558 WILLIAM S. STURGES AND LEO F. RETTGER 



of variation could never be sufficiently reduced, however, to 

 place much reUance on a single determination. At least two 

 determinations were always made, and when these gave results 

 differing from each other bj^ more than 10 per cent the process 

 was repeated until satisfactory checks were obtained. The 

 difficulties encountered were in all probability due to the presence 

 of bacterial cells and other suspended matter in the test material. 

 It soon became apparent that the number of oscillations had to 

 be regulated (240 per minute) and that more than five minutes 

 was required to obtain a complete reaction. Ten minutes was 

 finally adopted as the routine time. This was followed by an 

 additional shaking by hand for one or two minutes while the 

 motor was being used to shake the Hempel pipette. The addi- 

 tional gas Uberated at this time was also analyzed and its volume 

 of nitrogen added to the first detennination. This gave a means 

 of judging whether the reaction had been completed in the time 

 allowed. Because of the extra tim.e and labor required to ob- 

 tain rehable results, this method of determining the amino 

 nitrogen proved less practical than that of Sorensen. The 

 original method of Van Slyke (1911, 1912) was adliered to as 

 closely as possible, and the calculations of amino nitrogen made 

 by reference to the tables. 



EXPERIMENTAL DATA 



Early in the investigation the course of autolysis was followed 

 \vith the biuret and Sorensen tests. The results appeared to 

 indicate a direct relationship between proteolytic acti\'ity and 

 gelatin-liquefying power of bacteria, on the one hand, and autol- 

 ysis on the other. Thus, B. subtilis suspensions underwent 

 complete autolysis witliin two to three days, while Bad. coli 

 showed no changes. Ps. pyocyanea, while slower to undergo 

 self digestion than B. subtilis, gave biuret figures wliich ap- 

 proached 0. With both of these organisms the Sorensen figures 

 rapidly increased. In the study of Bad. coli material no changes 

 could be observed in the biuret or Sorensen values (see chart 1). 



Following these earUer observations, the plan of study was 

 considerably enlarged, and efforts were made to acquire as 



