Journal of Archaeological Science 40 (2013) 4331e4347

Contents lists available at SciVerse ScienceDirect

Journal of Archaeological Science

journal homepage: http://www.elsevier.com/locate/jas

An ethnoarchaeological study of cooking installations in rural

Uzbekistan: development of a new method for identification of fuel
sources
Shira Gur-Arieh a, b, Eugenia Mintz a, c, Elisabetta Boaretto a, c, Ruth Shahack-Gross a, *
a
The Kimmel Center for Archaeological Science, Weizmann Institute of Science, Hertzel St., Rehovot 76100, Israel
b
The Martin (Szusz) Department of Land of Israel Studies and Archaeology, Bar-Ilan University, Ramat Gan 52900, Israel
c
Weizmann Institute-Max Planck Center for Integrative Archaeology, D-REAMS Radiocarbon Laboratory, Weizmann Institute of Science,
Rehovot 76100, Israel

a r t i c l e i n f o a b s t r a c t

Article history: Mud constructed cooking installations such as ovens and hearths are common in modern, pre-modern
Received 15 April 2013 and archaeological domestic contexts in West and Central Asia. Archaeological cooking installations
Received in revised form are primarily identified using analogy of shape and size to ethnographic installations. The study pre-
31 May 2013
sented here establishes direct evidence to the use of fire within mud constructed cooking installations,
Accepted 1 June 2013
thus providing means for reducing ambiguity in identification of archaeological cooking installations. In
addition, we present here a newly developed method that enables a clear-cut distinction between wood
Keywords:
and dung ashes used as fuel materials in many modern and archaeological domestic contexts. The study
Cooking
Oven
is based on an ethnoarchaeological research in rural households at the Republic of Uzbekistan that was
Tandir followed by geoarchaeological analyses of installation walls, wood ash, dung ash and wood and dung
Ochock standards collected in the study area. Field work included ethnographic observations, interviews with
Ash pseudomorphs informants and temperature measurements during cooking experiments. We show that changes in the
Dung spherulites clay mineral structure due to exposure to high temperatures on the interior walls of cooking installations
Phytoliths can be detected using FTIR (Fourier Transform Infrared) spectroscopy, providing for the first time direct
Fuel evidence to the use of fire within such installations. We demonstrate that the temperature recorded by
PSR
clay alteration on installation walls as well as in the ashes left on installation bottoms does not corre-
spond to baking or cooking temperatures. We also show that the newly developed method, based on the
ratio of wood ash pseudomorphs to dung spherulites, separates between wood and dung ashes with very
high certainty. Yet, we identify a range of values where differentiation between wood and dung ashes is
uncertain, and suggest it results from intensive mixing processes. Lastly, we show that phytolith mor-
photype analysis is an insufficient tool, if used alone, for distinguishing wood from dung ashes in the
study area. The newly developed tools for temperature estimation within mud constructed installations
and for fuel origin determination contribute to better understanding of cooking-related practices, and
can be applied to archaeological contexts worldwide.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Palaeolithic ca. 400,000 years ago (Roebroeks and Villa, 2011).
Hearths can therefore be regarded as the earliest cooking in-
The manipulation of raw animal and vegetal materials into edible stallations. Mud constructed cooking installations are a much later
and nutritious food is a hallmark of human evolution (Carmody and development in human history, possibly dating as early as the PPNB
Wrangham, 2009; Organ et al., 2011; Wollstonecroft, 2011; culture in the southern Levant and Southeastern Anatolia, some
Wrangham, 2009). Cooking probably developed with the earliest 7000 years ago (Lyons and D’Andrea, 2003). The research presented
systematic use of fire by hominins, inferred from the identification of here focuses on mud constructed cooking installations, specifically
hearths associated with burnt animal bones as early as the Lower those related to west and central Asian types.
Mud constructed cooking installations in west and central Asia
have been described from pre-modern times by explorers, eth-
* Corresponding author. Tel.: þ972 8 934 3254.
nographers and ethnoarchaeologists (e.g., Avitsur, 1971; Dalman,
E-mail address: [email protected] (R. Shahack-Gross). 1964; Frankel, 2011; Kramer, 1982; McQuitty, 1984, 1993; Mulder-

0305-4403/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jas.2013.06.001
4332 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347

Heymans, 2002; Parker, 2011; Parker and Üzel, 2007; Weinstein, were used during operation of archaeological installations. While
1973). These installations can be broadly separated into baking in- ethnographic studies generally showed that wood and dung are the
stallations (dome-shaped ovens known as Tanur/Tandir, Tabun, and most widely-used fuels (Kramer, 1982; McQuitty, 1993; Miller,
two-storied ovens known as Wagdiah/Arsah/Mogadeh) and frying- 1982, 1984a, 1984b; Miller and Marston, 2012; Parker and Üzel,
roasting-boiling installations (hearths of various types). The Tanur is 2007; Pierce et al., 1998; Samuel, 1989; Zapata Peña et al., 2004),
a large (between 80 and 150 cm high and 50e80 cm in diameter) no study to date touched upon the proportions of such fuels within
cylindrical shaped installation built from soil with a ventilation hole specific installations. We therefore carried out an ethno-
at its base, and a large opening at the top. It is either dug into the archaeological research that was geared toward identifying phys-
surface of activity (Kramer, 1982), incorporated into a mud built ical and chemical criteria by which it could be determined whether
superstructure (Crawford, 1981) or is free standing with at least one a mud installation has been used for cooking, and toward under-
coating layer for insulation (McQuitty, 1984). It is usually fueled with standing what types of fuel can be identified from ashes found
wood, dung, and/or crop processing waste located at its base while within cooking installations. The ethnoarchaeological research
the dough for bread is placed on its inner walls. treated the modern installations and ashes as if they were
The Tabun is a hemispherical installation shorter than the Tanur archaeological, i.e., we carried out sampling in a manner similar to
(30 cm high and about 80 cm in diameter) with a wide opening at sampling archaeological installations and sediments in an archae-
the top from which the dough is inserted (McQuitty, 1984). Tabun ological context, for later geoarchaeological analyses.
shapes and sizes vary regionally (Frankel, 2011; McQuitty, 1993). In
northern Jordan, for example, the Tabun has a side opening from 1.1. Research area
which wood fuel and olive pulp after pressing (jift) are inserted,
and a built base where the dough is laid to bake. In southern Jordan The ethnoarchaeological study was conducted in Uzbekistan in
and in Israel Tabuns are laid on top of a shallowly excavated area collaboration with the Samarkand Institute of Archaeology, Uzbek
covered by river pebbles or potsherds. These installations have no Academy of Sciences.
side opening, and the fuel (mostly crushed dung and crop pro- The Republic of Uzbekistan is located in central Asia within the
cessing waste) is piled around the installation from the outside. Aral Sea basin. About 80% of its area is arid or semi-arid while the
Fuel is kept warm constantly all year long whereas refueling (and climate in the lowlands and mid-altitude is Mediterranean with
thus more intensive burning) occurs prior to every daily use. wet winters and dry summers (Gintzburger et al., 2003). More than
Ethnographic research on similar and other types of cooking 60% of the population lives in rural areas, of which about 34%
installations was also carried out in other parts of the world where subsist on agro-pastoralism (Payziyeva and Paiziev, 2012). The
cooking installations were (and are) still in use. For example, research reported here was conducted ca. 25 km south of Samar-
Greenhouse et al. (1981) studied the botanical residues from the kand, in the Karatyube Mountains area. The area is semi-arid with
use of earth ovens lined with pebbles by a group of Pima Indians in about 300 mm annual precipitation that carries mid-latitude
the Gila River reservation. Thoms (2008) studied cook-stone fea- steppe vegetation (Fig. 1).
tures used by North American Indians, and Amouretti (1986) The research was carried out in two villages, Sivasoy at the eastern
documented the use of a pit oven in the Sahara. Spatial distribu- part of the mountain ridge and Tolly at the western part of the same
tion patterns related to the use of hearths were studied in forager ridge, among two subsistence agro-pastoral families. Like other
societies in different parts of the world (Binford, 1998; Fisher and families in the research area, these families live in households that
Strickland, 1989, 1991), while studies on the botanical remains include a house for a nuclear family and separate houses for married
resulting from fueling hearths with dung were conducted by Miller sons and their families. Households include a courtyard where
in Iran (Miller, 1984b) and Reddy in India (1998). various economic and social activities are carried out daily, a garden
Archaeologically, hearths are prevalent in sites in west Asia since (that carries perennial vegetables as well as fruit trees) and animal
the Middle Palaeolithic while installations that somewhat resemble enclosures (Kandiyoti,1999 and personal observations). In these rural
Tanurs and/or Tabuns in shape and size are prevalent since the areas, houses are built from mud and vegetal materials, electricity
Early Bronze Age (4the3rd millenniums BC). A special hearth type, supply is scarce if existing, and although gas stoves are sometimes
consisting of a pavement of river pebbles, is known for example used, the main means of cooking, baking and heating are traditional
from Iron Age (ca. 3200 years BP) Israel and is ethnically associated installations. These traditional installations, locally called tandir and
with the Philistine culture (Gur-Arieh et al., 2012 and references ochock, are built from mud and vegetal matter (most commonly
therein). Field identification of these archaeological installations is straw). The tandir is a baking oven almost identical to the Turkish
based on simple analogy of shape and size between ethnographic installation bearing the same name. The ochock is a partially enclosed
and archaeological installations, sometimes coupled with attention hearth used for cooking, which includes roasting, frying and stewing.
to presence of burnt fuel (ash) within or on archaeological in- Cooking with the ochock is a daily task while baking takes place every
stallations, and somewhat also assisted by the general archaeo- two to three days. The general aridity of the environment, coupled
logical context (e.g., domestic context). While most of these with herd products due to the pastoral component in the economy of
identifications may be correct, we find that they are based on the society, dictates use of wood, dung and/or crop processing waste
insufficient and inconclusive data. The following questions are as fuel for cooking and baking activities.
crucial for better field identification of cooking installations:
2. Materials and methods
1. Is there evidence for heat within these installations?
2. Is the associated ‘ash’ indeed ash? 2.1. Field work
3. Is the associated ‘ash’ in situ, or a later dumping event into an
unused installation? Field work was conducted for a period of two weeks during
October 2011 and included ethnographic observations, interviews,
We could not find ethnographic documentation geared toward experimentation with various fuel types, and sampling of sedi-
answering these questions. Another research question that did not ments, raw fuel materials, ash, and installation wall parts. The study
receive much attention in archaeological research at historical sites in the Sivasoy household was made possible owing to the wife of
in west Asia relates to the identification of the type/s of fuel that the household’s owner. Additional information was obtained by
 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347 4333

Fig. 1. Maps showing Uzbekistan (inset) and the region of the study area. The location of the villages of Sivasoy and Tolly, in relation to the Karatyube Mountains and Samarkand, is
marked. Maps obtained from Ó2013 Google.

conversations with two of her daughters-in-law. The study in the 2.2. Field experiments
Tolly household was made possible with the help of the wife of the
youngest son of the household’s owner (Figs. 1 and 2a, b). In- Field experiments, designed to measure the temperature ranges
terviews were conducted with the help of an Uzbek interpreter. during cooking and baking in ochocks and tandirs were conducted
Material sampling, temperature measurements and experimenta- in both households. Five experiments were carried out in Sivasoy,
tion with various types of fuel were all conducted under full two in the tandir and three in the ochock. The variable tested was
permission. the type of fuel, i.e., wood was used as fuel in one experiment and

Fig. 2. (a) Sivasoy household, showing the main family house surrounded by fruit trees. (b) Tolly household, showing the general landscape and semi-arid environment. Arrows
point at the location of the kitchen structures. (c) An ochock in the final stages of cooking (Sivasoy). Note the placement of thermometer detectors, one within the glowing fuel and
another below the cooking utensil. (d) A tandir during initial stages of space heating (Sivasoy) showing placement of thermometer detectors, one within the glowing fuel and the
other inserted through a ventilation hole close to the tandir’s back wall. Note the heavily sooted walls behind both installations, and the soot close to the tandir’s main opening.
4334 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347

dung in the second, using the same installation and cooking/baking heated to 800e900  C calcite breaks down into calcium oxide
the same type of food. Four experiments were carried out in Tolly, (CaO), that immediately upon cooling absorbs humidity and forms
two in each installation, once with wood and once with dung. The calcium hydroxide (Ca[OH]2). This material is identified using
repetition of fuel type in the same installation was purposefully infrared spectroscopy by the presence of an absorbance band at
done in order to evaluate the reproducibility of results. 3643 cm1 (Weiner, 2010) (Fig. 3a). These changes in the infrared
In each experiment, the weight of fuel used during the cooking spectra of clay and calcite in soil thus provide means to determining
activity was recorded using a portable scale. Temperature along the the maximal temperature to which these minerals in a cooking
cooking activity was recorded every few minutes using two installation wall were exposed to.
portable thermometers (Newtron TM 5005, useful for temperatures
up to 1300  C with precision of 1  C  0.3%). One detector was 2.5. Microscopy
placed within the burning fuel at the bottom of each installation.
The second detector was placed between the installation wall and Microscopic remains that are expected to be found, and are
the cooking pot in order to measure the temperature just below the indicative of fueling materials include opal phytoliths, dung
pot’s bottom in the case of ochock experiments, while in tandir spherulites and wood ash pseudomorphs. All three were studied
experiments the second detector was placed in the back of the using a polarized light microscope (Nikon Labophot2-POL).
installation, hanging from the ventilation hole or from the main
opening, as close as possible to the tandir’s wall in order to record 2.5.1. Quantification of ash pseudomorphs and dung spherulites
baking temperature on tandir walls as best as possible (Fig. 2c, d). Both ash pseudomorphs and dung spherulites are micro re-
Samples for laboratory analyses were collected following the mains composed of calcite, each indicative of specific fuel material
experiments, after the installation walls and/or fuel cooled down. (Braadbaart et al., 2012; Canti, 2003; Lancelotti and Madella, 2012;
Installation walls were sampled by scraping or chipping small Matthews, 2010). Ash pseudomorphs are created when calcium
fragments from the bottom and top of the wall, inside and out. Ash oxalate crystals (CaC2O4$H2O) that form in living plants undergo
from the remains of fuel material inside the installations was heating to at least 450  C (see Shahack-Gross and Ayalon, 2013 for a
sampled as well. Different (unburnt) fuel materials were collected detailed description of their formation). Their typical shapes are
in both villages, including various plants and dung cakes. In Sivasoy, druses (crystal cluster), rombohedral and prismatic crystals that are
a household midden where ashes are regularly discarded was more common in dicotyledonous plants (hereafter dicots), and
excavated as if it was an archaeological feature, sectioned and ashy raphides or styloids that are more common in monocotyledonous
sediments were sampled vertically along its profile. The location of plants (hereafter monocots) (Franceschi and Nakata, 2005; Prychid
each sample was marked with a numbered note and photographed. and Rudall, 1999) and their size range is typically 10e50 mm. They
The following laboratory analyses were conducted in order to study are dark gray in plane polarized light (PPL) and have high order
the effect of heat on installation walls and sediments, and whether interference colors in cross polarized light (XPL) (Canti, 2003;
fueling practices can be deduced from ash. Shahack-Gross and Ayalon, 2013). Dung spherulites are spherical
bodies ranging 5e20 mm composed of radially crystallized mono-
2.3. FTIR (Fourier Transform Infrared) spectroscopy hydrocalcite that forms in the intestines of animals, especially ru-
minants (Canti, 1997; Shahack-Gross, 2011). Under cross polarized
FTIR analyses were conducted in order to determine the light they can be identified by their permanent cross of extinction
mineralogical composition of sediments and evaluate whether and low order white to second order orange interference colors
calcite and clay minerals have been transformed due to heating (Brochier, 1983; Shahack-Gross, 2011).
(Berna et al., 2007; Regev et al., 2010; Weiner, 2010). Spectra were In order to determine quantitatively whether fuel in cooking
obtained using Nicolet 380 FT-IR spectrometer (Thermo Electron installations was composed of wood, dung or both, quantification of
Corporation), collected between 4000 and 250 cm1 at 4 cm1 these microscopic remains is necessary. Oxalates in monocot plants
resolution using the KBr method. The results were evaluated based are often rare to absent (Prychid and Rudall, 1999), therefore, sty-
on a reference library (Weiner, 2010) and by conducting a heating loids and raphides are not expected to be found in high quantities
experiment specific to the soils from the study area (see Section 2.4 in dung ash. Rhombohedral and prismatic oxalate crystals are ex-
below). pected in high quantities in wood ash. Druses are known in both
monocot and dicot plants and are expected in both dung and wood
2.4. Soil heating experiment ashes.
While dung spherulites have been quantified recently from
Local soils around Sivasoy and Tolly are composed primarily of archaeological sediments (Albert et al., 2008; Berna et al., 2007;
clay (dominated by illite) and calcite. Samples of about 50 mg of soil Portillo et al., 2009; Simpson et al., 2003), their quantification
were heated for 4 h in a furnace oven (Adam Mandel T21Ó), each to was based on counting inhomogeneous samples prepared by
a different maximal temperature as follows: 200, 300, 400, 500, mixing a weighed aliquot of sediment sample with a high viscosity
600, 700, 800 and 900  C. After cooling, the heated soil was gluing material (usually Entellan New, Merck), on a microscope
analyzed using FTIR spectrometry and a reference of the spectral slide. We present here a new procedure that allows for homoge-
changes to the local soil with increased heating was built (Fig. 3a). nous spread of microscopic remains, based on the phytolith rapid
Two major criteria enable deduction of the temperature to which extraction method of Katz et al. (2010) that takes advantage of the
the soil has been exposed. The first is based on changes in the low viscosity of a heavy liquid solution prepared from sodium
infrared spectrum of the clay fraction in the soil which initially polytungstate.
undergoes significant change when burnt at 500  C noted by the The procedure is as follows: Between 5 and 40 mg sediment or
disappearance of the peaks at 527, 915 and 3620 cm1. When ash are weighed using an analytical balance (Sartorius
exposed to higher temperatures the main clay absorbance band BP121S  0.1 mg) and sieved through a 150 mm sieve in order to
shifts gradually from 1029 cm1 to 1048 cm1 (Fig. 3a). These eliminate grains larger than ash pseudomorphs and/or dung
changes occur due to irreversible structural changes in the clay spherulites (if the sediment contains aggregates they can be gently
mineral lattice (Berna et al., 2007). The second criterion is a change crushed with the edge of a spatula). When working with fresh ash,
in the infrared spectrum of the calcite present in the soil. When less than 10 mg should be weighed in order to avoid sample
 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347 4335

Fig. 3. (a) FTIR spectra of soil from the Sivasoy area heated to various temperatures in laboratory conditions. The spectral changes that occur at different temperatures allow using
these signatures for determination of heat exposure in cooking installation walls and ashes. Note that the absorption bands characteristic of clay (red labels) and calcite (blue labels)
change with temperature as follows: At 500  C the clay absorptions at 527, 915 and 3620 cm1 disappear and the calcite absorptions remain unchanged. At 600  C the main clay
absorption shifts from 1032 to 1037 cm1, a change that continues to 1046 cm1 gradually with increasing temperature to 900  C. From 600  C to 900  C the calcite absorptions
change as well, noted by lowering of the calcite characteristic absorptions relative to the clay absorptions, and appearance/formation of calcium hydroxide (Ca(OH)2) absorbance
band at 3643 cm1 (b) FTIR spectra obtained from: (1) Inner wall of tandir showing clay spectrum indicative to heating at about 600  C (red labels), and presence of the mineral
anhydrite (CaSO4; green labels). (2) Outer wall of tandir showing spectra characteristic of unheated clay (red labels) and unheated calcite (blue labels). (3) Dung ash from a cooking
experiment showing spectra of clay and calcite heated at about 900  C (main clay absorption at 1047 cm1 and presence of calcium hydroxide absorption at 3643 cm1). Note that
this ash spectrum also includes absorption bands characteristic of calcium phosphate (black labels). (For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this article.)

overloading, but in the case of archaeological samples between 30

and 40 mg sediments should be weighed due to their mixed nature. no: of micro remains counted
 area of the slide
The sieved sediments are then placed in a 0.5 ml conical centrifuge area counted
tube and 500 ml of 2.4 g/ml sodium polytungstate (SPT, ¼ no: of micro remains on the slide ðin 50 mlÞ ¼ A (1)
Na6(H2W12O40)$H2O) (Sometu Ltd., Berlin) are added. The tubes are
vortexed for about three seconds until the suspension is homoge-
nous and then sonicated for 10 min (ColeeParmer Ultrasonic
Cleaner) in order to ensure disaggregation. The tubes are then A10ð¼no:of micro remains in all solutionð500 mlÞÞ
1000
vortexed again and 50 ml of the solution are immediately pipetted initial sample weight ðmgÞ
out, placed on a microscope slide, and covered by a 24  24 mm ¼no:of micro remains in1g sediment (2)
cover slip.
Counting is performed using a polarized light microscope at Equation (1) calculates the number of microscopic remains
400 magnification. Counting is conducted cumulatively in 15 under the entire cover slip, i.e., within 50 ml of the sample solution.
random fields across the slide, once in PPL where ash pseudo- The “area counted” is calculated by multiplying the measured area
morphs are better identified, and once in XPL where dung spher- of one field with the number of fields counted. Equation (2) cal-
ulites are better identified. The results are calculated as follows in culates the number of microscopic remains in 1 g of sediment by
order to standardize the results presented as the amount of each also accounting for the fact that the result of Equation (1) (value A)
microscopic remain per 1 g of sediment: represents 10% of the total sample (50 ml out of 500 ml).
4336 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347

To evaluate the reproducibility of the method for ash pseudo- drying in the oven at 50  C overnight. Preparation of ash in the
morphs, fresh ashes of oak (Quercus calliprinus) and pomegranate laboratory from fresh Uzbek dung was preceded by homogenization
(Punica granatum) wood and bark were prepared experimentally in (light grinding) of about 6 g of cow dung and of sheep/goat dung
the laboratory by burning small branches in the furnace for 4 h at that was used by the informants to fuel the cooking installations. All
550  C after washing with distilled water and sonicating for 30 min these fuel materials were burnt at 500  C for 4 h in a furnace oven.
to remove external contamination. These two species were chosen The ashes were weighed after cooling to room temperature.
because initial observations and measurements of ash pseudo-
morphs showed that pseudomorphs from oak are larger (ranging 2.5.2. Phytolith analysis
15e50 mm) than those of pomegranate (ranging 10e25 mm). We Phytoliths are hydrated silica (opal) particles that form in living
thus tested the procedure’s reproducibility using pseudomorphs of plants. Thus they provide, to some extent, taxonomic information
different sizes. Counting triplicates of 8 sub-samples of oak ash and about the plant material brought to a site (Piperno, 2006). Two
13 of pomegranate ash gave an average error of 11% (Table 1). parameters in phytolith analysis may be used to differentiate dung
To replicate possible mixing of ash with other types of sediment from wood ashes. Phytolith concentrations in wood/bark are much
in an archaeological site, two samples from each species were lower than in grasses (i.e., dung), and phytolith morphologies in
prepared by mixing fresh ash with powdered mud brick in pre- wood/bark and grass are distinctively different (Albert et al., 1999;
determined weight ratios of 1:2 and 1:4. We therefore expect Piperno, 2006).
that ash pseudomorph concentrations from each mixture will be a Phytolith analysis was performed according to the rapid
third or a fifth of the pseudomorph concentration in fresh ash. Each extraction method of Katz et al. (2010). About 20 mg of sediment
sample was counted in triplicate. The results show that pseudo- were placed in a 0.5 ml conical tube and dissolved by adding 50 ml
morph concentrations reduced as expected, and the measure- HCl 6N. 450 ml of SPT (2.4 g/ml) were added, the sample vortexed
ment’s average error is 14% (Table 1). Fresh dung and wood ashes for about three seconds and sonicated for 10 min. The samples were
from the study area were used to determine the reproducibility in centrifuged for 5 min at 5000 rpm (Eppendorf centrifuge 5418).
dung spherulite counting. Three sub-samples were counted in Only opaline particles that could unequivocally be identified as
triplicates resulting in an average error of 31% (Table 1). We phytoliths have been counted. The samples included many multi-
conclude that the maximal error in ash pseudomorph and dung cell phytoliths (silica skeletons). Unlike conventional procedures,
spherulite quantification is in the order of 30%. the multi-cellular phytoliths were not counted as 1, but the total
We note that SPT has a pH of about 4.5 in the working density number of individual cells within them was counted. The reason for
(2.4 g/l) (Skipp and Brownfield, 1993). This raised a concern that this decision stems from taphonomical considerations. Sediments
some of the calcitic micro-remains might dissolve during the pro- with low phytolith concentrations sometimes result from poor
cedure. However since the results are reproducible and previous phytolith preservation (Cabanes et al., 2011, 2012). Thus counting a
work showed that reaction between calcite and SPT starts only after multi-cellular phytolith as one phytolith lowers the concentration
one hour (Skipp and Brownfield, 1993), it is methodologically safe results in a manner that a well preserved sample might erroneously
to use SPT for this procedure. be considered a poorly preserved one. Phytolith morphologies were
The method was applied to study ashes produced during the studied and identified according to the standard literature
field experiments, ashes produced in the laboratory from collected (Mulholland and Rapp, 1992; Madella et al., 2005; Piperno, 1988).
fresh fuel materials, and sediments from the Sivasoy ash midden. Wood/bark and dicot leaves phytoliths were identified based on the
Preparation of ash in the laboratory from fresh Uzbek wood fuel was presence of characteristic morphologies (mainly irregular, paral-
preceded by cleaning of about 10 g of twigs from adhering dust lelepipedal blocky and thin, platelet, spheroid, dicot hair, hair base,
through sonication (30 min), washing with distilled water and polyhedral, and tracheid phytoliths; Albert et al., 2000).

Table 1
Quantities of ash pseudomorphs and dung spherulites in Oak (Quercus calliprinus), Pomegranate (Punica granatum), weighed mixtures of these wood ashes with mud brick
material, and dung ash from ethnoarchaeological cooking experiments. For each material the number of samples tested and the number of repetitions on each sample is listed,
as well as the average result and average error (presented in percentage as the coefficient of variance between the standard deviation and average).

Material No. of samples Number of slides Total Average micro-remains Error (%)
counted from each (ash pseudomorphs/dung
spherulites) per 1 g ash (106)

Clean oak ash (Quercus calliprinus) 6 3 18 56  6.9 12.0

Clean pomegranate ash (Punica granatum) 11 3 33 191  17.9 9.1
Oak ash (Quercus calliprinus) þ mud brick 1:2 1 3 3 20  2.1 10.8
Oak ash (Quercus calliprinus) þ mud brick 1:4 1 3 3 10  3.0 30.1
Pomegranate ash (Punica granatum) þ mud brick 1:2 1 3 3 76  0.7 1.0
Pomegranate ash (Punica granatum) þ mud brick 1:4 1 3 3 31  3.9 12.8
Wood ash from cooking experiment in Uzbekistan SUZ 60 3 1 12 5  1.6 31.2
3 3
Wood ash from cooking experiment in Uzbekistan SUZ 70 1 1 10 1  0.4 35.7
3 3
Wood ash from cooking experiment in Uzbekistan SUZ 95 1 1 10 4  0.4 11.6
3 3
Dung ash from cooking experiment in Uzbekistan SUZ 78 7 1 16 37  5.4 14.6
3 3
Dung ash from cooking experiment in Uzbekistan SUZ 147 1 1 13 70  14.0 20.1
4 3
Dung ash from cooking experiment in Uzbekistan SUZ 147 1 1 10 21  6.4 31.3
3 3
Dung ash from cooking experiment in Uzbekistan SUZ 147 2 1 11 4  2.8 76.3
3 3
 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347 4337

3. Results In all households, a specific roofed structure is dedicated for

cooking (Fig. 2a, b). Such ‘kitchen’ always includes at least one
3.1. Ethnographic observations tandir and benches for food preparation and placement of utensils.
The tandir is usually built of two layers made of soil. The inner layer
The houses in the study area are traditionally built of unbaked which is made of fine carefully smoothed soil, is often bought from a
mud bricks or “pachsa” (local term for large blocks of casted mud), local professional builder, and brought unbaked to be placed in the
on stone foundations (di Cugno et al., 2013 and our own observa- kitchen. The outer layer, composed of local soil that includes coarse
tions). Most work is done by men, but we also observed women rock fragments, mixed with chaff and sometimes goats’ hair, is laid
participating in the building process. The households are usually by the resident woman/women. We observed two ways of placing
comprised of a main family house where the owner’s nuclear tandirs within ‘kitchens’ e one by orienting the main opening
family lives, usually in two rooms; one for storage and working and vertically so that it is facing the roof and the tandir’s ventilation
another that serves for sleep, hosting and other tasks such as hole/s is thus found close to the ground surface. With this type of
preparation of dough for baking. Additional houses may be added installation orientation the women have to bend over the opening
as the owner’s sons marry and build their own houses. Old in order to bake (Fig. 4a). The second possibility of tandir orienta-
damaged houses serve for storage or for animal keeping. In Sivasoy tion is by placing it so that its main opening is facing the room
the household included the owner’s house and two additional interior and the ventilation hole/s is at the back, close to the room
houses for his sons’ families. In Tolly the owners, an elderly couple, wall. The main opening of the installation is fixed at a height that
live in an old structure while the main house is used by their fits the chest height of an average person thus it does not require
youngest son’s family. Old structures in this household serve for bending in order to insert or take out the bread (Fig. 4b, c).
storage and animal keeping. The ochock is a partially enclosed hearth (key hole or horseshoe
Households are frequently parts of villages. Both Sivasoy and shaped) built from soil mixed with chaff. It may be located either
Tolly are well-spread villages thus neighbors are found only several indoors within a kitchen, or outdoors e either attached to a house
hundreds of meters from the main family house. In both Sivasoy wall or not (Fig. 5). The ochock includes two large openings, one at
and Tolly water supply was from a nearby stream channel. The its top on which cooking utensils are balanced and one in front
families subsist on herding (mostly sheep and goats, but also cows where fuel is added. Cooking on the ochock is mostly done using a
depending on the relative wealth of the family) and small scale large metallic bowl (Fig. 5).
horticulture (grapevines, fruit trees, seasonal vegetables). Poplar The kitchen in Sivasoy includes one functioning tandir, one old
trees, used as building materials and fuel, are tended as well. tandir that is used for meat roasting in special occasions, and one

Fig. 4. (a) A tandir in a village close to Tolly, oriented vertically. Note the ventilation hole at the ground level (arrow). (b) The main tandir in the Sivasoy household (arrow), bought
from a professional tandir builder, oriented horizontally on a mud bench. The installation on the left is an old tandir that was built by the informant on top of an old ochock which
went out of use. (c) Bread cakes being baked inside the Sivasoy tandir. Arrow points at the tip of the thermometer detector that measured wall temperature. Note that glowing
embers are piled away from the baking bread, to avoid accidental bread burning. Also note the natural cracks and broken areas in tandir walls. (d) Woman at Sivasoy sticking dough
onto the hot tandir wall wearing an arm protector made from an old coat sleeve. (e) A metal pot with water, placed on hot embers after baking in the tandir was over.
4338 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347

Fig. 5. (a) An ochock that was built attached to the kitchen exterior wall at Sivasoy. (b) Free standing ochocks at Tolly. Note the horseshoe shape of these installations, the placement
of cooking utensils (resembling Woks) on their top opening, and occasional mud benches that may, or may not, be associated with them.

ochock. Another ochock, built outside the kitchen but attached to protection made from a sleeve of an old coat (Fig. 4d). While the
the kitchen’s wall, is used primarily by one of the sons’ brides that bread is baking it is occasionally sprayed with water to prevent it
lives in the household. The kitchen in Tolly included a functioning from burning. Dough placed in the lower part of the tandir wall is
tandir and an old tandir. Several ochocks in Tolly are concentrated more prone to burning. The heat in this part of the tandir is
in a small area (3  2 m) outdoors. We noted that kitchen walls are controlled by covering the embers in certain localities of too high a
covered by thick black soot, especially above and around ochocks, temperature with cold ash or a piece of metal to decrease the heat.
probably due to the combination of oil and soot following frying In cases the heat at the lower part of the tandir is not sufficient for
activities. Installation internal walls and close to openings are also proper baking, embers were shifted closer to the dough to locally
blackened by soot (Figs. 4 and 5). elevate the temperature. Sometimes the main opening of the tandir
Fuel materials, primarily wood and dung, are stored indoors. was covered by a metal utensil in order to keep the heat within.
Wood fuel in Sivasoy is mostly poplar (Populus sp.) while in Tolly it When the bread is ready it is pulled out from the walls and slightly
is mostly mulberry (Morus sp.). Dung in different stages of prep- wiped with some water.
aration (into ‘dung cakes’) is present around the household, on We observed only preparation of meat stews in the ochocks. The
house walls or in the cow and sheep sheds (Fig. 6). When ready cooking of meat stew was very similar in both households and
(dry), dung cakes are stored in the fuel storage area together with started by heating some oil followed by addition of potato slices,
wood, and other materials not necessarily related to fuel (Fig. 6). In peppers, onions and meat. Water was then added and the stew was
Sivasoy dung cake preparation takes place inside the cow-shed left to boil for about 30e60 min. We noticed that once cooking or
during the winter where the cakes are stuck on the internal baking activities ceased a kettle was always placed on the embers to
walls to dry. During the summer the cows are herded away from boil water for tea (Fig. 4e).
the household, in grazing grounds at high elevations in the
mountains. Dung cakes are prepared on the ground at the summer 3.2. Field experiments
camp and are brought back to the village in a car. In principle dung
cakes are said to be stored for winter use only. The wife of the Table 2 details the type and amount of fuel used in each
house owner at Sivasoy complained that her brides tend to use experiment and the time elapsing between initial fire lighting and
dung also in the summer. The informant in Tolly prefers using end of cooking.
wood over dung because she perceives the former as cleaner and The temperature measured during the experiments in the
better smelling. ochocks shows a fast increase in the space (from ambient tem-
Baking and cooking are solely the domain of women. Dough is perature to 600e700  C in 10 min or less). The space temperature
prepared by mixing flour, water and some yeast (commercially was unstable when wood was used as fuel, fluctuating upwards
bought) in a large metal bowl. When the dough is ready it is shaped each time more wood was added (Fig. 7a.1). When dung was used
into balls and laid to leaven. The leavened dough is worked into flat as fuel the space temperature rose as fast as using wood, but was
“cakes” of approximately 30 cm in diameter. The dough cakes are kept steadier over time (Fig. 7a.2). The temperature measured
then perforated using a wooden device. Approximately 10e20 within the burning fuel rose slower than the temperature in the
bread cakes are prepared in a baking episode and last for about 3 space (from ambient temperature to 400e800  C in 10e30 min),
days in an extended family of ca. 10e15 people. and was kept relatively stable over time, slowly decreasing after the
Despite difference in the orientation of tandir ovens in the two initial increase (Fig. 7a.3). Note that temperatures were measured
villages, bread baking is similar. Fire is lit within the tandir using also after the cooking activity was completed.
dry grass or small branches followed by insertion of wood, dung There does not appear to be a difference in temperature
cakes, or both. The fire is fueled so that high flames are created as behavior in the tandir space when used with wood or dung
well as much smoke. At this point the tandir walls are covered by (Fig. 7b.1). The wood produces a fast rise of relatively high tem-
black soot. The fuel is mixed occasionally with a metal pole until it perature (700e800  C in 25 min at Sivasoy, and 10 min at Tolly).
is thoroughly turning into embers. According to the informants, the The use of dung also produced a fast increase of temperature but it
indication that it is time to insert the dough is when the soot on the did not reach more than 600  C. The temperature behavior
tandir internal walls is completely burnt away. The dough cakes are measured within the tandirs fuel also shows a fast increase, but
then sprinkled with water, placed on a round textile cushion and stays relatively steady between 400 and 500  C, without significant
the wet surface is stuck onto the tandir walls using an arm differences between wood and dung (Fig. 7b.2).
 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347 4339

Table 2
A list of the field experiments conducted using the two types of fuel in the two
installation types at the two studied households. Experiment #6 started with
glowing embers at the tandir due to previous use of this installation on the same day.
The thermometer that was placed in the space of the tandir in experiment #7 dis-
functioned after 20 min. The time column indicates time measured from the
beginning of fire to the end of cooking (i.e., until the food is taken out of the in-
stallations). Note that the temperature measurements presented in Fig. 7 continued
after the food was taken out of the installations.

Experiment Village Installation Fuel Weight Time Food

number (kg) (h)

1 Sivasoy Ochock Wood 6.5 2:08 Meat stew

2 Sivasoy Ochock Dung 6.5 1:36 Meat stew
3 Sivasoy Ochock Wood 6.8 1:09 Meat stew
4 Tolly Ochock Wood 3.1 0:54 Meat stew
5 Tolly Ochock Dung 4.2 0:50 Meat stew
6 Sivasoy Tandir Wood 4.5 0:57 10 Bread
cakes
7 Sivasoy Tandir Dung 19.5 1:20 10 Bread
cakes
8 Tolly Tandir Dung 8.6 1:00 16 Bread
(þwood) (þ1) cakes
9 Tolly Tandir Wood 5.7 0:40 12 Bread
cakes

heat-altered clay in the range of 500e600  C in both Sivasoy and

Tolly (Fig. 3b.1). Samples collected from the top part of ochocks
inner walls did not include heated clay, except for one sample from
Sivasoy. Samples taken from exterior walls of the various ochocks
showed no signs of clay alteration due to heat (Fig. 3b.2), even when
heavily sooted, except for one sample from Tolly. In all ash samples
collected inside the ochocks the clay fraction showed signs of
exposure to very high temperatures ranging between 700  C and
900  C. In three of these seven ash samples calcium hydroxide
(Ca(OH)2) was present, supporting exposure to temperatures as
high as 800e900  C (Fig. 3b.3).
Tandir walls should be intact in order to carry out baking. This
limited the number of samples collected and focused the sampling
mostly in areas already cracked within the tandirs. In Sivasoy,
tandir inner walls were exposed to 500e900  C both at the bottom
(close to the installation floor where the fuel was located), as well
as higher on the back wall. At Tolly the clay in the inner walls at the
bottom was not altered in one case and altered at about 600  C in
the other and a sample taken from the interior wall at the opening
of the old tandir contained unaltered clay. Five out of six samples
from the floors of Tolly’s tandirs included clay minerals that were
altered at temperatures in the range of 400e500  C.
All exterior walls of installations from both villages showed no
signs for heat-altered clay. The clay fraction in the ash samples
collected from inside the tandirs showed alteration due to exposure
to temperatures in the range of 700e900  C, one sample also
Fig. 6. (a) Dung cakes drying on a house wall at the village of Sivasoy. (b) Fuel storage
contained calcium hydroxide (Ca(OH)2) that supports exposure to
in an old kitchen at Sivasoy household, including dung cakes at the left and poplar such high temperatures. Anhydrite (CaSO4) was identified on the
twigs at the center. Behind the poplar twigs the outline of an old tandir is visible on the interior walls of all tandirs.
wall, delineated by black soot. Note the general un-tidy organization of this storage
room, including its un-swept floor. (c) A section in the Sivasoy household midden
where ash from tandirs and ochocks, as well as other food remains such as bones and 3.4. Quantities of ash pseudomorphs, dung spherulites and
feathers, old clothing and tools, are discarded. Note the layers formed by various phytoliths
discard episodes.
Quantities of ash pseudomorphs, dung spherulites and phyto-
3.3. FTIR liths were first determined in fuel materials collected in both
households. These materials, including wood, cow dung and sheep/
FTIR spectra of samples collected from walls of installations goat dung, were ashed in laboratory controlled conditions (500  C
were used to assess the level of clay and calcite alteration due to for 4 h) and served as standards in this study (Table 3). Using the
exposure to heat, based on the experimental data we produced in information of weight percentage left from each fuel material after
the laboratory (Fig. 3a). burning, the quantities of the three micro-remains were calculated
Two ochocks were sampled in Sivasoy and three at Tolly. Sam- per 1 g of organic fuel material, i.e., before burning (Table 3). Wood
ples taken from inner walls at the lower parts of the walls included standards had no dung spherulites, as expected, variably high
4340 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347

a.1
Temperature ºC
Ochock space wood b.1 wood/ dung
Tandir space wood

Temperature ºC
Time since lighting (hh:mm) Time since lighting (hh:mm)

a.2 Ochock space dung b.2 Tandir fuel wood

wood/ dung
Temperature ºC

Temperature ºC

Time since lighting (hh:mm) Time since lighting (hh:mm)

Ochock fuel wood

wood/ dung
a.3
Temperature ºC

Time since lighting (hh:mm)

Fig. 7. Plots showing the temperature measured during the various cooking experiments in ochocks (a) and tandirs (b). Green symbols represent the use of wood and black symbols
represent the use of dung as fuel materials. Squares represent experiments that took place in Sivasoy (note a square with black cross representing a second ochock experiment in
Sivasoy), and triangles represent experiments that took place in Tolly. (a.1) Temperatures measured in ochock spaces, just below the cooking vessel, in experiments using wood as
fuel. (a.2) Temperatures measured in ochock spaces, just below the cooking vessel, in experiments using dung as fuel. (a.3) Temperatures measured within burning ochock fuel in
experiments using wood and dung. (b.1) Temperatures measured in tandir spaces, close to the tandir wall, in experiments using wood or dung. (b.2) Temperatures measured within
burning tandir fuel in experiments using wood or dung. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

amounts of ash pseudomorphs (7 and 17 million in 1 g organic fuel Quantities of micro-remains in the standard fuel materials were
material), and low quantities of phytoliths. Pseudomorphs included used to calculate their expected amounts in 1 g of ash left after
rhombs, prisms and occasionally druses. Two cow dung and one burning theoretical mixtures of wood and dung, taking into ac-
sheep/goat dung standards contained low amounts of ash pseu- count the different weight loss of each sample (Table 4).
domorphs (1.2 and 1.7 million in 1 g organic fuel material), while
the other sheep/goat dung standard contained about 5 million ash 3.4.1. Modeling phytolith concentrations vs. a Pseudomorphs/
pseudomorphs. These differences reflect diverse dietary habits of Spherulites Ratio (PSR)
animals (Cowlishaw and Alder, 1960; Meyer et al., 1957). Pseudo- Plotting the three parameters of micro-remains present in the
morphs included rhombs, prisms and occasionally druses, and fuel materials was simplified by calculation of a Pseudomorphs/
notably no styloids or raphides (the latter typical of monocot Spherulites Ratio (hereafter PSR). This value takes advantage of the
plants). Both sheep/goat dung standards contain more spherulites facts that wood does not contain dung spherulites and has low
than cow dung. Phytolith quantities in the fuel standards range phytolith concentrations while dung contains variable amounts of
between 0.04 and 0.07 million per 1 g of wood and between 7 and wood ash pseudomorphs, dung spherulites and phytoliths
15 million per 1 g of dung (Table 3). depending on livestock diet (for a detailed explanation see Section
 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347 4341

Table 3
Quantities of phytoliths, ash pseudomorphs and dung spherulites in clean and pure (standard) wood and dung fuels from Sivasoy and Tolly. The weight loss due to burning was
used in order to calculate the amount of the micro-remains in 1 g of the organic matter. The last column gives the PSR values in each material. Note that although we rely here
on 4 dung samples only, the spherulite concentration in these samples is consistent with previous studies where spherulites have been quantified from modern dung ash
(Shahack-Gross and Finkelstein, 2008).

Sample Weight Phytoliths Ash pseudomorphs Dung spherulites PSR

Initial Final weight (g) Weight % Counted in Calculated for Counted in Calculated for Counted in Calculated for
weight (g) after 4 h 500  C left 1 g ash 106 1 g fuel 106 1 g ash 106 1 g fuel 106 1 g ash 106 1 g fuel 106

Cow dung SUZ 175 Sivasoy 5.717 1.101 19.3 78.4 15.10 6.2 1.2 9.9 1.9 0.60
SUZ 118 Tolly 5.719 1.548 27.1 39.1 10.60 6.4 1.7 94.5 25.6 0.07
Sheep/goat SUZ 79 Sivasoy 5.605 1.321 23.6 28.9 6.80 19.7 4.6 256.8 60.5 0.08
dung SUZ 104 Tolly 5.228 1.322 25.3 46.8 11.80 4.8 1.2 420.8 106.4 0.01
Wood SUZ 89 Sivasoy 1.861 0.046 2.5 1.7 0.04 272.1 6.7 e e e
SUZ 154 Tolly 9.487 0.511 5.4 1.3 0.07 321.9 17.3 e e e

4.4) (Table 3). A two-dimensional plot of PSR and phytolith con- using standard dung and wood samples from the study area
centrations was thus built in which the values of ashed standards (Table 4). Mixing wood with spherulite-rich dung forms a steep
and their theoretical mixtures (Table 4) can be easily appreciated mixing curve (Fig. 8b) indicating that very large amounts of wood
(Fig. 8a,b). It can be seen that pure wood ash does not have a PSR are required to shift the dung signature away from a PSR value
value because it does not contain spherulites, however, if we lower than 1, and phytolith concentrations decrease rapidly. Mixing
arbitrarily consider the presence of 1 dung spherulite in pure wood wood with spherulite-poor dung (Fig. 8b) forms a moderately
ash, PSR values can be calculated and they are in the range of sloping mixing curve that requires lower amounts of wood to shift
hundreds of millions. For such pure wood ash, phytolith concen- the dung signature from PSR value lower than 1 while phytolith
trations are low (below 2 million phytoliths per 1 g ash) and it plots concentrations do not change drastically. We note that large
at the far lower right of the graph (Fig. 8a). Pure dung, with high or amounts of wood relative to dung are required to obtain PSR values
low dung spherulite content, plots at the upper/middle left part of of 10 and above, thus any PSR value above 10 indicates wood as the
the graph having distinctively low PSR values (always below 1) and major fuel component. Interpretation of samples with PSR values
phytolith concentrations above 20 million in 1 g ash (Fig. 8a). For between 1 and 10 having low phytolith concentrations (below 20
comparison, the values obtained from a local soil sample are million in 1 g ash) is most difficult because it cannot be concluded
plotted as well, falling at the lower left part of the graph (i.e., low with certainty whether such values represent well-preserved
amounts of all micro-remains; Fig. 8a). A clear separation in values samples, predominance of wood ash, predominance of dung ash,
between pure wood and pure dung ashes is thus evident. or ash mixed with soil material. We therefore define this range as “a
When mixtures of wood and dung are used to fuel an installa- gray area” (Fig. 8b). The patterns presented in Fig. 8b thus serve as a
tion, mixing curves are expected to form between PSR values and quantitative model that allows approaching interpretation of un-
between phytolith concentrations. We calculated such mixtures known fuel sources.

Table 4
Calculation of the expected concentrations of phytoliths, ash pseudomorphs and dung spherulites in ashes resulting from mixing the standard wood and dung (the latter with
either high or low spherulite concentrations), at different ratios. The PSR values of each theoretical mixture are given as well. These data were used to create Fig. 8b.

Sample Initial Final weight (g) Calculated phytoliths Calculated ash Calculated PSR of theoretical
weight (g) after 4 h 500  C per 1 g ash (106) pseudomorphs dung spherulites results
per 1 g ash (106) per 1 g ash (106)

Wood: dung 1:30 31 7.55 47.1 7.1 423.0 0.02

(high spherulites) 1:10 11 2.55 46.5 11.6 417.5 0.03
SUZ 154: 104 1:5 6 1.55 45.6 18.0 409.4 0.04
1:2 3 0.55 43.1 36.0 387.1 0.10
1:1 2 0.30 39.7 61.9 354.8 0.20
2:1 3 0.35 34.2 102.6 304.2 0.30
5:1 6 0.50 24.4 175.8 212.9 0.83
8:1 9 0.65 19.1 215.3 163.8 1.30
10:1 11 0.75 16.7 232.8 141.9 1.60
20:1 21 1.25 10.6 278.4 85.2 3.30
40:1 41 2.25 6.5 308.8 47.3 6.50
50:1 51 2.75 5.6 315.7 38.7 8.20
100:1 101 5.25 3.6 330.5 20.3 16.30
Wood: dung 1:30 31 5.75 78.8 9.3 10.0 0.90
(low spherulites) 1:10 11 1.95 77.5 15.1 9.8 1.50
SUZ 154: 175 1:5 6 1.00 75.6 23.4 9.6 2.50
1:2 3 0.43 70.4 45.9 8.9 5.20
1:1 2 0.24 63.3 77.3 8.0 9.70
2:1 3 0.29 52.6 123.8 6.6 18.80
5:1 6 0.44 35.2 199.8 4.3 46.00
8:1 9 0.59 26.6 237.2 3.2 73.20
10:1 11 0.69 22.9 253.1 2.8 91.40
20:1 21 1.19 13.9 292.5 1.6 182.10
40:1 41 2.19 8.2 317.3 0.9 363.60
50:1 51 2.69 6.9 322.8 0.7 454.30
100:1 101 5.25 4.3 334.3 0.4 907.98
4342 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347

Fig. 8. (a) A plot of PSR vs. phytolith concentrations in standard wood and dung ash. This plot shows that pure wood ash and pure dung ash have very distinctive ranges of PSR and
phytolith concentrations. A soil sample plots in a distinctively different area relative to the two ash types. (b) PSR vs. phytolith concentrations in calculated mixtures of wood and
dung in changing ratios based on standard wood and dung samples from the study area (data in Table 4). Gray diamonds represent mixtures of wood with spherulite-rich dung, and
black diamonds represent mixtures of wood with spherulite-poor dung. Mixtures between these two end members create distinctive mixing curves. PSR values above 10 indicate
predominance of wood ash while PSR values below 1 indicate predominance of dung ash. PSR values between 1 and 10 associated with phytolith concentrations lower than 20
million in 1 g ash, are difficult to assign to a specific fuel origin (“gray area”). These data serve as an empirical model for interpretation of ashes from ethnographic installations and
middens, and consequently archaeological sediments. (c) A plot showing the PSR vs. phytolith concentrations in ethnoarchaeological wood ash (gray triangles), dung ash (black
triangles) and the Sivasoy midden sediments (gray circles). All ethnographic samples fall on the range between pure wood ash and pure spherulite-rich dung ash with relatively low
phytolith concentrations.
 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347 4343

3.4.2. PSR and phytolith values in ethnographic ashes of known fuel 3.5. Phytolith morphologies
materials
Quantities of ash pseudomorphs, dung spherulites and phyto- Differentiating wood from dung ash using phytolith morphol-
liths were determined in fresh wood and dung ash samples from ogies is based on the premise that monocot phytoliths will be more
cooking experiments, four control soil samples, and sediment abundant in dung as it is mainly composed of digested grass, while
samples collected from the household midden at Sivasoy (Table 5). dicot phytoliths will be more abundant in wood and bark. Past
Ash pseudomorph quantities are higher in installations fueled by research showed that monocots not only produce up to 20 times
wood, and dung spherulite quantities are higher in installations more phytoliths than dicot wood and bark and 16 times more
fueled by dung (Table 5). While there is a general higher quantity of phytoliths than dicot leaves, they are also present as an exterior
phytoliths in ethnographic dung ashes (8e17 million per 1 g of contamination on dicot bark (Albert et al., 2003). All samples
dung ash) vs. those in wood ash (5e10.5 million per 1 g of wood studied, including fresh wood and dung ashes as well as sediments
ash) these quantities overlap. from the Sivasoy household midden, contain phytoliths of both
Fig. 8c shows the results from the ethnographic ashes. All but 1 monocots and dicots. Both wood and dung ash samples are variable
of the wood ash samples (Table 5) have PSR values well above 10 in the relative proportions of monocots/dicots (Fig. 9). We note that
and low-medium quantities (5e10.5 million per 1 g ash) of phy- certain wood and dung ash samples may have similar phytolith
toliths, placing them in the expected range for wood ash. Dung ash morphotype compositions. The phytolith morphotype analysis in
samples have PSR values lower than 1, and medium quantities of this ethnographic case study does not contribute to distinguishing
phytoliths (8e17 million per 1 g ash). Despite the relatively low between the two fuel types (see explanation in Section 4.4).
phytolith concentrations, the PSR value clearly indicates a dung ash
origin. The results obtained from the ashes of known fuel thus 4. Discussion
support the usefulness of the PSR and phytolith concentration plot.
We note that all ethnographic wood ash samples contain dung The study presented here touches upon issues related to
spherulites and that phytolith concentrations in ethnographic dung archaeological inferences associated with cooking. It relates to the
ash samples are lower than expected. This indicates that certain mere identification of installations as used for cooking, to their use
other taphonomic processes occur (see Discussion). temperatures and fueling materials.

3.4.3. PSR and phytolith values in ethnographic ashes of unknown 4.1. Identification of archaeological cooking installations
fuel sources
The usefulness of the PSR to phytolith-concentration plot was A basic question often encountered during excavation is
tested on ashy sediments of an unknown fuel origin in the form of a whether installations that resemble cooking installations recorded
layered midden at the Sivasoy household (Fig. 6c, Table 5). One ethnographically were indeed used for cooking in the past. In most
sample had no dung spherulites and 1.7 million ash pseudomorphs publications we found that interpreting certain features as cooking
per 1 g sediment, and therefore was unequivocally identified as installations relies mostly on resemblance based on the form/shape
wood ash. One sample, having a PSR value well above 10 and low of archaeological installations to ethnographic ones (e.g., Parker,
phytolith concentrations is safely interpreted as wood ash. Two 2011). Sometimes these interpretations are supported by context.
samples have PSR values below 1 and the highest phytolith con- We would like to emphasize that ashes within archaeological in-
centrations indicating they originate primarily from dung ash. stallations cannot be used as an indicator for in situ burning that
Three samples fell within the “gray area”. occurred within the installation because ashes may have been

Table 5
Quantities of phytoliths, ash pseudomorphs and dung spherulites in 1 g ash/sediment from the ethnoarchaeological samples (local soil, wood and dung ashes from cooking
experiments, and the Sivasoy midden). Calculated PSR values of each sample are also given.

Sample Phytoliths per 1 g Ash pseudomorphs per 1 g Dung spherulites per 1 g PSR
ash/sediment (106) ash/sediment (106) ash/sediment (106)

Soil SUZ 62 0.52 0.24 0.00 e

SUZ 76 3.80 0.14 0.21 0.67
SUZ 80 3.72 0.14 0.00 e
SUZ 117 1.07 0.00 0.00 e
Fresh wood ash SUZ 60 9.13 14.98 5.16 2.90
SUZ 70 5.53 116.56 1.21 96.00
SUZ 89 4.99 85.33 0.91 93.55
SUZ 95 10.52 50.98 3.82 13.36
SUZ 131 4.81 75.58 1.41 53.61
SUZ 155 7.31 150.73 6.96 21.64
SUZ 168 6.55 98.61 0.31 319.63
Fresh dung ash SUZ 78 14.31 4.30 37.07 0.12
SUZ 147 16.69 3.60 69.78 0.05
SUZ 167 8.37 0.23 20.70 0.01
SUZ 174 11.20 2.66 3.76 0.71
Sivasoy household SUZ 63 6.95 1.24 0.49 2.55
midden SUZ 64 4.34 0.14 0.08 1.77
SUZ 65 3.87 1.71 0.00 e
SUZ 66 2.54 0.47 0.24 1.98
SUZ 67 15.26 1.75 15.80 0.11
SUZ 68 1.59 11.97 0.20 59.58
SUZ 69 14.82 0.69 5.75 0.12
4344 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347

Fig. 9. A diagram showing the relative percentages of phytolith morphotypes grouped into monocots (dark gray), dicots from wood/bark (gray), dicots from leaves (white),
unassigned origin (light gray), melted (black) and weathered (very light gray), in ethnographic ash samples from cooking experiments, the Sivasoy midden and one dung standard.
The midden samples are ordered from left to right according to stratigraphy, from bottom to top. Note the high variability in morphotype proportions in wood ash and the
resemblance in dicot and monocot proportions between certain wood and dung ashes.

discarded into unused installations that served for purposes not for short periods of time (no longer than 15 min per installation
related to fire, such as storage. operation) during the cooking activity. This observation indicates
In order to approach the question whether certain archaeolog- that the installation walls could not have been thoroughly heated
ical installations have been used with fire, the FTIR analyses of which may explain the lower temperatures recorded on ochock in-
installation walls, presented in this study, is very useful. We show ner walls relative to tandir inner walls. As cooking in the ochock
that in both tandirs and ochocks the outer side of the installation involved liquids (oil, water) the danger of food burning is lower than
walls were not affected by heat, while the inner walls were affected during baking. This allows cooking at high fire temperatures. We
by heat so that they contained evidence for alteration of calcite and noted that after the initial frying, at high temperature, simmering of
clay minerals in temperatures in the range of 500e900  C, similar stews occurred at lower temperatures accompanied by stopping the
to temperatures recorded by us during cooking practices. addition of fuel into the cooking installation, which allowed for a
We note that the degree of clay alteration is variable on the slow decrease in cooking temperature with time.
installations’ inner walls (i.e., 70% of the inner wall samples We noted that soot forms during the initial stages of combus-
included altered clay), possibly reflecting the inconsistent behavior tion, and is later partially oxidized. Soot is oxidized when in direct
of fire within installations during different use episodes. This contact with flames within the cooking installations, however,
observation indicates that when sampling an archaeological some soot may remain close to the openings of cooking in-
installation in order to test whether it was used with fire, sampling stallations e ventilation holes, tandir mouth, and ochock sides.
strategy should consider the amount of samples and their location Thick accumulation of oily soot was recorded on walls above
on the installation’s inner and outer walls. ochocks where frying activities took place. The chances of preser-
Once an archaeological installation has been proven to include vation of these patterns were not studied by us.
heat-altered clay on its internal walls, it may be interpreted as a
cooking installation also considering its archaeological context. 4.3. Fuel selection, preferences and efficiency
Note that kilns may also produce similar patterns, but their
archaeological context is expected to differ from that of domestic Several factors determining fuel selection in rural environments
cooking installations. Such positive identification allows going into have been identified in past research, among them availability, cost
the next stage of inference, i.e., estimation of cooking temperature (in time and/or money), fuel efficiency and fuel redundancy (i.e., if
and identification of fueling materials. the fuel material is not needed to other purposes as building ma-
terial or fertilizer) (Davis, 1995; Gwavuya et al., 2012; Hoeck et al.,
4.2. Estimating cooking temperatures 2007; Hosier and Dowd, 1987; Joon et al., 2009). Availability pri-
marily relates to the environment, whether it is wooded or arid,
Our results show that the temperatures recorded by clay alter- and to the economic base of the residents (i.e., whether or not
ation on tandir inner walls are in the range of 700e900  C. However, keeping livestock that supplies dung). Another interesting factor
we noted that the actual baking was done only after the temperature that affects fuel choice is cultural such as the Maasai general
dropped to 250e300  C. Thus it appears that in baking ovens clay avoidance from using dung for cooking as it is regarded ‘un-clean’
alteration records the initial heating of the oven space and walls (Shahack-Gross et al., 2004) or the taboo on using specific tree
prior to actual baking. The temperatures recorded on ochock inner species among the Fang villagers in Equatorial Guinea (Picornell
walls are in the range of 500e600  C while the maximum temper- Gelabert et al., 2011).
atures recorded at the base of the cooking pot during food prepa- Fuel selection in the study area seemed to have been primarily
ration were up to 700  C. We note that such high temperatures lasted guided by availability. The sparsely wooded nature of the
 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347 4345

environment coupled with herd keeping favors the use of dung over prevalent than spherulite-poor dung, making the latter’s mixing
wood. This tendency has been previously observed in other rural arid curve (Fig. 8b) irrelevant in this ethnographic case study. In order to
areas such as Iran (Kramer, 1982; Miller, 1984b). In the study area shift the PSR value characteristic of dung ash (lower than 1) into the
wood was mostly used during the summer while dung was prepared wood ash range (PSR higher than 10) quite intensive mixing is
as dry cakes and stored for use during the winter. Yet, there was no required (about 60 times more wood than dung; Table 4). The PSR
strict seasonal separation as personal preferences also play a role. values obtained from ethnographic dung ash are restricted in range
Fuel energetic efficiency may also play a role in decisions (between 0 and 1) while PSR values for ethnographic wood ash are
regarding which fuel to use during cooking. Looking at the results spread across a wide range (3e900) (Fig. 8c). These values reflect
of our cooking experiments it is hard to point out the more effi- variable degrees of mixing between wood and dung ashes. We
cient fuel (Table 2). When using the same amount of wood vs. defined a “gray area”, ranging between PSR values of 1 and 10 and
dung in the same ochock at Sivasoy, the whole process lasted phytolith concentration that is less than 20 million per 1 g ash,
30 min less with the dung, but then in the ochock in Tolly it took indicating a range of values that in an archaeological case will be
1.1 kg more dung for the same time of cooking (about 50 min). difficult to supply a clear-cut inference as to the fuel source. The
When looking at the tandir experiments we see that in both cases “gray area” reflects intensive mixing in which one cannot differ-
baking with dung as fuel required larger amounts (15 kg more at entiate whether both fuel materials have been used together or
Sivasoy, and 3 kg more at Tolly) and also lasted longer (23 min whether mixing occurred due to rake-out activities. We noted that
longer at Sivasoy, and 20 min at Tolly). Note that tandir volumes in rake out is intended simply to clear the installation space rather
both villages are about the same. On the other hand comparing the than keeping it clean, thus ashes from previous use episodes
span of time the temperature stays above 500  C we see that in all remain at the installation’s bottom and will mix with ash that forms
cases the dung is more efficient (11 min more in average with the in later use episodes. Other types of mixing may also result in
ochock, and 7.5 min more with the tandir). Taken together, the obtaining values in the “gray area” (see Section 4.5 below).
heterogeneity of the results suggests that from an energetic point Our results indicate that phytolith analysis by itself is not sen-
of view, there is not much difference between dung and wood as sitive enough to identify fuel origin in the case of fuel mixtures,
fuel materials. even when combining both phytolith concentration and morpho-
logical analyses. Theoretically, wood/bark can be distinguished
4.4. Identifying fuel materials from grasses based on morphotype composition. However, we note
that human behavior may obscure this theoretical separation. First,
Fuel identification in archaeological sites helps reconstruct the we observed that each cooking event begins with starting a fire
environment (climate and flora), and it can provide evidence for using grasses and small twigs. As monocots (grasses) produce more
animal husbandry. Previous studies either rely on the analysis of phytoliths than wood and bark, it is expected that in all ashes
charred macro-remains (e.g., Miller et al., 2009), on micromor- produced in cooking installations monocot phytoliths will be more
phological identification of wood ash (e.g., Mentzer and Quade, abundant, even if the main fueling material was wood. Second, an
2013 and references therein), or on morphological and quantita- important observation we made is the mixing that occurs at
tive analysis of phytoliths and dung spherulites (e.g., Katz et al., installation bases in relation to incomplete ash rake-out between
2007; Portillo et al., 2009). consecutive cooking events. A good example is the comparison
In this study we developed a method for the simultaneous between sample SUZ 155 originating from wood ash and sample
quantification of ash pseudomorphs and dung spherulites. The SUZ 167 originating from dung ash. Both contain ca. 7e8 million
method is accurate and reproducible (11% error for ash pseudo- phytoliths in 1 g ash, of which about 80% are from monocots and
morphs and 30% error for dung spherulites). Analysis of the eth- 10e15% from dicots (Table 5, Fig. 9). In the absence of quantitative
noarchaeological ash samples of known origin showed that in both spherulite and pseudomorph data, both samples will be interpreted
wood and dung ashes ash pseudomorphs and dung spherulites as originating from dung with some dicot contribution. However,
were present. The presence of wood ash pseudomorphs in herbi- when considering the PSR values of these samples, it is clear that
vore dung is explained as resulting from livestock consumption of SUZ 155 is predominantly wood ash with PSR value of 22, while
bark and dicot leaves as well as certain monocots that contain SUZ 167 is dung ash with PSR value of 0.01. Thus the PSR e phy-
calcium oxalate crystals (Prychid and Rudall, 1999). On the other tolith quantity combination is very powerful for identification of
hand, the presence of dung spherulites in wood ash can only be the predominant ash origin.
explained in relation to human behavior, i.e., that mixing between When applying the new method to ash of unknown origins in
wood and dung ashes occurs within the installations, despite the Sivasoy midden, we were able to assign with high certainty 4
(almost) daily removal of ashes. We observed that ash removal was out of the 7 samples analyzed to their specific fuel sources. Future
done haphazardly, without care for cleanliness but care for space research will be carried out in order to better understand the
for the next batch of fuel. Thus ash from former uses was often meaning of samples possessing PSR and phytolith concentrations
mixed with newly forming ashes, and in the case of tandirs these values that plot in the so-called “gray area”.
were manually mixed together by the cooks with a metal pole, as
part of the baking process and not due to the use of different fuels 4.5. Taphonomic considerations
together.
We note that Canti (1999) reported absence of spherulites in As noted above, the “gray area” between PSR values 1 and 10
dung of certain animals, including sheep. Theoretically, such indicates much mixing between wood and dung fuel/ashes. Can it
random examples of zero spherulites in dung will result in a PSR also result from taphonomic changes? Considering that these
value typical of wood ash, further leading to an erroneous identi- micro-remains are calcitic, they are soluble in acidic conditions. It is
fication of dung ash as wood ash. However, because dung cakes are possible that one of them is more soluble than the other. Future
produced from a mixture of sheep, goat and cow dung, it will be experiments are designed to test this taphonomic possibility. In
rare (if not impossible) to envision fueling a cooking installation addition, we postulate possible mixing of ashes with other types of
with dung that does not contain any spherulites. sediments e possibly clayey materials from the installations’ walls
Quantification of dung spherulites and ash pseudomorphs in and soil material in the Sivasoy dumped ashes. This possibility
ashes of known origin showed that spherulite-rich dung is more should also be studied in more detail in the future.
4346 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347

5. Summary and conclusions taken as a caution by phytolith specialists working at agricultural

and agro-pastoral contexts. We believe that for most sites pre-
In this study we have dealt with questions concerning the dating the Neolithic such caution is less significant (but see for
identification of cooking installations, reconstructing temperature example Miller et al., 2009).
ranges within them and identifying fuel materials that were used to
operate them. Archaeologists identify certain features to be cooking Acknowledgments
installations based on analogies of form and context of the
archaeological installations to ethnographic ones. These identifi- We thank Berdimuradov Amridin director of the Yahyo Gulomov
cations are usually supported by the discovery of ashes within in- Institute of Archaeology at Samarkand branch of the Academy of
stallations without considering alternative processes that might Science, Republic of Uzbekistan for permission to conduct the study
have led to the presence of non in-situ ash inside installations that and assistance during fieldwork. We thank Surat Kubayev and
were not originally related to fire use (such as ash discarded into Rahmatullah Rasulov for interpretation and field assistance, the
storage bins that went out of use). This study adds another tool for Beknazarov family that hosted us, Abdumanon Yerkulov who
identification of cooking installations, namely the identification of introduced us to the village of Tolly, and the Italo-Uzbek Project
heat-altered minerals (clay, calcite) on their inner walls. This can be “The Archaeological Map of the Middle Zeravshan Valley” directed
easily and efficiently done using FTIR analyses, which can also be by Maurizio Tosi, who introduced us to the study area in 2009.
done directly at the archaeological site using portable FTIR spec- Special thanks to Aren M. Maeir for his advice and support. The
trometers (Weiner, 2010). The study reported below contributes to study was supported by the Kimmel Center for Archaeological
the research of cooking installations worldwide, regardless of the Science (Weizmann Institute), and grants from the Israel Science
specific environment, specific installation shapes and specific Foundation, Focal Initiatives in Research in Science and Technology
subsistence practices of its inhabitants. This study focuses on the (Grant no. 527/09 to R. Shahack-Gross), the German Israel Foun-
physical and chemical attributes of materials used in a traditional dation for Research and Development (Grant # 1080-132.4/2009 to
society. The patterns that will be described below rely on universal A. Maeir and J. Maran), the F.I.R.S.T. (Bikura) Track of the Israel
laws in chemistry and physics of materials (a materials science Science Foundation (Grant # 32/11 to A. Maeir, E. Weiss and L.
approach). For example, the way wood or dung burn in Uzbekistan Horwitz) and the Kuchinsky Fund From the Martin (Szusz)
is just the same as they will burn almost anywhere else around the Department of Land of Israel Studies and Archaeology, Bar-Ilan
world (i.e., except for high altitude areas that are oxygen deficient). University, (to S. Gur-Arieh).
Similarly, the way that heat within an Uzbek installation affects the
chemistry of its walls is expected to be similar in cooking in-
stallations elsewhere around the world. Therefore, the patterns References
reported in this study can be applied to most other places on the
Albert, R.M., Bar-Yosef, O., Meignen, L., Weiner, S., 2003. Quantitative phytolith
globe. study of hearths from the Natufian and Middle Palaeolithic levels of Hayonim
Our results show that the temperatures recorded by clay alter- Cave (Galilee, Israel). Journal of Archaeological Science 30, 461e480.
ation on tandir inner walls (700e900  C) reflect the initial heating Albert, R.M., Lavi, O., Estroff, L., Weiner, S., Tsatskin, A., Ronen, A., Lev-Yadun, S.,
1999. Mode of occupation of Tabun Cave, Mt Carmel, Israel during the Mous-
of the oven prior to actual baking, the latter done at 250e300  C. terian Period: a study of the sediments and phytoliths. Journal of Archaeological
The temperatures recorded on ochock inner walls (500e600  C) Science 26, 1249e1260.
better match actual cooking temperatures (700  C and lower). We Albert, R.M., Shahack-Gross, R., Cabanes, D., Gilboa, A., Lev-Yadun, S., Portillo, M.,
Sharon, I., Boaretto, E., Weiner, S., 2008. Phytolith-rich layers from the Late
also found that wood and dung burn alike during the operation of Bronze and Iron Ages at Tel Dor (Israel): mode of formation and archaeological
the studied installations. significance. Journal of Archaeological Science 35, 57e75.
The method developed for the quantification of ash pseudo- Albert, R.M., Weiner, S., Bar-Yosef, O., Meignen, L., 2000. Phytoliths in the Middle
Palaeolithic deposits of Kebara Cave, Mt Carmel, Israel: study of the plant
morphs and dung spherulites was proved to be accurate and
materials used for fuel and other purposes. Journal of Archaeological Science 27,
reproducible. We introduced the plot of PSR vs. phytolith concen- 931e947.
tration in ashes and determined experimentally that almost pure Amouretti, M.C., 1986. Le pain et l’huile dans la Grèce antique. Les Belles Lettres,
Paris.
wood ash is characterized by PSR values above 10 and phytolith
Avitsur, S., 1971. Baking Installations and Implements. Machon Avshalom, Tel-Aviv.
concentrations below 1 million in 1 g ash. Pure dung ash is char- Berna, F., Behar, A., Shahack-Gross, R., Berg, J., Boaretto, E., Gilboa, A., Sharon, I.,
acterized by PSR values lower than 1 and more than 20 million Shalev, S., Shilstein, S., Yahalom-Mack, N., Zorn, J.R., Weiner, S., 2007. Sediments
phytoliths in 1 g ash. The ethnographic ashes studied did not al- exposed to high temperatures: reconstructing pyrotechnological processes in
Late Bronze and Iron Age Strata at Tel Dor (Israel). Journal of Archaeological
ways fall within these distinctive ranges, indicating that mixing Science 34, 358e373.
processes between fuel sources occur. Ashes with PSR values be- Binford, L.R., 1998. Hearth and home: the spatial analysis of ethnographically
tween 1 and 10 and phytolith concentrations below 20 million in documented rock shelter occupations as a template for distinguishing between
human and hominid use of sheltered space. In: Conard, N.J., Wendorf, F. (Eds.),
1 g ash fall into a “gray area” were intensive mixing occurred and Middle Paleolithic and Middle Stone Age Settlement Systems. A.B.A.C.O. Edi-
thus interpretation as to the dominant fuel source is highly zioni, Forli, pp. 229e240.
uncertain. Braadbaart, F., Poole, I., Huisman, H.D.J., van Os, B., 2012. Fuel, fire and heat: an
experimental approach to highlight the potential of studying ash and char
Mixing processes became evident primarily by the presence of remains from archaeological contexts. Journal of Archaeological Science 39,
dung spherulites in ash from installations fueled by wood only. We 836e847.
explain this phenomenon as primarily reflecting human behavior e Brochier, J.E., 1983. Combustion et parcage des herbivores domestiques. Le point
de vue du sedimentologue. Bulletin de la Societe Prehistorique Francaise 80,
either by consciously using different fuel types in the same burning 143e145.
event, or by unconsciously mixing the different fuel ashes in rela- Cabanes, D., Weiner, S., Shahack-Gross, R., 2011. Stability of phytoliths in the
tion to rake-out activities. archaeological record: a dissolution study of modern and fossil phytoliths.
Journal of Archaeological Science 38, 2480e2490.
Lastly, we noted that phytolith morphotype analysis was sur-
Cabanes, D., Gadot, Y., Cabanes, M., Finkelstein, I., Weiner, S., Shahack-Gross, R.,
prisingly not useful in distinguishing between fuel sources. The 2012. Human impact around settlement sites: a phytolith and mineralogical
reasons are probably (a) the original variability in livestock diet and study for assessing site boundaries, phytolith preservation, and implications for
thus in certain cases resemblance of dung phytoliths to dicot- spatial reconstructions using plant remains. Journal of Archaeological Science
39, 2697e2705.
dominated wood ash, and (b) the mixing that occurs within Canti, M.G., 1997. An investigation of microscopic calcareous spherulites from
installation bottoms. We suggest that this observation should be herbivore dungs. Journal of Archaeological Science 24, 219e231.
 S. Gur-Arieh et al. / Journal of Archaeological Science 40 (2013) 4331e4347 4347

Canti, M.G., 1999. The production and preservation of faecal spherulites: animals, Miller, N.F., Marston, J.M., 2012. Archaeological fuel remains as indicators of ancient
environment and taphonomy. Journal of Archaeological Science 26, 251e258. West Asian agropastoral and land-use systems. Journal of Arid Environments
Canti, M.G., 2003. Aspects of the chemical and microscopic characteristics of plant 86, 97e103.
ashes found in archaeological soils. Catena 54, 339e361. Miller, N.F., Zeder, M.A., Arter, S.R., 2009. From food and fuel to farms and flocks.
Carmody, R.N., Wrangham, R.W., 2009. The energetic significance of cooking. Current Anthropology 50, 915e924.
Journal of Human Evolution 57, 379e391. Mulder-Heymans, N., 2002. Archaeology, experimental archaeology and ethno-
Cowlishaw, S.J., Alder, F.E., 1960. The grazing preferences of cattle and sheep. Journal archaeology on bread ovens in Syria. Civilisations 49, 197e221.
of Agricultural Science 54, 257e265. Mulholland, S.C., Rapp, G.J., 1992. Phytolith systematics: an introduction. In: George
Crawford, H.E.W., 1981. Some fire installations from Abu Salabikh. Paleorient 7, Rapp, J., Mulholland, S.C. (Eds.), Phytolith Systematics: Emerging Issues. Plenum
105e114. Press, New York - London, pp. 1e13.
di Cugno, S., Mantellini, S., Berdimuradov, A., 2013. Traditional techniques, re- Organ, C., Nunn, C.L., Machanda, Z., Wrangham, R.W., 2011. Phylogenetic rate shifts
sources exploitation and archaeological record: an ethnographical study from in feeding time during the evolution of Homo. Proceedings of the National
the oasis of Samarkand (Uzbekistan). In: Lugli, F., Stoppiello, A.A., Biagetti, S. Academy of Sciences 108, 14555e14559.
(Eds.), Ethnoarchaeology: Current Research and Field Methods, Conference Parker, B., 2011. Bread ovens, social networks and gendered space: an ethno-
Proceedings, Rome, Italy, 13the14th May 2010. Archaeopress. archaeological study of Tandir ovens in Southeastern Anatolia. American An-
Dalman, G., 1964. Arbeit und Sitte in Palastina. G. Olms, Hildesheim. tiquity 76, 603e627.
Davis, M., 1995. Fuel choice in rural communities. Energy for Sustainable Devel- Parker, B., Üzel, M., 2007. The tradition of tandır cooking in Southeastern Anatolia:
opment 2, 45e48. an ethnoarchaeological perspective. In: Takaoglu, T. (Ed.), Ethnoarchaeological
Fisher, J.W., Strickland, H.C., 1989. Ethnoarchaeology among the Efe pygmies, Zaire: Investigations in Rural Anatolia. Ege Yayinlari, Istanbul.
spatial organization of campsites. American Journal of Physical Anthropology Payziyeva, S., Paiziev, A., 2012. Food security in Uzbekistan. In: Alpas, H., Smith, M.,
78, 473e484. Kulmyrzeav, A. (Eds.), Strategies for Achieving Food Security in Central Asia.
Fisher, J.W., Strickland, H.C., 1991. Dwellings and Fireplaces: Keys to Efe Pygmy Springer, Netherlands, Dordrecht, pp. 43e56.
Campsite Structure, Ethnoarchaeological Approaches to Mobile Campsites: Picornell Gelabert, L., Asouti, E., Martí, E.A., 2011. The ethnoarchaeology of firewood
Hunter-gatherer and Pastoralist Case Studies, pp. 215e236. management in the Fang villages of Equatorial Guinea, Central Africa: impli-
Franceschi, V.R., Nakata, P.A., 2005. Calcium Oxalate in Plants: Formation and cations for the interpretation of wood fuel remains from archaeological sites.
Function, Annual Review of Plant Biology. Annual Reviews, Palo Alto, pp. 41e71. Journal of Anthropological Archaeology 30, 375e384.
Frankel, R., 2011. Cooking installations in the Talmudic literature. Katedra 139, 79e Pierce, C., Adams, K.R., Stewart, J.D., 1998. Determining the fuel constituents of
114 (in Hebrew). ancient hearth ash via ICP-AES analysis. Journal of Archaeological Science 25,
Gintzburger, G., Toderich, K.N., Mardonov, B.K., Mahmudov, M., 2003. Rangelands of 493e503.
the Arid and Semi-arid Zones in Uzbekistan. Cirad, Montpellier. Piperno, D.R., 1988. Phytolith Analysis: an Archeological and Geological Perspective.
Greenhouse, R., Gasser, R.E., Gish, J.W., 1981. Cholla Bud roasting pits: an ethno- Academic Press, San Diego.
archaeological example. Kiva 46, 227e242. Piperno, D.R., 2006. Phytoliths: a Comprehensive Guide for Archaeologists and
Gur-Arieh, S., Boaretto, E., Maeir, A., Shahack-Gross, R., 2012. Formation processes in Paleoecologists. AltaMira Press, Lanham, MD.
Philistine hearths from Tell es-Safi/Gath (Israel): an experimental approach. Portillo, M., Albert, R.M., Henry, D.O., 2009. Domestic activities and spatial distri-
Journal of Field Archaeology 37, 121e131. bution in Ain Abu Nukhayla (Wadi Rum, Southern Jordan): the use of phytoliths
Gwavuya, S.G., Abele, S., Barfuss, I., Zeller, M., Müller, J., 2012. Household energy and spherulites studies. Quaternary International 193, 174e183.
economics in rural Ethiopia: a cost-benefit analysis of biogas energy. Renewable Prychid, C.J., Rudall, P.J., 1999. Calcium oxalate crystals in monocotyledons: a review
Energy 48, 202e209. of their structure and systematics. Annals of Botany 84, 725e739.
Hoeck, T., Droux, R., Breu, T., Hurni, H., Maselli, D., 2007. Rural energy consumption Reddy, S.N., 1998. Fueling the hearths in India: the role of dung in paleo-
and land degradation in a post-Soviet setting- an example from the west Pamir ethnobotanical interpretation. Paléorient 24, 61e69.
mountains in Tajikistan. Energy for Sustainable Development 11, 48e57. Regev, L., Poduska, K.M., Addadi, L., Weiner, S., Boaretto, E., 2010. Distinguishing
Hosier, R.H., Dowd, J., 1987. Household fuel choice in Zimbabwe: an empirical test of between calcites formed by different mechanisms using infrared spectrometry:
the energy ladder hypothesis. Resources and Energy 9, 347e361. archaeological applications. Journal of Archaeological Science 37, 3022e3029.
Joon, V., Chandra, A., Bhattacharya, M., 2009. Household energy consumption Roebroeks, W., Villa, P., 2011. On the earliest evidence for habitual use of fire in
pattern and socio-cultural dimensions associated with it: a case study of rural Europe. Proceedings of the National Academy of Sciences 108, 5209e5214.
Haryana, India. Biomass and Bioenergy 33, 1509e1512. Samuel, D., 1989. Their staff of life: initial investigations on ancient Egyptian bread
Kandiyoti, D., 1999. How to Get It Wrong in Rural Uzbekistan: an Ethnographic baking. In: Kemp Barry, J.K. (Ed.), 1989. Amarna Reports V. Egypt Exploration
Critique of Household Survey Categories (Discussion Paper No. 106). UN Society, London, pp. 253e290.
Research Institute for Social Development, Geneva. Shahack-Gross, R., 2011. Herbivorous livestock dung: formation, taphonomy,
Katz, O., Gilead, I., Bar, P., Shahack-Gross, R., 2007. Chalcolithic agricultural life at methods for identification, and archaeological significance. Journal of Archae-
Grar, Northern Negev, Israel: Dry farmed cereals and dung-fueled hearths. ological Science 38, 205e218.
Paléorient 33, 101e116. Shahack-Gross, R., Ayalon, A., 2013. Stable carbon and oxygen isotopic compositions
Katz, O., Cabanes, D., Weiner, S., Maeir, A.M., Boaretto, E., Shahack-Gross, R., 2010. of wood ash: an experimental study with archaeological implications. Journal of
Rapid phytolith extraction for analysis of phytolith concentrations and assem- Archaeological Science 40, 570e578.
blages during an excavation: an application at Tell es-Safi/Gath, Israel. Journal of Shahack-Gross, R., Finkelstein, I., 2008. Subsistence practices in an arid environ-
Archaeological Science 37, 1557e1563. ment: a geoarchaeological investigation in an Iron Age site, the Negev High-
Kramer, C., 1982. Village Ethnoarchaeology: Rural Iran in Archaeological Perspec- lands, Israel. Journal of Archaeological Science 35, 965e982.
tive. Academic Press, New York. Shahack-Gross, R., Marshall, F., Ryan, K., Weiner, S., 2004. Reconstruction of spatial
Lancelotti, C., Madella, M., 2012. The ‘invisible’ product: developing markers for organization in abandoned Maasai settlements: implications for site structure
identifying dung in archaeological contexts. Journal of Archaeological Science in the pastoral Neolithic of East Africa. Journal of Archaeological Science 31,
39, 953e963. 1395e1411.
Lyons, D., D’Andrea, A.C., 2003. Griddles, ovens, and agricultural origins: an eth- Simpson, I.A., Vésteinsson, O., Adderley, W.P., McGovern, T.H., 2003. Fuel resource
noarchaeological study of bread baking in highland Ethiopia. American An- utilisation in landscapes of settlement. Journal of Archaeological Science 30,
thropologist 105, 515e530. 1401e1420.
Madella, M., Alexandre, A., Ball, T., 2005. International code for phytolith nomen- Skipp, G.L., Brownfield, I., 1993. Improved Density Gradient Separation Techniques
clature 1.0. Annals of Botany 96, 253e260. Using Sodium Polytungstate and a Comparison to the Use of Other Heavy
Matthews, W., 2010. Geoarchaeology and taphonomy of plant remains and micro- Liquids. U.S. Geological Survey, Denver, CO.
archaeological residues in early urban environments in the Ancient Near East. Thoms, A.V., 2008. The fire stones carry: ethnographic records and archaeological
Quaternary International 214, 98e113. expectations for hot-rock cookery in western North America. Journal of
McQuitty, A., 1984. An ethnographic and archaeological study of clay ovens in Anthropological Archaeology 27, 443e460.
Jordan. Annual of the Department of Antiquities of Jordan 28, 259e268. Weiner, S., 2010. Microarchaeology. Beyond the Visible Archaeological Record.
McQuitty, A., 1993. Ovens in town and country. Berytus: Archaeological Studies 42, Cambridge University Press, New York.
53e76. Weinstein, M., 1973. Household structures and activities. Anatolian Studies 23,
Mentzer, S.M., Quade, J., 2013. Compositional and isotopic analytical methods in 271e276.
archaeological micromorphology. Geoarchaeology 28, 87e97. Wollstonecroft, M.l., 2011. Investigating the role of food processing in human
Meyer, J.H., Lofgreen, G.P., Hull, J.L., 1957. Selective grazing by sheep and cattle. evolution: a niche construction approach. Archaeological and Anthropological
Journal of Animal Science 16, 766e772. Sciences 3, 141e150.
Miller, N.F., 1982. Economy and Environment of Malyan, a Third Millennium City in Wrangham, R., 2009. Catching Fire: How Cooking Made Us Human. Basic books,
Southern Iran. In: Anthropology. The University of Michigan, Ann Arbor. New York.
Miller, N.F., 1984a. The interpretation of some carbonized cereal remains as rem- Zapata Peña, L., Peña-Chocarro, L., Esté, J.J.I., Urquijo, J.G., 2004. Ethnoarchaeology in
nants of dung cake fuel. Bulletin on Sumerian Agriculture 1, 45e47. the Moroccan Jebala (Western Rif): wood and dung as fuel. In: Neumann, K.,
Miller, N.F., 1984b. The use of dung as fuel: an ethnographic example and an Butler, A., Kahlheber, S. (Eds.), Food, Fuels and Fields -Progress in African
archaeological application. Paléorient 10, 71e79. Archaeobotany. Heinrich - Barth - Institut, Köln, pp. 163e175.