3. Results

Table 1 and Figure 3 show the stable isotope data for food crusts from the four sites. The two isotopes are apparently correlated, with high δ¹⁵N corresponding to more depleted δ¹³C values, consistent with our expectation that freshwater fish was one of the ingredients cooked regularly in these pots. The results must be interpreted with caution, however. Food crusts may represent the remains of a single meal, but still contain multiple ingredients. Experimental cooking suggests that some starchy plant food may be required to produce a thick layer of burnt food. In this case the ¹⁴C, δ¹³C and C/N values may primarily reflect those in a plant ingredient, while δ¹⁵N may be determined mainly by a fish ingredient. Our C/N values are typically 8–16, too high for most animal products and too low for most plants, suggesting a mixture of ingredients. Cooking, burning and diagenesis might also affect the concentrations of different components (protein, carbohydrate, fat, fibre), which could alter the stable isotope and C/N values – although in practice these effects seem to be very modest (Philippsen 2013b; Fernandes et al. 2014). Carbon in fat is depleted in δ¹³C compared to that in carbohydrate or protein, so high fat content should shift food-crust δ¹³C towards more aquatic values, but while it would not affect δ¹⁵N, it would dramatically raise the C/N value, and we see no evidence of this in our results. So we are confident that we can interpret δ¹³C values below c. -27‰ as evidence of freshwater fish, particularly if δ¹⁵N is above 6‰ (low trophic level aquatic foods (e.g. mussels) would be depleted in δ¹⁵N and δ¹³C).

Note 4: The three Funnel-Beaker sherds from Bebensee that could not be re-sampled for stable isotopes gave AMS δ¹³C values of -30 to -32‰, and ¹⁴C ages of 5800–6100BP, similar to those of the most depleted 'unknown' samples in Figure 3. We therefore cannot be certain that any of the Bebensee food crusts are associated with Ertebølle pottery.

The most terrestrial isotopic values (δ¹³C c. -24 to -26‰, δ¹⁵N under 6‰) are all associated with early Neolithic Funnel-Beaker pottery, and with the lowest ¹⁴C ages. It is of course plausible that terrestrial foods became more important during the early Neolithic, but the differences in ¹⁴C ages between Funnel-Beaker food crusts with aquatic and terrestrial stable isotope signatures could also be explained by the scale of ¹⁴C reservoir effects in the Trave and Alster [Note 4]. Aside from the date range of the assemblages concerned, variability in freshwater reservoir effects between individual aquatic organisms from the Trave and Alster (Philippsen and Heinemeier 2013) may explain why there is not a clearer correlation between stable isotopes and ¹⁴C ages (Figure 3): unlike isotope data from human bone, food-crust isotope signatures may reflect single cooking events, and should therefore capture more of the isotopic variability in an ecosystem.

Note 5: 95% confidence intervals for the calibrated dates were calculated using OxCal v4 (Bronk Ramsey 2009) and the IntCal13 calibration curve (Reimer et al. 2013).

Schlamersdorf LA 5 provides the best opportunity to investigate the use of pottery for cooking fish in the late Mesolithic. Terrestrial ¹⁴C samples span most of the Ertebølle period (Hartz 2011; Philippsen 2013b), from 5620–5300 cal BC (AAR-11399, 6480±90BP) to 4280–4030 cal BC (KIA-3024, 5320±65BP) [Note 5]. Fully aquatic samples (two fish-bones and one mollusc) have conventional ¹⁴C ages greater than 7500 BP. One sherd contains external sooting from the cooking fire, which would date the pot towards the end of the Ertebølle period (AAR-11481S, 5190±110BP). The food crust from the same sherd is 1660±160 ¹⁴C years older, despite its relatively terrestrial stable isotope values (δ¹³C=-28.0 and δ¹⁵N =3.4). Even if most of the carbon in this food crust is from aquatic sources, with reservoir effects of this magnitude, the Schlamersdorf fish-bone and mollusc samples would not be older than the fully terrestrial samples, and it would only require a 20–60% carbon contribution from aquatic sources (i.e. a reservoir effect of 330–1000 years, given the results for AAR-11481) for all six food crusts from this site to date to after c. 4500 cal BC (c. 5700 BP). The fish-bones and mollusc samples dated would, however, belong to the early-middle Ertebølle period, which is also represented by several charcoal and mammal bone radiocarbon samples.