Posted by Sappho on March 17th, 2013 filed in Genealogy
The faded photo shows a house in Bowling Green, Kentucky, the home of my great-great-grandmother, Carrie Burnam Taylor. Carrie Burnam Taylor, prominent dressmaker and entrepreneur, started her own company in the 19th century, when few women ran businesses. At least one thesis has been written about her, some of her designs are preserved at Western Kentucky University, and her portrait hangs in the Kentucky Women Remembered exhibit in the Kentucky State Capitol.
Nearly a century after Carrie Burnam Taylor died, her great-great-great-granddaughter, my niece, showed my mother a few sketches of fashion designs of her own. I don’t know if my niece has ever heard of her prominent great-great-great-grandmother, but her own designs, my mother said, showed an unexpected talent.
Has our line of DNA passed some creative spark from one fashion designer to another? Maybe, but at least as likely not. Such gifts seem to me to have at least as much to do with environment and plain old hard work as genetics. Besides, if there was DNA involved, it might just as well have come from my niece’s mother, my Chadian sister-in-law, who knows far more about beauty than most of the family.
I do know this, though. My Jim Crow era great-great-grandmother’s family looks, on the paper trail, as white as can be, even as Northern European as can be, until, in the latter half of the twentieth century, our family breaks that chain and marries every which way. But DNA says otherwise. Carrie Burnam Taylor would have had cousins who look much like my niece, and I’d give her, let’s say, at least 50/50 odds of carrying some non-European DNA herself that joined the family at some time after 1600. For the past few months, I have been on the trail of one of those connections.
M.’s 24 centimorgan share with me is one of the longest DNA segments I share with anyone. Its source remains a mystery. M. and I share no known surnames. I do know this. I have a white Southern line, and M. has an African-American Southern line. We started with not a lot more than that. My Southern line isn’t even, exactly, in the same state as hers. My family goes back to Kentucky. Hers includes Tennessee. So I have been trying to draw a line between Kentucky and Tennessee.
The connection between Kentucky and Tennessee, though, is the easy part. Ancestry.com has abundant records showing links between my Kentucky families and other branches of those families in Tennessee (there are even a few links to branches in M.’s other states). The harder part is the fact that M.’s ancestry seems less well captured by documentation than my own.
The abundant documentation on my mother’s family contrasts with my weak paper trail on my father’s side. On the New England side, especially, birth, marriage, and death records, all helpfully preserved at Ancestry, trace them back to the 17th century, supported by accompanying Sons of the American Revolution applications, to the point where I remarked on Facebook, after following a few of those trails, that every single one of my maternal grandfather’s ancestors seemed to have arrived together on the Mayflower, and then all of them fought in the American Revolution. I exaggerated only slightly. Southern records are a little less thorough than New England ones, but still cover a lot of ground if you are white. So, between the Northeastern colonial lines and the Southern colonial lines, I’ve pinned six DNA cousins to a definite paper trail, and have likely family lines that may connect several others.
African-American genealogy, naturally, has greater record hurdles to cross. I have, so far, found five DNA cousins who have at least one grandparent worth (sometimes more) of Sub-Saharan African ancestry. The ones with white lines all seem to have better documentation on their white lines than on their black lines. One of the reasons for this we all know. Slaves, as property, barely had their names documented at all. US census records go back to 1790, but the slave schedules gave no names, so most African-American families only appear in the census records starting in 1870.
I think it goes beyond this, though. It looks to me as if, even after slavery, my family got better documented than M.’s. And, like my father’s family, where some family stories seem to have been lost when the family was scattered by war, my DNA cousins seem to have, even in post-slavery years, more early deaths on black lines than white lines, in some cases taking away chances to learn family history. This could be chance, since the numbers are small, and any one person has individual reasons for dying young. But it could also be a matter of money. “If living were a thing that money could buy,” goes the old spiritual, “You know the rich would live, and the poor would die.”
So, how do you find a connection, when one side is short on documentation, and the relationship may not, when found, be one that your ancestors would have openly acknowledged?
In my last post, I told what I know so far about the basics of genetic genealogy. Now I want to talk a little more about what tools are available, both the ones that I’ve applied so far to try to connect myself with M., and the ones that may not yet be fully used. There are two sides to this: finding out where the paper trail can be found, and finding out everything that can be learned from the DNA.
Paper trail sources:
- Slave schedules: Some scant information is preserved about slaves in the slave schedule records. This includes, for each slave owner, a record of the sex and age of each slave, and whether that slave was deemed black or mulatto. You can also get some general idea, from the shift in slave records from the 1850 to the 1860 census, how far slaves may have been bought and sold between one slaveowner and another. There were also some “free people of color,” even prior to the Civil War, so there’s a chance that the ancestors you’re looking for may show up in the regular census.
- 1870 census records: At this point, African-Americans are fully featured in the census. In a few cases, it may be possible to get some idea (but probably not an exact one) of which person on the slave schedule matched which newly freed person by matching age, sex, and mulatto/black designation.
- Slaveowner records: Slaveowners had wills that listed slaves, and other family accounts that may give information about their slaves.
- Freedmen’s Bureau records: These have information about the newly freed African-Americans that the Freedman’s Bureau was meant to assist.
- Census, birth, marriage, and death records post-slavery: Pretty much, once African-Americans get to be free, you can do all the usual census and vital records searches.
- Afrigeneas is a web site designed to assist in African-American genealogy. This includes the Afrigeneas Slave Data collection, to which descendants of slaveowners can submit information like lists of slaves in wills.
- Ancestry.com African-American collection: Here you can find little videos advising how to use Ancestry to research your African-American ancestors, and links to collections like Freedman’s Bureau records, slaveowner petitions, records of black Civil War troops, records of Buffalo soldiers, African-American newpapers in the Library of Congress, etc.
- 23andme tools: 23andme tools include Ancestry Composition, which gives you an admixture analysis showing what populations make up your DNA, Ancestry Finder, which tells you what countries are represented on your DNA and includes a list of potential relatives associated with each country, Relative Finder, which tells you your closest cousins (all 600 or so of them, more if you have lots of American colonial lines), Family Inheritance Advanced, which lets you examine the segments you share with your relatives three at a time, a haplogroup mutation mapper that tells you how they figured your mitochondrial or Y haplogroup designation, and various other reports not so relevant to genealogy but just as relevant to why I personally joined the site, related to health and traits.
- Sharing DNA on 23andme: If you share DNA with a potential cousin on 23andme, you do not see each other’s health and trait reports (unless you extend your share to that, which most people only do with close family), but you do see the segment where you share, each other’s Ancestry Composition, and each other’s Ancestry Finder matches, and you can compare the different people with whom you share in Family Inheritance Advanced.
- GEDmatch: GEDmatch gives you access to some other tools, in exchange for making your data public (you can pick a pseudonym, but I don’t count any net pseudonym absolutely secure). The public data is genealogical, though, rather than the health and traits aspect of your genome, so there are a lot more people taking the deal than are willing to open source their whole genome. Tools at GEDmatch include comparing your autosomal or X DNA with anyone else on GEDmatch (the X DNA tool picks up more remote relationships than the X DNA matches reported by 23andme, which only show fairly close relatives), alternate admixture programs, a tool which lets you upload your GEDCOM and then finds people who both share DNA and surnames or locations with you, and a triangulation tool that shows you what cousins your cousins share.
- DIYDodecad: Dienekes Pontikos has written an admixture analysis program that you can install on your own computer, with various calculators, that, using different reference groups, break down your DNA in different ways. There is, for instance, an africa9 calculator specifically for Africans.
- Ancestor Projects: If you have uploaded your raw data to GEDmatch, you can also add it to the sister site Ancestor Projects, and join any ancestor project that you fit. The one that happens to fit the line I’m trying to link to M.’s is called the Tidewater project, and is for people with American colonial ancestry in the Tidewater states of Virginia, Maryland, and North Carolina (where my Kentucky line first landed). There are also some projects (that I’m not part of for obvious reasons) that may be of particular interest to African-Americans: an Afrolatino project, a project for Sally Hemmings descendants, a Melungeon project, etc.
- Interpretome: Unlike GEDmatch, this one analyzes your raw data while leaving it on your computer (no one else later gets to see it later). It tells you something about your health and traits and has its own admixture analysis.
- DNA Tools: A site for adoptees. What it does with 23andme is to process your Relative Finder data into a neat spreadsheet that you can view offline.
- 23 and You: A fuller list of 3rd party tools that can process your 23andme data, as well as suggestions on how to use the 23andme tools.
How I have been using the tools to try to find links:
- Triangulation: I can find out who shares the same segment with me as M. does in three ways. First, I can download my Ancestry Finder matches from 23andme and get a spreadsheet, which shows who shares on what chromosome, and look through that spreadsheet myself to find people who share in the same place. Second, if I upload both my downloaded DNA and my Ancestry Finder spreadsheet to GEDmatch, GEDmatch will display the same information in a more readable chart. Third, I can put my data on GEDmatch and run the “Show results that match on a given chromosome segment” tool there. (I’ve found that this actually works better if I look at a larger segment of the chromosome than the part I actually share, because the tool often shows nothing, even if people are there, if you give it too short a segment.) I can find out who shares regardless of whether they share in the same place by downloading both my and M.’s Ancestry Finder matches and seeing who shows up in both spreadsheets. To some degree, I can also find in this way when my cousins share common cousins who don’t share with me directly. I’ve found that finding these common cousins of my cousins is easier on my Balkan side, though, because there the names are more memorable. It’s hard to remember which Williams you may have seen before among another cousin’s relatives.
- Shared locations and surnames: You can check Relative Finder to see how many people share a particular surname. I also save surnames that I pull from certain profiles on a spreadsheet, so I can see common patterns. It would be too time consuming to do this for everyone, but I have a sheet for my father’s side, and one for the closer relatives on my mother’s side.
- Patterns of admixture among DNA cousins: About a third of my DNA cousins show in Ancestry Composition as 100% European. Others (myself included) show as varying mixes of European, Sub-Saharan African, Native American, East Asian, Middle Eastern, North African, South Asian, and in one solitary case Oceanian. I’ve been trying to track who has what admixture, to see if these results offered clues as to who was related to whom, and who was related to me how. I have also tried, for those people who showed as having some Native American ancestry (including M.) to find out if they knew what tribe.
What I learned from trying to find my relationship with M.:
- The slaveowner connection: I have slaveowner ancestors, and M., as an African-American, presumably has slave ancestors. We know that one of the reasons that African-Americans in general have some European-American DNA is the fact that some of the children born in slavery were the children of slaveowners by their female slaves. So the first obvious place to look was to see what I could find out about the slaves my ancestors had owned. My family records preserved the name of only one slave, “Aunt Lukey, the old cook,” who belonged to the family of John Burnam, of Bowling Green, Kentucky, treasurer of the not terribly successful Confederate government of Kentucky (Kentucky went Union, but some Kentuckians formed a government that affiliated with the Confederacy), at the time of the Civil War. Slave schedules gave some information about how many slaves each slaveowner had in 1850 and again in 1860, and their characteristics. I could see, for instance, that one slaveowner, Pleasant Hines, was particularly well supplied with slaves designated as “mulatto,” possibly a sign that someone in the Hines family was particularly given to siring said mulatto slaves. I also checked the 1870 census for likely former slaves of my ancestors, by looking for black people living in their households, living near them and sharing their surnames, or sharing particular combinations of characteristics (easiest with the “mulatto” slaves, since their were fewer of them than “black” slaves). I then summarized the results in a document which I attached to the relevant slaveowners in my Ancestry tree, so that it would be visible to anyone else trying to do slave genealogy in Bowling Green, Kentucky.
It proved to be very hard to identify any particular family slaves (likely or certain), and the number of people I identified was a small fraction of the number of slaves my various ancestors actually owned. Moreover, I couldn’t connect any of these people to M., since M.’s tree doesn’t go back to the Civil War, and she has no known relatives in Kentucky. I did, though, learn a couple of things about what the immediate post-slavery period looked like in this particular town. First, in 1870, five years after the end of the Civil War, white households of former slaveowners were full of black children, sometimes bearing the same surname as the head of the household, with no black parents in evidence for said black children. I’m not sure whether these were slaves who had been sold away from their parents and couldn’t be restored to their families when slavery ended, or children poor families had to place as servants to get by, but whatever was going on, it was common, not just a peculiarity of one family. Second, when I traced the “mulatto” former slaves forward to later censuses, I could see the families landing on different sides of the emerging Jim Crow color line, as some “mulatto” families became “white” while others became “black.”
- Shared surnames: My most obvious shared surname with M. belonged, in my case, to a woman in Virginia who is so sparsely documented that there’s disagreement about whether this even is her surname. Nobody has any idea where she came from. For all I know, she could be my Native American ancestor. It seems unlikely she’s related to M., since she’s in a completely different state from the people in M.’s family with that surname. Shared surnames between M. and other cousins may prove more fruitful, but haven’t born fruit yet.
- Triangulation: M. and I have several shared cousins. I haven’t learned much from them yet, mostly because most of them haven’t replied to me (a lot of people join 23andme for the health reports and stop following their accounts, or, even if doing genealogy, come and go as they do or don’t have time for it).
- Admixture analysis: As I said, a third of my DNA cousins are 100% European. These are not at all evenly distributed. Practically all of my DNA cousins who were actually born in England show as 100% European. Many of the DNA cousins who are proven or likely to connect to me through New England are 100% European. Few of my Balkan DNA cousins are 100% European, and few of my Southern cousins are. When I say that I believe Carrie Burnam Taylor to have had better than 50/50 odds of carrying some non-European DNA, I say this not because my sister and I show scraps of Sub-Saharan African and Native American (we might have gotten these from another line, and in fact I have some reason to believe the particular Sub-Saharan scrap that Ancestry Composition shows is some old admixture on the Greek side). Rather, I say this because it seems that almost none of my DNA cousins through Carrie Burnam Taylor, even the white ones, are 100% European. Remember when James Watson, co-discoverer of the double helix structure of DNA, made a flap by suggesting that black people were stupider than white people? Remember how, shortly after that, bloggers were gleeful about the fact that an admixture analysis of his genome showed that Watson’s ancestry was 16% African? I would bet, if that result stood the test of time, that Watson has some Southern ancestry. Not only are my own Southern DNA cousins less completely white than my Yankee ones, but Bryan Sykes, author of DNA USA, got the same result when he compared the ancestry of white people he recruited for testing in New England, and ones he recruited in the South. If you think about it, this result shouldn’t be surprising; there were simply more black people available in the South to be anyone’s ancestors, and people are pretty horny, imperfectly restrained from mating every which way. Though my Balkan and Southern DNA cousins are mostly less than 100% European, they’re less than 100% European in different ways. My Balkan DNA cousins have a little Sub-Saharan African, more than my English and Northeastern US DNA cousins, they have less Sub-Saharan African, on the whole, than my Southern line DNA cousins. Instead, they have, unsurprisingly, a lot more Middle Eastern DNA than the cousins on Mom’s side of the family.
- The Melungeon connection: I’ve turned up one DNA cousin, like M. with a connection to Tennessee, with Melungeon ancestry, and I suspect from my paper trail that I may have more Melungeon cousins (though probably not Melungeon ancestry myself). I don’t know, though, whether there’s any connection between the possible Melungeon lines and M.
- The Native American connection: Sykes, of DNA USA, found more Native American DNA in the South than in New England (which is actually kind of surprising). I think I see the same in my own DNA cousins, but less dramatically than with Sub-Saharan DNA. I did learn that the most common tribe among my DNA cousins is Cherokee, but there are several others. I can infer that my own Native American ancestor, 500 years ago, lived on the East Coast, in a house, growing corns and beans, and belonged to a community with a complex social structure, because that describes practically all the tribes my cousins’ ancestors belonged to. And M.’s Cherokee connection may (or may not) be related to the Cherokee connection of a couple of other cousins.
- Fanning out through siblings and expanding surnames and locations: This looks promising. It turned out that my Kentucky line was actually, by the beginning of the 20th century, a Kentucky/Tennessee/Virginia/North Carolina/South Carolina/Maryland/Georgia/Oklahoma/Texas line, which offers lots more opportunities for possible connections to M.
There’s another aspect of admixture analysis that I’m going to describe separately, because it’s a little more complicated to explain the results. I mentioned in my last post that L. had drawn inferences about the lines that might connect her and me, on my father’s side, from an admixture analysis of the segments that she shares with me and my sister (the same segment, actually, but for a longer stretch with my sister than with me). I tried the same with M., using a copy of DIYDodecad that I installed on my laptop. There’s a caution in the documentation about interpreting the “by segment” results that I should pass on first, before I describe the results.
A note of caution should be kept in mind for the interpretation of small admixture proportions: over a large region of your genome, such proportions may hint at your having a smaller segment from a particular population hidden within that larger region; you may use a smaller window size to discover such segments.
So my “by segment” results for my shared segment with M. may not be as reliable as my genomewide results (though at least 24 cM is a relatively large shared segment to work with. That said, here are my admixture results from DIYDodecad with the dv3 data set:
Here are my results, also with dv3, for the segment that I share with M.:
West European is elevated and Mediterranean reduced, relative to my genomewide results, as I’d expect for a relative on my mother’s side. Neo_African is also elevated. I’m not sure whether I should consider this result “real,” since the percentage is still small. If it’s real, then M.’s and my common ancestor was mostly white, but with some African ancestry as well (maybe from the colonial period?).
There’s also a data set, africa9, that’s designed for African populations. I suspect its results may be less useful for my own DNA, which shows anywhere from no African ancestry at all (Doug McDonald’s analysis) to a not very impressive high of barely over 2%. But, just out of curiosity, I will compare my genomewide africa9 results with my results for the segment I share with M.
The genomewide results are:
6.55% NW Africa
26.78% SW Asia
This is as un-European as I’ve gotten on any admixture analysis yet, but note that I get there mostly via SW Asia (probably that Asia Minor connection). Now let’s try looking at the segment I share with M. Will it show a higher percentage of any African population than my genome as a whole? To find out, I edit the africa9.par file that I downloaded, and replace the final line, “genomewide”, with
except that I use the real chromosome segment I share with M., which isn’t on chromosome 1, and which has a different starting and ending point. This time, I get my results more quickly:
Notice that my SW_Asia proportion is way lower on this segment, and my NW_Africa and Biaka percentages way higher (my shared ancestor with M. had some ancestry from a nomadic Mbenga pygmy people who live by hunting?). There’s a good chance, unfortunately, that this results isn’t all that meaningful, since Dienekes Pontikos, who developed this calculator, says of it that “It should be used only by Africans and African-West Eurasian admixed individuals,” something I definitely am not. But if you, unlike me, fit the profile for whom the africa9 calculator was devised, you can try this with better confidence in your results than I can have in these. And it is a bit interesting that both the very different calculators I tried like the segment that I share with my closest African-American DNA cousin for having a higher percentage of some sort of African than my genome as a whole (but still, even with this most African-oriented of calculators, more European than African – our common ancestor was more likely white than black).
At this point, I have a lot of information that doesn’t necessarily point in a single clear direction. So what’s my actual conclusion from all of this investigation? M. and I still don’t know for sure how we’re related. At the moment, we seem likely to be related through Tennessee, and we appear to both have some sort of family connection to Smoky Mountain, Tennessee (though it’s still not clear that’s actually where we’ll find our shared line). There’s also one shared DNA cousin who’s particularly closely related to me, who may be able to shed light, if he should get on and respond to my request. Till then, I’ve learned things about my Southern line in the process of trying to track this connection down (there is, for instance, a line that I had been thinking of as English that actually turns out to mostly go back to the Palatinate immigration from Germany). I don’t know where the gap is in the family tree, but I know a little more about the Kentucky/Tennessee line in general.