The value 0.6% was subtracted from the slopes observed for all samples of I to correct for the solubility-product effect, even though that exact value applied only to pure samples. However, this is justified since most samples are >97% pure, and in addition the impurities consumed essentially the same number of moles of acid as the pure compound. This is shown by titration of the sample with perchloric acid; the equivalent weight of impure samples is the same as the pure compound. It should be noted that any hydrochloride salt present in the sample will not show up as an impurity. Solubility analysis with methanolic HC1 as solvent is useful for measuring the purity of samples of I-HC1. The solubilityproduct correction, of course, is not used in this case.In Figure 3 are shown the analyses obtained on pure I, the impure sample shown in Figure 2, and a synthetic mixture of I and II containing 9.5 weight per cent of II. This mixture gives a slope of 12.7% after correction for the solubilityproduct effect. This is due to the fact, mentioned above, that the II forms a dihydrochloride salt in the system and hence the observed weight of II is increased. When the slope is corrected for the ratio of molecular weights 216/289, the predicted value of 9.5 % is obtained.The molecular weight correction was not applied in analyses of routine samples because the amounts of each impurity were not known. The molecular weight correction was negligible for essentially pure samples such as those usually assayed. Uncorrected slopes of several per cent might be several tenths of a per cent high.