The DNA

Her position in the descent is what makes any of it readable: she is one quarter Pablo Armando. Her other three grandparents are Mary Septima Taylor — English, of Devon and Suffolk — and two Booyzens, Cape Afrikaner.

A genetic group is not an ethnicity percentage. It is a cluster of people who share actual segments of chromosome and whose documented trees converge on one region. This one converges on the Río de la Plata.

It can only have come from him. English Devon and Cape Afrikaner do not produce a Río de la Plata cluster. Of her four grandparents, exactly one was born in Argentina.

So the Argentine origin now stands on three independent legs: his own 1940 passport; the 1908 army enrolment card found in a drawer in Buenos Aires; and a chromosome. A fortnight ago it stood on one.

And the group’s top places for 1900–1950 name Santa Fe. Not merely Argentina, and not merely Buenos Aires — Santa Fe, by name, alongside Buenos Aires, Corrientes and Entre Ríos. That is the first time anything outside this family’s own paperwork has independently pointed at the province the passport and the card both give.

There is no Spanish component whatever — at MyHeritage. If Pablo Armando had been the Spanish-descended criollo this archive has quietly assumed, a granddaughter should carry something in the order of 25% Iberian. MyHeritage gives her none; 23andMe gives her 2.1%, which is also far short of 25%, so the shape of the argument survives even though the flat zero does not.

What she does carry is Italian — and her other three grandparents are English and Afrikaner, so it has nowhere else to come from. At one quarter, 11.1% in her implies roughly 44% in him: about one Italian parent.

And at MyHeritage the split is northern-dominant, 6.1% against 3.6% — which this archive read as the signature of the Ligurian and Piedmontese emigration that made Rosario the most Italian city in Argentina. 23andMe points the other way and names Sicily. Two panels agree that the Italian is there and disagree about which end of Italy it came from, so the northern reading is withdrawn as a DNA argument. The Ligurian names in the Rosario register are still in the Rosario register; they were never a DNA claim.

A technical point first: the confidence slider governs the genetic groups only. The ethnicity percentages are identical in both views — the same thirteen values, the same figures. Nothing above changes.

The asymmetry is worth stating plainly. The Río de la Plata group is a positive result: a cluster formed, at high confidence, from 178 kits. The Italian claim rests on a negative — an absence of Iberian percentage — plus arithmetic on an 11.1% figure. Those are not the same class of evidence, and this archive should not treat them as though they were.

And one thing the eight groups do confirm. Every group except the Río de la Plata one is accounted for by her other three grandparents — two South African, two British Isles. That is what a clean four-grandparent reading looks like, and it makes the Argentine group harder to explain away rather than easier. The thing it proves, it proves more firmly than before.

A null is worth nothing until something known returns in the same form. Four surnames from the other three grandparents were run first, and all four came back — including matches for whom the surname appears only inside their family tree, never in their own name. The search reaches into trees. Only then were the nulls believed.

The MyHeritage side says the same thing, and says it over the whole corpus rather than a sample. MyHeritage publishes its own count of where all 11,581 of her matches live. Sixty-two countries are named — down to a single match each in Vietnam, Saint Lucia and the Cayman Islands.

Against that: 2,508 in the United States, 1,940 in the United Kingdom, 1,064 in South Africa, 788 in Australia, 515 in the Netherlands.

Spain shows 38 and Portugal 25, which looks like an Iberian signal and is not one. Read the names and they are van Heerden, Retief, Vorster, de Vos, Deetlefs, Hanekom — South Africans who retired to the Algarve. Country of residence is a trap, and this is what it looks like.

And MyHeritage has computed a documented relationship path — a Theory of Family Relativity — for 500 of her matches. Every one of the ten largest resolves to the Cape: Booyzen, Barry, Kolbe, Kruger, Pretorius, Botha, Jordaan, du Toit, De Villiers, Malan. Not one of them reaches Pablo Armando.

There is exactly one. A Uruguayan family — three kits sharing a single tree of 112 people. The eldest of the three shares 21.4 cM across two segments, largest 11.3; MyHeritage rates him a fifth cousin at medium confidence, and reports that the ancestral places common to him and her include Argentina. They are living and are not named here. Their tree was opened and read.

It is Basque-Navarrese and French-Basque, settled in Uruguay: Ciganda, Ziganda Erviti, De Goni, Rodriguez Salvatierra, Pereyra, Barneche, Larralde, Iraola, Echeverria, Cha, Arnaldi, Giriboni, Larrieu, Laclau, Perrone — back to Francisco Ziganda Erviti; and Pedro Maria Ciganda, 1822-1882. Those people are long dead, which is why they can be named here when the living cannot.

There is no Lerena in it. No Llerena, no Lenguas, no Traibel, no Aguilar, no Fernández. It is a real Río de la Plata connection and it does not corroborate the hypothesis. At 21 cM on a largest segment of 11 it is also close enough to the noise floor that this archive would not build on it either way.

The only other Lerena who has tested produced 19 clusters — and His chart is almost entirely ASHKENAZI JEWISH - Abrams, Litwack, Rothbart, Epstein, Goldman, Rosenzweig, Levitan, Novik, Aizenman, Graivier, Lumelsky, Mandel, Cohen, Goldberg, Weisberg, Waldman, Goldstein, Fishbein - with tree names in Hebrew and Russian. That is his MOTHER's family, not the Lerena line, and an endogamous population throws up far more high-centimorgan matches than an outbred one, so it crowds his chart out.

He shares exactly 3 clustered matches with his aunt. All three of the matches he shares with her sit in ONE of her clusters - the Cape one that holds Butler, Daniel and Daschner.

The only other tested Lerena contributes three clustered matches to this question, and all three are Cape. This is also the clearest statement yet of why he was never going to separate the lines: he is her NEPHEW, so the Lerena half of him arrives filtered through the same Cape grandmother.

Cluster 8, one match at 80 cM, a Kroon tree. A FALSE POSITIVE, and caught. Absent from the aunt's clustered chart, so the subtraction flags it - but MyHeritage's own Theory of Family Relativity for that match names the surnames common to them as KRUGER, VENTER and NEL. Those are Cape. It is maternal, and it simply was not clustered on the aunt's chart. Absence from a CLUSTERED list is not absence from a match list, and this is the case that proves it.

Its largest member is 116 cM - comfortably above the aunt's floor, so it had every opportunity to appear on her chart and did not. It is the one genuine paternal cluster the subtraction finds.

NOT Pablo Armando's - and NOT, on testing, demonstrably Brayley either. Mary Septima Taylor's mother was MARY ANN HELLEN BRAYLEY of Bideford. The surname resemblance was attractive and this archive wrote 'probably' on the strength of it. THEN IT RAN THE TEST. The real Brayley matches exist and were found - one Australian at 23.0 cM sharing BRAYLEY and BAKER, one Englishman actually surnamed Brayley at 16.2 cM - and NEITHER IS IN THIS CLUSTER. None of the three Baileys carries Brayley as an ancestral surname. The Taylors →

And the absence proves nothing either, which is why this is left open rather than closed. The Bailey trees hold TWO people and FOUR people, and MyHeritage cannot compute a shared ancestral surname from a tree with two people in it. THE LINK IS UNTESTED, NOT REFUTED. The one thing that would settle it is a message to the 116 cM match asking for two generations of her grandparents' names - and that is David's to send, not this archive's.

And a second pointer the same way. Below the aunt's floor, so the subtraction is silent on it - but WALKER runs through the Taylor line as a given name: Margaret Hellen WALKER Taylor and Agnes WALKER Taylor, two of Mary Septima's sisters. That naming pattern normally means a Walker grandmother. Independent of the subtraction, and pointing the same way.

It pulls in two directions, and both have to be said.

Stronger: the Italian. It now rests on two companies rather than on one company’s zero. That is a different class of evidence from where it stood this morning.

Weaker: the Uruguayan trunk. If Pablo Armando descended from Sancho Llerena’s Montevideo family — eight children born 1793–1815, and a visible nineteenth-century presence in the city — then somewhere among 11,581 matches with 11,114 trees, and 1,500 more at another company, a cousin of that family might be expected to appear. None does.

And the clusters put that in its hardest form. Her fourteen clusters are exhaustively Cape and British. Her son’s chart, run the same night, throws up two Croatian clusters for his father — so the tool plainly does find a foreign paternal line when one has tested. It found his father’s. It did not find hers.

And the honest counterweight to that. Twenty countries of grandparent-birth among 1,500 relatives, and none of them South American, is not really a fact about this family — it is a fact about who buys DNA tests. Uruguay and Argentina are barely present in either database. An absence measured in a place where nobody is standing is a weak absence, and this page will not pretend otherwise.

What would actually settle it is not more searching. It is Parental Inheritance, which would put the Italian on one side of her or the other; and a Rosario baptism for Pablo Armando naming his father. One costs a subscription. The other is in a Córdoba parish register that is not filmed.

An ethnicity estimate is not evidence of the same class as a record. At one quarter the noise is large, reference panels bleed into one another, and MyHeritage in particular is known to distribute Iberian awkwardly across French, North Italian and Sardinian. A 0% Iberian reading is suggestive. It is not a disproof of Spanish ancestry.

The genetic group is much the stronger of the two signals, because it rests on shared segments and documented trees rather than on a continental model. And note what it actually says: it says Río de la Plata. It does not say Italian.

1. See the other five genetic groups. Only three of eight are shown at high confidence. If an Italian group sits among the other five, this moves from arithmetic to evidence. One slider, one screenshot.
2. Work the matches, not the percentages. A single match with a documented Argentine or Italian tree is worth more than any estimate.
3. Re-read the sweeps for LERINA. The Rosario index sweeps recorded no Lerena, Llerena, Larena or Lorena — Lerina was never separately looked for, and it sorts adjacent.
4. CEMLA under LERINA and LARENA, treated as primary spellings rather than variants.