Determining prehistoric earthquake periodicity and magnitudes is important for risk assessments in seismically active areas. We evaluate a geochemical method, which has previously been used to identify prehistoric slips on normal fault scarps through an analysis of variations in the concentration of rare earth elements and Y (REE-Y) along vertical transects. Our study object is the Sparta Fault, a normal fault in southern Greece, developed in limestone and previously documented, and dated using ³⁶Cl, to have been last active 464 BC. From geochemical analyses of 39 fault rock samples, we conclude that REE-Y concentrations correlate strongly with the abundance of quartz and possibly other heterogeneities in the fault scarp. Because the sampled fault rock is a protocataclasite, formed at depth, variations in the abundance of quartz are not associated with prehistoric movements along the fault. We therefore conclude that geochemical evidence does not provide a reliable paleoseismic proxy for fault movement. We also present data indicating a co-variation between quartz and ³⁶Cl concentrations, which we suggest requires a re-examination of this widely used application of the cosmogenic nuclide surface exposure dating method.