SECT. 5] 



LONG OCEAN WAVES 



661 



or lower on the incident and lee shore, respectively, as compared to the un- 

 disturbed wave height. For smaller islands the distortion is even less. Largely 

 as a result of Van Dorn's and Donn's efforts, two dozen long-period wave 

 records were in operation during the International Geophysical Year, most of 

 these on islands. The station at Wake Island obtained a striking record of the 

 tsunami of 9 March, 1957. During the first two hours the frequency increased 

 wdtli time in accordance with the expected role of dispersion. The maximum 

 wave height was 1 ft. In contrast, at Hawaii (with roughly the same epicentral 

 distance) the height reached 15 ft. The spectral similarity between the tsunami 

 of 9 March, 1957, from an earthquake, with tsunamis from atomic tests pre- 

 viously recorded at Wake Island permitted the authors to estimate the total 



0.5 1.0 



Cycles per kilosecond 



Fig. 9. Spectra of background activity and of the tsunami of 28 July, 1957, at Acapulco. 



energy of the wave train: 10^2 ergs. The seismic energy had been estimated at 

 1023 ergs. 



Following the devastating tsunami of 1 April, 1946, a tsunami warning 

 system has been organized by the U.S. Coast Geodetic Survey. The basic 

 difficulty is that it has not been possible from the seismic record of an earth- 

 quake to determine whether or not a tsunami has been generated, and recourse 

 has to be made to the inspection of tide records near the epicenter. The observed 

 fact is that two earthquakes which are seismically undistinguishable with 

 regard to location and energy can generate tsunamis whose amplitudes differ 

 by an order of magnitude! Ewing, Tolstoy and Press (1950) have noted a 

 correlation between the T phase on seismograms and the occurrence of tsunamis, 

 and have suggested that this might be useful in removing some of the 



