Ultrafiltration of bone collagen, dissolved as gelatin (M ~100,000 D), has received considerable attention as a means to remove small contaminants and thus produce more reliable dates (Brown et al. 1988; Bronk Ramsey et al. 2004; Higham et al. 2006; Mellars 2006). However, comparative dating studies have raised the question whether this cleaning step itself may introduce contamination with carbon from the filters used (Bronk Ramsey et al. 2004; Brock et al. 2007; Hüls et al. 2007).
Here, we present results of further ultrafiltration experiments with modern and fossil collagen samples using Vivaspin 20™ and Vivaspin 15R™ ultrafilters. Evidently, the Vivaspin 20 (VS 20) ultrafilter with a polyethersulfone (PES) membrane retains more material in the >30 kD fraction than the Vivaspin 15R (VS 15R) filter with a regenerated cellulose membrane (Hydrosat), which may be related to increased retention of proteins due to suboptimal electrostatic conditions during ultrafiltration with the PES membrane. In addition, this filter type shows clear evidence for contamination with fossil carbon, presumably from membrane fibers, in the <30 kD fraction. Radiocarbon measurements on ultrafiltrated fossil collagen seem to indicate small contributions of modern carbon via glycerin left on and within the filter membranes of both types. Although SEM pictures show film remnants on the fibrous filter structure of cleaned filter membranes, EDX analysis on the VS 20 membrane to not support the assumption this may be glycerin. Our observations indicate the risks and benefits of the use of ultrafiltration in cleaning collagen samples for 14C dating need to be further quantified, especially for the cleaning of fossil bone collagen of good quality samples.