Natufian sickle blades (ca. 15,000–11,700 cal. BP) reveal cereal cultivation ca. 4.5 millennia before domestication

Abstract

This study reveals a hitherto undocumented cereal procurement strategy employed during the Late Epipaleolithic Natufian culture of the Southern Levant, from ca. 4.5 to 1 millennia before systematic crop cultivation and domestication are identified in archaeological records. A comprehensive microscopic use-wear analysis of flint sickle blades from the sites of el-Wad Terrace and Salibiya I detected patterns generated by the harvest of semi-ripe and ripe cereals that are attributed to the shattering mechanism of cereals. Many sickle blades also featured traces of stripping, shown by our experiments to act as a threshing operation that removes the spikelets and extracts the grains. Based on observations of current wild cereal fields and experimental harvesting, we reconstruct a protracted harvest that begins in mid-spring, when cereals are semi-ripe, and continues to the very late spring, when unshattered specimens with yellow-dry stems are targeted. We argue that cereal harvesting was an intensive and well-organized enterprise. It was based on a deep and intimate understanding of the shattering mechanism of cereals, where mixed stands were harvested, similar to those prevalent in many places in the Levant today, with plants of various properties, including early or late shattering or non-shattering cereals. The Natufian groups were aware of this mixture, exploited it for planning their harvest strategy, and used it to replenish storages. This was a form of early cultivation where it was necessary to ensure minimal seed loss and stock seed reserves for sowing next year’s fields.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-54740-5.

Introduction

The late Epipaleolithic Natufian culture of the Levant spanned about three millennia during the terminal Pleistocene epoch (ca. 15,000–11,700 cal. BP). Its early phase predates systematic crop cultivation and domestication characteristic of the following Neolithic period by ca. 4.5 millennia¹. Nevertheless, it featured numerous traits that seem to have anticipated the crystallization of Neolithic agriculture^2,3. This is the earliest cultural entity known to practice sedentism. The appearance of relatively large sites with durable architecture, rich artistic repertoire, the presence of commensal animals, intensified and diverse hunting patterns, and cemeteries exhibiting elaborate burial customs, all reflect a sedentary hunter-gatherer adaptation on the threshold of agriculture^3–9. Occasional evidence of storage installations^10,11, hewn bedrock features and ground stone assemblages, richer and more numerous than those of any previous prehistoric Levantine entity^12,13, attest to the rise of a more sedentary lifestyle. This is also the time when flint sickle blades become a regular component of the local flint tool assemblages, attesting to the practice of sickle harvesting, with rare examples of Natufian sickle blades still attached to bone-made hafts^14–16.

Sickle blades were widely regarded as a Natufian innovation, although rare occurrences are reported from earlier Epipaleolithic assemblages^17,18, and conceived as a momentous technological breakthrough tied to early cereal cultivation^19–21. In recent years, however, scholars dismissed this hypothesis because of the scarcity of supporting archaeobotanical evidence. Late Epipaleolithic cereal finds in south Levantine sites comprise a few wild barley, oat, and wheat grains from Hayonim Cave and Terrace^22,23, el-Wad Terrace^24,25, Raqefet Cave²⁶, and Wadi Hammeh 27²⁷. Also, in the Northern Levant they are quite rare^28–30. Unique finds comprise a bread-like product reported from Shubayqa 1, associated with a large and diverse botanical assemblage that included a few cereal grains³¹, and the possible use of cereals for the production of beer in stone mortars at Raqefet Cave³². All in all, there is little data to be considered anything more than random, episodic occurrences in the otherwise extensive Natufian archaeological record.

Consequently, it has been occasionally suggested that the Natufian sickle blades may have been used for harvesting other plants, such as reeds and sedges, or that cereal collection was aimed instead at other forms of gathering and processing, such as straw for construction, craft, or fuel^33–37. These suggestions have been reinforced by the appearance of Natufian sickle blades in low frequency compared to the Neolithic³⁵, and recurring indications that Natufian ground stone tools have been used not only for processing cereals^38–40 but also for a wide range of other substances, including ochre⁴¹, hides^42,43, various wild plants, legumes, sedges, and small-seeded grasses^42,44,45. With the appearance of domestic morphotypes in the archaeological record from the Early Pre-Pottery Neolithic B (EPPNB; 10,800–10,300 cal. BP) and beyond⁴⁶ it seems much safer to assume that these glossy artifacts were used for cereal harvesting and in some way played a role in the process of their domestication^47–49.

The present study, based on the use-wear analysis of an unprecedentedly large sample of Natufian sickle blades, approached the question of Natufian cereal harvesting through an independent proxy. Use-wear analysis is a high-resolution inspection of wear on the working edges of tools and the correlation of this wear with specifiable motions and materials. For sickle blades, the most distinctive traces comprise a type of surface polish produced by the abrasive action of cereal cutting^50–52. Studying this polish can disclose various aspects of cereal harvesting: the types of plants that were harvested, the manner and the intensity at which the harvest was conducted, the construction of the harvesting tools, the effect of tilling the fields and the condition of the harvested cereals, including their ripeness and the height at which they were cut^33,53–65.

We studied 158 sickle blades from two Natufian sites, el-Wad Terrace (EWT) and Salibiya I (SI) and conducted cereal harvesting experiments using 22 sickle replicas. These harvesting experiments addressed mainly the distinction between use-wear generated by cutting the cereals in semi-ripe and ripe conditions and the attribution of the resulting wear patterns to the cereals’ shattering mechanism^{47–49,66,67}.

The differences in the characteristics of the use-wear polish are quite prominent and may be defined through the qualitative method using the metallurgical microscope and in recent year these differences were defined successful by the application of the quantitative method using the confocal microscope^68–71. Past analyses have indicated that ripe cereal harvesting becomes dominant during the PPNB, which is consistent with the cultivation of domesticated non-shattering cereals^72–74. By contrast, Natufian sickle blades featured use-wear patterns generated primarily by harvesting semi-ripe cereals, which corresponds to the view that cereal domestication and the exploitation of non-shattering morphotypes is later^60,75–77. This position was challenged from an agronomic viewpoint by questioning the quality of cereals harvested in a semi-ripe condition⁷⁸, but a growing number of qualitative and quantitative analyses, and especially experiments, have verified this variation in the use-wear observed on archaeological tools^53,68,69,75, therefore it cannot be dismissed. Furthermore, data obtained from use-wear analyses is especially important for investigating the earliest evidence of cereal harvesting because archaeobotanical methods of measuring the proportions of abscission scar type (smooth vs. torn) or using grain size indicators, cannot be applied^79–81. The present study therefore breaks new ground in the understanding of Natufian harvesting strategies and behavior by examining use-wear produced by a harvest of semi-ripe and ripe cereals acting as an indicator for the prevalence of early or late shattering or non-shattering morphotypes. Our results provide new, hitherto unobserved data, including unique wear patterns undocumented before that were overlooked quite likely because of too small sample sizes. Our experiments also provided functional explanations of the complexity of the Natufian harvesting based on the detected patterns, paving the way for new insights about the manner and extent of harvesting and the collection of grains.

Natufian sites and sickle blades

The Natufian sites from which the sickle blades for the current research were obtained are located well within the Southern Levant, where some of the earliest evidence of Neolithic cereal domestication and farming was documented⁴⁶. These sites differ from each other in every aspect and, therefore, provide a unique platform for examining the Natufian cereal economy.

The site of el-Wad Terrace is located on the western slope of the Mount Carmel ridge, in the Mediterranean climate region^82–84. The site includes a large cave and an adjacent terrace. This major base camp exhibits one of the most complete sequences of Natufian culture, dating to ca. 15,000–12,000 cal BP, including the Early (EN) and Late (LN) Natufian phases⁸⁴. The front chambers of the cave and the central terrace underwent large-scale excavation by D.A.E. Garrod in 1929–1933, who defined the unique character of Natufian culture^19,85. Further excavations were later conducted on both the terrace⁸⁶ and in the cave⁸⁷. The materials studied here come from the recent excavations conducted by M. Weinstein-Evron, D. Kaufman, and R. Yeshurun at the northeastern part of the terrace on behalf of the University of Haifa. The renewed excavation uncovered a rich and well-dated sequence of Natufian habitations^84,88. The lower part of the EN occupation, dated to 14,950–13,820 cal BP, displays significant architectural remains, including dwelling remains and associated heaps and activity locales^9,88,89. The LN occupation on site is dated to ca. 13,765–12,135 cal BP. The overlap between the EN and LN calibrated ranges likely stems from some movement of charcoals along the stratigraphy, especially in the EN-LN transition⁹⁰.

The site yielded rich evidence of various everyday activities. The faunal assemblages attest to regular exploitation of a small ungulate, the mountain gazelle (Gazella gazella), and a diverse array of smaller game^91–94. The EN and LN ground-stone assemblages include grinding stones as well as vessels and pestles, made mostly of basalt and less of limestone, showing consistency in forms, shapes, and raw materials, with no noticeable differences between the two phases⁹⁵. The archaeobotanical assemblages include thousands of charred wood remains, especially varying abundances of evergreen and deciduous oak, as well as almond and buckthorn. Several hundred charred seeds represent mainly legumes and cereals of wild barley (Hordeum spontaneum)^25,96. The exploitation of local plant resources is further confirmed by the discovery, in the EN phase, of the earliest remains of the wild progenitor of faba bean⁹⁷.

The site of Salibiya I was discovered in the course of a survey conducted in the region of Nahal Fazael and Kibbutz Na’aran during 1977–1978^98,99 and excavated in 1978 and 1989^100,101. It is a single-phase site located at the low-lying Salibiya Basin in the arid southern Jordan Valley, at an elevation of 190–230 m bmsl, adjacent to other Natufian and PPNA sites^102,103. The Epipaleolithic and PPNA sites begin to appear in fluvial or colluvial deposits^{99, chap5} above the soft lacustrine sediments of Lake Lisan. No direct radiocarbon dating related to Salibiya I was produced, but a nearby associated deposit provided dates that span a long period of ca. 9000 years^{99, p439}. The detailed analysis of the Salibiya I flint assemblage¹⁰⁴ confirmed the affiliation of the site to the LN, compared with similar assemblages from LN levels at Hayonim Cave and Terrace^105,106, as well as at Eynan (Mallaha)¹⁰⁷.

The site was estimated to exceed 1200 m², and part of it has eroded into the Wadi Salibiya. Two main areas were excavated, one 1 × 1 m square area near the edge of the site, and one 2 × 2 m area at its center. The Natufian deposits reached a total depth of 95 cm and were subdivided into several distinct strata or interdigitated lenses, the lowermost exhibiting the most intensive occupation, although there were no unambiguous architectural features. Fragments of plaster, a section of puddled mud, and a concentration of small stones were recovered in the smaller edge area. A semi-circular concentration of artifacts and stones surrounded by a series of small, roughly circular areas of mud, exposed in the larger trench, suggested the possibility of a rudimentary structure^{100, p112}. Other finds included two fragments of human bones, bone tools, bone and stone beads, shells, and a limestone fragment with an incised decorative pattern. A rich collection of animal bones indicates a wide range of animal species that were exploited. The vast majority of the fauna assemblage was of gazelle and other large mammals, such as wild boar, fallow deer, wild goat, and aurochs. Remains of smaller mammals, such as fox and hare, were recovered as well. Salibiya I was compared to Beidha, which was interpreted by Byrd¹⁰⁸ to be a site of repeated short-term occupations over an extended period¹⁰⁴. Based on the fauna distribution, it was suggested that Salibiya I was a repetitive seasonal macroband encampment, where inhabitants practiced intensive hunting mainly of gazelle and wild boar during a single season, relying intensively on gathering cereals and storing food^100,109.

Results

All the sickle blades analyzed were found to bear use-wear consistent with cereal harvesting (Fig. 1), calling into question past claims that Natufian sickles were used for other purposes. Numerous sickle blades exhibit wide (< 4 mm), highly developed bands of polish along their edges (EWT, n = 34, 38%; SI, n = 35, 52%) exhibiting flat microtopography due to the continuous friction against the stems (Fig. 2a), indicating long-term use, compared to sickle blades exhibiting traces developed to a lower degree showing a more irregular microtopography (Fig. 2b). In our experiments we did not achieve such a developed polish, therefore we can estimate that the Natufian harvest clearly exceeded 10 h of work. Signs of re-sharpening and the application of retouch to strengthen the edges were also observed (EWT, n = 35; SI, n = 15), suggesting a level of maintenance during use, possibly associated with long harvests, perhaps even through several seasons.

Fig. 1.

Sickle blades from Salibiya I (a–f) and el-Wad Terrace (g–l), with the micrographs showing the cereal harvesting use-wear polish.

Fig. 2.

2D micrograph captured by confocal microscope showing a wide height range of ca. 8 μm for wear developed to a low degree (a), compared to a low height range of 3 μm for wear developed to a high degree (b), with corresponding micrographs by reflected light metallurgical microscope (a1–2, b1–2).

The use-wear analysis detected patterns associated with two distinct but closely related activities: harvesting and, for the first time in the analysis of Natufian sickle blades, the action of stripping. Cereal harvesting use-wear consists of a smooth polish, which when developed to a low degree (Fig. 2b1–b2), appears partially linked, spreading in a reticular pattern with occasional domed surfaces that are the polished protruding surfaces of the natural topography of the flint (consequently showing high height variability as can be seen in the scan by the confocal microscope in Fig. 2b). When developed to a high degree (Fig. 2a1–a2), it appears flat (showing slight height differences in the confocal microscope scan in Fig. 2a), spreading as a band along the edge with no evidence of the natural surface of the flint and with a smooth rounded edge of the tool. Also indicative to cereal harvesting polish are the features resulting from the longitudinal, uni-directional pulling motion of the sickle (Fig. 3a), including comet-shaped pits (Fig. 3b) and long, shallow striations extending parallel to the working edge (Fig. 3c).

Fig. 3.

The microscopic use-wear polish diagnostic of cereal harvesting observed on a sickle blade from Salibiya I (a), with focus on the comet-shaped pit (b) and striations (c) captured by scans by laser confocal microscope.

Stripping is a method used to separate the spikelets from the stem. Stripping with a sickle entails transversally shaving off the spikes (Fig. 4a, b). Until now, the task of stripping was known to be done by hand¹¹⁰ or with designated tools, such as the mesorias^72,111, and through occasional use-wear studies that revealed this activity on a notched bone^112–114 or a flint blade¹¹⁴. Past experimental examples^51,56,114 align with the results of our experiments, showing that the most indicative stripping use-wear consists of polish characteristic to the contact with cereals and especially the striations extending perpendicular or sightly oblique to the working edge (Fig. 4c). However, when used in combination with harvesting, the stripping wear pattern is erased and the only indication to the stripping action are the perpendicular or slightly oblique striations along the edge (Fig. 4d–f). Twenty-seven blades of Early Natufian EWT (40%), one of the Late Natufian EWT (4%), and 36 of SI (53%) exhibited stripping use-wear (Fig. 5, Table 2). The perpendicular striations appear in various patterns, mainly isolated on the actual edge (Fig. 5d, f), very close to it or slightly away (Fig. 5e), or they appear in clusters (Fig. 5a–c).

Fig. 4.

Experimental stripping: (a) stripping close-up showing the accumulation of the spikelets stripped-off the stem; (b) stripping domesticated wheat standing with the stems held against the body; (c) stripping -only polish and striations by confocal microscope scanning showing the transversal polish with arrow marking the striations on the edge; (d) harvesting and stripping combined with oblique striations near the edge; (e–f) harvesting and stripping combined with oblique striations near the edge by confocal microscope scanning.

Fig. 5.

Cereal harvesting polish with perpendicular and oblique striations associated with the activity of stripping (marked by the arrows) captured by scans of confocal laser microscope (a) and reflected light metallurgical microscope (b–f).

Table 2.

Distribution of sickle blades with and without stripping use-wear by cereal ripeness.

	EWT Early Natufian		EWT Late Natufian		S1
	With stripping	Without	With stripping	Without	With stripping	Without
Semi-ripe	11 (41%)	20	0	11	7	16
Ripe	16 (59%)	17	1	12	29	16
Total	27	37	1	23	36	32
	42%	58%	4%	96%	53%	47%

The co-occurrence of harvesting and stripping use-wear on the sickle blades indicates that the tools were used for both tasks, and provides evidence that cereals were harvested for their grains (straw may then be used for other purposes).

Variations in the use-wear polish differentiate between the harvest of semi-ripe and ripe cereals (Fig. 1) replicated in experiments (Fig. 6). Harvesting semi-ripe cereals produces a domed or undulating polish, smooth in texture and bright in reflectivity because of their contact with green, soft stems with high water content (Fig. 6a–d). Harvesting ripe cereals produces a flat polish with comparatively dull reflectivity with a pitted and rough texture given the low water content in the stems, which makes them dry, tough, and more abrasive (Fig. 6e–h. See also Fig. SUP1 for a preliminary result by confocal microscope scanning for microtexture measurements).

Fig. 6.

Experimental cereal harvesting and the resulting diagnostic polish by cutting semi-ripe (a–d) and ripe (e–h) cereals.

All studied specimens bear use-wear traces indicative of distinctive cereal ripening stages, except for four with mixed features semi-ripe and ripe cereals (Table 1), and all three studied assemblages feature use-wear typical of harvesting both ripe and semi-ripe cereals. Though this is a bimodal distinction, it is possible that the use history of some items differed, with some discarded following the early harvest, and some (or most) tools used for the ripe cereals were also used for semi-ripe earlier in the season. In the latter case, we expect that the traces would be erased during contact with the tougher ripe stems later in the harvesting season, and therefore the traces observed indicate the last action. Nonetheless, whereas the distribution of ripening conditions in the EWT assemblages features a ca.1:1 ratio, the distribution in the LN SI assemblage features a 1:2 ratio in favor of ripe cereals. Statistically, the SI distribution differs significantly from the EN EWT assemblage only (χ² = 4.33, DF = 1, p = 0.04). Furthermore, stripping use-wear is considerably more common on tools with ripe-cereal harvesting use-wear (Table 2; χ = 6.88, df = 1, p < 0.01).

Table 1.

Distribution of use-wear type on sickle blades by ripening condition.

	EWT Early Natufian	EWT Late Natufian	Total EWT	SI Late Natufian	Total
Semi-ripe	31 (46%)	11 (46%)		22 (32%)
Ripe	33 (50%)	13 (54%)		44 (65%)
Undiagnostic	2 (3%)			2 (3%)
Total	66	24	90	68	158
	42%	15%	57%	43%	100%

The high frequency of use-wear attributed to the harvest of ripe cereals is puzzling. Unlike domesticated morphotypes, the wild specimens collected by the Natufians must have shattered when ripe, therefore, they were unfit for sickle harvesting or stripping. To solve this quandary, we launched an experimental program to replicate the cereal harvesting use-wear observed on the archaeological specimens, paying close attention to the specificities of the timing of the harvest in a field of wild species (Fig. 8, SUP1 Table). Specifically, we used sickle blade replicas to harvest wild cereals throughout their ripening process and identified the time window when the stems of wild cereals had already dried up whereas their spikelets were still intact (unshattered). We achieved this by weekly surveying fields where wild barley (Hordeum spontaneum and Hordeum bulbusom) and oat (Avena sterilis) grow together. In mid-April, we began close monitoring the maturation of the grains, noting that by late April-early May, they turned from “milky-ripe”^110,115 to semi-ripe (i.e., dehydrated, green, and hard); oat began shattering at the beginning of April, followed by wild barley, in the second or third week of the same month. We observed that half the population in the field shattered by early May, retaining only the lowest spikelets, while the other half was still intact (Fig. 7a, b). Thus, the latter could be harvested without its spikelets shattering. This allows for an effective harvest by focusing on plant clusters with complete spikes, producing use-wear consistent with ripe cereals.

Fig. 8.

Section along the N–O line showing the sample squares cutting through the entire Natufian sequence of the NE Terrace, including the superimposed living floors at the lowermost attained strata.

Fig. 7.

The Hordeum spontaneum field in early May in ripe condition (a, b); the greenish-brown grain of Hordeum spontaneum stripped off the glume (c), close-up on a spike with empty glumes due to stripping (d) and general view of a complete spike after stripping (e).

The grains of these intact spikes are still immature—soft, quite moist, and greenish brown, having lost their milky appearance (Fig. 7c), while their stems are yellow dry. This is the time window when grains are almost ripe but have yet to lose all their water and shrink to their fully dehydrated state. Catching this stage depends on careful observation, mainly because the occasional hot spring easterly winds (khamsin) can accelerate the ripening process and shattering.

Our replication experiments demonstrated that sickles are effective tools for separating the spikelets from the spike by applying the stripping method (Fig. 4). We observed that using sickles to strip semi-ripe cereals separates the entire dispersal unit, requiring further drying and purposeful separation of the glumes with pounding tools, but applying them to dry cereals with moderately hydrated grains breaks the dry glume and extracts the seed (Fig. 7d, e). Thus, when applied to cereals in this advanced ripening stage, the sickle blade also functions as a threshing device, after which the grains are winnowed to separate them from the rest of the debris.

Discussion

Investigation of cereal cultivation and domestication in the Levant is complex, and questions of when and how these processes started are widely debated (for a recent review, see¹¹⁶). Our analysis of wear traces on Natufian sickle blades, supported by experimental harvesting, produced new data that contribute to this debate. Based on our results, we reconstructed a purposeful and knowledgeable exploitation of cereals by the Natufians, in a period from about four and a half to one millennia before the advent of agriculture and full cereal domestication. We rely on use-wear consistent with both semi-ripe and ripe cereals, indicating that the Natufians implemented a strategy of a protracted harvest; in our estimation, it began in mid-spring (late April), when the grains turned semi-ripe and soft, the ideal time for harvesting wild cereals, before the spikelets begin shattering and while the stems are still green. Grains at this point are perhaps poorer in their nutritional value^78,117, but this is gradually complemented with the continuation of harvesting for several weeks into late spring (early to late May), during which time the cereals ripen, the stems turn yellow-dry, and the harvester’s attention shifts to the unshattered specimens. Our experiments proved that this approach is possible and practical. We regard Natufian engagement with cereals as a carefully planned, extensive undertaking that continued through the ripening process further than previously thought. Together with the evidence of sickle blades with highly developed polish, a protracted harvest is a sign of intensive exploitation of stands of wild cereals that leaves no traces in the archaeological record, except for the “silent” evidence of use-wear polish on flint sickle blades.

Not all the quandaries can be answered through wear analysis of sickle blades, but our effort in reconstructing the process of cereal domestication may profit from examining several aspects of a protracted harvest in wild cereal stands. Most important in our view is the effect of a protracted harvest on the natural process of seed dispersal, e.g., involuntary sowing⁴⁹. In wild stands of mixed species, even if a portion of the field was left untouched to enable natural seed dispersal, we expect a severe reduction in next year’s crops. Our observation of the natural fields confirmed this. Initiating the harvest before natural shattering commenced in semi-ripe condition is essential to ensure minimal seed loss, therefore in order to maintain field capacity, stocking seed reserves for sowing next year’s fields is similarly essential. The evidence of a protracted harvest should be conceived in this case as a type of early cultivation that also requires some storage method, about which we know very little in the Natufian culture^10,11. This is not to say that other collecting methods were not used¹¹⁰, however the evidence of semi-ripe combined with ripe cereal harvesting reveals a more calculated and field-dynamic oriented method.

Our finds indicate that mixed wild stands, similar to those prevalent in many places in the Levant today, produced seeds with various properties, including early shattering, late shattering, or non-shattering and suggest that the Natufians were aware of this mixture and exploited it for planning their harvest strategy. Wear analysis cannot indicate whether the sowing was selective by genotype or mixed, and therefore the degree of Natufians’ engagement in crop breeding, if any. But a protracted harvest may be conceived as a method for producing sufficiently abundant yields, and it indicates that the Natufians possessed agronomic knowledge that lay at the bottom of relatively advanced methods of coping with potential crop failure and of manipulating productivity. Our study also demonstrates that sickles were indeed a technological breakthrough, even more than previously believed³⁵, serving both as cutting and threshing instruments, reaping and stripping devices.

To conclude, for the first time, our findings suggest a type of late Pleistocene cereal cultivation in the Southern Levant in the form of a protracted, intensive harvest. Cultivation was previously assumed only at Northern Levant sites because of preserved archaeobotanical remains^49,118. Our findings offer a glimpse into a cereal procurement strategy that builds on intimate knowledge and understanding of ripening and shattering patterns. We found this evidence at the EWT site, in a long occupation sequence, and at the more short-lived SI site, indicating that this strategy was used without relation to the amount of time spent in a certain camp. It was also without relation to site setting, at the area around Mount Carmel or in more desertic areas, at the margins of the Mediterranean zone. Further study of Natufian sickle blades is expected to reveal the scope of protracted harvesting and sickle-threshing in various locations, help consolidate site-environment relationships, and deepen our understanding of how full-fledged agriculture developed.

Methods

Typology and sample selection

In prehistoric toolkits, among types of flint tools, sickle blades are defined first and foremost as tools exhibiting glossy edges^19,21,119, which is the use-wear polish produced by contact with the cereals. Therefore, this is the only tool type defined by visible use-wear in combination with particular technological modifications. In the case of Natufian culture, sickle blades are mainly blades and bladelets, with modifications that may include a Helwan retouched back and truncations or an abruptly retouched back^19,120, although there are also unmodified blades and flakes with that specific gloss on them¹²¹. The artifacts selected for the use-wear analysis exhibit the gloss visible to the naked eye (Table 1).

The sample from EWT comprises 90 specimens, which amount to all the sickle blades retrieved from Squares O–P/6–7, excavated until the year of 2012. This was a clear and undisturbed stratigraphic sequence, encompassing the Early and Late Natufian phases (Fig. 8).

The SI sample consists of 68 specimens, comprising all the sickle blades found at the site.

The use-wear analysis protocol

Use-wear analysis determines the function of the tools by discerning and defining microscopic wear produced by their use. The underlying principle is that wear is indicative of the materials that came in contact with the edges of the tools, making functional reconstruction possible. Analysis protocols define abrasions, edge rounding, linear features and especially edge removals on the macroscale. Then, polish and features associated with the polish, including edge rounding, striations and pits are characterized on the microscale. The microscopic polish is the main concern here, analyzed following the conventional use-wear analysis protocol^{51,62,122,123}, commonly applied in sickle blades research (e.g.,^{17,18,54,61–63,74,77,124–127}. Cereal use-wear polish is a prominent wear type, distinctively different from other polishes, with characteristics that define the polish by its topography, distribution pattern, texture, and reflectivity. For the key aspect discussed here, namely, differentiating the ripeness of the cut stems as a proxy of shattering or non-shattering condition, the attributes of the wear produced a clear separation between semi-ripe and ripe (Fig. 7). This distinction was possible for almost all sickle blades, except for four tools with mixed features, probably because the sickles were used throughout the harvesting period for both semi-ripe and ripe cereals.

The functional interpretation is based primarily on a comparison between archaeological traces with experimentally produced ones. We conducted cereal harvesting and stripping experiments that produce a reference collection to be used for understanding wear formation process and interpret the traces observed on the archaeological tools (Figs. 4 and 6). Furthermore, we ruled out non-cereal harvesting use-wear based on the extensive reference collection of experimental flint tools at the Wear Analysis Laboratory at Zinman Institute of Archaeology, University of Haifa. The collection comprises hundreds of tools that were used for diverse tasks with various plants, animal materials, minerals, and more. Especially, our emphasis was to distinguish the cereal use-wear from other wear-types by siliceous plants such as reeds and small seeded grasses that may produce overlapping features.

The analyzed tools were washed with soap water, rinsed in an ultrasonic tank for one minute, and dried before observations. Occasional cleaning using alcohol was applied repeatedly during the analysis to remove external contamination. The sickle blades were then inspected under high magnification (100–200 ×) using a metallurgical microscope (Leica DM 1750 M) at the Wear Analysis Laboratory at the Zinman Institute of Archaeology, University of Haifa. To characterize the use-wear polish, we produced micrographs using the microscope camera and used the Helicon Focus © software (version 8.3.11, http://www.heliconsoft.com/heliconsoft-products/helicon-focus) to produce full focus images.

The confocal microscope (a Sensofar S neox microscope with the ISO 25,178 3D optical profilometer under a Nikon—EPI 50 × lens) was used for illustrational purposes only, to show the topographic characteristics of use-wear features (polish, striations, and comet-shaped pits; see Figs. 2, 3, 4, 5) with 2D images with height data, and to generate full focused images for both the experimental (Fig. 4c, e–f) and archaeological wear (Fig. 5a) located at highly angled surfaces. It was also used on representative experimental and archaeological sickle blades selected based on the results of the observations with the metallurgical microscope to show the differences between polish micro-texture of semi-ripe compared to ripe cereal harvesting polish by selected roughness measures described in supplemental material and Fig. SUP1.

Experimental cereal harvesting and stripping

The experiments involved replicas of Natufian sickles, based on archaeological specimens with flint inserts still attached to their hafts. The archeological sickles were recovered from Natufian sites, for example, Kebara Cave¹⁴, el-Wad Cave¹⁵, and Wadi Hammeh 27¹⁶. The replicas were used to harvest different cereal morphotypes. Each sickle had one to four flint blade inserts produced from local flint outcrops, fitted into bone or wood hafts. Our experiments sought to produce use-wear patterns consistent with those observed on the archaeological specimens, distinguishing harvesting semi-ripe cereals from ripe ones and their stripping (Figs. 4 and 6).

We conducted 17 experiments using 22 sickles (a total of 48 flint inserts) in various locations in Israel, including Mount Carmel, Galilee (northern Israel), the Western Negev (southern Israel), and the Shephelah (west-central Israel) (Table SUP1). The resulting experimental collection spanned the harvest of semi-ripe and ripe wild cereals (Avena sterilis, Hordeum spontaneum, Hordeum bulbosum) and their stripping. Experiments also included harvesting ripe domestic wheat (Triticum aestivum) to create comparative traces of the driest conditions (which can be achieved only by domestic morphotypes). The distinction between the two ripening states relies on the condition of the stem which is in direct contact with the sickle blade. Accordingly, the semi-ripe wild morphotypes were harvested starting from mid-April when the stems were green, fully grown and seeds clearly observed. The ripe wild morphotypes were harvested in early May, a state in which the stems are dry and the grains were green–brown but still moist (Fig. 6). Domesticated cereals were harvested in late May to the beginning of June, when the stems were fully ripe and the grains dehydrated and brown.

Each experimental harvesting was conducted by students and lasted between 30 min and 10 h. In some of the experiments, the yields and the rate of use-wear development were measured against the number of sickle strokes. The experimental stripping was also conducted by students, using the sickle held by hand with pressure against the body, but with no direct contact, ensuring the sickle slides on the cereals at all time. A cloth was placed on the ground where the grains and the spike debris accumulated.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Author contributions: I.G.Y., T.K., A.B.C., L.G., R.Y., D.K., M.W.E. designed research; I.G.Y., J.V., I.A., K.B, M.P. performed experiments; I.G.Y. analyzed the data; I.G.Y, A.B.C., L.G., R.Y., D.K., J.V., I.A., K.B, M.P., M.W.E., T.K. wrote and reviewed the paper.

Funding

The work was partially funded by the Israel Science Foundation (grant 511/22 to R.Y.) and Gerda Henkel Foundation (grant no. AZ 48/F/21 to I.G.Y).

Data availability

Data is provided within the manuscript file and supplementary information files.

Declarations

Competing interests

The authors declare no competing interests.