with distance and direction were also noted. Most recordings on standard equipment also showed definite evidence of at least a portion of the dispersive train for the four largest shots although the amplitudes were greatly reduced by lack of low-frequency response. Antipodes and second direct arrivals on VLF equipment also showed marked evidence of the dispersive train in cases of high signal-to-noise ratio. Horizontal-phase velocities were slightly lower than the normal velecity of sound at ground level (about 335 m/sec) and were nearly 2qual to the travel speeds for firs. arrivals at the same locations. Theoretical stucies predicted phase velocities equal ‘o ti speed of sound at ground-level, i.e., vertical, wave fronts. Horizontai-phase veiocities obtained from standard equipment at stations where the microphone spacing was, in general, small compared to the wave length of the acoustic signal showed a considerable range of values. However, practically every first-wave signal gave phase velocities covering some portion of the range from 318 to 360 m/sec. Signal amplitudes received were approximately as expected. A detailed study of the amplitudes recorded by VLF equipment was undertaken. Detectable signals for direct-wave arrivals on standard equipment persisted for a minimum of 8 minutes and a maximum of 369 minutes, the average being 74. Antipodes and later urrivals persisted for a minimum of 3, a maximum of 530, and an average of 140 minutes. For VLF equipment, the direct-wave signals persisted for a minimum of 9, a maximum of 240, and an average of 79 minutes. Antipodes and later arrivals gave a minimum of 83, a maximum of 339, and an average of 192 minutes. In general, signals from the megaton shots Started with an increase of pressure, foilowed by a larger negative pulse. The first measurable periods generally ranged from 200 to 450 seconds and were followed by decreasing periods at later time, at least for the first 30 minutes. Short-period arrivals characteristic of waves trapped by tempera- ture and wind gradients in the first few thousand feet of the atmosphere were observed at the beginning of some recordings at stations within 5,000 km of the explosion. Such waves had occasionally been observed at stations within 1,000 km of previous U. S. nuclear detonations, but never at such long ranges. Periods in these arrivals were of the order of 3 to 5 seconds and persisted for as long as 5 minutes. The characteristics of acoustic signals from the Castle detonations were similar to those observed for previous tests. All megatcn shots showed dispersive waves while the kiloton shot did not; horizontal-phase velocities showed considerable spread, but covered the same range of values previously observed. Amplitudes rangedfrom a tenth to several hundred dynes per square centimeter, depending on the equipment, yield of the shot, distance from source, and noise level. Signals persisted for a very-long time, and signal periods spread over more than 8 octaves, from 3 to 450 seconds. Castle data definitely proved that dispersive waves may be generated by shots having a yield as low as 1.7 Mt. These dispersive waves seemed to be modified by the atmos- pheric structure along the path from the source to the atation. 7.2.3 Travel Speeds. Travel speeds recorded by standard equipment were generally within a few meters per second of each other at all stations; however, there was a general trend shown toward decreasing speeds eastward and increasing speeds westward as the Castle series progressed from 28 February to 13 May. The average travel speed for first arrivals from the direct wave on VLF equipment ranged somewhat higher than speeds obtained from standard recordings. These higher speeds were due to the earlier arrival of the long period dispersive train recorded on VLF equipment. Greatest travei speeds were normally observed for the long-period dispersive waves, 96