COFFEY ORIGINS: EDWARD AND PETER GROUPS

By Fred Coffey

ONLINE: http://www.coffey.ws/FamilyTree/DNA/CoffeyOrigins-Edward&Peter.htm

There are several genetically distinct groups of Coffey/Coffee families that we have been following through our y-DNA studies. And recent developments have allowed better focus on two of those groups – those that descend from Edward Coffey who arrived in America before 1699, and those that descend from Peter Coffee, who arrived in the early 1700’s.

We have long known that Edward and Peter were related, and in the following we will explore what we can learn about the timing of their Most Recent Common Ancestor (MRCA).

We have also determined that Edward and Peter are part of a more ancient and extended family, which includes members of the Kehoe (and variant names) family. And we now believe that the Coffey and Kehoe names both evolved from a more ancient common name.

Finally, genealogists from both the Coffey and Kehoe lines have been examining an entirely different DNA approach, connecting our families to truly ancient migrations into Ireland.

We think what we have learned is fascinating, but the details of how we got there can get complicated. A lot of readers will be interested in the conclusions, but not in the details. Therefore we will write SHORT reports on what has been learned, and make reference to the document you are now reading as a source of DETAILS.

Let’s start with looking at the ancestry of Edward and Peter:

THE MRCA OF EDWARD AND PETER:

We have been working with two major lines of Coffey/Coffee ancestors, and DNA tells us that they are clearly related. There is a MRCA (Most Recent Common Ancestor) back there somewhere. We are of course talking about the "Edward Group" and the "Peter Group".

Do we have information on when this MRCA might have lived? Yes we do, but our perception has changed as new information develops. I want to lead you through the following exercise, exploring the development of our understanding:

Consider two of the tested individuals: We have "Luther Coffey" who descends from Edward, and we have "Carol Coffee" who descends from Peter. Both of these individuals have 67-marker tests, and they differ from each other at only two markers – a "genetic distance" of two. There is no doubt that they are related.

Can our testing service, Family Tree DNA, tell us anything about the time to the MRCA? Yes, they have a calculation tool that works with the probabilities of changes in individual markers to estimate how much time must have elapsed. They call it "Family Tree DNA Time Predictor", or "FTDNATiP".

So we can ask FTDNATiP to look at the 67-marker match for Luther and Carol. And it comes back with information like "There is a 5% chance they have the same father, and you can be 60% confident that the MRCA is within the last 5 generations."

Now, Luther and Carol were very good genealogists, and they would say "That is pure nonsense. We KNOW we don't have the same father, and we KNOW we're not related within 5 generations."

In fact they have each carefully researched their genealogy back 8 generations, and they KNOW there's no common ancestor there. The "most recent" possibility for a common ancestor is Edward's father, who could just possibly also be the father or grandfather of Peter. But they don't know his name or anything about him. And it could of course be someone much farther back that that!

For shorthand, we'll call this most recent possible common ancestor, in the 9^th generation, "FOE". That stands for "Father of Edward" (and doesn't "Foe Coffey" sound like a good 17^th century Irish name?)

FTDNATiP understands this issue, and it offers the option for you to instruct it "There's no possibility of a common ancestor within 8 generations". And it will happily re-calculate.

This time it comes back and says "OK, there is no chance of a MRCA in the last 8 generations. In that case, there is a 31% chance the MRCA is FOE (i.e. the 9^th generation). And there's about an 80% chance that MRCA is in the 4 generations including FOE."

But now it's MY turn to say "This is nonsense!" That's because I know that FTDNATiP did its calculations assuming that the two mutations that make Luther and Carol different could have occurred in ANY of the last eight to twelve generations, and also assumes the mutations could have been on EITHER of their lines. And I know better than that!

How do I know that? Let me take one line at a time:

For Luther's line leading back to Edward, be aware that Luther was NOT our only test subject. We have actually now tested about 50 men whose ancestry leads back to Edward, and a number of those have 37-marker and 67-marker tests. Now, there are indeed a lot of differences (i.e. mutations) in the DNA profiles of those 50 men. That's to be expected if you're talking 67 markers, 50 genealogies, over 8 generations.

But if you study the data carefully you quickly conclude that Luther's particular ancestral line is one that has NOT seen ANY mutations at all. Luther (and several other men) actually have the original Edward DNA, from which all others are derived. (Just try to postulate any OTHER starting point from Edward, and try to work out a sequence of mutations that would yield the actual results for the 50 men. It's impossible.)

So, if I could take my time machine back to year 1700 and take a sample from Edward (assuming of course he didn't shoot me for trying to practice witchcraft), that sample would yield identical results to Luther. There have been NO mutations in Luther's DNA line.

Now consider Carol's line: Here I'm not quite so absolutely confident, but we have now tested TWO men who descend via entirely different lines from Peter's son William (born about 1740). And those two men have absolutely identical profiles. In other words, they both differ from Luther in the SAME two specific markers.

Are the differences due to changes since the time of William, or did William already have the "two marker" difference? If the former, then there must have been two SEPARATE and IDENTICAL mutations of TWO specific markers, on each of the two descendant lines. The odds against that are VERY high. It's not impossible, but the odds strongly favor single mutations, if such an explanation can be found.

It is MUCH more likely that those mutations had occurred BEFORE the time of William, or in William himself. That only requires that they happen once – such single random changes are not unexpected. Thus it is VERY likely that ancestor William already had exactly the same DNA profile as does his living GGGG Grandson Carol.

(We do have a third descendant of William by a third, different line – but I haven't persuaded that person to upgrade from his 12-marker test. If he did so, and it matched, then I would KNOW I had William's DNA pinned down!)

So lets jump into our time machine again, and go back to Virginia, to the year 1750. We find William (he's about 10 years old) and get a DNA sample. And then we also find Luther's GGGG Grandfather John, son of Edward. John is about age 36. And we get a sample of John's DNA too.

Then we "return to the future" to the year 2012, and submit our two samples to FTDNA. And, as we would expect, they come back identical to the results for Carol and Luther.

So we can turn to FTDNATiP again, but tell it that THESE two samples have only TWO generations in which there cannot be a common ancestor. The results come back, and we correct for the six generations we skipped. This time FTDNATiP says there is only a 12% chance the MRCA is "FOE Coffey". But there's still a 58% chance that the MRCA is within the 4 generations including FOE.

To summarize, here's a graph of what we have just done:

To review: When we just told FTDNATiP to simply analyze the sample results, it jumped to the "unadjusted" conclusion that there could be a very early MRCA. We then told it there couldn't be one in the first 8 generations, and it shifted our curve down and to the right. When we also told it there was evidence of no mutations in the last 6 generations, it shifted the curve further down.

I believe the far right curve above is the best current representation of how many generations we have to look back in order to find the MRCA for all the descendants of Edward and Peter.

(MEMO: We do now have one test on the Edward line, and one on the Peter line, that go to 111-markers. These two do show 4 additional mutation differences above 67-markers. If this were typical, it makes a big difference to the “Unadjusted” line above because it’s almost impossible to accumulate that many differences in only 3 or 4 generations. But it makes only a modest difference to the “red” line above.)

EDWARD AND PETER’S DISTANT COUSIN FROM BALLYHALE, KILKENNY, IRELAND:

We have a recent y-DNA test on a descendant of an immigrant from Ireland, who arrived in America MUCH more recently than Edward or Peter, and who has an excellent match to Edward/Peter. The immigrant, Patrick, arrived in 1863 from Ballyhale, Kilkenny. This was a big discovery, because it suggests than Edward and Peter may also have had their origins in Kilkenny, or nearby.

The tested person was named Don Coffey. I would love to update the previous analysis with his data. Unfortunately, his y-DNA test was done on Ancestry.com, not on FTDNA.com. There is considerable overlap between the two tests, but there are several missing markers. I cannot do a full comparison, unless I can get Don re-tested on FTDNA. But at the moment I can’t establish contact with him.

THE COFFEY/KEHOE FAMILY:

Now let’s move on to our next topic, and see if we can “extend” the Edward and Peter families:

We have known for some time that many members of the Coffey DNA Project had some fairly good DNA matches with four individuals in the Kehoe DNA Project tested on FTDNA. And we have more recently learned there are two more individuals with similar DNA who have been tested on ancestry.com.

As shorthand, I’m going to refer to the Edward Coffey line as “ECoffey”, and the Peter Coffee line as “PCoffee”. As discussed earlier, we have a number of individuals with comparable DNA tested for each line. And we now also have the “DCoffey” line for the recent immigrant, tested on Ancestry. Further, I will also identify each tested “Keogh/Kehoe/Kaho” by initial and last name. One of these, R. Keogh, lives in Ireland. I think one of the others is Canadian, and three are from the USA.

There’s a mix of tests, but with considerable overlap. ECoffey, PCoffee, and RKeogh are tested on FTDNA to 111-markers. JamesKehoe, GKaho and PKehoe are tested on FTDNA to 67-markers. DCoffey and JKehoe have 46-marker tests on Ancestry, and DKehoe has a 33-marker Ancestry test. If you want to see the complete data results, these tests (plus those for every other tested person in the “Coffey DNA Project”) are included in either of the following: (Both have the same material, the “pdf” may be better for printing.)

http://www.coffey.ws/familytree/dna/CousinsData.htm

http://www.coffey.ws/familytree/dna/CousinsData.pdf

Most of the tested markers show identical values for all of the tested men. The following table shows ONLY those markers that have DIFFERENCES. Markers that were not tested for an individual are indicated with a “-“:

Here, the markers that are DIFFERENT versus the majority are shown in pink. (Some of the DYS numbers show markers in groups of 2, 3 or 4 – just look for any number within the group that has changed.)

First, note that in most of the cases, each “difference” only occurs in ONE of the tested men. This very strongly suggests that in each such case, that represents a mutation that only occurred in that ONE line, AFTER the various lines had separated. Therefore each of these does not reflect any change in any common ancestor. Given that, we should be convinced that the MRCA of the group, in each of these cases, must have had the most common allele value. The yellow band at the bottom shows my assessment of the value the shared MRCA must have had. (Note that in the table I showed differences in markers 73 through 98, but have not colored any. There were not enough people tested for those markers to be comfortable postulating what value the original MRCA held.)

In the end, there were only three markers that reflected a difference in more than one of the lines – Markers #21, #24 & #49. Consider Marker #21, where two people have a value of “30”. While it’s possible that this difference arose via two SEPARATE mutations in these two lines, this has low probability. Where possible, it is always more probable that it reflected a SINGLE mutation in a common ancestor. It thus quite probably marked the time that one of the sons of our overall MRCA (Most Recent Common Ancestor) had a single mutation. There’s no way of knowing if the MRCA had an allele value of “29”, which mutated in that one son to a “30”, or vice versa.

I have tried to construct a “tree” showing how a sequence of mutations in early ancestors could yield the above final result. I can do fairly well, but I keep tripping over the various “missing” markers. However, for what it’s worth, here’s one that would work IF Dkehoe had a “22” at Marker #49:

In the above, Marker #21 mutates first, and splits off GKaho and PKehoe lines. Then #49 mutates and splits off RKeogh, JamesKehoe, and DKehoe. And finally #24 splits off RKeogh.

Looking at the above, the Edward and Peter lines indicate very minimal mutations. But in part this is because we had already backed out all of the mutations that occurred in various lines in the 300 years since E & P arrived in America, selecting representative individuals who matched the immigrants’ DNA. We didn’t have enough samples to do this for the Kehoe Group.

It also looks like “DCoffey” has the DNA most consistent with the MRCA – if we ignore the missing data values.

Need to be careful about drawing conclusions about the men tested on Ancestry. Those tests involved fewer markers, and if we had all the same markers we would likely see more mutations in those two lines.

WHEN did all this happen? We don’t really know, but we can put boundaries on it. We KNOW that our Coffey/Coffee lines were separate from the Kehoe group by 300 years ago, because Edward and Peter had arrived in America by then. And we can observe SO MANY differences in the five Kehoe lines, that we can be pretty sure that the FIRST of their mutations occurred much more than 300 years ago.

At the other extreme, our testing service (FTDNA) has a “Time Predictor” (TiP) that does a probabilistic analysis on how long it takes for differences to accumulate. This seems to suggest a 90% plus probability that we all began to split NOT MORE than 700 years ago.

So a mid-point guess might be that our shared MRCA lived about 500 years ago?

And WHERE did this all happen? Well, the Kehoe group believe that their homeland is County Wexford. And the nearest heavy concentration of Coffey families in Ireland is Tipperary – which analysts have identified as also a probable Coffey homeland. Only County Kilkenny separates them, and Don Coffey’s ancestor comes from there. This suggests our combined family origin may be in the Tipperary/Kilkenny/Wexford area?

Miles Kehoe, the Administrator of the Kehoe Project, has done some study of Gaelic names. He concludes that Coffey/Coffee/ Keogh/Kaho/Kahoe/Keough could have readily originated from a single original name. Possibly the ones who ended up in Kilkenny came to prefer the variation that became “Coffey”? Or possibly some ancestor made a conscious decision to change his name slightly, perhaps to better fit into a new location?

COFFEY CONNECTIONS: The “Wilson” Connection

The above talks about the Coffey/Coffee and Kehoe connections. But there is one more curious match of “Coffey” to another family. There is a good DNA match to a “Wilson”. And since there is no overlap of the two families in America, the connection was probably back in Ireland.

And we can now examine this match in the context of the above discussion. This Wilson DNA match very clearly has the characteristics of the Edward Coffey ancestral line (i.e., Marker #16 is “10”), with two other differences (Markers #35 and #55) that match no Coffey or Kehoe (i.e., the mutations appear to have occurred after the “Wilson” line was separate). The Wilson does NOT have the Marker #30=17 indicative of the Peter Coffee line.

“Wilson” is “An English and Scottish Surname associated with the 16^th and 17^th Century settlement in Northern Ireland”. The tested Wilson says his ancestry is Scottish, but that his ancestors apparently travelled back and forth between Scotland and Ireland.

It thus would seem most likely that there was a relationship between a Wilson female and a Coffey male in Ireland, after the Edward Coffey line became separate from all the others. The details will never be known.

COFFEY ORIGINS: Deep Roots

(Again, my personal understanding of this topic is weak. I would be a lot more comfortable if Tim Peterman, Ed Southwell, and/or Miles Kehoe undertook to write an article.)

Anyway, there is another type of y-chromosome DNA study, that groups people into “haplogroups”, indicating a common ancient genetic origin. These studies show that humans evolved over time into a number of groups, all starting from a “y-chromosome Adam” that lived about 60,000 years ago in Africa.

About 25,000 years ago a branch called “R1b” had separated, that became the most common population group in Europe. Additional splits from R1b have been documented to date, and more continue to be defined. Both the Coffey and the Kehoe lines have done such testing, and since they are clearly one family, they clearly have the same haplogroup. At the present level of test depth, Coffey/Kehoe are members of a larger group “R1b1a2a1a1b5” (Now more commonly called “R-L176”, for the final tested SNP.)

Archaeologists know that Ireland was settled in several waves during the last 2000 to 4000 years. The present tests on our Coffey/Kehoe family suggest OUR wave may have come through Iberia or Southwest France. Future tests may confirm or refine this.

Where will this lead? Ed Southwell offers a thought: “The results represent an intriguing possibility…..according to ancient Irish legends, one of the founding peoples of Ireland were the Milesians, who were in Iberia prior to arriving in Ireland.  What if there is some truth to this ancient legend and a small percentage of modern Irish people still preserve this ancient lineage, including our Coffey (and Kehoe!) family?  It is fun to think about, and who knows…. perhaps further DNA testing of the population could confirm it!”