Monday, July 25, 2016

Are 111 Marker Tests Better at Predicting Relationships? A Case Study Perspective.

I have updated this post on July 28, 2016. 
The new sections are added in red.

Two years ago, I looked at whether Y-DNA genetic distance was an adequate predictor of relationships. Using a number of participants at both 37 and 43 markers, I concluded that genetic distance was an insufficient predictor of relationship range. This post examines the following question: “Would an STR marker test at 111 markers enhance the ability to predict relationships?”

In the past two years, the Owston/Ouston DNA project has had the opportunity to upgrade fifteen men to 111 markers. At this resolution, our 15 project members only have matches within our surname group, which indicates it is a sufficient tool to narrow the results to a particular family group. By comparison, each of us matched several men with a number of surnames at 67 markers. At 37 markers, we matched nearly 400 men with different surnames.

Although I previously stated that the 111 test had been sufficient enough to separate the wheat from the chaff in our one-name project, I failed to take into consideration that a number of individuals may have not upgraded to 111 markers. Thanks to John Lisle who encouraged me to assess this particular statement, as only two of the 11 non-Owston/ Ouston matches at 67 markers had upgraded to 111.

Our 15 subjects represent three families of a low-frequency surname group that totals an estimated 296 males. These 296 men and boys, yes I counted them, have residence in the UK, the USA, Australia, Canada, New Zealand, Finland, and France. Testing subjects represent all the aforementioned countries except France; however, the son of the lone Frenchman, who lives in England, has tested. A brief synopsis on these families is found below; they are designated by their location of origin. Charts indicating all extant lines and participants can be found at


The largest Owston/Ouston family hails originally from Sherburn in Hartford Lythe, North Yorkshire and its many members descend from Peter Owston who died in Sherburn in 1568. This group also includes individuals with Ouston surname variation who descend from James Ouston (1711-1785). The Sherburn family’s most recent common ancestor, William Owston, was alive in the 1550s and died in 1602. Seventy percent of all Owston/Ouston males descend from this family. The most distant relationship within the Sherburn family is that of 13th cousins.

A total of 19 men from the Sherburn family have tested: six exhibited ancestral non-paternity events, two remain to be upgraded from 43 markers, and 11 have been tested at 111 markers. Of those 11 subjects, the relationships range from siblings to 11th cousins, once removed.

The Sherburn family includes the Cobourg line, which descends from William Owston 1778-1857, and is named for the Canadian town near where the family located in the 1830s. While William lived in numerous locations in England, Scotland, Ireland, and Canada, he and his family lived the longest in Hamilton Township, which surrounds the Town of Cobourg, Ontario.

William was placed within the Sherburn line due to circumstantial evidence that led to this conclusion. Of the seven William Owstons born within a 10-year tolerance of his birth date, William Owston of Ganton, the son of Thomas Owston of Sherburn, met several criteria (birth date, birth region, and father’s name) to be considered likely to be one and the same as William Owston of the Cobourg line.

This particular line is well represented in the project with seven Y-DNA participants – six at the 111 level. The most distant relationship represented is that of fourth cousins, once removed. The most distant relationship among all males of this line is at the sixth cousin level. A total of 22 Owston and non-Owston relatives in the Cobourg line have also participated in autosomal testing. This is the author’s line and the reason for its overrepresentation was convenience sampling.

The abbreviated chart below shows the relationship of the Sherburn (and Cobourg) descendants who have upgraded to 111 markers. 


The second largest group of Owstons originated in the village of Ganton, which is located five miles east of Sherburn in North Yorkshire. Descent can be satisfactorily traced to Giles Owston who died in 1641. While an older connection cannot be firmly established, this family is probably descended from John Owston who was alive circa 1490 in nearby Staxton in Willerby. This supposition occurs because several unique first names exist in both lineages. While a few Ganton Owstons live in the UK, the majority of Owstons in the US are from this family. All surviving Ganton Owstons descend from Thomas Owston (1755-1823). The Ganton family represents 21% of living Owston males.

A total of four men from the Ganton family have tested; however, two were tested at 43 markers, but died before upgrading at FTNDA. The remaining two, who have tested at 111 markers, are fifth cousins. The family’s most distant relationship is that of seventh cousins.

The following chart shows the relationship of the two Ganton participants who have tested at 111 markers. 


Finally, a third group of Owstons from 15 miles south of Sherburn and Ganton can be traced to Richard Owston of the village of Thornholme in the parish of Burton Agnes, East Riding of Yorkshire. Richard Owston died in 1739. By using onomastic evidence, it is possible to theorize a connection to an earlier Ganton line fathered by Robert Owston who was born as recent as 1580. The Thornholme Owstons constitute the largest group of Owstons/Oustons in Canada.

A total of five men from the Thornholme family have tested and all four extant lines are represented in our project. Two men exhibited ancestral non-paternity events, one has not yet upgraded from 43 markers, and the remaining two have tested at 111 markers. These two individuals are seventh cousins once removed.

Being the smallest of the three families, the Thornholme family has only 25 males, which constitutes 9% of the total number of Owston/Ouston males. The most distant relationship found within the Thornholme family is that of 9th cousins. The following chart details the relationships of the two Thornholme family members who have tested at 111 markers.


Data for this one-name study comes from the individual and combined research of Timothy J. Owston, Roger J. Ouston, and James M. Owston. While each began researching the surname in the 1970s, their combined efforts began in 1990 when they crossed research paths.

As noted, a total of 15 men have tested at 111 markers; this represents 105 relationships. While intrafamily relationships are easily tracked, the difficulty arises in cross-family relationships, as records prior to 1550 are spotty. Matching Y-DNA has confirmed that the three families are related and they are from the same region; however, documentation on connections among the three families does not appear to exist.

To address the interfamily relationship problem, I have created a plausible tree based on naming conventions from the three current families and two extinct families who have originated in the Vale of Pickering that spans the historic border of the former North and East Ridings of Yorkshire. The first reference of the surname in this region appeared in 1452. I am confident that the relationships of these lines are within two generations (further distant) than I’ve charted. For this analysis, I used the closest possible relationship that could be presumed.


The following charts and table enumerate the known (and assumed) relationships in this family.

Click image for a larger version.

 2nd Cousins2
 2nd Cousins, Once Removed1
 4th Cousins7
 4th Cousins, Once Removed3
 5th Cousins1
 7th Cousins, Once Removed2
 8th Cousins5
 8th Cousins, Once Removed7
 8th Cousins, Twice Removed2
 9th Cousins6
 9th Cousins, Once Removed7
 9th Cousins, Thrice Removed 1
 10th Cousins1
 10th Cousins, Twice Removed7
 11th Cousins, Once Removed2
 12th Cousins3
 12th Cousins, Once Removed8
 12th Cousins, Twice Removed6
 12th Cousins, Thrice Removed1
 13th Cousins7
 13th Cousins, Once Removed16
 13th Cousins, Twice Removed1
 14th Cousins4
 14th Cousins, Once Removed2

Click image for a larger version.


By comparing the results of 15 subjects at 111 markers and the additional five participants at 43 markers, a modal haplotype has been constructed. Three participants shared the modal signature at 111 markers: Ganton03, Ganton04, and Cobourg08. The late Ganton01 also exhibited the modal haplotype at 43 markers. Several others who shared the modal haplotype at 37 and 43 markers did not at 111 markers.

There was a noted convergence with Cobourg08 who had a back mutation on DYS643 to 12 repeats, which was found in the modal haplotype. All of the other matching Cobourg line members have an 11 at this marker. This back mutation attributed to some of the outlying results in this analysis. The genetic distance (GD) for the 105 relationships at a 111 marker resolution ranges from 0-9. A GD of 2, however, was not recorded for any of the relationships.

Click image for a larger version.
The following table delineates the generational range, mean, the adjusted mean relationship, and the standard deviation for the results.

0114.56.06  5th Cousins        5.15
151511.4610th Cousins, Once Removed        4.68
3315.510.60  9th Cousins, Once Removed        4.88
4514.512.4811th Cousins, Once Removed        2.34
55.514.511.6810th Cousins, Once Removed        2.71
691510.53  9th Cousins, Once Removed        1.99
78.515.512.3711th Cousins, Once Removed        2.39
811.51412.8312th Cousins        0.98
99.51310.38  9th Cousins, Once Removed        1.80

Outside of a GD=0, the adjusted mean relationship for genetic distances of 1 to 9 ranges from ninth cousins, once removed to 12th cousins. The plot below provides a visual representation of the interquartile range and the outliers based on genetic distance (GD) and the time to the most recent common ancestor (TMRCA).


Largely due to two outlying relationships because of convergence and three very close relationships in the Cobourg line, those sharing a GD=0 have the greatest standard deviation (SD) of 5.15 generations. GD=1 and GD=3 are not far behind with standard deviations of 4.68 and 4.88 generations respectively. The relationships that are represented by these three genetic distances (0, 1, & 3) are more heterogeneous. This heterogeneity indicates that relationships at these levels are likely to be more different than similar.

Contrariwise, those with a greater genetic distance have a lower SD and are more likely to be similar in relationship. With a genetic distance of four through nine, the SD ranges from .98 to 2.71 generations. The most homogenous group is GD=8 with the lowest SD of 0.98 generations. Therefore, it appears that the greater genetic distance, relationships become slightly more predictable at least within our surname.


While 111 markers aided in fine tuning our connectivity to those sharing our genetic and genealogical roots, genetic distance was not an accurate predictor of most relationships. Outliers can and do happen, as experienced with a GD=0; however, 78% of the participants at a GD=0 fell within the predicted level of six generations or less with a p ≤  .01, Two did not, and as explained earlier, this was due to convergence. We have seen close relatives (5th cousins and closer) having genetic distances up to 5, while 13th cousins, once removed have a GD=0.

The caveat is that this is one family of meager size from one haplogroup I-M253 (fine tuned with SNP testing to I-A10207). It may not be representative of everyone’s experiences; however, this information can be used as a reminder to exercise caution in using genetic distance as an indication that someone is more closely or more distantly related than he actually is. A prediction based on genetic distance alone might just be wrong.


The move to 111 markers questioned our genealogical research methods, as all but one Cobourg line participant matched members of the Ganton family at 110 or 111 markers; Cobourg06 had more mismatches with most participants including his own family. In addition, matches between Cobourg participants and their Sherburn cousins were more distant. In fact, the closest related Sherburn family member matched most Cobourg members at 105/111. Matches between Cobourg and Thornholme families were distant as well. I questioned, “Did we place the Cobourg line in the wrong family?”

By analyzing the parish registers once more, we confirmed our original decision that the Cobourg line was a subset of the Sherburn family. Still, we wanted confirmation. In effort to be frugal, we decided to first test Ganton03, Ganton04, and Sherburn08 autosomally and hoped that at least one of the 22 Cobourg line members matched someone. While autosomal testing is generally reliable for up to five generations, this move was a gamble, as the new participants were related to the Cobourg line beyond that level in any scenario. Knowing this, we forged ahead and none of the three new autosomal subjects matched anyone. The two Ganton family matches, who were fifth cousins, even failed to match one another.

The next step was to utilize FTDNA’s BigY test and to contract with YFull for an analysis of the results. We originally tested Cobourg01, Ganton04, and Sherburn08 (Cobourg01’s closest related Sherburn family member). When funds were available, we added a Thornholme family participant. The results were surprising.

It has been suggested that SNP testing is a more accurate predictor; however, testing these four Owston/Ouston family members with the BigY has created more questions, as the results suggested a heretofore unforeseen and unpredictable scenario. Let me explain. 

I anticipated that Cobourg01 would closely match either Sherburn08 or Ganton04. I also assumed that Sherburn08, Ganton04, and Thornholme04 would match, but have some differences. If any of those three should have had closer matches, it would have been Ganton04 and Thornholme04, as onomastic evidence suggests that these two families are more closely related.

To summarize, we have three distinct families: Sherburn, Ganton, and Thornholme. Remember that STR results at 111 markers were the closest between the Cobourg line and the Ganton family, but parish registers indicated that the Cobourg line was part of the Sherburn family. However, there still was an outside chance that the Cobourg line could rather be attached to the Ganton family rather than to the Sherburn family. While all four had novel SNPs, they all had 16 new SNPs that matched. But differences arose in two SNPs.

Ganton04 and Sherburn08 tested positive with A10216, while Cobourg01 and Thornholme04 were negative. In addition, Cobourg01 and Thornholme04 were positive for Y22277; however, Ganton04 and Sherburn08 had no calls. A conclusion cannot be made regarding Y22277; however, A10216 placed Ganton04 and Sherburn08 downstream from Cobourg01 and Thornholme04’s A10207 terminal SNP.

With these results, Ganton04 and Sherburn08 were exact matches and Cobourg01 and Thornholme04  were exact matches. As a researcher who has devoted the last 38 years to primarily studying one surname, this created cognitive dissonance. What happened? Here are some possibilities that are listed in no particular order.

  1. The Cobourg family is actually descended from a heretofore unknown William Owston from the Thornholme family who was born in February 1778 in Yorkshire and who likely had a father named Thomas.This appears to be the logical conclusion, but we have yet to find this William Owston. In addition, the Thornholme family must be more distantly related to the Ganton family than we had previously predicted. It is doubtful any genealogical records exist to confirm or refute this.

  2. The A10216 SNP results shared by Ganton04 and Sherburn08 were examples of convergence. I’ve been told that this doesn’t happen with SNPs, but do we really know that?

  3. At least Cobourg01 and perhaps Thornholme04 had back mutations on A10216. I’ve also been told this doesn’t happen, but could this be possible?

  4. There were issues with the BigY BAM files from FTDNA or with the analysis from YFull. Anything is possible.
This creates a big mystery that impacts decades of research, and hopefully we’ll be able to confirm this anomaly in the near future. If the Cobourg line is from the Thornholme family and not Sherburn family, the results suggesting that genetic distance at 111 markers is not an adequate predictor of relationships still holds and makes the case even stronger as the Cobourg and Thornholme Owstons have a greater genetic distance. 

In any case, more research needs to be conducted. We’ve already upgraded Thornhole04 with an autosomal test and results should be in within a week. This is a gamble as the results may be inconclusive. Our plan is to also bring on several more participants for the BigY; but due to pricing, we will wait for sales to occur.

I hope these additions make some of my previous statements a bit clearer and my cognitive dissonance over the BigY results understandable. Thanks.


  1. Thank you so much for posting! As always, the cry is for more data, more data, more data!

    1. Thanks Stephanie. I appreciate it.

  2. I meant to get back to you re your initial version ... read your latest with amendments in 'red' and your newer analysis ... and sympathize with your efforts to make sense of your data ... I'm also the admin of a single surname project with two very very distinct genetic McKellar lines going back to the 1450s on Loch Awe in Argyll, Scotland ... both lines have a unique SNP that falls within genealogical time ... did you pass your Big Y results to Alex Williamson the owner of The Big Tree?

    1. No, but I'll look into it. Thanks. Actually, I kind of relish our placing in a new line, as I've always felt that the Thornholme family was a bit more adventurous. The only issue is now I have a missing link that I cannot find.

    2. Not sure why my name is "unknown" should be David McKellar ... Alex's email is - send him all the Big Y results you have (*.vcf not BAM) ... here are the two McKellar lineages in Alex's The Big Tree - they are genetically not even close as they were two families on Loch Awe that assumed the surname McKellar when they were first used in Argyll in the mid 1400s ... and ... both have a unique SNP in genealogical time that will help me figure out how existing family trees may be eventually merged ... David