Acceleration data from four air-burst tests of Operation Tumbler-Snapper® showan analogous relationship expressed by A = 0.32p"-®? (1.3) Differences in svil conditions at the Pacific Proving Grounds (Eq, 1.2) and Nevada Test Site (Eq. 1.3) are probably the primary cause of the differences in coefficients and exponents in the two equations. , The acceleration-time curves (Figs. A.1 to A.11) are of a complex periodic form. All show some distinct frequencies and apparent interference effects which irfluence the curve. Frequencies derived directly from the curves are included in Table 1.2. Frequencies enclosed in parentheses are those of apparently secondary importance. The ground-transmitted motion is characterized by two major frequency ranges: one of about 50 cps at Station 650.01 which, according to Fig. 1.1, involves principally transmission through the shallower materials characterized by a seismic velocity of 6300 ft/sec, and the other of from 2 to 4 cps which involves an apprectable proportion of travel over a refraction path within the 18,000 ft/sec basementrock, These frequencies are consistent with those observed in seismic exploration. The frequency of the air-shock induced motion is generally similar at all stations, ranging from 20 to 70 cps except for the higher ones in Station 603 measurements which probably result from response of structural elements. , Directions of the various components of motion are consistent. The initial vertical acceleration pulse is upward for the ground-transmitted signal and downwardfor the air-shock induced one at all stations. Initial radial pulse is outward from Ground Zero for all signals from both sources with the doubtful exception of the ground-transmitted signal at Station 650.03. Initial tangential pulses are less consistent; the pulse from the ground-transmitted acceleration is clockwise for the ground stations, 650.01, 650.02, and 650.03, but is reversed for the shelter station, 603, and a similar reversal occurs in the air-shock induced signal, which is counterclockwise at all stations except 603. 1.5.2 Velocities Velocity-time information was derived by integrationof data from.each measured acceleration component. A detailed description of the integration process is included in Appendix B. Integrations over time intervals of the order of 5 sec or longer were required even for strong motion portions of the ground-transmitted accelerations. Integration over periods of such length magnifies excessively the influence of very small low-frequency drifts in the primary recorded data. This effect becomes more serious where signal-to-noise ratio is low, even though the noise componentof the recorded data may be erased by the integration. The magnification is, furthermore, strongly enhanced when a second integration of the data is performed. The significance of small long-duration extraneous drifts in the primary data to the results of integration is evident if it is realized that an additive correction in acceleration data appears as a linear increase in velocity and as a parabolic increase in displacement. For ex- ample, slow sinusoidal or linear changes of the order of a fraction of 1 per cent of carrier voltage can, for long-duration integrations, introduce effects that distort the velocity curve badly and obscure small but real! displacements. Complete correction of data for integration was frequently neither feasible nor possible in this analysis, and all corrections involved some degree of arbitrariness. Consequently, in reviewing results of the first integration of the acceleration-time data, unrealistic or improbable results were often eliminated, and it was always necessary to recognize that precision had been lowered by the integration. Velocity-time curves for each accelerometer station are included in Figs. A.1 to A.11. These represent the corrected velocity data from which displacements were derived. Data from these velocity curves are compiled in Table 1.3. Maximum velocities enclosed in parentheses represent peak-to-peak values of the higher frequency signals which are reasonably independent of the spurious values introduced by instrument drift. Computed peak velocities and low-frequency components are in some cases introduced by extraneous sources. 24