Hill and de Fraine . — On the Classification of Seed-Leaves. 361 
were three whole cotyledons, the vascular bundles of which gave rise to 
a triarch root-structure which subsequently became reduced to diarch, 
a phenomenon not at all uncommon in the Coniferae. 
With regard to the other seedlings of Phacelia , the following analysis 
is perfectly obvious from Mr. Compton’s description. 
Tricotyl C and Hemitetracotyl D. Three whole cotyledons. Root- 
structure triarch throughout. 
Tetracotyl E. Four whole cotyledons. Root- structure tetrarch above, 
but successively reduced at lower levels to triarch and diarch. 
It is clear that Mr. Compton quite misunderstands our conception, and 
this, in part:, is possibly due to the fact that he worked from below upwards, 
whilst we, in common with most other investigators of seedling anatomy in 
this country, worked from the cotyledons downwards. This appears from 
the following quotation : 1 
‘ The most striking feature in the schizocotyls of Phacelia is the fact 
that the vascular strands [presumably of the root, below the collar] do not 
split to correspond with the split of the cotyledons. In all cases (except 
two, where the division occurred in the lamina) the number of xylem 
bundles is increased by the addition of new ones, not by the division of old. 
On tracing the strands from above downwards, we may say that in these 
cases the traces of certain cotyledons or half-cotyledons do not contribute 
to the structure of the root, but die out in the hypocotyl. This is the 
characteristic 2 of “ subsidiary cotyledons ” — a category of plant-members 
which apparently cannot be maintained.’ 
In all our work, the first organization of a root-structure, whether it 
occurred high up or low down in the hypocotyl or below the collar, was 
taken as a factor in our classification. 
This appeared the obvious course to take, for the fact that the number 
of poles might be altered at a lower level results in chaos, as Mr. Compton 
has found. If the structure of the root at a lower level is to be taken, what 
particular level is to be selected, and what is to guide us in our selection ? 
It is not at all infrequent, especially when the number of cotyledons varies, 
that the number of poles in a root, with the exception of diarch roots, 
becomes reduced and sometimes increased as we pass towards the apex. 
For instance, a pentarch root might sEow a reduction to tetrarch and finally 
to triarch just above the apex; it is possible that if the seedling had been 
allowed to grow for a week or so longer and then examined, a further 
reduction to diarchy might be shown. 
The difficulty could be overcome by referring the structures in question 
to a diarch organization, on the assumption that this is primitive ; and this 
no doubt would be excellent philosophy, but it does not appeal to us as 
science. 
1 Loc. cit., p. 812. 
2 As we have already remarked, it is not. 
