DNA Matches by Testing Company

Ancestry has recently announced that their DNA database now includes over 6 million individuals.  As Ancestry has the largest autosomal DNA database they are happy to announce these milestones, but accurate information on the database size of the other DNA testing companies is more difficult to obtain.

To provide some idea of how many matches you can expect if you test with the various DNA testing companies, I have compiled the data for my own DNA and my wife's DNA into a chart to show how many matches the major DNA testing companies have identified for each of us, and within that list of matches, how many matches I have been able to confirm match my paper research.

As Ancestry has the largest database, you would expect that it would generate the largest quantity of matches.  This is confirmed for both Marlene's sample and my sample.

You would also expect that Ancestry would generate the largest number of matches confirmable by paper research.  This is true for my sample, and would also be true for Marlene's sample, if she did not have a group of 5 matches on 23andMe that are all siblings / children of those siblings, which skews her data.

But the percentage of Ancestry matches that have been confirmed against paper research is certainly not as great as the percentage that have been confirmed for our samples on 23andMe and Family Tree DNA.  There are a number of reasons why this could be -

• With such a large number of matches on Ancestry, I simply have not investigated enough of them to confirm the match.
• Ancestry uses a lower threshold of the amount of shared DNA that constitutes a match, than the other testing companies do, but those more distant matches are more difficult to confirm against paper research.
• For those Ancestry users who have public family trees, I cannot view their family tree, as I do not have a paid monthly subscription to Ancestry's historical records service, making it more difficult to identify a common surname or even a region where our common ancestor may have lived.
• Ancestry does not identify which segments are shared on which chromosomes (23andMe and Family Tree DNA do), making it more difficult to see which ancestral line we may connect on.

My advise continues to be to test with as many DNA testing companies as you can afford, as they all will generate matches not found on the other testing companies' sites, but regardless, upload your DNA result to GEDmatch.com.

GEDmatch facilitates comparison of DNA samples from multiple DNA testing companies on one site, but requires those who have taken a DNA test to upload their raw DNA data file to the site (a free service).  GEDmatch limits the number of matches that are displayed to 2000, so you may not see a distant relative in your list of matches, if the amount of shared DNA puts them below number 2000 in your list of matches.

I complied the chart below to show the percentage of our GEDmatch matches that come from each DNA testing company.  More than half of our matches on GEDmatch are from Ancestry, but not the 90% that you might expect given the number of matches in each company's database.  

My best guess is that a lot of people take an Ancestry DNA test to obtain their ethnicity breakdown, but do not pursue searching for DNA relatives.

I will update these charts at some point in the future, perhaps when Ancestry hits 7 million DNA samples, but meanwhile I will continue trying to confirm the ancestral connection for our DNA matches from all testing platforms.

The Ryan Herd

The parents of my grandfather, William Sherlock (John Sherlock and Mary Ryan), both had Ryan ancestry.  John and Mary were granted a special dispensation to marry as they were blood relations in the second degree, meaning they were first cousins, and after much investigation I determined that Mary's father (Michael Ryan) & John's mother (Bridget Ryan) were siblings.  So began my involvement with the surname Ryan.

In 1891 Sir Robert Matheson issued a special report on surnames, which identified Ryan as the 8th most common surname in Ireland, and in North Tipperary it is extremely common, making it very difficult to identify one Ryan family from another, with so many Michaels, Patricks, Marys and Bridgets.  So much so that the Irish invented a system of nicknames to help identify the different families.

Chart of Ryan Nicknames (see at North Tipperary Heritage Centre)

My Ryan ancestors had the nickname "Herd", so the great grandparents mentioned above were known as Michael Ryan (Herd) and Bridget Ryan (Herd).  To make the situation more complex, Michael Ryan (Herd)'s daughter, Bridget Ryan (Herd) married Patrick Ryan (Honesty).  Needless to say, Ryans have not been a focus of much further research, as I get cross-eyed trying to keep track of the families.

So I was surprised a few weeks ago to see a new close DNA match on 23andMe, with a Thomas Ryan whose profile identified that he had both Ryan (Herd) and Ryan (Honesty) ancestry.  We corresponded, and confirmed that Tom is a grandson of Patrick Ryan (Honesty) and Bridget Ryan (Herd), making us 2nd cousins once removed.  Tom and I share 44 cM of DMA across 5 segments.

The following week, I noticed a new DNA relative on Ancestry, who turned out to be Tom's niece, Paula Crooks, who is my 3rd cousin.  Paula and I share 96 cM of DNA across 6 segments, which is more than I share with her uncle (one of the vagaries of DNA inheritance patterns).

It is amazing that DNA testing has allowed us to confirm our relationship, when just a few weeks ago we were not even aware of each others existence.

England GRO Website for BMD Certificates

The 1841-1911 decennial census returns for England and Wales are a wonderful tool for tracing a family back through time, but for married women the census returns do not provide a maiden name.

The FreeBMD online index to births marriages and deaths is another great tool, which can be useful for finding a marriage entry for these women, and thus a maiden name, but often there are several possible marriage entries, especially if the given name is Ann, Elizabeth, Mary or Sarah.  So after many years of researching, my database has accumulated a lot of women without surnames.

But the website of the General Record Office certificate ordering service is hugely helpful in finding these women's maiden names.  The search capability is intended to help you find the correct entry, when ordering birth or death certificates, but in the case of births, the search capability and the search results include the mother's maiden name, which on the FreeBMD website is only available from 1911.

If you know the mother's maiden name, it is easy to find all children born to a couple with surname x and mother's maiden name y over a 5 year period, which is useful for finding children who may have been born and died between census years, as long as the surname and maiden name are not too common.

If you know the names, district and approximate year of birth of a couple's children (obtained from the census), you can search for the birth reference of each child, and the search results will provide the mother's maiden name.  By searching for multiple children of the same couple, you can confirm that the mother's maiden name is the same for all children, pick up alternate spelling possibilities, and identify which children belong to which mother in cases where the father married more than once.

Once you have the woman's maiden surname, it is much easier to find her marriage entry in the FreeBMD online index.

I have just been through this exercise for all women without maiden names in my database, and have found the maiden name of at least 50% of them, which is a great improvement, considering that some were not in England or Wales and others did not have children born after 1837.

The GRO website has a similar helpful trick for early death entries.  The FreeBMD website identifies age at death from 1866, which is useful for isolating a death entry when there are multiple entries for the same name, but the GRO website provides age at death starting in 1837, so is helpful for earlier deaths.

Kudos to the GRO for providing this useful search functionality.

Coffin Stainer or Stainer Coffin?

Jane Hemblen, half-sister of my great great grandmother, Elizabeth Hemblen, married William Stainer Coffin, on 7 June 1867 at Bath, Somerset.  William's father is named as Thomas Coffin, a baker.  Their relationship produced 7 children, whose births were registered in Bath registration district as follows -
1862 William Stainer Coffin Hemblen
1864 Thomas Stainer Coffin Hemblen (died 1865)
1866 Thomas Stainer
1869 Elizabeth Stainer Coffin (died 1870)
1871 Elizabeth Coffin
1873 Emily Coffin
1875 Samuel Stainer Coffin
1877 Charles Stainer Coffin

I have not found William Stainer / William Coffin in the 1841 or 1851 census of England and Wales, but in the 1861 census he is listed as William Stainer, a knife grinder, born at Wimborne, Dorset; and Jane Hemblen is listed as his servant.  In the 1871 and later censuses he is listed as William C. Stainer or William Coffin Stainer.  His death is registered under the surname Stainer and also under the hyphenated surname Coffin-Stainer.  When his wife Jane, died in 1892, her death registration identified her as Jane Coffin Stainer.

When their children married and had children of their own, the events were registered under the surname Stainer, but Coffin was often used as a second given name, resulting in many "Coffin Stainer" registrations.

In researching the extended family, I found several related family trees on the Ancestry.com websites.  Many of these trees did not identify the parents of William Stainer Coffin, while others suggested that he was descended from a line of Stainers, not Coffins.

I believe it more likely that William Stainer Coffin was a son of Thomas Coffin and Ann Beale Stainer, who married in Wimborne, Dorset in 1819.  Thomas Coffin is listed in the 1841 census of Wimborne, as a 40 year old baker with 4 children, and appears to be a widower, as no wife is listed.  Thomas Coffin's listed occupation of baker, matches the information from William's marriage certificate.  Two of Thomas Coffin's children listed in the 1841 census are Samuel and Emily, names also given to two of the children of William and Jane Coffin / Stainer.

Why William Stainer Coffin changed his surname we will likely never know, but my guess is that it relates to William's life prior to meeting Jane Hemblen.  In the 1861 census Jane is listed as his servant, but there is also a 6 month old daughter, Mary Stainer, living with them.  It is unclear from the census entry whether Mary is Jane's daughter, or William's daughter from a previous relationship. William and Jane had 3 additional children together before they married in 1867, so perhaps William was still married, and not a widower as suggested by the 1861 census record?

The New 23andMe

Finally, after 18 months of waiting, my account was transitioned to the "new 23andMe" experience last week.  I had read a lot of negative comments about the new experience, so I was somewhat apprehensive about the transition.

The new experience certainly presents the information in a different way, and I wouldn't give the site any prizes for being user-friendly, but overall, I find the new experience is more useful to a genealogist than the old site was.

The new experience splits information into 4 broad categories, two of which are health related (Traits and Wellness) and two of which are ancestry related (Ancestry and DNA Relatives)

The New 23andMe Home Screen

The health-related reports are of some interest, but it's the ancestral stuff that is important to me, and this can be accessed by clicking on Ancestry Reports.  From here you can looking at your paternal and maternal haplogroup (to gain some insight into your ancient origins), see how much Neanderthal DNA you have, look at your genetic composition by region (I am 99.9% European, which is not unexpected), and view your DNA family.  These are all screens that you will likely ignore after your initial curiosity, if you want to use the site to find relatives with whom you share significant DNA.

Ancestry Reports Screen

The easiest way to access your DNA matches (those people with whom you share DNA), is by clicking on Tools on the top horizontal menu and then selecting DNA Relatives.

Tools Menu

The DNA Relatives screen is similar to the DNA Relatives screen on the old site, but with less visible information - you see only the name, predicted relationship, and amount of shared DNA.  Additional information can be viewed by clicking on an individual's name.  Two enhancements are the ability to mark matches as favourites (I use this to mark people with whom I have determined how we are related), and the use of coloured dots to indicate the sharing status of the individual (sharing, request pending, etc.).

When you click on an individual match, you see much more detail about the person (haplogroups, matching segments, ancestral surnames, geographic origins, and relatives in common).  It is this last feature that is most useful in determining which ancestral line someone may be connected through.

Some positive changes with the new experience -

• You can now contact anonymous users (a feature that was removed 18 months ago, when this transition began).
• The Relatives in Common feature is great for isolating which ancestral line someone may be connected through.
• Open Sharing - If you opt in to Open Sharing, all matches can see your Ancestral Composition and your matching DNA segments, without having to send you a sharing request and waiting for you to accept it.
• Favourites - Allows you to mark individual DNA matches with a star.  I use this to identify those with whom I have established how we are related.
• Sharing Status - The use of coloured dots on the DNA Relatives screen to show the sharing status of your matches is very useful, but the colours are not different enough to work well for colour blind people like myself.  The use of different shaped indicators (e.g. circle, square, triangle, and diamond) in addition to the colour, would be an even bigger improvement.

My only real dislike of the new experience is the lack of user-friendliness when using DNA Relatives and Share and Compare.  But each time I use the new site, I get more comfortable with it.

Overall, the new 23andMe is my second favourite DNA testing site, but FamilyTreeDNA remains my first choice.

Using Ancestry's Surname Filter

I decided to utilise the surname filter tool on my Ancestry DNA matches to try and determine which ancestral lines some of the 2500 matches might be on.  It should be noted that the filters will only identify other DNA test subjects, where the person tested has also created a family tree on Ancestry - if a person has not created an ancestral family tree on Ancestry, then they will not show up when you filter by surname.  When you enter a surname into the filter field, the quantity of matches will be limited to show only the DNA test subjects who have that surname in their ancestral family tree.

 

My Great Great Grandmother, Elizabeth Hemblen was transported to Tasmania in the 1850s for stealing a frying pan, and although we have strong circumstantial evidence that "our" Elizabeth was the daughter of a Stonemason in Bath, Somerset, named Isaac Hemblen, we don't have any firm evidence, so a DNA match with other descendants of Isaac Hemblen, would add more credibility to the paper research to date.

 

When I filtered my 2500 Ancestry DNA matches with the surname Hemblen, I had only one match - Diane R.  So far so good!

 

Hemblen Filtered Matches

 Investigating further, I found that Diane and I are predicted to be 5th - 8th cousins, and that we share 6.4 cM (centiMorgans) of DNA on a single segment of one chromosome. 

Shared DNA Details

This is the point, when looking at DNA matches on other DNA testing websites, that I would look at the details of how many cM we shared, on which chromosomes, and compare the matching segments to other known matches, to try and isolate which ancestral line the match is likely on.  Unfortunately, with Ancestry, this is as good as it gets.

 

But, again using the Ancestry surname filter, I tried various surnames, to see which other surnames also appear in Diane's ancestry.  The first surname that I tried was Meale, the surname of Elizabeth Hemblen's husband (my great great grandfather) - Diane R. was not in the list of matches.  Next I tried surnames associated with spouses of Elizabeth Hemblen's siblings and half siblings.  Diane R. appeared as a match when I filtered with the surname Stainer, suggesting that Diane is a descendant of Elizabeth's half-sister, Jane Hemblen, who married William Stainer.  By filtering with the surname Enefer, I was able to narrow down the potential match to Thomas Coffin Stainer, son of Jane Hemblen, who married Mary Ann Enefer, suggesting that Diane is a descendant of Thomas Stainer and Mary Ann Enefer.

 

I contacted Diane, and outlined how I believed we are connected, and within 24 hours she confirmed that Thomas Stainer and Mary Ann Enefer were her great grandparents.  Diane and I are actually 4th cousins, a little closer than Ancestry suggested.

 

So despite the limited information provided by Ancestry on which segments of which chromosomes I share with Diane, the Ancestry surname filtering tool provided a high level of confidence in identifying how we connected, and I now have much greater confidence that Elizabeth Hemblen, daughter of Isaac Hemblen is my great great grandmother.

Pet Peeves with DNA Matching

With the exception of those who test with 23andMe for health related information, the vast majority of people take autosomal DNA tests to identify familial connections.   But many people severely limit their opportunities to identify connections with others by not maximising the functionality of the testing service. 

My pet peeves related to DNA testing for family history are -

1. People who do not reply to messages related to a DNA match - Communication with others who share portions of your DNA is the best way to establish how you may be related, or to get help with achieving your objectives of taking a DNA test.  If someone contacts you, have the courtesy to reply.
2. No associated family tree - All of the major testing companies (and GEDmatch) provide a facility for publicising your ancestral family tree, and all provide protections for hiding information on living people.  Publish the information that you know, to make it easier for DNA matches to see how you might be related.
3. Associated family tree is not public - Why hide your research from others?  You may have identified an ancestor that others have not, or you may have made a mistake in identifying an ancestor.  If your online tree is private, you will miss the opportunity to help others and correct errors.
4. Anonymous people on 23andMe - I am still on "the old experience" so I get to see that I have a DNA match, and how much DNA we share, but I cannot contact people who have chosen to be anonymous.  Why did they bother taking the test?
5. DNA result not uploaded to GEDmatch - If you have tested with one of the major DNA testing companies, you can avoid a lot of the drawbacks of the individual companies' capabilities by uploading your DNA test result to GEDmatch, which facilitates matching between people who tested with different companies.
6. Not accepting genome sharing requests on 23andMe - 23andMe (the old experience) does not automatically show you which segments of which chromosome you share with another customer, so you have to send a genome sharing request to each potential match, to identify which segments you have in common.  If you do not accept genome sharing requests you can only identify that you share DNA with a person, but have no idea of which line you may be related on.
7. Ancestry and MyHeritage do not provide information on which segments of which chromosomes are shared - These two testing companies know exactly which segments of which chromosomes you share with every other person in their database, but they choose to not provide you with the information.  At least 23andMe lets the user decide to share or not share, but Ancestry and MyHeritage severely limit their DNA test capability by not allowing users to ever see this information.

Please, take a few minutes to -

1. Check the privacy settings on your DNA test account
2. Upload a GEDCOM file of your ancestors or manually add your ancestors to the family tree associated with your DNA test account, and make sure that viewing the tree is not restricted
3. Upload your DNA test results to GEDmatch, and include a GEDCOM file
4. Check for new matches every week or two
5. Communicate, communicate, communicate!