# Priestly tradition, chronological placement and surviving branches

9 September 2026. Local investigation authorised by Mike Russell. No website edits or publication.

## Conclusion

The supplementary descriptions support an old inherited priestly tradition in the major J1 cluster. They do not distinguish Aaronic inheritance from an early non-Aaronic founder whose descendants inherited the same tradition. E-BY8508 also favours ancestral priestly status in the reduced comparison. Accordingly, J1 remains a worthwhile first research target, but the status-change calculation does not establish a posterior ranking of Aaron candidates.

A separate check of the sanctioned local tree finds more old surviving divisions in the major J1 cluster than in the E or J2 clusters. That is positive evidence for its surviving diversity, conditional on the tree's dating. It is not a modern population estimate, and the present dump contains no priestly-affiliation data to identify those divisions as Aaronic.

## Inputs and scope

Inputs are the main Lipson et al. preprint and Mike's three Desktop supplements, already described in ../lipson-cohanim-review/SUPPLEMENT-REVIEW.md. No new empirical historical gain/loss rates have been supplied or assumed. We used the local sanctioned current_tree.json from the adam-y-phylogeny project; no requests to yfull.com.

The status exercise uses reduced witness trees, not the full sampled genealogy. A positive witness means the supplied discussion explicitly reports priestly tradition somewhere in a division. It does not mean every man in that division is priestly. Negative witnesses concern reported absence of tradition, not disproven Aaronic biology. Non-Jewish divisions without an explicit report of Aaronic tradition are treated as unknown, rather than automatically as negative. Unreported within-division variation is omitted, so scores are conditional minima for the reduced observations, not total changes in the real population.

The four witnesses under the eastern J2 division have incompletely specified mutual placement; their multifurcation in our reduced tree is an explicit convenience, not an inferred genealogy. This particularly limits numerical comparison involving J2.

## Investigation 1: how much ancestral priestly status explains

The executable analyse.py uses dynamic programming. For each node it computes the minimum sum of status-change costs conditional on that node being priestly or non-priestly. Unknown leaves can take either state without penalty. Gains and losses are first given equal cost, then gain cost is varied from 0.1 to 10 relative to loss cost 1. These weights are penalties, not probabilities or estimated historical frequencies.

At equal costs:

| Reduced tree | Root priestly | Root non-priestly |
| --- | ---: | ---: |
| Main J1 expansion, J-Z18271 | 1 change | 8 changes |
| Broader J1 CB-01, J-ZS222 | 1 change | 3 changes |
| E CB-02, E-BY8508 | 0 changes | 2 changes |
| J2 CB-03, J-FGC4992 | 1 change | 2 changes |

The eight positive J-Z18271 divisions are FT280381, FT303739, ZS2458, BY64521, Y146901, S17446, FT34605 and FT157560. ZS2434 supplies a non-Cohen witness; ZS237 and Y191714 remain unknown for this coding. At that reported multifurcation, a priestly root needs one loss; a non-priestly root needs separate acquisitions in eight child paths. This is an intelligible advantage for ancient inherited status, not eight-to-one odds.

The broader J1 comparison is less dramatic. A non-priestly ZS222 ancestor could have priestly descendants in the two major sub-branches before they subsequently diversified. The reduced minimum then has two gains and one loss. It is wrong to require eight independent acquisitions merely because a more remote ancestor was non-priestly.

At J-Z18271 the preference changes if a gain costs less than one eighth as much as a loss. That is an algebraic property of these scores, not an estimate of plausible rates. For the simplified J2 comparison, the equal-score boundary is a gain cost of one half. E favours the priestly root throughout the positive weights considered because all its retained observations are positive. Its zero is a consequence of the available observations, not proof of perfect preservation.

Polytomy uncertainty matters. If all eight positive J1 divisions were descendants of one additional, unobserved common branch, a non-priestly J-Z18271 root could explain them through one gain. The equal-cost scores become 1 versus 1. This alternative resolution is a sensitivity experiment, not a claim that the supplied tree contains that branch. To distinguish rapid genuine diversification from unresolved early divisions requires more complete genetic information.

Selection matters too: the study selected branches partly because they have high priestly affiliation. A test of priestly clustering across all Y lineages must account for that selection; this exercise does not calculate a significance level.

### Why this cannot identify Aaron

Two histories can produce identical observed states and topology:

1. The root is descended from Aaron and transmits priestly tradition.
2. The root is not descended from Aaron but already has priestly tradition, and transmits it in exactly the same way.

Any likelihood depending only on the subsequent status transitions gives these histories the same likelihood. One must add evidence or assumptions about how early non-Aaronic status could arise to distinguish them. Merely naming the root Aaronic does not change the calculation.

Taking the biblical genealogy seriously does not erase this distinction. It establishes what genuine Aaronic ancestry entails, but does not label a presently reconstructed genetic ancestor. Moreover, the same method favours ancient priestly traditions in multiple deeply separate genetic groups. Those cannot all be ordinary paternal descendants of one recent Aaron. The outstanding issue is the biological identification of the tradition, not whether Scripture must independently be proved.

## Investigation 2: chronological placement

The paper places CB-01's common ancestor about 833 BCE, CB-02's about 697 BCE and CB-03's about 539 BCE. All are later than Aaron at Mike's Exodus date of 1266 BCE and Noah around 2900 BCE. This is compatible with surviving subfamilies descended from those men. It does not require Aaron or Noah to equal any of these younger common ancestors.

The paper's upstream connection for J1 is about 2150 BCE, so its displayed ZS222 segment cannot represent the entire Adam-to-Noah interval. Noah would need to lie farther upstream under those dates. The local tree puts the corresponding ZS222/Y3088 block between 5000 and 2800 years before present, a materially different upstream boundary. Its marker membership also differs. These are two reconstructions to reconcile, not alternative dates from which to choose a preferred fit.

The J2 block is much longer in both descriptions, with a Neolithic upstream connection and an Iron Age common ancestor. A long unresolved segment permits possible placements but supplies no identity. Levi and Aaron may lie on unsampled portions of ancestry rather than separate labelled SNP nodes. Father-son links do not each require a distinguishable mutation.

The accepted Ezra 8 paternal reading includes both Eleazar/Phinehas and Ithamar descendants. One younger cluster can represent only part of that older family. It cannot establish that all surviving Aaronic ancestry passes through its younger root, nor does absence of a second identified cluster prove that the other son's line became extinct.

The unknown ancient-world mutation process remains a reason not to date Adam-to-Noah using the ordinary clock. It is not a reason to ignore ordinary-world mutation evidence in the much later priestly branches. No test here establishes compatibility of every detailed biblical chronology or identifies a node for Noah.

## Investigation 3: surviving divisions in the local tree

We independently cut the local tree at ages 1800, 2000 and 2200 in the dump's years-before-present units. The algorithm descends through a node if its point coalescence age exceeds the cut; otherwise it counts the resulting lineage once. These are age-threshold sensitivities, not confidence limits. No conversion of the dump's reference year to an exact calendar date is needed for this diagnostic.

At the 2000 cut, retaining every terminal representation yields J1 37, E 3 and J2 3. Requiring each counted path to lead to at least one nonempty-SNP terminal node gives J1 22, E 2 and J2 2. The latter is a conservative representation rule for this dump, not a biological rule that all empty-SNP terminals are false. Some may represent real basal samples.

| SNP-supported paths | Cut 2200 | Cut 2000 | Cut 1800 |
| --- | ---: | ---: | ---: |
| J-Y3088, carrying ZS222 | 19 | 22 | 22 |
| E-BY8508 | 2 | 2 | 2 |
| J-FGC4975, carrying FGC4992 | 2 | 2 | 4 |

These results substantiate substantial old surviving diversity in the J1 cluster using a separate available representation. They do not reproduce the paper's 60: the paper's tree, dates, marker boundaries and counting procedure differ, and its cumulative nodes/branching presentation needs a precise operational definition. Cumulative split counts and lineages intersecting a time slice must not be assumed identical.

Conditional on the dates and real terminal support, separate paternal lines represented at a past time have descendants represented by the data. No forward survival model is needed to establish that representation. Conditional additionally on an Aaronic root, they are Aaronic lines. But the dump is not an individual register, has no priestly status, and does not establish which sampled men remain alive today. Thus 22 is not an unconditional minimum count of living Noah descendants.

## What has advanced

There is a useful, computable coherence argument for old inherited priestly status, particularly at the principal J1 expansion. The supporting structure is clearer than a frequency comparison. However, the proposed exercise cannot turn continuity of status into identification with Aaron without additional information. We should express the preference for investigating J1 as a qualitative synthesis of its breadth, diversity and reported tradition, not as a posterior ranking produced by these minimum-change scores.

There is also a useful independent check that the J1 cluster contains many surviving old divisions. Its magnitude depends on tree representation, and should be presented conditionally. There is still no defensible worldwide headcount or probability distribution over Noah candidates from these materials alone.

Files: analyse.py, witness-trees.json and results.json contain the status exercise; chronology-survival.json contains the local-tree counts and ancestral paths. The original studies and prior reports remain unchanged.