70 Salisbury. ~ 0 n the Structure and Relationships of 
F. W. Oliver has called attention to the remarkable resemblance 
between the plans of T. Parkinsoni and P olylophospermum} one which is 
rendered complete if we assume that the secondary ribs of the former were 
the representatives of originally commissured structures. The conclusion, 
then, seems warranted that the six ribs of T. Shorensis and the non- vascular 
ribs of T. Parkinsoni are homologous with the six commissured ribs of 
Polylophospermnm (in other respects, of course, this seed is specialized). 
It is, moreover, probable that the sutures themselves represent the planes 
of lateral fusion between six originally free members — a view which is 
further supported by the recognition of six component units in the micro- 
pylar region of T. Shorensis . 
Newberry 2 has figured a cast showing the apex of a Trigonocarpus 
with its sarcotesta preserved, and in which the micropylar canal is seen as 
a star-shaped opening surrounded by six small but free sar cotes tal lobes. 
We have noted how, in the genus Trigonocarpus , ribs have become 
completely eliminated and each pair of units fused laterally, so that three 
sutures only remain. This condition obtains in the genus Pachytesta , and 
further reduction might well result in the production of a non-commissured 
seed, as in Stephanospermum. 
The remaining genera cited above are either six- or twelve-ribbed 
seeds, and in the latter case the additional members may well correspond 
to the six vascular ribs of Polylophospermum. 
Other genera there are, belonging to the Radiospermeae, which never- 
theless possess ribs that in number are some multiple of two, e. g. Eriotesta , 
Codonospermum. Our review has, however, shown that there existed a group 
of seeds agreeing in their internal structure and consistent with our hypo- 
thesis in their sclerotestal plan, moreover including all those forms in which 
an undoubtedly Trigonocarpean organization has been established. 
The possible origin of dimerous from trimerous forms is considered in 
the sequel, but the Radiospermeae is doubtless an artificial aggregate of 
which the Trigonocarpeae is one of the included natural groups ; a state- 
ment that is borne out by the constant association with these seeds of the 
Neuropteridian type of foliage . 3 
None of these genera militate, then, against the theory that the integu- 
ment originated as a whorl of six free members, each with a single vascular 
bundle, which subsequently became laterally fused. Such an interpretation 
necessitates the recognition of the sarcotesta and sclerotesta as constituting 
a phylogenetically homogenous structure, and in the next section will be 
given data that amply warrant -that conclusion. 
A striking feature of the Trigonocarpeae is not only the occurrence 
1 Ann. Bot., vol. xxi, 1907, pp. 303-4. 
3 Rep. Geol. Survey of Ohio, vol. i, Pt. 2, 1873, p. 336 and PI. XLTI, Fig. 5. 
3 See M. Grand’ Eury, Comptes rendus, vol. cxxxix, 1904, p. 3. 
