As much as DNA can help…
As much as The Legal Genealogist is a total DNA geek, and as much as DNA must be considered whenever any of us sits down to map out a research plan, we are going to be disappointed if we delude ourselves into thinking that DNA is always going to help us get to an answer.
It’s simply not a magic bullet.
That lesson was drummed home again yesterday when I was privileged to attend a chat session of serious genealogists who study articles published in the National Genealogical Society Quarterly and then discuss them — what worked, what didn’t, and how things might have been better.
The article that was yesterday’s focus was mine: “‘Don’t Stop There!’ Connecting Josias Baker to His Burke County, North Carolina, Parents,” NGSQ 99 (March 2011): 25-41.
A family Bible in the possession of descendants in Texas had said that Josias Baker was born 1 April 1787, and in 1815 he had married Nancy Parks, born 20 August 1791.1 And it recorded his and Nancy’s deaths in 1871 and 1878.2
Both Josias and Nancy left wills — Josias in Ellis County, Texas (he actually died in 1870, not 1871)3 and Nancy in Dallas County.4
But neither the Bible nor the probate files — nor anything in the family’s possession, for that matter — had given any hint to Josias’ parents. Where had the Bakers come from?
In that article more than five years ago, DNA wasn’t used at all. And the question came up yesterday: if I had to do it all over again today, would I use DNA today?
And the answer is the same: no, I wouldn’t.
Here’s why.
First of all, the Bible record noted that Josias had seven children, including two sons — both of whom had died young without marrying or leaving children of their own.5 That means that the easiest way of tracing Josias back to a family of origin — a YDNA test to try to find others in the very large and very active Baker surname project that he might match — can’t be used. There are no direct male descendants to test.
As to autosomal testing, it wasn’t used at the time because it wasn’t available at all when the research was completed — the article had first been written in 2009-2010 — and not widely available even when it was published in 2011. The first autosomal tests from 23andMe weren’t available at all until late in 20096 and the Family Finder from Family Tree DNA not until early 2010.7
And today, the question would be whether DNA testing could really help move the research further towards the goalposts. As one chat participant who’s truly a DNA expert put it, testing at that level would be “crazy, difficult and expensive.”
The reason for that is that the very small handful of known living descendants of the candidate identified for Josias’ father, Henry Baker, would be — at best — fifth cousins to any of the very small handful of Josias’ known living descendants. And because of the random mixing of autosomal DNA down the generations, fifth cousins share, on average, only about three centimorgans of DNA in common with each other — about 0.0488% of their total autosomal DNA.8
That average is well below the matching threshhold for any of the DNA testing companies,9 and as a result the vast majority of actual documented paper-trail fifth cousins will not share enough autosomal DNA to be detected by the kinds of tests we use for genetic genealogy. The odds projected by the testing companies are that no more than 10% of fifth cousins will have inherited enough autosomal DNA in common to show up as a match.10
So we could test a half-dozen different paper-trail fifth cousins — or more! — against known descendants of Josias — and still come up with no match.
That’s why, at that level, we say that DNA can help provide evidence to prove a family tie — but can’t disprove it. Because the reason why you’re not getting a match may well not be because you’re not related; it may simply be because you didn’t randomly inherit enough DNA in common with your relatives to show the relationship.
In short, here five years down the road, I still wouldn’t do anything differently, even with my natural bias towards DNA testing. I wouldn’t turn down testing as an option, and I’d certainly encourage anyone interested to get tested. Considering, in every case, whether DNA testing can help is always going to be part of the planning process.
But when testing is “crazy, difficult and expensive,” and the case can be made without it, then DNA testing isn’t going to be an essential part of the reasonably exhaustive research we conduct to meet the Genealogical Proof Standard.11
DNA isn’t a magic bullet.
The word “reasonably” is there in the Genealogical Proof Standard for a reason.12
SOURCES
- Baker family Bible record, 1787–1878; The Holy Bible (Philadelphia: Jesper Harding, 1846), births and marriages columns; Bible Records Collection; Dallas Public Library, Dallas, Tex. ↩
- Ibid., deaths column. ↩
- Ellis Co., Tex., Will Book A:132–34, Josias Baker; County Clerk, Waxahachie; FHL microfilm 1034623, item 4. ↩
- Dallas Co., Tex., probate cases, miscellaneous probate packet no. 769, Nancy Baker (1879); County Court, Dallas; microfilm 35, Dallas Public Library. ↩
- Baker family Bible record, deaths column. ↩
- See ISOGG Wiki (http://www.isogg.org/wiki), “DNA Relatives,” rev. 25 Jan 2015. ↩
- Ibid., “Family Finder,” rev. 8 July 2016. ↩
- Ibid., “Autosomal DNA statistics,” rev. 25 Aug 2016. ↩
- Ibid., “Autosomal DNA match thresholds,” rev. 4 June 2016. ↩
- See e.g. “What is the probability that my relative and I share enough DNA for Family Finder to detect?,” Family Tree DNA Learning Center BETA (https://www.familytreedna.com/learn/ : accessed 18 Oct 2016). ↩
- See Board for Certification of Genealogists, Genealogy Standards (Nashville, Tenn. : Ancestry, 2014), 1. ↩
- See Judy G. Russell, “DNA and the Reasonably Exhaustive Search,” OnBoard 20 (January 2014): 1–2, 7; HTML version, BCGCertification.org (http://bcgcertification.org/ : accessed 18 Oct 2016). ↩
Great post: i need these bite-sized tidbits of information!
This is a GREAT post Judy, thank you! I’m going to play devil’s advocate on this, because I struggle with this issue a lot.
This is a great post about a really important subject, and I am really, really torn. To play devil’s advocate, I don’t think cost can be a factor that weighs against using DNA in the specific case you mention, since 6 DNA tests would cost less than a trip to an archive or hiring someone to do some research, for example. I think most proofs using atDNA are going to have a MINIMUM of six test-takers and most will likely have MANY more than six.
The amount of DNA might be a reason not to pursue DNA evidence, although I think fifth cousins are pretty close to the border. I know that we all check record types that have much less than a 10% probability of shedding light on our questions. In fact, in some cases I would LOVE to have an evidence source with a 10% probability of finding anything helpful!
If the 6 people said no to testing, that’s a different situation; the courthouse has burned.
So the question is, does this case NEED the DNA evidence (or test-takers saying no)? I’m on the fence. For $500 with a 10% probability of good evidence, I’m having a really hard time saying that there’s no requirement for a consideration of the DNA evidence. I hope others weigh in too.
I used the possibility of six fifth cousins merely as a hypothetical, Blaine. In this case, we actually don’t have that many identified fifth cousins. So perhaps a better question would be: considering this case, and the evidence we had, how far would you go to find fifth or sixth cousins (the level we might realistically hope to locate) — and how many would you feel you needed? And what would you do if you tested six — and none of them matched? If you tested 10 and none of them matched?
Right. For example, what if there were 10 identified fifth cousins, and you’ve tested 5 with no luck. Obviously that doesn’t disprove your conclusion. But do we have to test the other 5?
If it were third or fourth cousins, which have a greater likelihood of success, I would have a much easier time saying “definitely!,” but with fifth cousins I’m struggling. I honestly don’t know what the answer is.
I posit, however, that at a minimum, an article published today relying on fifth/sixth cousin relationships or closer should provide either DNA evidence, or reason(s) why it wasn’t pursued. (test-takers said no, likelihood of success too low, too few test-takers available, etc.). Thoughts on that?
I’m not sure what you mean by “an article published today relying on fifth/sixth cousin relationships or closer.” ANY article that relies on relationships alone darned well better have direct record proof or DNA evidence, and if DNA evidence isn’t used, it better say why not. Where it’s the usual type of article that the NGSQ publishes, where the proofs are gathered from many sources, then I’d probably draw the line a generation closer: fourth cousin or closer, I’d want to know whether DNA was considered and, if it wasn’t used, why not. Fifth cousin is just too much of a crap shoot.
Sorry, I should have said “an article published today where the test-takers are fifth/sixth cousins or closer…”
I’m in total agreement with 4C or closer. I’m still on the fence about 5C and 6C, but I think maybe that’s the best we can do for now in these early stages of DNA evidence.
This was great, thank you!
I would agree that this may be the best we can do at this point, Blaine. I am 100% in agreement that DNA should be considered in every case, to determine whether it could be determinative, merely helpful, or not helpful/available/probative. In those cases, determinative would mean, to me, required; merely helpful, to me, would mean it could be foregone depending on other circumstances; not helpful/available/probative would mean, to me, it wouldn’t be necessary.
Great discussion, Judy & Blaine! Not only informative but I love that I “hear” each of your posts in your own voice. (And even Blaine’s little chuckle after this remark.)
Can we explore the “in total agreement with 4C or closer” comment? NGSQ articles deal with difficult research questions where there generally is no direct evidence answering our question. I totally get considering DNA as part of our reasonably exhaustive research on that kind of “no [reasonable] stone left unturned” problem. Fortunately, for many 4C and closer relationships in my tree, lots of direct, non-conflicting, readily available evidence exists to prove those relationships. Are either of you advocated that if the case CAN be made without it, we MUST include DNA evidence anyway for 4C and closer?
Thanks for your input.
Nope, not in my book. That would put it into the merely helpful category and, to me, would mean it could be foregone depending on other circumstances. But in that case I think the author should say why it was foregone.
I agree, although in addition I would say that it’s going to have to be a good reason. Just as it would have to be a good reason why you didn’t check the other obvious sources like census records and vital records.
And, I’ll add that many of our relationships based on paper – even the best paper trails ever constructed – are 100% wrong. Misattributed parentage is everywhere, and in many cases, only DNA can detect those events. Everyone thinks it didn’t happen in their tree, but trust me (!), it did. For me, that pushes DNA more toward the MUST category in many cases, especially 4C and closer where it is more definitive.
And I did chuckle! You know me too well!
Thank you both for your responses. Someone reading my question might perceive it as a reluctance to pursue DNA testing. You both know me better than that. I am a strong advocate of DNA testing and manage a lot of kits. (Maybe not a lot compared to either of you but still a large number.) But if we are leaning toward the MUST test category in many cases, and the required number of testers being “MANY more than six,”consider the impact on the complexity and page count of our proofs. If I have to prove how ten living people connect to my ancestor back five generations, explain what DNA tests were performed and why, provide the details of the results, and an analysis of what those results mean in order to make the statement, “the DNA test results are consistent with the other evidence in supporting the parent/child relationship” in my KDP, meeting the page limit might be very difficult. I guess I better start practicing.
Chiming in with a question after Skip’s KDP question. Currently for a KDP or case study, is DNA testing expected by BCG judges if applicable to a problem?
DNA testing still isn’t the norm, Jane, but that doesn’t mean a judge might not look for at least an explanation of why DNA wasn’t pursued — and at least one BCG trustee, writing in OnBoard, has taken the position that “standards and the mainstream availability of DNA testing strongly suggest that DNA testing be considered as one possible source in answering a genealogical question. That DNA testing should be considered, however, does not mean that it must be used, or that research that does not use genetic testing cannot meet the GPS.” Judy G. Russell, JD, CG, CGL, “DNA and the Reasonably Exhaustive Search,” OnBoard 20 (January 2014): 1–2, 7.
I don’t believe there’s a “MUST TEST” rule when six or more tests would be needed, Skip. That’s pushing the envelope, I think. I wish the science had advanced far enough that we could say what had to be there, but we can’t just yet. It has to be a rule of reason for now (and probably forever).
We need to update the Wiki page on the expected amount of sharing because the mathematical averages obtained by halving with each generation are not actually representative of the expected amount of DNA shared with cousins. Although the odds are loaded against having a match with a fifth cousin if we do have a match at this level we are likely to match on a much larger segment than 3 cMs. We don’t have good data on the expected level of sharing at this level but Paul Rakow recently did some computer simulations which provide some guidance:
https://drive.google.com/file/d/0ByiUck4kCwhnWUlWamQxcjdaMUk/view
From his work it seems that if we do have a match with a fifth cousin the segment is more likely to be over 7 cMs in size. At present we can’t accurately detect matches under 5 cMs without extensive generation-by-generation chromosome mapping. However, if were to have a legitimate match on a 3 cM segment in 90% of these cases the match is likely to date back beyond 10 generations. See this chart: http://isogg.org/wiki/IBD#mediaviewer/File:Speed_and_Balding_IBD_distribution.jpg
The range of values reported in Blaine’s Shared cM Project (http://thegeneticgenealogist.com/2016/06/26/update-to-the-shared-cm-project/), Debbie Kennett’s statement above regarding “the mathematical averages obtained by halving with each generation are not actually representative of the expected amount of DNA shared with cousins,” and many of the things I am seeing in my own family research, make me believe we need MUCH more published research incorporating well documented genealogies and DNA findings. Over the last sixteen years, many of our recommendations have changed as we see more real life DNA findings do not match statistical predictions. I am not convinced we will not be able to use DNA for those more distant cousin relationships in the future. As we learn more and have more data to compare to from more test-takers, we may be able to go farther than we can today with certainty. What is difficult to do today may become much easier and surer. I strongly encourage everyone using genetic genealogy to publish more on its use as evidence for relationships third cousins and beyond. Until we have real data we will all be guessing whether or not it can really be helpful.
Where the data exists, I couldn’t agree with you more. The question here is, would we need to go find the data beyond the research already done in order to add to the argument? Or is the argument good enough on its own?
May I revisit the original case and ask a related question? If there is a female ancestor with two granddaughters and a great granddaughter still living who might consent to testing, how does this change the likelihood that DNA testing might be useful to discover the ANCESTOR’s parents, grandparents, or cousins from the various databases? As mentioned in the blog chain, this ancestor’s paper trail clearly includes invented history so tracing her family has dead ended.
For the granddaughters we’d be talking about potentially locating great grandparents (parents of the grandmother); for the great granddaughter, 2nd great grandparents (parents of a great grandmother). That’s well within the range for autosomal testing, and sure worth doing.