388 MACHTA bility. Now, in addition to the intensity of mixing, the slope of the principal mixing surface also must be specified, This introduces another degree of freedom in specifying the transport processes, Reed*® has offered a method for determining the slope. The second approach is to allow a purely meteorological generalcirculation model to generate, insofar as possible, the turbulent mixing and organized circulations. The current model of the Geophysical Fluid Dynamics Laboratory of the Weather Bureau is such an example. It is expected that the radiotungsten will be inserted in the stratosphere at its source in this model and then transported by the turbulent and organized circulations. In this way, the likelihood that the air motions generated by the computer approximate reality may be tested. If they do transport the tracer to a realistic distribution and if they subsequently provide good patterns for other substances like ozone, cosmic- ray radionuclides, and “C, an increasing confidence in the reality of the independently determined transport processes and their magnitudes may be gained. So far it cannot be said that the prediction of any of the meteorological models is preferred over that obtained from empirical meth- ods. However, it is likely that there will be such a preference ina matter of a few years. If, as ishoped, a single model can reproduce the major features of the distributions of both artificial and natural tracers, the predictions of such a model will undoubtedly be correct in the mean. Only the year to year differences from the mean picture might then presentdifficulties. MOST IMPORTANT FALLOUT-FORECASTING NEEDS Given an injection of contaminants at any arbitrary point in the at- mosphere, how well can one predict its subsequent history? This forecast, with reasonable accuracy, is the central need in mostpollution problems, The discussion is made simpler by separating the trajectory from the global aspects. In the former case, most areas will have zero con- centrations, but those regions along the path will be finite, Accuracy of prediction of concentration at a given point may be meaningless in the presence of a “yes—no” situation. In global fallout, on the other hand, very large regions contain varying amountsof fallout, and prediction of air concentration or deposition at specific points or limited areas may be described in terms of Gaussian frequency curves. There have been several studies of the verification of free air trajectories. These, in general, suggest that the true position of an air parcel will lie in a circle around the forecast position with a radius about 20 to 30% of the path length. Unfortunately, there are virtually no verifications of long-range trajectories from ground-level or low-