The first thing we must understand in reading the results of our test is that Relative Genetics (otherwise known as "the lab") deals largely with averages and presents most-likely scenarios based on the available samples alone. There is often more than one interpretation of the DNA data, and further samples would help to fill in the picture and fine-tune the interpretation. Furthermore, the DNA findings are insufficient in themselves, and must be augmented with traditional genealogical documentation.
For example, when the wording of the report says that donor "V" may have a William-line ancestor but not Simmons as V believed, that's strictly a statistical possibility. If V's documentation is strongly in favor of a descent from Simmons, then he has no reason to discard that, just because the statistics initially point elsewhere. The most-likely isn't necessarily what actually happened, and the scope of possibility is still wide enough that a Simmons descent is still quite possible. We see from the other samples that V and the other 2 Simmons donors are still in the same general pool, as Simmons has proven to have what I'll call the "Fitzhead fingerprint." (More on his case later.)
So the DNA evidence doesn't necessarily trump one's solid genealogical evidence; the latter is just as important as ever. But the DNA evidence may point to cases where there must be some unknown "fatal error" in the genealogical evidence, as in the case of the 3 outlier samples. They do not show the required similarity to either the Fitzhead or Rockhold fingerprints. On a happier note, the Thomas branch samples are ruled out as having relation to the Fitzhead Rockwells, but match perfectly with the Rockhold group.
On reading the haplocation diagram, we need to understand that there's a standard protocol of putting together samples which show the same pattern, as in the central 11-sample circle I'm referring to as the "Core," and how to link up the other circles according to displaying identical mutations. So a mutation occurred on a certain site or locus labeled DYS439 among the common ancestry of donors K, L, and N; otherwise their DNA pattern is the same as those in the Core. All three are John-liners if we include sample N of the James of Scioto branch. It is still possible, however, that James of Scioto could be a William-liner and that one of his immediate ancestors experienced a mutation on locus DYS439 which resulted in the same reading. So the DNA evidence does not rule out the traditional pedigree current among some of James' descendants; but the more likely explanation, statistically, is that he was of the John line, as argued by H. Earl Close's proposed pedigree.
Now note that sample X shows a mutation on the same locus, and he's a John-liner. So why isn't he mapped into the same circle as K, L, & N? Because the value read at the locus is different, both from its value for KLN and for the Core. Likewise with sample C, whose genealogical data points to him being a William line. He has a fourth value at DYS439. [If by now you're wondering what these values are, or what a locus is for that matter, please be patient; and there will also be some explanation in the Relative Genetics report, and I'll dedicate a separate report to explaining the details.]
But while X appears differently than KLN, the form of the graph doesn't mean X is more distant from them than samples in the Core. In fact, according to our pedigrees, X should be the closest relative in the pool to my own sample, identified as "P." Both of us are descendants of Jeremiah Rockwell the Revolutionary soldier. This is where the genealogical evidence comes in to help explain what may be going on in the DNA data. A very possible explanation, according to Southard, is that DYS439 mutated TWICE in X's ancestry, first early on with the generations he shares with other John line samples such as K and P, and then a second time more recently, resulting in the current value. If this diagram were a genealogical chart, you'd expect a line to connect X not to the core but to KLN adjacent to P. But that's not the diagram's purpose; rather the lab charts out strictly what they see, as if it were one mutation removed from the core. The DNA data cannot reveal whether one mutation occurred or two, so they don't assume it while mapping out the samples. But our genealogical data can interpolate that such a double mutation may have indeed occurred. This demonstrates how genealogical documentation remains crucial to DNA interpretation.
Similarly, G is mapped as connected to the KLN circle because of the same value at DYS439. But this doesn't mean he's more closely related to John than William and Josiah. Indeed the 3 further mutations show that his common ancestor with the rest was probably back a few generations before the common ancestor of William, John, and Josiah. G is still enough alike to suggest the common ancestor came from the same neighborhood in England, and that G's line came to this country by a more recent immigrant. That line may have settled in an area where a family from the William line lived, since G's pedigree shows descent from William. Some confusion of genealogical data is suggested by the DNA results. The Rockwell Family Foundation's volunteer researchers may be able to help by examining G's genealogical data for points at which the confusion occurred. As for the common value for DYS439 between G and the KLN circle, this probably arose from an independent mutation event which happened to result in the same reading. (On the other hand, if G's pedigree proves solid, this means his line has been undergoing rapid mutation. G's pedigree would make him the only representative in the test of the descendants of John the grandson of William of Windsor via Samuel. If further samples from this line show some of the mutations found in G's sample, this would support G's pedigree over the above "later immigrant" hypothesis.)
Now I return to the 11-sample Core. William is well-represented here, along with a couple of Josiah samples, demonstrating the close links between them. But Simmons is here as well, suggesting superficially that he appears more closely related to William and Josiah than to John. We didn't expect him to appear here at all, due to the current understanding that he was an adoptee-yet he ended up clearly a Rockwell showing the Fitzhead fingerprint.
The report was worded to suggest he could be a descendant of William. This was based solely on the data values in the absence of much information on Simmons at the lab. So I discussed this with Southard, reviewing Rockwell geography in Connecticut, how the three old lines had their distinct regions, with John's line in Fairfield County where Simmons grew up. Adding this genealogical information to the DNA data, Southard revised here interpretation and said that Simmons is more likely a John-liner. Once again, DNA data is insufficient; it needs to be interpreted through genealogical knowledge. Furthermore, if Simmons is a John-liner, he could be expected to be of the Joseph branch which settled in Simmons' home area of southeastern Fairfield County (Norwalk, Wilton, Weston, etc.) as opposed to the Thomas or Jonathan branches, which concentrated in the Ridgefield and Danbury areas.
Now back to the DNA data: note how so many John samples show the DYS439 mutation, while Simmons does not. Since those John samples all have pedigrees showing Jonathan as their closest ancestor, this suggests that the mutation occurred early, either with Jonathan himself or before. If before (e.g., with John of Stamford or his father), then all the John liners should display that mutation. And if Simmons represents the Joseph line, then we have isolated the point in time when the DYS439 mutation occurred: with Jonathan. More samples, particularly from known John lines on the Thomas and Joseph branches, would help to confirm this, if they lack that mutation. Barring other mutations, they would display the same readings as the 11 samples in the Core, and would be mapped there.
Indeed, we may already have such a sample in M, if his ancestor John of Butternuts was the son of Uriah, a John-liner on the Joseph branch, as one construction of his pedigree has it. But note that if Simmons is of the Joseph branch as well, he and Uriah would be close cousins, and it may be impossible to resolve the issue of whether John of Butternuts was a son of Simmons or Uriah. That objective of our test was based on the expectation that Simmons would display a completely different pattern than the Fitzhead Rockwells-but that prediction proved invalid.
There is still more that can be done on this question, however. If a sample were found representing a line from Uriah's father John or grandfather Jonah, and it displayed a mutation on any of the 23 loci, this reading should also appear in any true line of Uriah's. That would show John to be more likely the son of Simmons, but almost certainly not the son of Uriah.
And those aren't the only alternatives. Since M also matches various William and Josiah branches, John of Butternuts could possibly be some wandering offshoot of one of those families who happened to settle in the same area as Uriah, where the land records show frequent interaction with Uriah's family. This alternative is just as possible, based on a strict reading of the DNA results. But the genealogical data again helps to interpret the findings and say it's less likely than that John of Butternuts was a Joseph-branch John liner, either through Uriah or Simmons.
The argument I have made about John of Butternuts may also be applied to John of Stamford. The project sought to see whether John of Stamford was related to William and thus could be the son of William's brother John, as Rockwell researchers have been hypothesizing for nearly twenty years now. The results tell us that, yes, this is entirely possible-but it doesn't prove it. If the Joseph and Thomas branches share the DYS439 mutation found in Jonathan's descendants, that would show that John of Stamford had it as well (and throw a monkey-wrench into the above Simmons argument). It could have arisen with John of Stamford or with John of Windsor (who has no other direct male lines to test)-or it could have happened a few generations further back. But not too far back, as the Jonathan samples show a match with William in all other respects, telling us that John came from the Fitzhead family as well.
Alternatively, if John (via samples through Joseph and Thomas) proves to be identical with William and Josiah, he could be the son of John of Windsor-but he could also be the son of Richard, as might Josiah. On the other hand, both John and Josiah may be close cousins to William. Again, the common generation wouldn't be likely to be too far back, or another mutation would likely have occurred. More than one interpretation is possible with identical DNA samples, lacking further genealogical documentation. The most important finding that would address this question would be to locate Richard's family in England. (He is believed to have lived away from Fitzhead as he isn't found there as an adult.) Record of his children might turn up a Josiah, a John, or neither. But based on the will of Honor Rockwell, which testifies that some child of Richard's (male or female we aren't told) did live in New England. So we may maintain the hypothesis that John of Stamford is the son of John of Windsor and Josiah is the son of Richard-but we can't say it is absolutely so on the basis of DNA data alone. But further DNA samples will shed light on many other questions, such as the Simmons puzzle. Further samples on under-represented lines (Thomas and Joseph branches for John, John branch for William, and any line from Josiah) will be particularly valuable.
I hope the above discussion is clear enough. If not, feel free to ask for clarifications. -Ken Rockwell
Return to DNA Project home page