Propagation of ultrasound through a porous body generates an electric signal, similarly to the well-known electroacoustic effect in dispersions of mobile particles. This obscure version of electroacoustics has been known since 1948, when M. Williams published his paper on electrokinetic transducers [M. Williams, Rev. Sci. Instrum. 19 (10) (1948) 640-646]. We observe this effect in a 1 wt% aqueous dispersion of carbon nanotubes. Magnitude and phase of the electroacoustic signal, as well as conductivity, are sensitive to sonication and mixing. Sonication with no mixing leads to phase rotation by up to 180 degrees comparing to the traditional colloid vibration current (CVI) in sols. This is explained by the fact that sonication terminates motion of the carbon nanotubes by building up a continuous network gel. Propagation of ultrasound through the immobile carbon nanotube network generates a streaming vibration current (SVI), but not a CVI, which requires free motion of the particles relative to the liquid. Theoretical analysis indicates that the SVI has 180 degrees difference in phase from the CVI. The magnitude of the SVI after sonication with no mixing depends on the shifts of the measuring probe position. Apparently this occurs due to inhomogeneity of the carbon nanotube get, which might have clusters with higher density and gaps with no solids at all. This effect can be used for testing homogeneity of the carbon nanotube gel. Sonication with continuous mixing also affects the electroacoustic signal and conductivity. However, the electroacoustic phase does not reach 360 degrees, which corresponds to the SVI in gel. The measured signal is the vector sum of the CVI and SVI under these conditions. It is possible to use data on the electroacoustic phase to monitoring the number of carbon nanotube segments that retain independent motion. (c) 2007 Elsevier Inc. All rights reserved.