J. ARTHUR HARRIS. 



The results are represented graphically in Figs. 4 and 5. In 

 the Italian series, flowers with fewer than 36 stamens and pistils 

 are only 29 in number, distributed among flowers with from 

 28-35 sporophylls per flower. The means at this end of the 

 range cannot, therefore, be expected to show great regularity. 



--2 



2? 3? 32 3,4 36 3,8 40 42. 44 46 48 SV 52 ^4 



FIG. 4. Regression of the deviation of the total number of pistils* from their 

 probable value on the total number of sporophylls. Italian series. ' 



Flowers with more than 61 sporophylls are only 10 in number 

 but are distributed among flowers ranging from 62-85 sporophylls 

 per flower. This portion of the range has not been included in 

 Fig. 4. The relationship is apparently not quite linear. 



In the material from Surrey, shown in Fig. 5, the increase in 

 the relative proportion of pistils associated with increase in the 

 total number of stamens and pistils could hardly be better 

 represented than by the slope of the straight line indicated by 

 the equation. 



