This sort of problem has come up quite a lot recently - and I am coming to believe that it applies to many distant matches. And, I do not really like the term 'false positive' - but prefer something like ''fixed segment'. Therefore, I do not deny there is a link, but the question becomes as to just how far back ... I think GEDmatch, FTDNA, Ancestry, 23andMe are fairly reliable when the generation distance is under 4, but after that the number can often be double what is suggested, if not a lot further. Just recently when looking at GEDmatch  One-to-many pages I have been using 11cM, instead of the default 7cM; but even this fails to exclude the larger 'fixed segments'. Good examples of this can be shown when looking on GEDmatch at African and Asian submissions; as suddenly they can turn up with large numbers of spurious American matches all of which look to have quite significant matching segments ! The answer, I suggest, is to concentrate on your closest matches and be very wary of all distant matches. Ian ------------------------------------------- On 07/12/2017 17:24, Brooks Family wrote: > I have a match on AncestryDNA.  AncestryDNA reports that it is a 6cM > match.  The match contacted me, and reported that although I don't > match either of her parents, the small segment "... means that both of > them are matched to you and together their DNA adds up to a match to > me but is not enough for them separately" > > It is true that we both have ancestry from a small parish in Cornwall > that intermarried for at least a century or two.  I didn't spot an > obvious connection on our trees, but I didn't inspect closely. > > I've suggested to my match that this segment is a false positive; she > is reluctant to give up her idea.  I've provided two links about false > positives. > > Anyone have thoughts about this?  Either direction. > > Thanks!