This paper reports the present stage in our research programme. We have used two very different models : (i) A numerical model treating the avalanche as a Newtonian liquid with a free surface; this model describes qualitatively the velocity fluctuations both of avalanche particles and of the front and gives some quantitative predictions, but much experimentation is needed to determine the large number of parameters involved. (ii) An analogue model in which powder avalanches have been simulated in an underground tunnel, taking account of the densimetric Froude number; this model reproduces the pressure rise which precedes the visible front, and also the pressure wave as the front passes which causes so much damage to engineering works—and human beings. These models have been applied to two cases—a bridge and a snow shed.

The interplay between models and experimentation is discussed and the need for improved data on the mass output of an avalanche, of its detailed velocity history, and of the force on large objects is emphasized. Methods being used to try to supply this information are discussed ; gamma-ray measurements for specific mass, high-speed stereophotogrammetry for velocity, and various pressure sensors for the force. The possibility of releasing real powder avalanches to enable such measurements to be made more reliably is being investigated.