Earth's Missing Billion Years, and the Mantle That Took Them

Read the episode (Narration script)

The Grand Canyon has a missing chapter, and you can put your hand on it. Down at the bottom, pale Cambrian sandstone about five hundred and twenty five million years old sits, in places, directly on dark schist close to one point seven billion years old. Between those two surfaces, a billion years of rock is simply not there.

Geologists call it the Great Unconformity, and it is not an Arizona oddity. The same gap turns up across North America, in northern Europe, in North China. Depending where you stand, two hundred million to well over a billion years of the record has gone missing. Something stripped the continents bare and then tidied up after itself.

A paper published in Nature yesterday argues that the culprit sits three thousand kilometres beneath your feet.

The idea is called dynamic topography. The mantle, the hot rocky layer under the crust, circulates, extremely slowly. Where it rises, the continent above is held up a little; where it sinks, the continent sags. Not mountains: a few hundred metres of lift smeared across thousands of kilometres over tens of millions of years. But a continent held high gets scoured, and a continent that sags collects sediment and keeps it.

Santosh Dhungana and Nicolas Flament at the University of Wollongong ran six models of mantle convection back a billion years, steered by reconstructions of where the continents actually were, and watched what rose and what sank. The recipe for a Great Unconformity, they argue, is one long rise followed by one long sink. Rise, and ice and rivers take hundreds of millions of years of rock off your back. Sink, and the sea tucks that scarred surface under fresh sediment.

Three ancient continental cores got exactly that single cycle: Laurentia, the old heart of North America, Baltica in northern Europe, and North China. Those are precisely the three places the Great Unconformity is best documented. South America and Australia went up and down twice, and a second uplift erases the first gap rather than preserving it.

This organises the older explanations rather than throwing them out. Snowball Earth glaciers still do the scraping; the mantle decides who is standing in the way. Reconstructing plate positions a billion years back is genuinely uncertain, so the authors lean on the shape of the trends rather than exact heights. Even so, the deepest part of the planet seems to have chosen which pages of Earth's history we get to read.

Next, a different missing record, and a culture that burned its own evidence.

Also in Nature yesterday: four and a half thousand ancient human genomes from Bronze and Iron Age Europe, assembled to test where the Celtic languages came from. Celtic survives now at the Atlantic fringe, in Irish, Welsh, Scottish Gaelic and Breton. Two thousand years ago its relatives were spoken from Portugal to the Balkans, and four theories compete over where that spread began.

The team, led from the University of Copenhagen with linguists at Leiden, found three large movements of people, and only one lands inside the window linguists estimate for Proto-Celtic, the unwritten parent language. Between about three thousand two hundred and two thousand eight hundred years ago, ancestry from eastern central Europe, associated with the Urnfield world, spread west into France, the British Isles, northeastern Spain and Italy. That fits the oldest of the four theories, Celtic from the east.

There is a lovely obstacle buried in this. The Urnfield people are named for their habit of cremating their dead and burying the ashes in pots, so from exactly the centuries that matter most, there are very few skeletons left to sequence. And genes are not languages: people switch tongues without moving, and move without switching. The authors are careful. This is the story that needs the fewest extra assumptions, not a proof.

Over on Mars, the European Space Agency's Trace Gas Orbiter has found something out of season. Hydrogen chloride in the Martian air normally peaks in northern winter and mostly vanishes in summer. Yet above the northern flank of Alba Mons, an enormous old volcano, the spectrometers keep catching it in early summer, right where the orbiter's camera sees fresh narrow cracks with the ground discoloured around them.

The suggestion, in Science Advances, is gas working up from magma still sitting below, through vents plugged with ice all winter at around minus a hundred and twenty degrees and opening when that ice turns straight to vapour in spring. Nobody has seen an eruption or new lava, and the paper's own title calls the detections enigmatic. It is still the sharpest hint yet that Mars may not be entirely finished.

Back on Earth, something smaller. Researchers at Rockefeller University report that males of Drosophila erecta, a West African cousin of the laboratory fruit fly, will not court at all unless they can smell food. Put a male and a perfectly willing female on a clean surface and nothing happens. Add the scent of fruit and he starts chasing, and the sight of other flies moving around him turns the dial up further, so a crowd on one piece of fruit works itself into a frenzy. The ordinary lab fruit fly treats food smell as encouragement rather than a precondition. The wiring is much the same in both; what differs is the threshold.

One last thing, about seeing. To photograph a planet beside its star you have to subtract the star, which is typically a hundred thousand times brighter. The usual trick is to use images of other stars as a template, and that fails as the telescope warms, cools and drifts between exposures. A team led from Northwestern tried the opposite: a physics model of the telescope's own optics, with around a million adjustable parameters, fitted to the image itself using automatic differentiation, the same gradient machinery that trains large neural networks. Thirty five minutes on a single graphics processor, sensitivity improved by two to fifteen times, and the young giant planet 51 Eridani b appeared at a wavelength where the earlier analysis had seen nothing.

Which is a cheering thought for a Thursday. Some of what is out there is waiting on better instruments, and some of it is waiting on better ideas about the instruments we already have.