Abstract
In the course of neutron scattering measurements on sodium near the martensitic transition at 36K, large single crystals were subjected to transformation cycles. On three occasions the original BCC orientation completely disappeared when the partially transformed crystal went through the reverse transition, and large grains (in one case a new excellent single crystal) with different orientations appeared. Analysis of relative orientations before and after a transformation cycle revealed no relationship akin to twinning. Reflections from the HCP phase indicated that it was very poorly developed. No significant changes in phonons in the BCC phase could be detected, neither above the transition point nor below it, apart from a large reduction in intensity in the partially transformed crystal. An explanation of this effect is suggested.