Google did not deign to tell me, so I am looking elsewhere. In somewhat related terms, is there a set mathematical formula for "data skew"? Folks talk about a "little skew" or "very skewed" data. Practically any column is going to have at least some skew, but I have to assume that there is a threshold whereby a given histogram (ie, insufficient number of buckets) is going to hurt your performance. charles schultz oracle dba aits - adsd university of illinois