Introduction

The domestication of plants and animals, predominantly wheat and barley, as well as goats and sheep, is believed to have commenced around 9500 BCE in the Near East, in the region commonly known as the Fertile Crescent, situated between the Levant, southern Anatolia, and the Zagros Mountains1. Subsequently, it is postulated that agriculture was disseminated from this point to numerous regions of Europe in the west and between the Iranian Plateau, Central Asia and northwest South Asia in the east, essentially from approximately 7000 to 5000 BCE1.

As stated by G. Childe nearly a century ago2, this “revolution” had a profound effect on the way of life of mankind, beginning with the subsistence economy, and marked the beginning of the Neolithic period. This transformation was accompanied by a number of other changes, including a significant shift in how humans conceptualized and interacted with the natural world. Additionally, the advent of permanent dwellings, predominantly rectangular in ground plan, and the subsequent dissemination of pottery vessels during the 7th millennium BCE are notable developments. It is now understood that this “revolution” was a gradual process that unfolded over millennia and encompassed the cultivation of wild plant species and the management of wild animals in the pre-Neolithic era1,3,4. Despite the plethora of models that have been proposed, there remain unanswered questions about the actual motives—the conceptual and tangible factors, including environmental conditions—and their chronological sequence that led to the emergence of farming villages in the Fertile Crescent1,3.

A further set of questions remains regarding the origins of agriculture outside the Fertile Crescent. It is well documented that the advent of the Neolithic era in Europe was the consequence of a diffusion process from the Near East. During the 7th millennium BCE, farmers from the Near East began migrating to Europe, bringing with them domesticated plants and animals. They also introduced and adapted their agricultural knowledge and practices to new lands, climates, and peoples1,5. On the other hand, the mechanisms by which agriculture emerged between the Iranian Plateau, Central Asia, and South Asia are currently less well documented and remain ambiguous. The data on the Neolithization and the Neolithic period are either lacking or, in many regions, extremely rare due to a lack of exploration. Furthermore, the majority of the currently available archaeological records6 indicate a later, demic or cultural diffusion from the West, while some records suggest the early, local and independent invention of agriculture5,6,7,8,9,10,11,12,13.

The archaeological site of Mehrgarh in the Pakistani province of Balochistan is illustrative of this situation, as it is the sole site in northwest South Asia (between Pakistan and northwest India) where Neolithic occupation deposits—specifically, Aceramic Neolithic Period I and Pottery Neolithic Period II—have been extensively excavated and radiocarbon dated (Fig. 1; Supplementary Information 1, 2 & 5). Accordingly, this site is of paramount importance for the comprehension of this period in the region. Nevertheless, the chronology of the Neolithic period at Mehrgarh remains a topic of considerable debate. J.-F. Jarrige, the director of the fieldwork at the site, proposed that the earliest Neolithic occupation period, Period I, commenced around 8000 BCE and was the consequence of a local process14 (Supplementary Information 2 & 3). In contrast, other scholars have proposed that the initial Neolithic occupation did not commence until the mid-to-late 6th millennium BCE, or at least that there are issues with the radiocarbon dating of Period I6,15,16,17, and that the Neolithic in South Asia resulted from the diffusion of practices from the West18. The previously published radiocarbon dates from the Neolithic period at Mehrgarh are a central point of contention in this debate. The majority of these dates were obtained from charcoal samples19,20, although some were derived from dental apatite and enamel21. Most dates fall within a narrow range, spanning from the mid-6th to the mid-5th millennia BCE, with relatively minor margins of error. However, several dates are older than this range19,20,21,22 (Supplementary Information 2 & 5).

Fig. 1
Fig. 1
Full size image

Map with location of Mehrgarh and additional sites mentioned in this article.

The excavations at Mehrgarh yielded a large amount of data about the incipient agricultural economy, crafts, architecture, and funerary practices during the Neolithic period that is qualitatively and quantitatively unparalleled (Supplementary Information 2). This record is of enormous interest for the study of early agriculture in the regions between the Iranian Plateau and South Asia. As a result, the dating of Mehrgarh Period I, which forms the basis of these data, has had a significant impact on current knowledge and the reconstruction of the chronology and processes involved in the onset of agricultural life between the Fertile Crescent and the Indus Valley.

In this paper, we present the results of new radiocarbon analyses of human tooth enamel from 23 graves in the nine stratified levels of the Mehrgarh Period I cemetery, which was excavated by Jarrige and his research team. The graves in question are representative of each of the aforementioned stratigraphic levels, from the earliest to the most recent ones recorded at the site. In contrast to most previously available radiocarbon dates, which were conducted on charcoal samples from settlement levels, these new analyses were performed on samples from exceptional, primary contexts with no stratigraphic disturbance. In the absence of preserved collagen, dating was performed on the inorganic (apatite) fraction of tooth enamel using the latest methodological advances23. The calibration and Bayesian modelling of these new dates, in conjunction with previously published tooth enamel dates, have yielded significant insights into the chronology of the Mehrgarh Neolithic period and the duration of Period I. These findings are presented herein and subsequently discussed in the context of the dating and process of the inception of agriculture in the Indus Valley region of northwest South Asia.

Results

Enamel samples were prepared using a micronizer to remove diagenetic carbonates more efficiently as described in23. Our results confirm that micronizing improves the quality of the radiocarbon dates. Indeed, we observe an age difference of 72 (T180), 89 (T162B) and 235 (T167A) 14C years BP between hand-ground and micronized subsamples, with the most ancient age obtained for the micronized powders. It is important to acknowledge that the result obtained for the T167A micronized tooth (5960 ± 60 BP) was derived from a minuscule, non-duplicated sample and was measured utilizing the gas source (larger uncertainty due to diminished counting statistics). The radiocarbon ages from these 23 new analyses range between 6092 ± 27 BP (T514, Level 1) to 5801 ± 25 BP (T105, Level 9) (Supplementary information 6). They are stratigraphically consistent as their ages are increasingly younger from Level 1 at the bottom of the stratigraphic sequence to Level 9 at the top. It is important to note that there is some variability in age within each level, with values typically ranging from 0 (Level 1) to 99 (Level 3) 14C years. This variability can be attributed, at least in part, to the residual presence of diagenetic carbon. The slight discrepancy observed between the two pre-treatment methods (typically less than 100 14C years) nevertheless indicates that the ages determined on micronized samples are highly accurate. This finding corroborates the hypothesis proposed by Zazzo21, which was based on a comparison of ages measured on enamel, dentin, and root samples from the same tooth at Mehrgarh.

Subsequently, the tooth enamel dataset was modelled using a Bayesian approach24, with the objective of proposing an estimate of the dating and duration of the cemetery. In the model, only the new dates derived from micronized enamel samples were considered. The aforementioned dates obtained from charcoal were deliberately excluded for two reasons. Firstly, most of these previous charcoal dates were performed prior to the advent of AMS, so their ages are associated with a larger uncertainty than AMS now permits to obtain6 (typically more than 100 14C years; Supplementary information 2 & 5). Secondly, in his previous publications (including the most recent one), Jarrige22 correctly identified that several charcoal dates were outliers and corresponded to intrusive samples. The model (Supplementary Information 8) is relatively straightforward, comprising a series of nine phases, represented by the nine successive cemetery levels. Each phase is separated by two boundaries, reflecting the presence of occupation layers between each cemetery. The time elapsed between each cemetery level is also modelled.

The v1.0 model is moderately supported, with a value of Amodel = 51. In the OxCal software, a model is deemed plausible as soon as its Amodel value exceeds 60. Given that the samples were gathered from solid archaeological contexts, it is plausible that the observed result may be linked to the fact that some samples have undergone slight rejuvenation as a result of diagenesis. In model v1.1, two dates in Levels 3 and 5 with the lowest Amodel (T509, Amodel = 25 and T551, Amodel = 32) were manually rejected. This model is much better supported (Amodel = 98). Despite the similarity in dating and duration between the two models, the main results of model v1.1 are presented here (Fig. 2; Supplementary Information 7).

The model suggests that Mehrgarh Neolithic cemetery started at the end of the 6th millennium BCE or the beginning of the 5th millennium BCE (5223 − 4914 cal. BCE) and ended during the first third of the 5th millennium BCE (4769 − 4679 cal. BCE), spanning between 186 and 531 years. The relatively large uncertainty in the estimated starting date is attributed to a calibration plateau between 5200 and 5000 BCE. In light of these considerations, two distinct probability estimates emerge: the first at 5200 BCE (with a 32% probability) and the second between 5000 and 4900 BCE (with a 62% probability). This plateau also explains that the Bayesian statistical process produced a bimodal span calculation with two distinct possibilities for the duration of Mehrgarh Neolithic Period I: the first one between 186 and 337 years (64% prob.) and the second one between 476 and 531 years (31% prob.) (Fig. 3). The time elapsed between two consecutive cemetery levels is usually very short, around a human generation for cemetery Levels 5–9 (less than 40 years) and slightly longer for cemetery Levels 1–5 (less than 82 years, except for Levels 3–4, although this could be caused by the calibration plateau). In general, the model tends to favour relatively short interruptions between two consecutive cemetery levels, with medians ranging from six years (between Levels 2 and 3) to 19–20 years (between Levels 3 and 4 and between Levels 4 and 5).

Fig. 2
Fig. 2
Full size image

Result of the v1.1 Bayesian model of Mehrgarh Period I.

Fig. 3
Fig. 3
Full size image

Probability for the duration of Mehrgarh Period I.

Discussion

The dating and duration of Mehrgarh Period I

The new radiocarbon dates provide indisputable evidence that Period I dates to a timeframe between 5250 and 4650 cal. BCE. These results are incompatible with Jarrige’s hypothesis that Period I started around 8000 BCE and ended before 6000 BCE. In fact, not only are these results internally consistent, but they are also consistent with the majority of the currently available archaeological data, while contradicting the primary arguments used by Jarrige to support his hypothesis (Supplementary Information 3).

Firstly, these results are consistent with and corroborate the majority of previously available radiocarbon dates from Neolithic contexts at Mehrgarh. Indeed, the existing set of radiocarbon dates from Mehrgarh Period I contains 19 dates between approximately the mid-6th and mid-5th millennia BCE. In contrast, the five radiocarbon-dated samples from Mehrgarh aceramic Neolithic Period I that Jarrige highlighted as supporting his view22 are inconsistent, ranging from approximately 9200 to 5500 cal. BCE, with only two dates partially overlapping with each other6 (see Supplementary Information 2 & 5).

Secondly, the new radiocarbon dates concur with the findings of field research projects conducted between the 1990s and 2010s in Iran and Pakistan. These studies have demonstrated that non-pottery Neolithic settlements existed in the Indo-Iranian Borderlands well after the advent of pottery in Mesopotamia and Iran during the 7th and 6th millennia BCE. Such settlements have been identified in southern Pakistan, specifically in the Kech-Makran region in the southwest and in the Las Bela region and Sindh province in the southeast. Excavations at Miri Qalat and Shahi-Tump in Kech-Makran have yielded aceramic deposits radiocarbon dated to the 5th millennium BCE6. The sites located in the southeast have only been subjected to surface surveys. The available radiocarbon dates from these sites indicate a period between the mid-6th and late 5th millennia BCE25. Furthermore, excavations at Kili Ghul Mohammed in the Quetta Valley, situated approximately 100 km from Mehrgarh, have yielded aceramic deposits radiocarbon dated to the mid-5th through the mid-4th millennia BCE26 (see Supplementary Information 1). This evidence indicates the existence of a late aceramic Neolithic (or even Chalcolithic) horizon in the Indo-Iranian Borderlands27. This proposition has recently been reinforced by findings from southeastern Iran, where excavations at Tell-e Atashi, situated approximately 850 km west of Mehrgarh, have unearthed the remnants of aceramic farming communities dated by radiocarbon analysis to a period spanning between the late 6th and the early 5th millennia BCE, coinciding with the majority of the aforementioned radiocarbon dates from Mehrgarh Period I6,15,28,29,30,31,32.

Thirdly, Jarrige identified a number of parallels between the material remains from Mehrgarh Period I and those from aceramic Early Neolithic sites in western Iran. These include, for example, the use of “thumb-impressed” mud bricks, quadrangular architecture, and clay figurines. Nevertheless, comparisons can also be made with remains that are geographically closer and chronologically compatible with the new radiocarbon dates mentioned above. These parallels include the construction of quadrangular rooms using “thumb-impressed” mud bricks in the 6th and 5th millennia BCE at Tell-e Atashi, as well as at Tepe Yahya and Tal-i Iblis in southeastern Iran. Clay figurines exhibiting a striking resemblance to those that characterize the Mehrgarh Neolithic period have also been documented at Tell-e Atashi, as well as at Tepe Zagheh in northern Iran. This site was occupied between 5200 and 4300 BCE6,15,33,34,35. These comparisons highlight a number of elements in Mehrgarh’s material culture that have exact parallels at sites more recent than the Early Neolithic sites Jarrige pointed out. Furthermore, a number of parallels he noted for other categories of remains, including walls painted with red ochre, circular hearths filled with stones, microlithic industry, blades, axes, shaft straighteners (Supplementary information 2 & 3), are observed over a wide area in Iran and beyond, and for a long time during the Neolithic and even the Chalcolithic periods. They are not exclusive to Early Neolithic sites. However, the aforementioned specific similarities observed at sites in Iran do not necessarily imply that these settlements are similar to Mehrgarh or were founded by the same cultural group(s). Many aspects of Mehrgarh remain original, and only a comprehensive reappraisal of all the Neolithic evidence at the site would help to answer this question.

A noteworthy finding pertains to the duration of Mehrgarh Period I. The Bayesian model indicates that the nine successive cemetery levels lasted between 186 and 531 years, with typically brief intermissions between two successive cemetery levels. The probability of the range being between 186 and 337 years (64%) is higher than that of the range being between 476 and 531 years (31%). This duration appears to be relatively brief when one considers that the sequence in question probably accumulated over nine to ten meters and consists of mudbrick dwellings that were then abandoned and whose ruins and abandonment layers were then used for garbage disposal and/or as burial areas, where new dwellings were later built (Supplementary information 2 & 3). In total, the same sequence of events was repeated on nine occasions, resulting in the formation of nine cemetery levels that alternated with as many architectural levels. Despite the recognition of an issue with the dating of Period I, the extended duration of this period assumed by Jarrige22 appeared plausible. Consequently, some researchers proposed an alternative scenario in which the extensive Neolithic sequence at Mehrgarh, spanning over a millennium, was repositioned to the 6th and 5th millennia BCE instead of the 8th and 7th millennia BCE6,15. However, the new dates demonstrate that Period I was at least two and potentially up to five times shorter. Given that this period comprises nine iterations of the same architectural, abandonment, cemetery and subsequent abandonment cycle, it seems more plausible to hypothesise that Mehrgarh Period I sequence of Levels 1 to 9 lasted for a longer period of time, commencing around 5223 cal. BCE instead of around 4914 cal. BCE. It is noteworthy that the upper part of the range (531 years) is consistent with the lower estimate suggested by C.A. Petrie17, which was “between 540 and 720 years”. This range of dates also suggests that each architectural phase during Period I may have lasted for a period of time much shorter than the 30–40 years that have been previously proposed by Jarrige36. In light of these findings, it can be posited that the inhabitants of Mehrgarh engaged with the mudbrick structures of Period I in a transient manner, contrary to the conventional wisdom. In addition, these findings highlight the necessity for a revision of the current understanding of archaeological site formation. This would entail a reassessment of the regularity with which buildings were rebuilt, the time required for alluvial deposits to form, and the time span between the initial deposition of human remains in a cemetery and its subsequent use as a dumping ground or construction site. In any case, it seems prudent to consider the possibility that recurrent flooding from the nearby Bolan River may have accelerated local sediment accumulation and, indeed, all of these processes.

These new radiocarbon dates have further implications for the dating of the emergence of pottery in northwest South Asia. Indeed, Mehrgarh Period IIA pottery vessels, which appeared subsequent to Period I, represent the earliest known pottery products from Pakistan. Although Jarrige paralleled them with material from Iran and Mesopotamia dating to the second half of the 7th millennium BCE (see Supplementary Information 3), the new dates suggest that they are no earlier than ca. 4650 BCE, which is relatively consistent with parallels observed in southeastern Iran6 as well as with the available radiocarbon dates from Period IIA (Supplementary information 2 & 5: the three available dates from Period IIA suggest 4704, 4494, and 4443 BCE as the earliest possible dates for this period). This new terminus post quem also implies that the ceramic vessels produced during the following Mehrgarh Periods IIB and III37,38 are more recent than previously thought. Furthermore, it suggests that different regional pottery traditions then emerged relatively quickly in northwest South Asia during the 4th millennium BCE. This is a shorter timeframe than deemed before, which essentially corresponds to the end of Mehrgarh Period III and the following Mehrgarh Periods IV and V.

As previously stated in relation to Tell-e Atashi, the absence of pottery in Mehrgarh Period I, at a time when pottery was widely produced, indicates that the populations at this site not only did not engage in pottery production, but also potentially did not store, cook, eat and exchange motifs in the same manner as ceramic Neolithic populations in other regions of the Iranian Plateau. The existence of parallels in the Neolithic subsistence economy and material culture between the aceramic and ceramic Neolithic occupations in the Indo-Iranian Borderlands suggests the existence of interactions in this area. It is therefore likely that the aceramic groups from Tell-e Atashi and Mehrgarh were aware of pottery technology. Furthermore, there is no evidence to suggest that the fuel required for firing the vessels was unavailable. Consequently, although a comprehensive explanation for this phenomenon remains elusive, the only hypothesis that can be posited is that these groups chose not to utilize ceramic vessels15,32.

The dating and process of inception of agriculture in Northwest South Asia

The archaeological deposits of Mehrgarh Period I contain the earliest evidence for agricultural life in northwest South Asia. Our results therefore suggest that agriculture emerged in this area around 5000 BCE. Moreover, the existence of numerous Neolithic sites in Iran and Central Asia dating from the 7th and 6th millennia BCE, with evidence of even earlier settlements in Iran, indicates that it emerged as a consequence of diffusion. This diffusion hypothesis is not in contradiction with the aforementioned parallels between the cultural remains from Mehrgarh and those from sites in Iran. Furthermore, it is consistent with the majority of currently available bioarchaeological data, particularly the fact that animals and plants were domesticated from the outset of Period I and were predominantly West Asian in character7,18,39 (see Supplementary Information 1, 2 & 3). Nevertheless, the hypothetical diffusion in question is challenging to trace at this time, as there is currently a dearth of Neolithic data between Mehrgarh and eastern Iran to the west, and this site and southern Central Asia to the north. In terms of distance, the nearest Neolithic sites to Mehrgarh are located in southeastern Iran. These sites include Tell-e Atashi (along with other sites surveyed in the Bam region)32 and Tepe Yahya in the Soghun Plain, as well as Tepe Gaz Tavila and several other sites in the Daulatabad Plain. Tepe Yahya and Tepe Gaz Tavila are situated in the western portion of the Kerman province, approximately 1,050 km and 1,080 km, respectively, from Mehrgarh. The archaeological records of these sites include settlements of agrarian communities engaged in the production of vegetal-tempered pottery vessels. The currently available data allows for a general dating of these remains between the early-mid 6th and early-mid 5th millennia BCE, although the chronological setting of these remains requires further investigation and reassessment6,40,41,42. Excavations at an additional site, Tepe Gav Koshi, also located in western Kerman, have yielded evidence of an even earlier Neolithic ceramic occupation. The levels exposed at this site are divided into two phases, the first of which dates to between ca. 7180 − 6690 cal. BCE, and the second to the second half of the 7th millennium BCE. The pottery from these phases is characterised by finely painted decorations, which is unusual in comparison with other early pottery vessels found in Iran at this time, especially those recorded in the first half of the 7th millennium BCE43.

To the north, between northeastern Iran and Turkmenistan, a number of archaeological sites have been identified that date to the Neolithic period, with the earliest evidence dating to the 7th and 6th millennia BCE. This evidence includes the aceramic and pottery Neolithic deposits at Sang-e Chakhmaq West Mound and East Mound, respectively44,45, and the pottery Neolithic occupation at Jeitun in Turkmenistan46,47,48. To the east, hundreds of kilometres distant, lie the archaeological deposits excavated in the vicinity of Ak Kupruk in Afghanistan and Obishir V in Kyrgyzstan, where the remains of domesticated sheep dating to around 6000 BCE have been identified6,49,50. The areas situated in the middle, between Uzbekistan and parts of Turkmenistan, are characterized from the 7th millennium BCE onwards by the presence of mobile groups of hunters, fishermen and herders, making incised, mineral-tempered pottery vessels. These groups are collectively known as the Kel’teminar archaeological complex51,52,53. Additionally, the Hissar archaeological culture in western Tajikistan, currently dated to between the 6th and 4th or 2nd millennia BCE, is purported to have included sheep and goat herders, although this assertion requires clarification50,54.

Thus far, no evidence has been uncovered to suggest that the Neolithic occupation of Mehrgarh was connected to the Neolithic settlements in China and northern India. However, now that Mehrgarh Period I (and II) has been securely dated to two to three millennia later than previously thought, it may be worthwhile to consider whether there was any relationship between the Neolithic occupation of this site and eastern settlements of the same period. Conversely, the existence of the aforementioned Neolithic sites and cultures to the west and north of Mehrgarh prior to Period I may suggest that these western and northern settlements exerted a direct or indirect influence on the development of agricultural village life at this site. It is possible that a demic or cultural diffusion originated in southeastern Iran, where the material and bioarchaeological remains of Period I (and II) at Mehrgarh have a number of parallels (see above; Supplementary Information 2 & 3). In addition to the southeastern Iranian sites, it is crucial to consider the northeastern Iranian and Central Asian sites, where connections with Mehrgarh are evident. For instance, links with the Kel’teminar archaeological complex are attested through the shared use of similar horned trapezes. The presence of turquoise at Mehrgarh also suggests a connection since sources for turquoise are located within the area of this archaeological complex. Furthermore, the lapis lazuli artifacts discovered within the Neolithic deposits at Mehrgarh also indicate a northern origin, particularly from northeastern Afghanistan, where a major source for this mineral is situated6 (Supplementary Information 2).

Although the hypothesis of a diffusion from somewhere between southern Iran and Central Asia seems reasonable, it is also important to remember that certain sites in these regions exhibit different configurations, suggesting that early, local, and independent inventions may have occurred. This is exemplified by Ghar-i Mar and Ghar-i Asp near Ak Kupruk in northern Afghanistan, where the hunting of wild caprines appears to have become a significant activity during the Late Palaeolithic period, eventually leading to their domestication sometime between the mid-11th and mid-8th millennia BCE49,55 (see also comparable examples at Komishani Tappeh near the Caspian Sea and in the Tang-e Bolaghi in Fars6,47,56,57,58). It is possible that G. Possehl59 was correct in his assertion that the southwestern Asian nuclear core zone for domestication was larger than previously assumed, encompassing not only the Near East but also the Iranian Plateau, Central Asia, and the Indus Valley. In light of this assertion, it is imperative to acknowledge the potential existence of numerous Neolithic sites within the regions between Mehrgarh and Kerman, Kerman and northeastern Iran, and Mehrgarh and southern Central Asia. It is possible that these sites have not been discovered due to a lack of exploration in certain regions, or because they have been covered by alluvial deposits or obliterated by fluvial or wind erosion. It is therefore possible that the most effective response to the question of how agriculture originated in northwest South Asia, whether through processes of diffusion or independent invention, or a combination of both, may be found in these areas, if not entirely lost. In a more general sense, the hypothesis that a general eastward impulse from the Fertile Crescent existed at the time of the emergence of agriculture between Iran, Central Asia, and South Asia is probably not sufficient to explain the full range of archaeological evidence. The very long timespan between the first Neolithic settlements in the Zagros Mountains and those on the Indus Plain suggests that the question of the origin and spread of agricultural practices should be conceived as a multidirectional process, not limited to a west-east direction, and staggered. It may, therefore, be appropriate to conceptualize this process as “Neolithization” in the plural6. Nevertheless, the extant archaeological record in the Indus region, or northwest South Asia, lacks sites that are comparable to the Early Neolithic sites in the Near East, including those in the Zagros Mountains and Tepe Gav Koshi in Kerman. It is possible that configurations similar to those recorded at these sites and in the Ak Kupruk region will be observed at a site in Pakistan in the future. However, the fact remains that Mehrgarh cannot be considered an Early Neolithic site, or more precisely, that such a configuration has not been observed in the levels studied so far at this site. The Neolithic occupation of Mehrgarh has now been demonstrated to be considerably more recent. Moreover, it is linked from its earliest stratigraphic layer onwards to other late Neolithic and even Chalcolithic sites.

In conclusion, although numerous aspects of Mehrgarh are undoubtedly distinctive, the most plausible hypothesis that accounts for the majority of the current chronological, material, and bioarchaeological data from this site and its parallels is that a group or groups of people settled at this site sometime between 5200 and 4900 BCE. They brought with them plants and animals that they had previously domesticated in other locations, as well as comprehensive knowledge of the diverse agricultural and hunting techniques they then employed in the cultivation and utilization of the plant and animal resources of the Kachi Plain. The settlement was established on an alluvial fan, a pattern also observed in southeastern Iran60. The enterprise was successful, as evidenced by the numerous grain storage structures that characterize Mehrgarh Period II. Additionally, the zebu, an animal native to the plain, was domesticated “soon” after the settlement’s establishment. This seems to be the most plausible explanation for the beginning of agriculture in northwestern South Asia based on the available evidence.

Materials and methods

The tooth material utilized for dating was obtained from 23 graves, representing the nine cemetery levels that comprise the sequence of Mehrgarh Period I (Fig. 4; Supplementary Information 2 & 4). A single individual for each grave (T162 and T167 are double inhumations) was sampled for dental enamel of a premolar or molar tooth. The enamel surface was cleaned, and the dentin was removed with a Dremel tool in order to isolate the enamel from the rest of the dental tissue. Subsequently, the enamel (approximately 100–650 mg) was pulverized using a steel mortar and pestle, followed by grinding in an agate mortar to a particle size of < 100 microns. Subsequently, the powder was subjected to further grinding using a McCrone Micronizer Retsch, in accordance with the methodology delineated by Wood et al.23. In summary, this approach facilitates a more effective separation of crystallite clusters and enhances the efficiency of the acetic acid pre-treatment, which is designed to remove diagenetic carbonates. To achieve this, the powder was ground three times for 10 minutes, with a 5-minute pause between each stage to avoid heating. In some cases, a portion of the hand-milled fraction (less than 100 microns) was retained for comparison with the conventional pre-treatment approach. Subsequently, the resulting powder was subjected to a pre-treatment process under a light vacuum for a duration of 17 to 20 hours, utilizing a solution of 1 N acetic acid (equivalent to approximately 1 ml of acetic acid for each 50 mg of powder). Due to relatively low extraction yields (approximately 50–75%), some particularly small samples (511-513-552-582-560) were pre-treated with a more diluted solution (0.5 to 0.33 N) in order to retain sufficient powder for dating purposes. Subsequently, the pre-treated powder was rinsed with milliQ water and dried at 50 °C in an oven. Subsequently, a quantity of powder ranging from 55 to 216 mg was subjected to a reaction under vacuum with orthophosphoric acid at a temperature of 70 °C for approximately 20 minutes. The CO₂ released was then separated cryogenically from the water produced and chemically purified, thanks to several passages through a trap filled with a copper-silver wool mixture. The quantity of gas thus obtained was then measured, and samples containing less than 100 micrograms of C were sealed in a glass microtube, while others were graphitized. The 14C ages were measured on the compact AMS ECHoMicadas using the GIS device for micro-samples. When possible, samples were measured twice, and the radiocarbon age was averaged.

Fig. 4
Fig. 4
Full size image

Plan of Mehrgarh archaeological area with location of the Neolithic graveyard (burials in grey) including location of the graves that were sampled for radiocarbon dating (in red).