400 MEMOIRS OF THE NEW YORK BOTANICAL GARDEN 



are not included in the table. The highest number of seeds for a 

 single head was 12. There were 12 cases of no seeds per head; 

 10 of these were on July 14, the same date when 3 heads of plant C 

 (see TABLE 15) failed to set seed. The weather records taken at the 

 Garden show that on that date it was cloudy with some rain. 

 Evidently on this particular day the weather conditions were 

 unfavorable to insect pollination or to fertilization. 



In comparison with plant C, the plant (C X Ej) no. 6 exhibited 

 a low degree of fertility; the daily percentage of seeds set was 

 lower and the number of seeds per head was consistently less. 

 Such results may be due to a marked cross-incompatibility on the 

 part of plant [A X -Ej) no. (5 to a greater number of plants in its 

 vicinity. The fact that there were quite constantly few seeds per 

 head is, however, quite in line with the conditions encountered in 

 selling and crossing. It is quite clear that the difference in the 

 total seed production of these two plants is not due to a difference 

 in flower number. Here the determination of flower number 

 enables one to compute accurately the percentage of seed produced 

 on the basis of the total that is possible. 



The data in tables 14 and 15 are quite conclusive that heads on 

 any part of the plant at any time in the season are productive 

 of seed; a point that seemed equally evident from the writer's 

 experience with the controlled pollinations. 



From the data of tables 15 and 16, Dr. J. A. Harris has com- 

 puted the correlation of flower with seed number per head accord- 

 ing to the method he has reported ClO). There is in both cases 

 a very slight negative correlation. This is so slight, however, 

 that it scarcely if at all affects the general results and con- 

 clusions, as it is clear that random pollinations at any time very 

 fairly represent the fertility of the combination, and that the 

 marked differences in fertility of self- and cross-pollinations are 

 entirely independent of diff^erences in flower number. 



While there was considerable fluctuation from day to day in 

 the percentage of seeds set to open-polHnation, there were usually 

 in any one day some heads tiiat were close to the maxinunn 

 number, especially if five or more heads were involved. In this 

 respect the plants C and iC X £j) no. 6 were quite uniform 

 throughout the season. 



In the case of the controlUd pollinations, pollen from known 



