Mud in marine sediments is a mixture of clay, sand, and silt particles. Present paper follows up on a suggestion by Holland and Dosso (JASA, 2013) that the variability of the measured frequency-dependent compressional wave attenuation may be caused by the variability of the amounts of sand and silt particles. The premise is that the porosity for the mud is high and that the sand and silt particles are in suspension. They do not settle out to the bottom of the layer because the card-house fabric of the clay particles tends to hold them in place. This supposition leads to a theory where the clay configuration gives a base-line attenuation, and the contribution from the individual sand and silt particles is additive. The estimation of the latter is distinguished from the existing theories of attenuation of sound in sandy/silty sediments in that the particles are presumed not to touch each other. Particles are assumed to be spherical and there is no slip between particle surfaces and the surrounding water. Earlier theories of attenuation in suspensions by Lamb (Hydrodynamics), Urick (JASA. 1948), and others are criticized because of their assumption that vorticity in fluid is zero. The present theory predicts that the attenuation contribution from a given category of particle is proportional to the square of the frequency and to square of particle diameter, and inversely proportional to the viscosity, in the limit of low frequencies. It approaches a frequency-independent constant at higher frequencies. [Work supported by ONR.]