better than others, but a comparison of the relative rate 

 values for the two tests brings out the fact that some 

 solutions showing mean rates above the average in one test 

 showed rates below it in the other test. The solutions that 

 gave mean rates more than 2 per cent above the average rate 

 are marked v^ith an X at the extreme right of table I, and 

 only three of them (marked with a double X) are thus designated 

 for both tests. These three sol^itlons (R3S2, R3S2'/ and R5S2) 

 might be considered as definitely better than the others of 

 this type and temperatiore, btit there is no app &rent- relation 

 between this apparent "goodness" of the solutions and the 

 corresponding salt proportions. T\,o of these ^^are in rov; 3 

 and the other is in row 5 of the triangular diagram, so that 

 they are not adjacent as to the potassi^um salt. Tv/o are 

 second in the row, the other being third, so that the former 

 two have tv7o-eighths of their salt molec\ales in the form of 

 the calcivim salt, and the latter has three-eighths in this 

 form. The intervening solution with two-eighths of its salt 

 molecules in the form of the calcium salt (R4G2) gave a value 

 somewhat below the relative value 1.00 in both tests. Finally, 

 one of these exceptionally "good" solutions has three-eighths 

 of its total niomber of salt molecules in the form of the 

 magnesium salt, another has tv/o-eighths. and the third has but 

 one-eighth in that form. It therefore seems that the except- 

 ional "goodness" of these three solutions is not clearly 

 T^elated to the salt proportions employed. 



