436 Ganong . — The Comparative Morphology of 
view of its distribution, the latter seems far more probable 
than the former as an explanation of the colour in these 
embryos- It is true the colour is present in the wet season, 
when the weather is clouded, but it is not unlikely that even 
then there are periods of vivid sunshine. But in the absence 
of exact observations of the subject in their native homes, 
these theories, like many others relative to the ecology of 
desert-plants, are mere guesses. 
The Size-Factors of the Embryos. 
Comparing the germinated embryos throughout the family, 
we find a wide range of size, from Pereskia on the one hand 
to Mamillaria on the other (see figures ; Pereskia and Opuntia 
are drawn natural size, all others magnified two and a half 
times). In seeking for the influences which determine the 
size in the different cases, we naturally turn first to ask 
whether size in the germinated embryo is related in any way 
to size of seed from which it comes. Excluding now certain 
cases presently to be considered, in some species of Cereus , 
Phyllocactus , and Opuntia , where there is a special growth of 
the cotyledons, it is plain that there is a relation between 
these two. It not only shows in a comparison of seed and 
embryo, but it follows also from the fact I have mentioned 
that the growth of the embryo to its fully germinated stage is 
little more than a swelling to full size of cells already laid 
down in the seed. The question then resolves itself into 
this — is it, in the main, the size of the seed which determines 
that of the germinated embryo, or is that of the embryo 
the more important factor, the seed being the size it is in 
order to accommodate it ? If one compares the size of 
embryo or seed with the size of the adult plant throughout 
the family, it appears at first sight as though there were no 
constant connexion between these two. Cereus gig anteus, for 
example, has a seed no larger than many of the smaller Cerei. 
But if, instead of particular species, we consider groups such 
as genera, we find that there is a relation between the average 
