182 NATURAL SCIENCE [March 



in this (calyx) wlioii, the two series agreeing most re- 

 markably. 



The total amount of variation from the mean being so small 

 (4*1 per cent.), it may be stated that there is practically no varying 

 in the number of parts. 



Corolla. 

 Number of specimens with 4 petals 



)> 5J '^ )■> 



!! 35 " >> 



7 



O 



q 



>> >) -'• ^ !> 



12 



3> 5J 1«^ 3) 1 1 



In the second series there is, on the whole, rather less tendency 

 in the petal whorl to vary from the mean than was shown in the first 

 series — 72*2 percent, of the second 500 having the average number 

 of petals, against 69 per cent, last year. The number of specimens 

 having 6 petals is, however, considerably greater than in the first series. 

 By a curious coincidence the number with 7 petals is the same in • 

 both series (36). The greatest difference between the two, lies in 

 the specimens having more than 7 petals, the first series having 52, 

 the second only 17. 



Stamens. . ; 



On reference to fig. 1, which represents the variation in the 

 first series of 500 specimens (those examined in 1896), it will be 

 seen that the curve is fairly regular, there being, however, a decided 

 hump at 49-50. Fig. 2 gives the corresponding curve for the 

 second 500 (1897). This is very much less satisfactory, there 

 being no decided maximum, as in fig. 1, and, also, the recorded 

 maximum 47 does not agree with the calculated mean 54. 



The range of variation is — 



Series I. — Maximum 88, minimum 21. > 

 Series II. — Maximum 84, minimum 23. 



Fig. 8 gives the curve obtained by combining the two series. 

 It must, I think, on the whole, be considered very satisfactory. 

 The recorded maximum is 55, the calculated mean 54. The hump 

 at 49, remarked in fig. 1, is still noticeable, although not traceable 



in fig. 2. 



