Skip to content
ArcanumA Fey Seam project
Speculation. Unverified interpretation, hypothesis or folklore. Presented as an open question only.Unverified interpretation, hypothesis or folklore. Presented as an open question only.

The Silurian Hypothesis

Researched and edited by Arcanum Research Desk · Archive last reviewed

Review dates apply to the archive as a whole, not to this entry individually.

Open library

Short answer

No — the Silurian Hypothesis does not claim an ancient industrial civilization existed. It asks a narrower scientific question: if one had existed millions of years ago, what traces would still be detectable today? The authors, a NASA climate modeller and an astrophysicist, state plainly in the paper that they know of no such evidence and consider the idea unlikely. It matters because it treats preservation bias as a measurable problem and proposes real tests, which is exactly the discipline this category is built on.

Why people ask this

The paper has a memorable name, a real journal, and mainstream authors, so it circulates widely — often as apparent scientific backing for lost-civilization claims. Reading what it actually argues is more interesting than the version that travels. It is a paper about the limits of our instruments, not about the contents of the past.

What we know

  • The paper is genuine, peer-reviewed, and authored by Gavin A. Schmidt (NASA GISS) and Adam Frank (University of Rochester).
  • The authors state explicitly that they are aware of no evidence for a pre-human industrial civilization and consider the hypothesis unlikely.
  • The paper's purpose is to assess detectability and to inform astrobiological searches, not to argue that such a civilization existed.
  • Direct artefact preservation over tens of millions of years is assessed as extremely improbable given erosion, subduction and the small footprint of urban areas.
  • Indirect markers considered include carbon and nitrogen isotope excursions, metal and rare-earth anomalies, synthetic organics, transuranics and biotic turnover.
  • Several past hyperthermal events, including the PETM around 56 million years ago, display broadly similar marker families with conventional geological explanations.
  • The authors propose concrete tests, including higher-resolution event-duration analysis and searches for synthetic and transuranic signatures.

What we don’t know

  • How far current stratigraphic resolution can actually distinguish a decades-long event from a multi-millennial one in deep-time sections.
  • Whether any proposed marker would be uniquely diagnostic of intelligence rather than of unusual but natural geochemistry.
  • Popular readings of the paper as support for lost civilizations — these contradict the authors' stated position.
  • How much weight the analogy to Earth's record should carry when applied to exoplanet technosignature searches.

What remains possible

Bounded unknowns. Nothing here is treated as evidence that the possibility is true.

  • Our stratigraphic resolution in deep time is often too coarse to distinguish an event lasting decades from one lasting tens of thousands of years.
  • Direct artefacts from a short-lived episode tens of millions of years old would be extremely unlikely to be found, whether or not any existed.
  • No proposed marker is uniquely diagnostic of intelligence on its own; it is the combination that would be hard to explain naturally.
  • None of this raises the likelihood that such a civilization existed. It only describes what we could and could not rule out.

What should have survived?

The traces this subject should have left, listed before we ask what we actually find.

  • A sharp negative carbon-isotope excursion from burning buried organic carbon.
  • Nitrogen-isotope changes from industrial fertiliser production.
  • Anomalous concentrations of metals and rare earth elements.
  • Synthetic organic molecules and their persistent breakdown products.
  • Transuranic elements, if fission were ever used.
  • An extinction pulse and abrupt biological redistribution.

What could disappear or be moved?

Preservation bias: the processes that erase, degrade or relocate the evidence.

  • Subduction destroys ocean-floor sediment on roughly 100-million-year timescales.
  • Erosion and metamorphism remove or overprint continental sections.
  • Urban areas cover a tiny fraction of the planet and consist mostly of non-fossilising materials.
  • Diagenesis smears sharp geochemical spikes into broader, less diagnostic signals.

What would we need to find?

Concrete findings that would strengthen, weaken or change this assessment.

  • Synthetic organic molecules with no natural production pathway in a securely dated deep-time layer.
  • Transuranic isotopes or their decay products where no natural reactor can explain them.
  • An event horizon resolvable to decades or centuries rather than millennia, ruling out the usual volcanic and orbital drivers.
  • A combination of markers that no single natural process produces together.

The longer research detail

In 2018 Gavin Schmidt, a climate modeller who directs NASA's Goddard Institute for Space Studies, and Adam Frank, an astrophysicist at the University of Rochester, published a paper in the International Journal of Astrobiology with a deliberately provocative title: "The Silurian hypothesis: would it be possible to detect an industrial civilization in the geological record?" The name is a joke borrowed from a Doctor Who species, and the authors say plainly in the paper that they know of no evidence that such a civilization existed and that they consider it unlikely. Anyone citing this paper as support for a lost prehistoric civilization is citing the opposite of what it argues. What the paper actually does is turn a question about the past into a question about method. Astrobiology needs to know what technological activity looks like at a distance and across deep time. Earth is the only planet with a well-sampled geological record, so the authors use it as a test case: if an industrial episode comparable to ours had occurred, say, fifty million years ago, what would remain, and would we currently be able to see it? Their answer is that direct evidence would almost certainly be gone. The fraction of Earth's surface preserving intact sediment older than a few million years is small; older surfaces are eroded, subducted, metamorphosed or buried beyond reach. Urban areas cover a tiny fraction of the planet, and the objects in them are exactly the things that do not fossilise. The odds of finding an artefact from a short-lived episode tens of millions of years back are, in their framing, extremely low — and they note the same logic makes our own artefacts unlikely to be found by a hypothetical successor. Indirect signatures are the interesting part. The authors catalogue what an industrial episode should imprint on the planet as a whole: a sharp negative carbon-isotope excursion from burning buried organic carbon, changes in nitrogen isotopes from industrial fertiliser, anomalous concentrations of metals and rare earth elements, synthetic organic molecules and their breakdown products, transuranic elements if fission were used, and biological markers such as an extinction pulse and abrupt species redistribution. These are precisely the kinds of signal that geologists already use to identify events in deep time. That leads to the paper's genuinely uncomfortable observation. The geological record contains several sudden hyperthermal events — the Palaeocene–Eocene Thermal Maximum around 56 million years ago is the best studied — that show exactly this family of markers: rapid carbon injection, isotope excursion, warming, ocean acidification, biotic turnover. The conventional explanations are volcanic, methane-hydrate or orbital. The authors are explicit that they are not proposing an industrial cause for the PETM. Their point is sharper: our current resolution is often too coarse to tell a decades-to-centuries-long event apart from one lasting tens of thousands of years, and duration is one of the few things that would distinguish an industrial spike from a geological one. They propose tests — finer-resolution isotope work, searching for synthetic molecules and unexpected transuranics, looking for compositional patterns natural processes do not produce. For this category, the paper matters for three reasons. First, it is peer-reviewed work by mainstream scientists that takes preservation bias seriously as a formal problem rather than as a rhetorical escape hatch. Second, it demonstrates the correct structure of the argument: define the claim, derive the traces, evaluate survival, propose a falsifying test. Third, it draws the line that most popular use of it erases. "We might not detect it" is a statement about our instruments. It is not evidence that there is something to detect, and it does not raise the probability of the claim by a single point. There is also a self-referential warning inside it. Schmidt and Frank note that the shorter and more intense an industrial episode, the harder it is to see in the record — and the more likely it is that the civilization responsible destabilised its own environment. That is a comment about us, not about the Silurian.

Origins and history

Published April 2018 in the International Journal of Astrobiology; preprint posted to arXiv the same month

Sources

Sources are labelled by type and access. Arcanum preserves claimant material where it documents the claim, while using academic, institutional and primary records to evaluate it.

Better questions to ask

Each question searches Arcanum’s own index — nothing is sent to an external search provider. Use “Copy question” to save the full wording anywhere you like.

Related threads

Your note

Notes are stored locally on this device only in this phase.