Estimation of atmospheric dispersion is one of the principal objectives of meteorological studies conducted at reactor sites, and it is often the most difficult one to achieve. methods ordinarily used involve either airborne tracer experiments or else inference from existing knowledge of atmospheric dispersion. Tracer experiments are, naturally, much to be preferred; bust since their use requires elaborate experimental arrangements, they have been conducted for the most part only at the larger nuclear complex. Observations of smoke plumes have frequently been used effectively as a qualitative substitute; these do give considerable insight into local diffusion patterns at reactor sites.