I suggest that radio-wave scattering by the interstellar plasma, in combina
tion with subsonic gradients in the Doppler velocity of interstellar H I, i
s responsible for the observed small-scale variation in H I absorption spec
tra of pulsars. Velocity gradients on the order of 0.05-0.3 km s(-1) across
1 AU can produce the observed variations. I suggest observational tests to
determine the relative importance of such gradients and AU-scale absorbing
cloudlets in producing the spectral variations.