It defines that the tuple in one relation that refers to another relation must refer to an existing tuple in that relation. This constraint is states on two relations (not essentially distinct). It uses a concept of foreign key and has been dealt with in more factors in the next unit.
In the example given above, the value of C^ in every R tuple is matching with the value of C# in some S tuple. If a tuple has values (A6, B2, and C4) is added then it is invalid as referenced relation S doesn't contain C4. Therefore, it will be a violation of referential integrity constraint.