We will assume here that P is a set of programs for computing a function F and that the programs in P are correct or contain computation errors. Each of the incorrect programs P in P are assumed to have computation errors relative to a particular correct program P* which is also in P. This is valid if we agree that the type of error an incorrect program contains is defined with respect to a particular correct program.
The paths in the incorrect programs P and the correct program P* are so indexed so that D(P_i) = D(P_i*) for all paths.

The incorrect assignment statement "J = MARKS(I) - 1/10 + 1" causes an incorrect value to be computed for the variable J.
P-testing is reliable for testing this program.