hui

According to the examples in this webpage ( http://en.wikipedia.org/wiki/Tajima’s_D ), it seems that pairwise difference between any two sequence pair never exceeds the number of segregating sites. How could their difference lead to a positive d, which results in positive tajima D? The example can look like this: >sequence1: TTTTT >sequence2: ATTGT >sequence3: ATTGT segregating sites: 2 average pairwise difference: (2+2+0)/3=4/3 Thanks a lot!