418 PROF. OLIVER ON THE STRUCTURE AND MODE OF DEHISCENCE 
one inch long, cut out of this layer, gained $ a line, or 44th of its length, when soaked. 
The fifth and outermost layer (e) consists of long coloured cells, disposed at right angles 
to those of d. I have not tried whether they possess any independent power of expansion 
when moistened, or not. The measurements given above were from single pieces cut out 
of each layer and thinned down, so that, if repeated, it is quite probable a slightly 
different result would be obtained, though not, I think, materially affecting their relative 
proportions. In different parts of the same valve also the several layers vary, no doubt 
in the proportion which they bear to the whole; the extent to which they vary I have 
not attempted to find out. í 
. In some legumes, especially with follicular dehiscence, the line of contraction is almost 
exactly transverse to the valves. In Platylobiwm, which dehisces in this way, the prin- 
cipal thickness of the valve is made up of thick-walled prosenchyma, the cells of which 
are disposed in the direction of contraction. The cells of the inner surface are of greater 
diameter in the direction of the thickness of the valve than are the outer cells of the 
contractile layer. The contraction of the latter, however, preponderates, and the valves - 
are consequently somewhat revolute, transversely, when dry. 
Fig. 12 represents a section, oblique to the valve, of a legume of Crotalaria incana, cut 
in the direction of the striation of its inner face. This striation is due to the surface- 
cells of the layer e; b is a thin layer of similar cells at right angles to those of the inner 
layer c. The principal contraction of the valves appears to be in a direction nearly, or 
directly, transverse to the valves. I apprehend that both c and b are contractile layers, 
and as they are at right angles to each other, and each oblique to the valve, the con- 
traction of the valve in a transverse direction may be regarded as the resultant of 
their united action. The outer layer (a) consists of thin-walled parenchyma. Fig. 13 
represents the layers û and c cut at right angles to the section fig. 12. I have made no 
measurements of the tissues of these valves. 
The horny hygroscopie contractile tissue of the valves of Pentaclethra increases in 
diameter as well as in length when moistened; but as this tissue is arranged in distinct 
bundles, the diameter of which is inconsiderable, the increase in breadth (or corre- 
sponding decrease when drying) is not practically of moment. I have but once measured 
a transverse section before and after moistening; the difference amounted only to about 
3 or 4 per cent. Von Mohl states that the longitudinal contraction of dicotyledonous 
_ Wood, in passing from the wet to a perfectly air-dried condition, amounts to about ‘07 to 
4 per cent., the contraction in the direction of its breadth amounting to from 4 to 9 per 
cent. In a large number of legumes in which the contractile tissue is disposed in 8 
single uniform layer upon or near to the inner face of the valves, as, for example, in 
Lathyrus or Pisum, it would appear that the twisting of the valves when dry is due to a 
preponderant contraction of this layer in a transverse direction; in other words, the 
layer contracts most in the direction of the breadth of the cells which compose it. 
Fig. 10 represents a section of a valve of Lathyrus cut transversely to the cells of its 
hygroscopic layer, but in the direction of contraction of the valve. The cells have their 
