The principles of the Gaussian distribution (bell shape curve) is valid only if n reaches a suitable size, and actually represent the general population....

In a cohort of a few students, it is never representative of the general population, and while ploting marks may get a similar curve, it will be a skewed curve.... therefore using it to assign grades becomes arbitrary....

So if this lecturer's cohort actually lies, say on the upper 2SD if a population curve, and he thinks they deserve an A grade, that's not wrong.....

In other words, if his students are all A students, what's wrong with giving them all the A grade accordingly?