Page Actions

Familial and genetic descendancy: conflict or complement?

From ISOGG Wiki

A consideration of seemingly contradictory genealogical conclusions when comparing traditional documentation and the results of DNA testing

By Donald Dickason

This article was written on 10 May 2005 and was originally posted on the ISOGG website in 2010:

To begin at the beginning: this situation arises when there is documented evidence of a lineage, but two or more lines do not have matching DNA. Or conversely there is strong genetic evidence that suggests line linkage, but traditional documentation is missing. The nub of the issue has been the acceptance of a “false dichotomy” between the two situations. It is not a matter of there being one to the exclusion of the other. In order to clarify my thinking, I have recently defined these two concepts as “Familial Descent” and “Genetic Descent.”

Opinion has evolved significantly on this issue since the onset of general genealogical DNA testing in 2002. The source of the first situation cited above, documented family linkage but not DNA matching, has been ascribed to non-paternal events (NPEs). When general genealogical DNA testing began a scant three years ago, a “non-paternal event” (NPE) was considered almost synonymous with illegitimacy. Considerable study since that time clearly indicates that “non-paternal” status has occurred for many reasons. There could be illegitimacy, but there also are other causes. The natural mother could die in child birth, and the child would go to live with an aunt or uncle or cousin with a different surname. The child’s father died, and the mother remarried, with the child taking on the new husband’s surname. In Scotland there have been many cases in which a husband simply takes his wife's name. Beyond all these examples, the reason for a non paternal event simply may be indeterminate.

I have not been able to find large scale scientific studies of the rates of non-paternality, but various estimates have been in the range of 3% to 10% per generation.

One review, published in April 2004 paper by Professor Kermyt Anderson, University of Oklahoma, reviewed and studied 66 published studies on the per cent of non-paternal events in father-son tests[1]. Anderson classified the results into three groupings, based on the test subjects’ certainty in their confidence of the knowledge of their own paternity. The non-paternity rates for the 3 groups were 1.9%, 3.9%, and 30.9%. Fathers with high paternity confidence seem to be correct in assessing non-paternity, the 1.9% rate. “High and unknown” levels of paternity confidence exhibited the 3.9% rate. Those who didn’t know, e.g., low paternity confidence, had the highest non-paternity rate, n.b. 30.9%.

For the purposes of this discussion I have concluded that the 3.9%, rounded to 4% per generation is at least representative of what we could expect. All non-paternal rates are highly idiosyncratic, depending on location, time, religion and other factors. But remember that a 4% non-paternality rate could be made up of about 2% illegitimacy and 2% other causes.

If other readers are aware of definitive studies, I ask that you share those citations with the group.

Some in the genealogical community feel strongly that a person who is not a “genetic descendant” cannot be a member of the “family.” Others have argued that DNA doesn’t matter; if the documentation is there the DNA must be wrong. Both these arguments were advanced early and often in genealogical discussions. And both these positions often but not always amounted to declarations of mutual exclusivity.

I argue that the mutual exclusivity premise is a false premise. Surely, if one has been a part of a particular family’s history and heritage, who are we to deny the reality of that family engagement or who are we to say that the geneticists have it wrong? In a case of familial identification and differing DNA information, there is still that living, breathing, real-life person who has all the richness of the familial relationship and the genetic descendancy revealed by the DNA.

Here is a real life case from a family surname project:

Source Friederich b. ca 1650 Germany
1. Son Able b. ca 1680 Son Cadwalader b. ca 1688
2. Grandsons Grandsons
3. G Grandsons G Grandsons
4. GG Grandsons GG Grandsons
5. GGG Grandsons GGG Grandsons
6. GGGG Grandsons Two GGGG Grandsons
7. Andrew - DNA TESTER #1
Basil - DNA TESTER #2
Clarence - DNA TESTER #3
Donald - DNA TESTER #4

In the this family history, Andrew is the source of the concern about a difference between familial and genetic descendancy. The divergence happens somewhere between Andrew and his GGGG Grandfather Able, b. ca 1650, Ireland. It is not at all certain where the familial/genetic disconnect occurs.

The line from Friederich to GGGG Grandson Andrew (DNA tester #1) has been documented. Friederich had two known sons, Able (b. ca 1680) and Cadwalader (b. ca 1688). Able’s known documented descendants include Andrew and Basil. Cadwalader's known documented descendants include Clarence and Donald, also GGGG Grandsons of Friederich.

Thus all four GGGG Grandsons descend by document from Friederich, and then the data become mixed.

As stated both Andrew and Basil descend from Friederich through Able based on family documentation. Importantly, they closely match each other on the DNA tests. Clarence and Donald also descend from Friederich but through Cadwalader, and they match each other on their DNA test.

But the DNA results for Andrew and Basil (call them group AB) don’t match the results for Clarence and Donald (call them group CD).

Group AB’s DNA does not match any other members of the overall Surname Project; thus group AB is defined as “ungrouped.” Their pattern of the DNA markers, or haplotype for Andrew and Basil, Able’s descendants, are similar to haplogroup “N” which distributed throughout Northern Eurasia. It is the most common Y-chromosome type in Uralic speakers (Finns and Hungarians)..

Group CD’s DNA matches very well with one of the overall project’s major family grouping. That pattern is closely associated with haplotype R1b, the most common haplogroup in European populations. It is believed to have expanded throughout Europe as humans re-colonized after the last glacial maximum 10-12 thousand years ago.

Depending upon how a reader of the Surname Project follows the various names and ancestors, one can come to different conclusions. The most general is that they all descend from Friederich. Only the keen reader and DNA enthusiast would ever be able to determine that the DNA match is missing.

And having discovered that there is a discrepancy and that there is no ready explanation for the discrepancy, some conclude absolutely that groups AB and CD are not related, period.

In this commentary we know that the distance from Andrew to GGGG Grandfather Friederich includes 7 generations, or SIX father-son events. 6 events times 4% NPE rate suggests a 24% probability of a non-paternal event in that many generations. We should not be surprised that “non-paternal” statuses do arise. In genealogical study spanning 300 years, not uncommon in the studied Surname project study, 12 generations (11 father-son events) would suggest a probability of a “non-paternal event” on the order of 44%!

Returning to this family’s situation, it is likely that the line from Friederich to Andrew is the odd one out in this comparison since both the Cadwalader descendants are close matches with each other, and are good matches with the well-established family grouping in his Family Surname Project. In the Friederich to Andrew line it could be argued that Able himself was the result of a non-paternal event, but that itself would be an assumption. The non-paternality could happen at any father-son event beginning with Friederich to Able and down the line to Andrew and Basil (more likely to their father since their own DNAs are comparable).

Thus it is reasonable to hypothesize that the ancestry of the non-paternal person in that line was of that ilk. What we cannot determine from genetic testing is where in the chain the haplotype “N” DNA input might have been introduced. which would cause further descendants not to have the “R1b” haplotype. In the end that doesn’t matter.

The Andrew line from Friederich via Able is of familial descent. This line is not of genetic descent from Friederich. No apologies nor arguments for mutual exclusivity need to be raised.

Special note: Even given my conclusions concerning the familial and genetic descendancies of this family, the family historian/geneticist must be cautious of the possibility of human or mechanical error. DNA labs do make mistakes, rarely but still possible. And who of us is to say that we have never made a mistake, or have not found a mistake in non-laboratory material that we have used for documentation. We all know the old saying that “If something seems too good to be true it probably it is.” I think we all ought to accept a “new saying” that “If something is too bad to be believable that it might be wrong.” We should recheck everything, including laboratories’ work. For example FamilyTreeDNA will rerun an analysis on request. I am aware of one rerun which turned up nothing different, and another that turned up a significant change. I continue to believe that mutual exclusivity is the wrong premise, but check and recheck all our information and be absolutely sure that the facts on both sides are correct.

Finally, I am the Group Administrator for the Dickason and Garrett Surname Projects. The Dickason studies in particular have a significant number of non-paternal event situations, some of which have been definable, some of which are indeterminate to date. But what has happened, significantly, is that I no longer refer to these projects as “DNA Surname Projects,” but rather, simply, “Surname Projects.” I am no less interested in those with non-matching DNA who have documentation back to a common ancestor than I am in those who have documentation and DNA matching. I make this very clear in my own reports. In my recent Dickason update I refer to each of about 12 ancestral groupings. I indicate in the opening summary for each descendant in an ancestor grouping those who have tested DNA, those who have not, and those for whom the NPE element seems to be present, and to the degree that we know, the other surnames with whom these folks are closely matched.

All these persons are familial descendants and, as individually identified, may or may not be genetic descendants of the common ancestor. And that is OK.


  1. Anderson, Kermyt G. (2006). "How Well Does Paternity Confidence Match Actual Paternity?". Current Anthropology 47 (3): 513–520.