J. Henderson Smith 



47 



of phenol), then when Rjlog T = 50, we obtain the following table of 

 survivors at the times stated (Table XXI), and the survivor curve 

 shown in Fig. 10. This shows no sigmoid character and approaches the 

 logarithmic type. It is not strictly logarithmic, as is evident from the 

 values of k given in the table; but if we had taken as our basis a true 



100 r 



IMimites 2 



Fig. 10. Survivor curve obtained from the frequency shown in Fig. 8 by allowing 

 for the change of rate with increased phenol. 



normal curve, instead of the frequency obtained from the experimental 

 observations, a much better approach to a true logarithmic curve would 

 have been obtained. A strictly logarithmic survivor curve can result 

 only if the frequency curve is shghtly skew ; but a survivor curve which 

 in experiment closely simulates a logarithmic curve can be got from a 

 strictly normal frequency. 



Table XXI. 



