This increment in power is designated by A E(K + 2,K)/N. 
The next step is to compute the power contributed by 
NA pw 6% 19z9t) with the use of equation (10.17). The phases 
drop out and since the #'s are different, by the arguments given 
in other parts of this paper, the first two expressions on the 
bottom line of equation (10.19) can be obtained. 
It is now necessary to investigate the limiting process more 
carefully. N and R must both approach infinity together. That 
is, each term in approaching the limit is found by picking fixed 
N and R and forming the net which has the property that each ele- 
mental area in the net contributes the same power (namely 
AE(K + 2,K)/NR). For the next N and R picked larger to approach 
the limit a completely different net will have to be found. 
Now as N and R become large, BaR becomes larger and larger 
but if it is divided by “R, the number Byp/VR has a probability 
distribution given by considering it to be a sample from a nor- 
mal (Gaussian) population with a zero mean and a unit standard 
deviation. This is stated by equation (10.20). Proof can be 
found in the statistical references cited elsewhere. Thus the 
sum over N of the (Byy)°/NR becomes more and more like the sum 
of N terms each of which is the square of a number taken from a 
normal population with zero mean and unit standard deviation. 
This sum of N terms is precisely the second moment of a sample 
of N values from the population, and by Tchebycheff's theorem 
and the law of large numbers, this sample second moment can be 
made to differ from one by as little as desired by picking N 
large enough. Therefore the limit as N and R approach infinity 
- 265 - 
