16 THE MEASUREMENT OF VARIATION. 



flowers, instead of the normal irregular ones) was as 

 follows : 



Number of flowers, 234 5 678 



Frequency, 1 2 43 810 52 2 1 



Percent., .109 .219 4.720 88.913 5.708 .219 .109 



From these two series a very interesting relationship 

 declares itself, w T hich may for convenience be referred 

 to here, though it properly comes under the heading of 

 " correlated variations." Thus, as the following fig- 

 ures show, we find that the probability of occurrence 

 of a peloric flower increases according to the amount of 

 deviation of the number of flowers on a stalk from the 

 normal pentamerous form, or that the less often a 

 particular number of flowers occurs, the more fre- 

 quently does it produce peloric flowers : 



.132 per cent, of the 5 flower form have peloric flowers. 

 15.19 " " 4 



23.53 " " 6 



22.22 " " 7 



33.33 " " 3 " " " 



100.00 " " 2 and 8 " 



Of English observers, J. H. Pledge * has determined 

 the variations in the numbers of petals, stamens, and 

 carpels in 1000 specimens of Ranunculus repens (creep- 

 ing crowfoot), the distributions of all but the numbers 

 of petals agreeing fairly closely with the probability 

 integral. For instance, the numbers of sepals varied 



thus: 



Sepals, 34 567 



Frequency, 1 20 959 18 2 



We see, therefore, that in the majority of the 

 characteristics of the various organisms investigated, 



* Nat. Science, vol. x. p. 323, and vol. xii. p. 179, 1898. 



