STRs are a big, tricky, and somewhat misunderstood topic. As I understand it, they were chosen over SNPs in the early days for several reasons. The SNP catalog was small. Only very ancient SNPs and their placement had been uncovered. They could not genetically define a lineage over the historical timeframe. Because STRs are long strings of repeating letters, they were quickly identified and had the benefit of being fast mutating. In other words, they would quickly evolve over the course of one or two thousand years, which allowed for the identification of branching, even within the last few hundred years.

STRs are comprised of short strings of genetic letters that repeat X number of times withing a given region. For example, I have 34 repeats of TTTC in the region known as DYS449 (allele position 21). It's likely that close relatives will have the same values.

However, the STR values can go either up or down. The next illustration demonstrates that the modal at that position is 33. Somewhere in my lineage, an ancestor gained an additional repeat of DYS449 while others lost a repeat. This makes STRs unreliable for a purposes. Yes, they will trend in a certain direction as they evolve and patterns will emerge. But you can't count on them with any certainty. Still, individual Y-STR results can be compared to one another and some insight can be gleaned. Just don't bank on it.

The following data was last updated on 17 Sept 2020. Click the image to go to the table.

This table is much the same thing but only those testers having 111 markers are looked at. Here we do some some patterns. For example, three Cochrans share 18 repeats at allele position 32. It should be noted that the relationship between this known. They're relatively close and one of them has been tested for the downstream YP5242 haplogroup. Likely, all three are of the same subclade.

But comparing individual STR markers can take us only so far. Genetic Distance (GD) is more representative of how the limbs of the tree hang in relationship to one another. GD is simply the number of differences between to sets of values. Here, each tester is compared to the modal.

The top line of markers (in gray) is the modal, which represents the most common marker found among the testers (not the average value). Theoretically, the modal reflects the expected values for the Most Recent Common Ancestor (MRCA) of all testers. But it can be accurate only to the degree that we have tested descendants. Should we ever dig YP4248 Man out of the soil, we should expect to see something rather different.

The highlighted markers on each line illustrate where the testers show a difference to the modal. The parenthetical numbers in blue are the total Genetic Distance (GD) between each tester and the modal.

Click the image to enlarge

A tester who might belong to the project can be compared to the modal (at top of the table) and to the min/max values (at the bottom of the table). For example, this tester is probably not YP4248. If he were positive for the SNP, however, the bar would be raised quite a bit.

Here the GD to the modal for each testers is compared those values among all testers. The kits are arranged this time based on the value in each cell, so the kit order varies from that in the table above. This is especially useful in predicting where someone might be placed on the SNP tree once advanced SNP testing is completed. Like results start clumping together by surname and subclade.

Here's a colorized version. The GD is simply replaced by a color value. I suppose this borders on the silly, but patterns readily emerge. The bright box at the lower right represents the Cooleys and Whitfields (YP4491). This strongly suggests large genetic and genealogical gaps above and below the subclade. Finding the right testers might fill those out. Other boxes could be just as telling, for example numbers 26 through 32. There are bands that appear to be out of the place. For example, 19, 24, and 25 should most likely be clumped together. But that's just the way the sort worked out. A better algorithm might come along, but better to just go along with the sort than to have to remember to revise the order each time the scripts are run.

And, again I point out, this is derived from STR values. They're not as precise as SNPs and can vary within lineages. They're best used in adjunct to SNPs.

The first thing to notice about the color version is that each side of the diagonal line is a mirror of the other. In other words, there's double the data. For the following, I removed the bottom half and turned the whole thing on its side, resulting in pyramids rather than squares. It's not coincidental, of course, that it looks a bit like the SNP tree at top, except that the hapolgroups are moving in the opposite direction. That's just the way the data is independently sorted.

This is a lot more ambiguous than the SNP tree. And that for the simple reason that we're using somewhat finicky STRs as opposed to stable and discrete SNPs. Certainly there are additional pyramids to flesh out, but STRs won't accomplish that. Only advanced SNP testing can help clarify questions of descent. STRs are the rough sketch. It's brought into form by SNPs.

It should be noted that the Cochrans are largely over-represented in this study. However, in the SNP study, the Ulster Cochrans presently have 13 novel SNPs (see the SNP tree, above). Quite a bit of parsing can yet be done. Perhaps that will be accomplished with as few as two Big Ys. They should be carefully chosen, though. The STR matrix might point the way to finding testers who are close, but not two close, to the two current Big Y testers.

Otherwise, it would be helpful to do some single SNP testing to confirm placement ($39 at FTDNA and $18 at YSEQ). FTDNA, however, has SNP tests available for this tree only for YP4248 (the top of the tree) and YP4491 (Cooley-Whitfield). They've stopped, for now, adding to their catalog.

Most of the other names, especially Hackett (and I'm sure there are more to come), are under-represented. The exception is the Cooley-Whitfield block, which is well defined. Still, those gaps above and below YP4491 need to be expanded. We need to find a way to expand our reach.

1

2

3

4

5

6

8

7

1. Speculation. Landing of YP4252, 200 BC.
2. Confirmed. Yorkshire 1655 (Story BY30796).
3. Confirmed. Derbyshire 1744 (Hackett YP4253)
4. Speculation. Birmingham c1750 (Cooley YP4491)
5. Confirmed. Ireland 1700s (Cochran YP5244)
6. Speculation. Renfrewshire 13th century (Barons Cochran).
7. Unconfirmed. Unk Scotland 19th century (Cummings BY27664).
8. Unconfirmed. Sutherland c1761 (Gray N/A).

YP5007

YP4253

Keep in mind that Great Britain is roughly the size of Minnesota and is 600 miles long. For hundreds and thousands of years, Britons have lived within a few hundred of miles of one another. Travel was not uncommon. Over the preceding thousand years and more, YP4248's descendants could have spread out from any point on the map.

But if we were to determine the "center of gravity," so to speak (and I don't really know how to do that), it's likely to be at or near the Scottish/English border. These placements are not likely to be totally accurate. They're derived from the extant data. The landing estimate for YP4252 comes from Hunter Provyn's phylogeographer.com. However, the results at FTDNA tell a different story. YP4252 has two known subclades, FT33200, to which our YP4248 belongs, and YP5598, which includes a Scandinavian-born descendant. FT33200 itself has two subclades, including YP4248. The other, BY63466, is presently comprised of two Scandinavian testers. The data is too small for making conclusions. Still, it appears most likely that YP4248, one of its member SNPs or an immediate parent SNP, hitched a ride to Britain, likely across the North Sea to the east coast, within each cell of its host. We can call the immigrant Björn.

I've barely started on this. More is to come. Help will be welcomed.

Hackett

Hackett Marriage Record Click for larger copy