Relatives in Sweden share about 25 cM more than the Shared cM Project

Relationship by relationship

Average shared DNA in centimorgans. The bar shows how much of what relatives in Sweden share is excess over the Shared cM Project average.

Each row is grouped by the number of generations between the two people, as in the Shared cM Project's own groupings. Half relationships sit in the row of the full relationship they share about as much with, so half first cousins are counted with first cousins once removed.

Why it matters most for distant matches

About 25 cM is nothing for first cousins, who share around 900. But for fifth cousins it is roughly half of everything they share! The further out you go, the larger the part of a match that comes from somewhere other than the relationship you are looking for.

The explanation is pedigree collapse in a stable population. Until the late 1800s most Swedes married within a few parishes of where they were born, generation after generation. Two people with deep Swedish roots are therefore almost always related along several lines at once, even when the parish registers show a single path. Each extra line leaves small segments behind.

This is the same mechanism as in well-known endogamous populations, only milder. And it is not unique to Sweden: the Shared cM Project's own averages already sit 10–22 cM above what the relationships alone should give from first cousins outward. Relatives in Sweden add a second layer of about the same size on top.

What the Shared cM Project table suggests

I took the typical total for each proven relationship and looked it up in the Shared cM Project table. This is the relationship the table points to.

Northern Sweden here means the nine provinces of Norrland, from Gästrikland northward. Southern Sweden means Skåne, Blekinge and Halland.

• Up to second cousins the table is right. That is where most lookups are made.
• From third cousins outward it points too close, everywhere in Sweden. In the north the error grows the further out you go.
• The error always goes the same way. If you have searched for a common ancestor born around 1800 and found nothing, you may have been looking a generation too recent.

The excess also varies between the Swedish provinces. I will report it province by province later, when more data has come in.

What does this mean for those of us researching Swedish roots?

The Shared cM Project is still a good tool, and for close relatives you can trust it. But when a match sits far out in the tree, it pays to think one more time. A match the table rates as a third cousin may in fact be a fourth cousin or even more distant. So look one or two generations further back than the table suggests, and be prepared that you may be related along more than one line.

My aim with this collection is to publish a table that fits Swedish conditions better: one built on relationships proven in the Swedish parish registers, and which takes into account the differences between parts of the country. The more people who report, the sooner we get there.

Who this applies to

The excess comes from shared background between two people whose ancestors lived in the same region for many generations. It applies in full when both people in a match have deep Swedish roots on the line in question. For someone in North America with one Swedish great-grandparent, matching a cousin in Sweden through that line, it will be smaller, because only one side carries the Swedish background.

How the numbers were made

• 5,034 pairs, all with the relationship documented in the Swedish parish registers, which record nearly every birth, marriage and household from the late 1600s onward. No names are collected, and the DNA file never leaves the reporter's computer.
• Almost all pairs tested at MyHeritage or FamilyTreeDNA, since the form reads segment data, which AncestryDNA does not provide.
• I compare averages with averages, because the Shared cM Project reports averages.
• The excess varies a little from row to row, partly with where in Sweden the reports come from. About 25 cM is my summary.
• Both collections are built on pairs that genealogists have managed to identify. Far out in the tree, only pairs who share enough to be found are reported, in the Shared cM Project as well as here.