1494 
Figures 2d and 2e illustrate this increasing com- 
plexity at very long ranges. (One theoretical curve in 
Figure 2d and the theoretical curve in Figure 2e are plotted 
on the assumption that all orders of reflection take place 
at glancing incidence, without entering into the detailed 
calculations required to account for the actually expected 
phase shifts.) It is evident that even in Figure 2e, where 
many processes have probably participated in changing the 
character of the wave, there is still a general correspondence 
between the principal features of the experimental and theor- 
etical curves. In the experimental record, the primary shock 
wave shows evidence of having been cut down by the finite 
amplitude effects discussed by Keil; viscous attenuation and 
gauge size have also played their parts, and as a result the 
initial portion is very much lower in amplitude than the linear 
theory predicts. The highér order reflections, however, are 
of the expected amplitude and show the gradually increasing 
amplitude of oscillation predicted by the theory. 
The marked rounding and smoothing of the wave which 
is apparent in the experimental record in Figure 2e is too 
strong to be ascribed entirely to viscous attenuations and 
gauge size effects. At least part of it must probably be 
ascribed to absorption or inelastic losses in the various re- 
flections, etc. 
= T8re2 
