0.00 



0.01 



Figure 13. Stage-A fals^alarm probability (corresponding to optimum Ng in 

 figure 12) vs single-scan detection probability. 



results indicate that the percentage of times a second stage is necessary in the 

 noise-only case is between about 0.5 percent and 4 percent when the two-stage 

 test is optimized for a Pq between about 0.2 and 0.8. It can be seen in figures 

 14 and 15 that the ratio of per-stage miss probabilities of an optimized test tends 

 to fall between about 0.04 and 0.1 (a large variance results from the binomial 



2 9 



