DISINFECTION STUDIES 211 



We find in these experiments striking examples of that pre- 

 liminary period in the disinfection process which we have referred 

 to as the period of induction. The bacteria do not begin to die 

 off at the maximum rate, but the mortahty increases to the 

 maximum gradually, depending upon the pH and the tem- 

 perature. The lower the temperature and the less extreme the 

 acidity, the greater is the duration of this period. We have 

 succeeded in this case, apparently, in magnifying that early phase 

 in disinfection which ordinarily is so small as to escape observa- 

 tion. Its significance will be discussed presently. 



DISCUSSION 



The probable error involved in the method of experimentation 

 leaves much to be desired. Under the most nearly ideal condi- 

 tions some of the results involve a probable error of only 2 to 3 

 per cent of the mean values indicated. In others, the probable 

 error amounts to 10 per cent of the mean or more. When the 

 bacterial numbers per unit volume are low, then the probable 

 error increases greatly. These considerations led to the choice 

 of an initial number of one million per cubic centimeter as involv- 

 ing the least manipulative error. 



It might perhaps be well to express here a word of caution 

 in regard to a too ready use of the calculated velocity coefficient, 

 k, in the comparison of disinfections under varying conditions. 

 This mode of expressing differences in disinfection has many 

 advantages, provided its use implies no unwarranted assumptions. 

 The general curve of disinfection, plotted logarithmically, is a 

 straight line in the main, but at each end there are characteristic 

 deflections. If these deflections are of small extent then an 

 average velocity constant calculated for the whole of the disin- 

 fection period will involve no serious error. If, however, there 

 is a lengthened period of induction before disinfection proceeds 

 at the maximum rate, then obviously, such a basis for the calcu- 

 lation of the velocity constant would involve a large error. The 

 same objection would hold for the condition in which the disin- 

 fection curve flattens out toward the end of the process. The 

 latter consideration prompted Phelps (1911) to recommend the 



