# Prior observations and disclosed knowledge

**Frozen:** 2026-08-10, before extraction or inspection of additional candidate outcomes.

This is not a pristine preregistration. The following observations and model choices were known before the new audit and cannot be treated as held-out confirmation.

## Required disclosures from the study brief

1. AADR v66.p1 has substantial 1240K coverage in Armenia and parts of Anatolia and Iran after the proposed Babel period, but very sparse sampling in Iraq/Shinar and southwest Iran.
2. Published ancIBD calls show Bronze/Iron Age Armenian cross-site family networks. Examples seen so far appeared compatible with ordinary regional kinship.
3. Tatika has close relatives around the proposed 2900 BCE arrival period. They are local baselines, not probable descendants.
4. An automated screen made Iranian samples BOE001/BOE002 appear close relatives separated by roughly 600 years. Source checking showed BOE002 lacked the supposed precise direct date and had low coverage. The anomaly was compiled-metadata artefact, not evidence.
5. A Shamanka II report described an approximately 350-radiocarbon-year difference between teeth and postcranial remains assigned to one person. Further work showed the mandible belonged to another person.
6. No verified positive kinship–chronology or multi-tissue candidate has been found.

## Earlier Cudi/Başur–Tatika work

- The local public Tatika transect is very small: three strict secure-pre, two QC-relaxed secure-pre, two usable secure-post and two boundary individuals.
- I4478/I4481 is a first-degree-candidate pair from a single READ-style mismatch estimator. I4478 is boundary dated and I4481 secure post. Segment-level IBD was not validated; this is a local-baseline candidate, not a chronology anomaly.
- Four Başur capture datasets were processed. SK36/SB726 failed authenticity; SK8/SB714 fell below the 50,000-SNP primary threshold; SK1096/SB710 and SK1080/SB731 had ordinary passing regional proxy models.
- A Tatika pre/post qpWave discontinuity at approximately p=0.045–0.048 failed the frozen p<0.01 rule and changed with sample definition.
- No Başur/Tatika individual passed the prior seven-part anomaly rule. The pilot had little power for a tiny family.

## Earlier maturation screens

- Five regional ancient shotgun genomes (I1631, I1632, I1633, I5742, I5743) and 23 out-of-region comparisons were screened in three prospectively controlled runs.
- A 1240K candidate-gene approach was shown structurally unsuitable; subsequent shotgun coding/regulatory screens were null.
- Run 1's 23 apparent target hits and 39 control analogues were post-mortem-damage artefacts; ESR1 R256* in I1633 was confirmed damage.
- Run 2's apparent 5.03-fold enrichment disappeared after haplotype collapsing and matched controls (target 1.30 versus control 1.22 findings/100 kb, p=0.65). The TACR3 block was an inherited population haplotype, not 43 observations.
- Run 3 found 26 target versus 19 control clusters, with controls having the higher rate; no target stop-gain, canonical splice, frameshift or ClinVar pathogenic/likely-pathogenic result survived.
- No specimen-level maturation variant or mechanism was validated. Modern polygenic/candidate-gene searching is outside the new audit.

## Conditional Genesis pedigree model already seen

- Website/model chronology: Cudi arrival approximately 2900 BCE; Shinar gathering approximately 2800; dispersal onset 2370; Peleg/Babel completion approximately 2033 BCE; route arrivals 2250–2033 BCE.
- The pre-Nahor boundary had already been noticed as the best numerical shock boundary. This is prior knowledge.
- LXX-derived fathering ages used in the model range from 79 to 135 years; the Shem–Serug cluster is approximately 130–135 years. Lifespans decline through the post-flood pedigree.
- A one-generation 50% retention pulse fit the Nahor/Terah held-out numbers better than uninterrupted retention in the conditional numerical model, but the effective-factor parameter was not identifiable. The result is internal numerical compatibility, not evidence for historical people or a biological mechanism.

## Methodological literature known before candidate extraction

- Sedig et al. (2021), DOI 10.1016/j.jas.2021.105452, reports extreme theoretical maximum death-date separations of 100 years for parent–offspring, 135 for siblings (despite a 130-year worked example), 180 for grandparent–grandchild, 195 for first cousins and 210 for aunt/uncle–niece/nephew.
- Sedig et al.'s historical/genealogical means are much smaller (approximately 26–35 years depending on topology); its ancient dataset contained known radiocarbon outliers resolved by redating.
- Ringbauer et al. (2024), DOI 10.1038/s41588-023-01582-w, provides public ancIBD segment calls and detects recent relatedness through approximately sixth degree, but exact topology becomes uncertain beyond third degree.

## Boundary statement

Before the freeze, no new AADR family-relation rows, ancIBD pair/segment rows, primary-paper candidate tables, candidate-specific radiocarbon distributions, or conventional-vocabulary search results were inspected. File headers, counts, schema and source inventories were inspected without pair outcomes.