produccs a curved segment a t lower pH values.8 In calculating the purity and ionization constant, only the linear portion of each plot was used. Nevertheless, it may be seen (Table 11) that the intercept obtained (and therefore the purity evaluated) is not markedly influenced by the choice of activity coeficient. It is rather the ionization constant that reflects the diffcrcncc, as may be seen by the two slopes in Fig. 1. A plot of pKa' uerstds d i d e m o nstrated an essentially linear relationship for both sets of data. These were extrapolated to infinite 433 3 In this area of the titration hydrogen ion makes its greatest contribution to Z' value's. so that an error in the activity coefficient would be most evident at these higher acidities. This is comparable to Benet and Goyan's type A curve in Fig. 1 (2) for the case where high erroneous pH values are substituted. In all cases the Kielland values are higher than the yf, and would result in a lower hydrogen ion concentration.
dilution to evaluate a thermodynamic ionization constant. The rcsults (Table 11) indicate that the pKa obtained with both values are consistent with the litcrature (5-7). Therefore, since the two sets of data are the same, the type of activity coefficient employed for calculations appears to be a mattcr of personal choice.