with respect to the minor axis and intersecting the coast. The nodes in 

 this case would approximate to radial lines from the center of the major 

 axis and the type of oscillation would effectively constitute edge waves. 



The fifth mode yielding a period T^ = 31.7 minutes is actually a 

 binodal oscillation of a different type which Involves a semi-ellipse as 

 the second node (see Figure 73 c and d). Since this mode would conform 

 well to the coastal configuration, it would seem to be the resonating 

 medium for stimulating the third harmonic of the. tsunami excitation. 

 None of the other modes of shelf oscillation had periods in conformity 

 with the tsunami period and its odd harmonics. The third mode (T2 = 52.1 

 minutes) would have been in good accord with the second harmonic of the 

 tsunami, but the implication of Figure U9 is that it was not excited. 



However, the fundamental eigenperiod for the shelf off Crescent City 

 (T = 79 minutes) is sufficiently close to the main tsunami wave period 

 (T - 108 minutes) to have provided a degree of amplification through par- 

 tial resonance, as seems to have been the case more completely, however, 

 at Port Alberni , Canada, and Lyttelton, New Zealand. 



The tsunami of May 23, I960, generated by the Chilean earthquake 

 (M = 8.U), also drew a strong response at Crescent City at a period of 

 32 minutes. Figure 7^, reproduced from Wiegel (1965), shows the response 

 and presents the energy spectrum for the waves calculated from the tide 

 record. Wiegel, seeking an explanation for the 32-minute peak as a pos- 

 sible shelf oscillation, investigated wave travel times from the edge of 

 the Continental Shelf, but found this approach unrewarding. The travel 

 time over the distance L of Figure 73b was about 21.2 minutes. Four 

 times this value would yield the approximate period of the fundamental 

 shelf oscillation, T-, = 8U.8 minutes, which compares favorably with our 

 result of T-[_ = 79.1 minutes. This period, according to Takahasi's data 

 in Figure 71, should have favored resonance of the main waves propagated 

 from the Chilean earthquake; Figure 7^, however, does not show the reso- 

 nance, although the accuracy of the spectrum at very low frequencies is 

 suspect. It should be noted that the harbor at Crescent City cannot 

 support resonance of tsunamis at periods greater than about 10 minutes. 



We conclude then that Crescent City's susceptibility to large 

 wave response from major tsunamis is, by its very name, related to its 

 crescent-shaped coast and bowl-shaped Continental Shelf. Because of 

 its dimensions, it will forever be a responsive echo-chamber for great 

 tsunamis since their periods will be always capable of exciting full 

 or partial resonances. 



We mentioned previously that the mechanism for transfer of energy 

 of long waves to higher frequencies when negotiating depth changes that 

 are sudden with respect to the wave length, is not yet fully understood. 

 Dean (196^1) may have uncovered the main elements of such changes by draw- 

 ing attention to the fact that the wave form, if resolved into its Fourier 

 constituents, would yield different amplifications and phase changes for 

 the components in their propagation over the continental slope. Long 



III 



