A Note on Methods and Language

What DNA Tests Actually Measure

Similarity, Not Ancestry Fractions — A Framework for Reading These Maps

The Question Behind the Percentages

When a DNA testing service tells you that you are “55% West African,” the natural reading is genealogical — as if roughly half your ancestors came from one place. That reading is intuitive. It is also incorrect.

What those percentages actually describe is how closely segments of your DNA resemble segments found in defined reference population databases. Similarity scores, not ancestral fractions.

I adopted a different framing for this research — one proposed by Kostas Kampourakis and Erik L. Peterson in a peer-reviewed paper published in Genetics in 2023, and one I find both scientifically more accurate and historically more honest.

Kampourakis K & Peterson EL. “The racist origins, racialist connotations, and purity assumptions of the concept of ‘admixture’ in human evolutionary genetics.” Genetics 223(3), iyad002 (2023). doi:10.1093/genetics/iyad002

What the Paper Argues

Kampourakis and Peterson trace the word “admixture” — the term used throughout genetic ancestry testing — back to 18th and 19th century discussions of racial purity in animal breeding and, eventually, in racist theorising about human populations. The concept carried with it an implicit assumption: that there exist “pure” populations that can then be “mixed.”

“The concept of admixture as currently used assumes the existence of pure or unadmixed categories. These do not need to have actually existed but to be able to exist in principle. We argue that this is a problematic notion that accrues from the racialist origins of the term.”

— Kampourakis & Peterson, 2023

Their argument is not that DNA testing is invalid — it is that the terminology obscures what is actually being measured. What genetic calculators genuinely measure is DNA sequence similarity to reference groups, not proportional descent from pure ancestral populations.

They propose replacing “admixture” with “DNA sequence similarity” or simply “similarity” — and suggest that the correct framing of a result like “66% European” is: “66% of the DNA markers tested show higher sequence similarity to the European reference panel than to any other panel used.”

How This Applies to These Maps

The DNA analysis documented in these maps was conducted using four independent genomic calculators on GEDmatch — EthioHelix K10 Africa Only, Dodecad Africa9, puntDNAL K8, and MDLP K23b — as well as Ancestry.com’s regional analysis. Each calculator compares your DNA against a set of reference populations and returns similarity scores.

Throughout these maps and summaries, the language has been updated to reflect the Kampourakis and Peterson framework. Specifically:

Old framing (avoided)	New framing (used here)
55% West African ancestry	~55% similarity to West African reference populations
24% North African admixture	~24% DNA sequence similarity to North African reference groups
Admixture results	DNA similarity profile
Ancestral components	Similarity signals / reference population matches
Your ancestry is X% Y	Your DNA shows X% similarity to Y reference populations
You have Z ancestry	Your DNA is most similar to Z reference populations

What This Does Not Change

Reframing the language does not change the patterns in the data. The extraordinary proximity of the Nigeria_Fulani reference population in the Oracle single-population results — closer than any other population by a factor of two or more across multiple independent calculators — is a genuine pattern regardless of how we describe it. The consistent pairing of West African and North African reference populations in every similarity model is real and coherent.

What the reframing changes is the claim being made. “You are 55% West African” implies a fixed genealogical fraction from a pure source. “Your DNA shows 55% similarity to West African reference populations” makes a narrower, more accurate, and more defensible claim — one that the data actually supports.

For an AST descendant whose paper trail was deliberately severed, this distinction carries particular weight. The goal of this analysis was never to establish a racial identity through percentage fractions. It was to find historical connections, geographic origins, and living relatives. The similarity framing is actually better suited to that purpose.

The Particular Significance for AST Descendants

There is a deep historical irony here. The conceptual framework of racial purity and admixture — which Kampourakis and Peterson trace to 18th and 19th century racism — was part of the same intellectual architecture that enabled the Atlantic Slave Trade. The tools now being used by AST descendants to recover what the slave trade erased were partly built on that framework.

This analysis is an act of reclamation using imperfect instruments. Knowing their limitations and their history makes the work more rigorous, not less meaningful.

The maps that follow document those patterns — described, as accurately as current science permits, as DNA sequence similarity to defined reference populations.

A Note on Reference Panels

Every calculator uses its own set of reference populations — people whose DNA is used as the comparison baseline. These reference groups are constructed by selecting individuals who (among other criteria) have multiple generations of ancestors from the same geographic region. As Kampourakis and Peterson note, this selection itself rests on assumptions — that such populations existed in relative genetic isolation, and that geography maps cleanly onto genetic similarity. Both assumptions are imperfect.

This is why running multiple calculators matters. EthioHelix K10 calls the North African signal “North Africa.” Africa9 splits it into “Europe + NW_Africa + SW_Asia.” puntDNAL calls it “Western_Semitic.” MDLP K23b decomposes it into ancient DNA streams. All four are measuring the same underlying similarity pattern through different reference lenses. When four independent systems using different mathematics and different reference populations all point to the same signals — the West African cluster, the North African/Semitic component, the Nigeria_Fulani proximity — that convergence is meaningful precisely because it transcends any single panel’s assumptions.

The maps that follow are best read as a set — a triangulation of signals across four imperfect but independent instruments, none of which is treated as authoritative on its own.

Kampourakis K & Peterson EL. Genetics 223(3), iyad002 (2023) · Fortes-Lima CA et al. Am J Hum Genet. 2025;112(2):261–275 · Sibomana O. Pharmgenomics Pers Med. 2024;17:487–496 · GEDmatch Kit NJ7476284

This note was prepared as a methodological preface to accompany a series of DNA similarity maps produced for genealogical research. The analysis was conducted by an AST descendant using publicly available consumer DNA tools. Language has been updated throughout to reflect the similarity framing proposed by Kampourakis & Peterson (2023).