Testing the Timeline
What this site claims, what can be checked, and what happened when I checked it
Research index · Updated 21 August 2026
Most of this site makes its case from the Bible and other ancient writings. A few of its claims, though, reach into the physical world, where they can be checked against hard evidence: ancient bones and DNA, and even the numbers in Genesis itself. This page is the full record: what I set out to test, what the evidence could actually show, and how each test turned out.
The results are mixed, and I have not tidied them up. One test read ancient DNA from around Cudi Dağı (the peak in south-eastern Turkey that many identify as the mountain where Noah’s ark came to rest), with its method fixed in writing before I saw any data; it found no sign of a new population arriving, though the few surviving samples were too small to have caught a single family anyway. Another combed the public DNA records for close relatives who lived too many centuries apart for normal lifespans to explain, and found no clear case (one Armenian family is worth dating properly). A third test measured terrain rather than DNA: with its scoring rule fixed in writing first, it graded the ground around all 184 catalogued language isolates, and found both that about a fifth have no shelter at all and that none has less than Sumerian. Two further checks hit the same wall: the evidence they would need has not been dug up yet.
One test came out more interesting than the rest. The idea it examines is this: after the flood, a small line of people went on living extraordinarily long lives, many centuries each, while the far larger population around them lived ordinary spans. If someone from that long-lived line married an outsider, their children’s lifespans should fall back towards the ordinary. And the recorded ages do fall away steadily after the flood, then drop sharply between Serug and Nahor. That sudden dip fits what you would expect if Nahor’s mother came from outside the long-lived line, better than it fits the earlier decline simply carrying on. It is a genuine “maybe”, no more: the telling number was already in view before the test was designed, and a writer shaping the text could have produced the same dip on purpose. So it counts as consistent with the idea, not proof of it, and certainly not DNA evidence that Babel happened. I have put every outcome here, the flat ones included, because you have no reason to trust the rest of this site if I quietly drop the tests that did not land.
What none of these tests can prove
No test on this page can prove that Noah, Cain, or Babel existed, or that any named event happened. Ancient DNA can show that the people in a region changed, or that two buried skeletons were relatives. A study of the Genesis numbers can show that one explanation fits them better than another. But none of them can hand you a name, a language, a motive, or a story of cause and effect: that always takes evidence from outside. Every study below runs into this same ceiling.
The studies
| Question | Status | Where it stands |
|---|---|---|
| Is Sumerian’s open, easily reached homeland unusual among language isolates? | Tested · outlier confirmed | A language isolate is a language with no known relatives; Basque is one, and so is Sumerian, from what is now southern Iraq. Isolates usually survive in hard-to-reach country, mountains especially, because expanding neighbours tend to overrun the easy ground and leave the rough ground alone. This study measured how hard to reach each of the 184 catalogued isolates’ homelands is, using a rule fixed in writing beforehand. Sumerian came last: its homeland is low, flat, open plain, with nothing over 200 metres within a hundred kilometres. The rank survives every variation of the scoring, though neighbouring Elamite is effectively tied for last. And about a fifth of the 184 sit in similarly open country; Sumerian’s homeland is simply the most open of all. The measure reads terrain, not history, and dates nothing: a language that arrived from elsewhere and one whose relatives were overrun on the plain would score the same. |
| Did a distinctive population appear near Cudi Dağı (where the earliest traditions place the ark's landing) around 2900 BC? | Tested · null | No sign of newcomers showed up in the DNA. But the surviving sample was too small to catch a group that made up only a tenth of the population, so the result argues against a large influx while saying little about a single family. |
| Did marriage outside the family after Babel shorten the patriarchs’ lives? | Exploratory · consistent, not proof | The patriarchs’ recorded ages fall steadily after the flood; carried straight forward, Nahor’s should land near 300. But this site’s idea is that a few people after the flood lived extraordinarily long lives, and that a child of one of them and an ordinary outsider would fall back toward a normal lifespan, which points to about 205. Genesis records 208. With so few figures to go on, the match cannot prove a genetic cause, or that any particular event happened. |
| Have ancient-DNA studies found close relatives who lived hundreds of years apart? | Tested · no secure example found | DNA can identify close relatives among excavated remains, while radiocarbon can estimate when they died. Nine high-quality, directly dated pairs could be compared; none was separated by more time than ordinary human lives and generations can explain. One Armenian family group is worth dating directly. |
| Did a distinctive sheep population arrive with Noah’s family near Cudi Dağı around 2900 BC? | Not testable | No directly dated local sheep genomes bracket 2900 BC. That is a statement about missing data, not about the claim. |
| Did a distinctive grapevine lineage appear? | Not testable | Ancient grapevine nuclear genomes barely exist anywhere, and none near Cudi Dağı in the relevant interval. |
| Did a distinctive population appear in the Zagros mountains (the region this site links to Cain’s line) around 5000 BC? | Underpowered | Individuals from about 4840 BC onward could carry the ancestry of an earlier arrival, and a few have been sequenced. But the nearest baseline before the proposed date is a single individual some eight hundred years older, so ordinary demographic change remains a rival explanation for anything found. |
Why some results are stronger than others
The middle column in the table above labels each study (tested, exploratory, not testable), and those labels are not all equally strong. What separates a strong result from a weak one comes down to timing: whether the rules of a test were fixed in writing before I saw the data, or whether I already knew what the numbers showed. A test whose rules were set in advance is much harder to fool, even by accident.
The Cudi Dağı analysis was locked down in advance: the question, the set of populations it would be compared against, the quality bar for the data, and the rule for what would count as a genuine anomaly were all fixed in writing before the new genome-wide results were examined. That constraint is uncomfortable by design. It removes the freedom to spot a pattern after the fact and then explain why it was the pattern I had been looking for all along.
The isolate comparison was locked down the same way, and further: the scoring rule, the comparison class, and what would count as an outlier were frozen and committed before any elevation data was retrieved, and Sumerian and Elamite were scored through exactly the same code as the other 182, masked until the end. Its numbers come from Glottolog 5.0 and the ETOPO 2022 60-arc-second relief grid, with terrain measured within 50 and 100 kilometres of each language’s catalogue point; that is one way of measuring refuge, not the only one. Both halves of its result are reported on the isolate pages: the extremity of Sumerian, and the fifth of the class that shares its open ground; the second is what keeps the argument within its evidence. Unlike the DNA studies on this page, it tests a corollary about language rather than the chronology itself; nothing in it dates anything.
The Genesis pedigree analysis has a different status. Its formal model was written down before the final scores were worked out, but the lifespan figures themselves (and the fact that the sharpest fall by far, almost half, comes between Serug and Nahor) were already known. Excluding Nahor and Terah from the fitting makes the comparison more demanding than fitting all ten figures together, but it does not turn the result into a fresh prediction of unseen data. That is why it is labelled exploratory.
The ancient-relatives audit fixed its time periods, regions, quality tiers and pass/fail rule in advance, before any new pair-by-pair results were opened. A few relevant examples had already turned up in earlier work, and I said so up front. So it sits in between: stronger than a hunch checked after the fact, weaker than a clean test of brand-new data. Its high-quality comparison came back empty.
It also has a limitation worth stating plainly. A test with its rules fixed in advance can only fail in the ways it was set up to fail. The Cudi Dağı pilot could not have detected a small family’s contribution no matter how real, because the surviving sample was too small, and I knew that before running it. Publishing the sum that shows how faint a signal the test could still have caught, alongside the empty result, is necessary for accurate reporting: the finding is not this did not happen but this design could not have seen it.
What would change the picture
For each open question, the requirement is new material rather than better analysis of the same few genomes.
- Cudi Dağı. Roughly sixteen good-quality unrelated individuals on each side of 2900 BC, from the same local sites, tightly dated. If the incoming population were only modestly different from those already present, even hundreds per side might not settle it.
- The Zagros. Any sequenced individual at all from western Iran between about 5600 and 4800 BC. The gap is a gap in sampling, not in the ground: excavated Chalcolithic (Copper Age) material from these sites exists in museum collections and has never been sequenced.
- Sheep and grapevine. Directly dated local specimens, screened for recoverable nuclear DNA, from contexts on both sides of the proposed date.
- Ancient relatives. Direct radiocarbon dates for the DNA-tested people in the four-person Armenian family group, together with renewed checks of their skeletal identities, burial contexts and exact family relationships.
- The genomic question. Shotgun sequencing rather than the targeted kind, that is, reading the whole genome even lightly (roughly half a fold of coverage or better), rather than just a fixed checklist of spots, from any individual at either site. Almost all ancient genomes are produced by that targeted method, which reads a set list of positions very well and the rest of the genome not at all: the wrong shape of data for this question, however good it is for others.
How new evidence gets noticed
Ancient DNA is a fast-moving field, and a claim tested against the data of one year may be testable against the data of the next. So I had Claude Opus 5 write a small program, which runs every week on a machine at home and searches the literature and the main public genotype archive for anything bearing on the questions above: new human genomes from the upper Tigris or western Iran, directly dated ancient sheep or goat genomes, or ancient grapevine nuclear data. It emails me when it finds something, and stays quiet otherwise. How this site was made sets out where else I have used AI, and where I have not.
It has so far caught two papers worth assessing, and neither changed anything. Seventeen new Mesopotamian genomes published in June 2026 come from Bakr Awa, some 250 to 300 kilometres south-east of Cudi Dağı in the wrong river system; they add comparison data but do not touch the sample-size problem, and their finding of ordinary Caucasus-related ancestry flow into Bronze Age Mesopotamia makes any future signal slightly harder to attribute. A 2026 grapevine study from France demonstrates that ancient grape pips (seeds) can yield usable DNA at the depth required, which is useful method but no help with geography.
I had it widened on 5 August 2026, after discovering that a 2025 paper adding an individual inside the Zagros window had gone unnoticed for a year: its title said “Northern Iranian Plateau”, and the only Iran-related term the program had been given was “Zagros”. A monitoring tool is only as good as the words behind it, and these were too narrow.
What a negative result is worth
It would be easy to present each of these outcomes as a delay: the data is thin, the samples are few, come back later. That is true, and it is also the kind of thing said by arguments that never intend to be caught.
So it is worth naming what would count against the proposals set out on this site. If dense, well-dated sampling accumulates around Cudi Dağı and the upper Tigris across the third millennium BC and continues to show unbroken local continuity with no room for an incoming group, the arrival proposed here becomes difficult to maintain in any form that matters. The same applies in the Zagros: a well-sampled sequence spanning 5500 to 4300 BC showing smooth continuity would count against the reading offered there. That range is chosen to bracket the proposed crossing around 5000 BC and to run to the end of the period this site attributes to Cain’s line, since a test that stopped earlier would finish before the thing it was meant to test. Those are the results that would require the argument to change, and they are more likely than the alternative.