(ref.doc)fjh 251193

(Note:
 Henderson)

Mutable actually solves a quite significant problem, which was the
lack of any way to distinguish between const-ness of the "abstract
state" and the "concrete state" of an object, except by deliberately
subverting const correctness using cast-away-const.  Bertrand Meyer
devotes eight pages to discussing this issue in his book
"Object-oriented Software Construction" (section 7.7, pages 132-139
Note:
 OOSC).  His solution in Eiffel is effectively the same as what
would result in C++ if every member was automatically declared as
mutable, i.e. const member functions were allowed to modify any member
without a cast.  I think the C++ "mutable" solution is better.