468 Ganong . — 77 ^ Comparative Morphology of 
and relatively diminished hypocotyl, allowed by exposure to 
more mesophytic habitat. Third, some minor details of out- 
line are due to the position of the embryos in the seed. 
The form-factors of the epicotyl are simpler than those of 
the embryos, and from what has been said in the preceding 
pages it will appear that we have in the epicotyls remarkably 
clear examples of the working back of the characters of the 
adults. The first leaves of the epicotyl (very small except 
in Pereskia and Opuntia ), and indeed often the cotyledons 
themselves, produce axillary buds which at once develop the 
characteristic spine-clusters. This is true in all of the species 
that I have examined throughout this family ; the very first 
formed axillary structures are true spines. There seems to 
be abundant evidence that the spines in this family really are 
metamorphosed leaves and not emergences or other dermal 
or epidermal structures ; and recalling the principle of repeti- 
tion of phylogeny in ontogeny, one would expect to find in 
these first formed axillary clusters some trace of a leaf-nature 
showing in the spines, which is never the case. This fact, 
however, is not necessarily evidence for a non-phyllome 
origin of the spines, since it may, and probably does, mean 
simply that the production of leaf-spines from axillary buds is 
very old, preceding the differentiation into genera, which 
indeed Pereskia proves to us is the case. It may be that 
this ancient habit of producing axillary spine-clusters is the 
oldest existent characteristic of the family, and may yet give 
the key to its still unknown phylogeny. The epicotyl of 
course tends to repeat its hereditary characters, and does so 
until these are obliterated by the working back of new 
ones, and it is the different stages of the working back of 
these which give us the very fine examples of repetition of 
phylogeny in ontogeny which appear in the climbing Cerei, 
Phyllocacti, Rhipsalis , &c. Taking the family as a whole 
we may picture successive waves, as it were, of characters 
acquired by adaptation in the adults sweeping back into the 
later seedlings, and wiping out earlier characters. The more 
superficial features of these waves do not however affect the 
