In case the hash function is rather good, and the volume of buckets is key, then anywhere up to at least one product for every a single bucket will result in a median look for depth of one and an average search depth just around 1.(notably Perl does this, however I hope other languages do likewise). Other languages elect to refer for their implemen