82 
Proceedings of the Royal Society of Edinburgh. [Sess. 
physiologist but usually eschewed by the morphologist, and would provide 
material for a study of the more recondite principles underlying animal 
architecture. 
As showing the suggestiveness of the problem, I may refer to a feature 
present in Amphipoda which distinctly gains in interest by comparison 
with the bony limb. 
Backward rotation of the fore-limb of the land mammal has the dis- 
advantage of making the digits point backwards, so that phalangeal 
flexion resists not a backward but a forward pull on the part of the 
limb. This defect is redressed by a subsequent secondary torsion of the 
forearm through two right angles, the dual bones here providing the 
Fig. 4.— Diagram to show liow one of the posterior thoracic limbs of Gammarus with 
anteriorly pointing dactylopodite is made to assist in forward progression. The 
limb is thrown over the back and the dactylopodite then points backwards. 
structural basis for rotation in the axis of the limb. In the Gammaridea 
forward progression occurs by means of the posterior three limbs. Here 
the terminal segments, in their original position hooking into the ground 
so as to resist a forward pull on the part of the limb, have become adapted 
to resist a backward thrust by undoing of the first major flexure of the 
limb to the extent of a right angle and exaggeration of the second major 
flexure to the extent of another right angle, as in the diagram, fig. 4. 
Torsion of limb-segments is here avoided, the limb is simply thrown over 
the back, to which rearrangement the animal accommodates itself by 
adopting a new mode of progression, viz. with one or other lateral aspect 
addressed to the walking surface. 
One naturally raises the question : Can extensive torsion of a segment 
of an arthropodan limb occur in the process of development or of 
