The seismograms of January 17, 1922, are the superimposed records of three distinct South American earthquakes: (1) in Brazil, O = 3h 50m 20s, M. G. T., ϕ = 4° 35.2′ ± 0.5′ South λ = 63° 56.3′ ± 0.4′ West; (2) in Venezuela, O = 3h 50m 22s, M. G. T., ϕ = 5° 11.5′ ± 5′ North, λ = 66° 45.2′ ± 4′ West; (3) near Ecuador, O = 3h 50m 24s, M. G. T., ϕ = 3° 12.1′ ± 3′ South, λ = 82° 50.2′ ± 3′ West.
The great amplitudes of reflected waves in comparison with the directly transmitted waves at the critical distances would appear to uphold the theory of discontinuous layers in the earth.
The period of the wave of maximum amplitude in the second preliminary group has no functional relation to the distance travelled by the wave. There is no dispersion of frequencies.
Taking the four maxima separately there is no one marked period preference which is shown by all of them. But taking them together there is a preference for periods of six, seven, and eight seconds.
In the earthquakes of January 17, 1922, in Venezuela and in Ecuador, the velocity of the maximum exceeded that of the group. For the California earthquake of January 31, 1922, the velocity of the maximum was less than that of the group. The position of the maximum in the group is a function of the epicentral distance, but not the same function for all maxima in all earthquakes.
The writer wishes to express his thanks to Professors Elmer E. Hall and James B. Macelwane. Their advice has made this investigation possible.