A fossil has an age and a lineage, and those are two different questions.
Most years, palaeoanthropology answers one at a time, and the answers sit
comfortably inside the existing calendar. 2026 is not most years.
In May 2026, two papers landed within twenty-four hours of each other — both
in Nature, both on hominin deep time, both pushing the calendar back rather
than forward — and they did it with methods that have almost nothing in
common.
The first was dated from the outside in. A molar from the Ledi research area
in the Afar, bracketed by volcanic tuffs, was dated by measuring the
$^{40}$Ar/$^{39}$Ar ratios in the ash layers above and below it. The fossil
itself was never touched. The bracket said 2.595 ± 0.035 million years —
and because the molar was confidently attributable to Paranthropus, the
Paranthropus–Homo split gets pushed back with the fossil, to at least
3.0 Ma. The 2014 consensus had the divergence near 2.5 Ma; the Afar bracket
says no, earlier.
The second was read from the inside out. Fu Qiaomei's group at IVPP recovered
enamel proteins — not DNA, never DNA, the molecules are too old — from
Homo erectus teeth at three Chinese localities (Zhoukoudian, Hexian,
Sunjiadong), at least 400,000 years old, using an acid-etching technique that
takes only a dust-mote of material and leaves the fossil's shape intact.
Two amino-acid variants in the AMBN gene did the work. A253G had never
been seen in any primate — a brand-new molecular marker, the strongest
possible evidence that these East Asian erectus populations were one
population, not three. M273V had previously been considered
Denisovan-specific. It wasn't. It was already in East Asian Homo erectus, and
the simplest reading is that it flowed from erectus into Denisovans and then
into some modern Oceanian and Southeast Asian groups — the first molecular
handle on a route that, until this year, was a diagram with no arrow.
Outside-in vs. inside-out. A bracket in the Afar vs. enamel proteins from
three Chinese sites. A volcanic layer vs. a gene. Different fossils,
different labs, different continents, different things being measured — and
the same direction of revision. Older. The new calendar is the older one.
This is what a correction phase in deep-time science actually looks like.
Not a single paper that overturns the consensus, but two or three independent
ones that converge on the same direction. When argon-argon says "older than the
2014 consensus" and enamel proteomics says "the erectus–Denisovan map has to
be redrawn" in the same week, the consensus was the artefact.
The third geometry is worth a mention. In February 2026, Tu et al. redated
the Yunxian cranium to 1.77 Ma using cosmogenic $^{26}$Al/$^{10}$Be burial
dating — measuring how long the sediment around the fossil had been shielded
from the cosmic-ray flux. A third method. A third geometry — neither bracket
nor molecule but a third independent clock. Older, again.
Three geometries. Outside-in via volcanic ash. Inside-out via protein.
Around-the-fossil via cosmogenic burial. Three clocks that don't tick on
the same mechanism, run by different groups, on different continents, in the
same calendar year — all three saying the hominin calendar was too recent.
That is not three papers; that is a metrology argument, and the metrology is
the point.
| Paper | Date | What it pushed back | Method | Journal |
|---|---|---|---|---|
| Ledi 1 (Afar, Paranthropus) | 14 May 2026 | Paranthropus–Homo split to ≥3.0 Ma | $^{40}$Ar/$^{39}$Ar on volcanic tuffs (outside-in) | Nature 655 (14 May 2026) |
| **Hexian / Zhoukoudian / Sunjiadong *H. erectus*** | 13 May 2026 | H. erectus–Denisovan map; new Asian erectus monogroup | Palaeoproteomics (LC-MS/MS on enamel, AMBN variants) | Nature 655 (13 May 2026) |
| Yunxian 2 (Tu et al.) | 20 Feb 2026 | H. erectus earliest Chinese cranium to 1.77 Ma | Cosmogenic $^{26}$Al/$^{10}$Be burial dating (around) | Science Advances 12(8), 20 Feb 2026 |
The durable observation for the next decade of hominin dating is small and
slightly uncomfortable: the 2010s–2020s maturation of high-precision
radiometric dating, ancient-protein recovery on thermally damaged or
chronologically out-of-range material, and cosmogenic burial dating each
produced headline-grade results in the same year. The headline is always
older. The calendar you learned ten years ago was too recent, and the three
methods that now exist to test it have started doing so in concert.
Sources
- 400,000-year-old teeth reveal a hidden link in human evolution — *ScienceDaily*, 9 Sep 2026
- Fu, Q. et al. (2026). Enamel proteins from six Homo erectus specimens across China. Nature 655 (8121): 141. DOI: 10.1038/s41586-026-10478-8
- Ledi 1 Paranthropus molar — Nature 655, 14 May 2026 (argon-argon on Afar volcanic tuffs; primary reference cited in the companion piece on the 2026 Ledi redating).
- Tu et al. (2026). Cosmogenic burial dating of the Yunxian Homo erectus cranium to 1.77 Ma. Science Advances 12(8), 20 Feb 2026. DOI: 10.1126/sciadv.ady2270