580 Proceedings of Royal Society of Edinburgh. [sess. 
not quite correspond with the numbering in the complete seismo- 
gram. In considering the possible significance of this apparent 
correspondence, we must remember that the succession of crests 
and troughs in the seismogram does not represent the vibration 
itself, but is due to a fluctuation in the intensity of the vibrations. 
For example, the well-marked double crest 6-7 is no doubt 
composed of a great many individual vibrations. This fluctuating 
intensity has long been known to he characteristic of earthquake 
motions, and has its origin in the complexity of the original disturb- 
ance. We see, then, that a definite fluctuating character belonging 
to one part of a drawn-out disturbance is almost exactly reproduced 
Fig. 2. 
in another part. Two explanations of this may be given. We may 
either suppose these similar groups of disturbances to be brought 
into existence simultaneously at the earthquake origin but to travel 
at different speeds or by different paths through the earth ; or we 
may suppose the later disturbance to be a reflection of the earlier 
at some distant barrier. 
If this - repeating ’ character of certain seismograms indicate a 
reverberation within the earth, then we must postulate a compara- 
tive uniformity of structure throughout the greater part of our 
globe, and a fairly abrupt boundary or transition surface. A train 
of waves of either type impinging on such a surface will, in general, 
give rise to trains of reflected waves of both types, and trains of 
refracted waves also of both types. The greater the differences in 
the elastic constants and densities of the two media separated by 
