search this blog

Monday, February 21, 2022

The Pict

KD001 is the first undeniable Pictish sample in my dataset, courtesy of Dulias et al. 2022. Thanks to Altvred for processing the files.

This is how KD001 behaves in my Celtic vs Germanic Principal Component Analysis (PCA). Looks kind of Irish, doesn't he?

To see an interactive version of the plot, paste the coordinates from here into the relevant field here.

See also...

Celtic vs Germanic Europe

Avalon vs Valhalla revisited

When did Celtic languages arrive in Britain?


Davidski said...


WeightofAudio said...

The Irish have the premium of Bronze Age Isles ancestry, so it seems intuitive for this sample to pull closer to Hibernia than to Britain, who’s inhabitants have more Celtic-speaking incursions (plus Germanic) ancestry.

ambron said...

Overall, I have the impression that the old genetic background of Northern Europe can be divided into two clusters - Balto-Slavic and Celtic. The Germans, on the other hand, are essentially the result of the interaction between the Celts and the Balto-Slavs. Of course, I omit later Ural and Roman admixtures.

John Thomas said...

David, just how much Bell Beaker affinity does this Pictish gentleman have?

Is it likely that the Picts were largely an unadmixed Bell Beaker folk?

Davidski said...


Iron Age Scandinavians don't have any Balto-Slavic-like ancestry.

It only starts showing up in Germanic areas during the Viking period.

Davidski said...

@John Thomas

This Pict doesn't look like a typical Bell Beaker. But he can easily pass for someone from the Bronze Age.

Target: Bell_Beaker_Scotland
Distance: 4.3516% / 0.04351601 | R3P
57.8 Yamnaya_RUS_Samara
27.4 Anatolia_Barcin_N
14.8 WHG

Target: Scotland_MBA
Distance: 3.5104% / 0.03510432 | R3P
51.0 Yamnaya_RUS_Samara
33.0 Anatolia_Barcin_N
16.0 WHG

Target: Scotland_Pictish_EMA:KD001
Distance: 4.2852% / 0.04285222 | R3P
50.6 Yamnaya_RUS_Samara
33.0 Anatolia_Barcin_N
16.4 WHG

Gaska said...

The picts spoke an IE language right?

The Celtic language must have entered Hibernia and Britannia during the Bronze Age. The same thing happened in Iberia because Spanish archaeologists have shown that the Celtiberians did not enter Iberia with La Tene culture, but rather descended directly from the Urnfelder (1300-1200 BC). Celtiberian was the first Celtic dialect to break away from continental (central) Celtic and this means that the Urnfield Culture (Hallstatt) spoke Celtic.

Celts were a very heterogeneous people, very similar to north Italians, Spaniards and French due to the high percentages of EEF admixture, they were very different from the Germanics or Balto-Slavs-Central and Western Europe maintained a very similar genetic makeup during the Bronze and Iron Ages (despite speaking Indo-European and non-Indo-European languages).

Wee e said...

Is this the Rosemarkie Cave body?

Few Pictish-era remains remain, and fewer still that can provide dna. There were a lot of different deposition types going on, even in the same graveyards at the same time: and this is unusual in being a burial that survived in good shape (at the back of a cave, if it’s the Rosemarkie one). Scotland mostly has acid soils.

A caveat that will apply to perhaps more than this burial. No-one knows (at least, as far as I know, no published isotopes) how local the bloke is. Scotland is bisected by the “great glen”, a trade and pilgrimage route that runs between the Moray Firth in the north east and the Argyll coast, via Loch Ness, Loch Oich, Loch Lochy, and Loch Linnhe. It’s a shortcut from Ireland to the continent-facing coast. So — this guy could be local, or equally could have come from up or down the east coast, from the Continent, the west coast, the Hebrides or from Ireland — they’re all within a couple of days’ travel.

I don’t think his dna is any surprise, but as to whether he is typical of the locals in that specific spot at that specific time, there is very little data to compare. There are few remains from this period and this area that have been able to be sampled.

Either roughly contemporary with, or shortly after this burial, the cave was used as a smithy. The possible dating also partly falls into the Christian proselytising period.

There is a blog description of the find with some good pictures of the skeleton, described as “robust”, particularly his arms. (Positioning also suggests brawny &/or fat.) It seems to have certainly been a violently inflicted death. There is also a lot of freewheeling speculation in the blog about other circumstances, which I’d take with a pinch if salt.

Davidski said...

@Wee e

This individual is indeed from the Rosemarkie Cave, and I can tell you that he's fairly similar to late Pictish samples from Dounreay and Westness.

Davidski said...


By the way, I've got a somewhat related question.

Is the sequel to Trainspotting worth watching, or will it ruin the original for me?

Wee e said...

I was careless: there has been an isotope analysis: in summary, the man probably locally raised, but on a more high-protein diet than the average for the time.

This from the above blog.
“The isotope analysis found that the source of protein in the diet of Rosemarkie Man was different to contemporary Pictish populations from elsewhere in Scotland. This may have included freshwater fish, water fowl, or higher-trophic level terrestrial fauna such as suckling (young) pigs. A similar, elevated value was found in a male from a high-status burial at the Pictish site of Lundin Links. This doesn’t prove Rosemarkie Man was Pictish royalty , but it’s certainly a theory that he may have been of high status. We also ascertained that he most likely originated in the local area.”

This spot has a notable waterfowl population to this day. It’s an internationally significant breeding ground for geese from the continent and from Iceland, as well as having a large resident goose population and hordes of Atlantic birds that otherwise don’t come ashore. There are sandbanks and skerries close to shore which were more extensive then. At the time there were also bird species that are now extinct, in huge abundance.

Tom said...


How much Germanic ancestry do Finns have? My impression is that it is sizeable

Did Estonians get some of their Germanic ancestry from Finns?

Wee e said...

I have not seen the sequel. The original is probably better seen first: it does have a definite flavour of the crummy 1980s where we “youths” had, statistically, a higher chance of getting diabetes than a job. (If you can get over posh boy actors versions of schemie accents.) cinematically it has some good moments.

Rob said...

What's the source of the (modest) rise in EEF in MBA & Picti Scotland ?

Davidski said...


How much Germanic ancestry do Finns have? My impression is that it is sizeable.

I don't know exactly, but quite a bit in the southwest, much less in the east.

Did Estonians get some of their Germanic ancestry from Finns?

I'd say so.

Grant said...


"Is the sequel to Trainspotting worth watching, or will it ruin the original for me?"

Personally, I preferred T2, but I may be alone there. In some ways it's even bleaker and the humour is even darker than the original.

(fwiw) The Picts, per se, are now believed to have spoken a P-Celtic language like the other Britons. (In fact, "Pict" may have simply been a Roman catch-all term for "Britons not ruled by us".) The evidence for this includes older placenames, e.g. Perth, which is cognate with Welsh _berdd_/_perdd_ "bush", "shrub", and more distantly, PIE *pert(h) “pole, sprout”. (Also the ancestor of Latin _pertika_ and Mod. English _perch_, btw.)

So, despite the obvious language difference (P-Celtic v Q-Celtic), it makes sense that there would be minimal genetic distance between a Pict and (i.) contemporary Irish and (ii.) ancient south Britons. (That is, Ireland, as an outermost island, would have been a migration terminus and source of back-migration, rather than somewhere from which novel autosomal material entered Britain.)

ambron said...

David, I think the Iron Age Ă–land sample from the Margaryan study looks like it has a Balto-Slavic admixture.

Carlos Aramayo said...

Yamnaya in Serbia?

"The bones of large men covered with red dye, discovered by archaeologists in two mounds in Vojvodina in the northern part of the Republic of Serbia probably belong to people who arrived there almost 5,000 years ago from the steppes of southern Russia or Ukraine."

KM said...

Orcadian VK201 and VK203 from ~500 AD are very likely Picts, right? They don't look much like modern Irish when I project them onto a PCA of many thousands of modern British and Irish people -- they look like an even more northern version of the Scottish IA samples in Nick Patterson's paper -- but the later Orcadian samples VK202 and VK207 samples land right on top of modern Irish, which made me conclude that those ones were Gaels.

Would be good to know if there are reasons to believe that what I said isn't correct!

Also re: Picts passing for Bronze Age groups -- as you probably know, this is sort of true even for modern Irish and Scottish people, in that the Bronze Age Rathlin Island samples look pretty much like a modern-day Irish or Scottish person, in my experience.

Davidski said...


VK201 and VK203 do look like modern Irish.

Have you checked for projection bias in your PCA?

You can run them here...


Grant said...

*Latin _pertica_ "pole", "rod", "staff"

Garvan said...

The Pict looks like North-West Irish in the first two dimensions of the Celtic vs Germanic PCA, but has close distances to bronze age populations of Britain, Scotland and Ireland across all dimensions. Not very close to me (Southern Irish), but a better match to people from the midlands with less drift.

Simon_W said...


"this means that the Urnfield Culture (Hallstatt) spoke Celtic"

Absolutely, at least its western part. The same can be seen in northwestern Italy and southern Switzerland where the western Urnfield-derived Canegrate culture evolved into the Golasecca culture of the Celtic Lepontii. Interestingly enough, unlike Celtiberian, Lepontic is p-Celtic.

KM said...


Well, here's the PCA I'm thinking of:

Matt said...

Ot: New Africa paper just dropped -

Samuel Andrews said...

I don't see thede Scotland samples in G25.

Davidski said...


Look for these IDs.

Queequeg said...

@ D and re: "I don't know exactly, but quite a bit in the southwest, much less in the east."

It seems to me that there's actually quite a lot Germanic ancestry in the East too, but it's older i.e. different from the modern Scandinavian type of ancestry people in the more western parts of Finland usually have.

Genos Historia said...


I find the mtDNA in the Africa study really interesting.

It has lots of basal L0. It also, lacks mHGs fundamental in other parts Africa. Overall, it makes it look like South Africa is less related to Eurasia than West Africa.

I suspect Y DNA E1b in West Africa is of African origin. For the same reason their mtDNA is more related to Eurasians than the mtDNA in East/South Africa.

Y DNA E probably originated in Africa just like mtDNA L3'4 did.

Davidski said...


Thanks, that's an interesting plot.

I think there may be a couple of technical and interpretive issues to take into account when considering whether these ancients are outliers:

- projection bias or shrinkage, in which the PCA space is smaller for the ancients than the moderns because the former are projected while the latter are not

- recent genetic drift and/or relatedness, especially among the Welsh samples, which seems to be masking other genetic substructures.

So to see what's going on, firstly I'd take a few samples from each of the modern populations and project them in the same way as the ancients to make sure that they're behaving exactly as the modern samples that aren't being projected.

Secondly, I'd check the Welsh samples for excessive IBS/IBD sharing, and take out the individuals that are too closely related, and/or reduce the number of the Welsh samples. This should flesh out more substructures in the Irish, English and Scottish samples.

I'd also try some near related populations in the analysis, like from France or Scandinavia, to check whether this can give the PCA more shape that reflects geography and deeper population substructures.

gamerz_J said...

@Genos Historia

If Ydna E is African and that's of course possible, then it will be from East Africa not West imo.From a population related to Mota. mtDNA L3'4 has always been African, it's very old.

Also, imo it's worth checking out their qpGraphs

KM said...


Shrinkage both due to missingness and inherent overfitting is already accounted for. If it hadn't, then scaling the samples away from (0,0) would just push them farther into the no man's land between Ireland and Wales anyhow. Of course in the grand scheme of things these samples look more Irish than anything else, but the fact that they don't land in the modern distribution is what I meant by "not much like modern Irish" - certainly not as much as VK545 and the later Orcadians.

I think degree 4 kin were pruned, possibly 3, can't recall. Welsh drift is real and downsampling the Welsh would just relegate it to a higher-indexed PC. PC3 and 4 are meaningful, both usefully help separate modern Irish and Scottish.

Positions are basically the same when I include the West Eurasian and North African pops from HGDP and 1000G.

Will get the bam for KD001 and project it too when I get round to it.

Do you not think VK201 and 203 are Picts?

Matt said...

@Genos, interesting point to focus on and thanks the explainer on the mtdna side of things.

@KM, is that PoBI data? If there are enough SNPs in common that might make sense for the extensive transect of samples from Britain we have now. Esp to see when the Welsh drift that only shows up in these small scales emerged (e.g. is the hypothesis that it emerged in the post-Roman period from some collapse and re-expansion of the native Brythonic population accurate, or is something else showing up?)... (Bearing in mind Davidski's comments about how to use SmartPCA to get the most meaningful modern-ancient projection on these things, which is definitely one of his specialities...)

Matt said...

(For context of my comment to @KM, that PCA is showing exactly the pattern that showed up in Galinsky's paper using PCA analysis of the White British subset in the UK Biobank - (see Fig 1).

Davidski said...


In that case, my interpretation is that almost all of the ancient samples are basically like modern Irish (except I0769).

But their deviation towards the Welsh cluster is interesting, and worth investigating further.

For instance, is it the result of these ancients not quite fitting into any of the clusters due to recent drift in Britain and Ireland, or is it because of more ancient substructures, like slightly higher levels of farmer ancestry?

As for the ethnic origins of VK201 and VK203, as far as I know, they're from the Pictish era in the Orkneys. But it remains to be seen whether they're typically Pictish or outliers, because there are different theories about the Orkney Picts.

Matt said...

(The most extreme illustration of the structure in Biobank and UK that I've seen is this by Kristjan Moore - )