The effect of hydrophone arrays in the recording of seismic signals during offshore Texas seismic marine experiments is judged by comparing traces of spatially tapered hydrophone array signals with traces that are combinations of simultaneously recorded wavetest hydrophone signals. Each spatially tapered hydrophone group array consists of 26 hydrophones nonuniformly spaced over 212 ft. The wavetest streamer section consists of 36 groups of two hydrophones, each pair connected in parallel and with hydrophones back-to-back for acceleration cancellation, with 5-ft spacing between groups.Reflections from deep subsurface interfaces are negligibly affected by hydrophone arrays except for very long arrays and/or long-range distances. Consequently, the report is primarily concerned with the effects of simulated and real hydrophone arrays on first-arrival signal and early subbottom reflections.Comparison of theoretical and actual seismic traces from an Aquapulse source for near range distances (835 ft) used in normal operations indicates that (1) near-simultaneous arrival of the direct wave and surface reflection result in their virtual cancellation, (2) the early event with largest amplitude is associated with constructive interference between source and receiver ghost reflections, and (3) the 'pseudobubble' period effectively fixed the predominant frequency of all seismic events at values near 28 Hz. At medium range distances (4755 ft), such comparisons indicate that (1) first arrivals are refracted waves traveling in subbottom layers; (2) the water-bottom reflection is beyond critical angle and is, therefore, complex; (3) the early events with largest amplitude are multiple reflections; and (4) at least two orders of water-bottom multiples are identified.The attenuation of the high-amplitude, first-arrival signal that includes the water-bottom reflection permits greater dynamic range in field recording and higher levels of 'true' amplitude for later reflections without overload distortion of early events on playback. However, if improved resolution of reflection from moderate depths ( approximately 4000 ft) is important, then arrays of length studied in this report ( approximately 200 ft) should not be used to record signals at range distances greater than about 2000 ft because frequencies above 50 Hz are attenuated severely.Spectral analysis of wavetest records in the absence of signals shows that the wavenumber distribution of the noise is located along a slope line equivalent to 5000 ft/sec between wavenumbers that imply a spectral distribution of 30 to 100 Hz. Theoretical array response studies show that both the 36-element Chebyshev array and the 26-element spatially tapered array are superior to a 36-element uniformly weighted array in rejection of seismic noise in the spectral range of 30 to 100 Hz.