Density is a measure of the average proportion of duplicate keys in the index. It varies between 0 and 1. An index with N rows whose keys are unique has a density of 1/N; an index whose keys are all duplicates of each other has a density of 1.
For indexes with multiple keys, density values are computed and stored for each prefix of keys in the index. That is, for an index on columns A, B, C, D, densities are stored for:
A, B, C
A, B, C, D