Perhaps you are being dense. However named, this is a hashing scheme. Division by a modulus m yields the remainders, m of them, 0, 1, 2, . . . , m - 1.
Long established number-theoretic results then predict clustering of these hash values at the prime divisors of a composite m. Without clustering the expected number of instances of a hash value H in a sample of size N is just N/m. When clustering occurs the values for the prime factors of m are larger. --jg
