+ The range over whicn these data mist be extrapolated in order to permit prediction of megaton craters is enormously greater than the ranges of extrapolation commonly accomplished in engineering or scientific fielas. The situation is roughly equivalent to an attempt to predict the penetration of the projectile from a new anti-tank gun through armorplate pased on observation of many measurements of tne penetration of BB's from an air rifle tnrough tin cans plus a few measurements of the penetration of .45 pistol bullets through pine. As a result of these facts any extrapolation procedure is inevitably associated with quite a large uncertainty in the final result. In making any extrapolation it is believed consequently, that it is of major importance to indicate the order of magnitude of the uncertainty involved as well as the extrapolation itsel®. At the outset of any attempt to develop extrapolation procedures, one is faced with a philosophical choice. On the one hand he may look critically into the mechanism of the phenomenon and on the basis of physical or, in this case, mechanical: analysis, study the causes, the effects, and the influence of specific parameters. Alternatively, he may adopt the attitude that, in a complicated phenomenon such as crater formation, the mechanisms by which causes and effects are interrelated are so illknown as to be for the moment. untmowable, and hence conclude that the appropriate approach is the empirical extrapolation of the existing data into the range of parameters where prediction is desired, It is the author's opinion that the second approach is the more realistic one under the circumstances involved in the present problem and that is the approach described in the remainder of this report. The most important deviation from past thinking cccasioned by this approach is that cube root scaling is on this basis discarded as a primary tool in the extrapolation and is used only for assistance in relatively minor aspects. In adopting an empirical approach, it would of course, be absurd to ignore the information, nowever meager, in regard to the physi- cal mechanism and particularly in the distinction between the mechanisms occuring in TNT and in nuclear explosions. On the other hand, it is believed that too much dependence on cuse root scaling is likely to give the illusion of a precision in prediction unjustified by the facts. The development described below was undertaken within the framework that the desirable result from a military standpoint is the construction of graphical or analytical relations such that knowledge of the yleld, soil, and depth will permit easy prediction of the crater dimensions. It is postulated that the snape of a crater for the craters of interest is primarily dependent on its size and hence the first attempt is to predict crater radius in terms of the three parameters just mentioned, with the expectation chat a later analysis can be made to predict depth and other shape aspects once the radius prediction has been accomplished. z . 4 2 x 7 f Fee orf - aacOR 4 let.