In regard to the attenuation of sound in marine sediments, Holmes, Carey, Dediu, and Siegmann [JASA Express Letters, 2007] report a nonlinear frequency dependence with frequency raised to a power n=1.8 (plus or minus 0.2); earlier literature gives results with non-integer values of n that are between 1 and 2. The present paper argues that the exponent should be exactly 2 in the limit of low frequencies. Plane wave propagation in a general medium allows specific modes for which wavenumber k is a given function of frequency. Attention is given to those propagating modes where the real part of k is proportional to frequency at low frequencies, so that the phase velocity approaches a finite value as the frequency goes to zero. General causality (all the past determines future) requires the exponent n to be greater than unity. A restricted type of causality, where a knowledge of only the present suffices (including a knowledge of possibly many hidden variables, corresponding to relaxation processes), with governing equations being linear partial differential equations with constant coefficients, leads to the requirement that the imaginary part of k must have a low-frequency dependence that is as the square of frequency or weaker.

This content is only available via PDF.