62 



VARIATION AND DIFFERENTIATION IN CERATOPHYLLUM. 



In the first case the constant is, of course, certainly significant in com- 

 parison with its probable error, and in the second case very probably 

 significant. 



It thus being clear that the relation between leaf-number and posi- 

 tion of whorl is not a simple linear one, it becomes necessary to find out 

 what it is. To do this we must first see what the exact form of the 

 observed regression is. In table 34 is given the mean leaf-number for 

 each array of whorls occurring in a given position, as indicated in tables 

 30 to 32. 



Table 34. — Mean nurtiher of leaves in successive whorls on primary branches. 



"It will be remembered that in these two series all branches are clubbed together, while In the 

 case of the otlier series we are dealing here with primary-branch whorls only. 



There is no doubt from the figures in this table that the form of the 

 regression line is essentially the same for all the series. In order to 

 show this graphically I have prepared fig. 11, in which the observed 

 regression lines for Series I, II, III, and IV are plotted. To prevent con- 



