The total of n integers will have a remainder of either 0, 1, 2, 3, …, or (n-1) upon division by n. This implies that the possible fractional parts for the average of n whole numbers are just: 0, 1/n, 2/n, 3/n, …, (n-1)/n.
E.g. If 1 subject: only can have decimal of 0.
If 2 subjects: only can have decimal of 0 or 0.5 (i.e. 0 or 1/2).
If 3 subjects: only can have decimal of 0, 0.333..., or 0.666..., (i.e., 0, 1/3, or 2/3).
If 4 subjects: only can have decimal of 0, 0.25, 0.5, or 0.75 (i.e. 0, 1/4, 2/4, or 3/4).
Etc.
Furthermore, if 0 < f < 1 and the smallest positive integer for which f occurs as a possible fractional part in the average is n0, then f can occur as a fractional part in the average of N subjects (N a positive integer) if and only if N is a (positive) multiple of n0 (easy exercise).
So for instance, a fractional part of 0.25 in the average means the no. of subjects used in the average must have been some multiple of 4, since 4 is the smallest positive integer for which 0.25 can occur as a fractional part.
(Assumptions: weights in the average are equal and subject marks are whole numbers. If one or more of these assumptions is violated, then other decimal values may be possible.)