23andMe Share your 23andMe Ancestry Composition

Very interesting map. British and Irish appears to be more Celtic than Anglo-Saxon. Don't understand why someone from Sardinia would be under 50%.
 
Speculative
I'm from East Poland

Eastern European 85.7
Scandinavian 2.4
British Irish 0.8
Broadly NW European 2.7
Balkan 3.9
Broadly European 3.9
 
Everything depends on 23andme's reference samples, which, other than the few academic samples they use, are not scientifically chosen, and therefore might not be representative of the areas involved. Many more testees at 23andme are Irish than English, because most clients are from the U.S. and there's a big population here that identifies as Irish. Any people of English descent usually consider themselves just "American".

In Italy, for example, the dot looks to be between Rome and Campania, which makes perfect sense as most of their testees are from southern Italy, but they have the big block of Tuscans from the 1000 genomes project which pulls the centrum northward. That dot does not mean that's where the most genetically "Italian" people can be found.

As for the "French-German" cluster, that is one of the least reliable of any of the clusters according to 23andme's own analysis. What they basically did is use the French Lyon sample, which is the only one that was available for a long time, and the Swiss samples. The southern French sample now available was not included. A high number for this basically tells you how "Swiss like" you are, because there are a lot more samples for that. Sardinian isn't much good either.

Scroll down to Testing and Validation at the following link:
https://www.23andme.com/ancestry-composition-guide/

The "map" results are totally predictable if you know how 23andme works.

By misinterpreting these results novices can be led astray.

Oh, in searching for the link for this white paper I discovered that in more recent white papers 23andme is telling people their results go back 500 years and more. Just google 23andme white papers and you'll find it. Of course, they're not going to acknowledge how often they were told that, including by me.
 
I have made a few maps based on the data available so far. Needless to say that the frequencies will need to be adjusted once I get more data, especially for Spain, southern France and Central Europe. I extrapolated in some regions with no data when the admixture could not possibly have been above 1%.

23andMe_British_Irish.png


Looks like the distribution of R1b-L21 combined with specifically Anglo-Saxon and Frisian clades of R1b-U106.

23andMe_French_German.png


If this component is as high in Austria as in South Germany, then it could represent Hallstatt and La Tène ancestry. The two peaks are Switzerland and Wallonia, which were two centres of the La Tène culture. The map matches fairly well the distribution of Celtic branches of R1b-U152 (once Italic ones are removed).

23andMe_Scandinavian.png


Reminiscent of the distribution of Y-haplogroup I1, but with a sharper gradient (higher in Sweden-Norway, lower outside Scandinavia).

23andMe_Italian.png


A distribution that seems to include mostly the non-Celtic/Germanic ancestry in Italy. Apart from the Aegean Islands, this admixture is quite distinct from the Greek one, despite the heavy Greek colonisation of southern Italy. Therefore 23andMe seems to have cut (inadvertently?) the depth of ancestry to about 2000 years ago, right during the Roman period. As a result there is a good chance that this Italian admixture outside Italy corresponds to Roman ancestry, and indeed it is only found with in borders of the Roman Empire. I have a few samples from Turkey but this component varies hugely between individuals, from 1.5% to 10% (near Trabzon).
 
That's great. My aim is to make distribution maps for each component once I get enough data for each country.

In my samples, the highest French & German is from Wallonia in Belgium (70%), and the highest Scandinavian are a Swede (89.6%) and a Norwegian (87%), who both exceed the figures in this map. The Norwegian is from Oslo, but I am not sure that's also where his ancestors come from. The southern Swedish-Norwegian border near Oslo may have the highest Scandinavian. Incidentally that's also where haplogroup I1 peaks (about 50% of the population).
Thanks, Maciamo! I'll update him.

From the same person:
(rounding off)
Portugal (n=30)
- Southern European: 81.2
italian: 5.7
balkan: 0.1
iberian: 59.0
broadly S.E.: 16.2
- Northwestern European: 11.8
french & german: 0.9
british & irish: 2.5
broadly NW.E.: 8.3
- Ashkenazi Jewish: 0.1
- Broadly European: 5.1
- MENA: 0.4
- Sub-Saharan: 0.2

Serbia (n=16)
- Southern European: 81.2
italian: 0.3
balkan: 75.5
iberian: 0.1
broadly S.E.: 5.3
- Northwestern European: 2.0
french & german: 0.3
broadly NW.E.: 1.6
- Eastern European: 9.0
- Broadly European: 7.4
- East Asian: 0.1
- South Asian: 0.1
Portugal, Serbia, and now Denmark, from the same person.

Denmark (n=27)
- Southern European: 0.3
italian: 0.1
broadly S.E.: 0.1
- Northwestern European: 97.5
french & german: 7.9
british & irish: 17.1
scandinavian: 38.5
broadly NW.E.: 34.0
- Eastern European: 0.4
- Broadly European: 1.6
 
Everything depends on 23andme's reference samples, which, other than the few academic samples they use, are not scientifically chosen, and therefore might not be representative of the areas involved. Many more testees at 23andme are Irish than English, because most clients are from the U.S. and there's a big population here that identifies as Irish. Any people of English descent usually consider themselves just "American".

In Italy, for example, the dot looks to be between Rome and Campania, which makes perfect sense as most of their testees are from southern Italy, but they have the big block of Tuscans from the 1000 genomes project which pulls the centrum northward. That dot does not mean that's where the most genetically "Italian" people can be found.

As for the "French-German" cluster, that is one of the least reliable of any of the clusters according to 23andme's own analysis. What they basically did is use the French Lyon sample, which is the only one that was available for a long time, and the Swiss samples. The southern French sample now available was not included. A high number for this basically tells you how "Swiss like" you are, because there are a lot more samples for that. Sardinian isn't much good either.

Scroll down to Testing and Validation at the following link:
https://www.23andme.com/ancestry-composition-guide/

The "map" results are totally predictable if you know how 23andme works.

By misinterpreting these results novices can be led astray.

Oh, in searching for the link for this white paper I discovered that in more recent white papers 23andme is telling people their results go back 500 years and more. Just google 23andme white papers and you'll find it. Of course, they're not going to acknowledge how often they were told that, including by me.

Thanks for the info. It was interesting to read the differences between precision and recall and how a lot of northwestern European gets tossed in the broadly category. I still don't understand how people in east central France would have 20% British and Irish with 90% precision. That's like almost one out of four grandparents. Is it because of ancient DNA markers that exist in both populations?
 
Some new people have shared with me, so here are some new averages:

Albania (n=12)
Balkan 86.86%
Italian 2.72%
Broadly South European 7.65%
East European 0.43%
Broadly Northwest European 0.44%
Middle Eastern & North African 0.02%
East Asian 0.1%

FYR Macedonia (n=4)
Balkan 71.38%
Italian 2.92%
Broadly South European 13.1%
East European 6.1%
Broadly Northwest European 1.05%
Middle Eastern & North African 0.02%
East Asian 0.15%
 
Thanks for the info. It was interesting to read the differences between precision and recall and how a lot of northwestern European gets tossed in the broadly category. I still don't understand how people in east central France would have 20% British and Irish with 90% precision. That's like almost one out of four grandparents. Is it because of ancient DNA markers that exist in both populations?

I think that's the explanation for a lot of it. For years 23andme stuck with the "only 500 years" in the past mantra, when it was patently absurd. The issue is that some of the admixture might be from very recent migrations, some in the past. There's no way of knowing, and in some cases it might "stop too short", i.e. not be giving you the whole story because it isn't going back far enough. That's the problem with using algorithms based on modern populations.
 
23andMe_Italian.png


A distribution that seems to include mostly the non-Celtic/Germanic ancestry in Italy. Apart from the Aegean Islands, this admixture is quite distinct from the Greek one, despite the heavy Greek colonisation of southern Italy. Therefore 23andMe seems to have cut (inadvertently?) the depth of ancestry to about 2000 years ago, right during the Roman period. As a result there is a good chance that this Italian admixture outside Italy corresponds to Roman ancestry, and indeed it is only found with in borders of the Roman Empire. I have a few samples from Turkey but this component varies hugely between individuals, from 1.5% to 10% (near Trabzon).

I am Cappadocian Greek and I score 36% Italian. Of course no Roman ancestry in my case (and many other Anatolian Greeks who tend to score really high Italian).

Cypriots score also high Italian.

Sent from my Robin using Tapatalk
 
What populations get the highest Broad results for Southern Europe specifically?
 
I am Cappadocian Greek and I score 36% Italian. Of course no Roman ancestry in my case (and many other Anatolian Greeks who tend to score really high Italian).

Cypriots score also high Italian.

Sent from my Robin using Tapatalk

Wow! That's interesting! Do you mind sharing your results with us? If you don't want to, that's ok with me and I won't quarrel.
 
Here are the results of my all Irish family on 23andMe.

Mine

European 100%

Northwestern European 100.0%
British & Irish 94.3%
French & German 2.6%
Broadly Northwestern European 3.1%
Broadly European < 0.1%

Mother

Northwestern European 99.6%
British & Irish 91.4%

French & German 4.6%
Broadly Northwestern European 3.6%

Southern European 0.2%
Broadly European 0.3%

Brother

European 99.9%
Northwestern European 99.9%
British & Irish 97.5%
Broadly Northwestern European 2.4%
East Asian & Native American < 0.1%

Daughter

European 100%
Northwestern European 99.7%
British & Irish 92.9%
French & German 3.9%
Broadly Northwestern European 2.9%
Broadly European 0.3%
 
Last edited:
I am Cappadocian Greek and I score 36% Italian. Of course no Roman ancestry in my case (and many other Anatolian Greeks who tend to score really high Italian).

How could you possibly know that you have no Roman ancestry? Nobody has a genealogy going back to Roman times.

Cypriots score also high Italian.

Would you happen to have a few samples to share?
 
How could you possibly know that you have no Roman ancestry? Nobody has a genealogy going back to Roman times.



Would you happen to have a few samples to share?
Nobody can know for sure, when we talk about such deep ancestry, but it would be really illogical for all this Greek population of Asia Minor and Cyprus to be descendants of the Romans, because they score really high Italian and at the same time not at all or very tiny percentages of Balkan.

There have been never such massive Roman settlements at those areas.

Every 23andme sample from Anatolian Greeks (or Cypriots), follow this pattern (more than 90% of the classified as "European" ancestry to be "Italian".

Have a look at mine (first) and then check some Cypriots:
4ef4b4f9f6348d197579b551764a794c.jpg
eca18532705a59eca77dee7d4a3c39d9.jpg
7334a349b5fb84b99c6c040f437b6891.jpg
98f2bfb1f2adda75831ae143c5404e03.jpg
285385cecf31044617403b141d22be69.jpg
1e4f38d432e09708f5228396b0882635.jpg


Sent from my Robin using Tapatalk
 
I now share with 2 people with 4 montenegrin grandparents (really rare to find)

Montenegro (n=2)
Balkan 85.75%
Italian 0.2%
Broadly South European 5.2%
East European 4.05%
British & Irish 1.05%
Broadly Northwest European 1.7%
 
Nobody can know for sure, when we talk about such deep ancestry, but it would be really illogical for all this Greek population of Asia Minor and Cyprus to be descendants of the Romans, because they score really high Italian and at the same time not at all or very tiny percentages of Balkan.

There have been never such massive Roman settlements at those areas.

Every 23andme sample from Anatolian Greeks (or Cypriots), follow this pattern (more than 90% of the classified as "European" ancestry to be "Italian".

That is odd indeed. Let's just say for now that 23andMe misreports some Greek ancestry (Aegean, Anatolia, Cyprus) as Italian. But in western Europe the Italian is probably of Roman origin.
 
I now share with 2 people with 4 montenegrin grandparents (really rare to find)

Montenegro (n=2)
Balkan 85.75%
Italian 0.2%
Broadly South European 5.2%
East European 4.05%
British & Irish 1.05%
Broadly Northwest European 1.7%

Perfect timing. I was just in the middle of making the Balkans map and only Montenegro was missing. ;)
 
Here are more maps.

23andMe_Balkans.png



This admixture is very well defined in term of region. There is a very sharp drop as one leaves the Balkans. It would be nice to have some percentages for Slovenia, Austria, Czechia, Slovakia, Moldova and Ukraine too. Not single sample so far.

23andMe_Northwest_European.png


This component might be linked to some sort of R1b ancestry, but without Scotland, Wales, Austria, Czechia, Slovakia and Spain it will be difficult to determine whether it really represents some definable ancestry or not. At present it looks like a blend of Gaulish and German, like a combination of R1b-U106 and R1b-U152, which could complement quite well the French & German admxiture.
 
Last edited:
That is odd indeed. Let's just say for now that 23andMe misreports some Greek ancestry (Aegean, Anatolia, Cyprus) as Italian. But in western Europe the Italian is probably of Roman origin.
Could be some shared Neolithic ancestry as well.

In any case looks really mislabeled and problematic.

Sent from my Robin using Tapatalk
 
This admixture is very well defined in term of region. There is a very sharp drop as one leaves the Balkans. It would be nice to have some percentages for Slovenia, Austria, Czechia, Slovakia, Moldova and Ukraine too. Not single sample so far.

I share with a person with 3 slovak and 1 hungarian grandparent:
Balkan 13.3%
Italian 0%
Broadly South 3.7%
East European 51.5%
French & German 2.7%
Britsh & Irish 0.8%
Broadly Northwest 10.5%
 

This thread has been viewed 238135 times.

Back
Top