CASH v. DAU

DNA testing - Cashel with the Baronets branch.
For background see DNA testing for the Blennerhassett problem.

This page is about the tests done to prove or disprove the theory of our descent from the Blennerhassett Baronets branch through a lost daughter.

On the very first sample (BART.1) the results said the Baronets theory is true. Our family is closely related to the Blennerhassett Baronets branch.

Groups

We compare individuals from the groups defined here.

This page will compare:

CASH v DAU

Total segment matches (minimum segment 9)

We start by considering total segment matches, using minimum segment size = 9 cM.
9 cM is generally considered significant.

We compare our family with the Blennerhassett Baronets family.

This looks really good.
We have done one to one comparisons.
We have two great matches, and they are both with the BART branch, which is where our theory predicts we connect.
We have a great supporting match with ROB.1, which descends from both ROB and YIELD.
We have good supporting matches in BUTLER, which descend from YIELD.
The results strongly support the Baronets theory.
Compare it with the 9 cM table for the Letitia theory (which is almost empty).

The results are even better than they look here

In fact, the results are even better than they look here, because we have more matches on Ancestry not shown here. They are not shown here because they are not on Gedmatch.

Is the BART.1 match evidence for the Letitia theory?

BART.1 descends from Arthur Blennerhassett of Blennerville and his wife Helena Jane Mullins. So this sample is also part of the MUL group in our DNA testing of the Letitia Blennerhassett theory.
So why not count this as evidence towards the Letitia theory?
Because it makes no sense to have a strong match in MUL and no match in REV or ED. We think the match to our family is through Arthur, not through his wife.
Also, of course, we have the BART.6 match, which is nothing to do with MUL.

False positives: BUTLER match with CASH.1

We actually have to remove some BUTLER matches with CASH.1 here.
BUTLER.3 and CASH.1 have a 12.4 cM match.
BUTLER.1 (nephew of BUTLER.3) and CASH.1 have an 8.5 cM match.
So why are they omitted? Because through Triangulation, we discovered they are related to CASH.1 through a non-Cashel line (the mother of CASH.1) and the matches are through that line. So we exclude them. Looking at them is misleading.
This does not affect any other CASH since none are related to the mother of CASH.1. So we can leave the others in.
To spell it out, any match of BUTLER to those other CASH cannot be explained by the line of the mother of CASH.1, since none of those CASH are related to the mother of CASH.1. So any match needs another explanation.
This amazing false positive shows that you can get a 12.4 cM match by chance, which is shocking.
It also shows that a DNA match is not enough. You need to triangulate to find out which line the match is through.

The proof that the Baronets theory is true, May 2020

The following 15 cM match between our family and the Baronets family basically proved that the Baronets theory is true.
This match proved the Blennerhassett link on 26 May 2020, after 35 years of trying.

The match is between these two people:

John Chute, descendant of Arthur Blennerhassett of Blennerville (born 1776), son of the 1st Baronet.
Meg Niemi, descendant of George Cashel (born 1807).

The second great match, Oct 2020

On 18 Oct 2020, we got an even better match with the Baronets family, of 21 cM.

The match is between these two people:

Sir Adrian Blennerhassett, 7th Baronet, descendant of Sir Robert Blennerhassett, 2nd Baronet (born 1769), son of the 1st Baronet.
Nuala Duggan, descendant of George Cashel (born 1807).

Total segment matches (minimum segment 7)

Let us try (as some sites do) reducing the minimum segment size and looking at total segment matches.

Here is total segment matches using minimum segment size = 7 cM.

With this metric, the Baronets theory looks even better.
Though caution that matches below 9 cM can happen by chance.
Compare this with the 7 cM table for the Letitia theory, which is much weaker.
ROB.1 and CASH.1 shows up as a "match" at 8 cM on FTDNA.
It is 7.5 cM on Gedmatch.

Largest segment in match

We now take a wider look at the data by showing for every match, what is the largest segment.
Largest single segment in match, in cM.
Small segments can happen by chance. Large segments much less so.

Compare this with the Largest segment table for the Letitia theory, which is much weaker.

Estimated MRCA

If there is a segment over 5 cM, Gedmatch estimates the number of generations to Most Recent Common Ancestor (MRCA).
This is an estimate from the DNA, not from the family tree.
Gedmatch can give different estimates depending on the minimum segment length you pick. We use the closest estimate, which is the estimate given under minimum segment 5 cM.

BART seem proven relations of our family, with matches across different branches of our family.
ROB.1 seems a proven relation of our family.
BUTLER seem proven relations of our family, with matches across different branches of our family.
Compare this with the MRCA table for the Letitia theory, which is dramatically worse.

Postulated cousins

Now we consider, under the Baronets theory, the postulated cousin relations of CASH with DAU.

How do we estimate the cousin numbers? This is difficult. We take the most likely theory, which is that we descend from the 1st Baronet. We assume George Cashel is somehow a grandson of the 1st Baronet. But depending exactly how he is a grandson, these cousin numbers could vary. e.g. If George Cashel is the natural son of the 2nd Baronet, then we have 3.1 and 4.0 cousinships with the 7th Baronet. If George Cashel is son of the 2nd Baronet's sibling, those become 4.1 and 5.0 cousinships.

Until we find out exactly how George Cashel descends, we will use the bigger numbers.

The CASH people that are closest to the origin of the Cashel family are the ones for which we can get:
- Any version of 4th cousin.
- 5.0 cousin.
These ones are all in the group CASH 3 to CASH 9 (as laid out here, not numerically).

Here are the Gedmatch estimated MRCAs:

The pattern looks different to the postulated cousins. None of the 4th cousins match.
It does not at first look great.
But in fact, looked at a different way, the DNA is actually a pretty good fit with the postulated cousins:

Consider the BART group. For each CASH sample, the closest postulated cousins are BART 6 and 7 and 1. And indeed the best DNA matches are found in BART 6 and 1.
Consider the BUTLER group. For each CASH sample, the closest postulated cousin is BUTLER.3. And indeed BUTLER.3 is where the best DNA matches are found.
Consider the group CASH 3 to CASH 9 (as laid out here, not numerically). This is where we have the closest postulated cousins to Blennerhassett (not all of them, but most of CASH.3 to CASH.9). And indeed it is where the best DNA matches are found.

Here are the largest segments:

Real cousins v. Postulated cousins

Let us write out the entire list of results for real cousins versus the same list for postulated cousins, separate into lists for each type of cousin, and sort the lists, to see if the postulated has the same pattern as the real.

Here are the lists, for each type of cousin, of largest segments for real cousins versus the same list for postulated cousins (for CASH crossed with DAU).
We start at "4.0" cousins (there are no closer living postulated cousins).

This looks bad.
But we have very few samples.
All these 4th cousin samples relate to just two people, who are siblings. If that branch got very little BART DNA, by chance, that could easily explain this.

It is the 5th cousins comparison that makes the Baronets theory look true.
The results are a little weaker than the known, but not by much. The general pattern looks good. These really are 5th cousins, or at most 6th cousins. The Baronets theory looks true.

The results seem to match the known results about right.

Conclusion - The Baronets theory looks true

The conclusion is the Baronets theory looks true.
We seem to be closely related to Blennerhassett, Yielding and Butler.
Note that the finding is actually stronger than appears here, because we have other matches on Ancestry which are not visible here. They are not visible here because they are not on Gedmatch.
And we also have some triangulation on Gedmatch indicating which exact lines the DNA matches are through.
As always with DNA, we need more samples to get the bigger picture.