AReasonableMan wrote:you realize you're talking about the separation b/w the #1 and #last person in a class, not a median? you're speaking about 4 people total, and extracting from the 4 something about hundreds of people? It's very possible the #1 kid in that class had an amazing test, and the #100 kid drew a smily face (I'm assuming a smily face was not the correct answer).
Sure, but a 16-point spread is still really narrow (yes, unless out of 16 points) to distribute a 1L curve over. The point is probably that it's hard to write an exam that distributes grades over a broader range, because at a decent school everyone does know everything so there isn't much to distinguish the different exams. But a professor should be able to write such an exam.