62 MODERN SEISMOLOGY 
time curve itself and not on the still less accurate curve ex- 
pressing 4 as a function of slope dT /d4. Different investiga- 
tors give smoothed time curves which differ sufficiently to lead 
to very different conclusions as to the interior of the Earth. 
Moreover, we have seen that a smoothed curve may really 
involve a quite wrong method of procedure. 
The primary curve itself is subject to many sources of 
error. Apart from actual errors of the time that do unfort- 
unately exist at seismological stations, we have to remember 
that the marking of the exact instants at which P and S occur 
is a matter of personal judgment, and depends also on the 
particular instrument used and the sharpness of the impulses. 
The first portion of the curve depends on the elimination 
of the effect of finite depth of the focus, and as that is a very 
difficult matter, I should doubt if it is often successfully accom- 
plished. Again for distances much beyond 10,000 km. S is 
often extremely indistinct. There are probable theoretical 
reasons for this as we have pointed out, but meanwhile it 
introduces uncertainty. Beyond 13,000 km. data are very 
meagre, and the determination of the incidence of P becomes 
increasingly difficult on account of the smallness of the hori- 
zontal movement. 
Thus there is room for progress both on the theoretical 
and the experimental side, but the growing activity of seis- 
mologists is a good augury for the successful improvement of 
time curves even to the semicircumference of the Earth. 
