186 
PACIFIC SCIENCE, Vol. I, July, 1947 
the epicenter and the entrance to Honolulu 
Harbor was plotted on a chart of the Pacific 
Ocean. The path was transferred to larger 
scale nautical charts and then divided into 
sections of 120 nautical miles each, except 
when rapid changes in depth required sec¬ 
tions of shorter length. The average depth 
in each section was taken and the time re¬ 
quired to pass over it was computed as 
follows: 
Let d represent the mean depth, in fath¬ 
oms, of the 120-mile section and t the travel 
time of the sea wave over that section. 
Since v = V gd = 8.23 V d nautical miles 
per hour* /= 120/8.23 V^=l4.58/V<7 
hours. 
The times thus computed were added 
cumulatively, increasing with distance from 
Honolulu, and the y>-hour points were de¬ 
termined by interpolation. 
The dividing of a path into small sections 
increases the precision of the determinations 
over those which are obtained by using a 
mean depth over a whole distance. The 120- 
mile points were usually 0.25 to 0.35 hour 
apart, thus allowing for reliable interpola¬ 
tion of 1^-hour points which would have 
been impossible were the path to be consid¬ 
ered as a whole. 
When a great circle course for a seismic 
sea wave is first laid off on large-scale 
charts, there are several details to be con¬ 
sidered. If the path crosses a large unit of 
land, the portion of the wave front which 
reaches Honolulu will have to go around 
that land. Because the part of the wave 
front in deeper water will advance more 
rapidly, the path is considered the combina¬ 
tion of arcs of two great circles joined in 
the deep water off the coast. If the path of 
a wave involves crossing a large section of 
shoal water, then the time of the wave front 
which diverges somewhat from the great 
circle course, but which travels over a deeper 
course, must be considered. An excellent 
example of the latter was found in the com¬ 
putation of the travel time of the wave from 
the Aleutian Trench on April 1, 1946, to 
Sitka, Alaska. The travel time along a great 
circle course, through shoal water most of 
the way, was computed as about 7 hours. 
By considering a path going to the south¬ 
east for about 90 miles and then moving 
along a great circle course from that point to 
Sitka, the time was calculated to be about 
3 hours, which was in almost perfect agree¬ 
ment with the observed time. 
The latter example is an extreme case. 
More frequently there was found a condi¬ 
tion in which part of a great circle course 
covered an area less deep than that covered 
by an adjacent path. Despite the additional 
distance covered in diverging from a great 
circle course to get into deeper water, the 
total travel time may be less than that along 
the original great circle path. Shallow areas 
of this type caused irregularities in the time 
curves and necessitated the computation of 
additional paths and the consideration of the 
bathymetric pattern. 
The problem of how to treat the compu¬ 
tation of the travel time of a wave whose 
path lies in a deep channel with compara¬ 
tively shallow water on both sides is a 
troublesome one. If the formula were fol¬ 
lowed rigorously in such a case, the front of 
the wave would gradually have to become 
more and more pointed as the wave moved 
forward, the shallow water on either side 
slowing down that portion of the wave front 
passing over it. The concept of such a 
pointed wave front did not seem reasonable, 
and the velocity was computed as just slightly 
faster than that over the shallow area. The 
observed travel times from several epicenters 
lying in deep channels were found to com¬ 
pare favorably with the times thus computed. 
Some of the epicenters used in comparing 
observed with computed travel times were 
found to plot on land near ocean deeps. 
There have been some differences of opinion 
among seismologists whether the true epi- 
