulled) or in the form of spikelets 235 (depending on whether they are of the free-threshing type or not). They can also be stored semi-236 clean, that is accompanied by small fragments of straw, rachis, awns or weeds. Most of the cleaning 237 was undertaken during the threshing process. Yet there remains a final, smaller scale cleaning that 238 can be added, as will be seen, to the steps of the overall operation of grinding. Therefore, the 239 operational sequence of processing each type of cereal depends, as noted above, on a variety of 240 characteristics.

242
The operational sequence of cereal grinding 243 244 Ethnoarchaeological analyses of the grinding of cereals rarely reveal detailed descriptions of each of 245 the steps of the process or of the different end products. Exceptions are the following series that a series of stages that vary depending on whether the species or variety is hulled or naked, the 249 degree of cleaning at the time of storage or the desired product or products. These stages lead to 250 processing operational sequences aimed at obtaining a product or a series of products that are to be 251 either immediately cooked or preserved over a medium term. Furthermore, although certain 252 operations are common to all species, others are associated specifically to distinct species. 253 Dehusking is a particular operation involving hulled cereals that requires a major effort. In this case, 254 the husk must be removed before the grinding process so as to be consumed whole and avoid 255 breaking or crushing. Another system, as noted below, is that of grinding before the removal of the 256 chaff fragments. 257 Husking in itself can comprise different stages. A pretreatment of the grain, for example is one of 258 the most widely documented measures put to use to remove the chaff. The most common 259 techniques are drying, roasting and soaking (Lundström- Baudais et al. 2002). Drying and roasting 260 render the chaff fragile, easier to break and pulverise. This is applied to hulled wheat such as barley, 261 millet and maize. In dry climates, spikelets can simply be dried in the sun (Hillman 1984, p the chaff become more flexible facilitating an almost intact separation from the grain (Nesbitt and  264 Samuel 1996, pp. 52-53). It must be noted that roasting (barley, for example) also leads to a better 265 preservation of the final product (Ferchiou 1985, p. 194). 266 The implements to carry out this operation are varied. G. Hillman (1984, p. 129) argues that the 267 mortar is the most effective tool to remove the chaff of hulled wheats, especially if they have a 268 strongly curved inner shape. This is and was the most frequent technique in many settings like 269 Ethiopia, Italy, Yugoslavia, Turkey, Iran or India (Harlan 1967, in Hillman 1984 A final cleaning of the grains is carried out before initiating any grinding operation. As noted, the 296 grains are usually stored in a semi-clean state and, before grinding or dehusking, all the unwanted 297 elements such as chaff fragments or weeds are removed. Generally this is carried out by sieving and 298 manually removing the larger unwanted weeds. In the case of roasting, the burnt elements are 299 removed. At this stage, the grains are either cooked, ground or prepared as bulgur, couscous or other 300 products that can be stored for later use. 301 We have presented several examples of consecutive operations undertaken with stored grains 302 (most often after threshing) to obtain one or several products. These operations in all cases involve 303 the transformation of a small amount of cereal and are tasks that require only a few hours of daily 304 work. Large quantities are not processed for storing except in the case of the preparation of certain 305 "pre-cooked" products such as bulgur or couscous (Hillman 1985). Grinding can yield from one to a 306 number of products, as well as a series of sub-products destined for animals or for other purposes 307 that are not included in published schemas (see the example of the potential products, sub-products 308 and residual products in the schema of Fig. 5, Alonso et al. 2014).

309
As illustrated in Figs. 3, 4 and 5, the sequence is made up of a variable number of operations (with 310 sieving, in particular, recorded in detail). The simplest processing, after removal of the impurities, is 311 that of naked wheat, in particular common wheat (Fig. 3a (1)). This species only requires a 312 combination of grinding followed by sieving by centrifugation. Centrifugation sieving demands a 313 rotary movement that retains the finer particles in the centre of the sieve while the coarser products 314 collect around the periphery. The two resulting products are fine flour and coarse semolina. Common 315 wheat is a soft grain with a whitish and floury endosperm yielding a fine flour of high quality. Durum 316 wheat, on the other hand, has a hard, less floury grain. In fact, when ground, it is broken down for 317 the most part into small yellow granules. It requires two sievings with different mesh sizes. These 318 combinations of sieves result in at least four products (coarse and fine groats, and coarse and fine 319 semolina) but never a fine flour like that of common wheat (Fig. 3a (2)) (Alonso et al. 2014, pp. 19-320 22). 321 In the case of hulled wheats, the goal of a large part of the process is to dehusk the grains. As 322 described previously, this can be carried with various implements including the rotary quern ( Fig. 3b-323 d). Obviously, dehusking after grinding or pounding the spikelets requires a series of tasks to 324 separate the dehusked grain from the chaff (in particular the glumes and the bases of both the 325 spikelets and the glumes). These operations to remove the impurities are carried out for the most 326 part by different types of sieving, winnowing (with baskets) and picking out by hand. Washing with 327 water can also be carried out to separate the bran (Fig. 3c and d). Furthermore, the number of times 328 the grain passes through the mill varies depending on the desired product (Hillman 1984 The husk of hulled barley, although easy to separate from the spike, is strongly bonded to the grain. 336 Dehusking involves two methods. The first takes place prior to grinding by pounding in a mortar and 337 then either sieving or winnowing, or winnowing or separating in an esparto open basket (Teklu 2012, 338 p. 69; Bruneton 1975, pp. 276-280) ( Fig. 4a and b). The second involves grinding the hulled grain 339 directly before separating the grain fragments from the other elements (Alonso et al. 2014, pp. 22-340 23) (Fig.4c).

341
In the first case, prior to dehusking, as in the case of wheats, the grinding operation involves only 342 one pass, either in a quern or a rotary quern, followed by sieving. In some instances, this stage 343 consists of a quick pounding in a mortar since the main purpose is to simply break the highly resistant 344 barley awns and glumes. The second case, by contrast, requires a second round of grinding. Roasting, 345 in turn, renders the glumes more fragile so they break more easily during grinding. This is followed by 346 a sieving whose main objective is to separate the fragments of grain from the fragments of chaff, the 347 largest of which are removed during a subsequent sieving. After the second grinding, resulting in the 348 elimination of the chaff, the different products can be classified as shown in the schema of the 349 grinding process by the Ouertani women of El Kef (Tunisia). The detailed schema of the Ouertani women 350 was possible to establish due to a systematic sampling and recording of each of the products, sub-351 products and residues after each operation (Alonso et al. 2014) (Fig. 5).

352
The grinding process of pearl millet in Mali (Hamon and Le Gall 2013b, pp. 111-112), in turn, serves 353 as an example of the processing of millet (Fig. 6a). The grains are usually dekerneled outside the 354 town in a low mortar specially designed for the task. Then the product is winnowed before soaking to 355 soften the glumes. This is followed by dehusking in a tall, deep mortar and washing to finish 356 separating the grains from the chaff. Depending on the desired product, it can then either be 357 pounded in a mortar or ground in a quern followed by sieving to obtain the desired texture.

358
Gast and Adrian (1965, pp. 22-24) record an alternative system of processing millet at El Ahaggar 359 (Algeria) similar to the previous system recorded in Mali requiring only mortars. According to the 360 locals, this system is the best means of attaining the most suitable flour. The procedure is initiated by 361 dehusking the soaked grains in a mortar followed by winnowing. Then the grains are soaked and 362 pounded a second time in a mortar before a final winnowing. 363 Maize, as noted previously, requires an alkaline treatment so as to take advantage of all of its 364 nutritional properties (Katz et al. 1974). The process consists of boiling in lime water (Adams 1999, p. 365 479; Duffy 2011, p. 9, 14), soaking and rinsing, followed by a final grinding while moist so as to obtain 366 a dough that is generally cooked immediately (Fig. 6b). Therefore, some cereals are ground after they 367 are soaked for a certain amount of time or by adding water when they are in the mill so that the 368 product gains the consistency of a thin dough. This practice concerns maize for the most part, as 369 noted previously, and other cereals such as sorghum in Sudan (Dirar 1993 In Yemen, the technique is applied to both sorghum and millet before overnight soaking in water. It is 372 then ground several times in a rotary quern until obtaining the required consistency (Bornstein-373 Johanssen 1975, p. 293). Similarly, grinding is also part of the operations required to produce 374 beverages such as beers deriving from cereals. A rapid and coarse grinding, for example converts 375 malted grains into groats or semolina (fragments) increasing the surface area of each grain allowing a 376 greater penetration of water (Samuel 2000, p. 550). 377 There are many ways of proceeding and it is essential to retain that the process also depends on 378 the experience and skill of the women millers, as well as other factors such as the quantity of grain to 379 grind, the types of sieves and the amount of available time. If grinding is carried out with rapid 380 movements, for example, the yield of grain will be higher and take on the form of groats. To 381 achieve fine flour, by contrast, grinding must be carried out more slowly while feeding the mill with 382 smaller quantities of grain. This observation is valid for both querns and rotary querns. 383 The importance of sieves in the grinding process cannot be stressed enough. Their meshes can be 384 made of tendons, tripe or metal. As seen above, sieves can serve different functions depending on 385 their calibre. Yet the aspect of mesh calibre, classified in general from 1 to 3 mm, is largely ignored in 386 most ethnoarchaeological work (Alonso et al. 2014, p. 200). 387 Moreover, the different operations of the sequence can be carried out in progression by one or 388 more than one person. In the American South-West, for example a woman initiates the sequence 389 with a rough grinding before passing the coarse semolina to a second woman that grinds it into a 390 finer material. This procedure continues until obtaining the desired texture (Adams 2002, p. 116 The manner in which querns are put to use (miller posture, position and inclination of the mill, type 396 of movements...) can vary depending on the type of product. The Minyanka women in Mali to grind 397 cereals with "overlapping" querns (a quern type where the length of the upper stone exceeds the 398 width of the lower stone) on either a cloth, or a natural or synthetic mat in order to collect the 399 product that falls to the sides of the lower stone's distal end. Yet when working with smaller querns, 400 they place a container (a gourd or a plate) under the quern's distal end to catch the product (Hamon 401 and Le Gall 2013b, p. 115). According to the same authors, the posture of the miller and the motion 402 applied to the upper stone may depend on several factors: the type of upper stone (size and weight), 403 the inclination of the lower stone, the length of the grinding surface, the type of movement, the 404 nature of the product and the desired final product. 405 The publications consulted indicate a diversity of means of driving a quern in an inclined or flat 406 position, either elevated or at floor level (Fig. 7). With the quern in a flat position at floor level, the 407 miller, most often a woman, can adopt a posture on the knees (Fig. 7a) or sit cross-legged (Fig. 7b). 408 The other positions, elevating the proximal end of the quern, result in an inclination away from the 409 miller. When adopting a posture on the knees at floor level, an inclination can be attained by placing 410 an elongated stone under the lower stone (Fig. 7c) or by using a quern equipped with legs (such as 411 the Meso-American metate) (Fig. 7d). The posture can involve kneeling and squatting on the heels so 412 that the grinding pressure derives from the arms, shoulders and torso. When on the knees, all of the 413 body leans forward generating and transmitting more traction (Gast 1968 , fig. 23).

414
The millers can also raise the mill so as to grind in an upright position. In this case the mill is set on a 415 raised earthen feature (Fig. 7e), on a wooden table (Fig. 7f), on a tripod fixed to the floor, or even on 416 a plastic drum turned upside down (Searcy 2011, pp. 119, fig. 5.11 and 120, fig. 5.12). The location of 417 the quern within the domestic unit may also depend on its type. Grain querns of the Dii of northern 418 Cameroon, for example are in the kitchen on a table. Condiment mills, although at times on grinding 419 tables, are often portable and either on the floor of the courtyard or in the kitchen and driven while 420 on the knees (Gelbert 2005, pp. 333-334).

421
An inclination of the mill is very common as it simplifies the grinding process as it increases the 422 weight on the upper stone and therefore it pressures on the grains (Horsfall 1987, p. 332 Roux 1985). These characteristics include "initial 431 properties" of the mill (dimensions, volume and raw material) and "modified properties" resulting 432 from a complex longer or shorter cycle of use and repeated dressings of the grinding surface (analysis 433 of the shape and dimensions of the grinding surface, convexity and concavity, state of the surface 434 and traces of wear). Furthermore, the typology is based on extrinsic characteristics such as the plant 435 species, grinding actions and the duration and intensity of use.

436
As noted previously, small mobile querns on some occasions can be used both in patios and 437 kitchens (Gelbert 2005, pp. 333-334) and often their location depends on the season, atmospheric 438 conditions, sheltering from the wind (Roux 1985, 439 p. 33) or if they are protected by specially built earthen features (Gelbert 2005, p. 322; Teklu 2012, 440 pp. 32-34). However, in other cases such as Ethiopia, they are usually in the kitchen and rarely in the 441 patios. In these cases, the larger grain querns are set on earthen structures to the side of the kitchen, 442 opposite the oven and fireplace (Teklu 2012, pp. 32-34). At times, two structures are raised side by 443 side resulting in two parallel mills: a larger mill for cereals and a second for smaller grains (Teklu 444 2012, p. 33, fig. 3.2). 445 The number of querns in a domestic unit normally depends on the number of adult women. The 446 approximate ratio of querns in societies with a diet based on cereals is two or three for grains and 447 one for the other condiments (Roux 1985 In Nigeria, for example each house has two "kits" to process a great variety of foods and substances 450 (David 1998, p. 23-25). The first kit, for grains, includes a quern, a quartz stone to dress the quern's 451 grinding surface, and a brush to collect the flour and clean the quern's surface. It is located inside the 452 house on a table with one to three other querns. The number of querns depends mainly on the 453 number of daughters. In the case of two querns, the grinding surface of one will be rough and serve 454 for coarse products (groats or semolina), while a second, with a finer surface, will serve for flour.

455
Women who possess a single quern can use a variety of coarser to finer upper stones to repeatedly 456 grind the same product. 457 The second kit includes concave querns used with elongated or globular-shaped upper stones 458 driven with a vertical, rotary or to-and-fro movement. Each house has one or more. Rotary querns, 459 by contrast, only saw occasional use due to access to more productive grinding mechanisms such as Rotary querns were driven at floor level ( Fig. 8a and b) or on elevated platforms beside a wall or in 470 a corner (Fig. 8c)  Berber women. The grinding takes place at floor level on the sheep skin that "catches" the product. 476 The skin is at times replaced by a cloth or an open sack. The woman works seated with one leg bent 477 back and the other stretched forward, and drives the rotary quern with one hand by means of a  478 vertical handle while the other hand feeds grain through the quern's eye. All the while rotation is not 479 halted (Fig. 8a). When tired, the woman has the option of switching hands. Similar systems are 480 documented in Greece (Parton 2011, p. 39) and Morocco (Bruneton 1975, p. 280). Yet in these cases, 481 both legs of the women are extended (Fig. 8b) Rotary querns can also drive at floor level as in the Moroccan High Atlas where they occupy a 488 specific place in the kitchen or the hallway, always covered and protected. The lower stone is fixed to 489 the floor with clay and a channel around the quern is fashioned to collect the product. When at rest, 490 the querns are covered with a cloth (Bruneton 1975, p. 279). 491 The Traditional manual cereal grinding: a task reserved mainly for women 501 502 Ethnographical studies indicate that the act of cereal grinding is very closely linked to women. The 503 preponderance of females in the tasks of processing and preparing cereals, in addition to other 504 foodstuffs, is well recorded. Although there are no detailed statistics regarding grain milling, it is 505 clear from ethnographical studies that the preparation of plant foods in almost all non-industrial 506 societies is the task of women (Murdock and Provost 1973;Alonso 2016). The graph in Fig. 9  507 illustrates the proportions of roles of women and men in 10 basic economic activities in agriculture, 508 stockbreeding, food processing and the procurement of basic materials. Grinding falls in the category 509 of processing plants for food (number 9 in the chart) that evidences the highest participation of 510 women.1 This perception is also ratified when taking into account not only the cases where this work 511 is exclusively female (83%), but also those where female participation is predominant, as well as 512 those in which female and male participation are equitable (97%). Cereal grinding, together with 513 water procurement (number 7 in the chart) and dairy product production (number 10 in the chart), is 514 therefore an almost exclusively female task. 515 This contrasts with the lesser role of women in tasks associated with cereal growing (in either slash-516 and-burn or intensive type of farming) than those in a system based on collecting roots and tubers 517 (Alonso 2016). Some authors have analysed this difference and considered that since cereals require 518 a more laborious secondary processing than roots or tubers, women in cereal-based systems were 519 compelled to dedicate less time to field work (Martin andVoorhies 1975 in White et al. 1981, p. 826; 520 Ember 1983). 521 Grinding, especially cereal grinding, in almost all the studies consulted for this article is a task 522 reserved exclusively for women even though men in certain nomadic groups such as the Tuareg also 523 are known to grind cereals with querns (Gast 1968 , fig. 23). Grinding of other types of non-cereal 524 materials in certain societies is also carried out by men. Men were responsible for the grinding of 525 ochre and snuff, for example in the Australian Western Desert (Hamilton 1980, p. 5) and other non-526 grain products in Nigeria (David 1998, p. 23). 527 Yet the querns themselves seem to be exclusively the property of women. This tool, whether in 528 gathering and farming societies, may be their sole lifetime possession (Hamilton 1980 Hamon and Le Gall 2011, p. 20). Querns could be obtained at marriage as a gift from the future 531 husband (along with mortars used collectively) or inherited from their mother. 532 Regarding millstone manufacture on the whole, it appears, by contrast, to be a task reserved 533 almost exclusively for men (Hayden 1987b; Hamon and Le Gall 2011; Searcy 2011, p. 32-65). 534 However, there are some examples where this task is undertaken completely by women. This is the 535 case, for instance, of the Fipa ethnic group in western Tanzania or other groups in other parts of 536 Tanzania and western Sudan (Haaland 1995, pp. 165-166). These women possess their own quern 537 quarries where they extract and fashion blocks into roughouts before finishing them in the village. 538 The women of the Dii of Djabe in Cameroon are also known to have manufactured their own querns 539 (Gelbert 2005, p. 324). In other cases, women participate in quern making in a partial manner. 540 Among the Lakia'a of the Tigray ethnic group in Ethiopia (Teklu 2012, p. 49), for example women 541 were consulted on the choice of the rock and intervened in the final stages of the fashioning process. 542 Hand grinding is an onerous, painstaking task even among gathering populations. Seed processing 543 and grinding among these communities is in fact clearly recognised as the most difficult task (Cane 544 1989, pp. 104-105). (Morocco) spend from 2 to 3 hours a day processing cereals (Bruneton 1975, p. 276 about 1 hour and 40 minutes of the women's time was dedicated to grinding while another half an 567 hour went to winnowing and other operations. There are even records that offer evidence that this 568 type of work was at times either undertaken by women specifically hired for the task or by slaves, as 569 in the case of Imperial Ethiopia prior to 1967 (Teklu 2012, p. 50 Grinding and pounding of different organic and inorganic materials, especially a wide variety of 574 cereals and non-cereal plant foodstuffs, are tasks closely related to human societies of all ages and, 575 in particular, to women. Grinding and pounding are procedures that are crucial to facilitate the 576 consumption of many plants that otherwise cannot be ingested. The procedures also provide 577 alternative culinary options as the different actions result in a multiplication of meals, avoiding the 578 monotony in certain culinary traditions based on a limited variety of foodstuffs. 579 In spite of the common factors linked to the characteristics of grains and the operations carried out 580 for their treatment, there is a large diversity of techniques and tools. This paper's review of the 581 general research regarding grinding, querns and related subjects provides a first broad vision of how 582 different societies put similar techniques to use. The interest of the current work focuses on the use 583 of hand-driven mills and other operations associated with grinding. 584 Among the variety of tools linked to grinding published in the literature, this study highlights that 585 the to-and-fro quern the object of more studies than the rotary quern in the ethnoarchaeological 586 literature. The study also outlines certain initial data on various aspects such as the factors that may 587 influence the choice of one or another tool, the choice of rock for manufacture, as well as the 588 relationship between raw material, functionality and more general socioeconomic aspects, mill use-589 life and multifunctionality (simultaneous or successive). 590 The operational sequences, nonetheless, go far beyond the mills themselves and involve mortars, 591 sieves and technical actions that can only be approached through ethnographic or 592 ethnoarchaeological records. Grinding with a quern is therefore only a link in a much more complex 593 sequence of operations. At the same time, the processing of cereals also provides a wide range of 594 products far beyond the simplistic view that mills served to grind flour to bake white bread. From this 595 point of view, an understanding of the diversity of products is essential to approach the wider 596 context of grinding grain in earlier times. This study, based on the few publications that deal with 597 these aspects, attempts to analyse the operational sequences and products of a number of cereals 598 including naked wheats, hulled wheats, hulled barley, pearl millet and maize. The search for these 599 products in the archaeobotanical record is therefore a very compelling line of research and 600 ethnographic sources must in all cases serve only as models to assist the inquiry.

601
This study has also taken into account the position adopted by millers, mainly women, as well as 602 the position of the mill and the actions undertaken during milling. All these aspects relate to the 603 work and daily life of women, their domestic space and family structure. In fact, the procedure of 604 preparing plant products for food, including the grinding of cereals, is the economic activity 605 undertaken almost exclusively by women who devote between 2 and 5 hours a day to the task. The 606 conditions in which these tasks are carried out are, therefore, of utmost relevance. Archaeological 607 research regarding early grinding and pounding tools and techniques requires new interdisciplinary 608 approaches taking into account a number of elements including quern lithology, the outcrops where 609 querns were quarried, the process of fabrication of querns, quern morphology, the types of materials 610 ground, grinding techniques, trace-wear, micro-residue, find-spot and tool re-use analyses, 611 archaeobotany of the products, distribution of work, and the socioeconomic relations of production. 612 This paper therefore represents a first overall attempt to extract and interpret ethnographic 613  Table 1).