Ancient DNA helps compare the proposed paternal families. It does not yet identify a patriarch.

The ancestry containing CB-01, the principal Cohen branch called J-ZS222, currently provides the strongest starting point for the connected family history. Its advantage comes from an independently recorded priestly tradition across several Jewish communities. The preference is provisional. CB-02 and CB-03 remain serious alternatives. Z642 has useful Samaritan and regional evidence, but its country pattern does not currently outweigh the priestly evidence. None of these results establishes an identification of Aaron, Jacob, Abraham, Noah or Adam.

Supplement, published 13 October 2026: new ancient-DNA checks from Akhziv, Iran, Sicily and Bakr Awa. The supplement does not change the conclusions of this report.

Existing ancient DNA does contain a strongly supported member of CB-01. The man is I37241. His published central date is about AD 100. I independently checked his DNA reads and found eleven defining CB-01 mutations that survive strict quality and end-trimming checks. Two additional mutations support a probable position inside its Z18271 branch. His archaeological location and community are withheld. He establishes ancient membership of the genetic branch; he does not establish priestly or Israelite identity. Akbari and colleagues, 2026, original tables

The published Erfurt individual I14846 offers a different kind of evidence. His medieval Jewish cemetery gives a meaningful community context. His precise CB-01 placement rests on one surviving informative DNA read, however. The older Megiddo individual I4517 does not supply the missing ancient identification: two reads retain the earlier base at a mutation defining CB-01. The distinction between a strong genetic placement and a strong archaeological context matters throughout this report. Waldman and colleagues, 2022 · Agranat-Tamir and colleagues, 2020

Alcuin directed the investigation, checked the decisive sources and calculations, and made this assessment. Sonnet 5.5 completed the detailed tree, archaeological and population-model research. Focused review assignments corrected errors and challenged the conclusions. The technical record describes the checks, unresolved evidence and model execution. The research files preserve the calculations and working evidence.

The investigation accepts Scripture as authoritative and uses the project’s particular historical interpretation. That interpretation places Noah’s arrival in an already inhabited world in 2903 BC. It does not require a worldwide human genetic bottleneck. It allows Adam’s chromosome to have been copied from an earthly source. These are the premises of this comparison, rather than discoveries made by DNA.

The chronology places Adam at about 5159 BC, Abraham’s birth in 1771 BC, Ishmael’s in 1685 BC, Isaac’s in 1671 BC, and Jacob and Esau’s in 1611 BC. It places the entry into Egypt in 1481 BC and the Exodus in 1266 BC. I have not altered genetic mutation rates to fit those dates. Estimates of genetic age retain their published uncertainty. The effects of the unusually late fatherhood in the interpretation remain uncertain.

The genealogy requires one father-to-son history from Noah through Shem, Abraham, Isaac and Jacob to Levi and Aaron. A proposed Aaronic branch therefore also proposes an ancestry for those earlier men. It does not give the same amount of evidence for each name. The arrival date, the date when a mutation appeared, and the date of the shared ancestor of tested descendants are different events.

A growing family need not become a larger share of its neighbours. Its sons must survive and have sons. Their families must reach the place later excavated. Other families may grow just as quickly. Marriages can spread genealogical descent widely while leaving a paternal line rare. Burial practice, preservation, excavation and readable mutations then determine which descendants can be recognised.

A son of an unrelated mother still inherits his father’s Y chromosome. His expected overall ancestry is diluted, but his paternal line is not. The pedigree model’s one-eighth threshold therefore removes some men who still belong to the paternal family. It also stops forming new marriages above 1,500 followed people in a region and stops generating births after 1736 BC. Its later decline cannot be interpreted as the disappearance of Noah’s male-line descendants.

Sonnet extended the calculation to compare declining, equal, slowly increasing and faster increasing paternal shares. The scenarios use the model’s staggered Levant arrivals and explicitly vary how many incoming men belong to Noah’s father line. They include population growth and, in the stochastic version, extinction of male lines. The calculations use a broad Levant-and-Cyprus region. They do not establish the frequency at Megiddo, Jerusalem or any particular cemetery.

Every range in this table compares an assumed 10% with an assumed 100% of arriving household males belonging to Noah’s father line. It is not a confidence interval. The calculation assumes half of household members are male, about 501,000 resident males near the first arrivals, and 30-year generations. The model’s principal arrivals occur near 2369 BC; other households arrive later. Its population growth is extrapolated beyond the period of its saved population estimates. These deterministic calculations omit chance extinction. The separate stochastic calculations in the technical record include it. A persistent reproductive advantage over many centuries is a strong assumption, not a finding.

For illustration only, Sonnet also simulated 100 unrelated men chosen without burial or recovery bias, with a 20% chance of reading enough DNA in each carrier to recognise the branch. Under the full-founder, equal-reproduction scenario, the probability of recognising at least one carrier near AD 100 is about 0.4%. Under the full-founder, 5%-greater-reproduction scenario it is about 19%. Those are probabilities for an invented unbiased collection under stated histories. They are not probabilities for the actual ancient collections, or for identifying Noah.

These alternative histories show why a fixed waiting period would be misleading. Under equal reproduction, more descendants can exist while their proportion stays small. Under declining reproduction, a later collection can be less likely to contain a descendant. Under sustained greater reproduction, later specimens become more informative. A positive early specimen can establish presence under any scenario. An early negative result requires a credible chance of detection before it can count against a history.

The following table places model outputs beside the evidence available for the same broad periods. “Not calculated” means that the model supplies no valid number. It does not mean zero descendants. The regional archaeological counts use the current AADR index and original papers. Intervals crossing a period boundary remain separate. The model’s Levant-and-Cyprus region and the four modern-country search area are approximations to different geographical populations. AADR, v66.p1, Dataverse release 14.0

These entries preserve source-model meanings rather than combine incompatible models. The later design-stage dates omit a separate 40-year wilderness interval and round generations; they are approximate stage labels. The detailed evidence file contains the longer output series. AADR, v66.p1, Dataverse release 14.0 · Agranat-Tamir and colleagues, 2020 · Feldman and colleagues, 2019 · Haber and colleagues, 2020 · Akbari and colleagues, 2026, original tables · Waldman and colleagues, 2022 · Pallarés-Viña and colleagues, 2026

The later community calculation gives Jacob-paternal shares of 91.4%, 49.9% and 12.6% among its present Ashkenazi male population under its low-, middle- and high-incorporation assumptions. I independently reproduced those three numbers. They are conditional model outputs. They cannot be compared directly with 47 CB-01 men among 104 recruited Cohanim. One denominator contains all modelled community men; the other contains recruited priests. A valid comparison also needs the community frequency of the larger branch that would include Jacob’s other descendants.

Increasing the model’s incorporation of outside fathers lowers its expected Jacob-paternal share. Choosing the top of its assumed male-incorporation ranges lowers the middle result from 49.9% to 33.8%. Choosing the high values for all incorporation inputs lowers it to 12.6%. These are explicit revisions after inspecting the results. They are not independent predictions or evidence that any particular branch belongs to Jacob.

The demographic calculation does not by itself choose a haplogroup. Assigning Jacob’s paternal family to the ancestry containing CB-01, CB-02 or CB-03 produces the same conditional paternal shares until evidence supplies different histories of survival, growth or incorporation. Each present priestly branch would represent only a subset of Jacob’s sons’ descendants. Predicting its frequency therefore also requires the proportion descending through Levi and Aaron. That proportion is not supplied by the model. Z642 presents the converse problem: a proposed Abraham-level branch could include descendants outside Jacob’s family. Its whole frequency cannot be substituted for Jacob’s share.

The proposed branches form alternatives within a connected tree. A haplogroup is a branch defined by shared mutations on the Y chromosome. The outline below omits many intermediate branches. An equals sign means alternative names for the same marker or mutation-defined block at the resolution used here. Sahakyan and colleagues, 2021 · Lipson and colleagues, 2025 preprint · YFull public tree data

The human paternal tree separates E, J and R long before the proposed patriarchs.

J
├─ J1 … P58 … YSC0000234
│  ├─ ZS241 … FGC13863 → ZS227
│  │  ├─ Earlier-diverging relatives, including I45358 (about AD 750)
│  │  └─ ZS222 = Y3088 block → CB-01
│  │     ├─ Z18271 → modern families; I37241 probably belongs here
│  │     └─ FT246135 → another part of CB-01
│  └─ Z1884 / L858 block
│     ├─ Other branches, including the ancestry containing CB-04
│     └─ Z640 → Z2292 / Z2293 block → Z642
│        ├─ ZS9008, ZS1387 and FT133907
│        └─ FGC30542 / ZS5746 … FGC62618
│           ├─ Published Samaritan and Druze representatives
│           └─ A later branch containing I43272 (about AD 1700)
├─ J2 … PF5087
│  ├─ M67 ancestry → the Danfi Samaritan family (broad placement secure)
│  └─ PF5160 ancestry
│     ├─ PF5169: a proposed finer Tsedaka placement remains unverified
│     └─ … PF5366 → FGC4975 / FGC4992 block → CB-03

E … M78
├─ V12 ancestry … Y2850
│  ├─ … CTS7178
│  │  ├─ BY8505 → BY8508 → CB-02
│  │  └─ … CTS9507 → Tàrrega ROQ2 and Jewish relatives
│  └─ … BY8536 / BY8528 → BY8529 → Levite-associated families
└─ V22 ancestry: the finer Samaritan priestly placement needs confirmation

R … R1a-M582 → Y2619 ancestry → the principal Ashkenazi Levite family

CB-01 and Z642 meet at YSC0000234. FamilyTreeDNA estimates that shared ancestor at 3352 BC, with a reported 95% interval of 4117–2682 BC. The central estimate is earlier than the proposed Noah arrival, and much earlier than Abraham. Their separate histories cannot both identify Abraham’s and Aaron’s paternal branch under the ordinary tree and clock. Their shared older ancestry does not identify Noah. The previously withdrawn identification of Noah with L858 is not restored. FamilyTreeDNA Discover

CB-01 and CB-03 meet much earlier, at the division between J1 and J2. The J1 and two J2 Samaritan families also have ancient separations. Loss of other sons’ lines cannot make these surviving, deeply different chromosomes descend from one recent paternal founder. At least some present tribal or priestly traditions require a different history of family membership. The same scrutiny must apply to Jewish and Samaritan traditions. Shen and colleagues, 2004 · Oefner and colleagues, 2013

A branch’s “formation” estimate generally refers to its separation from the parent branch. Its common-ancestor estimate refers to the tested descendants within it. Neither date names a historical person. CB-01’s published shared ancestor is around 833 BC, with a reported 95% interval of 1289–440 BC; Table 3 gives the slightly different 834 BC and 1291–440 BC. Descendants of an earlier Aaron could still share a later surviving ancestor. No clock adjustment is needed for that possibility. Lipson and colleagues, 2025 preprint

The comparison favours continuity of CB-01’s priestly family, but does not establish it. A coherent history must explain both the surviving branch and the other families whose traditions would need incorporation, mistaken attribution or loss of earlier lines. An unidentified paternal family must also explain why the recognised priestly lines survived and acquired their traditions. It is not an assumption-free alternative.

Lipson and colleagues, 2025 preprint · Shen and colleagues, 2004 · Oefner and colleagues, 2013 · Waas and colleagues, 2019, Figure 1 and Table 1 · Rootsi and colleagues, 2013 · Levy-Toledano and colleagues, 2026

The strongest modern observation is CB-01’s priestly association. Lipson’s researchers recruited 104 men with reported priestly practice in their own, their fathers’ and their grandfathers’ lives. They deliberately sought diverse origins. CB-01 contains 47 of those 104 men, including 27 of the 43 Ashkenazi recruits and 20 of the 61 others. These are recruitment-sample fractions, not estimates for all Cohanim or all Jewish men. The preprint remains version 1, posted in December 2025, with no journal version recorded in the checked bioRxiv metadata. Lipson and colleagues, 2025 preprint

Z642’s geographical pattern survives as an observation in the saved commercial data. Its purported statistical strength does not. I reproduced one qualifying result among 66 resolved rows, but the rows represent 65 distinct branches. At least 33 of those branches lie within other selected branches. The Arabian condition was at least 80% of known origins in six specified countries, rather than exclusively Arabian origins. The eight men assigned to the “Ashkenazi countries” category within the relevant Z642 child all reported Germany. They lie in one younger paternal family. The country does not establish Jewish membership.

The criteria file was written after the saved Z642 country pattern had already been printed. That makes the comparison an exploration of a noticed pattern. It cannot be presented as an independent prediction. Changing several count thresholds leaves Z642 distinctive in this dataset, but that does not repair the missing community labels or the overlapping branches. One result among 66 rows is not a probability that Z642 is Abraham’s family.

The Samaritan evidence strengthens the case for taking Z642 seriously. Waas’s published Figure 1 places a Samaritan-labelled man with a Druze-labelled man within FGC62618, below Z642. I inspected that figure and its original age table. Identifying that particular Samaritan sample as the Marhiv family remains an inference. Shen’s original study securely distinguishes the Marhiv J1 family, the Danfi J2-M67 family, and the Tsedaka J2 family outside M67. The public evidence does not justify treating all three traditions as equally precise genetic pedigrees. Waas and colleagues, 2019, Figure 1 and Table 1 · Shen and colleagues, 2004 · Oefner and colleagues, 2013

The E alternative has become more concrete. Levy-Toledano’s 2026 paper shows Levite-labelled Moroccan and Ashkenazi descendants within E-BY8529, below E-BY8528 (called BY8536 in the saved YFull tree). That family joins the CB-02 ancestry at E-Y2850. The medieval Jewish ROQ2 lineage joins CB-02 more recently, at E-CTS7178. I checked both relationships in the tree and inspected the paper’s figure. The paper combines surname and tradition evidence and gives inconsistent Moroccan subgroup counts, so I have not converted its labels into a population frequency or a Levi identification. Levy-Toledano and colleagues, 2026 · Pallarés-Viña and colleagues, 2026

Older studies independently support priestly and Levite associations, but they measured different markers. Hammer’s extended Cohen pattern used twelve repeating Y markers; it was not an exact J-ZS222 test. Rootsi’s Levite study found marked R1a-M582 enrichment, but some contributing collections deliberately recruited priests and Levites. Its 63/97 and 8/51 results describe those Levite recruits. Broad regional Middle Eastern sequencing helps explain shared ancestry and uneven growth of paternal families. It does not identify a particular modern Arabian branch with a named son of Abraham. Hammer and colleagues, 2009 · Rootsi and colleagues, 2013 · Almarri and colleagues, 2021

The earlier claim of an 85–100% chance of detecting another Jacob line does not yet supply a valid test of these candidates. The simulation gives every man a perfectly known tribal label. Real DNA does not supply those labels. Its result also depends on imposed modern family shares and on how it conditions on Aaron’s survival. Separately, its Erfurt calculation mistakenly drew 33 men: the paper studied 33 people, including 14 males, of whom ten received Y assignments. Correcting that denominator changes its Erfurt carrier probability; it does not by itself invalidate or repair the separate modern-tester calculation. Waldman and colleagues, 2022

The ancient individuals provide dated genetic relationships with different levels of contextual support. “Earlier base” below means that a readable position retains the state before the mutation defining the proposed branch. A position with no usable reads supplies no evidence either way. Dates from the anonymised 2026 dataset are published means; an individual interval or dating method was not supplied in the examined table.

The three medieval Jewish collections supply seventeen men with published Y assignments and no identified close male-to-male relationships that require their removal: ten at Erfurt, three at Norwich and four at Tàrrega. They show several paternal families in later Jewish communities. They do not represent seventeen independent chances to recognise a specific deep branch: coverage, mutations tested, cemetery selection and more distant kinship differ. Their existence is a reason to investigate persistence and spread, rather than stop at sparse Bronze Age evidence. Waldman and colleagues, 2022 · Brace and colleagues, 2022 · Pallarés-Viña and colleagues, 2026

The direct read checks are targeted reanalyses of existing aligned DNA, not new laboratory results. I checked mapping and base quality, removed problematic alignments, varied trimming of read ends, and distinguished independent molecules where possible. Published damage and mitochondrial-consensus measurements support ancient origin for I37241’s libraries. I did not estimate contamination anew. The technical record retains the precise mutations, qualities, missing positions and conflicting calls.

The amount that can be inferred falls as the proposed history moves back from the observed families.

The provisional preference for CB-01 is most stable when recorded priestly continuity receives substantial weight and when surviving branches can pass through later bottlenecks. CB-02 could gain substantially from a securely documented related Levite family with the correct branching time. CB-03 could gain from verified priestly continuity in its eastern branch. Z642 could gain from documented Jewish community identities and securely contextualised ancient descendants on different relevant branches. A low survival rate for original priestly lines would weaken all three Cohen candidates and make an unrecognised line more plausible. None of those assumptions has a measured probability here.

Existing material can improve this comparison before new laboratory work is proposed. The best next work has two parts. First, use the available higher-coverage or indexed DNA files for known-context Bronze, Iron, Roman and medieval men. Test the specific mutations that separate CB-01, its earlier relatives, Z642, CB-03 and the E alternatives. Second, verify the family and community records attached to modern men who may divide those branches. Neither broad haplogroup labels nor commercial page headings can substitute for those checks.

The most useful next medieval read checks are Norwich’s high-coverage J1 man SB604 and Tàrrega ROQ2’s E branch. The present Erfurt file supplies only one decisive CB-01 read; repeating the same check cannot create missing sequence. Recovering the original analysts’ exact mutation evidence could clarify their published manual assignment. The best earlier regional comparisons include Alalakh ALA026 and ALA018, enhanced Baq’ah data, Sidon and Beirut. Each can test a different part of the tree. A confirmed older relative can constrain ancestry without being a descendant of the patriarch proposed for a later branch.

The new anonymised men already improve genetic timing. Their withheld archaeological information prevents a corresponding improvement in historical identification. A search snippet does not establish the proposed Judahite burial identity of anonymised I14918. His public region label and point date do not supply that missing context. Mödling’s proposed CB-01 men remain useful leads, but their public raw reads need alignment and their cemetery context needs its own assessment. Those tasks require analysis, not fresh sampling.

The Ancient Genome Diversity Project was inspected rather than dismissed from its title or age. Its higher-coverage material includes an Early Bronze Age J2b man from ‘Ain Ghazal. It does not presently supply a known specimen that decides among the principal proposed families. Some enhanced older samples are now public through the 2026 study. Restricted files and archaeological metadata were not requested from researchers. Ancient Genome Diversity Project · Akbari and colleagues, 2026, original tables