Reproducibility— The reproducibility (R) of this test has been established as 34.5 %. Two single measurements (determinations) that differ by more than the tabulated (R) (expressed as percentage of their mean value) must be considered suspect, that is, having arisen from different sample populations. Such a decision dictates that appropriate action be taken.