Given these numbers of instances, how do we calculate the number of correlations between them? Note that association AB had initial cardinalities of A[2]-B[3], and in Single Associations, we calculated the number of correlations as card(A)*card(B)=6.
If we simply multiply sizes here as we did cardinalities, we'd get size(A)*size(B)=4*6=24, which is more correlations than we need. This is because the process of calculating sizes, resulting here in 4A and 6B, has simply created two groups of the original association A[2]-B[3]. Each group has the same number of correlations as the original association.
The total number of correlations is the number of correlations in one group, multiplied by the number of groups. The number of groups is simply the ratio of size to cardinality for an end. For example, the number of groups of A is 4/2=2, the number of correlations in a group (of A[2]-B[3]) is 6, so the total number of correlations is 6*2=12.
As a side note, we get the same result by taking the least common multiple of the sizes of the two ends.