In general, these marigrams are characterized by large and sometimes 

 pseudo-regular, sometimes highly irregular oscillations, announcing the 

 arrival of the tsunami. The extraordinary variety of effects recorded 

 is eloquent testimony to the profound changes that a tsunami is subject 

 to when propagating from deep water to the shallow-water locations of 

 most gaging stations. One logically wonders what aspects there are in 

 common, if any, between records such as those, say, of Prince Rupert, 

 Canada (Figure U6a), San Francisco, California (Figure 50a), and Hilo, 

 Hawaii (Figure 59a). The resemblances appear to be small, and possibly 

 they might be explained away on the grounds that these three receiving 

 stations are very far apart. However, when comparison is made between 

 Figures 58a and 59a for two stations close together in the Hawaiian 

 Islands, the differences are not so easily disregarded. 



Wave energy spectra might be expected to reveal certain fundamental 

 resemblances between marigrams, but except in the case of the records for 

 Honolulu and Hilo, Hawaii (Loomis, 1966), few comparisons of wave spectra 

 for the Alaskan tsunami have come to our attention. Figures 6Ta and 67b 

 present the results of Loomis' analyses, but the frequency resolution at 

 low frequencies is rather poor so that the existence of any waves having 

 periods of the order of two hours is not revealed. The best that can be 

 said of the two spectra for the early stages of the tsunami (O to 8.3 

 hours) is that periods around 33.3, I8.O to 20.8, and 12.1 to 13.0 

 minutes are prominent in both records and that the, dominance of the 

 periods changes with time at the two places and becomes different in 

 both. 



Spectral analysis, however, has the limitation of requiring a 

 specified time sequence of record for its elaboration, and that in that 

 time sequence changes in the dispersive wave system are continually taking 

 place. It can therefore never give a completely true picture of the com- 

 position of a tsunami at any given time, even though in the very nature 

 of an impulsive wave system the particular wave dominant at any time is 

 actually the resultant of a spectrimi of frequencies (cf. Wilson, et al, 

 1962; Van Dorn, I965). The computed spectrum can reveal a prominent, 

 often broad, frequency band, but it cannot usually identify whether, or 

 how, frequencies in the band may be interfering to form beats. Moreover, 

 in the range of frequencies covered in Figure 67, the obvious prominences 

 are likely to be characteristic of the oscillating modes of the locality 

 rather than the fundamental natiore of the tsunami (cf. Munk, et al, 1959). 



To discover more about the tsunami itself, the first author (Wilson) 

 has applied the same kind of subjective analysis employed in Figure 38 to 

 the marigrams of Figures U3 to 66. This process is a matter of judgment 

 and experience; perhaps something of an art. At present (1967), a purely 

 objective mathematical treatment of the data to achieve comparable results 

 has not been developed. The subjective analyses given in Figures U3 to 66 

 are subject to the human errors of interpretation and of drafting, but the 

 overall consistency of results in identifying characteristic signatures 

 of the tsunami is most encouraging. 



Text resumes on page 91 

 63 



