136 F. GOLDSTONE 
the time-distance graph; the first wave front to arrive is still the longi- 
tudinal wave propagated in V;, but now the longitudinal reflected 
wave and the transverse wave arrive simultaneously. At any point 
more remote than C from the origin of the disturbance, the reflected 
longitudinal wave arrives ahead of any transverse wave. 
Applying these considerations to a practical case it is obvious that 
reflected events always appear on a seismogram subsequent to the 
arrival of at least one other wave train, and we must therefore have 
instruments so designed that this first impetus will be suppressed suf- 
ficiently to permit the recognition of a subsequent burst of energy even 
though the latter be of much less intensity. It is possible to locate a 
seismograph sufficiently far from the origin of disturbance to insure 
that the reflected longitudinal wave from a particular depth will be 
the second event to occur on the seismogram. For instance under the 
conditions of Figure 1 this distance will be about 2,300 feet. When in- 
vestigating areas where reflexions are obtained from great depths it 
is frequently not desirable to use such large intervals between explosion 
points and recording station as would be imposed by this condition, 
and we consequently must be prepared to identify reflected events 
which arrive simultaneously with or subsequent -to the transverse 
wave propagated in the upper stratum and the Rayleigh wave travel- 
ling along the surface. I might mention that the velocity of propaga- 
tion of these two waves differs by very little and we are not able to 
differentiate between them; however, when a hard formation exists 
at the surface a wave having a velocity about in agreement with what 
might be expected from either of these forms is nearly always present 
and is extraordinarily persistent, the rate of decay in amplitude with 
distance being rather small. This suggests that propagation is taking 
place in two dimensions rather than in three dimensions and would 
bias one in favour of ascribing this disturbance to a Rayleigh wave. 
These waves constitute the phenomenon usually referred to as “ground 
roll.” It was on the stumbling-block of eliminating these extraneous 
wave trains that early efforts to apply this method to economic prob- 
lems failed. 
Experience has pointed the way to elimination of most of these 
unwanted wave trains. The “ground roll” has a frequency of about 20 
cycles per second, which is much removed from that usually observed 
in longitudinal waves and can be eliminated when electrical seismo- 
graphs are used by means of a wave filter or by the use of transformers 
in the amplifier circuit, which are very inefficient in amplifying the 
frequency band in which these waves fall; alternatively, a seismograph 
can be designed having a natural vibration frequency which is identical 
796 
