158 MORPHOLOaY OF SPERMATOPHYTES 



ales, and probably so in Ginkgo, although in the latter case the 

 supposed carpel is a very rudimentary structure. Among the 

 Cordaitales, Bennettitales, Gnetales, and Taxaceae, the ovules 

 seem to be clearly cauline ; while among the Pinaceae the carpel 

 is said to be " reduced to the ovule "; the latter therefore would 

 be essentially cauline. One is tempted to regard the foliar 

 ovule as the more primitive type, and to trace the reduction 

 of the sporophyll from Cycas, where it is well expressed, through 

 Ginkgo, in which it is much reduced, to the Conifers and 

 Gnetales, in which it is eliminated. To such a view, how- 

 ever, the testimony of history is opposed, for among the Cor- 

 daitales, the most ancient of recognized Gymnosperm groups, 

 the ovules were plainly cauline, and apparently so among the 

 Bennettitales, the most ancient of the Cycadean forms. The evi- 

 dence seems to be that the foliar ovule does not necessarily indi- 

 cate a more primitive group than the cauline, and that the dis- 

 tinction between these two types of ovules is more arbitrary 

 than real, often impossible to apply, and of no special signifi- 

 cance. 



The question of integuments is another Gymnosperm prob- 

 lem. The usual statement is that the integument is single in 

 all Gymnosperms except Gnetum, and this certainly expresses 

 the result of ordinary observation. If the definition of integu- 

 ment be restricted to siich structures as receive the name in 

 Angiosperms, then it can be claimed that one integument is the 

 rule among Gymnosperms. If it be allowed, however, that the 

 integument may become so modified as to lose its character of 

 an ordinary investment of the ovule, we are launched into a dis- 

 cussion whose merits can not be decided by demonstration. It 

 is not clear why it is essential for so ancient and diversified 

 a group as the Gymnosperms not to vary in the number of its 

 integuments. 



As opposed to the ordinary view that one integument char- 

 acterizes the group, and Lotsy would even include Gn-etum in 

 this generalization, is the view of Celakovsky that the whole 

 group is characterized by two integuments, the outer one of 

 which is often remarkably modified. This view certainly 

 sweeps into one category a number of problematical structures 

 and has the attractiveness of uniformity. The application of 

 this theory would result somewhat as follows: Among the an- 



