We base our definition of the testability of a class on the testability
values,
, of all the methods in the class. We define the testability of a
class, 
,
to be the minimum of the testabilities of the methods in the class.
Since the testing effort is maximum in the method that hides the most information, we have selected the lower bound of the testabilities of all the methods, to be the testability of the class.
We do not include constant objects and methods in our complexity calculations. The reason is that the metric considers the set of objects that have the potential to change their state, possibly erroneously, during the execution of the method. Since constant objects and methods are syntactically state invariant, at least during the execution of the method, this assumption makes the metric more sensitive to those objects that may have taken on erroneous values that will be propagated on to other methods, and possibly other objects.