In order to interpret strontium and oxygen isotope values in Neolithic human skeletons analysed previously, we begin to map the biologically available strontium, carbon and oxygen isotopic signatures of prehistoric southern Germany by analysing tooth enamel of pigs from archaeological sites distributed around the region. The mapping shows a marked upland–lowland difference in biologically available87Sr/86Sr values, ranging between 0.7086 and 0.7103 in the sedimentary lowlands, and from 0.710 to as high as 0.722 in the crystalline uplands of the Odenwald, the Black Forest and the Bavarian Forest. In addition, carbon isotopes in the carbonate fraction of pig enamel were generally about 1–2 more enriched in13C in the uplands. Despite the expected depletion of18O with altitude, oxygen isotopes in pig enamel showed little correlation with site altitude, although for pig samples not older than the Iron Age there was some geographical correlation withδ18O patterns in modern precipitation.


By identifying migrants through signatures in their very skeletons, strontium and oxygen isotope analysis can offer substantial evidence for characterizing prehistoric migrations. By comparing the isotope values in archaeological tooth enamel to expected signatures from the region, one hopes to identify the geographical area where the diet was obtained during childhood (when the enamel was forming).

In southern Germany, strontium isotope analysis has previously helped to identify non-local individuals in Linearbandkeramik (e.g., Price et al. 2001, 2003; Bentley et al. 2002, 2003b), Bell Beaker (Grupe et al. 1997) and Roman communities (Schweissing and Grupe 2003). To help identify the geographical origins of prehistoric individuals analysed previously and in the future, in this paper we provide details of biologically available strontium and oxygen isotopes in southern Germany through analysis of tooth enamel from archaeological pigs.

Isotopes and geographical origins

Unlike carbon, nitrogen and oxygen, strontium isotopes are conveyed, without the ratio of 87Sr to 86Sr measurably changing, from the weathering of rock minerals (with, importantly, different weathering rates and Sr concentrations) through soils and into the food chain. In mammalian tooth enamel, strontium substitutes for calcium in the hydroxyapatite mineral. Because 87Sr/86Sr values in rock minerals vary, depending on their original Rb/Sr content and geological age (as 87Rb radioactively decays to 87Sr with a half-life of 48 billion years), strontium isotopes can be measured as geological, and hence geographical, signatures within archaeological skeletons (e.g., Ericson 1985; Price et al. 2002; Knipper in press). As described below, the geology of southern Germany (Fig. 1) is excellent for Sr isotope analysis of skeletons because the upland areas, underlain by granites (old rocks with high Rb/Sr), exhibit substantially higher 87Sr/86Sr values than the sedimentary lowlands in which carbonates having low Rb/Sr ratios are predominant.

Figure 1.

The geology of southern Germany, showing sites mentioned in text, as well as the mean 87Sr/86Sr values measured in archaeological pig enamel from each site (see Table 2 and text). Main area: 1, Bruchenbrücken; 2, Schwanfeld. Odenwald: 3, Schnellerts. Rhine Valley: 4, Goddelau; 5, Sasbach; 6, Bischoffingen. Neckar area: 7, Ilsfeld; 8, Poltringen. Swabian Alb: 9, Heidengraben; 10, Urspring. Black Forest: 11, Münstertal; 12, Wieladingen. Bavarian Forest: 13, Wildstein; 14, Warberg. Pre-Alpine lowlands: 15, Mintraching; 16, Wittislingen; 17, Wang; 18, Altdorf; 19, Inzigkofen. Hegau: 20, Hilzingen; 21, Singen-Offwiese (Klein-Denkte not shown).

Oxygen isotope compositions (δ18O = 1000 ×[(18O/16Osam)/(18O/16OSMOW) − 1], all δ18O figures in this paper reported relative to Standard Mean Ocean Water [SMOW]) in the environment depend upon the fractionation of 18O versus 16O during evaporation, condensation and precipitation in the hydrological cycle, with 18O preferentially retained in the liquid phase. Since the fractionation factor for oxygen (on the Earth's surface) shows an approximately inverse dependence on temperature, the mean annual δ18O in precipitation depends largely on latitude and altitude (Bowen and Wilkinson 2002). Precipitation at higher altitudes is generally more depleted in 18O. In the Swiss Alps, for example, Schürch et al. (2003) report that mean annual δ18O in precipitation varies from −9.7 at Berne (∼ 500 m a.s.l.) to −13.3 at Grimsel (∼ 2000 m a.s.l.). Mean annual δ18O in precipitation also tends to become more negative with distance from the sea, as clouds moving inland preferentially lose 18O in precipitation. In southern Germany, mean annual δ18O in modern precipitation (ranging from −8.2 to −11.2 in 1991) becomes more negative moving eastwards away from the Atlantic Ocean, and southwards with increasing altitude towards the Alps (Fig. 2).

Figure 2.

A contour map, after Tütken et al. (2004), of mean annual δ18O (SMOW) in modern precipitation in southern Germany. At each site is shown the calculated mean annual δ18Omw (, SMOW) in meteoric water converted via equation (1) from pig enamel values (see Table 2 and text). White circles designate Neolithic sites, and solid black circles with underlined δ18O values designate sites from the Iron Age or later.

Because the values reported by Tütken et al. (2004) represent the year 1991 only, Figure 2 does not predict prehistoric δ18O. Not only have values of δ18O been different in the past, but they may have had different geographical patterns. Evidence from sediments in southwestern Germany (Mayer and Schwark 1999, fig. 4) and human bone phosphate from near Moscow (Iacumin et al. 2004, fig. 3) indicate that mean annual δ18O in precipitation was within about ± 1 of its present value during the Iron and Middle Ages, whereas it was about 2–3 more negative in the early Neolithic. Not only has δ18O changed with Holocene climate, but δ18O varies substantially within and between years. For example, δ18O in modern precipitation in the Swiss Alps is about 5–10 more negative during the winter than in the summer (Schürch et al. 2003, fig. 6 (a)). At the early Roman sites of Dangstetten (Baden-Württemberg) and Kalkriese (Niedersachsen), seasonal δ18O variations of up to 4 appear in enamel phosphate of single teeth from horses and mules (Paulus 2000).

As is well known, carbon isotopes (δ13C = 1000 ×[(13C/12Csam)/(13C/13CPDB) – 1], all δ13C figures in this paper reported relative to Pee Dee Belemite [PDB] carbonate standard) fractionate during primary production such that C3 plants have δ13C values between −23 and −34, whereas C4 plants (not expected for prehistoric southern Germany) have δ13C between −9 and −17 (O’Leary 1988). Since δ13C values in consumer bone carbonate generally reflect that of the diet (DeNiro and Epstein 1978), with all dietary and ecological factors being equal we would not expect strict geographical variation in δ13C values within southern Germany, as shown in measurements of archaeological bone (Van Klinken et al. 1994, fig. 2).There is a slight altitude effect, however, as high-altitude (over 1000 m) plants adapted to a lower partial pressure of CO2 show enrichment in plant 13C by a few per mil (Körner et al. 1991). Also, δ13C values become less negative with increasing temperature and isolation, and more negative with increasing precipitation and humidity (e.g., van der Merwe and Medina 1991; Koch et al. 1994; Van Klinken et al. 1994; Schoeninger 1995). An additional effect occurs in dense forests, where plants under the canopy photosynthesize less and take up recycled CO2, such that δ13C values decrease from canopy top to forest floor by about 2 in northern forests (e.g., van der Merwe and Medina 1991; France 1996; Heaton 1999).

Mapping prehistoric isotope signatures

Because biologically available 87Sr/86Sr correlates with geology and δ18O relates to hydrology, the two isotope systems together serve as independent indicators for geographical origin. Clearly, a regional map of biologically available isotope values is invaluable for interpreting isotope signatures in archaeological skeletons. To get a rough idea of the biologically available 87Sr/86Sr values, we can start with published geochemical studies of groundwaters and stream waters of southern Germany (e.g., Grupe et al. 1997; Tricca et al. 1999; Probst et al. 2000; Eikenberg et al. 2001; Douglas et al. 2002), from which we expect lower 87Sr/86Sr values (0.708–0.710) in the lowlands, and higher 87Sr/86Sr values (> 0.715) in samples from area uplands underlain by 300–400 Ma granites and metamorphic rocks, including (Fig. 1) the Vosges, the Black Forest and the Odenwald highlands, as well as the crystalline Bavarian Forest further east. More detail is needed, however, because 87Sr/86Sr varies between different rock minerals, different portions of an individual plant and different streams (e.g., Sillen et al. 1998; Tricca et al. 1999). For δ18O, we start simply with a published map of δ18O in modern precipitation (Tütken et al. 2004), as shown in Figure 2, but as mentioned above this is only a very rough guide for δ18O in prehistoric pigs, because the mean annual δ18O in local precipitation has changed through time.

One way to map regional 87Sr/86Sr is to analyse hundreds of soil leachate, water and plant samples, as Hodell et al. (2004) have impressively done for Mesoamerica. An alternative way, and a more direct measurement of biologically available isotope signatures, is to measure 87Sr/86Sr and δ18O in archaeological tooth enamel of a low-mobility animal species from different excavation sites (Price et al. 2002; Bentley et al. 2004). Through a lifetime of feeding, herbivores obtain a remarkably consistent average 87Sr/86Sr ratio that is representative of their catchment area (Burton et al. 1999; Blum et al. 2001; Price et al. 2002). Similarly, despite the fact that physiological factors complicate the relationship between δ18O in tooth enamel and that of an animal's drinking water (Kohn 1996), differences in enamel δ18O can reflect geographical differences in drinking water sources when comparing individuals of the same species, region and time period (D’Angela and Longinelli 1990; Luz et al. 1990; Kohn 1996; Balasse et al. 2002). The reasons to sample archaeological rather than modern material are that: (1) the modern environment is highly contaminated with strontium from human sources (e.g., Probst et al. 2000); and (2) the mean annual δ18O in precipitation will have changed since prehistory. For 87Sr/86Sr, one might define the ‘local’ range as within two standard deviations from the mean value in archaeological human bones (e.g., Grupe et al. 1997), but this has problems involving post-burial contamination of bones (Horn and Müller-Sohnius 1999). Hence, the reason to use archaeological tooth enamel and not archaeological bone is that enamel is much more resistant during burial to contamination from isotopes in groundwater (e.g., Horn et al. 1994; Koch et al. 1997; Chiaradia et al. 2003; Hoppe et al. 2003; Trickett et al. 2003).

Our study

In this study, we begin to map the biologically available 87Sr/86Sr and δ18O in prehistoric southern Germany, using geographically distributed samples of archaeological pigs. We measured δ18O in the structural carbonate (CO3) component of tooth enamel because the analysis simultaneously recovers δ13C, and the procedure (Koch et al. 1997; Balasse et al. 2002) is considerably faster and easier (in our laboratory, anyway) than measuring δ18O in the phosphate (PO4) component (cf., O’Neil et al. 1994; Stephan 2000; Vennemann et al. 2002). Fortunately, measurements from fossil and modern mammals show that the δ18O values recovered from phosphate (δ18Op) and structural carbonate (δ18Oc) are offset (δ18Op−δ18Oc) by a constant value of about 8.7 (Bryant et al. 1996; Iacumin et al. 1996), meaning that our measurements of the carbonate component can be compared to other measurements of phosphate component. In order to relate the value in pig bone carbonate, δ18Oc, with that of meteoric water, δ18Omw, we use the regression equation of Longinelli (1984; see also D’Angela and Longinelli 1990). By rearranging Longinelli's equation and incorporating the offset of 8.7, we have:


For humans, δ13Cc values in enamel carbonate reflect an average of the whole diet, offset by −9.4 such that a pure C3 vegetarian would have a value of about −13 (Ambrose and Norr 1993; Koch et al. 1994). Usefully, the amino acid requirements and metabolism of pigs are similar to those of humans, such that δ13Csc offsets in pig enamel should be similar; that is, 9–11 more negative than the diet (Hare et al. 1991; Howland et al. 2003; van der Merwe et al. 2003). Pigs are good candidates to represent local isotopes for humans, since pigs have lived locally around farming settlements since the Neolithic (Greenfield 1988; Benecke 1994, 248–60; Bentley et al. 2004), most likely eating human by-products such as rotting vegetables, crop wastes, table scraps and human and animal excrement (Gregg 1988, 118–22). Pigs’ diet would have been especially human-like during the winter if they were housed indoors at that time (Grigson 1982).

Although pigs can live for 20 years, most early Neolithic pigs were probably slaughtered within the first 2–3 years (Gregg 1988, 122). In domestic pigs, the tooth enamel mineralizes starting before birth until the second to third month for M1, between the first and eighth month for M2, P3 and P4, and between the third and 13th month for M3 (Hillson 1986, 207, fig. 3.9).


For 87Sr/86Sr analysis, about 5–20 mg of intact enamel was removed from each pig tooth, soaked overnight in 5% acetic acid (with an intermediate H2O rinse), dissolved in 5 M HNO3, and then purified by extraction chromatography using Eicrom® Sr-spec resin (for details, see Bentley et al. 2003b). We analysed the purified Sr samples on a VG-MicroMass Sector 54 thermal ionization mass spectrometer (TIMS) at the Southampton Oceanography Centre (SOC). During the period in which analyses were made, repeated measurements of the NBS SRM-987 standard (87Sr/86Sr = 0.710248) have yielded an average 87Sr/86Sr value of 0.710252 ± 0.000015 (2 s.d., n= 169).

In order to measure δ18Oc and δ13Cc in the carbonate fraction of the tooth enamel samples, each enamel sample was powdered in an agate mortar and pestle, and soaked overnight in 5% acetic acid to remove post-burial carbonate contamination (Koch et al. 1997). At the UCL Bloomsbury Environmental Isotope Facility, enamel powder samples of about 2 mg were reacted with 100% phosphoric acid at 70°C in individual vials in the Kiel III automated cryogenic distillation system, which is interfaced with a ThermoFinnigan Mat 253 gas-source mass spectrometer. The analytical precision, estimated from repeated analyses of the NBS-19 carbonate standard, is better than 0.1 (1 s.d.) for δ18O and 0.05 for δ13C.


Our pig teeth samples are from 22 archaeological sites (Table 1) that are distributed throughout the southern part of Germany and represent different types of bedrock, although a large proportion of the landscape is covered by loess (Fig. 1). Whenever possible, samples were taken from the M3 molar, with alternative options being P4, M2 and M1, in that order of preference. We collected pig teeth from the sites of Bruchenbrücken, Schwanfeld, Wang, Poltringen, Altdorf, Mintraching (all oldest LBK situated on loess: Stork 1993; Lüning 1997; Uerpmann and Uerpmann 1997), Wittislingen (oldest LBK on loam/loess), Goddelau (oldest LBK on Rhine gravel), Ilsfeld (Late Neolithic on loess) and Inzigkofen (Roman on gravel). The sites of Bischoffingen (Neolithic/Iron Age) and Sasbach (Iron Age), situated on loess, are near the Tertiary volcanic outcrops of the Kaiserstuhl in the Upper Rhine Valley. With very few Neolithic sites in the highlands, often with poor bone preservation, we sampled pig teeth from medieval sites on granites or gneisses, including Schnellerts in the Odenwald (Harre 1994), Münstertal and Wieladingen in the southern Black Forest, and Warberg and Wildstein in the Bavarian Forest.

Table 1. Isotope values in prehistoric animal tooth enamel from sites in southern Germany. Instrumental errors ( ± 1 s.d.) for each measurement are given in parentheses for the last digits
SitePeriodLab. IDSpeciesToothδ18Oc(, SMOW)δ13Cc (, PBB)87Sr/86Sr
Altdorf-AichLBKALTD 2PigYoung molarn.d.n.d.0.70982 (2)
Bischoffingen ‘Waldsberg’E. Lat.BIWA 2PigMand L M223.97 (6)−14.44 (3)0.70817 (1)
Bischoffingen ‘Waldsberg’E. Lat.BIWA 3PigMax L M224.32 (7)−13.40 (4)0.71272 (1)
Bischoffingen ‘Weingarten’NeolithicBIWE 1DogC25.83 (6)−12.58 (3)0.70848 (1)
BruchenbrückenLBKBRUC 4PigMand R M124.63 (4)−14.20 (2)0.71210 (3)
BruchenbrückenLBKBRUC 5PigMand L M125.74 (2)−13.28 (2)0.70907 (7)
GoddelauLBKGODD 13PigM324.47 (3)−13.28 (2)0.70863 (2)
GoddelauLBKGODD 20PigMand L M326.08 (3)−13.31 (3)0.71568 (1)
GoddelauLBKGODD 22PigMax R M323.39 (5)−13.69 (3)0.70953 (4)
HeidengrabenSHa/FLtHEID 1PigMand L M323.11 (6)−14.21 (3)0.70978 (1)
HeidengrabenSHa/FLtHEID 4PigMand R M324.21 (3)−13.07 (3)0.70972 (2)
HilzingenLBKHILZ 12PigMand L M325.85 (4)−12.45 (2)0.71141 (1)
HilzingenLBKHILZ 6PigMand L M324.62 (4)−12.50 (3)0.70633 (1)
IlsfeldMichsbg.ILSF 4PigMax L M225.08 (2)−14.23 (1)0.70991 (1)
IlsfeldMichsbg.ILSF 6PigMax L P422.77 (17)−14.70 (6)0.70961 (1)
IlsfeldMichsbg.ILSF 7PigMand L M223.58 (2)−14.25 (1)0.70977 (1)
InzigkofenRomanINZI 1PigMax M320.87 (4)−14.04 (2)0.71027 (1)
Klein-DenkteLBKKLEI 1PigM2n.d.n.d.0.70990 (3)
MintrachingLBKMINT 1PigMolarn.d.n.d.0.70862 (3)
Münstertal-MartinegelandeMAMUMA 5CattleMax M23.34 (4)−12.54 (1)0.71798 (1)
Münstertal-MartinegelandeMAMUMA 9PigMand L M324.68 (12)−12.47 (2)0.71718 (3)
Münstertal 29aMAMUNS 7PigMand M226.19 (7)−12.22 (2)0.71524 (1)
Münstertal 29aMAMUNS 8PigMax P424.9 (7)−12.72 (4)0.71539 (1)
Münstertal 29aMAMUNS 9PigMand M326.38 (7)−13.68 (2)0.71310 (1)
PoltringenLBKPOLT 1PigMand L M325.57 (2)−12.78 (2)0.71028 (1)
SasbachE. Lat.SASB 10PigMax R M223.32 (3)−13.70 (2)0.71283 (4)
SasbachE. Lat.SASB 2PigMand P423.54 (9)−14.80 (4)0.70925 (1)
SasbachE. Lat.SASB 8PigMand P423.53 (4)−14.40 (1)0.70992 (1)
SchnellertsMASCHN 1PigMax L P326.36 (3)−12.66 (2)0.71114 (1)
SchnellertsMASCHN 2PigMax L M325.1 (4)−13.30 (2)0.71023 (1)
SchwanfeldLBKSCHW 10PigMax M224.24 (2)−12.35 (2)0.70968 (1)
SchwanfeldLBKSCHW 7PigMax M223.58 (5)−12.67 (3)0.70949 (1)
SchwanfeldLBKSCHW 8PigMax L M224.62 (4)−12.75 (3)0.70944 (1)
Singen/OffwieseGroßch.SIOF 2PigMand L M323.38 (4)−13.71 (1)0.70724 (1)
Singen/OffwieseGroßch.SIOF 3PigMand L M222.76 (4)−13.05 (2)n.d.
UrspringMA/NZURSP 2PigMand C R26.83 (3)−12.15 (2)n.d.
UrspringMA/NZURSP 3PigMand L P226.47 (5)−13.80 (2)n.d.
WangLBKWANG 1PigMolarn.d.n.d.0.71033 (2)
WangLBKWANG 2PigMand M324.42 (2)−12.94 (3)0.70964 (1)
WangLBKWANG 3PigMand M223.94 (4)−13.20 (3)0.71046 (1)
WarbergMAWARB 1PigMand L M225.78 (5)−12.70 (3)0.71434 (2)
WarbergMAWARB 2PigMand L M224.31 (7)−13.40 (4)0.71948 (1)
WarbergMAWARB 4PigMand R M324.75 (2)−12.57 (2)0.72276 (1)
WieladingenMAWEIL 1PigMand P423.57 (6)−13.82 (2)0.71871 (1)
WieladingenMAWIEL3PigMand R P424.8 (5)−12.67 (3)n.d.
WildsteinMAWILD 1PigMand M223.49 (2)−12.51 (3)0.72217 (1)
WildsteinMAWILD 2PigMand R C22.74 (1)−11.48 (3)n.d.
WildsteinMAWILD 4PigMand P425.29 (2)−12.99 (3)0.72203 (1)
WittislingenLBKWITT 1PigMand M3n.d.n.d.0.70860 (1)

The Jurassic limestones of the Swabian Alb are represented by the Iron Age settlement of the Heidengraben near Grabenstetten and the medieval village of Urspring. From the Hegau, a small area west of the Lake Constance (Bodensee) with abundant volcanic outcrops and glacial moraines, we have samples from Hilzingen (LBK; Fritsch 1998) and Singen-Offwiese (middle Neolithic).


The results for 87Sr/86Sinline image, δ18O and δ13C in pig enamel are listed individually in Table 1 and averaged in Table 2. The mean 87Sr/86Sr results from each site are also shown on the geological map (Fig. 1). Table 2 also lists the mean δ18Oc values for each site converted via equation (1) to predict δ18Omw in precipitation, and the calculated δ18Omw values shown geographically in Figure 2. In short, the upland–lowland difference is clearest with strontium and carbon isotopes (Fig. 3 (a)). We now discuss the results in context with what is known from the geology.

Table 2. Average isotope values in pigs’ teeth (unless otherwise indicated) for each site. Oxygen isotope values are given as measured in the structural carbonate (δ18Oc) and as calculated via equation (1) for meteoric water (δ18Omw). The standard deviation of each mean (of n values), given as the last digits for the87Sr/86Sr values, is shown in parentheses. Each asterisk (δ) represents one outlier omitted from the calculation of the means
AreaNAverage 87Sr/86SrAverage δ18Oc (, SMOW)Average δ18Omw (, SMOW)Average δ13Cc (, PBB)
 Rhine Valley 0.7090 (7)***24.1 (0.9)***11.0 (1.1)***13.9 (0.8)***
Goddelau30.7091 (6)*23.9 (0.8)*−11.2 (0.9)*−13.5 (0.3)*
Sasbach30.7096 (5)*23.5 (0.1)*−11.7 (0.1)*−14.6 (0.3)*
Bischoffingen (Kaiserstuhl)30.7083 (2)*24.9 (1.3)*−10.1 (1.5)*−13.5 (1.3)*
 Main area 0.7094 (3)*24.5 (0.9)*−10.5 (1.0)*−12.8 (0.4)*
Bruchenbrücken20.7091 ()*25.7 ()* −9.1 ()*−13.3 ()*
Schwanfeld30.7095 (1)24.1 (0.5)−11.0 (0.6)−12.6 (0.2)
 Neckar Valley 0.7098 (3)24.7 (1.0)*−10.3 (1.2)*−13.8 (0.8)*
Ilsfeld30.7098 (2)24.3 (1.1)*−10.8 (1.2)*−14.2 (0.1)*
Poltringen10.7103 ()25.6 () −9.3 ()−12.8 ()
Vaihingena100.7095 (2)n.d.n.d.n.d.
 Pre-Alpine lowlands 0.7097 (8)23.1 (1.9)*−12.2 (2.2)*−13.4 (0.6)*
Altdorf-Aich10.7098 ()n.d.n.d.n.d.
Inzigkofen10.7103 ()20.9 ()−14.8 ()−14.0 ()
Mintraching10.7086 ()n.d.n.d.n.d.
Wang30.7101 (4)24.2 (0.3)−10.9 (0.4)−13.1 (0.2)
Wittislingen10.7086 ()28.2 ()* −6.3 ()*−13.9 ()*
Dillingen (modern snails)b50.7084 (3)n.d.n.d.n.d.
 Black Forest 0.7163 (21)24.8 (1.2)−10.2 (1.4)−12.9 (0.6)
Münstertal (both sites)40.7152 (17)25.5 (0.9) −9.4 (1.0)−12.8 (0.6)
Wieladingen10.7187 ()24.2 (0.9)−10.9 (1.0)−13.2 (0.8)
 Odenwald 0.7107 (6)25.7 (0.9) −9.1 (1.0)−13.0 (0.5)
Schnellerts20.7107 (7)25.7 (0.9) −9.1 (1.0)−13.0 (0.5)
Various sites (modern snails)b70.7090 (7)n.d.n.d.n.d.
 Bavarian Forest 0.7202 (35)24.4 (1.1)−10.7 (1.3)−12.6 (0.6)
Warberg30.7189 (42)24.9 (0.8)−10.1 (0.9)−12.9 (0.4)
Wildstein30.7221 (1)23.8 (1.3)−11.3 (1.5)−12.3 (0.8)
Hegau 0.7083 (27)24.2 (1.4)−11.0 (1.6)−12.9 (0.6)
Hilzingen20.7089 (36)25.2 (0.9) −9.7 (1.0)−12.5 (0.1)
Singen/Offwiese20.7072 ()23.1 (0.4)−12.2 (0.5)−13.4 (0.5)
 Swabian Alb 0.7097 ()25.2 (1.8) −9.8 (2.1)−13.3 (0.9)
Heidengraben20.7097 (1)23.7 (0.8)−11.6 (0.9)−13.6 (0.8)
Urspring2n.d.26.7 (0.3) −8.1 (0.3)−13.0 (1.2)
Figure 3.

Isotopes in Neolithic pig enamel plotted by site region. Solid symbols are for samples from crystalline mountain sites, grey symbols are for calcareous upland sites, and open symbols are for lowland sites. C and O isotopes are shown as measured in structural carbonate, with δ13Cc in versus PDB, and δ18Oc in versus SMOW.

87Sr/86Sr in the lowlands

Our sample sites in the Neckar area include Ilsfeld and Poltringen, with a mean 87Sr/86Sr value of 0.7098 ± 0.0003 (n= 5), indistinguishable from the previously determined average in Neolithic pig teeth at Vaihingen of 0.7095 ± 0.0002 (Bentley et al. 2004). Values from the Main River area are similar, with 87Sr/86Sr averaging 0.7094 ± 0.0003 from Bruchenbrücken and Schwanfeld, omitting one outlier value (0.7121—a traded pig?) from Bruchenbrücken. As these sites are all situated on loess, these averages may reflect the windblown sediment in the area.

The pre-Alpine lowlands contain shallow Oligocene/Miocene marine sediments overlain by freshwater sediments, glacial moraines and loess deposits. Pig teeth from Altdorf-Aich, Inzigkofen, Mintraching, Wang and Wittislingen yielded 87Sr/86Sr values (n= 8) with a narrow distribution of 0.7097 ± 0.0008 that is on the upper end of the range (0.708–0.710) calculated for mobile 87Sr/86Sr in studies of Neolithic human bones (Grupe et al. 1997; Horn and Müller-Sohnius 1999). The 87Sr/86Sr values for Mintraching and Wittislingen (both 0.7086) are close to the average in modern snails (0.7084 ± 0.0003) from the Danube site of Dillingen (Price et al. 2002).

87Sr/86Sr values recovered from pigs’ teeth in the Upper Rhine Valley (n= 6, three outliers omitted) were generally similar to the other lowlands, with a mean of 0.7090 ± 0.0007, but somewhat more variable. This is close to the expectation for the Triassic/Jurassic sediments from the Northern Calcareous Alps, where 87Sr/86Sr in modern groundwater and stream water ranges between 0.7085 and 0.7095 (Tricca et al. 1999; Eikenberg et al. 2001). Two of the three values at Goddelau (Table 1) in the northern Upper Rhine Valley compare well with Neolithic human bones from Flomborn and Schwetzingen, averaging 0.7100 ± 0.0002 and 0.7094 ± 0.0004, respectively (Price et al. 2001; Bentley et al. 2004).

The teeth from Sasbach and Bischoffingen average 0.7090 ± 0.0008 (n= 4, two outliers omitted). This is consistent with stream water values from the southern Upper Rhine Valley (Tricca et al. 1999), and contrasts with the whole-rock 87Sr/86Sr of the Tertiary basalts of the nearby Kaiserstuhl (0.7034–0.7507; Calvez and Lippolt 1980).

The two outlier values omitted from calculating the mean 87Sr/86Sr in the Rhine Valley include one at Goddelau (0.7157) and one at Sasbach (0.7128) that are considerably higher than the rest of the values from the Upper Rhine Valley. As Table 1 shows, the δ18O for the Goddelau pig (GODD 20) and the δ13C of the Sasbach pig (SASB 10) are also anomalous for the respective sites, suggesting that these pigs were traded in from elsewhere. At Sasbach, the high value could come from 20 km eastwards, where granites and gneisses begin to outcrop in the foothills (Fig. 1).

In summary, having omitted a few outliers from possibly traded pigs, the loess and alluvial deposits of the Rhine, Main, Neckar and pre-Alpine lowlands exhibit biologically available 87Sr/86Sr values between 0.7086 and 0.7103.

87Sr/86Sr in the uplands

The Black Forest and the Vosges mountains of France are both part of the same horst formation of Palaeozoic granites, granodiorites and metamorphic rocks that are about 300–400 million years old. We sampled archaeological pigs’ teeth from the medieval sites of Münstertal and Wieladingen in the Black Forest, which exhibit the expected high 87Sr/86Sr, ranging between 0.713 and 0.719 and averaging 0.7163 ± 0.0021. These values overlap with respective water values from the Black Forest and Vosges, which range from 0.714 to 0.725 (Horn et al. 1994; Tricca et al. 1999; Aubert et al. 2002).

Although similar granites underlying the Black Forest and Vosges also form the Odenwald, the two 87Sr/86Sr values from Schnellerts are 0.7102 and 0.7111, consistent with values in modern snails from the Odenwald (0.7090 ± 0.0032; Bentley et al. 2003a, table 1).

Our sites from the Bavarian Forest include Warberg and Wildstein, which yielded the highest 87Sr/86Sr values, averaging 0.7202 ± 0.0035 (n= 6). This probably reflects weathering of granites and gneisses north of the River Danube (Grupe et al. 1997).

The two lowest 87Sr/86Sr values in our study occurred at Hilzingen (0.70633) and Singen-Offwiese (0.70724) in the Hegau, west of Lake Constance. This may reflect a mixture of volcanic bedrock, with whole-rock 87Sr/86Sr between 0.7038 and 0.7046 (Calvez and Lippolt 1980), and glacial moraines with 87Sr/86Sr between 0.709 and 0.710.

Our two archaeological pig teeth samples from the Iron Age site of Heidengraben, in the Swabian Jura, had 87Sr/86Sr values of 0.7097 and 0.7098, which is higher than expected for these Jurassic limestones, which ought to range from about 0.707 to 0.7086 (Veizer et al. 1999; Tütken 2003). As opposed to reflecting the limestones themselves, these values might reflect the loam (an accumulation of insoluble material from former weathered calcareous layers) overlying the limestones.

Patterns in δ18O and δ13C

In addition to having generally higher 87Sr/86Sr values, upland pigs also show about 1–2 less negative δ13Cc (Fig. 3 (a)), with relatively low variance in the upland δ13Cc values. The most likely explanation is that our Neolithic samples from the lowlands, where settlements on small clearings existed within a largely forested landscape (e.g., Lüning 2000; Strien 2000, fig. 1.3), reflect a greater forest canopy effect (i.e., more negative δ13C) than the medieval sites in the uplands, which were intensively settled and used in the Middle Ages (i.e., Rösch 2000). Another possibility is that the vegetation consists of high-altitude plants (Körner et al. 1991). Among the less likely explanations, there is no obvious reason why upland pigs should have eaten more meat, as δ13C increases by about one per mil per trophic level (e.g., DeNiro and Epstein 1978; Post 2002); nor is it likely that the uplands were ever warmer, drier or sunnier, which also correlates with increased 13C.

Counter to the expectation for δ18Omw to decrease with altitude, the upland–lowland difference is not apparent in the δ18Oc results. This is not so surprising, however, since the monthly variation in δ18Omw in precipitation leads us to expect variability between pigs that grew up in different years, as well between molars of the same pig, which have different eruption ages (e.g., Gadbury et al. 2000; Balasse et al. 2002). To explore this, we analysed several M1 molars (0–3 months of age), which produced some of the highest δ18Oc values in our study (Sasbach, 28.1; Heidengraben, 27.2). The relative 18O enrichment in the M1 samples may be partly an effect of suckling, as δ18O values in first molars of humans average 0.7% higher than third molars (Wright and Schwartz 1998), but the more extreme differences are probably seasonal effects. For pigs born in the spring, M1 will have formed during the summer, when δ18Omw in rain may have averaged about 4 higher (cf., Schürch et al. 2003, fig. 6 (a)). Figure 4 shows δ18Oc and δ13Cc differences between different molars of the same pig, showing why we analysed M3 whenever possible and have not otherwise reported results from M1.

Figure 4.

δ13Cc and δ18Oc in the pig enamel of different molars from the same pigs.

Somewhat surprisingly given all this variability, we actually find a modest correlation between δ18Oc in modern precipitation and δ18Omw calculated from the pigs’ teeth of the Iron Age and later. When we plot these converted values versus those inferred from the 1991 map (Fig. 5 (a)), the best-fit regression (r2= 0.58) through the points has a slope greater than one (about 2.7 as shown, or about 1.5 if the point at the bottom left is removed). Although these are only a few data, this positive slope could indicate that the contours of δ18O in southern Germany were closer together in the first millennium ad than they are now. However, we have no guarantee that the spatial pattern of δ18Omw was similar in the past—notably, there is absolutely no correlation (r2 < 0.01) between δ18Omw calculated from the Neolithic pigs and the 1991 precipitation (Fig. 5 (b)). As discussed above, it would be ideal to create separate contour maps of δ18O for different time slices of the past, but that would require many times more data than we have presented here.

Figure 5.

A comparison of the values predicted versus values inferred from the contour map, for (a) sites from the Iron Age or later, and (b) Neolithic sites. The error bars are r.m.s. 1σ for the mean at each site (no bar for single-sample sites).

In conclusion, although our motivation for this project has been to help interpret previous 87Sr/86Sr analyses of Neolithic humans and domestic animals (Price et al. 2001, 2003; Bentley et al. 2002, 2003b, 2004), we hope our contribution will also be useful for any other researchers measuring biologically available strontium, oxygen or carbon isotopes from prehistoric Germany.


The archaeological samples for this study were generously provided by Professor Dr H.-P. Uerpmann (University of Tübingen), Dr E. Stephan and Dr P. Schmidt-Thomé (State Heritage Department of Baden-Württemberg), Dr N. Harre (Groß-Gerau) and H. Schaller (Pfreimd). Their help and assistance is gratefully acknowledged. For a student project, Martina McArthur prepared most of the samples for oxygen and carbon isotope analysis, for which we also thank Professor Tim Atkinson (Geology, University College London) for facilitating our use of the GC–MS in the UCL Bloomsbury Environmental Isotope Facility. We also thank Dr Rex Taylor, Dr Matthew Cooper and especially Tina Hayes of the School of Ocean and Earth Science, Southampton Oceanography Centre, where the TIMS analyses were performed for this study.