I've been looking at DYS413a and DYS413b in DF13- groups. The modal values for these are 23 in L21.

There are 43 kits that have tested DF63+. These consist of 19 surnames when I count McFarlane, McFarland, MacFarlane as one and Broome, De Brome, Broom as one, etc.

Only one surname (Bertrand) has 22 alleles at DYS413b, the rest have 23, so it seems fairly stable within this group. But when I look at DYS413a:
6 surnames have 20 alleles:
Allan, Brown, Bush, Passmore, Roma, Sims.
2 surnames have 21 alleles:
Bertrand (x3), Broom/De Brome (x2).
1 surname has 22 alleles:
The rest (11 surnames) have 23 alleles. Interestingly, this group also includes the majority of the Broome family.

Seeing as there are so few with DYS413a=22 and so many with DYS413a=20 or 21, I'm tempted to say that the original DF63+ person had the DYS413a=20 or 21, DYS413b=23 and that these markers quite often have RecLOHs. However, looking at kits that are L21+, DF13- (7 surnames), 5 surnames have 23 at both markers and 2 surnames have 24 at both markers. So they do not suggest the same for the original L21+ person.

So to my questions:
Does this make sense at all, or am I twisting the data to show something which isn't there?
Is the data at all big enough to make such assumptions?
Have I understood RecLOHs correctly?

Any comments, help, suggestions appreciated. :-)