More than just a convoluted table? Discussion of 'Mediterranean tephra stratigraphy revisited: Results from a long terrestrial sequence on Lesvos Island, Greece' by Margari et al. [J. Volcanol. Geotherm. Res. 163 (2007), 34–54]

Margari et al. (2007) recently published a range of major and trace element geochemical data (EPMA, bulk tephra solution ICP–MS) as part of a study of a series of tephras from a lake sediment core on Lesvos Island, Greece. In their paper they propose correlations for some of their deposits with more widespread tephra deposits in the Mediterranean. However, some issues stemming from their sample preparation methods, a confused table of data, an incorrect statement about the incomparability of analytical methods, and a possible miscorrelation in the paper by Margari et al. led us to the decision to write this comment.

In their discussion of trace element analysis and correlation of tephras from their core, Margari et al. (2007) state that published tephra data obtained by neutron activation analysis cannot be directly compared with their ICP–MS analyses (bottom of p. 46). This is a somewhat traditional misconception i.e. that concentration data produced by different methods are not comparable. It is true however that, depending on the elements' concentrations, different methods produce data of differing quality, largely as a consequence of different detection limits. Independent of the method applied however, a value for an element's concentration has to be valid (within the given analytical error for that technique) and should thus be comparable with data acquired by other methods, provided standard analytical protocols are followed. Several papers specifically compare tephra compositional data obtained by different analytical methods and show the validity of this approach (e.g. Westgate et al., 1994, Pearce et al., 2002, Pearce et al., 2004 and Steinhauser et al., 2006).

To check the alleged incomparability of the ICP–MS data from Margari et al. with neutron activation analysis (NAA) from other studies, we decided to plot the tephra data from Margari et al. against data from our extensive data base of Mediterranean tephras (obtained by NAA, see Peltz et al., 1999 and Steinhauser et al., 2006). After overcoming several problems with the data presented by Margari et al. (discussed later), we found a perfect agreement between samples ML-3 and ML-4 from Margari et al. and the chemical fingerprint of the Giali main eruption (“Yali D” of Bond, 1976) from our database. This comparison is presented in Fig. 1, where the concentrations of elements present in both data sets were normalised to mean upper crust (from Taylor and McLennan, 1985), and plotted according to their atomic number. The data from Margari et al. show high concentrations of Cr, Fe and Co, which can be explained as contamination of their sample by sediment derived from the local, peridotitic basement, or even of the coring tools, but all other elements show a near perfect agreement with the Giali lower pumice (i.e. the Giali main eruption). Had Margari et al. compared these data sets, they would have undoubtedly observed this agreement immediately.