"% Viking DNA" from Genetic Tests

Factual Questions
1

I was chatting with a British friend on the weekend and she told me that her sister had done a DNA test from Ancestry.com and they reported she had 70% “viking” (Norse) DNA.

She commented that she wasn’t surprised by this since her family tree could be traced back to the same village for at least 400 years (somewhere in the west, near Wales) and she “believed” that vikings had settled in the area 1000 years ago.

I don’t dispute the “viking” ancestry, but I’m struggling to believe that after 1000 years, it’s showing as high as 70%. It seems to me that as soon as someone in her family tree mated with a non-viking ancestor, it would drop by 50%. It would be tough to go back up unless pure 100% Norse DNA was reintroduced.

I realize the tests can be wildly inaccurate but is it possible that the percentage of Norse DNA still can be that high after so long?

I’ve never seen an ancestry.com test result myself, so I wonder if another option is that her sister somehow misinterpreted the results she was given. She said the test also showed they had ancestry from Normandy and Celts.

2

As soon as someone mated with a 100%-non-Viking ancestor you mean.

In any case I agree 70% seems unlikely.

3

The Saxons were a northern Germanic people. Before emigrating to the British isles they lived in what is now northern Germany, the Netherlands, and Denmark. The Angles were also from northern Germany. Throw on the Jutes that were on the northern tip of the Jutland peninsula, and the Germanic peoples that blended into what we call the Anglo-Saxons would look an awful lot like they had Norse DNA.

Then there is the Norman part of her lineage. The Northmen…er Normans… were the result of people from mainly Denmark, but also Norway and Iceland raiding and settling in the area. There was a lot of Norse DNA on the boats with William the Conqueror in 1066.

A lot of those potential non-Viking ancestors could still have been dominantly carrying Norse DNA.

4

@Oldguy

Yes, exactly what I mean.

@ Dino R
Agreed, but ancestry.com explicitly split anglo-sax and Normandy groups out from the “Norse” DNA in her total, with separate scores for those.

As I said, I’m not familiar with the tests: Are you saying that it would roll the Normandy (Norse portion?) back into the total British Norse as well? That seems like a duplicate counting to me, but I guess it’s possible.

5

How many of her gt-grandparents came from that “village in the west”?

Great Britain may have the most thorough genetic studies of anywhere. This article has a series of maps showing migrations over a long period of time. The 5th and final map shows the Norse in St. George’s Channel (though without any arrow showing landings in Wales).

The 4th map, BTW, shows Celtic people fleeing the Anglo-Saxons ca 450 AD by traveling Cornwall to Brittany. Some of their descendants accompanied the Normans back to England in or after 1066. (The agnatic line that became Scotland’s powerful Stewart family followed this Cornwall–>Brittany–>England–>Scotland trajectory IIUC.)

6

(The “Viking” thing is of course garbage. I’m just going to use “Norse” here.)

If one of your grandparents is non-Norse and the rest are, then that’s 75%. So a “one off” outside ancestor will not permanently produce 50% ratio.

So going a ways back if there’s a some 75% people and a 50% person whose descendants intermarry, then you could end up with ~70%.

But, in any case, these “You have x% y ancestry.” results range from approximate-ish to worthless. Just give them a quick glance and move on.

7

@ftg

I understand that, but my point is that as soon as that single ancestor at 75% is created once, the only way to get it back above 75% is to introduce someone above 75%. In the short term absolutely this would happen, but given the population of norse arrived once +250 generation ago, the chance that highly pure genetic norse (over 75% in your example) would continue to be around for so long seems to be very low.

I completely agree and suggested that, but ancestry.com wins the debate based on DNA expertise versus a Joe Lunchbucket like myself.

8

I administrate four DNA-tests at Ancestry.com, for me and some close relatives. We’re all actual Norwegians with roots mainly here, with smatterings from other parts of Scandinavia and Northern Europe.

Nowhere does it say we have 70% Norse DNA. For Norway, and my test, it says, if you click through from the overview:

It also shows 5% for “England, Wales & Northwestern Europe” and that region is described as:

It might be that I haven’t received the latest update, as Ancestry works hard at honing both their estimates and the way the present them, but I highly doubt there isn’t a few major misunderstandings in this game of telephone about your British friend’s sisters DNA.

9

Ancestry.com isn’t in this debate though, just the clearly misreported second hand statements of your friend.

10

Do different DNA test sites report the same or similar results?

11

Lots of places near the coast will have a fairly high percentage of “Viking”. And Great Britain is an island; it’s all near the coast.

12

They don’t publish their algorithm. It is impossible to evaluate whether they are accurately sorting or even double counting. It is not a simple problem for them to try and solve accurately.

13

@RioRico

There was an investigative segment on a Canadian TV show last year(?) where the reporter sent their own DNA to various different testing firms. The results varied wildly. When they interviewed the companies they all “stood by their testing methods” and algorithms.

I also read a story where the sent identical twins DNA to the same companies under different names and they had very different results as well.

@ naita

This was exactly the number they got but since they have no known modern norse connections, they read that as being “viking”. That sounds better than “norse” I guess.

As far as the game of telephone goes, this was at a party, so it was not a serious discussion of ethnicity implications and inherited DNA. It was an entertaining party story, I just happened to question the validity of their result.

@Chronos

Agreed, it’s just the definition of “fairly high” after +250 generations.

14

It would not be any near 250 generations. At roughly four generations per century, the Vikings were landing in Britain roughly 40 to 50 generations ago. In most of Europe, 250 generations gets you back to the Stone Age (the Neolithic period lasted until roughly 1700 BC in Northern Europe, which is still “only” about 200 generations).

15

The overall percentage won’t change with time, if people in those partly-Viking villages are all mixing with people in similarly-partly-Viking villages.

Individuals will vary randomly above and below that average.

16

Different DNA test sites? Try the same DNA test site.

There was an infamous case a few years ago where 3 sisters sent their DNA off to 23andme. The “ancestry” results between them varied widely. (While confirming they were full siblings, so nothing funny going on there.)

17

Curious you bring this up. Recently, I had my DNA checked through Ancestry.com. To my surprise I was 8% Norwegian. Never have I seen any non English/French (my coonass mother) names in my linage. I’m wondering if the Viking influence in England/Scotalnd/Ireland/French still shows through in DNA tests.

18

Doesn’t surprise me. I’m no geneticist, but I do read a lot of history. England was ruled by several Danish (“Viking”) kings in the 10th and 11th Centuries. Besides the Norman invasion, there was also the Danelaw, a large region of England where Vikings settled, intermarried with the locals, and added a lot of Old Norse words to the English language (for example, “skirt” and “shirt” used to be the same word – “skirt” was Norse and “shirt” was Anglo-Saxon). The Danelaw was the North and East of England, not west in Wales. But there was also a lot of Viking settlement across the Irish Sea in Ireland, especially around Dublin. I could see a Norse influence coming from that direction too.

That said, DNA testing like Ancestry and 23andme are bogus gimmicks if you ask me. Your DNA can’t tell you where your family is from. Mostly they’re just telling you where the people with similar DNA live now. And with a long history of nonstop human migration, I don’t think that’s a very useful metric. If you really want to know where your family came from, you’re better off doing regular genealogy, chasing down birth and marriage records as far back as you can. But that does take a lot more work than spitting into a cup.

19

It’s not possible to have a modern Norse connection and reading “we estimate you have Norwegian DNA” as “It’s the Vikings!” is common, but not warranted.

These are not precise tests, but the true horror is the ad hoc hypotheses people come up with to “make sense” of results that are just not that accurate in the first place.

A gross simplification is to compare it to phenotypes. Although it’s done on the SNP level, what the companies does it take selected reference groups in various areas, recruiting people with at least a couple of generations specifically in the area they live in, feed the data from those individuals into machine learning algorithms that try to make distinct groups based on the frequencies of the features, and then group their other customers into the same groups based on which group they resemble the most.

So if you are a brit with blue eyes, fair hair and a strong preference for herring, you get dubbed a Norwegian, because those features are most dominant in Norway, but in reality you could have gotten each of the three from a different European area.

If you happen to be quite similar to one of the groups, you are likely to get a result that reflect your ancestry, but the groups are not that many or that large, and likely miss a lot of real world regional variation, so if you are just in-between a bunch of them, it’s a bit random what you are assigned.

You can see this is true by comparing results for the same individual between companies, by seeing how much some results change with updates, and by looking at ethnicity estimates that do not match ancestry that has been shown correct by the part of these services that actually work, cousin matching.

And if you don’t do an actual systematic review you can easily only pick the people, who might even be in the majority, who get results that match expectences, and do ad hoc “it’s the vikings” for every discrepancy.