grading on the curve

where assessment is conducted so that there is a normal distribution of learners' results, with most around the mean and a few at either pole.