Because of the scale, the five percentage points that can be gained from the unit tests have a dramatically lower impact on students who get average grades in the class than on those with either very low or very high grades.

Each student is ranked and given a percentile in the class based on problem sets, exams and section participation according to teaching fellow David Blackburn.

The curve based on the ranking system places the greatest number of students in the middle letter grade groupings, with fewer students falling to the far ends of the spectrum.

Each unit test a student passes raises their grade by one 'percentile,' and the new percentile determines the final grade based on the original curve.

As the syllabus states, "unit tests act like an insurance policy. They certainly don't guarantee that your letter grade will be higher, but if your pre-unit test scores put you very close to the cut-off between two letter grades, passing all of the unit tests will insure that you get the higher grade."

The tests thus disproportionately impact students on the extremes of the class.

Teaching fellow Matthew Choate explains that because so few students get lower than a B- anyway, if everyone in the class passed all the unit tests, virtually no one would get lower than a B-.

But that also means a disproportionate percentage of students at the high end of the scale will benefit from the unit tests. If 15 percent of students would receive an A- without unit tests, a third of them will get bumped up to an A if they pass all five unit tests.