This problem is somewhat similar to an earlier question of mine, but the issue is more general than before. In one tree I am researching, I can find a marriage record for Nathaniel Claughton and Martha Howgate in 1791 in Guisely in both Ancestry.com and FamilySearch.org (the date is off, but it's the same people). Based on his death record as well at the date of his marriage, he was probably born around 1770. They had several children, including another Nathaniel born 1802, before Martha died in 1822. This can all be traced through both Ancestry.com and FamilySearch.
Here's where it gets weird. There is a record for a Nathaniel Claughton marrying a Sarah Smith in December 1822. I had this as a marriage involving the younger Nathaniel, based on the Ancestry.com scan of the parish record, and the fact that there don’t seem to be other Claughtons in Guiseley at the time – though there are others in nearby towns in West Yorkshire. But the FamilySearch.org index listing lists both members of the couple as being 50 years old, which would be consistent with the elder Nathaniel remarrying soon after Martha died.
In her answer to my earlier question, ColeValleyGirl indicated that Familysearch.org tends to be based on Bishop’s Transcripts not the original parish record. In contrast the Ancestry.com scans are clearly the original parish register.
I suspect that the marriage to Sarah Smith was incorrectly annotated in the Bishop’s transcripts: in the marriage to Martha Howgate, Nathaniel made a mark rather than signing. But in the marriage to Sarah Smith, that Nathaniel signed the register.
So I'm going to conclude that the original parish register had no ages recorded, that the illiterate and literate Nathaniels were different people (but father and son), and that the ages in the FamilySearch.org index were erroneous additions to the Bishop’s Transcript, probably because the elder Nathaniel was a property owner and therefore better known.
Is it common for FamilySearch.org indexes of marriages (and burials, I suppose) to have extraneous information added compared to the original parish register, and if so, how much weight should be put on that extra information if it conflicts with a scan of the parish register?
I am trying to construct a question of general use here, rather than one specific to this particular case.