HITHERTO it appears that when size distributions of tobacco smoke droplets have been made by light or electron microscopy, the droplets have been caught on glass surfaces or one of the usual kinds of specimen support film, and upon deposition they have spread on the sampling surface (Fig. 1a and c). As a result, published size distributions obtained by these methods have been of the spread diameter of the droplets instead of their true diameter as they existed in the smoke at the time of sampling.