Archiver > GENEALOGY-DNA > 2005-06 > 1118837407

From: "John McEwan" <>
Subject: RE: [DNA] R1b some haplotypes & Possible Geographical Origin
Date: Thu, 16 Jun 2005 00:10:07 +1200
In-Reply-To: <008a01c5711a$8d155100$01ccae51@oemcomputer>

Dear Kevin, Mark, John, Peter, Larry and list

Thanks for your many comments and queries on and off the list regards
the McEwan surname cluster diagram. I see the graph has also spurred the
debate along regards "native Britons that spoke p-celtic" as well.

Below I have tried to reply to your main queries
1) Why was XXX surname not included?
a) I was using Ybase and it often does not have many entries for some
surnames even if they are available elsewhere. I was also looking for 10
or more 25plus marker entries per surname in Ybase except where the
surname had a particular relationship to the McEwen name.
b) I was looking primarily for surnames reputed to be of Irish, dal
riadic, or "native/aboriginal Briton located north of Hadrian's wall"
ancestry i.e. predominantly R1b to see if surname patterns could be
observed within this "group" using this number of STRs which is somewhat
greater than published papers.
c) The graphics begin to break down above about 50 names
d) However the graph does need updating and when I do it I will try to
ensure that I include more names from those interested which have more
than 10 samples satisfying the criteria.

2) I thought XYZ surname was of a different historical origin?
a) The classification was based on searching for the origin of the
surname (or group of surnames) on the web. This is "fast but prone to
error". Apologies and thanks for corrections.
b) Some surnames appear to have several historical origins or disputed
origins (eg Campbell as has been discussed in some depth on this list).
c) In the case of Kincaid I actually used the Kincade variant and it
returned [(origin: Gaelic.) From ceann, head, and cath or cad,
battle--the head or front of the battle.]. However Kincaid variant
returned a "native Briton/Pict origin". Given I used 9 "Kincades" of
which 5 were spelt Kincaid this is probably an error. As I stated
originally, identifying surname origin via a plot of frequency by county
in the 1840 census would be much better.

3) XYZ surname has several separate origins based on DNA!
a) Steven Perkins and Mark McDonald suggest two distinct separate
origins for the McDonalds stating: "You need two entries for the
McDonalds. The Chiefs are in R1a1, descendents of Somerled, and the
majority of the clan is R1b. I believe this has also been found in a
few other clans." I am sure some other surnames are similar and I noted
this in the original post as a source of error. In fact very internally
diverse surname groups via this clustering procedure can be detected
because the have short branches emerging from the origin and are located
mid way between extreme examples of near pure haplotypes. In the case of
McDonald surname it consists mainly of a mixture of R1a (eg Orr) and R1b
(eg Fennessy) haplogroups and on the opposite side of York which has a
component of I1a ancestry.

There are several possible replies to this: the first is if there is
evidence for separate origins they should be separated. In the extreme
we should plot each individual using Network and ignore the surname
information. The other extreme is to retain this variation, because we
are trying to use surname only as a proxy for geographic location 1000
years ago. In practice I guess most people are like me, we want
something in between as we are trying to find evidence of DNA
relationships between surnames that we think are related based on
historical genealogy. To do this we need some finer and multileveled
grouping strategy better than simply haplogroup. Where a surname has
obvious significant subgroups that definitely originate prior to 500AD
(eg haplogroups) then there is a case for dividing them into separate
surname subgroups.

4) Finally I am no historian but know the dangers of using cultural
markers like language as evidence of independent paternal Y chromosome
origins as they tend to often be adapted by the population in situ
rather than by replacement of individuals. In this case p-celtic (aka
Brythonic which is actually derived from the p-celtic word for Briton!)
and q-celtic or Goidelic. However, I understand the latest evidence is
that p-celtic and q-celtic languages split a lot earlier than is often
supposed i.e. perhaps up to several thousand years BC (prior to being
first used in Briton and Ireland) and therefore there is at least some
substance to the testable hypothesis that if sufficient Y STRs are used
(or SNPs identified) it may be possible to identify "aboriginal"
Britonic Celtic Y haplotype strains different from those that came from
Ireland within the current R1b haplogroup. In the graph presented (with
all its flaws) I was also trying to determine if the Dal Riadic celtic
surnames clustered more closely together than those from Ireland (they
appear not too) and if they were different from "aboriginal" inhabitants
of northern Britain. Even with the flaws the evidence from the graph
suggests that there is a potential case for further investigation.


John McEwan

This thread: