1892.] NATURAL SCIENCES OF PHILADELPHIA. 375 



ously but advance from stage to stage by leaps — making consider- 

 able rests between the stages — is well illustrated by the flowers of 

 this plant. After the flower bud has reached a stage ready fof 

 expansion, it rests for a day but the style continues to grow and 

 pushes through the closed flower bud to the extent of about two 

 lines. Then it rests, and the corolla opens and assumes an erect 

 campanulate form. The stamens grow as the petals lengthen, but 

 continue growing for a day after the corolla is at rest, continuing till 

 they exceed the style, the anthers forming a close circle just above 

 the stigma, when they discharge their pollen over it. As every 

 flower is fertile, and the plant produces seeds profusely, I surmised 

 that the flowers must self-pollinate but the advance of the pistil, 

 with its evidently receptive stigma, so long before the maturity of 

 the anthers, seemed theoretically against this view\ In a large area 

 of these flowers, where some plants in bloom would matui'e in 

 advance of the others, insects might convey the gelatinous pollen to 

 the exposed pistils on other plants. In this one specimen, however, 

 there were no insect visitors observed except an occasional sand 

 wasp, and the effect was only to help the stigma to its own pollen. 

 This plant was certainly self-fertile, though the conditions seem to 

 be such that it might be cross-fertilized under favorable circum- 

 stances. 



Campanula rotundifolia. 



A large branch of specimens, placed in water for a week, pre- 

 sented some remarkable variations. The lobes in most instances 

 were about one-fourth the depth of the campanulate corolla, in 

 some instances one-third. In one case the corollas on the stem were 

 cut to fully one-half their length, and the lobes spread so that with 

 a little more efibrt the corollas would have been rotate ! The flowers 

 of this species of Campanula are centrifugal, the terminal one 

 opening first. A number of these terminal flowers were 10-lobed 

 with ten stamens, still more with six lobes and six stamens, but the 

 majority were normal with five lobes and five stamens. In one flower 

 with five lobes, the five stamens had been transformed to petals, and 

 it is worthy of note that these five were separate and not united into 

 a monopetalous corolla similar to the outer series. On the same 

 stalk, another flower had two of the stamens somewhat petaloid 

 instead of antheriferous. Two other flowers on the same stalk were 

 normal. 



