Insights from a tribological analysis of the tribulum

Abstract

Since the 1980s, “strange” microwear traces were found to occur on flint blades from sites in the Near East from the late Neolithic and occurring in great abundance by the Early Bronze Age. Although these were considered by archaeologists to be sickles because they had visible gloss on their edges, their use-traces could not be reproduced in harvesting experiments carried out in the field. Subsequently, several lines of evidence were used to study the blades, including not only direct observation of microscopic wear traces, but also Near Eastern cuneiform texts from the third and second millennium BC describing agricultural instruments and analogy with ethnographic and experimental reference material. We found that these tools and their traces best matched traces on flint used to arm the underside of a tribulum (threshing sledge) for threshing grain and cutting straw. We built a replica of the tribulum described in cuneiform texts from the Bronze Age, using copies of the Bronze Age blades, and used this instrument in experiments.

Full text

Insights from a tribological analysis of the tribulum
P.C. Anderson
a,
*, J.-M. Georges
b
, R. Vargiolu
b
, H. Zahouani
b
a
CEPAM UMR 6130 and GDR 2517 CNRS, 250 rue A. Einstein, Sophia Antipolis, 06560 Valbonne, Alpes-Maritimes, France
b
LTDS UMR 5513 CNRS, Ecole Centrale de Lyon 69131, Ecully, France
Received 30 June 2005; received in revised form 15 February 2006; accepted 20 February 2006
Abstract
Since the 1980s, ‘‘strange’’ microwear traces were found to occur on flint blades from sites in the Near East from the late Neolithic and
occurring in great abundance by the Early Bronze Age. Although these were considered by archaeologists to be sickles because they had visible
gloss on their edges, their use-traces could not be reproduced in harvesting experiments carried out in the field. Subsequently, several lines of
evidence were used to study the blades, including not only direct observation of microscopic wear traces, but also Near Eastern cuneiform texts
from the third and second millennium BC describing agricultural instruments and analogy with ethnographic and experimental reference
material. We found that these tools and their traces best matched traces on flint used to arm the underside of a tribulum (threshing sledge)
for threshing grain and cutting straw. We built a replica of the tribulum described in cuneiform texts from the Bronze Age, using copies of
the Bronze Age blades, and used this instrument in experiments.
The publication of S. A. Semenov’s work concerning traceology of ancient implements [S.A. Semenov, Prehistoric Technology (M.W.
Thompson, Trans.), Cory, Adams and Mackay, London, 1964] awakened considerable interest in the possibilities of directly inferring tool func-
tion from the microscopic traces left by use on ancient implements. However, the goal and purpose of microwear studies is to reconstruct, as
completely as possible, the economic activities of prehistoric groups, requiring a methodology that embraces all aspects relevant to the inter-
pretation of microwear traces [L. Keeley, The methodology of microwear analysis: a comment on Nance. American Antiquity 39 (1974) 126e128].
In this study, we apply the science of tribology, which studies the friction and the wear of solid bodies in contact, to the threshing sledge and
its blades.
Tribological analysis of the tribulum has thus far provided new explanations for the formation of traces of wear on flint inserts in threshing
sledges, while also revealing features on a far smaller scale, involving the role of a film deposit. We determine the mechanisms by which the
tribulum threshes grain, and, particularly, fine-cuts straw from the sheaves of cereal laid on the threshing floor. The superior functioning of
the tribulum compared to other methods of threshing and cutting straw is due to its great control of the rheologyethe deformation and
floweof the straw layer on the threshing floor. The study lead to insights as to the mastery, of what may be considered today to be complex
tribological principles, in the Bronze Age and probably beginning in the Late Neolithic [P.C. Anderson, Observations on the threshing sledge
and its products in ancient and present-day Mesopotamia, in: P.C. Anderson, L.S. Cummings, T.K. Schippers, B. Simonel, (Eds.), Le traitement
des re
´coltes: un regard sur la diversite
´du Ne
´olithique au pre
´sent, ADPCA, Antibes, 2003, pp. 417e438].
Ó2006 Elsevier Ltd. All rights reserved.
Keywords: Tribulum; Tribology; Rheology; Threshing sledge; Microwear; Phytoliths; Bronze Age; Middle East; Gloss; Cereal threshing; Flint blades; Agriculture;
Experiments in archaeology; Ethnoarchaeology
1. Introduction
The threshing sledge, which we shall refer to here by its
Latin name, tribulum, first used in writings of the second
century BC [26] and best known from Pliny [20], was an agri-
cultural tool common all over the Mediterranean until a few de-
cades ago [6,16,18,27] and according to our direct observation,
is still in use in limited areas of the Middle East [2], Spain and
Tunisia. The tribulum is made up of assembled planks or staves,
which make up a board or a raft that is studded on its underside
with stone or metal inserts, which usually have cutting edges.
* Corresponding author. Tel.: þ33 493 95 41 54; fax: þ33 493 65 29 05.
E-mail address: anderson@cepam.cnrs.fr (P.C. Anderson).
0305-4403/$ - see front matter Ó2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2006.02.011
Journal of Archaeological Science 33 (2006) 1559e1568
http://www.elsevier.com/locate/jas

Working on sheaves of harvested plants laid on a threshing
floor, the tribulum simultaneously and efficiently removes the
grain from the stem and seed head without damage or crush-
ing, while separating the stem into fine-cut segments. Tribulae
in use have been frequently documented in religious iconogra-
phy showing the annual agricultural cycle, for example, in
a Spanish Medieval frieze [19], and in historical [18] and
ethno archaeological [5,6,16,27] studies from all over the
Mediterranean, Balkan and Black Sea areas. Concerning the
inserts pounded or glued into the underside of these of thresh-
ing boards or threshing rafts, there is great variability in their
size [6,16], nature (flint, basalt, iron) [2,16,18], number and
pattern of insertion (usually checkerboard-fashion or random).
The dimension and type of surface of the threshing floor are
also variable [2,16,27]. Nonetheless, certain general threshing
activity ‘‘principles’’ hold true over time to the present. The
sledge is dragged by draft animals around in a circle, as the
instrument slides over harvested sheaves of cereals or of
pulses, such as lentils, chickpea, broad bean or vetch that
have been deposited on the threshing floor (Fig. 1).
The material is turned on the threshing floor using winnowing
forks, and once the grain is separated from the stalks and the
straw is cut into sufficiently small fragments, all the material
is separated into distinct piles of the various components by
size and weight. This separation occurs first by wind winnowing,
lifting the materials into the wind using winnowing forks,
shovels or basketry, causing materials to fall into separate heaps,
then these fractions are put through sieves or riddles in order to
better isolate the grain. These various products are then, accord-
ing to their size and their nature, destined for many different
uses, including human or animal consumption, tempering of
ceramics and mud-brick, or as fuel. We found that these uses
of grain and straw still persist [2] when we observed the tribulum
being used to process grain and pulses for use by families and
their animals in 2001 and 2004 in villages in S. Syria (Jebel
Hauran region), and in 2005 near Be
´ja, in N. Tunisia (Anti-Atlas
region). The tribulum’s great antiquity is attested to by numer-
ous literary references including the Bible, and in particular,
cuneiform texts from the second and third millennia BC
(2500e2200 BC) referring clearly to its structure and number
of inserts, called ‘‘teeth’’ (in inventories), and to its use (in the
form of prayers, as instructions of God the Farmer to his disci-
ples) [11]. Its oldest identified depiction to date is on a rolled-
out imprint of a cylinder seal from fourth millennium BC
Proto-urban levels, at Arslantepe, in Turkey [9]. Here, a person
is shown sitting on a chair or possibly a throne, on a tribulum, (as
was done in recent times in Spain and Cyprus) and cutting ele-
ments are shown protruding beneath the instrument (Fig. 2). Al-
though possibly depicting a ritual scene, this cylinder seal shows
a tribulum with an up curved front, surrounded by workers with
threshing forks and an animal driver, resembling its use today.
The particular situation evaluated in this article involves
a study of the threshing sledge referred to in Sumerian cunei-
form texts from the third and second millennia B.C. as having
existed during the Bronze Age of Mesopotamia. We have re-
constructed this threshing sledge, tested it and analysed its
functioning principles using tribology. Both descriptions in cu-
neiform texts from the third and second millennium BC, [10]
and our observations of use-traces seen on numerous flint
blade inserts [3] were used in the reconstruction of the Bronze
Age tribulum. A small number of microwear analysts [3,4]
have studied these special blades with particular microscopic
traces of wear, which lead to the hypothesis that these blades
constituted the ‘‘teeth’’ of the threshing sledge of the third mil-
lennium BC. The usual method of observation of wear traces
on such blades is optical microscopy at 100e200, and occa-
sionally the scanning electron microscopy [17,23]. Prior to
this, archaeologists had generally classified these blades as be-
ing inserts for sickles for harvesting grain, because of the vis-
ible gloss from use on their edges and sides. The object of this
paper is not to review debate concerning mechanisms of for-
mation of such traces [i.e. 1]. However, microwear analysts
(traceologists) experimenting with sickles found, that the mi-
crowear traces that were produced on the tools when they har-
vested various kinds of grain in different locations, were not
the same kind of traces as those seen on the ‘‘special blades’’
described above [2e6,16]. Conversely, examination of flints
both from recently used threshing sledges and of new blades
used in experimental sledges constructed according to the an-
cient documents, shows a different kind of microwear trace
formation, and this pattern closely resembles the ‘‘strange’’
microwear traces seen on the ancient blades discussed here.
Although traces like those on the archaeological blades began
to form after a minimum of 5e10 h of use, the extent of devel-
opment of traces on many of the Bronze Age blades and on
many blades from recent tribulae shows that they were used
for far longer than this, indeed over generations, as interviews
with recent users attest [2e4].
The distinctive pattern of microwear traces on tribulum
blades particularly concerns the distribution of the gloss, and
the orientation and the appearance of what were thought to
be abrasive phenomena, when compared with sickle use: Al-
though sickles develop tiny dotted striae during use, visible
as though superimposed on the gloss, (presumably caused by
soil grains on stems rolling over the flint surface as the tool
Fig. 1. Recent use of a threshing sledge in the Jebel Hauran region, Southern
Syria.
1560 P.C. Anderson et al. / Journal of Archaeological Science 33 (2006) 1559e1568

cuts), threshing sledge blades, on the contrary do not develop
these microscopic features, rather they show randomly ori-
ented lines and particularly wide comet-shaped grooves or ir-
regular darker areas that give the impression that parts of the
flint surface have been torn away as a result of use [3,4,25,
i.e. Fig. 3b]. We asked the following question: should this dis-
tinction hold true in a tribological sense?
Most of these blades determined to have been used as trib-
ulum inserts are large ‘‘Canaanean blade’’ segments from
Early Bronze Age sites in Syria, as well as Iraq [2e5].We
have verified that, at least these blades were part of an exten-
sive trade network in the third millennium BC involving the
tribulum. Indeed the blades were not made in the sites where
they were used, but were imported from elsewhere, made by
specialised artisans who also broke them into straight, stan-
dardized segments ready for use [3]. Possibly they were traded
already inserted in the tribulum structure, although we do not
have means of verifying this point. Agricultural products pro-
duced by the tribulum also seem to have been involved in this
trade network in Northern Mesopotamia [3,4]. This network
was probably in operation already during the prior Uruk period
(4th millennium BC) [4], and smaller numbers of other blades
with these special tribulum-type wear traces have been identi-
fied in earlier sites, beginning in the late Neolithic, in this
same region [2].
We have also verified that the chopped straw produced by
the tribulum on the threshing floor, was used in the fourth
and third millennia in a manner similar to recent times: to tem-
per mud-brick and pottery, for animals, and as fuel [2e4]. This
observation was made because microscopic remains of amor-
phous silica cell imprints, phytoliths [14]econtained in the
finely-cut plant stems and glumes surrounding the graine
have been found in the remains of dung fuel, granary and
silo residues, threshing floor surfaces and temper in mud-
brick, in the same Bronze Age sites as those producing the
blades with traces with tribulum-type wear [4]. These remains
show that the original straw was cut in such a way as to sug-
gest the passage of the blades of a tribulum (see below), as op-
posed to breaking, cutting, chopping, or natural crushing. This
pattern is also seen in material processed on present-day
threshing floors and in our experiments with the tribulum
[2e4;Figs. 4, 10].
The cuneiform clay tablet texts, describing agricultural in-
struments roughly contemporaneous with these sites, provide
a number of references as to the use and the construction of
the tribulum [10,11,21]. Although its use is like that today,
Fig. 2. Threshing sledge image on a cylinder seal stamp imprint from Arslantepe, Turkey, fourth millenium BC (from [9]: p. 67, fig. 16.1).
Fig. 3. (a) A typical fragment of a Canaanean blade with microwear patterns
attributed to use in a threshing sledge. Three zones are shown: zone I is gen-
erally covered by an adherent layer of bitumen; zone II is the surface affected
by the straw flow; zone III is the irregular cutting edge. (From [3]: 266
fig. 12b). (b) Photomicrograph of zone III, observed using a reflected-light
microscope at 100magnification.
1561P.C. Anderson et al. / Journal of Archaeological Science 33 (2006) 1559e1568

its structure differs somewhat from the most common forms
seen in the recent past: The Sumerian poem Georgica [21],
or the Instructions of God, a song dating from the second
millennium BC but undoubtedly originating in the third
millennium [11] gives divine advice that, in order to thresh
cereal grain, it is necessary to prepare the tribulum, consisting
of wooden staves lashed together using leather straps, firmly
attaching the ‘‘teeth’’ with leather straps and bitumen (tar).
This tribulum was, therefore, a raft-like structure, with ‘‘teeth’’
glued into its underside into the interstices between staves
using tar or bitumen [11]. An Akkadian cuneiform tablet
(2350e2200 BC) with references to a tribulum with ‘‘teeth’’
for example, mentions 40, then 80 ‘‘teeth’’ [10], which is
few if compared with modern tribulae that usually have small
inserts [6,16,27]. However, this low number of ‘‘teeth’’
would coincide nicely with the very large Canaanean blade
segments found in the archaeological sites [3,5], and with
the general size of the tribulum drawn on the cylinder seal
in Fig. 2 [9].
A number of questions have arisen in the course of seven
years of experiments in relation to this instrument, its function,
and what mechanisms come into play as it works [25]. What
insights can this tribulum provide concerning the relationship
between technique and social logic? Is iconography concern-
ing the threshing sledge of the third and fourth millennia BC
reliable? Why does this instrument appear to be so efficient,
as compared to manual methods for processing grain and
straw, or animal trampling? The Canaanean blades of the third
millennium were generally large and standardized, were made
using optimal techniques by specialists in central sites, and
were distributed over a large region [4]. Why do these partic-
ular blades appear to have been so important to the social and
economic networks of the time? Later forms of the tribulum
use smaller but more numerous stone or metal inserts, in ran-
dom or checkerboard patterns [16,27], but the technological
know-how for making large stone blades had been lost [6].
Therefore the aim of this study is, first, to investigate the de-
sign principles of the tribulum and understand how it works,
and second, to deduce the methodology for observation of
the Canaanean blade segments used in the instrument.
2. A physical presentation
As a component essential to the use of the threshing sledge,
the threshing floor (A) is prepared, and a layer of plant sheaves
(C) is laid down so as to cover the floor over a circular area S
F
,
to a thickness h
f
of about 20 cm. Thus, if S
F
is assumed to be
20 m
2
, then the cereal volume is 4 m
3
, corresponding to 60 kg
in the case of barley or wheat, for example. The cereals, either
harvested by cutting close to the ground or by pulling up by
hand, leaving long stems, are randomly distributed. In the
working area of the threshing floors, which measure Dz8e
10 m in diameter, one or two traction animal(s), either oxen
or equids, pull the sledge (B) at constant low speed
(Vz0.5e1 m/s) in a circular trajectory.
For a tribologist, A (the threshing floor), B (the tribulum)
and C (the plant sheaves) are the constitutive elements, given
in Fig. 5, which determine the behaviour and the efficiency of
the sliding and cutting processes. It is important to first de-
scribe the three bodies or entities functioning together [8,12].
The threshing floor A can be made of beaten and hardened
clay, of paving stones or of a plain earth surface with grass cut
as short as possible. It is a relatively hard surface, whose
roughness height h
roughness
is small (<2 cm) in comparison
with h
inlet
(see below).
The tribulum B (Fig. 5) forms a rectangular structure with
typical dimensions of length 1.2 <l<2.0 m and width
0.5 <w<0.9 m, which gives a minimum area S
T
¼0.6 m
2
and a maximum of 1.8 m
2
. Generally built with pieces of
wood (planks or of staves from tree branches as is described
for the Bronze Age), the weight is approximately 25 kg or
greater. During the cutting process, a person stands or sits
(with a weight of roughly 75 kg) on the sledge. Therefore
a normal force (W¼1000 N) is applied on the tribulum, which
compresses the straw layer. This force Wsubmits the contact
zone containing the straw layer between the sledge and the
ground to a low-pressure regime. The pressure p¼W/S
T
is
the ratio of the force normal to the surface (unit: N) with
the surface area (unit: m
2
), so the unit of pis N/m
2
¼Pa
(Pascal). Here, the pressure is 555 <p<1670 Pa. This pressure
value is low and not enough to irreversibly deform the straw
layer. But ppermits the straw to flow, if the layer is sheared.
A more detailed description of the tribulum is shown in
Fig. 5. Its first part is a relatively smooth inclined plane or
curved piece (length 0.4 <l
inlet
<0.8 m), its second part,
a flat plane (length 0.8 <l
cutting
<1.2 m) is covered with in-
serted blades of flint (or more recently, other stones or metal).
There is frequently a third section, a flat plane without blades
(length 0.1 <l
outlet
<0.2 m). The inserted blades are usually ar-
ranged in 10e15 rows, with a usual distance of dwz5 cm be-
tween rows (see Fig. 4, top). In the present study, the blades have
a typical length of 3 <l
blade
<7 cm and their functional height
h
blade
, which extends beyond the tribulum plane, is 1 <h
blade
<
2 cm. In the cuneiform inventories, from 50 to 80 ‘teeth’ are
mentioned as being needed to arm threshing sledges [10].
The role played by the two areas of the sledge can be better
understood after a description of the mechanical behaviour of
the layer of plant sheaves (C). This layer, with a thickness of
Fig. 4. Stems of barley straw collected from a threshing floor after threshing
the crop using the tribulum. Note visible indentation marks (short arrows)
due to the impacts of the blades against the stems.
1562 P.C. Anderson et al. / Journal of Archaeological Science 33 (2006) 1559e1568

20 cm overall, consists of randomly oriented stems. Each stem
has a cylindrical shape, and is approximately 40 cm long and
0.4 cm in diameter (Fig. 4). So, the thickness of the layer can
consist of a pile of 20/0.4 ¼50 stems. This relatively great
number of components suggests that an overall description
of the material constituting the straw in the plant layer is use-
ful. In this description, the layer (C) is considered as a contin-
uum of material that flows between the tribulum and the
threshing floor that has particular mechanical characteristics,
in particular viscosity and elasticity, which makes it a visco-
elastic material [13].
2.1. The viscosity concept [7,8,24]
It is important to understand the concept of viscosity, be-
cause viscosity is central to the performance of lubricated ma-
chinery. The simplest definition of viscosity is resistance to
flow. Sir Isaac Newton defined it as ‘‘the resistance that arises
from lack of slipperiness in a fluid.’’ Cold maple syrup is thick
and not slippery, but cold water is thin and slippery.
When scientists talk about fluids, concepts of shear stress t
and shear rate _
gare used. In order to visualize these two con-
cepts, consider two rigid and parallel plates separated by a fluid
layer of thickness h. (i) The tangential force Tis applied at the
edge of the top plate to move the plate at the speed V. The
shear stress tis the ratio of this tangential force T(unit N)
with the surface area (unit m
2
), so the shear stress unit is N/
m
2
¼Pa (Pascal). (ii) The concept of shear rate can be visual-
ised by assuming that between the two plates, the movement
of the fluid is laminar, like the flow of cards in a card game.
The shear rate _
gis related to the speed with which the layers
of fluid between the plates move. The top layers, the ones clos-
est to the moving plate, move the fastest and the layers nearest
to the stationary plate move the slowest. Thus, there is a veloc-
ity gradient from fastest to slowest. This gradient is the shear
rate _
g¼V=h. The unit of _
gis an inverse of time so equal to 1/s.
If the shear stress tand shear rate _
gare plotted, for a perfect
fluid called a Newtonian fluid, a straight line starting from zero
is found. The slope of that line is the viscosity hfor that fluid
at a given temperature. Mathematically, viscosity is the ratio
of shear stress to shear rate (viscosity h¼shear stress/shear
rate), therefore:
h¼t=_
gð1Þ
The unit of viscosity his Pascal second (Pa s).
This property of viscosity is due to the dynamic behaviour
of the molecules. Their size is less than 0.01 mm, and their
contact duration less than 0.01 ms. The local friction between
the molecules statistically induces a mechanical property on
a greater scale of length (mm) and time (s). This property
can also be considered for the material making up the depos-
ited straw. The straw or plant layer consists of elements whose
size is much greater than that of the maple syrup molecules.
However, on a centimetre scale, the same viscous behaviour
is found for the maple syrup as for the straw layer. Experiments
conducted with barley and wheat stems indicate that the appar-
ent viscosity of this layer is in the range of 50e200 Pa. This
value is, of course, much higher than that for water
(0.001 Pa) but it is in the range of that of viscous maple syrup.
Fig. 5. Schematic representation in cross section of a tribulum (B), which moves on a film of straw (C) deposited on the threshing floor (A). A normal force W,
corresponds to the tribulum weight and the weight of a person sitting or standing on it. Three zones are present: the inlet zone, the cutting zone and the outlet zone.
In the inlet zone, the flow of layer C creates pressure that regulates the thickness h
cutting
of the layer under the cutting zone. Note that no blades are inserted in the
inlet and outlet zones. The blades do not touch the threshing floor h
blade
<h
cutting
.
1563P.C. Anderson et al. / Journal of Archaeological Science 33 (2006) 1559e1568

2.2. The three contact zones
Seen from today’s vantage point, the design of the tribulum
structure found empirically over its long history can be ex-
plained by tribological analysis.
In the tribological analysis, the behaviour of the layer is
considered to be fluid. Thus, the inlet zone (Fig. 5) corre-
sponds to a dynamic bearing zone. This means that the flow,
in order to be maintained throughout the zone, develops
a flow gradient and an internal pressure in the straw film
(C). This pressure, accumulated in the inlet zone area, bal-
ances out the normal applied load W
inlet
. We can see that W
inlet
is part of the total load Wapplied to the tribulum. In fact, for
an efficient lubrication, W
inlet
has to be an important part of W.
Without this inlet zone, or with a zone that has a low W
inlet
value, the flow of the layer can be destroyed. More detailed ex-
planations are found in classic lubrication textbooks [7,12].An
efficient lubrication W
inlet
, as shown in the following equation
(Eq. (2)), is first, related to the speed V, to the width of the
inlet zone wand to the viscosity of the layer h, second, it is
strongly related (power two) to the size dimensions of the inlet
zone, the length l
inlet
, and to the mean film thickness h
inlet
, and
third, it is independent of the angle of the inlet (if it is small
<15):
WinletzhVwlinlet
hinlet2
ð2Þ
Numerical calculation leads to 300 <W
inlet
<1200 N. This
load compared with W(1000 N) explains the utility of the bear-
ing zone. It fact, Eq. (2) is the result of a dynamic, auto-
controlled process between the design of the inlet area (length
l
inlet
) and the properties of the layer (thickness h
inlet
, viscosity
h). Because the contact pressure pvalue is low, the viscosity h
is independent of p. Now two points can be noted. First, accord-
ing to Eq. (2), the greater the length l
inlet
of the inlet zone, the
greater is W
inlet
. So, for a given W
inlet
, a tribulum with a large
l
inlet
induces a thicker straw layer. A tribulum without an inlet
zone destroys the straw layer rapidly. Second, there are usually
no blades in the inlet zone of the tribulum. The main reason for
this absence is to avoid any interference with the straw flow.
Moreover, as observed in our experiments, the straw is ran-
domly oriented before reaching the inlet. But in the inlet
zone, which has a relatively smooth surface, two effects are
possible: (i) the stems will be oriented parallel to the flow, or
(ii) if the length l
inlet
is great enough compared to the stem
lengths, the friction of the stems sliding against the tribulum in-
duces an orientation perpendicular to its movement, which is
favourable for the cutting process. Nevertheless, the flow of
straw is present along the blades, and the friction of the straw
against them creates a deposit transfer, as will be shown later.
The ability of the tribulum to produce massive quantities of
chopped straw in such a short time is particularly remarkable.
Indeed, the total volume of straw, which flows through the in-
terface, is equal to the mean entrance section h
inlet
w, which is
multiplied by 0.5 Vt, where tis the duration of use. The total
volume of straw to be worked is equal to S
F
h
inlet
¼4m
3
.
Therefore using these two relationships, tis found to be in
the range of few minutes. Of course this crude analysis does
not take into account the efficiency of the cutting process.
2.3. The viscoelastic material concept [24]
If the tribulum does not slide but, rather, remains stationary
on the straw layer, it compresses this layer. The layer does not
squeeze down completely during the process, however, so the
initial layer thickness is roughly restored, or rebounds, if the
tribulum is removed. This simple experiment allows us to eval-
uate a compression elastic modulus Eof the layer. Eis given
by the equation:
E¼W
ST
hf
dhð3Þ
where h
f
is the initial thickness and dhis thickness variation
during loading and unloading. Numerical values for Eof
a layer of barley straw are in the range of Ez2500e
9000 Pa, in relation to the straw’s pre-compressive state.
During a shear experiment, the material can also show elas-
tic behaviour, and the shear elastic modulus Gfound is related
to Eby the simple equation [12]:
GzE=3ð4Þ
Therefore for the straw, it is postulated that Gz800e
3000 Pa.
For viscoelastic material, it is useful to consider the ratio of
the viscosity hto the shear elastic modulus G, because this ratio
is a time T
mat
, which characterises the material. For the barley
straw, T
mat
z0.02e0.2 s. Next, it is important to compare this
time T
mat
with the characteristic time in the flow process. In the
flow process the time T
flow
is shown by the equation [24]:
Tflow ¼1
_
g¼hinlet
Vð5Þ
Numerically, T
flow
is close to 0.4 s, and this value is greater
than T
mat
. Therefore, it is concluded that the material has
time to flow in the interface C as shown in Fig. 6.
According to this lubrication theory [7,12], based on the
fluid behaviour of the straw, the overall thickness h
cutting
of
the straw film in the cutting zone is determined by the inlet
zone. This thickness h
cutting
prevents any inadvertent contact
between the blades and the hard surface of the threshing floor,
if the blade height h
blade
is less than h
cutting
. So if the tribulum
is well designed, in particular with a long inlet zone, it can
function using a greater cutting edge height.
In the cutting zone (Fig. 5), the straw layer impacts the
blades. The impact duration can be evaluated by saying that,
during the process, a small part of the collective material
(straw), made up of the stems is in contact with a part of the
blade. In this case the impact duration range T
cutting
corre-
sponds to the stem diameter (4 mm) divided by the sliding
speed V (0.5 m/s). So the impact duration is T
cutting
¼0.008s
(Fig. 6), a value which is less than T
mat
. Therefore, during
1564 P.C. Anderson et al. / Journal of Archaeological Science 33 (2006) 1559e1568

the impact of the stems against the blades, it is concluded that
the layer behaves as a solid and therefore the stem is fractured
as a brittle material. Fig. 4 shows an example of stems, where
the indentation due to the impacts of the blades are visible and
where the distance (dwz5 cm) between two fractures corre-
sponds to the mean blade inter-row distance, suggesting that
the stems are perpendicularly oriented to the rows of blades.
To conclude, it is very interesting to note that this structure
of the tribulum corresponds perfectly to the design of modern
machinery as recently analysed by lubrication theories [7,8].
The long history of this instrument has been driven by efficient
transformations linked to practical experience.
An illustration of how these concepts are still used in trib-
ulum design was found by one of us (PA) when interviewing
and filming the work of the carpenter who makes threshing
sledges today in the Anti-Atlas region of Northern Tunisia.
He spontaneously invoked the absolute necessity of many of
the functioning principles aboveealthough his sledge is
made of pine planks and now metal inserts have replaced
the flintein particular, the design of the inlet zone and its im-
portance for drawing in and turning the straw perpendicular to
the blade, then drawing it to the cutting zone, the value of the
blade height protruding from the sledge frame, and the length,
width and weight of the sledge as cited above.
Although as illustrated above, it is not necessary to know
all the above concepts to design the tribulum, they are useful
for knowing how to examine its components, and in particular
the blades used.
3. The friction surfaces of inserted blades
The aspect of microwear studies of such stone inserts that
has attracted the most critical attention has been the method-
ology involved in inferring use from the traces observed.
Discussion has centred on questions of the representativity
of the number of zones and objects sampled, as well as ade-
quacy of experimental controls and the size of the area ana-
lysed [25].
Three scales of length are used in the study of the blades,
and here we examine the case of flint Canaanean blade
segments.
First, on a scale of a few centimetres, the general shape of
the blades, oriented as they are in the tribulum, shows three
zones. Zone I in Fig. 3a, is generally covered by an adherent
layer of bitumen, used to glue the blade into the tribulum
structure. Zone II corresponds to the flint surface, which par-
ticularly reflects the effects of the straw flow. Zone III is the
irregular cutting edge.
Second, on a millimetre scale, an enlarged area of zone III
was observed, using optical microscopy (Fig. 3b). The surface
appears rough, with dark depressions, bright plateaus and
grooves more or less parallel to the cutting edge, correspond-
ing to the sliding direction of the straw.
Third, the details of the surface roughness were analysed
with the micrometric method, using a vertical-scanning inter-
ferometer microscope, which has been developed only re-
cently [13,22,28]. The goal is to obtain a light section
showing the micro profile (i.e. height and depth) of the sur-
face. Using the optical microscope with a 20 objective, the
light reflected from a reference mirror combines with light
reflected from a sample to produce interference fringes, where
the best contrast fringes occur at the best focus. During the
measurement, the reference arm containing the interferometer
objective moves vertically to scan the surface at varying
heights. A linearised piezoelectric transducer precisely con-
trols the motion. Therefore each position of the XY plane of
the observed surface corresponds to an image of a given light
intensity. An algorithm converts the intensity of light in the
normal height Zof the analysed surface. The maximum
Fig. 6. A logarithmic scale of time is plotted, showing the characteristic time T
mat
of the barley straw. Two other experimental times are reported. T
flow
, which is
related to the general flow of the straw layer under the tribulum, noting that when T
flow
>T
mat
, this produces a fluid behaviour. T
cutting
which is related to the cutting
process of the straw with the blades, noting that T
cutting
<T
mat
, therefore it is behaving as a solid.
1565P.C. Anderson et al. / Journal of Archaeological Science 33 (2006) 1559e1568

measured amplitude DZis 2 mm, with a resolution of dZ¼
0.000002 mm, for the two lateral resolutions of dX¼dY¼
0.002 mm.
Fig. 7 presents typical results obtained using the interfer-
ometer. The flint surface of a freshly knapped blade is exam-
ined. During the knapping process, a fracture wave creates two
new surfaces on the blade. Although flint is a brittle material,
the fracture surface is not highly smooth, and in particular this
roughness is dependent upon the nature of the flint. Therefore
it is necessary to make a comparison for a given flint material.
Fig. 7a shows the XY plane that was examined. The rectangu-
lar surface (X¼1.5 mm, Y¼0.5 mm) shows areas that appear
dark and white, depending upon their height Z. A cross section
ab of the surface (Fig. 7b) clearly shows that the difference in
altitude values between summits and valleys is in the range of
0.025 mm. The difference between the scales of magnification
used for the Zand X-axes magnifications (Zbeing amplified by
a factor of 10 in relation to X) shows the surface as rougher
than it is in reality. For instance at the point M, the slope forms
an angle of, not 70 with the horizontal plane, but rather of
15. This method of representation is used uniformly and al-
lows for easier visualisation and comparison of the asperity
transformations, as is shown later.
The friction simulation of the layer of plant sheaves (C)
against the blade was carried out in order to characterise the
wear mechanisms under specific tribological constraints, and
to understand the formation of traces of wear on flint inserts
in threshing sledges. The surface of a flint blade was pressed
against parallel segments of cereal stems with a small load
of 5 N, and submitted to a back and forth, reciprocal motion
at a low speed (0.025 m/s), with an dXamplitude of 10 mm.
The apparent contact pressure between the flint surface and
the straw, evaluated as a few MegaPascal, implies that the
straw stems were not statically deformed. Evolution of the flint
surface damage could be traced, because after each phase of
the friction experiments, it was possible to relocate the surface
under the microscope and therefore to follow the evolution of
exactly the same surface area. The data before and after the
friction tests are compared. Fig. 8 shows the same area as in
Fig. 7. (a) 3D examination with the optical interferometer of a freshly knapped
flint surface. (b) Cross section ab showing the surface profile. Notice the dif-
ference between the scales for the Zand X-axes magnifications (a ratio of 10),
done to aid visualisation of the profile, which gives a difference between the
observed and the real slope at the point M (15).
Fig. 8. 3D examination with an optical interferometer of the surface shown in
Fig. 7a, but after its sliding against a straw layer for 3000 cycles. The surface
appears smoother than before, but has some pronounced valleys.
Fig. 9. Cross sections ab showing the surface profiles (unused flint surface and
after friction against a straw layer). The purpose of relocating the two surfaces
is the same surface to follow the evolution of the wear. The grey-coloured
cross section represents the unused surface prior to the friction test.
1566 P.C. Anderson et al. / Journal of Archaeological Science 33 (2006) 1559e1568

Fig. 7a, but after the friction process against the straw had oc-
curred. The surface appears to be smoother than before, in par-
ticular because some smooth plateaus are visible, making the
valleys appear more pronounced. Because it is possible to pre-
cisely relocate the XY surface at each stage, the two cross sec-
tions ab can be compared, and the differences in the exact
local heights before and after the friction test evaluated.
Fig. 9 shows, in the same diagram, the difference between
the two cross sections, before and after a friction test with
3000 bi-directional cycles. It appears clearly that the hard flint
asperities are not deformed or fractured by the soft straw mat-
tress. Therefore the wear process of the flint is not due to a loss
of material, but rather to a gain in matter. A heterogeneous
film covers the flint surface. Film islands have sizes 150e
200 mm, with a mean thickness dZmof 0.008 mm or 8 mm
(Fig. 9, white cross section), corresponding to straw deposits
in the form of silica phytoliths.
Complementary observations of phytoliths having been
cut through by the threshing sledge blades, when mounted
on slides and viewed with a transmitted-light microscope
at 200e400, show that the spodogrammes (linked
phytoliths) of stem or chaff epidermis have sharp straight,
convex or complex cuts, as shown in Fig. 10 (see also
Fig. 4, short arrows). Due to their great thinness, these
phytoliths behave as small, very flexible sheets [2,14]. This
flexibility induces intimate contact between phytoliths and
flint and increases the van der Waals adhesion forces [15].
This implies that, during the friction process of straw against
flint, the straw is gently fractured into many fragments, which
adhere to the flint. This is typical of the creation of a transferred
layer. Therefore, the real friction occurring is not of straw
against flint, but of straw against transferred straw on flint.
Additional experiments, not reported here, showed two
phenomena [25]. First, if the friction process is increased in
duration, the thickness of the deposit layer stays constant but
the flint surface coverage by the film increases. Second, the
layer is in a continuous removal process, therefore as
a phytolith sheet is removed from the straw, it adheres to the
flint surface, then is again partially removed, etc. This is a clas-
sical adhesion wear process [12].
In conclusion, the flint surfaces show a more polished
appearance after friction against straw. This is not due to
the removal of the asperities created by an abrasion or
fatigue process. On the contrary, it is an adhesion process of
particles of the siliceous sheets (phytoliths) in the straw. The
filling in of the relatively shallow valleys is able to create
smooth plateaus on the flint surface. But this filling-in process
rarely occurs for the deep depressions. Therefore, the flint
surface shows plateau areas and deep depressions, as in
Fig. 3b. Microwear literature has often treated the question of
whether or not observed microwear traces on flint tools are
due to deposits on the surface or removal of the flint through
abrasion [1].
4. Conclusions
This study uses state-of-the-art tribological measurement
and theory to understand what occurs during use of the thresh-
ing sledge. It underlines the importance of the three bodies in
tribological interaction and the role of a film in the interaction
between materials.
This study explains, for blades having been used in the
threshing sledge, the features described in microwear analysis
using the optical reflected-light microscope, customarily at
100e200magnifications, while also revealing features on
a far smaller scale. It shows that what has been imagined to be
depressions caused by the tearing out of the flint surface, are
in fact residual features of the original flint surface, deeper val-
leys that have not been filled in by the film deposit on the blades.
However, these finds have farther-reaching implications for
understanding of the wear process on stone tools, at least for
those uses where friction with silica-rich cereal stems are con-
cerned [4].
The principal reason for the efficiency of the tribulum is
that the large inlet zone generates a thick film of straw in
the cutting zone, effectively controlling the straw rheology.
This study shows that this structure of the tribulum, still in
use today, corresponds to the design of machinery as recently
analysed by lubrication theories [7,8], and that the long, prob-
ably 8000 year, history of this instrument [2] has been driven
by efficient modifications based upon practical experience.
Acknowledgements
Jean-Pierre Gre
´goire (CNRS, Strasbourg), Jacques Pelegrin
(CNRS, Paris), and particularly Anthony Handley are thanked
for their invaluable roles in the conception of the Bronze Age
instrument reconstructed here. The field experiments with the
Bronze Age tribulum were funded in part by a special grant
from the CEPAM, CNRS and by the GDR 2517 of the
CNRS. Finally, we are grateful to the carpenter who makes
sledges in Be
´ja, Tunisia today for showing and describing
the functioning principles in their design.
Fig. 10. Photomicrograph of a phytolith spodogramme (amorphous silica
sheet) using a transmitted-light microscope at 400. The phytolith was
extracted from cereal chaff cut by the threshing sledge, and shows a distinctive
double concave cut in the lower part of the image.
1567P.C. Anderson et al. / Journal of Archaeological Science 33 (2006) 1559e1568

References
[1] P.C. Anderson, A testimony of prehistoric tasks: diagnostic residues on
stone tool working edges, World Archaeology 12 (2) (1980) 181e194.
[2] P.C. Anderson, Observations on the threshing sledge and its products in
ancient and present-day Mesopotamia, in: P.C. Anderson,
L.S. Cummings, T.K. Schippers, B. Simonel (Eds.), Le traitement des
re
´coltes: un regard sur la diversite
´du Ne
´olithique au pre
´sent, ADPCA,
Antibes, 2003, pp. 417e438.
[3] P.C. Anderson, J. Chabot, Functional analysis of glossed blades from
Northern Mesopotamia in the Early Bronze Age (3000e2500 BB): the
case of Tell ’Atij, Recherches Arche
´ome
´triques, Cahiers Arche
´ologiques
du CELAT, Quebec, Universite
´Laval (2001) 257e276.
[4] P.C. Anderson, J. Chabot, A. van Gijn, The functional riddle of ‘Glossy’
Canaanean blades and the Near Eastern threshing sledge, Journal of
Mediterranean Archaeology 17.1 (2004) 87e130.
[5] P. Anderson, M.L. Inizan, Utilisation du tribulum au de
´but du troisie
`me
mille
´naire: des lames ‘‘canane
´ennes’’ lustre
´es a
`Kutan (Ninive V) dans la
re
´gion de Mossoul, Iraq, Pale
´orient 20 (2) (1994) 85e103.
[6] K. Ataman, Threshing sledges and archaeology, in: P.C. Anderson (Ed.),
Prehistory of Agriculture: New Experimental and Ethnographic Ap-
proaches, UCLA Institute of Archaeology, Los Angeles, 1999, pp.
211e212 [Monograph 40].
[7] A. Cameron, Principle of Lubrication, John Wiley and Sons, London,
1966.
[8] D. Dowson, History of Tribology, second ed., Professional Engineering
Limited, London and Bury St. Edmunds, UK, 1998.
[9] M.A. Frangipane, 4th millennium temple palace complex at Arslantepe-
Malatya. NortheSouth relations and the formation of early state societies
in the northern regions of greater Mesopotamia, Paleorient 23 (1) (1997)
45e73.
[10] I.J. Gelb, Old Akkadian inscriptions in Chicago Natural History Mu-
seum: texts of legal and business interest, Fieldiana Anthropology 44
(2) (1955) 275e277 [Chicago].
[11] J.-P. Gre
´goire, in: P. Anderson, M.L. Inizan (Eds.), Utilisation du tribu-
lum au de
´but du troisie
`me mille
´naire: des lames ‘‘canane
´ennes’’ lustre
´es
a
`Kutan (Ninive V) dans la re
´gion de Mossoul, Iraq, Pale
´orient 20 (2)
(1994) 85e103.
[12] J.M. Georges, Frottement, usure et lubrification, CNRS Editions, Eyr-
olles, Paris, 2000.
[13] A. Harashki, J. Schmit, J.C. Wyant, Improved vertical scanning interfer-
ometry, Applied Optics 39 (2000) 2107e2115.
[14] D.M. Hayward, D. Parry, Scanning electron microscopy of silica deposits
in the culms, floral bracts and awns of barley (Hordeum sativum jess.),
Annals of Botany 46 (1980) 541e548.
[15] J.N. Israelachvili, Intermolecular and Surface Forces, second ed., Aca-
demic Press, London, 1992.
[16] P.N. Kardulias, R.W. Yerkes, Microwear and metrical analysis of thresh-
ing sledge flints from Greece and Cyprus, Journal of Archaeological
Science 23 (1996) 657e666.
[17] L. Keeley, The methodology of microwear analysis: a comment on
Nance, American Antiquity 39 (1974) 126e128.
[18] G.-H. Luquet, P. Rivet, Sur le tribulum, Me
´langes offerts a
`Nicolas Iorga,
Librairie J. Gamber, Paris, 1933, pp. 613e638.
[19] J.-L. Mingote Calderon, Tecnologia agriocola medieval en Espana. Una re-
lacion entre la etnologia y la arqueologia a trave
´s de loos aperos agricolas,
Ministerio de Agricultura, Pesca y alimentacion, Madrid, 1996.
[20] Pliny. Histoire Naturelle, XIV, XV, XVI, XXIV, Les Belles Lettres, Paris,
1958 (J. Andre
´Compiler, Trans.), 18, 30, 72, x298.
[21] A. Salonen, Agricultura Mesopotamica nach Sumerisch-Akkadischen
Quellen, Annales Academiae Scientarium Fennicae 149 (1968) 202e
212 [Helsinki].
[22] J. Schmit, K. Creath, Interferometric Techniques. The Encyclopaedia of
Applied Optics (1998).
[23] S.A. Semenov, Prehistoric Technology (M.W. Thompson Trans.), Cory,
Adams and Mackay, London, 1964.
[24] R.I. Tanner, Engineering Rheology, Clarendon Press, Oxford, 1988.
[25] R. Vargiolu, H. Zahouani, P.C. Anderson, Etude tribologique du proces-
sus d’usure des lames de silex et fonctionnement du tribulum, in:
P.C. Anderson, L.S. Cummings, T.K. Schippers, B. Simonel (Eds.), Le
traitement des re
´coltes: un regard sur la diversite
´du Ne
´olithique au pre
´s-
ent, ADPCA, Antibes, 2003, pp. 439e454.
[26] Varron, Economie Rurale (Heurgon J. Trans.), Les Belles Lettres, Paris,
1978, 1, 52, 1.
[27] J. Whittaker, Alonia and Dhoukanes: the ethnoarchaeology of threshing
in Cyprus, Near Eastern Archaeology 63 (2) (2000) 62e69.
[28] H. Zahouani, R. Vargiolu, J.L. Loubet, Ph. Kapsa, T. Mathia, Effect of
Lateral Resolution on Topographical Images and 3D Functional Param-
eters, World Tribology Congress, Mechanical Engineering Publications,
London, 1997.
1568 P.C. Anderson et al. / Journal of Archaeological Science 33 (2006) 1559e1568