Recent work on robust estimation has led to many procedures, which are easy to formulate and straightforward to program but difficult to study analytically. In such circumstances experimental sampling is quite attractive, but the variety and complexity of both estimators and sampling situations...