Rationale behind requiring ampersand and fully qualified name to form a member function pointer
 

I hope Mr. Stroustrup can give an answer to this question:

What was rationale behind the requirements to use an ampersand and a fully
qualified name of a function, to form a pointer to a member function? Is
there any syntax ambiguity without those? MS C++ compiler is quite happy
when a plain function name is used (I haven't tried it in ANSI compliance
mode, though), so for long time I've been pretty sure that it's OK...