Herriott Surname DNA Project
A 5,000-Year Tree for (Most of) Our Family Lines
By Scott R. Herriott
Summary of the DNA Analysis in the Last Two Articles
In the December 2016 article, we reported the results of an attempt to use DNA data to show, in a single family tree, the relationships among all the Heriots,Herriots, Herriotts, Harriots, and Harriotts whose male-line DNA is clearly Scots.
The June 2017 article contained a detailed explanation about the two types of DNA analysis that we are now using: the STR analysis that we began ten years ago and the SNP analysis that we turned to only a year ago. We noted that if historial research is like looking back in time with our bare eyes, about 500 years, then STR analysis is like looking with opera glasses, seeing 500-2,000 years back in time, and SNP analysis is like using a telescope, looking back 1,000-50,000 years. The June 2017 article showed a family tree for humankind that goes back 60,000 years.
The SNP tests give a very clear picture of where different family groups lie on humankind's family tree, because a SNP is a type of change to DNA that happens very, very rarely, and if it happens, it is very rarely reversed. So a SNP change is essentially a permanent change to the DNA that is inherited by all subsequent generations. The accumulation of specific SNP changes shows the path that a person's ancestry has taken down humankind's family tree. Genetic researchers discover a new SNP by noticing a common genetic change within a subgroup of people who are otherwise (via known SNPs) identical. For any particular line in the human family tree, researchers may not have yet discovered SNPs that distinguish small branches of the tree, recent family groups close to the present in time. That is why, even if one pays a lot of money for a "Big Y" SNP test, that test may not yield known SNPs closer than a couple thousand years from the present. The smaller, and cheaper, SNP tests give results that are at best a few thousand years old.
The result of a SNP test is presented as a series of SNPs that define the smaller and smaller (more and more recent) branches of the human family tree down which a person descended. Every branch-point in the human family tree is defined by a SNP, and each SNP has been given a code name such as P25 or M269. Unfortunately, the code names are not related to each other in any way that shows which SNPs follow others down a branch of the human family tree. One has to look at certain web sites (e.g. www.yfull.com) to see which SNP is above and which others are below it.
For many of the older SNPs, genetic researchers have been able to estimate the century in which the SNP change took place in the DNA molecule, so some SNPs are presented with a date (century) for the time to the most recent common ancestor of all people who have that SNP.
To see the twigs of the family tree that are closer to the present in time, we have to look at parts of the human DNA that have a faster rate of change. Those are the STR changes (single tandem repeats), where we can see changes back in the range of 500-2,000 years. Beyond a few thousand years, STR changes that happened a long time ago can be reversed by later changes, masking the original change and confounding the analysis of family tree lines.
The period of 1,000-3,000 years ago is a time where the STP and STR analyses overlap. Part of our research in the Herriott Surname Project is to use STR data, of which we have a lot from the past ten years, to give us a good idea about the haplotype (SNP) of a family member without having to spend a lot of money on a Big Y SNP test.
SNP Analysis of the Various "Herriott" Family Lines
In this article, we focus on the branches of the surname that are of Scots origin, not Scandinavian. This excludes the "Belfast" Herriott/Heriot/Harriett group, the Berwickupon-Tweed Herriott/Harriott/Heriots, and a Trinidad Harriott. It still leaves us with 10 family groups of Scots origin: the "New Jersey" group who are descendants of the 1685 immigrant David Herriot, the Heriots of Longniddry whom we believe to be related to George Heriot (founder of the Heriot School in Edinburgh), a group of Herriots from Berwickshire in the UK, the Herriots from East Lothian, a pair of "New York" Harriott/Harriets, and five individuals who each seem distinct: a Harriott from Jamaica, a Herritt, Michael Herriott of the UK, John Herriott of the UK, and a Harrod.
The cost of a SNP test depends on how extensive the test is. Smaller, cheaper tests look only at the common SNPs that define fairly old branches of the human family tree, from 5,000 to 15,000 years ago, so they don't help us discriminate among branches of the family tree that diverged recently. The most complete test, called BigY by Family Tree DNA, costs $575 per person and examines about 25,000 SNP locations on a person's DNA in a shotgun approach. Through sheer volume, it is more likely than any other to find a recent SNP that narrowly defines a person's haplotype.
The tests we were using for the past ten years, the STR tests, use a combination of slow-changing and faster-changing DNA sites. The 67-marker STR test includes one site that is very slow to change (about once ever 5,000 generations) and that turns out to be a marker for a branch of the family tree for Western Atlantic Europe peoples called U106. So our 67-marker tests can tell us whether a person is in the U106 branch or not. We noticed that the New Jersey group is in U106, but we have not found any other of our surname groups that are.
We have BigY reports for the Longniddry Heriot group and for the Jamaican Harriott. A small SNP test that had been ordered a few years ago by an East Lothian Herriot showed that they are in the branch called M222 (from 1900 BC).
The Longniddry Heriots BigY test got positive results down as far as haplotype BY20462, but that is down only four SNPs from the L21 "Scots Modal" of 2400 BC. We don't have dates on any of those four SNPs.
The Jamaican Harriott BigY test showed that line to pass through the L1065 "Little Scots Cluster."/p>
Because we knew that the New Jersey group is in U106 (from 2800 BC), we bought the U106 SNP package from Family Tree DNA to get a more refined set of SNP tests down the U106 branch.
Those are the only formal SNP test data we had. But through our participation in the surname project of Family Tree DNA, we got access to databases of STR data (which we have) for groups defined by SNP branches. We noted above that STR data can change over time as additional genetic mutations take place at a particular STR site on a person's DNA molecule. STR tests cannot look very far back in time but if we can find a person in a particular SNP database whose STR report looks fairly similar to one of our own, then we can make a fairly good guess that our person has that SNP type. This is the (cheap) approach we took with several of our other lines.
It worked very well for the UK Michael Herriott STR report. We found a match in a database of people who belong to the L21 "Scots Modal" haplotype (from 2400 BC), and the match was so close (only 8 differences across 67 STR markers) that we feel very confident that our UK Michael Herriott is on a branch of the human family tree defined by the sequence of SNP markers L21 > DF63 > CTS6919 > A92 > Z16506 > Y674 > A7803. The only SNP in that sequence that had been dated was CTS6919, from 2100 BC. The last-known SNP in the sequence is called the person's haplotype, which here is A7803.
Our analysis of STR data within SNP databases also worked out quite well for the East Lothian Herriots. We looked within the M222 database and found a few people who were fairly close (19-22 changes in a 67-marker report) from our East Lothians, so we now guess that their haplotype is PF2028, which dates to 1100 AD.
We were not as lucky with UK John Herriott. We looked in the L21 database for STR reports similar to his, but the closest we got was a distance of 50 changes, so we doubt that he is in L21, and we don't yet know where else to look.
For the Berwickshire Herriots, we have only 37-marker STR data, and we had only a very old guess about their haplotype (P312, from 2500 BC). We looked in the L21 SNP database and found some entries that were a distance of 20-30 from Berwickshire's STR report, but that's on only 37 markers, so we don't have much confidence that we have located their haplotype.
We have not yet investigated the New York Harriott/Harriet group, because we would have to extend their 37-marker STR test to 67 markers and probably run some small SNP tests even to get some old SNP branches. Nor have we investigated the unclassified Herritt for whom we only have a 37-marker STR report.
See the chart for the family tree for the Scots "Herriott" surname groups (SNP branches), as best we know it now. Interestingly (sadly?), the New Jersey group is quite distant.