38 LUTHER BURBANK 



when attempting to cross the ordinary canna 

 with the Florida species already mentioned. I 

 worked for eight years with that purpose in view 

 before succeeding. And even then the seedlings 

 were greatly lacking in fecundity, producing 

 very little seed, notwithstanding the fact that 

 cannas in general usually produce seed abun- 

 dantly in California. 



The infecundity of the canna hybrids suggests 

 that the species in question are almost at the 

 limits of affinity. But the seeds produced, al- 

 though few in number, were some of them fertile, 

 and the hybrid progeny showed possibilities of 

 development, as already suggested. Most of the 

 later generations, however, are almost or quite 

 sterile, refusing to seed. 



The chief difficulty in growing seedlings of the 

 canna is to insure the germination of the seed. 

 The familiar name " Indian-shot plant" by which 

 the canna was first known suggests the character 

 of its seeds, which are not unlike small bullets in 

 appearance and in hardness of texture. The old 

 plan of germinating the seeds used to be to file 

 off part of the thick shell, in order that the seed 

 might absorb moisture. This works very well, 

 but can hardly be applied on a large scale. 



My own method has been to disinfect the canna 

 seed with a solution of bluestone (sulphate of 



