478. BOTANICAL GAZETTE [DECEMBER 
It has already been pointed out that those species in which 
hybridism is habitual generally show a well marked variation. In 
this connection it may be well to mention the genera Rubus, Rosa, 
and Crataegus, whose fluctuation is very pronounced, in proof of 
which the lack of agreement in regard to their classification among 
systematic botanists is significant. This type of variability like- 
wise is very noticeable in a study of any large amount of herbarium 
material of Epilobium, such as was observed by the writer in his 
_ examination of specimens at the Gray Herbarium. It is equally 
important in this connection to state that HaussKNECHT (17) men- 
tions 14 formae of E. angustifolium, to of E. latifolum, and 6 of E. 
Dodonaei. Thus it would appear that this subgenus presents 
capricious conditions of change similar to the genera just mentioned, 
which possess known hybrids and likewise present morphological 
evidences of pollen sterility as an indication of past interbreeding. 
Epilobium Dodonaei Vill. is apparently subject to similar sterile 
conditions, since 2 out of the 3 specimens obtainable showed abor- 
tive pollen. This feature was also observed by Miss HoLpEN in 
Cambridge, England, as a result of an investigation of microspore 
conditions in relation to hybridism in this species and in E. angusti- 
folium from Europe. The probability that crosses of this nature 
have taken place is strengthened by the fact that a hybrid from E. 
DodonaeiX E. spicatum is noted by HaussKnecut as described by 
HENNIGER. 
Before turning to other species of the Onagraceae it seems 
advisable to summarize briefly the behavior of the species already 
discussed. It is apparent that the section Chamaenerion presents 
features which are characteristic of genera subject to hybridization 
in nature, as indicated both by sterility of the pollen and by a well 
marked variation. It is also evident that there is a distinct 
relation between this morphological proof of intercrossing and the 
geographical distribution of the species. In those regions where 
E. angustifolium has an opportunity to cross with its close allies, 
the anatomical proof appears conclusive that such crosses have 
taken place. Evidences of this phenomenon are still further 
strengthened by the fact that whenever this plant occurs sufficiently 
far south for crosses with other species of the subgenus Chamaene- 
