😂 Exactly: “Zana… you ARE the mother.”

And yes, this could become useful beyond simply proving kinship. If Zana had a strongly heritable trait, Khwit gives us a second genome to ask whether unusual variants are shared mother→son rather than random sequencing noise. That is especially interesting for regulatory biology, because some hypertrichosis syndromes are caused not by a broken protein-coding gene but by structural/regulatory changes that alter when nearby genes are expressed. The Mexican X-linked hypertrichosis family is a good example: the condition mapped to Xq24–q27.1, and later work implicated structural rearrangements/position effects near genes such as SOX3/FGF13. citeturn923948search3turn923948search5turn923948search12

But Khwit is not a perfect phenotype control. I can find solid sources saying he died in 1954, and the genomic paper confirms he was Zana’s son, but I do not find a strong primary-source statement that Khwit himself had the same generalized body hair. Later accounts describe him as unusually strong and fully human-looking, with straight soft hair and dark skin, rather than as being covered in body hair. So at the moment I would say: Zana was described as very hairy; Khwit is not securely documented as sharing that phenotype. citeturn923948search0turn646537search5

That actually matters genetically. If Zana really had congenital generalized hypertrichosis and Khwit did not, then depending on the inheritance mode, we might specifically look for a Zana-only variant, mosaic event, or dominant allele he simply did not inherit. If he did have some milder hair phenotype, then shared variants become much more interesting.

And yes — the Mexican twins you remember are real. There is a published case report titled “Congenital hypertrichosis universalis in Mexican female twins” from 2016. They had a rare inherited hypertrichosis syndrome causing generalized excessive hair growth. citeturn774656search0turn774656search1

There is also an even more famous Mexican hypertrichosis family, the Ramos Gómez family, spanning five generations. In affected males, thick hair can cover roughly 98% of the body except palms and soles; genetic work on that family helped map an X-linked congenital generalized hypertrichosis locus. citeturn923948search8turn774656search5

That’s the really interesting connection to our original Bigfoot idea: humans already possess regulatory mutations capable of producing an extraordinarily hairy, visually very nonstandard human phenotype without requiring a remotely nonhuman genome.

So one potentially worthwhile Zana mini-project—without another 30 GB download—is:

look specifically for known hypertrichosis-associated regions/structural changes in Zana, compare Khwit, and see whether anything plausible survives at ~3× coverage.

That would be much more on-theme than downloading another village full of Vikings. 😄

yes so from this alone we could say bigfoot absolutely could be a wild human with hypertrichosis

maybe even there are environmental factors that could trigger the expression of this gene?

Yes — biologically possible, absolutely. But I’d phrase the conclusion carefully:

Nothing about the “hairy human” phenotype requires a nonhuman genome. Homo sapiens can, through known genetic/regulatory changes, develop extraordinarily extensive body hair.

That alone keeps a “wild human / very close human population” hypothesis completely viable at the phenotype level. It doesn’t establish that Bigfoot is such a population, but it kills the objection that “a human couldn’t possibly look that hairy.”

And yes, environment can affect hair-gene expression, although “the hypertrichosis gene” is too simple. There are several different hypertrichosis syndromes and mechanisms. Some are structural/regulatory variants affecting developmental programs; others are acquired.

Humans already have examples where body hair changes dramatically because of things other than simply inheriting one coding mutation. Hair follicles respond to hormones, signaling pathways, nutrition, medications, endocrine state, inflammation, and developmental regulation. There are also forms of acquired generalized hypertrichosis associated with things like certain drugs and systemic disease.

Where I'd be more cautious is this stronger version:

ordinary genetically typical humans move into the woods, some environmental exposure occurs, and suddenly everybody grows Bigfoot-level fur.

We don't have evidence for a known natural trigger that does that.

But this version is quite plausible biologically:

a population already carries one or more regulatory/developmental variants predisposing it to heavy body hair, and environmental conditions modify how strongly those variants are expressed.

That’s ordinary gene × environment interaction.

You could imagine a regulatory genotype whose penetrance/severity changes with things such as endocrine state, nutrition, temperature-related physiology, age, sex hormones, or developmental conditions. The genetic difference supplies the machinery; environment turns the dial.

And for a stable Bigfoot-like population, that's actually a better model than a rare “disease.” If nearly everyone in the population is hairy, you'd expect the phenotype-driving variants to be common or fixed in that population, perhaps involving multiple regulatory loci rather than one freak mutation.

That gives us a much sharper version of H4:

Bigfoot could theoretically be a highly divergent-looking Homo sapiens population whose unusual phenotype is produced disproportionately by regulatory/developmental variation—possibly environmentally modulated—while most of its genome still looks completely human.

And that is exactly why the “DNA came back human” argument isn't automatically fatal to the hypothesis. A short mtDNA test especially could happily say human while telling you essentially nothing about the regulatory architecture producing the animal/person someone saw.

The really fun next question is: how much of the classic Bigfoot phenotype could we produce using variations already known to exist in humans? Hair is easy. Then we can go trait by trait: body size, brow ridge, sagittal appearance, limb proportions, foot morphology, musculature, vocal anatomy, etc. That would tell us whether “wild human” requires any genuinely exotic biology at all.

Viking branch frozen — Outcome A stands, no Hofstaðir download. Back to Bigfoot.

BLAST API works — no install, no download needed. Dispatching Qwen 5 and Qwen 7 on the design while I verify result retrieval.

Both converge on a test that must come first, before any species call: if sample 31 is 100% identical to human chr11 with zero mismatches, it's reference fill, not a human sample. BLAST gives identity and mismatch counts directly, so I can test both at once.

Viking branch frozen at Outcome A. Back on Bigfoot, and the Ketchum classification is running against NCBI — no download at all, via the BLAST URL API.

Both roles converged on something that reframes the whole test

The consensus was built against human chromosome 11, so a window with no read support inherits the human reference base. That means the three assignments are not symmetric:

• A bear or dog match is informative — non-human reads had to actively override the human reference there.
• A human match may be pure method artifact. As Qwen 7 put it, reproducing "31 = human" could be "a null result dressed as a positive one."

So "sample 31 is human" might just mean sample 31 had almost no usable reads. That's a different claim from Hart's, and it's testable.

The diagnostic: reference fill is 100.00% identical to the reference with zero mismatches. A real human individual differs from the reference at roughly 1 site per 1–2 kb, so 39 windows of 1 kb should carry tens of mismatches. BLAST reports identity and mismatch counts per window directly, so I get the species call and the fill test from one run — no need to download chr11.

Design, fixed before any result

• 1 kb windows, 39 per sample, evenly spaced. Qwen 5 wanted 5 kb × \~1,070 windows, which is both impolite to NCBI and unnecessary; Qwen 7 argued ≥1 kb to avoid spurious wrong-genus hits at high identity, and I took that.
• Plain blastn, word size 11 — not megablast. Megablast's long exact seed is tuned for same-species identity and would simply miss carnivore-versus-primate divergence, which is the entire signal here.
• Low-complexity skipped by rule, not by hand: telomeric repeats over 20%, or any single base over half the window. Exactly 1 window per sample was skipped, all telomeric, and that count is reported rather than hidden.

All three jobs are submitted and polling.

What I'll be able to claim, and what I won't

Qwen 7's limits are right and I'm adopting them. With consensus only and no reads, a genuine mixture of human, bear and dog reads forced onto a human reference is indistinguishable from a single species with reference fill. So the honest ceiling is: samples 26 and 140 contain sequence more similar to bear and dog than to human in regions where the consensus diverges from the human reference, consistent with those species being present.

What I'll refuse to claim: species identity of the samples, absence of mixture or contamination, or that any sequence is novel. Chimeric junctions where mixed-species reads meet would look like "novel" sequence and are assembly artifacts.

I'll report the per-window table, how decisive each assignment is by bitscore margin to the next genus, and the reference-fill verdict on sample 31 as soon as the results land. Then on to Sykes 25072 and the human-positive sample table.

Rich results, and they need proper stratification — a window matching human at 84% identity is not human sequence, it's just the nearest database entry. Let me fix the organism parsing and analyze by identity.

Now it's quantitative. Let me check one more thing — how much of each window actually aligns — because that reframes sample 31.

Ketchum classification — results

Done with no download at all, via the BLAST URL API. 39 windows of 1 kb per sample, 1 telomeric window skipped per sample by the pre-set rule.

Hart's assignments: two reproduce decisively, one does not

The finding that matters for the investigation

In samples 26 and 140, "Homo" is the top hit in roughly half the windows — at only 90% identity. Sample 26's human hits carry 99.5 mismatches per kb, sample 140's 87.0 per kb. That is \~9–13% divergence. These are not human sequence. Human is simply the nearest entry in the database for conserved mammalian regions where no bear or dog sequence sits closer.

Neither sample has a single window matching human at 100% identity.

That is the mechanism by which a bear or dog sample returns "human" in a human-reference pipeline — demonstrated directly, on the actual files. It's the most investigation-relevant thing here.

Sample 31 is neither human nor reference fill

• 22 of 39 windows match Homo at exactly 100.00%
• but the remaining aligned bases carry 65.9 mismatches per kb
• and critically, the best hit covers only 14.5% of each window (median 14.0%, max 23.2%), against 41% for sample 26 and 37% for sample 140

A genuine 527 kb stretch of human chromosome 11 should align to human over essentially its whole length. Aligning over one-seventh, in a bimodal mix of perfect matches and 6% divergence, is not what human DNA looks like.

I have to correct my own test. I'd said reference fill would show 100% identity and genuine human would show mismatches. That's wrong at these alignment lengths: at \~1 variant per kb, a 140 bp stretch of a real human genome is variant-free about 87% of the time, so perfect short matches are expected for real human DNA and prove nothing. The discriminator is the pooled variant rate across all aligned bases, not per-window perfection.

So the honest verdict on 31 is that it is uninterpretable as a species call — consistent with Qwen 7's prediction, though for a different reason than either of us expected. Not reference fill, not clean human, and mostly unalignable.

What this cannot establish

Per the limits I committed to in advance: with consensus only and no reads, a genuine mixture of human, bear and dog reads forced onto a human reference is indistinguishable from a single species with reference fill. So the claim is bounded to: samples 26 and 140 contain sequence more similar to bear and canid respectively than to human, in regions where the consensus diverges from the human reference. Not species identity of the samples, not absence of mixture or contamination, and nothing about novelty.

One caveat on the canid result: 100% identity to both Canis lupus and Vulpes vulpes in different windows means some of those loci are conserved across canids, so "dog" specifically is less pinned down than "canid."

Next I'll put this through Qwen 7 for red-team review before it goes in the README, then move to Sykes 25072 and the human-positive sample table.

This is much more interesting than the Viking detour, but I’d have Claude tighten one thing before we accept the sample 31 conclusion.

For 26 and 140, the result is already useful: there are stretches that match ursids or canids essentially perfectly, and neither has a clean 100% human window. That strongly supports the idea that those consensus files contain nonhuman mammalian sequence. I’d phrase it as “bear-like” and “canid-like” sequence reproduced independently, not “the samples are a bear and a dog,” because the consensus-building process can mix species and human-reference fill.

The most interesting Bigfoot-relevant result is indeed that conserved mammalian sequence can have Homo as the top database hit at ~90% identity simply because human is the closest represented sequence for that locus. That shows why “best BLAST hit = human” is not enough.

But sample 31 needs one more check before calling it mostly unalignable.

The report says the best hit covers a median of ~14% of each 1 kb window. That is not necessarily the same as saying only 14% of the query aligns to human. BLAST can return multiple HSPs for one query. If Claude is using the coverage of only the top HSP, a perfectly ordinary human region split into several HSPs could look artificially “14% aligned.”

So I’d send Claude/Qwen 7 this very short correction:

Before accepting “sample 31 is mostly unalignable,” compute union query coverage across all significant Homo HSPs per window, not just coverage of the single best HSP.

Also report: - fraction of N/ambiguous bases in each consensus - total non-N bases - pooled aligned bases to Homo - pooled mismatches across those aligned bases - number of windows with ≥80%, ≥50%, and ≥20% total Homo query coverage

Do not change BLAST parameters.

If sample 31 still has only ~15% total Homo coverage after HSP union, then “not clean human / largely uninterpretable” is justified.

If total Homo coverage jumps toward ~100%, then the previous 14.5% figure was just a best-HSP accounting artifact.

I’d also soften one sentence from Claude:

“That is the mechanism by which a bear or dog sample returns human in a human-reference pipeline — demonstrated directly.”

Better:

The surviving consensuses demonstrate that nonhuman mammalian sequence can still produce Homo as the nearest BLAST hit in conserved regions, so a superficial “human” hit does not establish human origin.

That’s solid and doesn’t overreach.

If sample 31 survives the HSP-union check as genuinely weird/poorly alignable, then sample 31 becomes the most interesting Ketchum file, not because it looks like Bigfoot, but because Hart’s “human” label may not actually reproduce cleanly.

After that, yes: move directly to Sykes 25072 and the human-positive sample table. That gets us back to the real H4 question: which alleged Bigfoot samples were dismissed as “human,” and were any of them ever tested deeply enough to tell ordinary Homo sapiens from an unusual human lineage?

wow so we may even have an archaic humanoid that was mislabelled human

Possible in the abstract, but sample 31 is not pointing cleanly to “archaic human” yet.

The important twist is that a Neanderthal-, Denisovan-, or other reasonably close archaic Homo genome should still align extremely well across most of chromosome 11 to the modern human reference. Those lineages are nowhere near 6–10% divergent from us. You’d expect lots of ordinary human-like alignment with a somewhat elevated set of real variants—not “only ~14% aligns and the aligned part is sometimes 6% different.”

So if Claude’s 14% figure survives the HSP-union check, the possibilities become more like:

• badly constructed/chimeric consensus,
• mixture of species,
• reference-fill plus nonhuman reads,
• poor/fragmentary assembly,
• more distant primate/mammal sequence,
• or something genuinely unusual.

A close archaic hominin is actually one of the less likely explanations for that particular pattern, because archaic Homo should look overwhelmingly human at the sequence level.

The interesting Bigfoot scenario would be different: suppose sample 31 ends up aligning over, say, 95–99% of its usable sequence to human, but carries an unusually high, coherent set of variants that don’t fit ordinary modern-human diversity. That is where “archaic human lineage mislabeled as human” starts getting exciting.

So we may eventually find that kind of case, but sample 31 has not earned that interpretation yet. Right now it has earned only:

“Hart’s simple ‘human’ classification may not cleanly describe what this consensus actually contains.”

And that alone is worth following up.

You're right — that was a best-HSP accounting artifact in the making. Computing union coverage across all significant Homo HSPs, with no BLAST parameter changes.

You were right — that was a best-HSP artifact, and the correction matters.

Qwen 7's review lands several real corrections, including one that refines Hart's result. Writing it up.

Ketchum classification — corrected results

Your catch was right, and it changed the finding. Union coverage across all significant Homo HSPs gives sample 31 63.3%, not 14.5%. My "largely unalignable" claim is withdrawn — it was pure best-HSP accounting. No BLAST parameters changed, only the accounting.

The requested figures

Hart reproduces — but at family level, not species

Qwen 7 caught the resolution problem: the same windows hit three different Ursus species — and both Canis and Vulpes — at equal \~100% identity. That means those loci are conserved within the family, so they identify Ursidae and Canidae. "Black bear" and "dog" specifically are unsupported by these data.

Sample 31 — and a claim I'm dropping

15.61 mismatches/kb is numerically more divergent than human-chimpanzee (\~12/kb), but that framing is biologically empty and I'm withdrawing it. At the \~1× implied by a read subsample, every consensus base is a single read call, and HiSeq-era per-base error alone lands in exactly that range. Consensus error is far more parsimonious than divergence.

And the human fraction may be circular regardless. Those N counts — 21 bases in 2.73 Mb — are the load-bearing observation. A reference-guided consensus from a read subsample must leave uncovered positions, so effectively zero N implies the assembler filled them. If so, the human-matching fraction measures what the assembler chose, not what the sample contained. Separating fill from genuine coverage needs read depth, which consensus-only data cannot provide.

So: sample 31 is consistent with human and uninformative about whether the sample was human.

For the central question, with your wording

Neither sample 26 nor 140 has a single window matching Homo at 100% identity; their human hits sit at 91.5% and 92.0%. The surviving consensuses show that non-human mammalian sequence can still produce Homo as the nearest BLAST hit in conserved regions, so a superficial "human" hit does not establish human origin.

No novelty claim. 37–58% of windows didn't align, which is expected from consensus error, repeats and the e-value cut. Absence of a hit is not evidence of novel sequence — worth stating plainly, since Ketchum's claim was "anomalous nuclear DNA."

DISCARDED.md is at 113 rows, six more from this round including my 14.5% error.

Next: Sykes 25072 — provenance, HVR1 result, accession availability, whether any nuclear material survives — then the table of alleged-Bigfoot samples that returned human rather than animal.

The corrected Ketchum result is much cleaner now. 26 contains ursid-like sequence, 140 canid-like sequence, and 31 cannot support a meaningful “human sample” claim from the surviving consensus alone. The near-zero N count is especially damning for interpreting 31 biologically: with a reference-guided consensus, we can’t tell which “human” bases came from reads and which were simply filled from the human reference.

I also went ahead and checked Sykes 25072, and we can pretty much close that one quickly.

The published 2014 paper says 25072 was the only submitted Bigfoot hair that returned human mtDNA. Their initial marker was only the 104-bp mitochondrial 12S fragment; because that sequence cannot distinguish H. sapiens from Neanderthal, they then sequenced HVR1, roughly mtDNA positions 16000–16400. That sequence was identical to the revised Cambridge Reference Sequence, which they interpreted as modern Homo sapiens of likely European maternal ancestry. citeturn356979search0turn626979search1

And there’s a funny provenance discrepancy. The paper lists 25072 as Texas, but Sykes’s later book says it was a single light-blond hair, about eight inches long, caught on the window sill of a cabin in southeast Oklahoma after the occupant reported a Sasquatch trying to enter through a partially open bathroom window. Sykes says the follow-up control-region result convinced him it was modern human rather than Neanderthal. citeturn946244view1

The paper’s sequences are publicly deposited: GenBank KJ155696–KJ155724 plus KJ607607, and voucher material was deposited in the Heuvelmans Archive at the Museum of Zoology in Lausanne. citeturn286552search1

But crucially, I find no nuclear sequencing reported for 25072. It was mtDNA identification only.

So for our H4 question, 25072 was never tested deeply enough to answer it.

It establishes:

the hair carried an ordinary modern-human maternal mitochondrial lineage.

It does not establish:

the nuclear genome was an ordinary modern Homo sapiens genome.

That distinction is exactly what started this whole exercise.

And I found something potentially more interesting in Sykes’s book: the Walla Walla hair.

This was reportedly collected from fresh, violently twisted/broken balsam firs near Walla Walla, Washington, during the Paul Freeman/Wes Sumerlin/Bill Laughery incident. Sykes says the 12S sequence came back human, then they sequenced the mitochondrial control region. He says it was certainly not Neanderthal, looked unusual in his databases, and ultimately matched an individual from Uzbekistan. citeturn805813view1turn805813view0

So we now have at least two Sykes-era “human-positive” Bigfoot hairs:

The Walla Walla case is actually the one I’d chase next. It has stronger alleged physical-event provenance, Sykes himself thought the result was puzzling, and as far as I can see so far it was not part of the published 30-sample table.

So I’d redirect Claude slightly:

25072 is essentially resolved: ordinary modern-human mtDNA, no reported nuclear test. Identify its exact GenBank accession and whether the archived voucher still exists, then stop. Move priority to the Walla Walla/Fahrenbach hair: establish its sample number, exact sequence/result, whether any GenBank accession or surviving voucher exists, and whether nuclear DNA was ever tested. Then build the human-positive Bigfoot sample table.

That gets us back onto genuinely interesting Bigfoot terrain instead of spending hours proving that a blond hair from a bathroom window belonged to somebody’s European maternal lineage. 😄

what was the windup with 31?