362 MEMOIRS OF THE NEW YORK BOTANICAL GARDEN 



can be properly used in manipulating cross- and self-pollinations. 

 In making controlled self-pollinations heads that have opened 

 under a bag are used. One head is removed with a pair of scissors 

 and the petals are cut away as shown in figure 5. The group of 

 broken anther-sacs with the protruding stigmas is used to brush 

 thoroughly the inner surfaces of the expanded stigmas of attached 

 flower heads that have opened under a bag on the same plant. 

 Heads thus treated are tagged and the bags are replaced until a 

 later day. This method insures the distribution of a plant's 

 own pollen on the stigmatic surfaces at a time when these are 

 fully exposed. The chief source of error is the chance that 

 winged insects, which are sometimes very active, may gain access 

 to a head while the bag is removed. To decrease such error two 

 persons have cooperated in the manipulations, one keeping a 

 constant watch while a bag is removed. When insects are ob- 

 served to alight on a head, which occasionally occurs, the head is 

 removed. Pollen ma^^ also be distributed by wind. That such 

 errors cannot be entirely prevented is obvious, and in some cases 

 the results seem to show that such an error has occurred. The 

 scissors used in cutting away the petals are dipped in alcohol 

 after a plant is worked and the hands of the operators are washed 

 in water. 



In making crosses involving a seed parent known to be self- 

 incompatible, the method used is quite like that of selling except 

 that flower heads from one plant are used as a source of pollen 

 applied to the flowers of another plant. Crossing onto a seed 

 parent that is self-fertile necessitates the depollination method 

 described by Oliver ('10). In testing the effects of selling and 

 crossing, it was always the aim to use several heads on one date, 

 and if conditions allowed, to make the same combination on a 

 different date in the same season or even in different years until 

 at least ten heads had been treated. 



Some cases of failure to set seed, especially when an entire head 

 soon shrivelled, were no doubt due to injury such as bruising in 

 handling or by contact with the paj^er bags or with twigs. Such a 

 source of error was operating in all cases of fertility. The data 

 for heads of branches that obviously became broken or otherwise 

 injured were discarded. 



In regard to the immediate development after pollination, there 



