124 
verse of the usual change. So rare is this metamorphosis of 
petals to stamens that Masters, in his Vegetable Teratology, re- 
cords only two or three perfect examples of it. In the common 
shepherd’s purse, Capsella Bursa-pastoris, he says such a change 
“has been observed,” but he quotes no authority. The one pos- 
itive case, which seems to rest on his own observation, he states 
in these words: “ There is in cultivation a form of Sarifraga 
granulata, wherein the petals are replaced by stamens, so that 
there are fifteen stamens.” 
S. granulata is a common European species, with large and 
handsome white flowers, and has been a garden plant for many 
years, and it is pretty certain that the variation occurred in some 
florist’s establishment, and, being an oddity, was maintained in 
cultivation for a while by division, quite independent of seeds. 
In the present case, however, the new form appears in the wild 
state, and the plant must produce good seeds “after its kind” in 
order to be perpetuated and diffused. To my mind the evidence 
is considerable that it is already doing this.* 
Now, why has this change occurred in two species of the 
self-same genus, separated as: to station by over a thousand 
leagues of land and water? Is this identity of variation in S. 
granulata and S. Virginiensis a mere coincidence? Or have we 
here a striking case of atavism ? Is this variation the recurrence, 
_ in the descendants, of the peculiar and long obsolete structure of 
their common ancestor? Was the progenitor of the hundred 
and sixty or more distinct Saxifrage of to-day a plant with 
apetalous and fifteen-stamened flowers? We shall never know 
with certainty, but two of its descendants testify strongly in the 
affirmative. I would even go a step farther, and hazard the con- 
__ jecture that the original of the Saxiffages was dicecious, or at least 
polygamous. In the two perfect plants I have described the 
stamens were remarkably vigorous and well developed. In the 
ten or more others the stamens, as I have said, were singularly 
imperfect, and numbers of them were curiously ovary-like in ap- 
pearance. Is it possible that these plants were blindly struggling 
to reproduce a primitive pistillate form ? A certain confirma- 
* In case the new form should perchance prove permanent, I propose for it the 
name: Saxifraga Virginiensis, var, PENTADECANDRA.—E. E. S. 
