FIRST RADIOCARBON DATING OF NEOLITHIC STONE CIST GRAVES FROM THE AOSTA VALLEY (ITALY): INSIGHTS INTO THE CHRONOLOGY AND BURIAL RITES OF THE WESTERN ALPINE REGION

Abstract

Previous research on the Neolithic cist graves of the Western Alpine region—also known under the term Chamblandes type graves—mostly focused on sites located in western Switzerland and eastern France. For the adjacent Aosta Valley (Italy), only a little information is available. Within the framework of our research project, it was possible to identify about 120 stone cist graves from 10 sites in the Aosta Valley. Due to the lack of distinctive grave goods and missing absolute dating, however, their chronological position has been unclear until now. Here we present the first extensive series of radiocarbon dates from Neolithic stone cist graves of the Aosta Valley. We analyzed 31 human bone samples from four sites, and most dates indicate an unexpected early chronological position around the first half of the 5th millennium BCE, in particular, the site of Villeneuve, dating to 4800–4550 cal BCE. This identifies these burials from the Aosta Valley as belonging to the oldest known Neolithic cist graves of the Western Alpine region discovered so far. Altogether, our study provides new evidence allowing the first time to clarify the chronology of these sites and trace the evolution of this burial practice in the Western Alps.

INTRODUCTION

In the Western Alpine region Neolithic cist graves are named after the eponymous site of Pully, Chamblandes (Moinat and Studer 2007) and have a surprisingly long period of use of around 1000 years and date largely between 4800–3800 BCE (see discussion). Jeunesse et al. (2019) placed the emergence of this burial type in the regions of Savoy and Ain (eastern France). Regionally, the study of Chamblandes type graves has been traditionally centered within the surroundings of Lake Geneva and the Upper Rhône Valley, in Switzerland and France respectively (e.g., Gallay 2008; Honegger 2011; Stöckli 2016; Baudais et al. 2017). These two regions describe an area to the north of the main Alpine ridge in the northwestern Alps and in the catchment area of the river Rhône. The adjacent Aosta Valley (Italy) lies south of the main Alpine ridge and drains into the Po River. All these regions are connected by alpine passes, which usually have altitudes of over 2000 m a.s.l.

While the northern areas of Lake Geneva and the Upper Rhône Valley are relatively well studied, only a few analyses have covered the stone cist graves in the Aosta Valley. In particular, there was a complete lack of radiocarbon (¹⁴C) data in this region, which made a proper chronological contextualization of the archeological findings challenging. This knowledge gap is particularly unfortunate given the promising scientific potential of this region, as the Aosta Valley is connected via high-Alpine passes and therefore an ideal area for the analysis of transalpine Neolithic exchange networks. The distribution of stone cist grave sites (Figure 1a) shows the broader intra-regional significance of this burial type. They are situated on both sides of the Alpine ridge, in the border areas of larger Neolithic cultural zones in eastern France, northern Italy, and southern Germany. About 120 stone cist graves pertaining to 10 different sites have been identified in the Aosta Valley. These consist of two large cemeteries (Vollein and Villeneuve) and several old discoveries or unexcavated sites (Figure 1b). Neolithic graves of the Chamblandes type usually consist of four lateral stone slabs forming a rectangular box or cist of varying small dimensions. Individuals were placed on the grave floor, generally on their left side and with the lower limbs flexed (Figure 2a). Most stone cist graves include single burials, whereas multiple burials of at least three individuals have been described only sporadically in the unpublished documentation of the Vollein excavation. Chamblandes graves typically include only a few grave goods, for the mentioned sites in the Aosta Valley, these are mainly large Glycymeris shell bracelets (one example visible on Figure 2a) and small jet disc beads. This scarcity hampers the typochronological and cultural historical interpretation of these finds (Denaire et al. 2011; Stöckli 2016). Accordingly, serial ¹⁴C dating becomes essential, not only for a chronological reconstruction of the contexts, but also for a better understanding of the complex social networks characterizing both sides of the Alps during the periods of emerging farming societies.

Figure 1 (a) Examined sites of Neolithic cist graves in the Western Alps and (b) sites identified in the Aosta Valley (Italy).

Figure 2 (a) Tomb 31 from the stone cist necropolis of Quart, Vollein (Photo: Soprintendenza di Aosta), (b) excavation of the necropolis in Villeneuve, Champ Rotard in 1917 (Corrain 1986:20).

MATERIAL AND METHODS

In this study, we consider four recently sampled sites:

• Villeneuve, Champ Rotard (Italy): this necropolis consists of 33 stone cist graves; 25 excavated in 1917 and 8 additional graves discovered in 1987 (Figure 2b; Barocelli 1919; Barocelli 1951; Mezzena 1997). So far, only the human remains excavated in 1917 (T1-T25) were anthropologically analyzed (Corrain 1986).

• Quart, Vollein (Italy): the necropolis includes at least 66 stone cist graves (T1-T66) and stands as the largest in the Aosta Valley. The human remains excavated between 1968 and 1983 (Mezzena 1981, 1997) were stored without anthropological examination. However, in the context of our project, the bones were cleaned and will be analyzed soon.

• La Salle, Derby (Italy): a single stone cist grave from this previously unknown site was discovered in 1952 and the human remains of three individuals have been collected and partially examined (Fumagalli 1955).

• Montjovet, Fiusey (Italy): 5 stone cist graves (Sep. I-Sep. V) with the remains of at least 8 individuals were discovered and excavated without anthropological examination below an Early Medieval cemetery in 1910 (Rizzo 1910).

Selection of Individuals

For each site, we first screened the available archaeological and anthropological documentation. We then proceeded to a re-evaluation of the basic demographic parameters for each individual: We estimated adult age-at-death based on the morphological changes of the pubic symphysis and auricular surface of the ilium (Brooks and Suchey 1990; Buckberry and Chamberlain 2002). For nonadults, we estimated age-at-death based on the development and eruption of deciduous and permanent teeth and the degrees of epiphyseal fusion (Ubelaker 1989; Scheuer and Black 2000; Schaefer et al. 2009). We determined sex only of adult remains following standard anthropological methods based on the sexually dimorphic features of the pubic symphysis, coxal bone, cranium, and mandible (Buikstra and Ubelaker 1994). To avoid double sampling, only human remains that could be clearly assigned to a specific individual and grave number were considered for this study. The specimens were selected in such a way to minimize possible age- or sex biases.

Sample Collection

For each selected individual we collected a bone sample of ca. 3 g, taken from the petrous portion of the temporal bone. We targeted this anatomical region since (a) this region presents a good portion of highly mineralized compact bone which makes it ideal for sampling (Lösch et al. 2020), and (b) it allows to minimize the risk of double-sampling. In the few cases where more than one individual was buried in a grave, double sampling was avoided by considering only temporal bone from one side and/or belonging to individuals of obvious different age-at-death (i.e., adults vs. nonadults). These criteria led to the following sample composition collected in September 2019:

• 11 individuals from 10 graves of the Villeneuve necropolis corresponding to approximately a third of all the graves (n=33),

• 12 individuals from 8 different graves of Vollein could be sampled, which corresponds to about 12% of the necropolis,

• Among the rediscovered bones labeled to be from the stone cist grave from Derby, 3 individuals could be distinguished and sampled.

In July/August 2021 the few remaining human bones from the site of Montjovet (stored in the Musei Reali in Torino, Italy) were examined by the laboratory of ancient DNA of the University of Bologna, and 1 g of 4 different petrous bones could be sampled. It was however not possible to assign these bones to specific individuals or graves mentioned in the archaeological documentation. The human remains were heavily disturbed, and 3 of the 5 graves were destroyed during the Second World War (Fumagalli 1955). Further, as a surprise, in April 2022 one complete stone cist grave with the human remains of 1 individual in original position was rediscovered in the archive of the Musei Reali in Torino and subsequently sampled for this study.

Treatment

The 31 samples were processed in the laboratory of the Department of Physical Anthropology (IRM) of the University of Bern and transferred to the Laboratory for the Analysis of Radiocarbon with AMS (LARA) of the University of Bern for dating. The sample preparation was slightly modified from Szidat et al. (2017) by implementation of an ultrafiltration step. In brief, the bones were cleaned by ultrasonication in ultra-pure water and ground to 0.5–1 mm with a ball mill. The chemical treatment included the following steps: 4% hydrochloric acid (HCl) for 60 hr, 0.1 mol/L sodium hydroxide (NaOH) for 1 hr, 4% HCl for 1 hr, followed by a gelatinization in diluted HCl at pH 3 and 60°C overnight. The warm solution was filtered using precleaned Ezee-Filters, ultrafiltration was performed with Vivaspin™ 15 30 kDa molecular weight cut-offs (MWCO) ultrafilters (Sartorius) and the high-molecular-weight fraction was lyophilized. The extracted collagen was combusted and graphitized with an automated graphitization equipment (AGE). The ¹⁴C measurements were performed with the accelerator mass spectrometry (AMS) system MICADAS using ¹⁴C-free sodium acetate and the NIST standard Oxalic Acid II (SRM 4990C) for blank subtraction, standard normalization, and correction for isotope fractionations (Szidat et al. 2014).

Exploratory Check for Freshwater Reservoir Effect

In order to check for possible freshwater reservoir effect on our ¹⁴C estimates, we run a preliminary analysis of stable isotopic ratios of carbon (δ¹³C) and nitrogen (δ¹⁵N) on a subset of seven individuals (three each from Villeneuve and Vollein, and one from Derby). Isotope ratios were measured by isotope ratio mass spectrometry (IRMS) at Isolab GmbH, Schweitenkirchen, Germany, using the average of three measurements per sample. Results are reported in δ-notation as units per mill (‰) according to the international standards of Vienna Pee Dee Belemnite (V-PDB) for carbon and Ambient Inhalable Reservoir (AIR) for nitrogen. Moreover, the laboratory internal standards STD R (collagen from cowhide from the EU project TRACE) and STD BRA (collagen from Brazilian cowhide) were used. Internal analytical errors were recorded as ± 0.1‰ for δ¹³C, ± 0.2‰ for δ¹⁵N (standard error of the means calculated from 3 or 4 measurements).

Isotopic ratios were first explored visually and then included in Bayesian models of dietary composition using FRUITS (Fernandes et al. 2014), including as dietary resources C₃ plants, terrestrial animals, and freshwater fish. Previous isotopic research strongly suggest that millets were, in the Italian peninsula, part of human diet only starting from the Early Bronze Age (Varalli et al. 2022). For this reason, we did not include C₄ plants in the model. Faunal and paleobotanical samples are at present not available from the analyzed contexts. The model was therefore run with no priors and using as proxies published isotopic values for C₃ plants and terrestrial herbivores/omnivores (averaging values for bovids, caprinae, pigs, and red deer) from Bronze Age western Switzerland (Varalli et al. 2021), and from Bronze Age Savoy for freshwater fish (pike) (Varalli et al. 2021). Given the location of the sites (the nearest coastline is some 250 km away), we assume that the dietary exploitation of marine resources was quite unlikely for prehistoric times and was therefore not considered.

Calibration and plots of the raw ¹⁴C data were conducted with the Oxcal 4.4.4 software (Bronk Ramsey 2009, 2021) using the IntCal20 atmospheric curve (Reimer et al. 2020). The two-sigma probability interval was used discussing the ¹⁴C results (as recommended by Millard 2014). Quality control for the data focused on the atomic C/N ratio and the collagen yield (%w/w) (van Klinken 1999; Szidat et al. 2017). Sufficient collagen preservation of bone samples is indicated by a C/N ratio between 3.10 and 3.30 and collagen yields ≥0.5%. In contrast to earlier work (Szidat et al. 2017), we narrowed the range of the C/N ratio conservatively to 3.10–3.30 due to the observation that the distribution of both good and still acceptable bone samples widely falls into these boundaries, whereas C/N ratios above 3.30 may occasionally show biased results. A similar approach was recently applied by Zazzo et al. (2019).

RESULTS

Table 1 shows all the obtained ¹⁴C data. Considered valid ¹⁴C dates are displayed in Figure 3 as a multi-plot (a) and summarized using overlapping sum and kernel density estimation (KDE) plot (b) with the purpose of showing the most likely use time of the burial sites. While the sum distribution often exhibits sharp drops and rises associated with features from the calibration curve (Bronk Ramsey 2017), the frequentist method KDE can be applied to characterize the overall age range while removing much of the frequency variability of sum distributions (Loftus et al. 2019).

Table 1 ¹⁴C results of the analyzed bone samples (n=31).

* All temporal bone samples were taken from the petrous portion.

† Invalid data, indicating where it does not meet the quality control criteria.

Figure 3 Calendar ages of the considered valid ¹⁴C dates (n=21) displayed as (a) multi-plot and (b) overlapping sum and KDE plots using the software OxCal 4.4.4 with default settings (Bronk Ramsey 2017) based on the data from Table 1.

Villeneuve

The LARA was able to generate a valid result for 64%, or 7 out of 11 samples, the remaining 4 did not meet the quality control criteria and were rejected. Thus, a valid ¹⁴C date is now available for 21% of the 33 stone cist graves from the Villeneuve necropolis. The raw dates (n=7) are almost identical dating in the period between 4720 and 4550 cal BCE, with the exception of the slightly older individual from T17 (4798–4681 cal BCE).

Vollein

For all samples (n=12) a valid result could be generated. The bones were stored unwashed and untreated right after their excavation, and we assume that these circumstances may explain the good collagen preservation. The data of the bone samples show a larger spread. The oldest dating yielded the individual from grave T31 (4611–4455 cal BCE; Figure 2a). The youngest result is from individual 1 of grave T50 (4442–4265 cal BCE). Based on the age determination of individual 2 from T50 (4445–4332 cal BCE), assumed to have been deposited simultaneously with individual 1, a dating to the 43rd century BCE is rather unlikely. With this exception addressed, the graves date in the period between 4600 and 4350 cal BCE.

Derby

A valid result could be generated for all 3 samples. The dates of individuals 1 and 2 are almost identical, dating in the period between 4700 and 4450 cal BCE (chronologically situated between the dated individuals of Villeneuve and Vollein; Figure 3b). Individual 3 yielded an Early Medieval date, it can be assumed that this third skull fragment does not originate from the Neolithic stone cist (therefore this dating was not considered for the investigation).

Montjovet

Three of the five analyzed samples yielded a valid ¹⁴C result. However, with dating ranges between 435–775 cal CE these were not considered for the investigation. Based on the archaeological record, Late Antique or Early Medieval dating for these graves seems highly unlikely and a mix up with bones from the overlying cemetery during the long and disturbed storage of these human remains is most likely.

Stable Isotopes of Carbon and Nitrogen: Preliminary Results and Explorative Analysis

Table 2 shows the isotopic values of the 7 human samples. All samples fit the quality criteria. Stable carbon and nitrogen isotopic ratios range respectively from –19.43 to –19.99 (average: –19.69 ± 0.15) and from 9.17 to 10.64 (average: 9.75 ± 0.47). Figure 4 illustrates the results of the Bayesian modeling of dietary composition. These are consistent with a mixed diet including C₃ plant products (64.9%). Conversely, the exploitation of terrestrial animals and freshwater food resources seems quite low (respectively 15.5% and 19.5%).

Table 2 Stable isotope ratios of carbon and nitrogen from bone collagen of the 7 analyzed individuals.

Figure 4 Estimates of dietary composition based on the average δ¹³C and δ¹⁵N values from human bone collagen. Horizontal continuous and dashed lines show the mean and median values respectively.

DISCUSSION

Our ¹⁴C dates are the first available for Chamblandes type burials in the Aosta Valley. They significantly expand our knowledge of the chronological timespan for the use of this type of grave. Previous hypotheses placed the stone cist graves from this region in the Late Neolithic, or even the Bronze Age (Mezzena 1997). Our data, especially those from Villeneuve, allow to substantially revise these estimates, rather pointing to an earlier chronology. The Bayesian results based on an overlapping phase model (Figure 5; A_model = 84.2, A_overall = 80.5), frame the burial activity within the Villeneuve necropolis to 4737–4606 cal BCE (68.4%)/4785–4555 cal BCE (95.4%) and within the Vollein site to 4502–4320 cal BCE (68.4%)/4555–4305 cal BCE (95.4%). This result is of special interest since it places the finds from the Aosta Valley among the oldest Chamblandes type graves ever discovered.

Figure 5 Independent bounded phase model of ¹⁴C dates from the necropolises of Villeneuve and Vollein using OxCal 4.4.4 software (Bronk Ramsey 2009).

Regarding the burial activity, contemporaneity is statistically unlikely for all available dates of the two necropolises; modeled in one phase they span 331–387 (68.4%)/286–426 (95.4%) years. Regarding the sites themselves, the span of dates from Villeneuve is 0–63 (68.4%)/0–131 (95.4%) years and as it includes zero, they could be isochronous. In contrast, the dates from the site of Vollein span 36–151 (68.4%)/18–177 (95.4%) years. However, excluding the oldest sample (T31), the span of dates from Vollein is 6–72 (68.4%)/0–109 (95.4%) years and contemporaneity is statistically possible. This indicates that the data from Vollein should be divided into an older, and a younger main phase. Finally, the combination of the Villeneuve-dates with sample T31 from Vollein, results in a span of 0–112 (68.4%)/0–194 (95.4%) years. This indicates that the oldest date from Vollein could statistically be isochronous to the burials from Villeneuve. Taken together, the sample T31 is difficult to contextualize—both within Vollein and compared to Villeneuve—and further ¹⁴C dates would be required to dismiss or confirm the older phase from Vollein.

Table 3 Selection of the most important sites of Neolithic cist graves in the northwestern Alpine region. Published ¹⁴C dates were recalibrated with the Oxcal 4.4.4 software (Bronk Ramsey 2009) using the IntCal20 atmospheric curve (Reimer et al. 2020), data with uncertainties over ± 65 yr were not considered.

Figure 6 KDE plots derived from independent bounded phase model of ¹⁴C dates (n=112) from the sites listed in Table 3 using OxCal 4.4.4 software (Bronk Ramsey 2009). The colors refer to the geographical area (green: Aosta Valley, blue: Tarentaise Valley, purple: Upper Rhône Valley, red: Lake Geneva South, orange: Lake Geneva North, gray: Swiss Plateau, teal: Bugey).

Table 3 lists the most important Neolithic stone cist sites in the Western Alpine region featuring multiple ¹⁴C data. The considered ¹⁴C dates (n=112) were modeled in independent bounded phases (A_model = 99, A_overall = 89.7). Figure 6 shows the KDE plots ordered by their respective median to visualize the likely period of use for these necropolises. From this overview it appears that the only context presenting a ¹⁴C age as old as those from Villeneuve is the vast necropolis of Thonon-les-Bains, Genevray (Haute-Savoie, France). In addition, a single date each from the two sites of Bourg-Saint-Maurice, Le Châtelard (Haute-Savoie, France) and Sion, Sous-le-Scex (Valais, Switzerland) indicate burials starting in the 48th century BCE. However, only few data are so far available from Bourg-Saint-Maurice, Le Châtelard (Rey et al. 2018) and the current ¹⁴C data from Sion, Sous-le-Scex exhibit relatively high uncertainties of ± 65 yr (Honegger 2011), resulting in a modeled median date of 4510 cal BCE (see Figure 6). Taken together, the comparison of these data suggests the intriguing hypothesis of an almost contemporaneous emergence of the use of stone cist burials in the southern shore of Lake Geneva (Thonon-les-Bains, Genevray) and several inner-Alpine Valleys (mainly Aosta, but most probably also the Tarentaise and Upper Rhône Valley).

Further, we observe the wider inter-regional spread of stone cist graves occurring after 4500 BCE, mainly to the north (Swiss northern shore of Lake Geneva), northeast (Swiss Plateau) and west (Ain, France). Moreover, our new data show that the use of stone cist graves for burials in the Aosta Valley ceased around 4300 BCE. This is earlier than in neighboring areas, where the demise of this funerary custom can be placed around 4000/3800 BCE. Isolated burials from the late 4th or 3rd millennium BCE (identified within the sites of Thonon-les-Bains, Genevray and Lausanne, Vidy) indicate the sporadic reuse of stone cist graves in younger periods, an interesting phenomenon which deserves further investigations.

This assessment of the chronology and emergence of Neolithic cist graves in the Western Alpine region is supported by the fact that the earliest pile dwelling sites on the southern foothills of the Alps were found at Lake Varese (Lombardy, Italy), where the beginning of farming activities was dated around ca. 5000–4850 cal BCE (Antolín et al. 2022), whereas the earliest settlement of this type north of the Alps was found in Egolzwil (Lucerne, Switzerland) dating to 4300 cal BCE (Stöckli et al. 2013). Therefore, it was suggested that parts of the Neolithic community from Lake Varese could have migrated to north of the Alps, possibly across the Upper Rhône Valley (Antolín et al. 2022). In that sense, ¹⁴C data of Neolithic settlement layers in the region of Sion (Valais, Switzerland) date to after 4750 cal BCE and increase in the second halve of the 5th millennium BCE (Piguet 2011). In addition, evidence for high Alpine mobility between the Upper Rhône Valley and the Swiss Plateau during this period was found on top of the Schnidejoch mountain pass in the western Bernese Alps (Bern, Switzerland) at 2750 m a.s.l, where 5 wood samples of arrow shafts date to 4800–4500 cal BCE (Hafner 2015). Therefore, these finds are even older than the famous Iceman from the more eastern Italian Alps, who dates to 3370–3100 cal BCE (Kutschera and Rom 2000). In this context, also the few available grave goods found within stone cist graves in the Aosta Valley indicate far reaching exchange networks of these inner Alpine societies, as the mentioned Glycymeris shell bracelets originate from the Mediterranean Sea (Borrello et al. 2009), whereas the raw material of the disc jet beads probably came from southern Germany (Rochna 1962; Baudais et al. 2017).

Furthermore, a large-scale comparison with the new data shows that the Western Alpine cist graves are largely contemporaneous with settlements and graves of the Upper Rhine area (southern Germany/eastern France). Here, the oldest Neolithic settlements of the region with Linear Pottery (LBK) are followed by a phase of related ceramic styles—Hinkelstein, Grossgartach, Rössen, Bischheim occidental du Rhin supérieur (BORS)—dating to 4800–4000 cal BCE (Denaire et. al 2017). Further connections to the Upper Rhine area exist through the discovery of a typical shoulder cup in one of the stone cist graves at the site of Däniken, Studenweid (Solothurn, Switzerland), number 44 on Figure 1a (Steuri and Hafner 2022).

Based on our preliminary analysis of stable carbon and nitrogen isotopic ratios a freshwater reservoir effect seems unlikely in the present specimen. Naturally, these estimates need in any case to consider three important caveats: (a) the small size of the analyzed sample (n=7, i.e., 23% of the sampled individuals), (b) the lack of faunal samples from the analyzed sites, which strongly limits our interpretation, and (c) freshwater food webs vary greatly in stable carbon and nitrogen isotope ratios depending on the species and additionally the species values show large standard deviations (e.g., Bösl et al. 2006). These three reasons combined make it difficult to estimate a potential freshwater reservoir effect. However, given the long distance of the sites to the nearest coastline, we assume the dietary exploitation of marine resources as quite unlikely.

It should additionally be noted that we investigated almost exclusively petrous bones for ¹⁴C dating in this work, whereas the earlier studies (listed in Table 3) applied different bone material such as other skull parts or long bones. All of these bones were selected for the sake of collagen preservation, although none of these allows the determination of the time of death directly due to a slow carbon turnover, which typically results in ¹⁴C ages of the bones that are a few decades older than the date of the death (Meadows et al. 2020; Indra et al. 2022). As the petrous bone is already formed in early childhood (Jørkov et al. 2009; Meadows et al. 2020), this offset may even be a few decades larger than for both other skull parts or long bones, which should be taken into account for the comparison of the results from the Aosta Valley from this work with results from earlier studies. As the modeled age ranges span about a century (Figure 5), however, we regard this possible difference of a few decades as negligible.

CONCLUSION

This study provides new evidence about the emergence of a specific burial practice of 5th millennium BCE Neolithic cist graves of the Western Alpine region, and new data about the development of social and cultural transalpine networks. It should be noted that anthropological remains from cist graves of Aosta Valley are typically few and badly preserved. As such, the data presented in this contribution might represent some of the only absolute dating potentially available from the stone cist graves of this region. Our project includes more than 120 new ¹⁴C dates from cist graves in the Aosta Valley and neighboring alpine regions and preliminary data does not show major deviations from the chronology provided in this study.