The CV at the detection limit.

lent.” However, this was true only if the results from the determination of diluted (range 75-680 zg/L) and undiluted samples (range 0-75 tgfL) were pooled. Using the data in Table 2 of Hall et al. (1), we can compare the regression lines for the concentration ranges 0-75, 75-680, and 0-680 ugfL (Figure 1). It is obvious that there are discrepancies between the two methodswithin the range 0-75 g/L, which is an important range when small AM! is to be diagnosed. The authors state that there was a good linearity for the Hybritech method within the range 275 ugIL, and probably this method was most appropriate within this range. This accords with our unpublished results. According to Hybritech (personal communication), 1 UIL equals 1 1zg/L, but no conversion factor is mentioned in the article (1). Using this conversion factor, we see that the precision of the two methods [Hall et al., Table 1 (1)] was determined at different concentration ranges. Precision, both betweenrun and within-run, was determined near the reference limit (12.3 and 13.8 pg/L) for the Tandem-E method, but far from this limit (85 and 87 UIL) in the electrophoresis method. No other reference to the accuracy of the electrophoresis method is mentioned in the article. Their Table 1 indicates that the electrophoresis method is less precise for values near the reference limit, but no such comparison was done. We agree with Hall et al. (1), that the Tandem-E method provides useful data for the diagnosis of AM!, but the design of their study favored the electrophoresis method, especially in cases of small AM!s. References