This is a question I recently posed at 23andMe, but perhaps someone here can answer it as well.  In messing with my raw data file (a text file of the 500K tested SNPs under the 23andMe V2 chip), something caught my eye.  Specifically, the number of CC pairs was almost identical to the number of GG pairs (i.e., within 0.5% of each other, with both at roughly 100K), the number of AA pairs was almost identical to the number of TT pairs (roughly 88K), the number of AG pairs was almost identical to the number of CT pairs (roughly 72K), and the number of AC pairs was almost identical to the number of GT pairs (roughly 16K). AT and CG pairs hardly appeared at all, but the counts for both were broadly similar, well under 1000 in each case.  Can someone explain the mechanism that makes these counts themselves occur in pairs?  No doubt it's something basic, but the answer is not immediately obvious to me. Thanks. Ruy Cardoso