Making sense of your Ancestry DNA results

Introduction

My dad recently did an Ancestry DNA test. It made me think a little - it's actually very easy to be underwhelmed by it if you aren't a seasoned genealogist. His comment was something like "there are no surprises in my ethnicity, and I can't see how the people in the list relate to DNA". After a bit of work together, he's now a lot happier.

Ethnicity estimate

Now don't get me started on ethnicity. It is exactly what it says it is - an estimate. Mine looks like this:

Don't be fooled here - the database isn't giving the East Midlands bit. That comes from your tree and your connections, so if your ethnicity is from the British Isles, your ethnicity is broken down between England, Wales & Northwestern Europe and Ireland & Scotland. Furthermore, if I click on the England... link then it tells me it's in the range 79%-100%. If it were an opinion poll, I'd bin it with that range.

But wait. It can be useful. If you are illegitimate and you don't know who your father was, your mother isn't telling or has passed away, and you suspect you may have Irish/Jewish/African ancestry from your appearance, it may help. However, for most of us it's a bit of fun. In general, my advice is to notice it and move on.

Move on to what, Ian?

I'm coming to that. The first thing you should do is build a tree going back six generations to the best of your ability. That is 64 fourth great grandparents. If you have an Ancestry subscription and a FindMyPast subscription (the latter is particularly good for Lincolnshire) it isn't too hard to flesh out your tree, but you don't need a subscription to create a tree, just an account. But don't get too upset if you can't manage 64 - I only have 56 at present. You just should attempt to go further back than second great grandparents or you will struggle to link your DNA results to any particular ancestor.

It's fairly obvious, but the tree you attach to your results doesn't need to have all your fifth cousins once removed in it - it only needs direct ancestors. Mine has my main tree attached, but dad's has a small tree we built of ancestors only. Here are the ancestors of one of his grandparents. I've chosen this one deliberately, as it illustrates the reality that you sometimes cannot create a tree where you have 64 fourth great grandparents due to illegitimacy. But you do the best you can.

ThruLines

What next? You turn on the beta, as below (although this will become default behaviour, so in time no need to do this). Then you wait a few days, as Ancestry need to run an indexing job between your new tree and your DNA matches.

What you see after turning on the beta is a menu like this.

It's really tempting to click on the ThruLines page and then attempt to work out your relatives. At the time of writing (June 2019) this is really hard to do. The approach is, after waiting the few days for Ancestry to reindex, to click on your nnn 4th cousins or closer, then search for common ancestors. Click Add a filter->Common Ancestors as below.

Hopefully you will see a number of people in the resulting list - here is what my dad's results would have looked like after first creating his tree:

You will see on the right hand side there is a link Add to group. What I am going to suggest that you do at this stage is to attempt to group people only by your eight great-grandparents. Remember that if you do have second cousins or closer in the list, they will belong in more than one group. Obviously, icnore them for this purpose.

So here are Dad's eight great grandparents (redacting his mother's maiden name for security reasons):

You may want to ask why I suggested creating a tree going back six generations when I am only going to use three of them. Simple. Without the extensive tree, Ancestry wouldn't have been able to create the common ancestor hints. Note the word hints - they come from other people's trees, so they may not be correct, and they do need to be researched. However, the chances of someone matching you by an incorrect ancestor is remote (but it does happen).

Adding the people with common ancestors to groups

For each of the people in the list, we add the people to the appropriate great grandparent(s). The person we see is a second cousin once removed, so will need to be attached to two great grandparents:

On the right hand side of the screen, there is a link where we can the person to groups. To start with, we'll create as we find them groups for the eight great grandparents.

We create our first custom group, Parkhouse, and add this relative to the group. I see a yellow dot, and if I hover over it I see the name Parkhouse. You don't need to remember the colours - simply hover over the dot to see its description.

We then add the person to group Turner, and we have two dots against them, as below.

We then repeat the process for all of the people in the common ancestor list. Yes, it's laborious, but you'll thank me for it afterwards as it gives a good start at classification. When you complete, and it might take a few hours, turn off the Common Ancestor filter by clicking on it, then click All matches. You will see the numbers in each of your groups, as below.

The first thing you see is that there is one great grandparent (Johnson) missing of the eight. That is nothing to be concerned about. It is when you have an entire side of your family missing that you might think that your paper trail doesn't match your DNA. Even then, don't panic or open uneasy discussions with family until you have more evidence.

Also note that there is only one person in the Surtees group, and that person is below the 'Distant Cousin' 20cM or fourth cousin cut-off. Usually experts advise not to research below that 20cm line. Here is an example where you really should do so to obtain the best evidence from the matches with common ancestors.

Grouping the rest of the people above the 20cM line

If you thought the part you've just done was laborious, the next thing to do is to say that relatives of people related to great grandparent x are also related to great grandparent x with the objectives of giving great grandparent dots to as many people as possible. We start from the top of the DNA matches with the first person with only one dot. Don't worry too much about anyone above them in the list - they are likely to be given dots from numerous places. We look at this person's shared matches, and we give the all the same dot as the person with the shared ancestor, as below.

You then work down with the next person with a single dot until you reach the 20 cM line. Some people will be missed out - don't worry about that at this stage.

You'll hopefully end up with matches which look like this - not too many gaps.

And a considerable number in the various groups:

Now what do you do? You group the rest! You have 24 dots in total, so you have 16 or more left. If you find no shared matches, add a note with a date saying 'No shared matches dd/mm/2019'. It shows you've been there. Also, if you have a dot for groups of two, you'll run out of dots. Use sensible names for your groups: I'm using 'Group D1' onwards, as they are for my dad. A typical page when finished looks like this:

And the grouping look like this. Remember I said that you would miss some people in your great-grandparent groups? Well I did, quite a few, but the mopping up exercise puts everyone in a group, or with no shared ancestors.

Next steps

It's probably time for a cup of tea, but the next thing I would do is to look to break down some of the larger groups to second great grandparents. How to do this? It's a much smaller job than you've done thus far. You delete the dot against everyone for that great grandparent, and then add back the common ancestors at second great grandparent level, then map everyone without a group back to one or the other second great grandparent. Why not do this for a start? Well, you quickly run out of dots (there are only 24) and it's arguably barely worth it when breaking down a group of 5 people.

Conclusion

This technique, while laborious, is likely to give some indications of groupings within DNA matches. Please try it and send feedback.