582 Proceedings of Royal Society of Edinburgh. [sess. 
only can exist in a fluid, there is never more than one ray in the 
lower medium. In all the figures, C represents a condensational- 
rarefactional wave, and D a distortional. The first two figures 
show how an incident condensational wave breaks up into two 
reflected waves and one refracted wave. The angles of incidence 
are 36° and 80° respectively. In both cases the greater part of the 
energy is reflected in the distortional form, and in the second case 
the reflected condensational wave is practically non-existent. The 
numbers attached to the different rays indicate roughly the amounts 
of energy associated with them. The accurate values will be found 
in the article already referred to. 
In the four remaining figures the behaviour of an incident distor- 
tional wave at various incidences is shown. The condensational 
wave travels faster than the distortional wave, and is therefore 
in all cases reflected at a greater angle. When the angle of inci- 
dence approaches 35° (see fourth diagram), the reflected conden- 
sational wave is sent off at a very high angle, and carries away with 
it most of the energy, about 80 per cent., while the amount of 
energy associated with the reflected distortional wave is excessively 
small. At a slightly greater angle of incidence, namely, 35° 16', 
the reflected condensational wave passes off parallel to the surface 
with zero energy, and at higher incidences has no existence. 
With the vanishing of the reflected condensational wave at this 
critical angle, the refracted condensational wave also vanishes — a 
very remarkable result. Consequently at this angle all the energy 
is reflected back into the rock in the distortional form. See the 
fifth diagram, in which the direction the refracted ray would have if 
it existed is indicated by a dotted line. For higher incidences the 
refracted ray comes strongly into evidence, accounting for approxi- 
mately half the energy, but becoming less important at very high 
incidences, until at grazing incidence nothing is left but the reflected 
distortional wave. 
Bearing in mind these broad facts regarding the behaviour of 
waves at surfaces separating two elastic media, and especially the 
fact that, in general, each type of wave arriving at such a surface 
brings into existence the other type as well, we have little difficulty 
in understanding how an earthquake disturbance may be drawn 
out in time as the various vibrations, started directly by it or 
