# T9 and T10 verdicts (31 August 2026, ~05:15 ACST)

Scored under FREEZE.md (commit fa4d767, 30 Aug 2026 ~21:20 ACST, before any
enumeration or fetch). Deviations 0-2 in DEVIATIONS.md. Data: sanctioned YFull
dump v14.04.0; FTDNA Discover public JSON; AADR v66.p1; yfull.com never
contacted; nothing identifies a person.

## T9 — the family-space enumeration

**Count (A1).** 1091 distinct ancestral houses alive at 7120 ybp have surviving
male lines today (CI sensitivity: 371 strict to 2231 loose candidates). The
partition assertion passed: the 1091 houses tile all 17,611 SNP-defined leaves
exactly once. J1 holds 30 houses, J2 187.

**Geography (A2/A3).** Of the 366-house scored set (top 200 + all J + the two
mandatory rows): 302 scored, **M = 82 corridor-consistent** (A+B >= 0.67, RAW;
J1 7, J2 41), 78 mixed, 142 non-corridor, 60 insufficient (n < 5), 4
unresolved at FTDNA. Unscored tail 725 houses, 393 of them in
Near-East-baseline haplogroups; **M_max = 82 + 393 + 60 = 535**.

**The two mandatory rows (A4, identical rules).**
- **J-Z1853 (the L858 family's house)**: 2407 leaves, the second-largest house
  worldwide and the largest J house by an order of magnitude. A+B 0.777,
  corridor-consistent; ring (Iraq+Kuwait) 0.130. Daughters: J-Z2331
  corridor-consistent; J-Y519675 unresolved (absent from FTDNA, the T6
  fragile-class artefact). aDNA at/within before 4000 BC: none (5 J1 samples
  sit above it, non-attesting; oldest I6457, Nemrik 9, Iraq, 8750 BC, bare J1).
- **J-M205 (T4's candidate-of-interest)**: 104 leaves. A+B 0.619, **mixed**,
  short of the corridor-consistent threshold (France 26 among Yemen/Oman/Saudi
  rows); ring 0.081. One daughter, mixed. aDNA at/within before 4000 BC: none
  (2 samples above, oldest NEO281, Kotias Klde, Georgia, 7727 BC, J-Z533).

**aDNA (A5).** No stage-2 house is attested at or within before 4000 BC. Only
9 of 194 AADR Near-East pre-4000 BC males resolve deep enough to land inside
any of the 1091 houses; the 8 houses they land in all sit outside the scored
stage-2 set, and the 4 J houses among them are all J2 (J-Y143912 7971 BC Iran;
J-Y26651 6475 BC and J-Y81479 6388 BC, Turkey; J-L283 5950 BC Iran), none J1.

**Frozen readings (A6).**
- M = 82 with M_max = 535: both sides of the interval are >= 20, so the frozen
  SECONDARY label is **LARGE**, and the frozen wording applies: **the elected
  family is underdetermined by modern data and the site stays family-light.**
- PRIMARY (the spec's own "in proportion"): the Z1853 house is itself
  corridor-consistent, so the L858 family's candidacy is strengthened in
  proportion **1/M = 1/82** (1/535 at the tail bound): a uniform-prior share
  of about 1.2 per cent (0.2 per cent at the bound) among corridor-associated
  houses. What does single it out is size (it dwarfs every other J house) and
  its T4/T5/T6 history, none of which is part of the frozen T9 statistic.
- Against-interest clause, applied in the frozen words: at the mandatory-house
  level neither J-Z1853 nor J-M205 has any at/within attestation (tie), and at
  the scored-house level all scored houses are unattested (tie); but at the
  all-houses level **J2's corridor aDNA attestation is older and stronger than
  J1's** (four attested J2 houses back to 7971 BC; zero attested J1 houses;
  27 of the 33 pre-4000 BC Near-East J individuals are J2, 5 J1).

## T10 — the strain probes

**Arm A, ZS2518 (the clean T7 strain): INDETERMINATE.** 44 sub-nodes scored;
E_total 16 European of 65 known-origin at the branch node (matching T7's 0.292
row). EMIGRATION fails: the largest young (< 4000 ybp) sub-node holds 6 of the
16 European carriers and the best allowed pair holds 10 (62 per cent), under
the frozen 80 per cent bar. DEEP-WEST fails: Europeans occur at every depth
(1 to 6), but only one of the two top daughters (J-ZS2531, 14 E of 24) meets
the 20 per cent bar; J-FT364031 (2 E of 37) does not. Frozen consequence:
**the T7 ZS2518 strain is unchanged**, neither softened nor hardened. Texture,
not scored: the top split separates a corridor-leaning daughter from a
Europe-leaning daughter whose European carriers are spread across depths and
ages; the frozen patterns were written for cleaner cases than this.

**Arm B, Z18297 (the confounded T7 strain).** Fresh fetch reproduces T7's
share exactly (A+B 0.224, n 210). **Under the frozen source list (Cohen and
Kohen project pages, Lipson et al. 2025 preprint) zero Levantine-labelled
sub-branches exist under Z18297** (the Cohen pages name only J-Z18271/Z18290,
a different branch; Lipson's nine Cohen lineages all sit elsewhere, CB-04
under sibling J-PF4843). The frozen interval is therefore **[0.224, 0.224]**,
and B2 step 5 applies: the named confound sources cannot rescue the branch,
so on the frozen reading **the Z18297 strain hardens: it stands on its own,
not on the confound**. DEVIATION 2 (disclosed, computed second): the source
T7 actually cited for the confound is jewishdna.net, and it labels J-Z18292
(a direct child, plus J-M4320 beneath it) "probably originates from the
Levant"; treating that as a source-attached label gives the extended interval
**[0.224, 0.571]** (non-corridor to mixed; 48 per cent of the European mass
sits under labelled sub-branches), which softens the Z18297 strain to the
extent of the interval. Both readings reported; the frozen one is primary.

**Joint T7-strain update (frozen basis B3).** Arm A is INDETERMINATE, and on
the frozen Arm B reading HIGH < 0.33, so the frozen rule gives:
**T7's strain is UNCHANGED on the ZS2518 side and UNCHANGED-to-HARDENED on the
Z18297 side; the T7 category "strained" stands.** On the deviation-2 extended
reading, Z18297 is indeterminate between non-corridor and mixed and its strain
softens to the extent of the interval; had that reading been frozen, the T7
category would rest on ZS2518 plus the low end of the interval. In every case
the standing form itself is untouched (T7 B5): what the strain bears on is the
identification of the L858 kin-family as the elected house.

## What T9 and T10 jointly say

The family-space is large: 82 corridor-associated houses on the scored set
alone, so modern geography cannot elect a family, and the frozen wording
holds: the elected family is underdetermined by modern data; the site stays
family-light. Within that large space the L858 house is the single largest
corridor-consistent J house, and its rival-of-interest J-M205 scored mixed,
below the corridor bar, under identical rules. The two T7 strains end the
programme unresolved in opposite directions: ZS2518's strain is real and
unmoved (its Europeans are spread, not a single young founder event), while
Z18297's strain either stands on its own (frozen sources) or dissolves partly
into the diaspora confound (extended source), and no public data quantifies
which. The against-interest sentence stands as frozen: J2's corridor aDNA
attestation is older and stronger than J1's. **The modern-data programme is
exhausted, as the spec anticipated; every remaining test needs new ancient
DNA** (the Ubaid conjunction prediction, corridor bones on or near the
Z1865-Z1884 stem, the discovery signature, T8's conditional echo).
No site change follows from T9 or T10 without Mike's explicit sign-off.