If you look at the original page of the 1871 census it clearly shows Rhoda as D-in law, not daughter (so has been wrongly transcribed) listed immediately after husband Charles. Then comes Harry - shown as son i.e. since relationships are to the head of the household, Harry is son to John and Harriet - this has already been proved - birth Jun.qtr.1853 Ecclesall Bierlow, mmn Swiby (should of course be Twiby) as shown on GRO's own index.
George H. 2 is correctly shown as 'grandson' to John and Harriet BUT was born the year before his parents Charles and Rhoda (nee Bell) married in 1870. Again, as already mentioned, his birth is registered as George Harry Bell in Mar.qtr.1869 Barnsley.
So, there is no mystery! Rhoda was wife of John and Harriet's son Charles, George Harry was their son who was born the year before they married so registered as 'Bell' and Harry was the son of John and Harriet.
Don't know how to make it any clearer for you.
Annette