Reliability of measurements from photocopies of study models.