INVERSE does not preserve multiple translations. Suppose table 1 has two (or more) different characters translated to the same value; for example, "A" and "B" are both translated to "1". For table 2, INVERSE uses the last translated character for the value. "1" is always translated to "B" and not "A", assuming that "A" appears before "B" in the first table. Sort programs in SAS require an inverse table for proper operation.