ORGANIC VARIABILITY 545 



or bias, in one direction or the other. This was the older con- 

 ception of variability, and it was held to be " normal " and 

 analogous to a physical curve of error. But very little experi- 

 ence of biometric work shows that frequency distributions 

 described by the " normal curve of error " are quite excep- 

 tional. We may indeed describe distributions in this way, 

 but analysis shows that they are, as a rule, far better described 

 by other types of the general equation. (2) The second case 

 arises when the coefficient c 3 vanishes, leaving c and c x . The 

 contributory causes of variability are still very numerous, and 

 they are independent of each other, but they have bias and 

 the distribution is asymmetrical, on one side or other, of the 

 mean. This case is more frequent than the first one. (3) 

 The third case arises when all three co-efficients, c , C\ } and 

 c-j, are present ; the numerous contributory causes of vari- 

 ability are then no longer independent of each other, but are 

 correlated. They give asymmetrical distributions, that is to 

 say, series in which the variations exhibit a definite tendency, 

 and this is by far the most common type of organic variability. 



