Previous Page Table of Contents Next Page

Part 3. Archaeobotanical Evidence for Agricultural Transitions

Identifying Pre-domestication Cultivation Using Multivariate Analysis - S. Colledge
Pre-Pottery Neolithic A and Pre-Pottery Neolithic B Lithic Agricultural Tools on the Middle Euphrates: The Sites of Tell Mureybit and Tell Halula - J.J. Ibáñez, J.E. González, A. Palomo and A. Ferrer
History of Harvesting and Threshing Techniques for Cereals in the Prehistoric Near East - P.C. Anderson
Problems in Correlating Pollen Diagrams of the Near East: A Preliminary Report - R.T.J. Cappers, S. Bottema and H. Woldring
Investigations of Botanical Remains from Nevali Çori PPNB, Turkey: A Short Interim Report - R. Pasternak
Crop Water Availability from a Pre-Pottery Neolithic Site on the Euphrates, Determined by Carbon Isotope Discrimination of Seeds - J.L. Araus, A. Febrero, M. Catalá, M. Molist, I. Romagosa and J. Voltas

Identifying Pre-domestication Cultivation Using Multivariate Analysis - S. Colledge


An analytical method for assessing the composition of archaeobotanical assemblages is presented in this paper. The method highlights the differences between suites of plant taxa found on certain early sites in the Levant and the establishment of 'vegetational fingerprints' is proposed. Explanations for the differences in the taxonomic compositions of the assemblages are given in terms of the differences in the vegetation at the sites and, in particular, in the changing 'management' of the landscape by cultivation.

The work is based on a large body of published data from several sites in Syria (van Zeist and Bakker-Heeres 1982, 1984, 1986), and it stems from an investigation of plant exploitation on early prehistoric sites in the Levant (Colledge 1994).

Modern experimental work has confirmed that certain methods of land preparation and harvesting are prerequisites for an increase in numbers of wild cereals with the 'domestic-type' genetic mutation (i.e. in the gene controlling the toughness of the cereal rachis), such that if these techniques were practised over a considerable period of time they would lead to the dominance of fully domestic crops (Hillman and Davies 1992). In antiquity, therefore, pre-domestication cultivation (i.e. the cultivation of wild cereals) would have been a necessary stage in the evolution of domestic cereal species. It has also been proposed that a progressive increase in human energy input into the land would have been necessary to maintain productivity at and after the inception of cultivation (Harris 1996).

Changes in the material culture (including differences in types and abundance of groundstone and flint artefacts) in the Natufian period have been attributed to a greater use of plant resources.

“Even though the primary source of subsistence in the Natufian seems to have been animal resources, it was possibly the consistent utilization of plant resources that produced the ecodemographic and material cultural changes from the Palestinian Upper Palaeolithic tradition.” (Henry 1973:189-190)
The Natufian period is the cultural entity immediately preceding the period during which the first domesticated cereals have been found in the Levant (the period known as the Pre-Pottery Neolithic A, or PPNA). It is thought that during this time the first efforts at plant management, or cultivation, took place. There has been a long history of speculation about the possibility of pre-domestication cultivation of wild cereals during the Natufian period, and speculation has been accompanied by attempts to look for evidence (whether direct or indirect) of it having taken place at sites in the Levant. Two examples of potential sources of evidence in the archaeological and archaeobotanical records are described here.

Experimental use of modern flint sickle blades to harvest wild cereals seemed to demonstrate that the microwear (i.e. alteration to the surface of the flint in the form of polish or grooves) produced on the tools was distinctive if the soil the plants had grown on was cultivated (i.e. tilled) (Unger-Hamilton 1989:101). After comparison with the microwear on ancient blades, Unger-Hamilton concluded that wild cereals were being cultivated in the Early Natufian. Recent studies have shown, however, that there is not necessarily a direct correlation between surface alteration of flint and specific worked materials (Levi Sala 1996:68-69), and so the conclusion that Natufian sickles had been used to harvest cultivated wild cereals must be questionable.

It was suggested that there may have been histological changes to the cells in the outer layers of the grains of wild cereals which had been exposed to cultivation (Hillman et al. 1989:242-243), prior to the evolution of the domestic species (with the obvious domestic characteristics). Micromorphological investigations proved that no such changes were manifest in modern populations of wild cereals that were found growing alongside domestic crops (Colledge 1988; Hillman et al. 1989).

Despite these attempts and others, however, no clear, undisputed evidence has been found in the archaeological and archaeobotanical records that would favor pre-domestication cultivation having taken place.

Identification of 'vegetational fingerprints' in the archaeobotanical record

The published data from five Syrian tell sites have been used in this study. The sites are Tell Mureybit, Tell Aswad, Tell Ghoraifé, Tell Ramad and Tell Ras Shamra. Figure 1 shows the geographic location of each of the sites and Figure 2 gives the 14C dates. Tell Mureybit is situated on the northern bank of the Euphrates River, in the 'elbow' of the river course before it flows eastward to join the Belikh River (van Zeist and Bakker-Heeres 1986:171-173). The site lies within the Irano-Turanian phytogeographical region, in an area of heavily denuded steppe and desert vegetation. At the time the site was occupied the Euphrates Valley would have been covered with poplar forest. Tell Aswad, Tell Ghoraifé and Tell Ramad are all situated in the Damascus Basin, in southwestern Syria (van Zeist and Bakker-Heeres 1982:168-173). The basin is bordered on the west by the eastern escarpments of the Anti-Lebanon Mountains, to the north by the limestone hills of the Palmyra range, and to the south and east by the basalt outcrops of the Hauran Plain. Tell Aswad and Tell Ghoraifé are both located in the eastern part of the basin, within an area bounded by the shores of an ancient Pleistocene lake. They are in the vicinity of two extant lakes, Lake Aateibé and Lake Hijjâné, fed by perennial streams that flow from the west. The Damascus Basin lies within the Irano-Turanian phytogeograhical region. The vegetation in the area has become degraded because of overuse of the natural resources. Tell Ramad is situated farther to the west, outside the area formerly occupied by the ancient lake. In these higher elevations the natural vegetation is typical of the Xero-Mediterranean almond-pistachio forest steppe. Tell Ras Shamra is located in northwestern Syria, close to the Mediterranean coastline (van Zeist and Bakker-Heeres 1984:151-153). The site lies within the Eu-Mediterranean forest region.

Fig. 1. Map of Syria with locations of the archaeological sites.

Occupation at the five sites spanned the time during which it is thought cultivation began. The periods represented at the sites are also those for which there is the first evidence of domestic crops and signs that cereal-based agriculture spread throughout the Levant. The published data for the sites were in the form of lists of plant taxa, with numbers of taxa recorded per sample and per occupation phase. The authors undertook simple numerical analyses and made comments based on subjective assessments of the data sets about the relevance to early cultivation of the differing proportions of crops and wild species (van Zeist and Bakker-Heeres 1982:234-241, 1984:167-169, 1986:192-194).

From the outset the aim of this investigation was to look for patterns and trends in the archaeobotanical data that might reflect developmental changes associated with the control and use of plant resources through time. Multivariate statistical analysis was used to investigate any covariational relationships between the taxonomic composition of samples and chronological phases.

Multivariate analysis is described succinctly by Lange as a means by which for large data sets:

“Redundancy of information is summarised, noise is reduced, outliers can by identified and relations are brought to light.” (Lange 1990:41)

Fig. 2. Radiocarbon dates of the archaeological sites in Syria.

In this study it was relevant to use simple correspondence analysis as the multivariate technique, it being an analytical method ideally suited to data of the abundance-type (the data sets for the sites in question were large, comprising counts of over 200 taxa from 244 samples). Many statistical analyses are applicable only to data with a normal distribution, which data of the abundance-type seldom have, but correspondence analysis makes no assumptions concerning the distribution of variables in the data set. Of importance to this study also was the fact that correspondence analysis is symmetrical with regard to the samples and taxa, i.e. they are analyzed in the same multidimensional space and the results can thus be shown on the same plots (Høiland-Nielsen 1988:37). Lange (1990:43) comments:

“In graphical form the results of a Correspondence Analysis bring out the position of each sample relative to all other samples and to all the species, and of each species relative to all other species and to all the samples in the analysis.”
Exploration of the data sets using correspondence analysis was essentially an exercise in 'pattern searching', looking for any meaningful grouping of samples from the sites by phase or geographical region. The analyses comprised a long series of progressional steps, each step being based on the outcome of the previous one, and at each stage the data sets were refined (e.g. sites were included or excluded where appropriate, non-frequently occurring taxa or groups of taxa were eliminated, etc.) in order to define more clearly any observed patterns. The most informative analyses, i.e. those which produced the most meaningful patterns of clustering of samples within site groups and phases, were those which used the composition of the wild/'weed' taxa only (i.e. the 'crop' components were non-discriminatory; in this paper all the discussions refer to analyses of the composition of the wild/'weed' taxa in the archaeobotanical assemblages). Where there were obviously demarcated clusters of samples it was an indication that those samples within a cluster comprised similar suites of taxa, and significantly different suites to those in other clusters. The term 'weed' is applied here with caution, given that for these periods and in these locations there is bound to be considerable ambiguity concerning the status of taxa that are now designated obligatory segetals.

Figure 3 shows a correspondence analysis (hereafter denoted CA) output plot for the wild/'weed' taxa in the samples from the three sites in the Damascus Basin: Tell Aswad, Tell Ramad and Tell Ghoraifé. The plot shows that the samples fall into three distinct groups. In the plots CA the strongest separation will appear along the first principal axis (horizontal) and then along the second (vertical), thereafter along axes 3 and 4 (all plots in this paper show the first two principal axes). The Tell Ramad samples are separated from those of Tell Ghoraifé and Tell Aswad along axis 1, and the latter two sites are separated along axis 2. This indicates that the taxonomic compositions were significantly different in all three sites (although the sites have a majority of taxa in common). The implication is that the archaeobotanical record has preserved a Vegetational fingerprint', presumably as a result of diverse plant-related activities which led to the transport and deposition of the different suites of taxa at each of the sites.

The sensitivity of the archaeobotanical record

Figure 4 shows a CA plot with two additional sites in the data sets; Tell Mureybit and Tell Ras Shamra, but in this case only samples dating to the Pre-Pottery Neolithic B (PPNB) period are included in the analysis. Again the samples from each of the sites form definable and distinct clusters: Ramad, Ras Shamra and Mureybit are separated from Aswad and Ghoraifé on the first principal axis, with further separation along the second axis (rather less clearly for the Mureybit samples in this plot). It seems, therefore, that - whatever the nature of the activities that led to the deposition of plants at the sites - they were repeated and consistent, such that the Vegetational fingerprints' were established at each of the sites within relatively short periods of time (e.g. a little over 1000 years for the PPNB).

Analyses run for individual sites showed even more clearly how the different phases of occupation were defined by the wild/'weed' taxa in the archaeobotanical assemblages. In the CA plot for Tell Ras Shamra (Fig. 5) the three phases of occupation are separated along axis 1 (to a lesser extent along axis 2). Further confirmation of the sensitivity of the archaeobotanical record is evident in this plot. An aberrant 'Halafian' sample appears to nestle among the Pottery Neolithic/Late PPNB clusters (i.e. it has a taxonomic composition characteristic of the earlier phases), implying perhaps that the archaeologist has assigned it wrongly to the latest phase. Of note is the apparent time trend in this plot, with the earliest phase (Late PPNB) at the extreme right hand side and the latest phase (Halafian) at the left.

Fig. 3. Correspondence analysis of the Damascus Basin samples for wild/'weed’ taxa only, after taxa occurring in <10% of the samples have been omitted.

Fig. 4. Correspondence analysis of the Syrian Pre-Pottery Neolithic B samples for wild/'weed' taxa only.

It seems reasonable to propose that the 'vegetational fingerprints' are representative in some way of the floristic composition of the vegetation on land surrounding the sites. The differences between the sites and phases may reflect differences in the natural, undisturbed landscape or in the 'managed' land (i.e. the cultivated fields), or they may reflect either the relative availability of wild plant resources or the predisposition of people to choose certain taxa for use (or any combination of these possible explanations). It also follows, therefore, that the differences through time (coincidental with culturally defined boundaries) may relate to changes in any one of these aspects (whether caused by natural phenomena, e.g. climatic fluctuations, or by 'human-induced' factors). Presence of the wild/'weed' taxa on the sites would have been the result either of collection for use or of accidental inclusion with other plant material.

Interpretation of the 'vegetational fingerprints'

To shed some light on how and why the 'vegetational fingerprints' were established it was necessary to attempt to explain the patterns of samples in terms of the suites of plant taxa that characterized them. This involved classification of the plants; several different methods were attempted. The classification was considered to be successful (i.e. to be an appropriate explanation of the site or phase grouping) if it produced taxonomic patterning coincident with the relevant patterns of samples. The wild/'weed' taxa were put into categories according to (1) their growth habit (e.g. whether they were annuals, perennials, trees or shrubs, etc.), (2) the plant part which may have been used (e.g. whether leaf, stem, wood, fruit, flower or seed was useful), (3) the possible usage of plant (e.g. for food, fuel, construction, fodder, etc.), and (4) the plant ecologies (many other categorizations of the plant taxa could be appropriate but within the time constraints of this investigation only those listed could be applied). The most successful was the ecological classification. For this it was only possible to assign the taxa to very general ecological classes because of the limitations of nonspecific identifications and also the restricted information on habitat preferences in the published floras. The wild/'weed' taxa were grouped according to whether they grew in maquis, batha, field (those taxa which were obligatory segetals), wet and steppe/desert regions. The term batha refers to dwarf shrub communities in the Eu-Mediterranean zone where the aboreal cover has been denuded or has failed to develop (Zohary 1973:532).

Tell Mureybit and pre-domestication cultivation

The analytical exercise was perhaps most successfully applied in the case of Tell Mureybit. The CA plot (Fig. 6) shows the clear grouping of samples according to the periods of occupation. The samples from the Khiamian phase, an intermediate archaeological entity between Natufian and the PPNA, fall into two groups (van Zeist and Bakker-Heeres 1986:193-194). Interestingly, in this analysis, one allies closely with the PPNA group and the second allies with the PPNB.

Fig. 5. Correspondence analysis of the Tell Ras Shamra samples for wild/'weed' taxa only, after taxa occurring in <10% of the samples have been omitted.

The easiest way to portray the ecological grouping of taxa within the samples was to use pie charts, where each pie chart represents a sample and the segments of the pie charts represent the relative proportions of the taxa in the different ecological groups. Figure 7 shows the pie charts for the equivalent sample plot, and there is clear patterning coincidental with that of the periods of occupation.

As discussed above, the suites of taxa are likely to represent in some way the floristic composition of the vegetation on land surrounding the sites, and so for Mureybit the dominant ecological groups for each period of occupation may be indicative of the most prevalent vegetation type at that time. If the assumption is made that 'human-induced' factors are at least as likely an explanation for the changes as the effects of natural phenomena, then the following interpretation may be appropriate. Wild einkorn wheat was the most commonly occurring 'crop' found at the site. A likely mode of transport of the wild/'weed' taxa to the site was as inclusions with the wild cereals, contaminants of the 'harvest'. It follows that the ecological grouping may reflect the composition of the vegetation in areas where the cereals were growing. So the dominance of 'wet-loving' taxa in the PPNA samples (for which high numbers of wild cereal grains were recorded) may indicate that at this stage the wild einkorn was growing on the alluvial margins of the Euphrates. A component of 'field' taxa appears in some of the PPNA samples and this may reflect the presence of disturbed ground that the 'weedy' plants would favor (as commented upon by van Zeist and Bakker-Heeres 1986:198). 'Wet-loving' taxa are not as apparent in the PPNB samples and this may be an indication that the wild cereals were growing on drier ground at this time. 'Field' taxa and taxa common in batha are represented in these samples. In this case the taxa common in batha were predominantly annuals, and it is this aspect of the group which may be relevant to the study. The presence of taxa from both ecological groups could be an indication of tillage, in preparation for the growth of wild cereals. The 'weedy' taxa would favor the newly opened ground, and there may have been an increase in the proportions of annuals as this growth habit is favored (as opposed to perennials) if the soil is disturbed in cultivation.

Fig. 6. Correspondence analysis of the Tell Mureybit samples for wild/'weed' taxa only, after taxa occurring in <10% of the samples have been omitted.

Fig. 7. Pie chart of the Tell Mureybit samples for wild/'weed' taxa only showing ecological categories.


It appears, therefore, that the application of correspondence analysis to the wild/'weed' taxa in the samples from Tell Mureybit, together with appropriate ecological classification of those taxa, indicates pre-domestication cultivation of wild cereals near the site (or in this case non-domestication cultivation, where the methods of land preparation and harvesting practices did not favor the selection of wild cereals with the 'domestic-type' genetic mutations, and so did not therefore result in the establishment of domesticated crops. It is proposed that the increasing energy applied to the land, in the form of tillage, causing interference and alteration of plant communities, is reflected in the archaeobotanical record.

Van Zeist concluded that the weed seed frequencies lent no support to the hypothesis of 'protoagricultural' practices and that the interpretation of the plant material must remain inconclusive (van Zeist and Bakker-Heeres 1986:198). I argue, to the contrary, that statistical manipulation of the numerical data in the form of correspondence analysis may allow pre-domestication cultivation to be recognized on early multiperiod sites such as Tell Mureybit.


Colledge, S.M. 1988. Scanning electron studies of the cell patterns of the pericarp layers of some wild wheats and ryes. Methods and Problems. Pages 225-236 in Scanning Electron Microscopy in Archaeology (S.L. Olsen, ed.). BAR International Series 452, Oxford, UK.

Colledge, S.M. 1994. Plant exploitation on Epipalaeolithic and early Neolithic sites in the Levant. PhD Thesis. University of Sheffield, UK.

Harris, D.R. 1996. Domesticatory relationships of people, plants and animals. Pages 437-463 in Redefining Nature: Ecology, Culture and Domestication (R. Ellen and K. Fukui, eds). Berg, Oxford, UK.

Henry, D.O. 1973. The Natufian of Palestine: Its material culture and ecology. PhD Thesis. Southern Methodist University, USA.

Hillman, G.C. and M.S. Davies. 1992. Domestication rate in wild wheats and barley under primitive cultivation: preliminary results and archaeological implications of field measurements of selection coefficient. Pages 113-158 in Préhistoire de l'agriculture: nouvelles approches expérimentales et ethnographiques. Monographie du CRA No. 6, Centre de Recherches Archéologiques (PC. Anderson, ed.). CNRS, Paris, France.

Hillman, G.C., S.M. Colledge and D.R. Harris. 1989. Plant-food economy during the Epipalaeolithic period at Tell Abu Hureyra, Syria: dietary diversity, seasonality, and modes of exploitation. Pages 240-268 in Foraging and Farming - The Evolution of Plant Exploitation (D.R. Harris and G.C. Hillman, eds.). Unwin Hyman, London, UK.

Høiland-Nielsen, K. 1988. Correspondence analysis applied to hoards and graves of the Germanic Iron Age. Pages 37-54 in Multivariate Archaeology, Jutland Archaeological Society Publications XXI (T. Madsen, ed.). Aarhus University Press, Denmark.

Lange, A.G. 1990. De Horden Near Wijk Bij Duurstede [Plant remains from a native settlement at the Roman Frontier: A numerical approach]. Nederlandse Oudheden 13, Kromme Rijn Projekt 3, ROB, Amersfoort [in Dutch].

Levi Sala, I. 1996. A Study of Microscopic Polish on Flint Implements. BAR International Series 629, Oxford, UK.

Unger-Hamilton, R. 1989. The Epi-Palaeolithic Southern Levant and the Origins of Cultivation. Curr. Anthropol. 30(1):88-103.

van Zeist, W. and J.A.H. Bakker-Heeres. 1982. Archaeobotanical studies in the Levant, 1. Neolithic sites in the Damascus Basin: Aswad, Ghoraifé and Ramad. Palaeohistoria 24:165-256.

van Zeist, W. and J.A.H. Bakker-Heeres. 1984. Archaeobotanical studies in the Levant, 2. Neolithic and Halaf levels and Ras Shamra. Palaeohistoria 26:151-170.

van Zeist, W. and J.A.H. Bakker-Heeres. 1986. Archaeobotanical studies in the Levant, 2. Late-Palaeolithic Mureybit. Palaeohistoria 26:171-199.

Zohary, M. 1973. Geobotanical Foundations of the Middle East. Volumes I and II. Gustav Fischer Verlag, Amsterdam, Netherlands.

Pre-Pottery Neolithic A and Pre-Pottery Neolithic B Lithic Agricultural Tools on the Middle Euphrates: The Sites of Tell Mureybit and Tell Halula - J.J. Ibáñez, J.E. González, A. Palomo and A. Ferrer


The study of the origins and first developments of agriculture must be approached in an interdisciplinary way, taking into account fields of analysis such as the identification of archaeobotanical remains, plant genetics, soil micromorphology, etc. Within this endeavor to gain an understanding of the characteristics of primitive agriculture, a fundamental aspect is provided by the study of prehistoric implements. New productive tasks demanded a marked technological change, which is at the very root of the economic transformations that then took place. It was necessary to innovate and to perfect tools used in the preparation of the soil and in the gathering and processing of crops. The study of tool culture is, consequently, key to gaining a knowledge of the development of agricultural techniques.

Analysis of the function of implements found at the sites of Tell Mureybit (J. Cauvin 1979), levels IIIa, IIIb and IVa, and from the sequence at Tell Halula (Molist 1996), has allowed us to identify some aspects of the development of agricultural techniques between the beginning of the PPNA (10th millennium BP) and the late Neolithic period (first half of the 8th millennium BP) in the middle Euphrates Valley. In this study we shall concentrate on two types of agricultural implement: sickles and hoes.

The morphology of sickles

The sickle is the tool characteristically used in cereal-based agriculture. Research undertaken by Anderson (1988, 1992) and Unger-Hamilton (1989, 1992) has provided interesting information concerning the use of these instruments, particularly during the early days of agriculture.

The main aspects that characterize sickles from this period are the shape of the shaft and the manner in which the pieces of flint were inserted into it. These two factors determined the way the implement was used and how effective it was. Natufian bone sickles are well documented, such as those found at Umm ez-Zoueitina (Neuville 1951), and at El Ouad and Kebara (Turville-Petre 1932). These are straight shafts containing blades of flint inserted in a parallel fashion.

Excavations at Tell Mureybit from PPNA discovered flint pieces used for cereal-harvesting consisting of straight blades that reveal a use-wear polish parallel to one or both edges of the piece. This shows that the pieces were inserted parallel to the shaft. One of the sickle pieces from level IIIa at Tell Mureybit is a broad blade of relatively large dimensions (135 x 42 mm) which carries the signs left by cutting cereal all along the edge, the traces being most intense in the middle and at the far end of the edge (Fig. 1). On the side opposite the blade the piece reveals a cortical back with a series of irregularities that were not removed by retouching. The distribution of marks and the shape of the back of the tool show that this piece was actually hand-held when used. It was, therefore, a harvesting knife, used to cut cereal stalks which would be held in the hand opposite to the one that held the knife. At this level of the site a limestone shaft turned up with a groove for the insertion of a flint blade (Anderson 1992). The shaft and blade together would constitute a kind of implement similar to a large-dimensioned blade. It is possible that apart from these harvesting knives there existed other tools distinguished by having one area for the insertion of the blades and another for grasping. Some of the blades showing traces from harvesting are quite long, between 8 and 12 cm, which indicates that the shaft into which they were inserted had to be straight. Nevertheless, the greater part of the sickle components are smaller, between 3 and 5 cm. In these cases it is difficult to ascertain whether the shaft they were inserted into was straight or bent.

Fig. 1. Flint blade with marks from harvesting, probably used without a shaft and held by hand. Level IIIa at Tell Mureybit, PPNA. Drawing: A. Deraprahamian.

The data obtained from the sickle pieces found in the old and mid-PPNB levels at Tell Mureybit and Tell Halula point to systems of shafting similar to those observed in the PPNA (Fig. 2). The morphology of the pieces and the distribution of the use-traces are similar to those from the preceding period. Long blades exist that must have been inserted into straight-shafted sickles, alongside others of lesser size that could be used in both straight and bent shafts (Fig. 3).

Fig. 2. Mid-PPNB sickle elements at Tell Halula.

Fig. 3. Late Neolithic sickle elements at Tell Halula.

However, from the Late PPNB at Tell Halula important changes can be observed in sickle morphology. At this point distributions of traces appear in an oblique position, relative to the edge. During the late Neolithic period, the sickle elements with traces in oblique disposition amount to 70% of the glossed tools (Table 1, Fig. 4). Fractured blades were inserted obliquely into the shaft. Since the sickles were made up of various insertions, the outline of the edge was toothed. Moreover, the distribution of bitumen residue used for hafting the pieces with a shaft, very common in the sickle pieces from the site (Table 1), shows that it is a question here of curved shafts, rather than straight ones. The curvature of the shaft would draw an arc with a radius of around 10 cm. The curved sickle with oblique insertions also has been identified on the site at Tell Aswad, in PPNB levels (Cauvin 1973). This type of sickle continues to be dominant in the late Neolithic period at Tell Halula, although it must be pointed out that both in the Late PPNB and in this period some sickle elements show use-traces which are parallel to the edge (Table 1).

While we can state that the use of oblique insertion began, at least in Tell Halula, in the Late PPNB, it is more difficult to situate chronologically the beginning of the curved shaft, since some of the pieces from the PPNA at Tell Mureybit and from the PPNB at Tell Mureybit and Tell Halula could have been inserted into this kind of shaft. The sickle from the cave at Nahal Hemar, from the second half of the 9th millennium BP, has a curved shaft with parallel pieces (Bar-Yosef and Alon 1988). From the 8th millennium BP onwards, curved sickles from Çatal-Höyük VI-V and the later ones from Hacilar VI-II are known (Mellaart 1963, 1970; Cauvin 1983).

The change in sickle shape, evidence of which dates at least from the Late PPNB at Tell Halula, must have meant an important transformation as far as the cutting motion and its productivity were concerned. The harvesting experiments we have carried out with straight and curved-shafted sickles, whether with parallel or oblique insertions, have revealed the peculiarities of the work associated with each kind of implement. The harvesting work was done in Zuheros (Cordoba, Spain), where einkorn (Triticum monococcum) is still cultivated in the traditional way (Peña-Chocarro 1996), i.e. it is harvested using sickles. The experience of harvesting which the farmers in the area transmitted to us allowed us to understand the relationship between the shape of the tool and the technical movements which the cutting work requires (González et al. 1997). With the straight shafts one has to gather the stalks with one's free hand, then make a series of tangential cuts on the bunch of stalks. However, the curved shape of the sickle allows one to group together the stalks that are to be cut, using the tool itself. The farmers of Zuheros emphasized the importance of this movement, which they literally describe as 'calling the stalks'. They are grouped together with the sickle, held in the free hand and cut with the tool. With this technique it is possible to carry out the task of gathering the stalks by moving the sickle inwards, and to cut them by making the opposite movement with the tool. Thus, a continuous technical movement is made that allows the harvesting to be done with greater agility.

Moreover, from our field experience, we observe that with the use of oblique insertions, which produce edges with a toothed outline, a certain transverse component is offered when cutting which makes it possible to harvest bunches of stalks of a greater size.

In conclusion, the oblique insertions and curved shape of the sickles must have meant a certain advance in harvesting tasks, since they must have been accompanied by changes in the technical movements used in this work, with a corresponding increase in productivity.

Use-wear traces

As regards the characteristics of the traces left from harvesting, microscopic observation has shown that throughout the sequence at Tell Mureybit and Tell Halula the use-traces tend to be more and more intense with a more striated use-wear polish. These two aspects had already been mentioned by other researchers (Anderson 1992; Unger-Hamilton 1992). The greater intensity in the use-traces must be placed in relation to the greater importance of the work in the more recent period. The presence of more grooved use-wear polishes has been explained by cereal-cutting, the cultivation of which meant tilling the land. The cereal stalks would be impregnated with dust from the worked soil, which would produce the striations (Unger-Hamilton 1989, 1992). Nevertheless, this hypothesis has been rightly questioned (Anderson 1992; González et al. 1997). The presence of striations in the polish is connected with: (1) the degree of dampness in the material being worked on, (2) the presence of abrasive elements in the working context, and (3) the intensity of use. The lower the degree of dampness (1), the greater the presence of abrasive elements (2) and the greater the intensity of use (3), the greater the number of grooves that will appear in the polish. In our opinion, the increase in striations in the tools of the most recent periods must be situated in relation to:

· the cutting of ripe cereals, which means a lower level of dampness in working conditions

· the increase in the tendency to cut the plant from below, near the ground, with the aim of making use of the straw, and

· more intense use of the tools.

Why sickles?

In the Near East, flint implements were used for the harvesting of cereals from the Natufian period. However, cereals can be gathered in different ways without resorting to cutting. Wild cereals can be pounded with a stick and the grain gathered in a basket, or can be harvested by hand, as demonstrated by ethnographic examples (Harlan 1992; Hillman and Davis 1992; de Moulin, pers. comm.). Hulled cereals under cultivation can be gathered by hand, or by using systems involving nipping the ear, as is practised with mesorias (Sigaut 1978; Peña-Chocarro 1996). So, one has to ask what the motive was that led to the generalized use of the sickle for this task. Some authors have related the use of the sickle to the need to make use of the straw (Sigaut 1996). In this way, harvesting with sharp implements would be related to contexts where there is a need to use the grain and the straw. Nevertheless, it is not reasonable to suppose that during the Natufian and the PPNA the need for straw for construction and the manufacture of objects was as great as might be suggested by the number of sickle pieces encountered in the levels corresponding to these periods. In fact, in the adobes of the PPNA (Tell Mureybit, Jerf el Ahmar) fragments of cereal stalk are not found, since only the chaff from the ears was used. Stalks begin to be in evidence in the adobes from the PPNB onwards (D'jade, Tell Halula) (Hillman and Davis 1992; Anderson, pers. comm.). On the other hand, in some contexts involving domestic cereals, when sickles are used a cut is made in the stalk close to the ear, with no intention of making use of the straw. In Egyptian drawings in the Book of the Dead, in which harvesting tasks are depicted, it can be observed that the stalks were cut with sickles some 20 cm below the ear. In the present day too, in Zuheros, einkorn is harvested employing a high cut, thereby avoiding contact with the thistles that make their appearance in the fields like weeds, and thus making the worker's job more comfortable. In this case the exclusive aim is to obtain grain, since the straw is not used.

Fig. 4. Lithic implement probably used as a hoe. Mid-PPNB at Tell Halula. Drawing by A. Deraprahamian

Table 1. Technical characteristics of mid-PPNB and late Neolithic sickle elements at Tell Halula.

Late Neolithic






Flaking type

















Type of blank
















Type of flint












Distribution of gloss


































The technical justification for the use of the sickle, therefore, resides not only in the need to make use of the straw; there must also be other reasons to justify its use, also including when the intention is exclusively of a grain-gathering nature. The ethnographic work we have carried out in Zureda (Asturias, Spain) and Zuheros (Córdoba, Spain) allows us to suggest some hypotheses in this regard.

Zureda is in Asturias, in the north of Spain, a region with a damp Atlantic climate. Spelt (Triticum spelta) and emmer (Triticum dicoccum) are cultivated in these parts and mature at the beginning of September. Harvesting is carried out with mesorias, an implement consisting of two wooden sticks between which the ears are pincered and then pulled off in a tugging movement (Sigaut 1978; Peña-Chocarro 1996). The ears fall into a basket while the straw remains in the field. Why are sickles used for harvesting in Zuheros whereas mesorias are used in Zureda?

In Zureda we were able to verify that harvesting with simulated Neolithic sickles was up to three times faster than the gathering process using mesorias. However, with the sickle system the workload is reduced, since it is not necessary to separate the straw from the ears. In contrast, in Zuheros, threshing work is relatively expensive. What can have led to choosing between these two options in each context? In Zuheros (Andalucía) which has an arid mediterranean climate with very hot dry summers, the available time during which harvesting has to be done is limited to about a week. If the reaping work is delayed, as the farmers told us, the wheat becomes too dry and the heads fall down on the ground when cut and there is a loss of production. In Zureda, on the other hand, in a damp Atlantic climate, harvesting can be drawn out for up to three weeks without the losses seen in more dry areas. Moreover, the fields are of relatively small dimensions. So, in Zureda they opt for a technical solution that involves prolonged working in the fields because the atmospheric conditions and size of the fields allow this. In Zuheros harvesting must be carried out in a short space of time, for which it is necessary to use a tool that allows the work to be completed quickly.

It is assumed that the relatively dry atmospheric conditions in areas of the Near East at the beginning of the Neolithic period meant that the time available for gathering a wheat crop was relatively short. In the experimental fields at Jalés, the harvesting time available for wild wheat (Triticum boeoticum) is very limited, actually less than a week (G. Willcox, pers. comm.). This is the interval of time during which the cereal grain is physiologically mature but the plant is not yet completely ripe, and it is consequently possible to cut it without losing too much grain. During periods when the need for cereals was low, it was possible to revert to slower or less productive systems of harvests (Hillman and Davis 1992) but once cereal exploitation increased, a reaping implement had to be chosen that would exploit work time in the fields to the fullest, as is the case with the sickle. Thus, the technical value of the sickle would be not just as a tool to obtain straw but also a highly efficient implement at harvest time. Once the maximum amount of cereal had been gathered in the minimum time possible, the cereal could then be processed further on the settlement at leisure.

Technical characteristics of sickle elements

The study of the technical characteristics of sickle elements found at Tell Halula has allowed us to pinpoint certain changes throughout the sequence. At no point do polished pieces become dominant among the retouched implements (Table 2). During the Middle PPNB, the most numerous retouched tools are the tips of projectiles, and in the late Neolithic, retouched flakes and blades.

During the Middle PPNB, a form of bipolar knapping was used, and this made it possible to obtain blades, some of which were used as sickle components. In the majority of cases these blades were used without retouching. In the late Neolithic the knapping was unipolar, and medium-grain flint, with worse cutting qualities, was most commonly used (Table 1). In this period, not only blades but also stone flakes were used as sickle elements.

Both in the Middle PPNB and in the late Neolithic, 40% of the sickle elements reveal residues of a gluing substance that had been used in fixing the elements into the shaft. However, on other types of implements (points, scrapers and bores) there are no such traces, which either means that these other tools were used with no shaft or that systems of fixing into the shafts were used that did not require the use of a gluing substance.


Apart from sickles, during the Middle PPNB at Tell Halula we have found another kind of tool which we believe to be associated with farm work. It consists of extended pieces worked in limestone, displaying an active edge at one end. A dozen of these objects were found in the Middle PPNB layer of this site. The functional study of these tools is in progress, but the results obtained to date are sufficient to justify the following preliminary remarks. The stone from which the objects were made is relatively soft which enabled them to be shaped using flint tools. At a first glance one can see the striations left by such tools in the manufacturing process.

The fundamental morphology of the pieces, of an axe or adze-type, suggests percussion work. However, the stone they are made from is relatively soft, therefore its application would not be effective on materials of a consistency such as wood or bone. Moreover the active edges are relatively blunt, which provides another argument to rule out work on the above-mentioned materials.

On one side of the active edge a black layer can be seen. It was debated whether this might be the residue of the substance worked on which had stuck to the tool. On some of the pieces it was evident that use has left partial abrasions on that layer, especially in the areas nearest to the edge from which it was concluded that the layer was added prior to use. One of the pieces, whose edge was not used intensively, clearly reveals the way in which the layer was attached (Fig. 4). The edge of the object was dipped into a black liquid so that one of the sides became substantially impregnated, whereas the other was only affected in the form of a narrow strip stuck to the edge. The position of the black layer and the small black drops on the objects that can be observed with a magnifying glass reveal the liquid nature of the black product before it was impregnated. From the appearance of this product, it is very likely that the substance is bitumen.

The use marks indicate that the tool was used for a percussion action. In two of the four objects studied the edge is partially fractured from impact use. One of the pieces was used despite the break. Moreover on this same piece several microscars due to impact also can be observed. Both the fracture and the scarring have been subject to partial abrasion by later use. Apart from fractures and microscarring, use of the tool left abrasions in the active area, as can be seen in the rounding of the edge and the use-wear polish on the active part, particularly noticeable on the side with the black layer.

Table 2. Retouched tools from mid-PPNB and late Neolithic at Tell Halula.



Late Neolithic










Glossed tools















Retouched blades





Retouched flakes





Pointed blades


















Consequently, we are dealing with percussion work such as that performed by an adze, as demonstrated by the presence of use-marks which are more intense on one side than the other. The material cannot have been hard but must have been abrasive, and the tool would have entered deep into it. These being the characteristics, our view is that the material being worked on must have been the soil. These tools would work like a hoe. What function would the black layer (very probably composed of bitumen), which appears on one of the sides of the active edges, perform? We have established that the black layer and the most intense abrasion marks are present on the same surface. The layer of possible bitumen would be aimed at reinforcing the resistance of the surface of the tool to abrasion from the earth. Indeed, the limestone from which the tools are made is quite soft. Moreover, bitumen offers enormous resistance to traction and abrasion. It is spread over one of the sides, since in hoeing one of the sides of the edge penetrates the earth and is therefore subject to abrasion.

Despite the protection afforded by the bitumen, use must have considerably altered the state of the active part of the tool, since the edges of the pieces show marks of reconditioning in the form of scraping with lithic tools. This reconditioning work would involve the active edge being impregnated with a new layer of bitumen.

Following the above assumptions we have designed an experimental program to test the efficiency of the tool and to compare the marks left by the experiment and the archaeological ones. We must state that the experimental work done until now is of a preliminary nature, and was carried out in Spain. Although we used limestone taken from near the Tell Halula site to make the experimental tool, we are aware that this limestone is softer than that which was selected for the archaeological tools. Also, the hoe was used on farming lands in Vizcaya (Spain). Therefore, the experience will have to be repeated in autumn of this year at Tell Halula, during the excavation season. Despite the preliminary nature of the experiment, it has produced interesting results. Once the piece had been made, using percussion and scraping with flint blades, bitumen was added to the active side. To do this the bitumen was heated until it became liquid and then the edge of the tool was dipped in it, with the active side facing downward. The piece was fitted with a wooden elbowed shaft and used to work the land, to clear away the vegetable growth on the surface and subsequently dig up the earth. The tool showed itself to be efficient for this task.

In our opinion, from the data presented so far it can be considered highly probable that these tools were used to work the land, but were they used for farming tasks? The houses on the site are made from adobe, in which case they might well have been tools for extracting the clay with which to make the adobe. However, one factor leads us to believe that they were more likely to have been used for farming work. The layer of bitumen and the abrasion marks indicate that the working angle of the tool was very low. This working angle is more apt for work that involves digging over the surface of the earth than for deep digging work, where the implement needs to strike the earth in a more perpendicular way.

Up to the time of writing, no mention has been made of the existence of tools for working the land in such ancient periods. Nevertheless, a number of signs seem to indicate that it was during the PPNB that the hoe made its appearance as a tool for that purpose. On several PPNB sites in the Zagros area, limestone implements of a very similar nature to the ones found at Halula have cropped up (F. Hole, pers. comm.), and these may have had the same function. At the PPNB levels of the Beidha site, some tools have been found that are fashioned from the extremities of cattle bones. The active part, situated on the diaphysis of the bone, is manufactured by making a bevelled cut in the diaphysis. The active areas reveal microscarring, rounding, grooves and use-wear polish that would seem to imply that they had been used to work the land (D. Stordeur, pers. comm.). As regards later periods, stone hoes have been documented on the Tello site, from the Obeid period (M.-C. Cauvin 1979) and perhaps also from the Moroccan Neolithic period (Bensimon and Martineau 1990).

Very few data are available to us concerning the characteristics of land-working tools in the early stages of agriculture. It is probable that, at first, wooden implements such as digging sticks or spades were used. However, as agricultural production intensified and the same fields began to be used year after year, it became necessary for the job of turning the earth over to be done more intensively, to mitigate the effects of the loss of productivity of the soil. In this context it would have made sense for the hoe to make its appearance, since it is a tool that allows the soil to be dug over more intensively and systematically (Sigaut 1978). Archaeobotanical studies carried out at Tell Halula (Willcox and Catalá 1996) point to the appearance of a wide variety and diversity of weeds alongside the crops. This evidence implies that continuous crops were cultivated in the same working fields. So, the appearance of hoes from the PPNB at Tell Halula, and probably also at Beidha, would reflect the growing intensification of farming work.


The study of tool culture among the Neolithic human communities provides information on the first agricultural techniques and implements. In fact, the development of the tools used reflects the growing importance of agriculture within the group economy between the 10th and 8th millennia BP.

During the PPNA, sickle material is relatively abundant. The generalized use of this implement for gathering wild cereals must have been motivated by a desire to get the maximum profit from this resource. Compared with other harvesting systems such as pounding with sticks or collecting by hand, the sickle made it possible to harvest more over a relatively short period of time. The speed of the work must have been a key factor when selecting the technical process for the exploitation of wild cereals, since the time period during which harvesting can be carried out is very short.

A development can be seen in sickle morphology, with a tendency toward greater working efficiency. It is difficult to establish precisely when curved-shafted sickles began to be used, though they were used, at least, in the Late PPNB at Tell Halula, and even somewhat before at Nahal Hemar. The adoption of the curved shaft and the appearance of obliquely inserted flint elements permitted the development of more continuous and efficient harvesting techniques.

Use-wear polish on tools used for harvesting, which becomes more and more intense and striated, reflects the progressive tendency to cut cereals when more ripe, as the process of domestication of these plants occurred. Reaping work progressively intensified, whereas in the more recent periods, ground-level harvesting would tend to become more important since it yielded a considerable amount of straw for roofing, cattle feed or bedding.

The data available so far seem to point to the existence of hoes during the PPNB. The intensification of agricultural work, repeated year after year on the same piece of land, would seem to imply an increasingly marked need for digging over the soil, and this was undertaken with the generalized use of hoes.


Anderson, P.C. 1992. Experimental cultivation, harvest and threshing of wild cereals and their relevance for interpreting the use of Epipalaeolithic and Neolithic artifacts. Pages 179-209 in Préhistoire de l'agriculture: nouvelles approches expérimentales et ethnographiques. Monographie du CRA No. 6, Centre de Recherches Archéologiques (P.C. Anderson, ed.). CNRS, Paris, France.

Anderson-Gerfaud, P.C. 1988. Using prehistoric stone tools to harvest cultivated wild cereals: preliminary observations of traces and impact. Pages 175-195 in Industries lithiques, tracéologie et technologie (S. Beyries, ed.). BAR International Series 411, Oxford, UK.

Bar-Yosef, O. and D. Alon. 1988. Nahal Hemar Cave. Atiqot [English Series], Jerusalem, Israel.

Bensimon, Y. and M. Martineau. 1990. Une houe en silex dans le Néolithique Marocain. L'Anthropologie 94 (3):587-590.

Cauvin, J. 1979. Les fouilles de Mureybet (1971-1974) et leur signification pour les origines de la sédentarisation au Proche-Orient. Bull. Am. School Oriental Res. 44:19-48.

Cauvin, M.-C. 1973. Problèmes d'enmanchement des faucilles au Proche-Orient. Paléorient 1:103-106 [in French].

Cauvin, M.-C. 1979. Tello et l'origine de la houe au Proche-Orient. Paléorient 5:193-206.

Cauvin, M.-C. 1983. Les faucilles prehistoriques du Proche-Orient. Données morphologiques et fonctionnelles. Paléorient 9(l):63-79.

González, J.E., J.J. Ibañez, L. Peña-Chocarro, B. Gavilan and J.C. Vera. 1997. Harvesting tasks in the Neolithic levels of 'Los Murcielagos' Cave: An archeobotanical and functional approach. Helinium 34 (in press).

Harlan, J. 1992. Wild grass seed harvesting and implications for domestication. Pages 21-28 in Préhistoire de l'agriculture: nouvelles approches expérimentales et ethnographiques. Monographie du CRA No. 6, Centre de Recherches Archéologiques (P.C. Anderson, ed.). CNRS, Paris, France.

Hillman, G.C. and M.S. Davies. 1992. Domestication rates in wild wheats and barley under primitive cultivation: preliminary results and archaeological implications of field measurements of selection coefficient. Pages 113-158 in Préhistoire de l'agriculture: nouvelles approches expérimentales et ethnographiques. Monographie du CRA No. 6, Centre de Recherches Archéologiques (P.C. Anderson, ed.). CNRS, Paris, France.

Mellaart, J. 1963. Excavations at Catal Hüyük. Anatolian Studies 13:43-103.

Mellaart, J. 1970. Excavations at Hacilar. British Institute of Archaeology, Ankara, Turkey.

Molist, M.M. 1996. Tell Halula (Syria). Un yacimiento neolítico del valle medio del Eufrates. Campañas de 1991 y 1992. Ministerio de Educación y Cultura, Madrid, Spain [in Spanish].

Neuville, R. 1951. Le Paléolothique et le Mésolithique du désert de Judée. Archives de l'Institut de Paléontologie Humaine 24. Masson, Paris, France.

Peña-Chocarro, L. 1996. In situ conservation of hulled wheats species: the case of Spain. Pages 129-146 in Hulled Wheats. Promoting the Conservation and Use of Underutilized and Neglected Crops. 4. Proc. of the First International Workshop on Hulled Wheats, 21-22 July 1995, Castelvecchio Pascoli, Tuscany, Italy (S. Padulosi, K. Hammer and J. Heller, eds.). IPGRI, Rome, Italy.

Sigaut, F. 1978. Identification des techniques de récolte des graines alimentaires. J. d'agric. traditionnelle et de botanique appliquée 24(3):145-161.

Sigaut, F. 1996. Crops, Techniques and Affordances. Pages 417-436 in Redefining Nature, Ecology, Culture and Domestication (R. Elben and K. Fukui, eds.). Oxford, UK.

Turville-Petre, F. 1932. Excavations in the Mugharet el-Kebarah. J. Royal Anthropol. Inst. 62:271- 276.

Unger-Hamilton, R. 1989. Epipalaeolithic Palestine and the beginnings of plant cultivation: the evidence from harvesting experiments and microwear analysis. Curr. Anthropol. 30:88-103.

Unger-Hamilton, R. 1992. Experiments in harvesting wild cereals and other plants. Pages 211- 224 in Préhistoire de l'agriculture: nouvelles approches expérimentales et ethnographiques. Monographie du CRA No. 6, Centre de Recherches Archéologiques (PC. Anderson, ed.). CNRS, Paris, France.

Willcox, G. and M. Catalá. 1996. Análisis paleobotánico. Pages 135-142 in Tell Halula (Siria) un yacimiento Neolítico del valle medio del Eufrates campañas de 1991 y 1992. Informes arqueológicos (M. Molist, ed.). Instituto del Patrimonio Histórico Español, ministerio de Cultura, Madrid, Spain [in Spanish].

History of Harvesting and Threshing Techniques for Cereals in the Prehistoric Near East - P.C. Anderson


This paper looks at evidence for harvesting and processing methods from two periods: when wild cereals were used in archaeological sites, and after the appearance of morphologically domestic cereal grains. We discuss whether these data show 'agriculture' to be present and evolving, increasing in quantity across these periods, or whether agricultural techniques make a rather more sudden appearance in a defined context. The dynamics of change leading to agriculture are often depicted as bipolar: a progressive reversal over time in dependency on food from the wild vs. food obtained through cultivation, and in production of wild plant food vs. crops (Harris 1996). In Harris' model the “wild plant food procurement” stage is followed by a lengthy and widespread stage of “wild plant food production”, encompassing the duration of the domestication process. Similarly, Hillman and Davies (1992) present a model suggesting a specific behavioral pathway to domestication of the wild cereal population by a human group, which breaks down this phase into several theoretical scenarios of cultivation of morphologically wild cereals. However, like Harris, they assume that tilling of the soil and sowing, the same basic techniques as for domestic cereals, were regularly and increasingly occurring.

My experience of ten years or so of cultivation experiments harvesting wild cereals in Syria and Turkey, and cultivating them in France, indicates that agricultural practices described above may have been used in exceptional circumstances, but make little sense as habitual methods for wild cereal exploitation. Differences between wild and domestic cereals impose use of different effective exploitation techniques for each. Wild cereals with shattering rachis and uneven-ripening habits tend to re-seed spontaneously, as enough spikelets always seem to escape harvest to regenerate the stand, unlike domesticated cereals, which owing to their semi-solid or solid rachis, and even-ripening habit, need to be perpetuated by cultivation. What could be the reason for regular practice of cultivation of wild cereals, when they could be harvested year after year from wild stands, without sowing? It would appear that efficient, non-agricultural strategies of exploitation and management may well have been used over most of the time of exploitation of wild cereals, much as ethnoarchaeological data shows that management systems of wild plants important in the subsistence base today use intensive and often sophisticated exploitation patterns over very long periods, without cultivation (see Ertug, this volume). Nonetheless this point continues to be overlooked; for example, Sheratt (1997) reviews arguments for climatic change as important in precipitating 'farming' of morphologically wild cereals, but assumes this would be the chosen form of intensive exploitation of these plants. Leach (1997), however, has correctly emphasized the lack of agreement on terminology and concepts concerning agricultural origins and food production systems. Did an 'agricultural threshold' actually occur during the period of wild cereal exploitation, or rather only later in time, near the beginning of cultivation of morphologically domestic cereals?

Material and methods

This research uses data that show which techniques and tools were used to harvest and treat cereals from 12,000 to 5000 BP (see Table 1). Using a system of 'feed-back', we have moved between research questions dictated by archaeological material, experiments in the field, and interpretation of archaeological material from the field data. The archaeological material we studied consists mainly of stone (and rarely, bone) tools made at the sites, which potentially were (parts of) plant-working tools. Wood and other organic material, often forming the handle or the body of tools, is rarely preserved. Therefore overall structure of instruments must be deduced from traces of use, residues of adhesive, and other kinds of analogy derived from ethnographic and experimental data. Three basic categories of tools will be discussed here: sickles for harvest, and two other tools for post-harvest treatment or 'threshing'.

We have also checked pieces of daub and mud brick from structures in the sites, where available, for imprints of stem and glume fragments which would have been used to temper the clay. We have studied such data from about 40 archaeological sites from the Natufian to the Bronze Age in the Near East thus far, spanning the time from the first use of wild cereals, to village and urban agricultural systems (see Table 1 for examples), as part of an ongoing research program. Over the years we have used experimental copies of prehistoric tools (Anderson-Gerfaud et al. 1991), to harvest or to thresh, chop, etc. wild cereals, or domestic cereals, as appropriate, given the nature of archaeobotanical remains from the sites having yielded the original tools (Table 1). These experiments produce data concerning either the effect on plants and tools of a given process, or actual traces of use on the tool's active area. Such data are necessary to demonstrate ancient tool use. Initially inspired by Harlan's (1967) wild cereal harvesting experiments in Turkey, we have harvested stands of wild wheats and barley in Syria, in the Jebel Arab region in the south, and near Ain el Arab in the north, and wild wheat and rye in southeastern Anatolia.

We also planted a collection of wild einkorn wheat which was collected in 1986 from 'primary' stands in southeastern Anatolia, found near sites in Turkey where early cultivation of domestic einkorn occurred. This wheat formed the basis for eight years of experiments, carried out by myself and George Willcox, in southern France, Jalès, Ardèche region (Anderson-Gerfaud et al. 1991; Anderson 1992), in order to understand possible methods of exploitation for wild einkorn in prehistoric contexts. Morphologically domestic wheats and barley were also used in experiments, and here interpretations of tools and processes from the past were helped by observing traditional practices in present-day Syria and Turkey. Table 1 shows presence of three kinds of tools we identified, in relation to their chronological and archaeobotanical contexts.

Table 1. (Facing page) Cereal treatment methods from the Natufian to the Bronze Age. ina = Information not available.? = Presence not certain. Blank = Plant, tool or attribute not found at site. Square: wild cereal remains identified. Circle = remains of domestic cereal grains identified. Triangle: presence of sickle use for harvesting of cereals identified in site from use-traces on blades. Diamond: Evidence of threshing using tools identified in sites from use-traces. Solid diamond: Bronze Age threshing sledge with Canaanean blades (structure described in footnote C). Last column: from examination of remains of mud brick or of phytoliths.

Most plant-harvesting and processing tools from archaeological sites are preserved in the form of flint blades (sometimes obsidian). If they show microscopic traces of use, they may be interpreted according to our experiments, which allow differentiation of the actual use of the ancient tools (motion, material, etc.). The tool surfaces are observed using reflected-light optical microscopy at 100 and 200 X (Keeley 1980) and compared with a large data bank of potential characteristic traces. These kinds of observations of wear traces in optical microscopy are sometimes supplemented by identification of residues of the worked material preserved on the tools.

The scanning electron microscope and microanalysis have often been used to identify residues on the basis of characteristic morphology and elemental composition (Anderson 1980; Stordeur and Anderson-Gerfaud 1985; Anderson-Gerfaud 1986; Anderson and Formenti 1996). Other methods, not discussed here, are increasingly used to identify organic compounds preserved as residues, particularly for finding what material was processed with querns and other 'groundstone' objects (Juan-Tresseras 1997; Formenti and Procopiou 1998). Here we discuss only tools which were part of instruments for processing cereals, excluding grinding and pounding tools, but these are only a part of the overall picture of tool use and activities at this time, which includes function for hunting and butchery, for harvesting reeds, or for making objects of skin, wood and stone (Anderson 1994a; Anderson and Formenti 1996), etc.

Discussion and results

Harvesting tools

Method of harvest and field situation

Beginning in about 10,000 BC or 12,000 BP, each site (with the exception of those from the 5th millenium BP) has produced data showing numerous examples of blades (Fig. 1), undoubtedly mounted in handles (Fig. 2), which were used for harvesting cereals. These form the first of three categories of tool function that will be discussed here, whose occurrence is shown in Table 1. Sickles were generally used to harvest by cutting groups of cereal stems, pulling the tool blade back through the stems toward the user, cutting near the ground (Fig. 3). A visible 'gloss' on the cutting edge of such tools, seen using the optical microscope, shows features characteristic of harvesting: smooth, bright surface traversed by fine, short scratches or striations, probably from contact with soil grains, aligned parallel to the cutting edge (Fig. 4).

Harvesting tools identified (Table 1) in each case provide positive evidence for the use-situation described above, as well as for harvesting cereals at a precise time in their ripeness, and where groups of stems occurred in a density of over 300/m2, judging from our experiments. This is the density we observed of wild cereals growing in natural, rather pure stands in eastern Turkey and Syria (Fig. 5), only 30 km from a site in the North Euphrates basin where wild cereals were used. Wild cereals are uneven-ripening, and are most effectively harvested during a period lasting only a few days at a given altitude and latitude, when seeds are still unripe and top spikelets are just beginning to dry and shatter (Fig. 6). Blades in sickles, used for cutting plant stems at this phase of grain ripeness, developed the microscopic traces having a bright, smooth appearance, described above.

Fig. 1. Flint blade, with gloss on edge and use-traces from harvesting, from Syrian Neolithic site with remains of wild cereal. Scale = 1 cm.

Fig. 2. Reconstruction of sickle used in experiments to harvest wild cereals, with flint blades (e.g. Fig. 1), glued in a groove of a wooden handle using bitumen.

Fig. 3. Harvesting of wild wheat in stands in the Jebel Arab, Syria, using the experimental sickle (Fig. 2).

Fig. 4. Magnification (100X) of the cutting edge of a blade in sickle in Figures 2 and 3. Traces of harvest include smooth bright surface and short striations (arrows) oriented parallel to the cutting edge direction.

Fig. 5. Relic stand of wild wheat near Ain al Arab, Syria, about 30 km from D'jade, middle Neolithic site with wild cereal remains, which may have extended nearer sites at that time. Note its density and 'purity' (lack of adventices or other plants).

Fig. 6. Uneven ripening habit of wild einkorn wheat, showing shattering of upper spikelets of seed head, which escape the harvester and self-seed to spontaneously regenerate the stand the following year.

Fig. 7. Reconstructed use of a bone tool (see Table 1) to separate ears of domestic cereal (here, emmer wheat) from the stem. The procedure also worked for einkorn and barley.

Fig. 8. Microwear traces (100X) seen on edge of blade from traditional threshing sled from Turkey (Fig. 9), showing heavy abrasion of the flint surface with large grooves and pits (arrow: comet-shaped) and bright areas, characteristic of dragging blade under pressure over threshing floor covered with sheaves of cereal (Fig. 10).

The polishing of the blade edge producing this appearance is influenced by the action of water as a lubricant, because at this time, stems are green to half-green, and still contain significant amounts of humidity, according to our measurements (Anderson 1992). The striations correspond to soil grains on stems rolling over the tool edge during harvesting use, as well as perhaps abrasion from silica phytoliths contained in the stem epidermis (Anderson 1980). The longer the use, the more microscopic 'polish', marked by scratches and striations, becomes visible. Using these criteria, seen on experimental sickle inserts we used from 5 to 15 hours, most sickle inserts from archaeological sites with wild cereals represent harvesting of over 20 hours, which cannot for various reasons be converted into an estimate of surface area harvested.

Although harvesting wild cereal with a sickle was unlikely to have been the only method used (pulling up the plant or pulling off the seed head being preferable methods in sparser stands or when the plants are ripe and shattering), it is the only one leaving behind recognizable traces of tools. This means that it is impossible for us to compare quantities of cereals harvested at different periods just by counting numbers of sickles found in sites. However, the presence of sickle inserts in nearly all the periods discussed shows that harvesting in the manner described above must have been a very common practice, as only repetitive actions carried out over a period of time would reach our levels of detection today, given selective abandonment patterns, preservation and recovery represented by the archaeological contexts.

The importance attributed to sickles and their use from the beginning, during the Natufian, is shown by the fact that the blades show standardized morphology, undoubtedly to ensure calibration of multiple blades assembled end to end in grooves cut in handles and glued with adhesives. Experimental attempts to use similar tools underline the importance of balance between components of the sickle if it is to function over long periods of time (Anderson 1994a, 1994b). Their presence shows that durable and efficient tools were used in the same technological process over millenia, representing a technological continuity, as well as attesting to harvest in dense stands, even during the time when archaeobotanical remains correspond to use of morphologically wild cereals.

Sowing and tilling

We observed that simple harvesting of natural stands (or of fields left to re-seed spontaneously) stimulates their growth. The idea that harvesting required re-sowing of grain to replenish the field is contradicted by experiments and observations in Syria showing that self-seeding was always adequate to produce a dense new field the following year, however efficient the harvesting was in taking grain (Anderson-Gerfaud et al. 1991). Therefore it is difficult to imagine how any cultivation of wild cereals could have been frequent and widespread enough to constitute an evolutionary phase of wild plant food production, as most models propose. Other exploitation patterns can involve intensive use and even displacement, however. We were able to harvest stands after they were moved to and planted in southern France from Turkey, without turning to annual cultivation to enhance their propagation. This could simulate a situation occurring during the Natufian or the Neolithic whereby wild cereals were taken from original stands, to be transplanted as a 'new' stand nearer human habitations, but without further need for maintainance by cultivation (planting, tilling) after their initial installation in the new area, particularly as various data show that sites are already sedentary at this time (Cauvin 1994).

Other data

Archaeological data presently available do not appear to resolve the question of whether regular cultivation of wild cereals occurred. Although hoe-shaped stone tools were found in sites where wild cereal grain occurred, study of traces of use shows they were used as adzes for woodworking, not for working the ground (Coqueugniot 1983). No tools for working the soil have been found for which the study of traces demonstrates their use, before irrigation contexts in the 5th millenium BP, such as Tello (Cauvin 1979), although a possible hoe made of limestone was found at Halula (Ibáñez et al., this volume; excavations M. Molist) which nonetheless involves a domestic cereal context, not one associated with wild cereals.

Two persons who studied stone tool function by observation of microscopic wear traces - Unger-Hamilton (1989) and Korobkova (1994) - saw traces on harvesting tools that led them to deduce that cultivation (working of the ground) occurred as early as the Natufian. They claimed that the striation or scratching they saw on flint blades with traces of harvesting was caused by contact with soil which was loosened, by tilling of ground for cultivation of wild cereals, to an extent that affected the future plant stems during harvest one year later, in a different way from plant stems of wild cereals growing in a stand. This theory was disproved by our experiments, where we harvested wild cereals on a larger scale, over both tilled and untilled ground in the same environment, as well as in various natural stands in the Near East, as described above. All of the above produced striations on the flint and obsidian tool edges, but amount or density of striations increased only the nearer the ground the tool cut and the longer it was used. Therefore no correlation was found as to whether or not the ground had been worked or the plants had grown in natural stands, which discounts striations on stone sickles as criteria able to address the question of whether or not pre-domestic cultivation occurred (Anderson 1992, 1994a, 1994b; Anderson-Gerfaud et al. 1991).

Pre-domestic cultivation has been postulated by some where seeds of plants found in archaeological sites along with morphologically wild cereal grains could be adventices or weeds brought in by soil disturbance, possibly representing tilling (see Willcox, this volume). Colledge (this volume) proposes that changes in macrobotanical remains may reflect changes in field composition. We believe that these correspond to spatial management and tending strategies, or occasional transplanting or displacement of wild cereals, as opposed to annual preparation of fields and sowing.

Unconscious selection toward domestication

Using calculations from genetic data and limited field trials, Hillman and Davies (1992) established that domestication could have occurred due to repeated use of particular harvesting and cultivation methods leading to an unconscious selection for 'domestic-type' mutants, the rare cereals with a non-fragile rachis hidden among the spontaneous, fragile-rachis wild cereal population. This is because the methods, harvesting with sickles (or by pulling up the plants), followed by annual sowing of the grains gathered that year and annual re-tilling of the previous year's field (or movement away from it), would have the effect of, first, selecting for all grains present of 'mutant-type' plants which do not spill grain, and leaving behind in the field grain spilled from true 'wild' cereals, and second, preventing mixing of this harvest, increasingly consisting of mutants, with earlier crops or with wild stands. In this way, if this occurred annually over 20-200 years, selection for morphologically domestic (non-fragile-rachis) cereal would occur; the length of time needed was later extended by Willcox's (1991) estimates from our experiments. Aside from the estimated duration of the process, these practices, including sowing of wild grain away from prior fields or tilling them under, would need to occur without interruption yearly, otherwise the selection effect would be erased and the domestication process halted.

It appears unlikely that the particular combination of practices described by Hillman is characteristic of contemporaneous sedentary groups for which archaeobotanical data are preserved, such as those in Table 1. Indeed, no site known today shows a sequence of wild cereal followed by domestic cereal remains. Thus domestication's significant and apparently irreversible impact on the evolutionary process may have actually involved an isolated occurrence of regular use of an unusual combination of methods, over a limited time, near the end of the period of wild cereal exploitation. If these practices were used by a group moving regularly with wild cereal grain, in such a way that they distanced themselves from former fields or natural stands, and planted only the products of its last harvest each year, their yearly harvests would constitute greater and greater amounts of 'mutant' domestic-type grain, leading to domestication of the cereal population (Anderson 1992; Hillman and Davies 1992). The domestic grain would then spread rapidly throughout the area, and our data have shown that new techniques and instruments for cereal-processing appear at this time, adding to harvesting tools.

Domesticated cereals: harvest

We consider that sickles, about the only remains bearing witness to harvesting methods, show technical and functional continuity throughout the period discussed here, because the harvesting motion described above for wild cereals continues throughout the Neolithic and beyond for harvest of domestic cereal (Anderson 1994b, 1994c) with changes which occur in sickle morphology, for example increased length of blades making up the edge, or more curvature of the handle (see figure in Table 1), which undoubtedly improved their performance. A difference is that the time of harvest is open to greater social choice, as the ripening pattern of domestic cereals will not involve disarticulation of the spikelets and grain loss over a relatively short period, although seed heads of einkorn and emmer have been seen to disarticulate when very dry (Miller 1992). The appearance of traces on the blades from sites with domestic cereal remains is more variable than those from harvesting of wild cereals, corresponding to the greater choice of harvesting periods (up until the plant is very dry and brittle) where there is more or less humidity in the stems, which we have seen affects development of the 'polish' traces. Otherwise, as a technical process, the motion and manner of use show no significant difference from the prior period, and for that matter, from harvesting motions with curved metal sickles we have seen used today in Syria and Turkey.

Domesticated cereals: threshing

Contrary to observations above concerning harvesting tools and their use, changes occur in technical processes and tools used for cereal treatment at the moment of exploitation of morphologically domestic cereals. Furthermore, these new techniques and tools do not appear to result from a gradual evolutionary process having its origins in wild cereal exploitation, but rather, represent significant technical innovation (Anderson 1994c). Threshing, used here to encompass all treatment after harvest of the plant other than breaking and grinding grain, corresponds to adoption of agricultural processes in the context of the first morphologically domestic cereals. This is apparently related to the change of cereals to domestic morphology (semi-solid or solid rachis from fragile rachis) and to the need, using tools, to produce certain products on a large scale at the site, for dietary and technical purposes (i.e. threshed grain, long length of stems, chopped stems). Wild cereals shatter into spikelets as they dry in the sun, needing no special tools for separation of grain from stems, whereas morphologically domestic cereals do in fact need manual intervention of some sort to separate the seed head from the culm or stem.

A tool has been identified, associated with the earliest dates for domestication of two-row hulled barley in the Kermanshah region in Iran, from the 8th millenium BP, which was used to strip seed heads from stems. These peculiar bone tools have a notched, V-shaped working area produced on scapulas of sheep and goats (see Table 1). About 100 of these tools were identified, of which several were studied in detail for function and use-traces (Stordeur and Anderson-Gerfaud 1985). These tools show gloss in used areas within notches, on straight areas of the same edges which, under the microscope, showed gloss but also striations oriented radially in the notches and parallel to the V-shaped edges. These traces of use are sufficiently characteristic of a certain motion that the use of the tool was able to be narrowed down. Use-traces on the ancient tools (the striation pattern above, but also gloss, and silica phytolith residues from grasses/cereals identified on the tools) were reproduced on the modern, experimental tools by pulling groups of stems, after harvest, through the notched V in the tools, producing clean separation of seed heads from stems (Fig. 7).

Domestic hulled barley appeared to be the most likely silica-rich plant processed using this tool type of those found on or around the site, and experiments showed that many thousands of barley stems would have been processed by each tool to produce the high degree of use-wear observed on the tools found at the site. Also, experiments made it clear the tool was indeed for cereal treatment, not harvest, because harvesting of barley in the field did not create the same wear traces as those on the archaeological tools. Other tools of this shape have been seen by the author at Cayönü in Turkey and reported from Tepe Abdul Hosein, Iran, and study of their use-traces may verify they also had this use. Study of use-traces (Skakun 1993) showed that a very similar tool was used in Bulgaria during the Chalcolithic, or 5th millenium BP, perhaps representing a 'diffusion' of this tool use from the Near East over several thousand years as part of a toolkit associated with early domestic agriculture. Finally, a scapula with a shallower and narrower groove cut than for our archaeological tools is still in use today in Japan for stripping seeds (not whole seed heads) from sorghum after harvest (F. Sigaut and E. Takei, pers. comm.).

The threshing sledge or tribulum

Certain blades, associated with only morphologically domestic cereals, also show gloss, but under the microscope their traces exclude use for harvesting, as they show other features, e.g. abrasion and long deep grooves of comet shape, long striations corresponding to a continuous motion in one direction (Anderson and Inizan 1994), that are characteristic of use as inserts in instruments functioning as threshing sledges or 'tribulums' (Fig. 8). The latter have been described by several researchers who have looked at use of these instruments and traces on their flint inserts (Whallon 1978; Ataman 1992; Skakun 1992; Anderson and Inizan 1994; Kardulius and Yerkes 1996). Such wear is apparently due to contact, by dragging, with dry cereal stems and seed heads, but also with a threshing floor surface of clay or stone. This instrument achieves threshing of grain and chopping of straw on a threshing floor, and is pulled by various animals, depending upon the context - oxen, horses or donkeys. Its overall construction is variable, but today and in the recent past it usually consisted of joined planks forming a board, with cutting elements of stone or metal inserted into the surface of its underside by hammering or gluing into holes, grooves or slots. Tribulums or threshing sledges, of various dimensions, have been used until recently (or are still used) throughout the Mediterranean region. Often the instrument is weighted during use by a person or stones.

The tribulum, armed with anything from 50 to 2000 fragments of stone on its underside (Fig. 9), is pulled around in a circle over the plant material, which for hulled grain breaks up the seed head into spikelets or actually releases the grain from naked cereal, as well as chopping stems into fragments (Fig. 10). All this material is then separated by winnowing and used for various purposes. Today, as in the past, these uses include temper for mud brick (attested to in the past from imprints or phytolith remains from stems in remains of clay walls of structures, see Table 1, last column). Other uses include food and bedding for animals, fuel, etc. Obviously threshing of grain accomplished by the instrument is essential to both human and animal food, but arguably this can be accomplished by any means of producing pressure and friction, such as trampling with animal hooves. The threshing sled, armed with sharp stone, would be important in chopping of straw for various uses.

Our clearest archaeological picture of the use of this instrument is for sites from the Bronze Age in the Levant (Table 1, second-last column). However, according to cuneiform texts describing the instrument of this time in Mesopotamia (Anderson and Inizan 1994; Civil 1994; J.-P. Grégoire, pers. comm.), it shows major differences with the construction of the present-day tool, involving an assemblage of planks or boards, because it was made like a raft, of logs of small diameter which were assembled by lashing with leather straps, with bitumen used to glue the blades between the interstices in the logs (Fig. 11). The blades with traces characteristic of use as inserts dating from the Bronze Age (Fig. 12), are large, purposely fragmented Canaanean blades which are standardized in width and thickness, and for which the hundreds we have seen (Anderson 1994c) all carry the same kind of traces (Fig. 13). These blades often have thick deposits of bitumen (tar) adhering to them, preserving the imprint of the wood, in a few cases of insertion between logs. From 50 to 80 blades are mentioned in one text as making up the instrument, which corresponds well to the large size of the blades from the period having these traces. We have reconstructed this instrument (Anderson 1994c) and tested it over two seasons, and found its use does produce traces comparable to those on the blades from the Bronze Age. Flint inserts in modern sledges may become greatly rounded, unlike archaeological ones, because they contact the threshing floor directly at some points, or may be protected, today by wheels added to the sides of the plank which lift the blades up slightly off the threshing floor. Our reconstruction used wooden 'skis' to prevent blades from wearing too quickly on the threshing floor (Fig. 14).

Fig. 9. Underside of threshing sled from Turkey, showing cutting edges of flint (Fig. 8) inserted in grooves in the two assembled planks.

Fig. 10. Example of use of the threshing sled or tribulum in Turkey. Note weighting of board, pulled by oxen in a circular motion over cereals spread on a clay threshing floor, to chop straw and thresh grain.

Fig. 11. Reconstruction of threshing sled used in the Bronze Age, from descriptions in cuneiform texts. Small logs are lashed together with leather straps and blades (50) are set between the logs in rows, and fixed using a mixture of bitumen and fine sand. 'Skis' were added to create a small gap between the blade edges and threshing floor, both allowing efficient evacuation of chopped straw and threshed grain from under the sled, and reducing wear formation on blade edges.

Fig. 12. Both sides of a Canaanean blade fragmented and used in a Bronze Age threshing sled, according to use-traces (Fig. 13). Note gloss and bitumen traces corresponding to the set of the blade in the instrument. Wear on the edge is slightly toothed.

Fig. 13. Use-traces on the edge of a blade section like that in Fig. 12. Traces are like those in Fig. 8, with depressions including comet-shapes traces (arrows). Smooth bright areas show greater contact with plant material than with the threshing floor, corresponding to the results of the experiments (Figs. 11, 14).

Fig. 14. Experimentally made Bronze Age threshing sled being pulled over clay threshing floor covered with sheaves of bread wheat. Only stones were needed to weigh down this relatively small instrument for efficient chopping and threshing of the plant material. The photograph in the last column in Table 1 shows chopped straw on threshing floor after this experiment.

Fig. 15. Fragment of mud brick from a Neolithic site with domestic agriculture, showing imprints (arrow) of finely chopped straw used to temper the clay. This straw was probably chopped using an early form of threshing sled, then winnowed. Blades with wear similar to those in the experiment are found in the site.

Fig. 16. Silica phytolith from a cereal glume, magnified 400X, from an ashy deposit of chopped cereal straw and chaff, found in a building structure in a late Neolithic site in North Euphrates, Syria. Note sharp cut (arrow) of the silica sheet of phytoliths, very common in this deposit, which is neither a natural break nor the usual way these phytolith sheets break up in the soil. This cut corresponds probably to cutting with a blade set in a threshing instrument. Phytoliths cut in this way have been found in deposits on threshing floors after use of the tribulum, and the frequence of this observation at the site suggests this deposit represents plant material chopped using some early form of threshing sled (see Fig. 14).

We found that a quantity of sheaves harvested in 30 hours with a sickle, stacked on the threshing floor, was threshed and chopped by this relatively small instrument in only 3 hours! The product (figure in last column of Table 1) not only shows perfect threshing of the grain, but also corresponds to sizes of chopped straw we have seen used to temper mud brick constructions from archaeological sites. Table 1 shows that none of the stone tools examined from sites from the Bronze Age (5th millenium BP) carries traces indicating use for harvesting. Harvesting may have been accomplished using metal sickles, relatively rare objects at the time, which texts show were stored in central areas or melted down, so we would not expect to find them left on site from one season to the next (J.-P. Grégoire, pers. comm.). Alternately, plants were pulled up without using tools, a harvesting technique that we found presented difficulties for threshing and winnowing, but which may be suggested by presence of culm bases among the archaeobotanical remains in one instance (J. McCorriston, pers. comm.).

A Neolithic instrument?

Cereal glumes, not chopped stems, tend to be used as tempering material for clay from structures in sites with wild cereal grain, and in these sites we do not find any tools with traces showing use for chopping of stems. This situation changes for sites with domestic cereals, where temper tends to be of chopped cereal stem fragments, and blades with traces from a tool like a threshing sledge begin to be identified (Table 1). These imprints in mud brick (Fig. 15) sometimes correspond to use of a size fraction of chopped straw which has been sorted (e.g. through winnowing), much as today for the straw threshed mechanically in Syria, where a particular size fraction is used to temper mud brick and other sizes are used for fuel and animal food. Another indication of the use of such an instrument at this time is provided by our analysis of a 1-m2, 20-cm thick deposit, found in the corner of a late Neolithic structure from the site Halula in the Upper Euphrates (Anderson, unpublished). The deposit was made up of cereal phytoliths, not grains, representing burnt pieces of stem and glumes, cut (Fig. 16) in a manner characteristic of processing with the threshing sledge, according to analyses of material left on threshing floors after use of a threshing sled in the traditional way (Juan-Tresseras 1997) and our experiments described above (Fig. 14). The deposit, overlain by woven stem material, may correspond to a pillow or mattress and its stuffing, or to an area where chopped fragments were stored (in a bag?) for future use.

Indeed, a variety of uses of plant material processed by the tribulum is attested or likely, but it is not known whether this new tool was initially conceived to process increased amounts of grain, for human and animal consumption, or (also) to deal with a need for massive quantities of chopped stems (for architecture of greatly expanded habitation sites, but also as fodder, bedding, etc.). This instrument is pulled by animals traditionally, and the earliest traces of blades which could have armed it, from the late Neolithic, coincide with the earliest domestication of oxen (Anderson 1994c; Anderson and Inizan 1994; D. Helmer, pers. comm.).


Present data do not point to the existence of a universal 'wild plant food production phase' at this time, where humans would have regularly used all the basic agricultural techniques (of tilling, sowing, harvest, re-tilling and sowing) comprising agriculture of morphologically domesticated crops, to exploit morphologically wild cereals. This should not have the negative connotation of pushing the date for the beginning of agriculture ahead, but rather highlight the fact that Neolithic peoples were evidently adapting to and effectively exploiting cereal resources. On the other hand, once domestic agriculture began, new tools for threshing apparently came into use, resembling the traditional threshing sledge or tribulum. They correspond to a great increase in scale, both in quantity of cereal products produced and used, and in the complexity of the sequence of treatment of cereals. These data also serve to underline the ancient origins of the traditional threshing sledge. Appearance of the Neolithic agricultural assemblage in the Middle East is shown using criteria in Table 1, and its travel or diffusion may be mapped using locations where this instrument is attested [blades (Skakun 1992, 1993), remains of plants chopped with it, mud brick imprints, phytolith data, etc.] as well as clearly identified sickle blades, groundstone tools and cereal grain.


I wish to thank the Institut Français de Archéologie au Proche Orient (IFAPO) among others for financial support, and individuals from the Department of Antiquities at Damascus, Derra and Aleppo, Syria, for facilitating and assisting with the ethnographic, archaeological and experimental work. Thanks are also due to Gordon Hillman, Daniel Zohary and Jack Harlan, for their valuable discussions and guidance. Figures 1 and 12 were taken by Benoit Bireaud, Figure 7 by Elizabeth Willcox, and Figures 9 and 10 by George Willcox.


Anderson, P.C. 1980. A testimony of prehistoric tasks: diagnostic residues on stone tool working edges. World Archaeol. 12(2):181-194.

Anderson, P.C. 1992. Experimental cultivation, harvesting and threshing of wild cereals and their relevance for interpreting the use of Epipalaeolithic and Neolithic artefacts. Pages 179-210 in Préhistoire de l'agriculture: nouvelles approches expérimentales et ethnographiques. Monographie du CRA No. 6, Centre de Recherches Archéologiques (P.C. Anderson, ed.). CNRS, Paris, France.

Anderson, P.C. 1994b. Reflections on the significance of two PPN typological classes in light of experimentation and microwear analysis: Flint “sickles” and obsidian “Cayönü tools”. Pages 61-82 in Neolithic Chipped Stone Industries of the Fertile Crescent (H-G. Gebel and S. Koslowski, eds.). Studies in Early Near Eastern Production, Subsistence, and Environment I. Berlin, ex oriente.

Anderson, P.C. 1994c. Interpreting traces of new Near Eastern Neolithic craft activities: An ancestor of the threshing sledge for processing domestic crops? Helinium 34:306-321.

Anderson, P.C. and F. Formenti. 1996. Exploring the use of abraded obsidian “Cayönü tools” using experimentation, optical and SEM microscopy and EDA analysis. Pages 553-566 in Archaeometry 94 (S. Demirci, A.M. Ozer and G.D. Summers, eds.). Tübitak, Ankara, Turkey.

Anderson, P.C. and M-L. Inizan. 1994. Utilisation du tribulum au début du IIIe millénaire: des lames “cananéennes” lustrées à Kutan (Nivive V) dans la région de Mossoul, Iraq. Paléorient 20(2):85-103 [in French].

Anderson, P.C. 1994a. Insights into plant harvesting and other activities at Hatoula, as revealed by microscopic functional analysis of selected chipped stone tools. Pages 277-293 in Le gisement de Hatoula en Judée occidentale, Israël (M. Lechevallier and A. Ronen, eds.). Mémoires et travaux du Centre de Recherche Français de Jérusalem 8, Paris.

Anderson-Gerfaud, P., G. Deraprahamian and G. Willcox. 1991. Les premières cultures de céréales sauvages et domestiques primitives au Proche-Orient néolithique: résultats préliminaires d'expériences à Jalès (Ardèche). Cahiers de l'Euphrate 5-6:191-232 [in French].

Anderson-Gerfaud, P. 1986. A few comments concerning residue analysis of stone plant-processing tools. Pages 69-87 in Technical Aspects of Microwear Studies on Stone Tools (L. Owen and G. Unrath, eds.). Early Man News 9/10/11, Tübingen, Germany.

Ataman, K. 1992. Threshing sledges and archaeology. Pages 305-320 in Préhistoire de l'agriculture: nouvelles approches expérimentales et ethnographiques. Monographie du CRA No. 6, Centre de Recherches Archéologiques (P.C. Anderson, ed.). CNRS, Paris, France.

Cauvin, J. 1994. Naissance des divinités. Naissance de l'agriculture. CNRS Editions, Paris, France [in French].

Cauvin, M.C. 1979. Tello et l'origine de la houe au Proche-Orient. Paléorient 5:193-206 [in French].

Civil, M. 1994. The Farmer's Instructions. A Sumerian Agricultural Manual. Aula Orientalis-Supplementa-5. Ausa, Barcelona, Spain.

Coqueugniot, E. 1983. Analyse tracéologique d'une série de grattoirs et herminettes de Mureybet, Syrie (9e-7e millénaires). Pages 163-172 in Traces d'utilisation sur les outils néolithies du Proche-Orient (M.-C. Cauvin, ed.). Travaux de la Maison de l'Orient no. 5, Lyon, France [in French].

Formenti, F. and H. Procopiou. 1998. Analyse chromatographique de traces d'acides gras sur l'outillage de mouture. Contribution a son interpretation fonctionnelle. Cahiers de l'Euphrate 8 (in press).

Harlan, J.R. 1967. A wild wheat harvest in Turkey. Archaeology 20:197-201.

Harris, D.R. 1996. Domesticatory relationships of people, plants and animals. Pages 437-463 in Redefining Nature: Ecology, Culture and Domestication (R. Ellen and K. Futuki, eds.). Berg, Oxford, UK.

Hillman, G.C. and M.S. Davies. 1992. Domestication rate in wild wheats and barley under primitive cultivation: preliminary results and archaeological implications of field measurements of selection coefficient. Pages 113-158 in Préhistoire de l'agriculture: nouvelles approches expérimentales et ethnographiques. Monographie du CRA No. 6, Centre de Recherches Archéologiques (P.C. Anderson, ed.). CNRS, Paris, France.

Juan-Tresseras, J. 1997. Processado y preparacion de alimentos vegetales para consumo humano. PhD Thesis. University of Barcelona, Spain [in Spanish].

Kardulius, P.N. and R.W. Yerkes. 1996. Microwear and metric analysis of threshing sledge flints from Greece and Cyprus. J. Archaeol. Sci. 23:657-666.

Keeley, L.H. 1980. Experimental determination of tool uses. University of Chicago Press, Chicago, USA.

Korobkova, G.F. 1994. The study of flint tools and the origin of farming in the Near East. Archaeol. News 3:166-180.

Leach, H.L. 1997. The terminology of agricultural origins and food production systems-a horticultural perspective. Antiquity 71/271:155-148.

Miller, T.E. 1992. A cautionary note on the use of morphological characters for recognizing taxa in wheat (genus Triticum). Pages 249-254 in Préhistoire de l'agriculture: nouvelles approches expérimentales et ethnographiques. Monographie du CRA No. 6, Centre de Recherches Archéologiques (P.C. Anderson, ed.). CNRS, Paris, France.

Sheratt, A. 1997. Climatic cycles and behavioural revolutions: the emergence of modem humans and the beginning of farming. Antiquity 71/272:271-287.

Skakun, N.N. 1992. Evolution des techniques agricoles en Bulgarie chalcolithique (d'après les analyses tracéologiques). Pages 271-280 in Préhistoire de l'agriculture: nouvelles approches expérimentales et ethnographiques. Monographie du CRA No. 6, Centre de Recherches Archéologiques (P.C. Anderson, ed.). CNRS, Paris, France.

Skakun, N.N. 1993. Agricultural implements in the Neolithic and Eneolithic of Bulgaria. Pages 361-368 in Traces et fonction: les gestes retrouvés (P.C. Anderson, S. Beyries, M. Otte et H. Plisson, eds.). ERAUL 50, Liege, Belgium [in French].

Stordeur, D. and P. Anderson-Gerfaud. 1985. Les omoplates encochées néolithiques de Ganj Dareh (Iran). Etude morphologique et fonctionnelle. Cahiers de l'Euphrate 4:289-313 [in French].

Unger-Hamilton, R. 1989. Epipalaeolithic Palestine and the beginnings of plant cultivation: the evidence from harvesting experiments and microwear analysis. Curr. Anthropol. 30:88-103.

Whallon, R.J. 1978. Threshing sledge flints. A distinctive pattern of wear. Paléorient 4:319-324.

Willcox, G. 1991. La culture inventée, la domestication inconsciente: le début de l'agriculture au Proche-Orient. Pages 9-31 in Rites et rythmes agraires (M.C. Cauvin, ed.). Travaux de la Maison de l'Orient, Lyon, France [in French].

Problems in Correlating Pollen Diagrams of the Near East: A Preliminary Report - R.T.J. Cappers, S. Bottema and H. Woldring


Studies dealing with the transition from hunting/gathering to farming in the Near East focus on different aspects of this process. The main questions, which are closely related to each other, are why and how this transition took place.

An important contribution to the discussions dealing with these two questions comes from the study of pollen, which can be used for the reconstruction of past vegetation and climatic changes. To gain insight in these environmental aspects, many core samples from the Near East have been investigated during the last decades (Van Zeist and Bottema 1991; Bottema 1993).

The interpretation of pollen diagrams may be seriously hampered by problems in dating, which in turn may have consequences for both the correlation of pollen diagrams with each other (Rossignol-Strick 1995) and with the correlation of diagrams with occupation periods of archaeological sites.

Usually, the interpretation of radiocarbon dates of organic material from archaeological contexts does not give serious difficulties. The reason is that those dates are based on the measurements of 14C that has been incorporated by organisms from the atmosphere. It is even possible to correct for fluctuations of the concentration of 14C over considerable periods by calibration.

Dating of pollen diagrams, on the other hand, is more problematic. First, it is not always possible to obtain a reasonable number of dates from a single core because of insufficient amounts of organic material. This is especially true for the Near East, where most of the sediments consist of clay and, moreover, the organic fractions originate predominantly from the local vegetation. A related problem is that carbon may be incorporated by organisms originating from older sediments.

The concentration of 13C differs in the various parts of the biosphere. This is partly the result of the differential uptake by plants. Organisms discriminate in the uptake of carbon isotopes, with a tendency for the lightest isotope. Thus 12C will be taken up in preference to 13C, and 13C in preference to 14C. This differential uptake is referred to as fractionation and a correction for this process is a standard procedure in calculating the radiocarbon dates. A further correction, however, is possible by taking into account the variation in 13C of different plant types (e.g. submerged and emerged water plants) in which 14C activity can differ from 100% modern carbon (pMC).

Unlike 14C, 13C is a stable isotope, which makes it possible to measure its concentration irrespective of the age of the sediment. The concentration of 13C is expressed as d13C, being the 13C/12C ratio in the sample relative to a standard (Mook 1987).

Another process which can make radiocarbon dates too old is contamination with older carbon-containing material. A correction is possible with the extrapolation method. This method is only applicable if a reasonable number of radiocarbon dates are available, a constant sedimentation rate can be assumed and no sediment is missing.

The correlation of pollen diagrams is hampered not only by problems in dating, but also by the complexity of the data matrices themselves. Pollen diagrams of the Near East represent altogether more than 500 different taxa. Of course, only a part of these taxa are present in a particular diagram. Still, the large number of taxa in a data matrix and the relatively large number of taxa with low counts make it difficult to incorporate all data in a pollen diagram. As a result, taxa which may be of importance because of their ecological information, and the appearance or disappearance of taxa with low percentages, may be poorly represented.

Correspondence analysis (CA) can be used to reduce the complexity of a data set. Carrying out this analysis on pollen data makes it easier to compare fluctuations in the composition of the spectra. Moreover, it is possible to arrange the taxa in a pollen diagram so that the vegetation history can be better visualized (Turner 1986; Birks 1993).

This article presents preliminary results dealing with the correlation of the pollen diagrams of Hula, Ghab and Eski Acigöl (Fig. 1). The analysis is limited to a selection of taxa, in which special emphasis is laid on the tree pollen. The pollen diagrams of Hula and Ghab have been published and discussed by Niklewski and van Zeist (1970), van Zeist and Woldring (1980) and Baruch and Bottema (1991). The publication of the pollen diagram of Eski Acigöl is in preparation (Woldring and Bottema) and the publication of the complete Hula diagram is in press (Baruch and Bottema 1998).

Material and methods

Correspondence analysis

Applied to pollen diagrams, correspondence analysis (CA) assumes a unidimensional model for describing the frequency distribution of each taxon as a function of depth. This relationship between frequency distributions of taxa and depth is assumed to be unimodal. Taxa may dominate at a specific depth, while their presence is decreasing in lower and upper parts of the diagram.

The CA was used to construct a tree curve for each of the pollen diagrams. This curve is based on the CA scores of the first axis. To facilitate the comparison between the three pollen diagrams, the curve for each diagram was calculated on the same scale. Thus, the data matrices were joined together for the analyses.

In first instance the complete pollen diagrams of Eski Acigöl (spectra 1-72), Ghab 1 (spectra 1-74) and Hula (spectra 1-138) were analyzed. Next, it was decided to exclude those parts of the pollen diagrams that were considered to be not relevant to the Pleistocene-Holocene transition. This reduction excluded the lower part of the Ghab 1 diagram (spectra 1-39) and the upper part of the Hula diagram (spectra 116-138).

A further reduction of the data set for this study concerned the exclusion of non-woody plants, with the exception of grasses (Gramineae), chenopods (Chenopodiaceae), mugworts (Artemisia) and members of the sedge family (Cyperaceae).

Fig. 1. Location of Eski Acigöl, Ghab and the Hula sites.

To investigate the influence of differences in dominance, analyses were carried out with both absolute counts and presence/absence scores. A further differentiation was made by analyzing both the complete data set of trees and a reduced data set in which only trees that are present in all three pollen diagrams were taken into account. Some taxa were converted to other taxonomic levels, for example, Quercus cerris-type, Q. robur-type and Q. ithaburensis-type were considered as Q. cerris-type.

Corrections of radiocarbon dates

In addition to the standard correction of radiocarbon dates, a second correction was introduced which was calculated from:


where Am is the measured 14C activity and corresponds with the calculated BP value and Ac is the initial corrected activity and can be described by the equation:

Ac = X · A0 + (1 - X) · 80


in which A0 = 100 (related to emerged plants) and the assumed value of 14C-% = 80 (related to submerged water plants), X is the contribution to emerged lake plants (assumed d13C = -16‰) and (1 - X) is the contribution to submerged water plants (assumed d13C = -34‰). Thus, the measured d13C can be expressed as:

d13C = - 34(1 - X) + -16(X)


and X can be expressed as:

X = (d13C + 34)/18


The radiocarbon date of the core section 129-137 cm from the Ghab diagram (GrN 5810) was obtained from freshwater mollusc shells. Like water plants the molluscs may have incorporated fossil carbon. Therefore, a corrected date of 8580 BP is proposed, based on a reduction of the original activity to 85% (H.J. Streurman, pers. comm.).

With respect to the extrapolation method, a regression line was calculated through the original BP values. Extrapolated to the sediment surface, the intercept gives the deviation in relation to the original date. A correction is not applicable to dates from which the regression line has a negative intercept. This would indicate that the dates concerned are too young, which is not likely. Therefore, with respect to the Hula pollen diagram the extrapolation method was restricted to the five youngest radiocarbon dates (spectra 67-129).


Information on the radiocarbon dates is summarized in Table 1. Because this overview is limited to the pollen spectra that were chosen for comparison, the two youngest dates of the Hula are not presented, although they were used for calculating the regression equation.

If an additional correction for d13C is applied, the radiocarbon dates of Hula are ca. 900 years too old and those of Acigöl ca. 1000 years. Using the extrapolation method, the corrections become even larger: the dates of the upper part of the Hula core are ca. 1350 years too old and all dates of Acigöl are ca. 3850 years too old.

Figure 2 presents a selection of the curves of the pollen diagrams of Hula, Ghab and Acigöl. The arboreal pollen (AP) curves in Fig. 2 show the fluctuations in the proportion of the tree pollen in comparison with the non-arboreal pollen which reflects the alternate expansion and reduction of forest and open vegetation.

The tree curves as calculated with correspondence analysis (CA curves) are shown in Figure 3. Contrary to the AP curves, they express both qualitative and quantitative information. A spectrum with an average composition of trees, as calculated from the data of all the spectra, has value 0 (e.g. solid line in spectrum 4 of the Hula diagram, Fig. 3). If all spectra had exactly the same tree pollen, that is not only the same pollen types but also the same number of pollen from each type, the curve would coincide with the vertical axis which is centralized on the X-axis. The more the composition of a certain spectrum deviates from the mean composition, the more it is projected to the left or to the right. The direction of the projection (i.e. a negative or positive score on the X-axis) has only a relative meaning. Because the data from Hula, Ghab and Acigöl were treated as one file, the scores on the X-axes are comparable.

Table 1. Radiocarbon dates of Hula and Eski Acigöl. Radiocarbon dates are uncalibrated and presented without correction (1), with correction based on d13C (2) and with correction based on linear regression (3).

Lab. no.


Depth (cm)


Age BP (1)

Age BP (2)

Age BP (3)

Hula (spectra 1-115)

GrN 22396







GrN 22397







GrN 22398







GrN 17067







GrN 17068







GrN 14986







GrN 22399







GrN 14463







GrN 22402





GrN 22403





GrN 22404





GrN 22405





Eski Acigöl (spectra 1-72)

GrN 21037







GrN 22881







GrN 21036







GrN 22447







GrN 22605







GrN 21035







GrN 20633







GrN 19988







GrN 22882





GrN 22448





GrN 19989





Fig. 2. Selection of curves from pollen diagrams for the Hula, Ghab and Eski Acigöl sites. (AP = arboreal pollen, GRA = Gramineae, CHE = Chenopodiaceae, ART = Artemisia, CYP = Cyperaceae, CER = Cerealia, CHE/ATR = Chenopodiaceae and Atriplex).

From the four curves that have been calculated using correspondence analyses, only two are presented for each pollen diagram (Fig. 3). The two curves based on presence/absence (dotted lines) fluctuate considerably, indicating that the composition of tree taxa is quite variable. It was therefore decided to present only the curve showing the fluctuation of the 27 taxa that are present in all three diagrams. This curve represents 61% of the total number of trees in Hula, 55% of the trees in Acigöl and 71% of the trees in Ghab. A less fluctuating pattern is shown by the two curves based on the occurrences (straight lines).

Fig. 3. Tree curves calculated with correspondence analysis for the Hula, Ghab and Eski Acigöl sites.

The Ghab diagram takes an intermediate position in terms of the tree pollen composition of the spectra (Fig. 3). Only in the lower part of their diagrams do both Hula and Acigöl resemble the composition of the Ghab diagram to some extent. In the middle and upper part, however, the curves of Hula and Acigöl increasingly diverge from the Ghab curve in an opposite direction, indicating that in both areas a different combination of tree taxa becomes dominant, which can be explained by differences in latitude and altitude.

The pollen diagrams of Hula and Acigöl have been subdivided into five zones based on the curves in Figures 2 and 3, showing a similar pattern in the vegetation development. A short characterization will be given below of each of these zones. The correlation with the Ghab diagram will be discussed in the next section.

Zone 1 is characterized by relatively low tree pollen percentages. Gramineae, Chenopodiaceae, Artemisia and Cyperaceae, on the other hand, have fairly high pollen percentages.

Zone 2 is characterized by increasing AP values to a maximum of 76% in Hula and 56% in Acigöl. In both diagrams Quercus cerris-type is the predominating tree, in which probably Q. ithaburensis accounts for most of the deciduous oak pollen in Hula and probably Q. pubescens and/or robur for those of Acigöl. The increasing values of both Pinus and Juniperus in the second part of this zone in the Acigöl diagram are responsible for the strong fluctuation of the CA curve (Fig. 3). The relatively high percentages of Chenopodiaceae, Artemisia and Cyperaceae drop quickly to low values at Hula and Acigöl (Fig. 2). In Acigöl, where these herbs were represented by higher values, this decrease is compensated by the Gramineae, although trees expand at the expense of grasses toward the end of zone 2.

A decline of tree pollen percentages is shown in zone 3, mainly compensated by Gramineae and in Acigöl also by Artemisia. Again the composition of the tree assemblage is more variable in Acigöl than in the Hula (Fig. 3). Especially Corylus and Cedrus increase in pollen percentages, a process initiated even before the transition to the third zone. In both Hula and Acigöl the percentage of Cerealia pollen increases.

In zone 4 the percentages of the tree pollen increase again. In the Hula Q. ithaburensis-type, Q. calliprinos-type and Olea are dominant. The considerable increase of the last one from the upper half of this zone onwards is responsible for the shift in the CA curve (Fig. 3). In Acigöl this zone is dominated by Q. robur-type, Pinus and Cedrus.

The last zone starts with an increase of the tree pollen percentages in the Hula and with a decrease of those in Acigöl. The CA curve of Acigöl shows that despite fluctuations in tree pollen percentages the composition has been strongly changed. This can be attributed mainly to the increase of Pinus at the expense of Q. robur-type. Pinus also becomes one of the dominant trees in the Hula, though its percentages remain considerable lower than that of Q. ithaburensis-type, Q. calliprinos-type and Olea.


On the basis of radiocarbon dates that have been corrected by the d13C values, the increase of the tree pollen in the Hula diagram can be dated at ca. 14,500 BP (boundary zone 1-2). This increase continues until ca. 11,000 BP, after which the tree pollen starts to decrease until ca. 9500 BP (zone 3). This climatic deterioration could be coupled with the Younger Dryas, though it has to be stressed that this period is less recognizable in pollen diagrams from the Near East than in diagrams of Europe (Bottema 1995). From ca. 9500 to 6100 BP (zone 4) the tree pollen percentages increase only gradually and from ca. 6100 BP onward (zone 5) this increase is more substantial. Taking into account a correction on the basis of the extrapolation method, the gradual increase in zone 4 could have lasted several hundred years longer (Table 2).

This pattern of increase and decrease of tree pollen percentages is also represented in the Acigöl diagram, although the zones distinguished are dated much younger. The boundary between zones 1 and 2 (first increase of tree pollen) is dated ca. 2100-2300 radiocarbon years later. Taking into account the extrapolation method, the discrepancy between Hula and Acigöl is ca. 4900 radiocarbon years for this boundary. And even if the uncorrected dates of Acigöl are compared with the corrected ones of the Hula, there is still a difference of ca. 1000 radiocarbon years.

Table 2. Radiocarbon dates of the boundaries of the five zones that are distinguished in the pollen diagrams of Hula and Eski Acigöl. The standard dates have been corrected on the basis of d13C and the extrapolation method.

BP (standard)

BP (d13C corrected)

BP (extrapolated)


Boundary 4-5




Boundary 3-4




Boundary 2-3




Boundary 1 -2




Eski Acigöl

Boundary 4-5




Boundary 3-4




Boundary 2-3




Boundary 1 -2




Whereas the extrapolation method is responsible for only relatively minor corrections of the radiocarbon dates of the Hula, it seriously affects those of Acigöl. A possible explanation for the larger corrections of the radiocarbon dates of Acigöl could be that, besides the uptake of old carbon by water plants, the sediments also became contaminated with older sediments. Eski Acigöl is a small, recently drained crater lake at the foot of the Karapinar volcano. The older sediments could have originated from the slopes of this volcano.

Another possibility is that volcanic activity was responsible for the age shift. Volcanic gases do contain carbon, but this is too old to be radioactive. Consequently it dilutes the concentration of the carbon isotopes in the surrounding air for which reason local plants are dated 1000 or more radiocarbon years too old. There is, however, scholarly disagreement on the effect of volcanic activity on radiocarbon dates. Despite older dates of recent plants near volcanos, radiocarbon dates for the eruption that destroyed Akrotiri (Santorini, Greece) are surprisingly close to dates given by other techniques (Bowman 1990).

It is also possible that in Eski Acigöl the upper part of the sediment is missing, resulting in a misinterpretation of the intercept. An indication for its absence could be the presence of strongly disintegrated peat at the edge of the drained lake, although this explanation seems not very likely. To make sure of this possibility, additional radiocarbon dates will be obtained from the upper sediment. Also the counting of varves can be used for an additional check on the radiocarbon dates.

The correlation of the Ghab diagram with that of the Hula and Acigöl is problematic because only one radiocarbon date is present in the part of the Ghab core that is under discussion. The dated sediment consists of freshwater mollusc shells which makes an inaccurate dating very likely, as was mentioned before (Niklewski and van Zeist 1970). Without a correction for this radiocarbon date the pattern of the tree pollen curve is conflicting with that of the lower part of the Hula diagram. Using the corrected date of 8580 BP it is possible to coincide the tree pollen curve of Ghab with that of the Hula, especially if a slower sedimentation rate is taken into account for the lower part of the diagram, as was proposed by Hillman (1996). This lower part of the Ghab diagram remains problematic, however, with respect to the percentages of Gramineae, Chenopodiaceae and Artemisia, which have relatively high values comparable with zone 2 and 3 of the Hula diagram.

One radiocarbon date in the upper part of the Ghab diagram offers a high degree of freedom in correlating it with the well-dated Hula diagram. Using the corrected radiocarbon date, it is shown that a correlation is possible. The well-dated Acigöl diagram, on the other hand, demonstrates that a correlation based on the vegetation development can be problematic. A similar development is dated much younger in the Acigöl diagram, indicating that a shift in this development may have occurred in a south-north direction.


The authors thank U. Baruch for permission to include some unpublished curves of the Hula diagram, G. Delger for preparing the map, N. Veldhuizen for his suggestions on the use of the correspondence analysis and H.J. Streurman for information on the correction of the radiocarbon dates.


Baruch, U. and S. Bottema. 1991. Palynological evidence for climatic changes in the Levant ca. 17,000-9,000 B.P. Pages 11-26 in The Natufian Culture in the Levant (O. Bar-Yosef and F.R. Valla, eds.). International Monographs in Prehistory, Archaeological series 1. Ann Arbor, Michigan, USA.

Baruch, U. and S. Bottema. 1998. A new pollen diagram from lake Hula: vegetational, climatic and anthropogenic implications in ancient lakes: Their Biological and Cultural Diversities (H. Kawanabe, ed.) (in press).

Birks, H.J.B. 1993. Quaternary palaeoecology and vegetation science - current contributions and possible future developments. Rev. Palaeobot. & Palynology 79:153-177.

Bottema, S. 1993. The palaeoenvironment of prehistoric man in the Near East: some aspects of palynological research. Japan Rev. 4:129-140.

Bottema, S. 1995. The Younger Dryas in the Eastern Mediterranean. Q. Sci. Rev. 14:883-891

Bowman, S. 1990. Radiocarbon Dating. The Trustees of the British Museum, London, England.

Hillman, G. 1996. Late Pleistocene changes in wild plant-foods available to hunter-gatherers of the northern Fertile Crescent: possible preludes to cereal cultivation. Pages 159-203 in The Origins and Spread of Agriculture and Pastoralism in Eurasia (D.R. Harris, ed.). UCL Press, London, England.

Mook, W.G. 1987. Stabiele isotopen en hun natuurlijk gedrag. Natuurkundige Voordrachten 66:71-79 [in Dutch].

Niklewski, J. and W. van Zeist. 1970. A late Quaternary pollen diagram from northwestern Syria. Ada Bot. Neerl. 19:737-754.

Rossignol-Strick, M. 1995. Sea-land correlation of pollen records in the Eastern Mediterranean for the Glacial-Interglacial transition: biostratigraphy versus radiometric time-scale. Q. Sci. Rev. 14:893-915

Turner, J. 1986. Principal components analyses of pollen data with special reference to anthropogenic indicators. Pages 221-232 in Anthropogenic Indicators in Pollen Diagrams (K.E. Behre, ed.). Balkema, Rotterdam, The Netherlands.

van Zeist, W. and H. Woldring. 1980. Holocene vegetation and climate of northwestern Syria. Palaeohistoria 22:111-125.

van Zeist, W. and S. Bottema. 1991. Late Quaternary vegetation of the Near East. Beihefte zum Tübinger Atlas des vorderen Orients Reihe A (Naturwissenschaften) 18:1-156.

Investigations of Botanical Remains from Nevali Çori PPNB, Turkey: A Short Interim Report - R. Pasternak

The site

Nevali Çori (NÇ) is located near the Turkish town of Urfa (37°60'N, 38°70'E, 490 m above sea level) on the slope of a Euphrates side valley. The 14C analyses, only available from the early levels, provided dates of ca. 8400 BC (calibrated). The settlement was occupied from the Early PPNB to the early Bronze Age (Hauptmann 1984, 1987, 1988, 1997). The PPNB layers included stone houses and a temple building of around 200 m2. On the slope in front of the houses a 'roasting-pit area' was found with a number of hearths in close proximity to each other. Most of these hearths were in use during the earliest phases.

Samples, methods and results

In 1991, 267 samples were taken, with a hand sieve of 0.35 mm, from the PPNB layers. The average sample volume was 10 liters. One-half of the samples were taken from areas on the site that appeared to have good potential for archaeobotanical results. These were essentially areas associated with hearths. The hearths themselves contained only a small number of poorly preserved plant remains. The other half were random samples collected from different parts of the excavation. The material used to fill the space between the stones of the big temple wall contained a lot of charred botanical remains. The composition of remains in these samples was similar to those from the areas associated with the hearths. The results are presented in Table 1. This database is available from the author in electronic form.


Triticum spp., wheats. There were 26,792 items identified as wheat. The wheats in all PPNB layers of NÇ are morphologically domestic. Nearly all spikelet forks show the typical tear-off points of non-brittle spikes.

Problems of determination

On the whole, hulled wheats have distinct phenotypic flexibility. Because the diploid T. boeoticum and the tetraploid T. dicoccoides both produce one - and two-seeded spikelets, the determination of a grain from a one - or two-seeded spikelet gives no information about the species and thus no information about the ploidy level. After extensive comparative research on modern grains and spikelet forks it was decided that, with the exception of the terminal spikelet forks, the ancient material gives no information about the species and the ploidy level.

From the ancient wheat grains, 661 one-seeded and 129 two-seeded grains were recorded. Only a small number of the grains are complete, so measurements were not possible. The approximate size is given in the drawings. There appear to be two different types of grains: small (Fig. 1) and large (Fig. 2). Most of the grains belong to the smaller type. Only 30 grains belong to the larger type. Four of these are obviously two-seeded and could be identified as T. dicoccum, the others as T. monococcum. But I decided that because of the overlap resulting from morphological variation the identification cannot be certain.

Table 1. The results of the analysis of 267 samples of PPNB layers of Nevali Çori (for the counts of finds fragments of seeds, fruits and shells were put together to a unit and then counted as one). All data are available as a dBase 3+ file.

Glume bases/spikelet forks/rachis fragments:







Triticum spp., one-seeded type


Triticum spp., two-seeded type


Spikelet forks of both types


Unidentified cereals







Vicia ervilia


Lathyrus 'sativus'


Vicia faba


cf. Cicer


Unidentified pulses


Fruits and nuts

Pistacia, nut shells


Pistacia, complete




Prunus spp.




Cornus mas





Hordeum distichon/spontaneum


Rachis fragments


Bromus arvensis-type


Bromus, long-seeded


Lolium, small-seeded


Stipa spp.


Lolium cf. temulentum


Aegilops, glume-bases


Phragmites, fragment of culm


Unidentified Gramineae


Other plants



Verbena officinalis


Fabaceae, Trifolium-type




Polygonum persicaria-type


Galium, large-seeded


Galium, small-seeded


Bupleurum cf. rotundifolium


Glaucium corniculatum


Lithospermum cf. tenuiflorum


Solanum nigrum




Apiaceae cf. Bupleurum


Rumex spp.






Portulaca oleracea


Papaver argemone-type


Asteraceae cf. Achillea


Ranunculus spp.


Helianthemum salicifolium


Equisetum, Sporophyllae


























Unidentified seeds


Concerning the spikelet forks nearly 70% of the charred remains are spikelet forks of Triticum. They occur consistently in all layers. Around 60% of the spikelet forks are broken. Most of them are small. Around 500 complete spikelet forks were found in the fraction below 0.5 mm (Fig. 3). Twenty-three terminal spikelet forks were found and these indicate the presence of tetraploid wheats (Fig. 4).

Fig. 1. Triticum grains from NÇ, small-type. From the top: first is probably two-seeded, second is two-seeded, third and fourth one-seeded, and fifth probably one-seeded.

Fig. 2. Triticum grain from NÇ, large-type. From the top: the first one is a complete grain, second, third and fourth are typical broken grains.

Fig. 3. Triticum spikelet forks from NÇ with the typical tear-off markings of non-brittle spikes.

Fig. 4. Terminal spikelet forks of Triticum from NÇ.

Fig. 5. Fragments of Hordeum spp. from NÇ.

Fig. 6. Well-preserved rachis fragments of Hordeum from NÇ (probably Hordeum spontaneum).

Fig. 7. Typical seeds of Lens from NÇ.

Approximately 89 broken grains of Hordeum, barley, were recovered. They were small and slender with distinct edges (Fig. 5). It was not possible to make a distinction between Hordeum distichon and H. spontaneum. More than 1500 rachis fragments were found occurring in all layers. With a small number of exceptions they are very small; one-third was found in the fraction below 0.5 mm. Most of them are small, badly preserved fragments. In comparison with those of Çayonü (van Zeist and de Roller 1991/1992), the poorly preserved fragments were identified as Hordeum subsp. The well-preserved rachis fragments appear to belong to H. spontaneum (Fig. 6). Whether they are cultivated or wild is difficult to ascertain.


The number of pulses, totalling 1726 seeds, is nearly as great as the number of cereal grains, which totalled 1946. This clearly shows the importance of vegetable protein during this period. At least five species were present in the PPNB layers.

Lens spp.: Lentil is the most frequent species. Most of the 342 seeds of NÇ are very small. Only one reached a diameter of 3 mm; the average diameter of all the others is about 2.2 mm with an average thickness about 1.2 mm (Fig. 7).

Pisum spp.: 137 seeds of pea provide evidence for the use of this pulse during Early PPNB. About a dozen seeds show typical traces of shrinking, probably from drying prior to charring (Fig. 8). The average diameter of the seeds is around 4.5 mm. Most of the peas were found in low frequencies associated with hearths in the 'roasting-pit-area'. Hence they may represent seeds lost during cooking.

Lathyrus subsp., cultivated grass pea: The 137 seed remains show the similar importance of grass pea as compared with pea. Seeds showing a hatchet-like lateral view predominate (Fig. 9). The grass pea seeds show several morphological types. Hence a definite identification as Lathyrus sativus is not possible.

Another small-seeded pulse with a diameter of 3 mm was found. It closely resembles Cicer or Lathyrus cicera. Since this single seed is poorly preserved, no precise identification was possible.

Vicia ervilia, bitter vetch: Despite the small size of the seeds and the possible toxicity, bitter vetch was obviously in use. Nearly 100 seeds were found and probably a lot of the small unidentified pulses were also bitter vetch (Fig. 10). Bitter vetch is a small plant without tendrils. In the field it is difficult to eliminate when it grows with lentils. The plant tolerates very dry conditions and has a short vegetative period of 120 days.

Vicia faba, horse bean: Only 15 poorly preserved seeds of horse bean were found (Fig. 11). Most of them are broken.

Fruits and nuts

Pistacia spp., terebinth: Pistacia nuts were frequently consumed. Of the 470 nutshells found, most were broken, only a few of them completely preserved. Pistacia is often found on Neolithic sites in the Near East. The botanical remains from NÇ show that these nutrient nuts were roasted. Numerous blistered-charred fragments were found. Roasting reduces the bitter aspects in taste and changes the carbohydrates partially into sugar. This is evidence for food-processing, rarely found in archaebotanical remains.

Amygdalus spp., almond: Almonds were probably eaten during the Neolithic NÇ. Fragments of nutshells (Fig. 12), together around 30 nuts, were found near the hearths. In small amounts almonds are not poisonous for adults. The kernels contain cyanogene glycosides that break down to hydrocyanic acid, up to 1 mg per nut. Prunus dulcis var. dulcis is poor in these substances and can be consumed in greater quantity. The fragments found do not allow identification of the species, so it is not known if the almonds of NÇ were toxic or not.

Vitis spp., grape: Two of the four grape pips are completely preserved (Fig. 13). The pips indicate that the fruit was small, probably gathered early in the unripe state, before birds or other animals ate them. The wild grape vines probably grew in the Euphrates Valley.

Fig. 8. Three different seeds of Pisum from NÇ. The bottom sample is a small seed of Pisum showing depressions, probably due to drying.

Fig. 9. Lathyrus 'sativus' from NÇ. The top first and second are seeds with hatchet-like shape from the proximal part of the pod. The bottom one is seed from the middle of a pod.

Fig. 10. Vicia ervilia from NÇ.

Fig. 11. Vicia faba from NÇ.

Fig. 12. Amygdalus stone from NÇ.

Fig. 13. Vitis seeds from NÇ.

Fig. 14. Celtis stone from NÇ.

Fig. 15. Stipa from NÇ.

Cornus mas, Celtis australis, Prunus spp.: Only one item of each of these was recorded. The seed of Celtis is mineralized (Fig. 14). Cornus and Celtis might have been used not only as fruits, but could have been helpful as medicinal plants as well. The remains of these two species and those of Prunus complete a picture of a distinct diversity of plants in use at this site.

Other seeds

More than 300 seeds of Gramineae were found representing at least 10 different species. Bromus and seeds resembling Alopecurus were the most numerous. The presence of Aegilops is recorded by 27 glume bases. Five seeds of Lolium indicate an early association of this species with cereal cultivation. Surprisingly a very brittle grain of Stipa (Fig. 15) was completely preserved.

Nearly 1275 remains of Helianthemum salicifolium and 210 remains of Artemisia species indicate dry conditions.


The plant remains of NÇ. represent early evidence of food production. Wheats and diverse species of pulses were cultivated. The work on modern material casts doubt as to whether the two-seeded wheats found on Neolithic sites of the Near East are tetraploid or diploid. The only conclusive proof we have is terminal spikelet forks which record the existence of tetraploid wheats in the material. A wide diversity of different wheats appears to have been in use during the early Neolithic. We still do not know how long the domestication process took.

Probably the botanical remains from the Early PPNB layers of NC do not represent the very beginning of food production. Given the advanced architecture, even in the earliest levels, it is not improbable that food production dates from before the occupation of the site and was introduced from elsewhere.


Hauptmann, H. 1984. Hauptmann, Nevali Çori. Anatolian Studies 34:228.

Hauptmann, H. 1987. Hauptmann, Nevali Çori. Anatolian Studies 37:206-207.

Hauptmann, H. 1988. Nevali Çori: Architektur. Anatolica 15:99-110.

Hauptmann, H. 1997. Nevali Çori. The Oxford Encyclopedia of Archaeology in the Near East 4:131-134.

van Zeist, W. and G.J. de Roller. 1991/1992. The plant husbandry of Aceramic Çayonü, SE Turkey. Palaeohistoria 33/34:65-96.

Crop Water Availability from a Pre-Pottery Neolithic Site on the Euphrates, Determined by Carbon Isotope Discrimination of Seeds - J.L. Araus, A. Febrero, M. Catalá, M. Molist, I. Romagosa and J. Voltas


The beginnings of agriculture in the Old World, which occurred sometime during the period of transition from hunting/gathering to farming communities, seem to be associated with cereal domestication. Indeed, experimental studies have given a theoretical indication of a lapse in time between the beginnings of cultivation and morphological domestication (Hillman and Davies 1990; Willcox 1991, 1992), but up to now cultivation prior to morphological domestication has not been identified from plant remains in the Near East (Willcox 1996). The Pre-Pottery (aceramic) Neolithic site of Tell Halula is the oldest reported archaeological site on the Euphrates where domesticated crops have been reported (Molist et al. 1995; Willcox 1996). Thus, 9th millennium BP levels at Halula see the appearance of domesticated crops such as naked wheat (Triticum aestivum and/or T. durum), hulled wheat such as emmer (Triticum dicoccum) and barley (Hordeum vulgare) (Molist et al. 1995; Willcox 1996; Willcox and Catalá 1996). The cultivars appear to have been introduced from elsewhere. With regard to edible pulses Ladizinsky (1987, 1989) argues that domestication of grain legumes such as lentils was before that of any other crop. Anyway, as in the case of cereals, domestication of grain legumes would have been possible only under cultivation (Zohary 1989).

Information about the environmental conditions during the beginning of agriculture is scarce and indirect. Recent data on climatic conditions within a more precise chronological framework (Baruch and Bottema 1991; Moore and Hillman 1992) are helping to explain the environmental setting during the transition from hunting/gathering to farming communities. The presence of wild rye (and also wild einkorn) at a number of early sites, including Euphrates sites, during the 10th and 11th millennia BP would appear to indicate cooler and moister climatic conditions than today (Hillman et al. 1993; Willcox 1996). However no evidence of rye appears in the 9th millennium BP levels at Halula when agriculture was present (Molist et al. 1995; Willcox 1996). Indeed the occupation of the aceramic Euphrates sites coincides with the latter part of the period of global warming following the Younger Dryas (Baruch and Bottema 1991). Thus it is probable that present-day temperatures are higher than those which occurred during the Neolithic.

Other archaeobotanical evidence also supports the possibility that environmental conditions during the beginnings of agriculture were more favorable (i.e. cooler and moister) in this region than nowadays. They are based on the former presence of gallery forest species, identified by charcoal analyses, such as ash, vine, elm and plane which, today, are only found much farther north in Turkey and Greece. Similarly, almond, Pistacia and deciduous oak, which nowadays only occur at higher altitudes in the region, also suggest that conditions during the Neolithic were cooler and moister (Harlan 1995; Willcox 1996). However, vegetation degradation through human impact, particularly grazing by livestock, rather than climatic change must account for some of the differences between the vegetation of the early Neolithic and that of the present day. This makes it difficult to evaluate precisely the effect of climatic change on the vegetation (Willcox 1996). Indeed archaeobotanical data from Tell Halula levels (less than one millennium younger than other PPN sites of the Euphrates) indicate a shift to the contemporary steppe/weed vegetation.

Summarizing, even when present palaeoclimatology data and archaeological discoveries strongly suggest a more humid environment prevailing in the Near East at the beginnings of the Holocene, much remains to be discovered about the precise nature of the climate during these crucial millennia (see F. Hole, this volume) as well as the cultural conditions of agriculture at the beginnings of domestication. For example there are no reliable data on precipitation (or even water status) for this period (Rossignol-Strick 1993).

For C3 plants, such as most of the crop plants first cultivated in this region, carbon isotope discrimination (D) in crop grains constitutes an integrated record of the ratio of intercellular to atmospheric partial pressure of CO2 (pi/pa) and thus of the water status during the growth of these grains (Farquhar and Richards 1984; Araus and Buxó 1993; Araus et al. 1997). Both decreased water availability and increased evapotranspiration cause lower pi/pa and thus A in grains because of their effects on stomatal transpiration or photosynthetic capacity (Farquhar and Richards 1984; Romagosa and Araus 1991; Condon et al. 1992). Therefore, from the analysis of A of grains it should be possible, in principle, to infer water status during grain-filling (Fig. 1). In this context the measurement of A from seeds found in archaeological sites has been proposed as a method to evaluate the water status during the growing of these crops (Araus and Buxó 1993; Araus et al. 1997). Nevertheless an inherent limitation of the D approach would be the difficulty of distinguishing with certainty low (i.e. suboptimal) levels of irrigation from the existence of a wetter climate in the past or the utilization of naturally wetter soils. In the present study the A method is applied to elucidate the water status of different crop plants grown in Tell Halula during a period of about 1000 years between Middle Pre-Pottery Neolithic B (PPNB) and late Neolithic (ca. 8700 to 7700 BP uncalibrated).

Material and methods

Archaeological site and plant material

Tell Halula is situated in northern Syria about 100 km east of Aleppo and 30 km southeast of Membij. This site is on the west river bank, in a subsidiary valley, 4 km from the main Euphrates Valley. The archaeological site comprises Middle and Late PPNB, and late Neolithic (pre-Halaf) and is being excavated by the Universitat Autonoma de Barcelona. The present-day natural vegetation in the region is a degraded steppe, with a total annual rainfall of about 250 mm (Willcox 1996). At present the land above the valley floor is extensively used for dry (i.e. rain-fed) farming of barley, whereas wheat and horticultural crops are only cultivated where irrigation is available.

Samples of seeds of naked wheat (T. aestivum/durum), lentil (Lens orientalis/culinaris) and flax (Linum spp.) were used for stable carbon isotope analysis. They were found in a carbonized state and were gathered from domestic fires, cooking ovens and floors of rooms. Six different stratigraphic levels from the archaeological site were studied. Soil samples were treated using a standard flotation tank in the field with 0.3 mm (flotation) and 2.5 mm (wet) sieves. Plant remains were then dried slowly before transport and sorting of seeds. Material was compared with modern reference material which had been gathered from various locations in the Near East. Palaeobotanical determinations were performed at the Institut de Prehistoire Orientale of the CNRS, Jalès (France) under the supervision of G. Willcox. The three species considered in this study were present throughout the stratigraphic sequence studied. Chronology of archaeological samples, in years before the present (BP), was based on stratigraphic dating and radiocarbon ages. Radiocarbon determinations were performed at Beta Analytic Inc. (Miami, Florida, USA). Dates ranged from ca. 8700 ± 60 BP to 7690 ± 130 BP uncalibrated. Calibrated ages were determined according to Stuiver and Reimer (1986) by using the computer program CALIBTH3. After calibration the range of dates for the material studied was 9550 to 8465 BP. In addition samples from the present time (1997) were taken from hulled barley cultivated near the site.

Carbon isotope analysis

Prior to stable carbon isotope analysis, seed samples were cleaned as reported elsewhere (Araus and Buxó 1993). The 13C/12C ratios were determined by mass spectrometric analysis at Isotope Services, Inc., Los Alamos, New Mexico, USA. Results are expressed as d13C values, where:

d13C(‰) = [(R sample/R standard) - 1] x 1000

R being the 13C/12C ratio. A secondary standard calibrated against Peedee belemnite (PDB) carbonate was used for comparison. Sample sizes of 5-10 mg were used. The precision of analysis was less than 0.10‰. The potential effect of carbonization on d13C of archaeological grains was ignored as reported before (Marino and DeNiro 1987; Araus et al. 1997).

Discrimination (D) against 13C relative to air was calculated from da and dp, where a and p refer to air and plant respectively (Fig. 1); as follows (Farquhar et al. 1989):

On the PDB scale, da currently has a value of approximately -8.00‰. For calculation of D values of grain samples from the archaeological site, da values were inferred from the work of Marino et al. (1992) as reported elsewhere (Araus and Buxó 1993).

Fig. 1. Stable carbon isotope composition (d13C) of plant tissues is the result of two different kinds of factor: (1) the isotopic composition of the carbon source (surrounding air, da) used by the plant during photosynthesis, and (2) for most plant species (C3 plants), the water status during growth. The effect of da on d13C can be eliminated by calculating the carbon isotope discrimination (D). Therefore D of plant material reflects the water regime during its growth. In principle this is applicable either to samples from currently growing plant material or to plant remains from archaeological sites.

Results and discussion

Except for two samples (one of wheat and the other of lentil) all the seeds studied showed carbon isotope discrimination (D) values greater than 16‰, reaching 20‰ for some samples of flax (Fig. 2). These values of D are very high for rain-fed crops growing under Mediterranean conditions. For example, whereas D values for wheat kernels were around 17‰ during the period studied (Fig. 3), it has been reported that kernel D of 17.5 ‰ and higher would indicate growing conditions resembling those grown under full irrigation (Araus et al. 1997). Indeed the water regime during grain-filling strongly affects the D values of mature kernels (Romagosa and Araus 1991; Condon et al. 1992; Araus et al. 1997, 1998). Thus, under the usual rain-fed conditions in the western Mediterranean basin, the relatively low precipitation and high evapotranspiration at the time of grain-filling led to significantly lower A values in mature kernels than in those grown under irrigated conditions. In the same way, the D values of wheat were much higher than those reported in wheat crops cultivated under rain-fed conditions in northwest Syria during the 1995/96 season (Araus et al. 1998) in environments with similar (Breda) or somewhat higher (Tel Hadya) rainfall than Tell Halula. Thus, mean ± SD values for Breda and Tel Hadya were 14.0 ± 0.4‰ and 15.3 ± 0.4‰, respectively. These values were clearly below those measured in the archaeological seeds (Fig. 2).

Samples from hulled barley kernels cultivated in seven different places during the 1996/97 season around the archaeological site were also analyzed. These barley crops grow without significant (if any) chemical nitrogen fertilization, as inferred from the low total nitrogen content and the high proportion of stable isotope 15N (d15N) in kernels (Amaro et al. 1995), a situation similar to that expected in ancient agriculture. Mean ± SE of D was 16.02±0.19‰. This value is again clearly below D values for all the archaeological seeds. Moreover, under present conditions, D of wheat would be even lower. Considering that barley attains maturity at least 2-3 weeks earlier than wheat, the grain-filling period for barley usually takes place under wetter conditions than those that could be expected for wheat. Indeed this difference in phenology between both crops is the reason why, among cereals, barley is more commonly cultivated under rain-fed conditions in areas with conditions comparable to those which predominate in Tell Halula at present.

These results strongly suggest that, during the period studied, wheat was cultivated at Tell Halula under much wetter conditions than could be expected from present-day conditions. The presence of flax and its very high D values also support this conclusion. Cultivation under more humid conditions could have been possible owing to more humid environmental conditions prevailing at that time or by planting in alluvial areas (Bar-Yosef and Kislev 1989). Another possibility lies in the existence of irrigation practices. Indeed one indirect method to assess the presence of ancient irrigation takes into consideration the development of weeds; for example, from the size of charred flax seeds which occurred in an archaeological assemblage of plant remains (Helbaek 1960). However, even when flax seeds were very abundant in all the stratigraphic levels studied at Tell Halula their length did not reach 3.0 mm which has been proposed as the minimum size for domesticated plants.

In the case of wheat, although D values of archaeological kernels were higher than those of kernels cultivated at present because of the wetter conditions, they were still lower than typical values of irrigation. In fact, D seems to depend on the water accumulated during grain-filling on a logarithmic basis (Araus et al. 1997). Therefore, D values slightly lower than those reported as typical of irrigated crops under Mediterranean conditions will be associated with considerably lower amounts of water during grain-filling. For example a value of D 0.5 ‰, lower than that typical of irrigated wheat (17.5 ‰), would correspond to an amount of water during grain-filling about 30% lower than that of irrigated crops (Araus et al. 1997). Of course any condition between rain-fed and full irrigation (i.e. irrigation support) would produce kernels with A values below those of full irrigation.

Fig. 2. Carbon isotope discrimination of durum wheat, lentil and flax seeds found in six different stratigraphic levels of Tell Halula ranging from Middle PPNB (ca. 8700 ± 60 BP uncalibrated) to late Neolithic (ca. 7690 ± 130 BP uncalibrated). The D values from present-day (1996) mature kernels of durum wheat cultivated under rain-fed conditions in northwest Syria in environments with similar (Breda) or somewhat higher (Tel Hadya) precipitation than Tell Halula are presented for comparison. For these two environments values presented are means ± SD of 144 genotypes and two plots each (Araus et al. 1998).

Fig. 3. Mean ± SE of D values of durum wheat, lentil and flax seeds found in stratigraphic levels of Tell Halula corresponding to Middle PPNB (M-PPNB), Late PPNB (L-PPNB) and late Neolithic (L-N). Means with different letters are significantly different (P £ 0.05) by Duncan's comparison test.

Alternatively, as pointed out above, the exploitation of naturally moist soils (without evidence of irrigation) is largely evidenced in early agricultural sites (Bar-Yosef and Kislev 1989). A high water level in the soil could explain, in turn, such high D values in the absence of irrigation. Indeed, if the information of early agricultural sites in the Near East (Bar-Yosef and Kislev 1989) may be considered as a guide, it could be assumed that cultivation whenever possible was based on sowing on alluvial fans and terraces as well as on the edges of freshwater swamps where the water table was always high and the soil fertilized by silt deposited by periodic floods. However, as yet there is no evidence for irrigation (Bar-Yosef and Kislev 1989; Hillman and Davies 1990).

Willcox (1996) notes that the presence of charcoal of ash, vine, maple, plane, alder and elm from the gallery forest, and wild rye, wild einkorn, deciduous oak, wild almond, Pistacia and Pyrus from the hinterland, indicates cooler conditions in the middle Euphrates region at the beginning of agriculture. Particularly in Halula the presence of Quercus, Amygdalus, Pistacia and Olea europaea during the PPNB suggests that climate was much more humid than at present (Willcox and Catalá 1996). However the present-day natural vegetation corresponding to a degraded steppe could be the consequence of human effect on the landscape. Thus, annual weed plants usually associated with open landscape, due to the cultivation of cereals or other plants, were already present in the same soil samples where wheat, lentil and flax appeared. In addition they occur throughout the stratigraphic sequence studied. Among these annual plants it is worth mentioning the genera Centaurea, Astragalus, Glaucium, Galium, Lithospermum, Teucrium, Sherardia, Fumaria and Rumex. Expansion of cultivation as well as the great development of extensive goat and sheep pastoralism, could be responsible for this increase in landscape degradation. Nevertheless, in comparison with the contemporary steppe vegetation, the archaeobotanical results indicate a moister, cooler, more continental vegetation found today at higher altitudes and more northerly latitudes (Willcox 1996).

Regarding the possible effect of past climate on D, not only an overall higher precipitation (and perhaps a lower evapotranspiration), but also the seasonal distribution of precipitation, would be involved. Thus at the site of Mureybit on the Euphrates it has been reported that about 8000 BP Poaceae pollen was, in relative terms, much more abundant than today (El-Moslimany 1994). This suggests that in the middle Euphrates region summer precipitation was much higher than today. A shift throughout the region to dominant winter rainfall (typical of Mediterranean climates) would occur after 6000 BP (El-Moslimany 1994). Because for durum wheat and flax, seed formation takes place in late spring and early summer, any increase in precipitation during this period, even when total accumulated rainfall during cultivation does not vary, would eventually lead to higher D on seeds.

For each plant species, means of D for the samples found in the different stratigraphic levels corresponding to Middle PPNB and Late PPNB and late Neolithic (pre-Halaf culture) were plotted (Fig. 3). Flax seeds showed higher values of D than wheat kernels and lentil seeds during the Middle and Late PPNB. Lentil seeds also tended to show higher D values than wheat during the late Neolithic. This could reflect the growth habit, lentils being less determinate plants than cereals (Araus et al. 1997).

As a general trend D values tended to decrease with time, particularly those of flax. Flax is reported to be an indicator of either high precipitation (see F. Hole, this volume) or irrigation practices (Helbaek 1960). This decrease in D suggests a gradual change in climate with an evolution to drier conditions. The first evidence of domesticated plants (cereals) on Tell Halula (ca. 10,000 BP calibrated) appears a full 1000 years after the Climatic Optimum was already established (Becker et al. 1991). The progressive attenuation of the Climatic Optimum until around 6000 years ago, when climatic conditions not far from those prevailing at present were established throughout the Near East (Roberts and Wright 1993), would agree with this tendency on D values.

To summarize, present results strongly suggest that during the period studied, wheat and lentil were cultivated at Tell Halula under much wetter conditions than today. The presence of flax and its very high D values also support this conclusion. Cultivation under wetter conditions could have been possible owing either to more humid environmental conditions prevailing at this time, or planting in alluvial areas, or a combination of both.


This work was supported in part by the grants DGICYT93-0903 and CICYT AGF95-1008-C05-03 and a fellowship DGICYT PR95-006 (Spain).


Amaro, T., C. Bergareche and J.L. Araus. 1995. Changes in stable C and N isotope composition in barley as affected by N nutrition and sowing date. Interdrought. Pages 111-113 in Proc. Int. Congress on Integrated Studies on Drought Tolerance of Higher Plants. Montpellier, France.

Araus, J.L. and R. Buxó. 1993. Changes in carbon isotope discrimination in grain cereals from the north-western mediterranean basin during the past seven millennia. Aust. J. Plant Physiol. 20:117-128.

Araus, J.L., R. Buxó, A. Febrero, M.D. Camalich, D. Martín, F. Molina, M.O. Rodríguez-Ariza and J. Voltas. 1997. Identification of ancient irrigation practices based on the carbon isotope discrimination of plant seeds: a case study from the southeast Iberian peninsula. J. Archaeol. Sci. 24:729-740.

Araus, J.L., T. Amaro, J. Voltas, H. Nakkoul and M.M. Nachit. 1998. Chlorophyll fluorescence as a selection criterion for grain yield in durum wheat under Mediterranean conditions. Field Crops Res. 55:209-223.

Baruch, U. and S. Bottema. 1991. Palynological evidence for climatic changes in the Levant ca. 17,000-9,000 B.P. Pages 11-20 in The Natufian Culture in the Levant (O. Bar-Yosef and F. Valla, eds.). International Monographs in Prehistory, Archaeological Series 1. Ann Arbor, Michigan, USA.

Bar-Yosef, O. and M. Kislev. 1989. Early farming communities in the Jordan Valley. Pages 633-642 in Foraging and Farming - The Evolution of Plant Exploitation (D.R. Harris and G.C. Hillman, eds.). Unwin Hyman, London, UK.

Becker, B.B., B. Kromer and P. Trimborn. 1991. A stable isotope tree-ring time scale on the Late Glacial/Holocene boundary. Nature 353:647-649.

Condon, A.G., R.A. Richards and G.D. Farquhar. 1992. The effect of variation in soil water availability, vapor pressure deficit and nitrogen nutrition on carbon isotope discrimination in wheat. Aust. J. Agric. Res. 43:935-947.

El-Moslimany, A.P. 1994. Evidence of early Holocene summer precipitation in the continental Middle East. Pages 121-130 in Late Quaternary Chronology and Palaeoclimates of the Eastern Mediterranean (O. Bar-Yosef and R.S. Kra, eds.). Radiocarbon, Department of Geosciences, University of Arizona, Tucson, USA.

Farquhar, G.D. and R.A. Richards. 1984. Isotopic composition of plant carbon correlates with water-use efficiency of wheat genotypes. Aust. J. Plant Physiol. 11:539-552.

Farquhar, G.D., J.R. Ehleringer and K.T. Hubick. 1989. Carbon isotope discrimination and photosynthesis. Ann. Rev. Plant Physiol. & Plant Molecular Biol. 40:503-537.

Harlan, J.R. 1995. The Living Fields: Our Agricultural Heritage. Cambridge University Press, Cambridge, UK.

Helbaek, H. 1960. Cereals and weed grasses in Phase A. Pages 540-543 in Excavations in the Plain of Antioch I. (R.J. Braidwood and L.S. Braidwood, eds.). University of Chicago Press, Chicago, USA.

Hillman, G. and S. Davies. 1990. Measured domestication rates in wild wheats and barley under primitive cultivation, and their archaeological implications. J. World Prehistory 42:157-219.

Hillman, G., S. Wales, F. McLaren, J. Evans and A. Butler. 1993. Identifying problematic remains of ancient plant foods: a comparison of the role of chemical, histological and morphological criteria. World Archaeol 25:94-121.

Ladizinsky, G. 1987. Pulse domestication before cultivation. Econ. Bot. 41:60-65.

Ladizinsky, G. 1989. Pulse domestication: Fact or Fiction? Econ. Bot. 43:131-132.

Marino, B.D. and M.J. DeNiro. 1987. Isotopic analysis or archaeobotanicals to reconstruct past climates: effects of activities associated with food preparation on carbon, hydrogen and oxygen isotope ratios of plant cellulose. J. Archaeol. Sci. 14:537-548.

Marino, B.D., M.B. McElroy, R.J. Salawitch and G.W. Spaulding. 1992. Glacial-to-interglacial variations in the carbon isotopic composition of atmospheric CO2. Nature 357:461-466.

Molist, M., J. Anfruns, M. Catalá, D. Helmer, J.M. Faura, J. Mateu, A. Palomo, M. Sana and G. Willcox. 1995. Tell Halula (Siria), un poblat dels primers agricultors i ramaders del VII milleni a la vall de l'Eufrates. Tribuna d'Arqueologia 1992-1993:49-61 [in Catalan].

Moore, A.M.T. and G.C. Hillman. 1992. The Pleistocene to Holocene transition and human economy in Southwest Asia: the impact of the Younger Dryas. Am. Antiquity 57:482-494.

Roberts, N. and H.E. Wright, Jr. 1993. Vegetational, lake-level, and climatic history of the Near East and Southwest Asia. Pages 194-220 in Global Climates Since the Last Glacial Maximum (H.E. Wright Jr., J.E. Kutzbach and T.I. Webbet, eds.). University of Minnesota Press, Minneapolis, USA.

Romagosa, I. and J.L. Araus. 1991. Genotype-environment interaction for grain yield and 13C discrimination in barley. Pages 563-567 in Barley Genetics VI (L. Munck, ed.). Helsingborg, Sweden.

Rossignol-Strick, M. 1993. Late Quaternary climate in the eastern Mediterranean region. Paléorient 19:135-151.

Stuiver, M. and PJ. Reimer. 1986. A computer program for radiocarbon age calibration. Radiocarbon 28:1022-1030.

Willcox, G. 1991. La culture inventée, la domestication inconsciente: le début de l'agriculture au Proche Orient. Pages 9-31 in Rites et ryhmes agraires (M.C. Cauvin, ed.). Travaux Maison L'Oriente 20, Paris, France [in French].

Willcox, G. 1992. Archaeobotanical significance of growing Near Eastern progenitors of domestic plants at Jalès (France). Pages 159-178 in Préhistoire de l'agriculture: nouvelles approches expérimentales et ethnographiques. Monographie du CRA No. 6, Centre de Recherches Archéologiques (P.C. Anderson, ed.). CNRS, Paris, France.

Willcox, G.H. 1996. Evidence for plant exploitation and vegetation history from three early Neolithic pre-pottery sites on the Euphrates (Syria). Veget. Hist. Archaeobot. 5:143-152.

Willcox, G. and M. Catalá. 1996. Análisis paleobotánico. Pages 135-142 in Tell Halula (Siria) un yacimiento Neolítico del valle medio del Eufrates campañas de 1991 y 1992. Informes arqueológicos (M. Molist, ed.). Instituto del Patrimonio Histórico Español, Ministerio de Cultura, Madrid, Spain [in Spanish].

Zohary, D. 1989. Pulse domestication and cereal domestication: How different are they? Econ. Bot. 43:31-34.

Previous Page Top of Page Next Page