26 
PACIFIC SCIENCE, January, 1947 
harbor bottoms for distances up to 500 feet 
or more from the normal strand line. The 
outflow of the water was rapid and turbu¬ 
lent, making a loud hissing, roaring, and 
rattling noise. At several places houses were 
carried out to sea, and in some areas even 
large rocks and blocks of concrete were car¬ 
ried out onto the reefs. Sand beaches were 
strongly eroded by the outgoing water. 
People and their belongings were swept to 
sea, some being rescued hours later by boats 
and life rafts dropped from planes. 
At a few places, generally but not exclu¬ 
sively on the sides of the islands away from 
the wave origin, the first wave was reported 
to have been the highest. At those places, 
the rise was generally of the quiet sort. 
There are, however, no instrumental records 
showing the first wave to have been the high¬ 
est, and it is possible that at places reporting 
the first wave as the highest, earlier waves 
may have been overlooked. Much more gen¬ 
erally the third or fourth wave was reported 
to have been the highest and most violent. 
The third crest was the largest at the Hono¬ 
lulu tide gage (Fig. 2). At other localities 
the sixth, seventh, or eighth waves were said 
to have been the highest. At Waimea River, 
Kauai, the sixth crest was higher than any 
other, both in absolute level and in its height 
above the preceding and succeeding troughs. 
In general, if not everywhere, the size and 
violence of the waves increased to a maxi¬ 
mum with the third to eighth waves. The 
oscillations then gradually decreased in am¬ 
plitude over a period of at least 2 days, but 
with occasional waves which were larger 
than those just before and after them. Such 
temporary increases in wave height prob¬ 
ably resulted from mutual reinforcement by 
the essentially simultaneous arrival, in phase, 
of waves which had traveled different paths, 
or from the coincidence of tsunami waves 
with storm waves or seiche oscillations. 
Measures of the height of the waves ap¬ 
proaching shore in shallow water, but before 
they dashed up on shore, are poor. At 
Kawela Bay, Oahu, Shepard estimated the 
height of the waves advancing across the reef 
to have been as much as 18 feet, and observ¬ 
ers estimated the height of the waves cross¬ 
ing the reef off Lanikai, on Oahu, to have 
been about 7 feet. Photographs taken at Hilo 
show the top of the breakers to have been 2 5 
feet above the normal bay surface where they 
struck Cocoanut Island, but the waves may 
have increased considerably in height in 
crossing the breakwater, and the effect of 
dashing up on the shore was probably already 
present, further exaggerating the height. 
Photographs of some of the late waves at the 
mouth of the Wailuku River, in Hilo, show 
them to have been 6 to 8 feet high (Plate 
8), and early waves undoubtedly were 
higher. In general, these heights correspond 
fairly closely with the measured heights to 
which the water dashed on the shore at those 
localities. At any rate it appears clear that 
the waves not only slowed down, but in¬ 
creased in height on entering shallow water. 
George Green (1838: 457-462) states that 
the wave height varies inversely as the fourth 
root of the depth of the water. 
Most observers reported the first move¬ 
ment on Hawaiian shores to have been a 
withdrawal of the water. However, the only 
available instrumental records, at Honolulu 
and Waimea, both indicate the first move¬ 
ment to have been a rise. The instrumental 
records are probably more reliable than the 
reports of untrained observers. The initial 
rise at Honolulu was small (Fig. 2), and a 
similar small rise at other localities may eas¬ 
ily have been overlooked. Certainly it would 
have been less impressive than the large 
withdrawal of the water from shore as the 
succeeding trough approached. It is inter¬ 
esting to note that the records of tide gages 
along the coasts of North and South America 
obtained by C. K. Green (1946: 497) all 
show the initial movement to have been a 
