GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2005-06 > 1119739647

From: "Chris" <>
Subject: TMRCA
Date: Sat, 25 Jun 2005 23:47:27 +0100

John very kindly answered my questions last night. I've just been re-reading and am still a little perplexed. so sorry John, I'm going to labour the subject again. In two sections:

Section one:
I wrote regarding two subjects. One subject A) with a well-documented line back to an individual alive in 1341. Results are that they are a 33/37 match. Ive searched Ybase & Ysearch and the nearest matches are a 31/37 and 6x29/37 matches. FTDNA give the following TMRCA:
>
100yrs: 4.54%, 200 yrs:30%, 300 yrs: 61.80%, 400 yrs 83.20%, 500 yr:93.66% & 600yrs: 97.85%

John wrote: "The FTDNA time scale is based on the problematic Arizona study of mutation rates, which seems to be off by a factor of two or so.To convert to the real world, you should multiply those times by two.I.e., the 200yr probability is 4.54%, and so on."

I wrote: If I adjust the figures for the knowledge that they couldnt be connected within 16 generations (ie Subject B)'s earliest known ancestor Thomas: then we get the following: 400-500 yrs: 60.95%. 600 yrs: 86.45%.*

John Wrote: "This gets a little complicated. Since there is a scale factor of two,you need to go back and redo that second calculation and specify only 8 generations of exclusion. That way, when you then multiply the time spans by two, you'll be excluding the right amount."

From which I concluded that:
>
A conclusive link exists. The probability is in favour of this being in the 4 generations prior to Thomas (ie 60.95%), but that this may have taken place in the previous 4 generations (ie probability of 25.50%). The odds are that they both descend from Hugo. But that there is an outside chance that this link was earlier.

John Added: "After applying the scale factor, you lose the conclusiveness, but you still have a 37/37 match!" Sorry John a 33/37 match." I've done as you suggested and input only 8 generations. This gave the following figures:
200-300 yrs: 41.64% x 2 = 400-600yrs: 41.64%
400 yrs: 73.11% x 2 = 800yrs: 73.11%
500 yrs: 89.50% x 2 = 100yrs: 89.50%
At 1600 yrs it still not 100%

I can't help but feel that these figures are pessimistic? Based on a) the same Surname b) a geographic distance between the two earliest known ancestors of 55 miles as the crow flies c) no-one else on Ybase or Ysearch come anywhere near matching them. I still feel I can conclusively say they are connected but now with much less confidence that this is since Hugo (1341).Indeed based on the above their is a 58.36% probabilty that this was before Hugo? Indeed if we take 800yrs ie 1100 as the beginnings of surnames this gives a 26.89% possibility before this.What do people think?

Section Two:
*The latter loses me, why has that gone down by 11.40%? increasing the 2.15% to 13.55%.

John Wrote: "The revised calculation is a set of conditional probabilities obtained by blanking out the cases from 0 to 16 generations and then scaling the remaining probabilities up to total 100%. Since you're cuttingout 83.20% of the cases this way, there's a major rescaling needed".I still can't understand this. Is this starting the 100% at 16 or 17 generations? If so why? Shouldn't it be like a cake with slices taken out and you deal with the remaining percentage? ie if the 100% mark is say 1000yrs, that should remain at 1000yrs. I'm no mathematician and this may a) be beyond me or b) I'm just misunderstanding. The blanking out of x numbers of generations/yrs, to me would indicate that the remaining generations/yrs up to that 100yr point when recalibrated, would result in the 2.15% having gone down further not up to 13.55%? Is my logic totally wrong?

Sorry to labour this but I'm trying to get to grips with things.

Many thanks folks for your patience. Thoughts appreciated,
Chris