6 Bulletin 392 



cutting a small hole thru the seed coat to allow the water to enter. In 

 1916 seeds were successfully germinated in petri dishes on damp filter 

 paper, arid the seedlings were then put directly into the soil of the garden. 



Early frosts at the end of two seasons killed the cultures before com- 

 plete notes had been taken, so that only very meager data were obtained 

 on most of the pedigrees and in many instances no seeds were produced 

 for carrying certain pedigrees further. In the summers of 191 5 and 1916 

 the plants in the garden were so late that no crossing was attempted. 

 Some of these plants were cut back severely and removed to the green- 

 house, where crosses were later made with them. 



The stigma is receptive for pollination on the same morning that the 

 flower opens. Pollination must be done early, as the flowers wither 

 very soon, especially on hot days. Emasculation can be performed as 

 late as the afternoon previous to the day of the flower's opening. The 

 flowers are so readily self -fertile that when selfing was desired in these 

 experiments it was found sufficient to tie the tip of the mature bud with 

 a bit of cotton thread and leave it alone, or else to shake and pinch the 

 corolla a little the next morning in order to insure the pollen's reaching 

 the stigma. Except in the warmer months, however, when the venti- 

 lators and doors of the houses were open, or when the plants grew in the 

 garden, this precaution was not necessary. No trouble was given by 

 insects interfering with the pollinations. In the garden all pollinations 

 were guarded by bagging. 



In the winter of 1912-13, 488 cross- and self-pollinations were made in 

 the greenhouse, of which 55 per cent set seed. In the simimer of 1912, 

 between August 14 and September i over 600 pollinations were made 

 in the garden, of which 51 per cent of the self-pollinations and 35 per 

 cent of the cross-pollinations set seed. 



The colors of the flowers were at first described and recorded by means 

 of the Repertoire de Couleurs.^ Later it was found more convenient and 

 more nearly accurate to compare each flower with a colored sketch and 

 refer it to a certain type. The light purple type varies from approxi- 

 mately dark blue to dark purple. Dark purples on withering break 

 down into light purple and magenta. Indeed, all the blue series are 

 subject to such a change in withering; the blue pigment breaks down 

 and the flower appears transformed into one of the red series. Thus, 

 in cool, cloudy weather flowers may remain open for two days, blues on 

 the second day appearing as red. On cool mornings in early autumn 

 the pigments do not develop properly, and it was found necessary at 

 such times to give up all attempts to classify the types. Records were 

 always checked at least once when possible, and often many times. 



' See under Societe Francaise des Chrysanlhi'misles el Rent Oberth'tlr, in References cited, page 38. 



122 



