It is physically challenging to manufacture a waveplate with a thickness precisely corresponding to a path difference of lambda/2. First, estimate the typical thickness (hint: consider the difference in 'n' for the slow and fast axes). how do manufacturers circumvent this problem?
