Chapter 7: Air, Ocean, Continent

There is a way to prove that the air came out of the rock, and it takes one measurement.

Argon makes up a little under one per cent of the atmosphere. Almost all of it is a particular isotope, argon-40, and argon-40 has an unambiguous origin: it is produced by the radioactive decay of potassium-40. It is not primordial, it was not delivered by comets, and it cannot be manufactured by any chemical or biological process. Every atom of it was made inside a mineral grain, by a potassium atom that fell apart.

Potassium-40 is distributed through the rock of the Earth's interior. Argon-40 is now sitting in the sky. There is no route between those two places except outgassing.

And the quantity is not marginal. Roughly half of all the argon-40 that Earth's potassium has ever produced is now in the atmosphere rather than still locked in the interior — which means the planet has, over four and a half billion years, ventilated a substantial proportion of its own interior gas inventory to the surface. Not as a trickle. As the dominant fate of that gas.

I open with argon because it is the cleanest available demonstration of the thing this chapter is about, and because it is not an inference from a model. It is arithmetic on an isotope. The atmosphere is, in significant part, exhaled planet.

This is the load-bearing chapter of the book. Chapter 6 gave something back; this one will not have to. Register: predominantly empirical, with theoretical sections flagged where the science is genuinely open — and one section near the end which I include specifically because it is the weakest link in the argument.

What the planet exhaled

Earth almost certainly had an earlier atmosphere, captured from the gas of the solar nebula while it was still forming: hydrogen and helium, in the proportions of the Sun. It is gone. We know it is gone because the noble gases we have are wrong — the pattern of neon and the other unreactive gases in our air does not match a solar-derived envelope, which it would if we had simply kept what we caught. That first atmosphere was lost, to some combination of hydrodynamic escape while the young Sun was violently active and the enormous energy of the impacts that finished building the planet.

What we breathe is a second atmosphere, and it came from inside.

The gas that comes out of a volcano today is dominated by water vapour — typically the great majority of it — followed by carbon dioxide, then sulfur species, then chlorine and fluorine compounds and nitrogen. Run that mixture continuously for billions of years, condense the water, dissolve the soluble acids into it, and let the nitrogen accumulate because it is unreactive and has nowhere else to go, and you get, in outline, the composition of the air.

The oxygen is the exception, and it is worth being clear about. Free oxygen is not volcanic. It is biological waste — the product of photosynthesis, accumulated over billions of years. Volcanism supplied the atmosphere; life then poisoned it with a highly reactive gas that most of the organisms then alive presumably found catastrophic, and which we now depend on. That is Chapter 6's ecosystem argument running at planetary scale, and it is a useful reminder that the engine built the stage rather than the whole play.

There is a second line of evidence I find as satisfying as the argon. Helium-3 is primordial: it was present when the solar system formed and is not produced in any quantity by decay. It is light enough to escape to space from the top of the atmosphere, so any that we find must have arrived recently. And it is found, leaking out of the mantle at mid-ocean ridges and at hotspots, at measurable rates.

Which tells you the degassing has not finished. This is not an ancient event that happened and stopped. The planet is still breathing out gas that has been inside it since before there was an Earth to be inside of.

Where the ocean came from

Now the ocean, and here I have to be careful in a way that Chapter 6 taught me to be careful, because the popular version of this claim overstates it in exactly the manner this book is trying to avoid.

The tracer is the ratio of deuterium — heavy hydrogen — to ordinary hydrogen. Different reservoirs in the solar system carry distinguishably different D/H ratios, and water largely keeps the signature of where it came from. So you can compare Earth's ocean against the candidates.

Comets were long the favoured source, and for a while the measurements looked unhelpful: several of the comets measured had roughly twice the deuterium of seawater, which is difficult to reconcile with them having delivered most of it. Later measurements of other comets came out closer. The picture is not clean.

Carbonaceous chondrites — a class of primitive meteorites — match Earth's water considerably better. And the current mainstream position is that most of Earth's water arrived as part of the material the planet was built from, chemically bound into hydrous minerals in the accreting rock, rather than being delivered afterward by a bombardment of ice.

So here is the honest statement, and note that it has the same shape as the Lost City correction. The water was not created by volcanism. It was accreted inside the rock, and volcanism is the mechanism that got it out. Heat the hydrous minerals, drive off the water, erupt it as steam, condense it. The engine is the release mechanism, not the source.

That is a smaller claim than "volcanoes made the ocean," and it is the true one.

What is genuinely remarkable, and less widely known, is that the ocean is not a closed surface layer sitting on top of the rock. Water goes back down. Subducting plates carry hydrated minerals into the mantle — this is the same process that drives arc volcanism, from Chapter 1 — and some fraction of that water is not returned at the arc but continues deeper. The mantle contains water, dissolved in the crystal structures of its minerals, and current estimates for how much range from a fraction of an ocean to several oceans' worth.

The most direct evidence is the one I ended Chapter 1 with: a speck of ringwoodite, trapped inside a diamond from Brazil, containing water, from the transition zone hundreds of kilometres down. That single inclusion is the closest thing anyone has to a physical sample of the deep water cycle.

The ocean is not on the planet. It is circulating through it, on a timescale of hundreds of millions of years.

Why there is anywhere to stand

Continental crust is a peculiar substance and its existence requires explanation.

Oceanic crust is basalt: made at ridges by decompression melting, dense, and — this is the key property — dense enough to be pushed back down into the mantle when it gets old and cold. It is temporary. The oldest oceanic crust anywhere on Earth is a couple of hundred million years old, which on the planet's timescale is nothing. The ocean floor is continuously manufactured and continuously destroyed.

Continental crust is different. It is granitic to andesitic in bulk composition — considerably richer in silica, aluminium and the alkalis — and it is buoyant. It cannot readily be subducted. Once made, it stays, and the oldest surviving continental material is billions of years old.

Where does it come from? From the distillation described in Chapter 1. Flux melting at subduction zones produces magmas already enriched in silica relative to the mantle; those magmas stall in the crust, crystallize, and the residual melt becomes more silica-rich still; that melt rises and the process repeats. Every cycle drives the composition further from peridotite and further toward granite, and the products are too light to sink.

Continents are the accumulated scum of four billion years of arc volcanism. They are the tailings of the engine.

The consequence for anything living is direct. Without buoyant crust, a planet with Earth's water inventory would very likely be a waterworld — ocean everywhere, no exposed land. And I want to draw out one specific consequence of that, because it is the hinge of this chapter and it is usually left implicit.

Exposed continental rock is the weathering surface. The thermostat I am about to describe works by CO₂ dissolving in rainwater and attacking silicate minerals. That reaction needs silicate rock in contact with rain. Seafloor weathering exists and contributes, but the continental surface — mountains, uplifted terrain, freshly exposed rock — is where the bulk of the drawdown happens, and mountains exist because tectonics builds them.

So the same engine that supplies the carbon dioxide also manufactures the surface that removes it. Both halves of the loop are its products. Hold that thought.

The thermostat

This is the most important argument in this book. Everything else in Part III is supporting structure.

In 1981, James Walker, Paul Hays and James Kasting published a paper in the Journal of Geophysical Research proposing a negative feedback mechanism for the long-term stabilization of Earth's surface temperature. It is not a long paper and it is not famous outside the field, and I think it is one of the more important things written in the twentieth century about why this planet is habitable.

The loop runs like this.

Volcanoes and metamorphic decarbonation release carbon dioxide into the atmosphere. Carbon dioxide dissolves in rainwater, making it weakly acidic. That acidic water attacks silicate rocks, breaking them down and releasing calcium and magnesium ions along with bicarbonate, which rivers carry to the sea. In the ocean, those ions combine — biologically, mostly, though the process predates the organisms that now dominate it — to precipitate carbonate minerals, which settle to the seafloor and are buried. Subduction carries that carbonate down. At depth it is heated, the carbon is driven off as CO₂, and arc volcanism returns it to the atmosphere.

Carbon in, carbon out, over a full circuit of the planet's interior.

Now the part that makes it a thermostat rather than merely a cycle. The weathering step is temperature-dependent. Chemical reactions go faster when hot. And a warmer planet also has a more vigorous hydrological cycle — more evaporation, more rainfall, more runoff — which delivers more water to more rock. Both effects push the same way.

So: if the planet warms, weathering accelerates, CO₂ is drawn out of the atmosphere faster than volcanism supplies it, the greenhouse weakens, and the planet cools. If the planet cools, weathering slows — and in the limit, freezes out altogether — while volcanic outgassing continues regardless, so CO₂ accumulates, the greenhouse strengthens, and the planet warms.

It is a negative feedback with a volcanic supply term and a climatic sink term, and it has been running for billions of years.

I want to state its response time immediately and emphatically, because this mechanism is periodically invoked in bad faith. The loop operates on timescales of hundreds of thousands to millions of years. It is not going to help with anthropogenic climate change. It cannot help. On any horizon a human institution can plan for, the thermostat is a fixed background, not a corrective. Anyone citing the carbonate–silicate cycle as a reason for equanimity about present emissions has either not understood the timescale or is relying on you not to.

The proof that it matters: a Sun that got brighter

Here is why the thermostat is not merely elegant but necessary.

Stars brighten as they age. The Sun at the time of Earth's formation was substantially fainter than it is now — around seventy per cent of its present output, an increase since then of roughly thirty per cent — and it has been rising steadily ever since. This was pointed out as a problem in the 1970s, and it is genuinely a problem: run the calculation with today's atmosphere and a young, faint Sun, and Earth's surface should have been frozen solid for the first couple of billion years of its existence.

It was not. There is evidence of liquid water on the surface from very early on — from ancient zircon crystals that appear to have formed in the presence of water, and unambiguously from sedimentary rocks that could only have been deposited underwater.

So the early Earth was warm under a faint Sun, and has stayed within the liquid-water range as the Sun brightened by a third. A fixed atmosphere cannot do that. Something has to have been adjusting the greenhouse downward at roughly the rate the Sun was turning up.

The thermostat does exactly this. A weathering sink that responds to temperature will hold the surface near a set point while the input changes, drawing atmospheric CO₂ steadily down over geological time as the solar input rises. High CO₂ early, low CO₂ now, liquid water throughout.

Register: theoretical, with an honest complication. The thermostat is the framework, and I do not think the framework is in doubt. But whether it is the whole answer is argued. Some reconstructions of ancient atmospheric CO₂ from fossil soils come out lower than the simple version requires, which has led people to invoke supporting contributions — methane, which is a potent greenhouse gas and would have been more stable in an oxygen-free atmosphere; or a lower planetary albedo early on, with less exposed continent and fewer of the biological aerosols that seed clouds. The debate is about what else was involved, not about whether the thermostat was.

Snowball Earth: the thermostat's proof of work

And then there is the episode where the mechanism was pushed to its absolute limit, and the record of what happened is written in the rocks.

Roughly six to seven hundred million years ago, in the Neoproterozoic, the Earth froze — twice, in two long glaciations. This is not an ordinary ice age. Glacial deposits from these events are found in places that palaeomagnetic evidence places at tropical latitudes, at sea level. Ice at the equator, at the coast.

The physics behind this is a runaway: ice is reflective, so as ice advances the planet absorbs less sunlight and cools, which advances the ice further. Past a critical latitude the feedback becomes unstoppable and the ice sheets close on the equator. That much has been understood as a theoretical possibility since the 1960s. The evidence that it actually happened accumulated through the 1990s.

The obvious objection is that this should be permanent. A frozen white planet reflects most of what the Sun sends it. There is no obvious way back.

Unless something keeps adding greenhouse gas regardless of how cold it is.

Volcanoes do not care whether the planet is frozen. Outgassing depends on the interior heat engine, not on the surface temperature. So through the glaciation, carbon dioxide continued to be supplied at roughly its usual rate — while the sink was switched off, because weathering requires liquid water, rain and exposed rock, and there was ice over everything.

Supply continuing, sink disabled. Carbon dioxide accumulates. Over millions of years it builds to concentrations vastly above anything in the recent record, until the greenhouse forcing finally overwhelms the ice albedo, and the planet deglaciates — extremely fast, because the same runaway now runs in reverse.

And then the aftermath, which is the part that turns this from a story into evidence. A newly deglaciated planet has an enormous CO₂ load, a hot wet climate, and vast expanses of freshly exposed, finely ground rock. Weathering goes into overdrive, ions pour into the ocean, and carbonate precipitates in enormous quantity.

Which is precisely what is found. Directly on top of the glacial deposits, all over the world, sit distinctive layers of carbonate rock — laid down immediately after the ice, in exactly the way the mechanism predicts. They were an anomaly before the theory. They are a signature after it.

The details remain argued — whether the ocean froze completely or retained open water, the precise durations, the role of the biosphere. I do not think any of that touches the point being made here.

This is the thermostat's proof of work. Take the mechanism, break the sink entirely, freeze the planet from pole to pole, and the volcanic supply term still hauls it back out. There is no clearer demonstration anywhere in the geological record that a planet's habitability is actively maintained by its interior, rather than being a passive property of its distance from a star.

The weakest link

I have one more component to offer and I am going to argue against it, because it is the place where the standard version of this chapter overreaches and I would rather do the overreaching myself and then withdraw it.

The usual next paragraph runs: Earth's core convects, that convection generates the magnetic field, the magnetic field deflects the solar wind, and without it the atmosphere would be stripped away — as happened at Mars, which lost its dynamo early and then lost its air. And since core convection depends on the mantle carrying heat away from the core, the magnetic field belongs on the list of things the interior heat engine provides.

The first half of that is solid. The geodynamo does require the core to be losing heat, and the rate at which it loses heat is controlled by mantle convection, which is the same circulation that drives volcanism. The link between the engine and the field is real.

It is the second half — that the field is what protects the atmosphere — that has been substantially revised, and I think honesty requires saying so.

Venus has no intrinsic magnetic field and has an enormously massive atmosphere. That single fact should have made everyone more careful than they were. Bodies without intrinsic fields develop induced magnetospheres through the interaction of the solar wind with their ionospheres, and those provide some protection. And there is modelling and spacecraft work suggesting that an intrinsic field can, under some conditions, facilitate atmospheric loss by channelling ions out along open field lines at the poles — so the sign of the effect is not even unambiguous.

The current position, as best I can characterize it, is that magnetic shielding is one factor among several — alongside gravity, atmospheric composition, and the intensity of the stellar wind — and that the confident textbook version, in which the field is the reason Earth kept its air and Mars did not, is not well supported.

I include this section precisely because it would have been easy to leave in. It is a satisfying argument, it points the right way, and almost nobody would have challenged it. It is also, on the current evidence, weaker than it sounds — and a chapter that has just made a very strong claim about the thermostat should demonstrate that it knows the difference.

The system, assembled

Let me put the pieces together, because the argument is cumulative and its force is in the assembly.

The air was outgassed from the interior, and we can prove it with an isotope that has nowhere else to have come from.

The ocean was accreted inside the rock and released by heating — the engine as mechanism rather than source — and it continues to circulate through the mantle rather than merely sitting on it.

The continents are the accumulated, un-subductable product of arc volcanism: the reason there is dry land, and the reason there is a weathering surface.

The thermostat couples the last two, with volcanic outgassing as the source term and weathering of tectonically exposed rock as the sink, and it has held the surface in the liquid-water range across a thirty per cent increase in solar output. Snowball Earth is the demonstration that the supply term keeps running when everything else has stopped.

And underneath all of it, chemical disequilibrium — the condition Chapter 6 established as the prerequisite for any origin of life, and which a geologically dead planet does not sustain.

Here is what I want you to notice about that list. It is tempting to read it as five separate services rendered by volcanism, like an itemized bill. That is the wrong reading and it undersells the argument. They are not five benefits. They are five aspects of a single process: a planet exporting internal heat, and doing so through a system that moves material as well as energy — up through eruption, sideways through spreading, and back down through subduction.

Break any part of the circuit and the rest degrades. Stop the outgassing and the thermostat loses its supply. Stop the subduction and the carbon does not come back, and the continents stop being built, and the weathering surface stops being renewed. Stop the interior convection and all of it ends at once.

Which raises the obvious test, and it is the right way to check an argument of this kind: what does a planet look like where part of that circuit is missing?

We do not have to speculate. There are four of them within reach of a telescope, and one of them has volcanoes in abundance, no subduction whatsoever, and a surface hot enough to melt lead.

That is Chapter 8.

Draft notes — verification status

Standing convention. The Walker/Hays/Kasting citation was resolved during the thesis-paper verification pass. This is the book's load-bearing chapter and it should get the most rigorous checking of any chapter — several claims here carry the entire argument.

Already verified — do not re-check:

• Walker, J.C.G., Hays, P.B. & Kasting, J.F. (1981), "A negative feedback mechanism for the long-term stabilization of Earth's surface temperature," Journal of Geophysical Research 86(C10), 9776–9782.

Pending verification — high priority:

• The argon-40 argument. Atmospheric argon abundance (~0.93%), the ⁴⁰Ar/³⁶Ar ratio, and the fraction of Earth's total radiogenic ⁴⁰Ar now residing in the atmosphere. The draft says "a very large fraction" — this is the chapter's opening proof and the number must be right. Commonly cited figures are around half; confirm.
• Solar luminosity increase. Draft says "a quarter to a third" early on and "thirty per cent" in the summary — make these consistent and source. Standard figure is ~70–75% of present at 4.5 Ga.
• Thermostat response time ("hundreds of thousands to millions of years").
• Sagan & Mullen (1972) as the origin of the faint young Sun problem.
• Neoproterozoic glaciation ages (Sturtian ~717–660 Ma; Marinoan ~650–635 Ma) — the draft says "six to seven hundred million years ago," which should be checked and possibly tightened.
• Palaeomagnetic evidence for low-latitude sea-level glaciation; Kirschvink (1992) for the "snowball" formulation; Hoffman & Schrag (1998) for the synthesis.
• Cap carbonates as the predicted post-glacial weathering signature.

Pending verification — standard:

• Loss of the primary nebular atmosphere and the noble gas (neon) evidence for it.
• Typical volcanic gas composition by volume (H₂O dominant, then CO₂, then S species).
• ³He as primordial, its escape from the atmosphere, and measured mantle outgassing at ridges and hotspots.
• D/H ratios: Earth's ocean (VSMOW), Oort-cloud comets (~2× terrestrial), Jupiter-family comets (103P/Hartley 2 near-terrestrial), carbonaceous chondrites. Confirm the current balance of evidence favours accreted hydrous minerals over cometary delivery.
• Mantle water inventory estimates ("a fraction of an ocean to several oceans").
• Pearson et al. (2014) ringwoodite inclusion — already flagged in Chapter 1; reconcile the two mentions.
• Age of the oldest oceanic crust (~180–200 Ma) and oldest continental material (Acasta gneiss ~4.03 Ga; Jack Hills zircons ~4.4 Ga as detrital evidence, not intact crust — be precise about this distinction).
• Bulk composition of continental crust (andesitic/granodioritic, ~60% SiO₂) versus oceanic (~50%).
• The waterworld claim — that Earth without buoyant continental crust would have no exposed land given its water inventory.
• Relative contribution of continental versus seafloor weathering to CO₂ drawdown.
• The Urey reaction as commonly written, and whether the draft's schematic description is accurate.
• Archean pCO₂ constraints from palaeosols, and the methane and albedo arguments as supplementary resolutions to the faint young Sun problem.
• The geodynamo section. Venus's lack of an intrinsic field and retention of its atmosphere; induced magnetospheres; the polar-outflow argument that a field may increase loss; MAVEN results at Mars. This section makes a revisionist claim against the textbook account and must be sourced carefully — it is the one place in the chapter arguing against the consensus a general reader will have absorbed.