of Fertilization in Angiosperms. 697 
sperms. We have no series of forms to link the structures 
produced within the embryo-sac of such a plant as Lilinm 
Martagon with those derived from the macrospore of the 
higher Cryptogams. The connexion of the Gymnosperms 
with the Vascular Cryptogams can be traced in their gameto- 
phytes. But though the ancient and very distinct groups 
included under the general term Gymnosperms are in 
some sense intermediate between Vascular Cryptogams and 
Angiosperms, yet our knowledge of embryo-sac structure in 
anomalous Angiosperms is so incomplete that we can hardly 
do more than hazard a guess as to whether any particular 
deviation from the best-known type is to be considered a 
specialized, a reduced, or a primitive form. Thus the attempt 
to derive Angiosperms from one or more primitive forms 
related to one or more groups of existing Cryptogams must 
be made at present without the aid of independent evidence 
as to the origin and progressive reduction of the Angiosper- 
mous gametophyte. But this is the very class of evidence which 
has been found most valuable in tracing the origin of any 
group of Gymnosperms, for example, from some ‘ generalized ’ 
form of Vascular Cryptogam. The gametophyte of some 
species of Gnetum is little less reduced than that of the 
typical Angiosperm. But the work of Karsten ( 28 ) and 
Lotsy ( 30 ) enables us to trace its reduction step by step 
through the successive forms of Gnetum Gnemon , Welwitschia, 
Ephedra , from a.massive tissue such as that found in Cycas or 
the Coniferae. From that point the connexion with the 
prothallus of the higher Cryptogams is clear. 
Until some causeway such as this has been built up stone 
by stone to connect the Angiosperms with their neighbours, 
all theories as to the race-history of the structures contained 
in the embryo-sac of Lilium —to take the classical example— 
must be considered as at best working hypotheses, useful 
indeed to stimulate and direct research, but without authority, 
and in their very nature tentative. 
Without such hypotheses research could not proceed. 
Where two or more explanations of the same facts are equally 
