Usually absence of evidence is not evidence of absence. However in this case, there is a conspicuous absence of records for John Weir and Mary Welles; particularly a peculiar absence of records for Mary Welles.
There is a sizable amount of evidence that Sarah of Iowa who married Geo Jenks of New York had her last name misspelled repeatedly including one record where her last name is listed as Idore which matches no other records pertaining to her, her husband, or her children. There are a few one-time in the record variations like Warr or Mace.
Her later year records consistently list her as Sarah S. Ware, and I presume that as time went by she became more literate than she was when she was 16 years old at the time of her marriage. I trust her later records over her earlier records, and I presume the earlier records are almost exclusively filled out by Geo. Jenks or city, county, or state clerks.
At present, I have records for Sarah S Ware from 15 to burial that constructs a fairly consistent timeline leading to the ancestor I have the most direct knowledge about, my maternal 2nd great grandmother, Bertha May Matthew nee (Jenks).
There may very well be a John Weir and Mary Welles somewhere in the world that were a couple and had documented lives, but they are probably not the parents for my direct ancestor, Bertha. There has to be an accounting of records from birth to death that includes an intersection with Bertha May Jenks and Mansfield Bruce Matthews each born in Iowa. Sarah Ware and George Jenks are best fit for the area and time, by triangulation of relationships, we can see that some records are inconsistent and more records are consistent with Sarah Ware's parents being John Ware and Mary or Polly, Rhuday, Ruday, Rudi, Rudy, or Rhudy.
Human error or deceit in transcription is the most likely explanation.