Abstract
This paper theoretically derives a general rule that while the slope of the semi-logarithmic plot (Y vs. log X) of a calibration curve varies depending on analyte concentration, X, the slope takes a specific value at the detection limit (LD). This rule holds good irrespective of the shape of the calibration curve (linear or non-linear) and in this paper, is applied to competitive ELISA (enzyme linked immunosorbent assay). The following relationship is deduced: slope of log-dose B/B0 at LD = [relative standard deviation (RSD) of blank responses] ÷ 0.13. The LD obtained from the above-mentioned slope corresponds to the dose at which the RSD of dose estimates is 0.3 (= 30%). A commercial kit for 17α-hydroxyprogesterone is taken as an example.
