Comments about the proposal to adapt the WASP programme to fibre counting tests.

(i) From figure 4 (Grzebyk et al., 2005), the French method (Method 1) of appraisal of the quality of the results of analysts involved in fibre counting by PCOM is, at least theoretically, quite a lot more stringent than the other two methods discussed. However, in practice in the field (table 2, Grzebyk et al., 2005), the French method is a little bit less demanding (‘selective’) than the RICE method (Method 2). One can thus conclude the CV criterion (CV < 0.4), characterizing the French method, does not play any significant role in the analysts’ world. (ii) The proposed adaptation of the WASP method gives results close to those obtained by traditional methods only if the reference relative variance a = 0.14–0.18 [a = (CV), where CV is the inter-analyst coefficient of variation. Note that this CV is not the coefficient of variation of the normalized results of a given laboratory, checked by the French method]. In their paper, the authors refer to the inter-experienced laboratory coefficient of variation (0.45) given in the NIOSH 7400 Method (NIOSH, 1994). However, one can expect the laboratories to have improved their performances since the publication of the NIOSH 7400 Method, 10 years ago, and that the inter-laboratory CV (including the variability of the possibly required replicas) has been reduced. Jones et al. (2005) mention a geometric standard deviation (GSD) around 1.2, confirming data (figure 4) of Arroyo and Rojo (2001) for the laboratories taking part to the Spanish scheme, corresponding to a = 0.034. According to figure 5 (Grzebyk et al., 2005), with this value of a in the adapted WASP method, the number of counters classified in category 3 (meaning ‘not passing the test’) would more than quintuple in comparison with the present levels and the adapted WASP method will turn out as a lot more ‘selective’ (demanding) than the traditional methods presently used.