-33- 



The "apparently best" and "aioparently poorest" solutions 

 ■shown in tables II and III are clearly the ones that did 

 give, individually and empirically, respectively h^%iyer^ and 

 3-ower crowth rates than the averages in the several series. 

 One group of solutions (tablell) v^ere poor, by actual test. 

 Had the^^ v.'hole series of 126 solutions been selected at 

 random, without reference to the physico-chemical scheme of 

 the triangular diagram, then the list of good solutions v/ould 

 have to be taken at its face value, as showing v/hich solutions 

 had been found best by test. B\it the solutions of this study 

 were not selected at random, they represent a certain definite 

 series of different sets of salt proportions and different 

 salts, '.'.'ithin the limits set by the chosen total concentration 

 and by the nine salts used, the series is so selected as to be 

 equally distributed throughout the entire range of possibilities. 

 They are somewhat like a set of soil samples secured one from 

 each of a number of stations frequently and equally spaced 

 over a broad terrain comprising many kinds of soil. This being 

 the case, it follows that evidence for any significant influence 

 exerted by the makeup of any given solution should be shown not 

 only by that particular solution itself btit also by the solutions 

 adjacent to it on the triangular diagram. A study of the salt 

 proportions of the "apparently best" solutions and of the growth 

 rates given by the adjacent solutions fails generally, in the 

 present study, to bring forth any evidence that one set of salt 



