STATISTICAL METHODS IN PROTOZOOLOGY 



447 



TABLE 3 

 CALCULATION OF THE EXPECTED FREQUENCY OF 



OF TROPHOZOITES OF Giavdia 



MEASUREMENTS OF LENGTHS 



Bradypi 



It vi^ill be seen that the theoretical frequency agrees satisfactorily 

 with the observed and we may therefore consider that this material 

 is normally distributed. A more thorough test of the agreement 

 between the observed and theoretical frequencies will be given 

 in the section on the Chi = square test. 



It cannot be too strongly emphasized that the use of the normal 

 curve in such cases is purely empirical, the curve being in no 

 sense a "law of frequency." Its value rests rather on the fact 

 that it does fit the distributions of measurements in many problems 

 and due to the tables that have been prepared furnishes a very 

 convenient form for the graduation of a distribution, to determine 

 the proportion of cases that would be expected between any par- 

 ticular limits. 



CORRELATION 



up to this point, we have attempted a description of measure- 

 ments on a single variable, but problems often arise in which we 



