ON TECHNIQUE 57 



nection with a study of morphologic variations during the death 

 phase described in Chapter IX. 



The statistical part of the work was greatly facilitated by the 

 use of an adding machine, measurements being recorded on that 

 instrument as they were observed, so that the total and consequently 

 the average w^ere immediately available. In the earlier w^ork with 

 B. megatherium recorded in Chapter IV, 300 cells were measured 

 from each sample except during the period of great variability when 

 500 were measured (1,000 in the case of the five and one-half-hour 

 sample from Culture I) ; 300 cells were also measured from each 

 sample in the case of the diphtheroid organism described in Chapter 

 VI; but it was apparent that nothing was gained in the way of 

 smoother curves by using such large samples, and the number of cells 

 measured was reduced to 250 in the case of the colon bacillus (Chap- 

 ter V), and to 200 in the succeeding studies. Cell measurements 

 were made to the nearest millimeter of the projected image, and this 

 unit was chosen in most cases for the class intervals of the fre- 

 quency curves; the class intervals are therefore smaller with the 

 higher magnifications. The various statistical constants have been 

 computed by standard methods save the modes and medians. Modes 

 have been determined simply by inspection of the smoothed frequency 

 curves, tnedians from the (unsmoothed) cumulative percentile curves. 

 The frequency curves presented in the succeeding chapters have all 

 been smothed twice by the method of three-point averages. None 

 of the other curves have been smoothed save those shown in Figure 

 25 and Figure 27, which have been fitted by simple inspection. 



