INTRODUCTION
Bone collagen is the most often targeted molecule for the direct radiocarbon dating of skeletal remains. This is mainly because the integrity and abundance of this protein can be assessed through a variety of tests including extraction yield, C/N ratio, and more recently, FT-IR analysis (Brock et al. Reference Brock, Geoghegan, Thomas, Jurkschat and Higham2013; Lebon et al. Reference Lebon, Reiche, Gallet, Bellot-Gurlet and Zazzo2016). However, there are a number of contexts where collagen is not preserved. It can be rapidly degraded in arid and tropical environments under the combined influence of heat and water (Hedges Reference Hedges2002; Zazzo and Saliège Reference Zazzo and Saliège2011; Dal Sasso et al. Reference Dal Sasso, Maritan, Usai, Angelini and Artioli2014; Maurer et al. Reference Maurer, Person, Tütken, Amblard-Pison and Ségalen2014). Anthropogenic modifications of bones, through burning, also lead to the destruction of bone collagen, leaving the archaeologist with the mineral phase of bones and teeth, bioapatite (Shipman et al. Reference Shipman, Foster and Schoeninger1984; Stiner et al. Reference Stiner, Kuhn, Weiner and Bar-Yosef1995). 14C dating of the carbonate in bioapatite has long been considered unreliable, but progress in the pretreatment protocols, together with a better understanding of bone crystallographic properties and post mortem diagenetic processes, helped to renew the debate. In the late 1990s, it was discovered that calcined bone apatite could be reliably dated due to the intense crystallographic reorganization occurring when bones are exposed to high (>600°C) temperatures (Lanting et al. Reference Lanting, Aerts-Bijma and van der Plicht2001). This allowed archaeologists to date the remains of cremated humans (Nakamura et al. Reference Nakamura, Sagawa, Yamada, Kanehara, Tsuchimoto, Minami, Omori, Okuno and Ohta2010; Olsen et al. Reference Olsen, Hornstrup, Heinemeier, Bennike and Thrane2011; De Mulder et al. Reference De Mulder, Van Strydonck, Annaert and Boudin2012), but it was later found that 14C dating of calcined bones is not without limitations. First, a series of laboratory (Hüls et al. Reference Hüls, Erlenkeuser, Nadeau, Grootes and Andersen2010; Van Strydonck et al. Reference Van Strydonck, Boudin and De Mulder2010) and field (Zazzo et al. Reference Zazzo, Saliège, Lebon, Lepetz and Moreau2012; Snoeck et al. Reference Snoeck, Brock and Schulting2014) experiments demonstrated that a significant part (35–95%) of the carbon in calcined bones does not originate from the bone itself but from the combustion environment. Therefore, modification of the 14C content of the bone during cremation is possible if the fuel and the bone initial 14C content differ from each other (Olsen et al. Reference Olsen, Heinemeier, Hornstrup, Bennike and Thrane2013). Second, if postburial modifications appear limited for Holocene samples, the stability of calcined bones for more ancient (Paleolithic) periods remains questionable (Zazzo et al. Reference Zazzo, Lebon, Chiotti, Comby, Delque-Kolic, Nespoulet and Reiche2013). Despite these caveats, calcined bone dating has become routine, at least for Holocene samples. The situation is less clear for unburnt bone and tooth enamel apatite. Pioneer work showed that bone apatite dating could provide reliable ages in arid environments (Saliège et al. Reference Saliège, Person and Paris1995), allowing the reconstruction of the chronology of human occupations in the Sahara based on direct dates (Paris and Saliège Reference Paris and Saliège2007; Sereno et al. Reference Sereno, Garcea, Jousse, Stojanowski and Saliège2008; di Lernia et al. Reference di Lernia, Tafuri, Gallinaro, Alhaique, Balasse, Cavorsi, Fullagar, Mercuri, Monaco, Perego and Zerboni2013; Berkani et al. Reference Berkani, Zazzo and Paris2015). However, large-scale surveys across the globe suggest that isotopic exchange between bone apatite and soil carbonate starts rapidly following burial, leading to age shifts increasing as a function of time (Zazzo and Saliège Reference Zazzo and Saliège2011; Zazzo Reference Zazzo2014; Cherkinsky et al. Reference Cherkinsky, Glassburn and Reuther2015). Age shifts (always towards younger ages) appear limited (<300 14C yr) in samples younger than 8000 BP, confirming that bone apatite can provide good estimates of the sample age for recent (mid-Holocene) samples. However, this was demonstrated by the comparison of 14C ages obtained from coexisting apatite carbonate and collagen, and it is possible that samples that lack collagen would offer a different (higher or lower) surface area for exchange with the burial environment, leading to different conclusions. Unless the precise calendar age of the remains can be assessed independently, this approach is limited to skeletal remains for which collagen is available for dating, and therefore cannot be applied to calcined or burnt bones.
This study aims at comparing the reliability of different types of apatite fractions for which collagen cannot be dated. We focused on the cremated remains of individuals found at the Roman necropolis of Porta Nocera near the city of Pompeii, and for which the date of burial can be assessed independently.
MATERIAL AND METHODS
The necropolis of Porta Nocera is located in Pompeii, Italy. It was excavated between 2003 and 2007 and consists of several enclosures belonging to free slaves (Van Andringa et al. Reference Van Andringa, Duday and Lepetz2013) (Figure 1). The tombs are aligned to the southeastern exit of the city, alongside the road going to the gate of Stabies and to the harbor located to the mouth of the River Sarno. A funerary monument was built in the northern part of the enclosure, facing the road (Figure 2). It is composed of a high podium on top of which sits an edicule with a pediment presenting three statues. A vaulted niche was laid out on the podium, facing the interior of the enclosure. The stele epigraph indicates that this space was destined to receive the remains of the enclosure’s holder, Publius Vesonius Phileros (tomb 1). The tomb’s dedication specifies also that the deceased was an emancipated slave who previously belonged to Gaia Vesonia, whose cremated remains were deposited and discovered in the same enclosure (tomb 2). The concession (enclosure 23 OS), which was founded by Phileros around AD 60, is located within a funerary area well documented by epigraphy. This concession was installed on an area where several graves were already present since 40–30 BC. A detailed stratigraphic analysis, together with the analysis of the associated archaeological material (ceramics, coins, oil lamps, etc.), allowed to precisely date the burials (tombs 1 to 48) sometimes within a 10-yr uncertainty. This analysis revealed that the most ancient graves (T38 and T48) date from between 20 BC and AD 30, with the most recent date from AD 79, i.e. the day of the destruction of Pompeii following the eruption of the Vesuvius. The details of the phase determination and dating of the graves are given in Van Andringa et al. (Reference Van Andringa, Duday and Lepetz2013: 125–7). Next to this enclosure, another enclosure (enclosure 21 OS) was excavated. This enclosure also received one of Gaia Vesonia’s emancipated slave (Stallia Haphe) as well as other deceased persons (including tombs 201 and 203). Pyres located in the immediate vicinity were also excavated.
Map showing the extension of the excavated area within the Porta Nocera necropolis (Pompeii, Italy) and the position of the sampled features (pyres and tombs).
I. Location of Pompeii; II. Funerary monument from the enclosure of Publius Vesonius phileros (Enclosure 23 OS); III. Tomb of Gaia Vesonia (tomb 2); IV. Tomb of Bebryx, young 6-year-old slave (tomb 201); V. Examples of the sampled material. A: Calcined human bone from tomb 25. B: Unburnt human teeth from tomb 29. C: Calcined human teeth from tomb 25. D: Burnt Donax sp. from tomb 10 (SU 232124-6). E: Unburnt Murex sp. from occupation layer (SU 132501).
Eleven individuals were selected for 14C dating (for a detailed description of the samples and contexts, see the Online Supplementary Material). Individual ages of the individuals were determined, when possible, following the methods of physical anthropology and are provided in Van Andringa et al. (Reference Van Andringa, Duday and Lepetz2013: 862–5). Among them, three children were not incinerated, allowing dating of both the enamel and collagen fraction of the bone for two of them (T29 and T38). During the Roman period, very young children were usually not incinerated even if the cremation of adults was the usual practice. For the remaining eight individuals, degraded collagen from burnt bone, as well as bone apatite from burnt and calcined bone, was dated. Charcoal and shells found in association with the tombs or the pyre were also dated in order to estimate the local reservoir effect as well as the influence of the old-wood effect. Bone samples were powdered (<100 μm) prior to pretreatment using a mortar and pestle. For enamel, two different protocols were tested. Enamel samples were pretreated first in the form of millimeter chunks, then in the form of fine (<100 μm) powder. Charcoals were prepared following the classical acid-alkali-acid treatment. They were first immersed in 0.5N HCl for 3 hr, then in 0.05N NaOH for 0.5 hr, then in 0.5N HCl for 16 hr. Degraded collagen from burnt bone was isolated using a similar treatment: 10% HCl for 0.5 hr; 0.1N NaOH for 0.5 hr; 0.1N HCl for 0.5 hr. Bone and enamel apatite were pretreated using 1N acetic acid under weak vacuum at room temperature for 18 to 22 hr. The surface of the shells was acidified in 10% HCl for 10 s. The shells were rinsed, then finely powdered. Carbon dioxide (CO2) derived from carbonate in shell aragonite and bone and enamel apatite was extracted using orthophosphoric acid (H3PO4) for 15–30 min at 70°C under vacuum. Charcoal and degraded collagen was combusted at 500°C in the presence of pure oxygen. Two charcoal samples were large enough to be dated using the classical liquid scintillation method. Preparation of benzene and counting were performed at the LOCEAN lab (UPMC, Paris). For the other samples, sealed tubes containing between 1–2 mg C in the form of CO2 were sent to the 14C laboratory of Tucson, Arizona, USA, for graphitization and 14C measurement. Carbon isotope values were measured using isotope ratio mass spectrometry (IRMS). Pretreated shell (0.05 mg), unburned bone and enamel apatite (0.6 mg), and calcined bone apatite (1.8–2.7 mg) were reacted with 100% orthophosphoric acid at 70°C in individual vessels in an automated cryogenic distillation system (Kiel IV device), interfaced with a Delta V Advantage isotope ratio mass spectrometer. Over the period of analysis of the bioapatite samples, the analytical precision estimated from 16 samples of the laboratory internal carbonate standard (LM Marble, calibrated against NBS-19) was ±0.03‰ (1σ).
RESULTS
The results are presented in Table 1 and Figure 3. Charcoal ages (n=4) range between 1960±27 and 2328±50 14C yr BP. Shells (n=6) show a very large range in 14C age, from 2400±39 to 4035±41 14C yr BP. The dated human bones show an even larger age range, between 1805±49 and 5570±120 14C yr BP. Multiple sampling on some of the individuals shows large (up to 1200 14C yr) age variability. Carbon content in calcined bones is very variable and ranges between 0.06 and 0.60 wt% (0.28±0.19% on average). Although based on a lower number of specimens, the carbon content of burnt bone (0.82%) and enamel (0.71 and 0.82%, respectively, for powdered and crushed enamel) appears higher and less variable. The carbon isotope value of calcined bone also is variable and ranges between –13.2 and –23.9‰. Calcined bone carbon content is negatively correlated with δ13C values (R 2=0.51) and positively correlated with 14C age (R 2=0.36). Correlations increase to 0.80 and 0.67, respectively, if only samples with 14C age higher than 2200 14C yr BP are considered.
Relationship between the amount of carbon, the 14C age (A) and the carbon isotope value (B) of the apatite samples from the Porta Nocera necropolis.
Carbon content, carbon isotope value, and radiocarbon age of the Porta Nocera samples.
* Dates measured by liquid scintillation at the LOCEAN Lab (UPMC, Paris).
DISCUSSION
The ages of the burials range between 20 BC and AD 79. If we apply the IntCal13 14C calibration curve (Reimer et al. Reference Reimer, Bard, Bayliss, Beck, Blackwell, Bronk Ramsey, Grootes, Guilderson, Haflidason, Hajdas, Hatté, Heaton, Hoffmann, Hogg, Hughen, Kaiser, Kromer, Manning, Niu, Reimer, Richards, Scott, Southon, Staff, Turney and van der Plicht2013) together with the OxCal v 4.2 software (Bronk Ramsey and Lee Reference Bronk Ramsey and Lee2013), we can convert these calendar ages into 14C ages. By doing so, we calculate that the 14C ages of individuals buried in the Porta Nocera necropolis should range between about 1900 and 2035 14C yr BP. The most ancient graves (Grave 38 and 48) are from children, so no further correction due to the individual age of the deceased is necessary. While one date on degraded collagen from Grave 6 is more recent possibly due to contamination with young organic carbon (1805±49 14C yr BP), most (18/21) of the 14C ages measured on the skeletal remains from Porta Nocera are therefore too old. Deviation from expected age is small for collagen (0–150 14C yr) but can reach 3500 14C yr for calcined bone. In the following, we discuss the respective impact of diet, old-wood effect, and diagenesis to explain this unexpected 14C age offset.
Marine diets are responsible for the incorporation of a marine reservoir age in skeletal remains (Yoneda et al. Reference Yoneda, Tanaka, Shibata, Morita, Uzawa, Hirota and Uchida2002; Zazzo et al. Reference Zazzo, Munoz and Saliège2014). In the Roman Empire, marine food was common in the diet but probably more restricted to people with a high social status. At Herculaneum, the coupling of 14C and stable isotopes performed in bone collagen of individuals who died during the AD 79 eruption showed that variable amounts of marine resources were included in their diet, affecting bone collagen 14C ages by up to 84 14C yr (Craig et al. Reference Craig, Bondioli, Fattore, Higham and Hedges2013). In our study, the collagen of two individuals could be dated. These two individuals were 1–2-yr-old children; therefore, most of their dietary input would have come from their mother’s diet through breastfeeding. The first one (Grave 38, 1977±37 14C yr BP) shows no evidence of marine food in its diet as its calibrated age (36 BC–AD 65, 1σ) in is perfect agreement with the estimated burial date (20 BC–AD 30). The other one (Grave 29, 2127±27 14C yr BP) is in strong disagreement with the estimated burial date (AD 70–79), suggesting that marine food could be responsible for a ~150 14C yr offset. This represents approximately 29–36% of the marine reservoir age of the Mediterranean near Naples (415–515 14C yr according to Siani et al. Reference Siani, Paterne, Arnold, Bard, Mativier, Tisnerat and Bassinot2000), suggesting that the mother’s diet during and after her pregnancy was composed for a large part of marine food.
In the case of cremated bones, diet is not likely to play a major role because most of the biogenic carbon is replaced by fuel-derived carbon during calcination (Hüls et al. Reference Hüls, Erlenkeuser, Nadeau, Grootes and Andersen2010; Zazzo et al. Reference Zazzo, Saliège, Lebon, Lepetz and Moreau2012; Snoeck et al. Reference Snoeck, Brock and Schulting2014). This raises the possibility of the transfer of an inbuilt age from the wood to the bone if large trees (or trees cut several decades earlier) are used for the cremation (Olsen et al. Reference Olsen, Heinemeier, Hornstrup, Bennike and Thrane2013). An anthracological study was performed at Porta Nocera and revealed the presence of wood of various sizes, including small branches, bundles of wood, and larger trunks. The presence of nails and marks of squared in the pyres may also indicate the use of waste wood coming from elements of furniture (Coubray Reference Coubray2013). Thus, an inbuilt age is to be expected and its magnitude must be evaluated. The age of the contexts in which the charcoals were found is known with a precision of 10–20 calendar years, allowing to estimate the expected age offset due to inbuilt age. These upper and lower age limits were first converted into a 14C age range via the IntCal13 calibration curve (Reimer et al. Reference Reimer, Bard, Bayliss, Beck, Blackwell, Bronk Ramsey, Grootes, Guilderson, Haflidason, Hajdas, Hatté, Heaton, Hoffmann, Hogg, Hughen, Kaiser, Kromer, Manning, Niu, Reimer, Richards, Scott, Southon, Staff, Turney and van der Plicht2013) (Table 2), following the methodology described in Olsen et al. (Reference Olsen, Heinemeier, Hornstrup, Bennike and Thrane2013). Testing the converted dates against the charcoal 14C dates results in age differences ranging between 24±43 and 398±69 14C yr (160±164 14C yr on average, n=4). Open-air experiments have shown that carbon isotope exchange between the bone and the fuel ranges between 35 and 95% (Hüls et al. Reference Hüls, Erlenkeuser, Nadeau, Grootes and Andersen2010; Zazzo et al. Reference Zazzo, Saliège, Lebon, Lepetz and Moreau2012; Snoeck et al. Reference Snoeck, Brock and Schulting2014). This leads to calculate that, on average, calcined bone dates can be shifted by 56–152 14C yr due to exchange with old-wood carbon. The eight cremated individuals died between AD 40 and 79. This translates to 14C ages ranging between 1896 and 1978 BP. If we take in account the average age shift calculated above, we can estimate that the maximum age in calcined bone due to the old-wood effect should be about 2130 BP. Only 3 out of the 10 cremated samples are below this threshold. Burnt bone apatites are beyond, together with the two enamel samples (non-cremated) for which exchange with wood carbon cannot be invoked.
Estimation of the impact of wood inbuilt age on Porta Nocera calcined bone age.
Instead, we propose that the 14C depletion measured in archaeological apatites is due to the influence of the 14C-free CO2 emissions of the nearby Vesuvius volcano and the Campi Flegrei volcanic system at the Bay of Naples (Pasquier-Cardin et al. Reference Pasquier-Cardin, Allard, Ferreira, Hatté, Coutinho, Fontugne and Jaudon1999; Chiodini et al. Reference Chiodini, Caliro, Cardellini, Granieri, Avino, Baldini, Donnini and Minopoli2010). Although never measured before in humans, this pattern has already been observed in soils and plants living at the vicinity of a volcano (Pasquier-Cardin et al. Reference Pasquier-Cardin, Allard, Ferreira, Hatté, Coutinho, Fontugne and Jaudon1999). This “volcano effect” could have been recorded in vivo (through diet) or during fossilization (through isotope exchange with CO2 dissolved in percolating waters). Degassing of dead CO2 at the Bay of Naples could also explain the anomalously high 14C ages measured on some seashells (Table 1). Several lines of evidence suggest that diagenesis, rather than diet, can explain the pattern observed in archaeological apatites. First, while expected 14C activities were measured in bone collagen, anomalously high 14C ages were measured in enamel from the same individuals (T29 and T38). Since the two sampled individuals were newborns, differences in enamel apatite and bone collagen turnover cannot explain the difference in 14C activity between the two tissues. Second, the fact that we measured a different 14C activity in the same enamel samples treated according to two different protocols (powdered vs. millimeter chunks) also points to a contamination issue. Enamel 14C activity increased by 2 pMC when treated in the form of powder, and became closer to 14C activities measured in bone collagen. Powdered samples also contained less inorganic carbon following treatment (0.68 vs. 0.73 and 0.74 vs. 0.89 weight% for Grave 29 and 38 enamel, respectively). This confirms previous findings showing that the acid acetic treatment is more effective at removing secondary carbonates on finely powdered enamel than on small fragments (Zazzo Reference Zazzo2014). Thus, it appears clearly that diagenesis, i.e. post mortem exchange of carbon between apatite and dissolved inorganic carbon, is responsible for the measured age shift. In this context, diagenetic carbon is depleted in 14C compared to the buried skeletal remains. This conclusion appears to contradict the findings from a previous study showing that during diagenesis, the carbon source available for isotopic exchange is always enriched in 14C (i.e. younger) compared to the fossil remains (Zazzo Reference Zazzo2014). This conclusion was based on the 14C dating of more than a hundred fossils buried in different contexts including 14C-free sediments such as caves (Hedges et al. Reference Hedges, Lee-Thorp and Tuross1995; Zazzo et al. Reference Zazzo, Lebon, Chiotti, Comby, Delque-Kolic, Nespoulet and Reiche2013) and limestone rocks (Zazzo et al. Reference Zazzo, Munoz and Saliège2014). In surface or near-surface finds like archaeological burials or habitats, carbonates dissolved in percolating waters are close to the isotopic equilibrium with the atmosphere and have therefore a younger 14C age than the fossils, leading to their apparently younger age as times goes by. Volcanic contexts, by providing a constant input of 14C-free CO2 to the soil, appear to be the exception that proves the rule.
The largest age shifts are measured in calcined bones. This could lead to the conclusion that calcined bones are more prone to diagenetic alteration than burnt bone, or enamel apatite, contradicting previously published evidence (Zazzo and Saliège Reference Zazzo and Saliège2011; Zazzo Reference Zazzo2014), but a closer look at the data indicates that this may actually not be the case. A mass balance calculation allows estimating the contribution (in %) of dead carbon to the measured age. We considered two different end-members for initial (i.e. pre-diagenetic) 14C age, depending on whether or not apatite ages are influenced by an old-wood effect (for calcined and burnt bones) or a marine diet (for enamel). As discussed above, both factors will lead to a maximum age of 2130 BP, which corresponds to a fraction modern carbon (F) value of about 0.7674. The results (Table 3) show that on average, contamination represents 8–10% of the total carbon in calcined bones depending on the initial F value chosen for the calculations. This is intermediate between burnt bones (4–6%) and powdered enamel (10–12%). However, large (0–35%) interindividual variations are calculated, and we note that calcined bone samples that deviate the most from the expected age are usually the ones that present the lowest carbon contents (Figure 3). These percentages were then multiplied by the carbon concentration of the apatite samples and converted in amount of dead carbon. By doing so, we calculate that calcined bones contain on average 0.8–1.4 μg dead C, i.e. 3.6–4.1 times less than burnt bones (3.3–5.0 μg dead C) and 6.1–8.6 times less than powdered enamel (7.0–8.3 μg C). This conclusion is more in line with previous work showing that postburial carbon isotope exchange is much lower in calcined bone than in enamel, unburnt and charred bone altogether, due to recrystallization during heating (Zazzo and Saliège Reference Zazzo and Saliège2011; Zazzo Reference Zazzo2014). Our results show that a very small amount of contamination can impact the 14C age of a sample that contains very little carbon (typically less than 0.3%). One of the side effects of bone exposure to high temperatures is the lowering of the carbonate content of bone (Person et al. Reference Person, Bocherens, Mariotti and Renard1996; Zazzo et al. Reference Zazzo, Saliège, Person and Boucher2009). Thus, there is a tradeoff between the advantage (recrystallization) and the disadvantage (lower carbon content) of calcination. Because volcanic CO2 and calcined bone carbonate δ13C values differ by about 20–25‰ (Pasquier-Cardin et al. Reference Pasquier-Cardin, Allard, Ferreira, Hatté, Coutinho, Fontugne and Jaudon1999; Hüls et al. Reference Hüls, Erlenkeuser, Nadeau, Grootes and Andersen2010), contamination can also be monitored using calcined bone apatite δ13C values. As a general guiding rule, we propose that a prescreening using both carbon content and δ13C values with thresholds set at 0.3% and –20‰, respectively, could be used to select the best samples for 14C dating in volcanic contexts.
Estimation of the amount of dead carbon in Porta Nocera apatites.
CONCLUSION
This study demonstrates that in volcanic contexts, large 14C age shifts can be measured in relatively recent archaeological apatites, even calcined ones. We show that while a marine diet or an old-wood effect could explain the smallest age shifts, they are not able to explain the largest ones, and we thus propose diagenesis as the main factor. This unexpected result was likely caused by the large amounts of dead C diffusing in the soil following the AD 79 eruption, which created a strong gradient in 14C concentration between the biogenic and diagenetic end-members and caused an increase in the apparent age of fossil apatites. It emphasizes the need to be cautious when working in volcanic environments. This study also has implications for apatites found in nonvolcanic contexts. Even if large age offsets were sometimes observed in calcined bones from Pompeii, most of them showed 14C ages that could easily have been interpreted as “normal” without the strong stratigraphic and chronological constraints available for this historical site excavated with great care. Our quantitative estimate of the amount of dead carbon in the different apatite fractions provides a way to quantitatively assess the intensity of C isotope exchange in different types of apatite materials. It confirms the overall strong resistance of calcined bones relative to burnt bone and enamel apatite and suggests that 14C dating of calcined bones should be restricted to samples with high (>0.3%) carbon content and low (< –20‰) δ13C values.
ACKNOWLEDGMENTS
The authors wish to thank H Duday for the determination of the anatomical pieces. Stable isotope analysis was performed by J Ughetto (SSMIM, MNHN). We are grateful to C Pierre for access to her lab at the LOCEAN (UPMC, Paris), to J-F Saliège for sample preparation, and to the staff of the Tucson lab for graphitization and AMS measurements. The interpretation of the results benefitted from discussions with J-F Saliège. We thank also Alex Cherkinsky as well as two anonymous reviewers for their insightfull comments.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/RDC.2016.28
