Remarks on the determination of optical constants from ATR measurements