4 Oliver and Salisbury . — On the Structure and 
Our object was to reconstruct our seed in the form of an enlarged 
model upon which the planes of the various sections could be plotted, and 
then, by cutting along the appropriate plane, any given section could be 
reproduced. We began by plotting a sketch of a provisional median longi- 
tudinal section, using for the purpose such data, not exaggerated by obliquity 
of plane, as could be derived from the most favourable preparations avail- 
able. As each successive section was handled the sketch of course under- 
went gradual modification. 
Thus, every section contributed something ; and in the end, after 
repeated correction, a sketch was obtained upon which all our sections could 
be approximately plotted. This ‘ ideal ’ section could at best be no more 
than an approximation, because the sections upon which it was founded had 
been cut from a large number of specimens which from the nature of the 
case would show some variability of dimensions, even if the extreme 
assumption were made that they all belonged to the same developmental 
phase. 
With the ‘ ideal 5 longitudinal section as basis, a model of the halved 
seed on a scale of 40:1 was constructed in papier mache, together with 
a number of identical plasticine models of the complete seed on the same 
scale from which to reproduce our original sections. The papier mache 
model, which gave internal details as well as surface relief, was of course of 
great utility in orientating the planes to be followed in sectioning the plas- 
ticine models. For accuracy in cutting the sections of the models a simple 
guillotine or microtome was devised. It consisted essentially of an oblong 
frame hinged at its narrower end on to a board which formed the support 
on which the model was placed for sectioning. The plasticine model pro- 
j ected through this frame (which could be adjusted at any required angle), 
and the section was cut by sliding a taut copper wire along the smooth 
upper surface of the frame. By the use of this contrivance we have found 
it possible to reproduce in essentials any given section of our series. 
Repeated trials have often been necessary, for the slightest differences in 
either the plane of cutting or in the relation of the point of entrance of the 
stretched wire to the surface bring about very marked differences in the 
contours of the sections produced. In other words, form, as expressed in 
a section, is an extremely sensitive thing, subject to very striking fluctua- 
tions from apparently trivial causes. What is true of the form of an object 
like a seed holds also in the case of minute histological detail. 
When reliance has to be placed on the study of oblique sections (an 
everyday occurrence in palaeobotanical work) we have found wrong infer- 
ences, as to the three-dimensional figure of the elements cut, to be almost 
inevitable without checking by means of models. 
In the case of the seeds described in this paper, the plotting of almost 
every section has been verified by following the method outlined above, and 
