Proving Grounds — probably mostly tropospheric. In both cases, the close-in fallout has been omitted. The upper part of Fig. 2 shows the superposition of all the tropospheric fallout patterns as the heavy line as well as the individual contributions based on the lower curves. The heavy line is now offered as the probable world-wide tropospheric fallout. In order to find the amount of stratospheric fallout, it now is possible to subtract the cumulative tropospheric fallout curve in the upper part of Fig. 2 from the cumulative total fallout curve of Fig. 1. Unfortunately, the dates for the various sets of data do not coincide. The observed curve contained points for many different times in 1956, and the tropospheric fall- out curve has been computed as of the end of 1957. Hence, in Fig. 3, the actual observed curve has been adjusted to the end of 1957, by adding the pot fallout at stations corresponding to the latitudes of the arrows, A certain amount of extrapolation is necessary Since not all pot sta- tions were operative from mid-1956 to the end of 1957. Though some error is added by performing this extrapolation to the end of 1957, this is not more than 5 or 10 per cent of the 1956 value at each latitude, in our opinion. Figure 4 now shows the tropospheric and total curves with the difference curve as the heavy Solid line. This difference curve is now considered to be the stratospheric fallout. It shows rather clearly that the stratospheric fallout is pronouncedly nonuniform with a peak in the north temperate latitudes. The errors in this curve stem from three sources: first, the determination of the true fallout profile shown in Fig. 1. It is evident that one has considerable leeway in the details of the construction of the observed fallout curve although it is contended that the peak in the temperate latitude fallout will, in any analysis, be at least twice the world mean, The second error, for which we quoted 5 to 10 per cent is in the conversion of 1956 to December 1957 fallout. And finally, the tropospheric profile determination, which is dominated by the uncertainty in the source strength of an unknown magnitude rather than from the shape of an individual tropospheric fallout profile. It is our view that unless the U.S.S.R. tropospheric fallout has been underestimated by a factor of about 3, the data cannot be reanalyzed to yield anything but a nonuniform stratospheric fallout pattern. 4 RAINFALL Rainfall data have been collected at many stations in the United States, the United Kingdom, and elsewhere. First of all, the rainfall data confirm the north-south profile in the yearly observed fallout, found from the soil data. The rainfall data also show a seasonal variation in the amount of deposition per unit time for stations in the north temperate latitudes. Figure 5 shows a graph for New York City, the station with the longest continuous record in the United States. The abscissa is time on a linear scale, and the ordinate the amount of fallout per month. It is seen that the fallout is greater in the spring and less in the autumn. One can also see that the heavier fallout in the first half of the year is not associated with heavier rainfall. Until 1957, almostall of the United States tests occurred in the spring of the year, more or less. One could have ascribed the increased spring fallout to tropospheric fallout from these spring tests. It still was difficult to understand the lack of fallout from the U.S.S.R. tests in the fall, but, since information of the tropospheric contribution was lacking, this omission was not unrea- sonable, However, in the summer and early fall of 1957, a test series was conducted in Nevada, Plumbbob, in which we tried, if anything, to minimize local fallout. This meant that tropospheric fallout should have been at least as great as in previous Nevadatests, if not greater. Despite this, the fall of 1957 showed the same drop in fallout experienced in previous years. Data from other stations confirmed this autumn drop. It has been the position of Machta and Stewart et al., even prior to the Plumbbobtests, that this seasonal variation was caused by stratospheric air being brought into the tropospherein the late winter and spring. This air would bring with it the high Sr* content of the stratosphere. Measurements of ozone (Fig. 6 in the preceding paper), which has its origin at about 30,000 or 40,000 ft in the stratosphere, show that the ozone in the troposphere also exhibits the same seasonal variation shown by the Sr*°, Further, Canadian observations of ground level Be’ concentrations, as seen in Fig. 6 (this paper), also show the same variation as the Sr®’, One can stretch one’s imagination to find an alternate explanation for this seasonal variation in the Be‘, but too the most reasonable explanation is the injection of stratospheric air into the 330 tte.