Erosion of snow by pyroclastic flows and surges presumably involves mechanical scour, but there may be thermally driven phenomena involved as well. To investigate this possibility, layers of hot (up to 400??C), uniformly sized, fine- to medium-grained sand were emplaced vertically onto finely shaved ice ('snow'); thus there was no relative shear motion between sand and snow and no purely mechanical scour. In some cases large vapor bubbles, commonly more than 10 mm across, rose through the sand layer, burst at the surface, and caused complete convective overturn of the sand, which then scoured and mixed with snow and transformed into a slurry. In other cases no bubbling occurred and the sand passively melted its way downward into the snow as a wetting front moved upward into the sand. A continuum of behaviors between these two cases was observed. Vigorous bubbling and convection were generally favored by high temperature, small grain size, and small layer thickness. A physically based theory of heat- and mass transfer at the pyroclast/snow interface, developed in Part 1 of this paper, does a good job of explaining the observations as a manifestation of unstable vapor-driven fluidization. The theory, when extrapolated to the behavior of actual, poorly sorted pyroclastic flow sediments, leads to the prediction that the observed 'thermal-scour' phenomenon should also occur for many real pyroclastic flows passing over snow. 'Thermal scour' is therefore likely to be involved in the generation of lahars.