The Silurian Hypothesis — The Faint Signal
Station 01 · Open water
Detection survey — deep time
Schmidt · Frank
Int. J. Astrobiology 18(2)

If they existed, what would be left?

The Silurian
Hypothesis

A peer-reviewed thought experiment: if an industrial civilization rose on Earth millions of years before us — would anything remain that we could actually find?

The industrial layer
~300 years ≈ 1 cm of sediment

Station 02 · The question · T−0

A thought experiment,
asked with a straight face

In 2018 Gavin Schmidt, director of NASA's Goddard Institute for Space Studies, and Adam Frank, astrophysicist at Rochester, published a paper asking a question usually left to late-night television: would it be possible to detect an industrial civilization in the geological record?

It started as a conversation about other planets. Frank was describing how a civilization alters a world; Schmidt stopped him — how do you know that hasn't already happened here? Neither had an answer. The name is a wink at the 1970 Doctor Who serial about reptiles waking beneath the English countryside. The geochemistry is not a joke.

And here is what almost everyone gets wrong: neither author thinks such a civilization existed. They wrote the hypothesis to find out what our evidence can and cannot rule out.

"The Silurian hypothesis cannot be regarded as likely merely because no other valid idea presents itself." Schmidt & Frank, 2018

Station 03 · The window · 400 Mya → today

Where the question
is even allowed

Note what they cut to keep it honest. The analysis runs from the Devonian, ~400 million years ago, to the mid-Pliocene, ~4 million years ago — and stops. Everything after that is young, shallow, heavily excavated ground where metallurgy, farming and chemistry would show up. That single choice separates this paper from every documentary that misquotes it.

Fig. 01 · Deep-Time Ribbon

The Searchable Window

The searchable window: a deep-time ribbon, 400 million years ago to today The Paper’s Search Window Record too poor, here, to rule a civilization out. Excluded Young, shallow, heavily excavated ground. Oldest Surviving Crust · ~170 Mya No seafloor older than this survives anywhere. 400 300 200 100 50 20 10 4 0 MYA Kellwasser 378 Mya Hangenberg 359 Mya Carb. Collapse 305 Mya End-Permian 252 Mya Toarcian 183 Mya Weissert 132 Mya OAE 1a 120 Mya OAE 2 93 Mya Strongest-Dated Candidate PETM 56 Mya Industrial Era · 300 Yrs Invisible at this scale.
The Search Window 400–4 Mya

Nine hyperthermal and mass-extinction horizons sit inside the record gap. Hover or tap a gold marker to read what each layer is — and why nobody has yet screened it for an industrial fingerprint.

A piecewise-scaled ribbon from the Devonian to today, compressed so recent time stays legible. The cyan band is where the fossil record is too incomplete to rule out a prior industrial civilization; the coral sliver at the right — the last four million years — is not. Gold ticks mark real hyperthermal and extinction layers, none yet screened for artificial signatures.

Station 04 · The eraser · How Earth forgets

The planet is a machine
for destroying evidence

Instinct says a civilization would obviously leave ruins. On geological time, instinct is wrong.

~170Myr — oldest ocean floor

Crust is born at ridges and swallowed at trenches. Two-thirds of Earth's surface self-erases faster than complex life has existed.

~1.8Myr — oldest large land surface

The oldest big continuous surface you can simply walk onto — the Negev — is 1.8 million years old. Older exposure is orders of magnitude rarer.

<1%of Earth urbanised

Every city together covers under one percent. A deep-time city must survive burial and surface exactly where someone digs.

A fewfossil individuals per 100,000 yr

Across thousands of dinosaur taxa. A species 300 years into industry could leave no bones at all.

~1 cmour whole era, in future rock

Sediment builds a few cm per thousand years. Three centuries of industry compresses to a hairline. Not a chapter — a line.

Fig — Seafloor Recycling

The Conveyor

Nothing survives the round trip Max age of ocean floor ≈ 170 million years · two-thirds of Earth's surface
Oceanic crust is born at the ridge, carries a thin sediment record for at most ~170 Myr, then bends into the trench and is subducted — erasing whatever it held before it can be found.

Station 05 · The core · T−170,000,000 yr

Everything we ever built,
to scale

Pull a sediment core and read it downward. Our entire industrial history is the thinnest thing in it — and the record itself bottoms out at 170 million years, because below that there is no seafloor left to read.

The Core

A sediment column, read top to bottom — from today's seabed down to the oldest ocean crust still in existence. Select a layer.

Seabed mudToday
Quaternary0–2.6 Mya
Plio / Miocene2.6–23 Mya
Cretaceous chalk66–145 Mya
Oldest crust~170 Mya
Subducted — no record survives
Ocean sediment locks in whatever falls on it, one layer at a time, for as long as the seafloor survives beneath it — and two-thirds of it eventually doesn't. Everything industrial civilization has produced compresses to a hairline near the top.

Station 06 · The fingerprints · What survives

Industry doesn't leave ruins.
It leaves chemistry.

This is the paper's real contribution: stop looking for statues. What lasts a hundred million years is a geochemical signature — a thin band where the chemistry of a whole planet briefly went strange. Six things to hunt for.

Carbon isotopes

δ¹³C · permanent

Fossil carbon is poor in ¹³C. We have burned over half a trillion tonnes of it, shifting the global ratio about −1‰. That number copies itself into every living thing, then into the mud. The loudest mark industry makes.

Orphan isotopes

Pu-244 · 80.8 Myr half-life

Plutonium-244 and curium-247 have no natural source on Earth. Outside a supernova, they mean a reactor or a weapon. Closest thing in the whole paper to a smoking gun.

Molecules with a handedness

Chirality · biology can't fake it

Life is almost entirely single-handed. Industrial synthesis usually makes both mirror forms in equal measure. A 50/50 mix in ancient sediment is a signature chemistry-without-a-factory does not produce.

Fixed nitrogen

δ¹⁵N · as long as the mud

The Haber–Bosch process now fixes more nitrogen than the entire natural land cycle. Runoff distorts nitrogen isotopes and starves coastal water of oxygen — recording as something almost indistinguishable from an ancient anoxic event.

Metals out of place

With the host rock

Human flux now exceeds the natural background for lead, chromium, antimony, rhenium, the platinum group, the rare earths and gold — smeared across sediment in combinations no ore body or volcano produces.

Polymers

Uncertain — needs testing

Plastics break down mostly under sunlight. Buried in anoxic mud, out of the light, that clock nearly stops. How long a polymer horizon actually survives is one of the paper's honest unknowns.

Orphan isotopes — what's still there to findfraction remaining · schematic floor
Plutonium-244 · half-life 80.8 Myr Curium-247 · half-life 15 Myr Detection floor
Neither isotope has a natural terrestrial source. Find plutonium-244 in a hundred-million-year-old layer with no impact crater and no nearby supernova to blame, and you are looking at a reactor. It is the cleanest test in the paper — and the only one that stays legible this deep.
The handedness testwhy a mirror image gives it away
Life builds this one
A factory builds both, equally
Same atoms, same bonds, opposite handedness — like a left and right glove. Biology picks one and sticks to it with near-total consistency. Industrial chemistry has no such preference, so it turns out roughly fifty-fifty. Find that fifty-fifty mix locked in ancient rock and you have found chemistry that life did not do.

Station 07 · The twin spikes · T−56,000,000 yr

Earth has seen this
signal before

Fifty-six million years ago the Paleocene–Eocene Thermal Maximum dumped thousands of gigatonnes of carbon into the air. Temperature rose 5–7 °C, oceans acidified, 30–50% of deep-sea foraminifera species died, and carbon isotopes swung over 3‰ negative — the exact shape of signal industry makes.

Nobody thinks it was a civilization. The unnerving part is the rate: the PETM injected carbon at roughly 1–4 petagrams a year against our ~10 — some ten times faster than anything in 66 million years. But sediment blurs time. Compressed into rock, a frantic three-century civilization and a slow multi-millennium belch can look like the same thin smear.

Carbon release — the shape of the signalschematic · not to scale
PETM · 56 Mya · millennia wide Us · now · ~300 years

The record holds a whole shelf of these — Toarcian, Weissert, two Cretaceous anoxic events, the Kellwasser events, the end-Permian. Several are still unexplained. Almost certainly all natural. The paper's actual ask is narrower and more interesting than any conspiracy: nobody has ever gone back and screened those layers for the other fingerprints.

Station 08 · The paradox · The best idea in the paper

Lasting longer
makes you harder to see

A civilization only survives a long time if it stops wrecking its planet. But the geological signal is the wreckage.

Civilization lasts longerits practices must be sustainable
More sustainablesmaller footprint on the planet
Smaller footprintless signal in the rock
Footprint against survivalthe detectability paradox
Geological signal left behind Chance the civilization survives
The two things move in opposite directions. Anything loud enough to find in the rock was, almost by definition, not living within its planet's means — which is why it stopped. The civilizations worth becoming are the ones that leave almost nothing to find.
"The longer a civilization lasts, the more sustainable its practices would need to have become in order to survive… But the smaller the footprint, the less of a signal will be embedded in the geological record." Gavin Schmidt, RealClimate, 2018

So if we ever do find an industrial layer in deep time, we have most likely found a civilization that failed — because a successful one would be almost invisible. The same logic points at us. Our layer is loud right now precisely because we are still in the reckless part.

Station 09 · The verdict · Reading the instruments honestly

So — were there Silurians?

The straight answer: no. There is no evidence for any pre-human industrial civilization. No anomalous isotope layer, no orphan plutonium, no polymer horizon has ever been found.

What the paper actually establishes

Absence of evidence is weaker than we assumed

Before 2018, "we'd obviously know" was an unexamined default. Their conclusion, in their own words: for potential civilizations older than about 4 million years, the chance of finding direct evidence via objects or fossils is small. Not because it's likely — because Earth keeps such poor records. That is a real, uncomfortable result about the limits of what rock can remember.

What this is

A detectability study in deep time — tens to hundreds of millions of years, long before anything human.

A field guide for reading strata and exoplanet atmospheres for industrial traces.

A mirror. Its true subject is how visible we will be.

What this is not

Support for a lost human civilization — Atlantis, an advanced Ice Age culture, Younger Dryas survivors. That whole era sits inside the 4-million-year window the paper deliberately cut out, in ground where metallurgy, agriculture and chemistry would all show. They don't.

Nor does it need to be. Göbekli Tepe's carved enclosures, raised by hunter-gatherers around 9600 BCE, and the drowned world of Doggerland are astonishing and fully documented. Wonder and evidence are not in competition.

The neighbouring debate has moved too: the Younger Dryas impact hypothesis — the comet usually invoked to explain a vanished civilization — lost more ground in April 2026, when a Science Advances paper argued volcanic forcing better explains the cooling in North American sediments. The Silurian Hypothesis is frequently drafted into service for these claims. It does not support them, and its authors have never said it does.

Station 10 · The deep signal · Why it matters anyway

The question was never
really about the past

The search for life elsewhere

The cruellest term in the Drake equation is L — how long a technological civilization lasts. If they are brief planetary events, their traces are thin everywhere in the universe. Knowing what industry writes into a planet tells astronomers what to look for.

Civilizations as planetary events

Frank's Light of the Stars (2018) reframes any energy-hungry civilization as a geophysical process — something a planet does, like an ice age or a flood basalt. Every such process feeds back on its own climate. The open question is whether any of them learn to persist.

The mirror

Run it forward and it points at us. In a hundred million years everything ever built resolves to a centimetre of odd chemistry on a seabed nobody has visited. Schmidt even wrote a companion short story, Under the Sun, about the moment of discovery.

Station 11 · Seabed · T+100,000,000 yr

A hundred million years from now, all of this is a line one centimetre thick.

The Silurians almost certainly never existed. But the instrument that could have found them, we only just built — and the paradox at the heart of the paper says the civilizations we could detect are the ones that didn't make it. Right now, ours would be easy to spot.