This example attempts to demonstrate a testing strategy, which we may term H. In some cases, H may be 'almost reliable' for a program P.

This example is correct except for the missing initialization assignment SUM=X which should occur before the DO-loop. If the testing strategy H involves the generation of a sequence {T_i} for i=1 to i=n, causing the loop to be executed i times, and the selection of a set T containing one element chosen at random from each non-empty T_i. It is almost certain, using principles of further decomposition of T_i, that all of the elements of T will belong to the subsets T_i prime, and that the error in P will be revealed when P is tested on T.

H is therefore 'almost reliable.'