Agraham's answer seems to universally fit all situations.
I wrestled with this question myself this morning, before I could find this great answer. My approach was to create a method in the class (test in your above example) which accepts a single integer parameter. The new method is simply a DIM of the array (thisarray in your above example). Just another way to do it. Both work. Agraham's approach is a bit tighter, while mine encapsulates it into the class, if you aspire to that. Mine may not be quite so tidy if there were lots of such member arrays.