So I read the paper the guy linked to and it is a cool methodology, but I don't think it can be used as a Borderer count.
Briefly, they try to find names distinctive to each minority group, then estimate the frequency of that group based on the abundance of those names. If you know that 1/X Irish people have the last names Murphy, Kelly and Sullivan, and you know that these names are rare outside of Ireland, then you can figure out how many Irish people there are in a group by counting the Murphys, Kellys and Sullivans and then multiplying by X. Using things like baptismal records from the old country, they try to find as many pure Irish names as possible to make X as small as possible so the count is more accurate. Ambiguous names (e.g. Hayes can be Irish or Scottish) are purposefully excluded. There's some tradeoff you have to navigate where adding more, less pure Irish names will make X smaller but will lead to you counting some non-Irish as Irish. For some of the smaller ethnic groups on that table (e.g. Swedish) founder effect is a serious issue so they choose names from records after emigration to the New World but before mixing. Their estimate of English is just 100% minus the white minorities.
The problem, which they fully acknowledge, is that there is no way of distinguishing Scottish from Scotch-Irish (of Scottish descent). They estimate based on outgoing immigration records a ~2:1 ratio of Ulster:Scottish emigration, which is why on that table their Scotch-Irish estimate is double their Scottish estimate for every state. IMHO they should have just merged those columns rather than create an estimate of false precision. The bigger issue, I would think, is that people of English Borderer (including some Scotch-Irish) descent are classed as English here. They weren't trying to estimate Borderers so I can't complain, they explicitly say that their analysis ignores distinct English subgroups, but without that number we can't trash Albion's Seed.
So I read the paper the guy linked to and it is a cool methodology, but I don't think it can be used as a Borderer count.
Briefly, they try to find names distinctive to each minority group, then estimate the frequency of that group based on the abundance of those names. If you know that 1/X Irish people have the last names Murphy, Kelly and Sullivan, and you know that these names are rare outside of Ireland, then you can figure out how many Irish people there are in a group by counting the Murphys, Kellys and Sullivans and then multiplying by X. Using things like baptismal records from the old country, they try to find as many pure Irish names as possible to make X as small as possible so the count is more accurate. Ambiguous names (e.g. Hayes can be Irish or Scottish) are purposefully excluded. There's some tradeoff you have to navigate where adding more, less pure Irish names will make X smaller but will lead to you counting some non-Irish as Irish. For some of the smaller ethnic groups on that table (e.g. Swedish) founder effect is a serious issue so they choose names from records after emigration to the New World but before mixing. Their estimate of English is just 100% minus the white minorities.
The problem, which they fully acknowledge, is that there is no way of distinguishing Scottish from Scotch-Irish (of Scottish descent). They estimate based on outgoing immigration records a ~2:1 ratio of Ulster:Scottish emigration, which is why on that table their Scotch-Irish estimate is double their Scottish estimate for every state. IMHO they should have just merged those columns rather than create an estimate of false precision. The bigger issue, I would think, is that people of English Borderer (including some Scotch-Irish) descent are classed as English here. They weren't trying to estimate Borderers so I can't complain, they explicitly say that their analysis ignores distinct English subgroups, but without that number we can't trash Albion's Seed.