vol in_both the Mike-t _ vice | / :and Castle-type devices larger quantity of H? left over or un- ‘burned in the Castle-type devices should be expected. This difference appears to be suggested by the data in a broad, qualitative sense. A study of the H3/Kr®> atom ratios determined experimentally shows an intolerable variation, with most values being unreasonably higher than theoretical expectations. Some results also appear to be too low, Many explanations can be offered for these variations, although none is com~ pletely satisfactory. The high values can be caused by tritium rain-out at time of sampling, while the low values could represent sampling immedi- ately after rainout where the atmosphere may be momentarily scrubbed of the tritium. an attempt was made to correlate the H?/Kr®5 ratios with respect to time of sampling. A plot of this correlation is included for what it is worth in Figure 4.1. data available. No specific conclusions can be drawn based on the The C'¥Kr® ratio exhibits the sam variation within samples collected from the same shot and throughout the entire test series. variations with altitude or sampling time are observed. No consistent For each sample in which a successful separation and assay of a de- tectable amount of Kr®5, c!4, and H®? were found, the C!4and H? formed per fission have been calculated. Only a small number of these computed ratios appear compatible with expected theoretical ratios computed for nuclear reactions of this type. of these ratios. No general observations resulted from an analysis Average values of the C'¥a3’ ratio calculated for each shot of the series are; Pa. MI peleted. it 40