# How Volcanoes Work: Following the Pressure From Melt to Vent

URL: https://sciencestruck.com/geology/volcanoes
Category: Earth Science
Published: 2026-09-02T12:45:27
Updated: 2026-09-02T12:45:27
Image: https://sciencestruck.com/_astro/volcanoes.DzXKmG0f_1HDoj6.webp
A volcano erupts for the same reason a shaken bottle of soda sprays the ceiling. Pressure that was holding gas dissolved in a liquid drops, the gas comes out of solution as bubbles, the bubbles expand, and the whole frothing mess is pushed out through the only opening available. Swap the soda for molten rock at around 1,000 °C, and swap the bottle cap for a few kilometres of solid crust, and you have the mechanism in one sentence.

The analogy earns its keep and then it fails, and the place it fails is worth knowing. Nobody shakes a volcano. The gas in magma comes out of solution because the magma _rises_, and rising means the pressure squeezing it drops. The bottle opens itself, slowly, from below.

One more thing before the machinery. Most people picture a volcano as a mountain sitting on an underground lake of lava, a fiery basement waiting for a crack to spill through. Drill anywhere on Earth and you will not hit that lake. The rock beneath your feet, all the way down through the mantle, is solid. Molten rock is rare, local, and temporary, and every eruption starts with the question of how any of it got molten in the first place.

## What actually makes a volcano erupt?

Three things have to line up. Rock has to melt somewhere. The melt has to be buoyant enough to climb. And a path to the surface has to open faster than the surrounding rock can squeeze it shut.

The third condition is the one that turns a slow geological process into an event. Magma sitting in storage a few kilometres down is under enormous confining pressure from the weight of rock above it, and that pressure keeps its dissolved gases (mostly water, with carbon dioxide and sulfur dioxide behind it) invisibly in solution, the way carbon dioxide stays invisible in an unopened bottle. Start that magma moving upward and the confining pressure falls. Bubbles nucleate. Bubbles take up far more room than the dissolved gas did, so the magma expands, which makes it less dense, which makes it more buoyant, which makes it rise faster, which drops the pressure further. That feedback loop is the eruption.

What kicks the loop off varies. A fresh batch of hotter magma can arrive from below and reheat a stalled reservoir. A cooling magma can crystallize out its solid minerals, concentrating the remaining gas in the shrinking pool of liquid until it forces its own way out, an unglamorous process geologists call second boiling. Or the roof simply comes off. On 18 May 1980, an earthquake dropped the entire bulging north flank of Mount St. [Helens](https://sciencestruck.com/science-facts/facts-about-mount-st-helens) off the mountain in the largest landslide in recorded history, and the magma that had been pushing that bulge outward found itself with several hundred metres less rock on top of it than it had a minute earlier. It flashed to gas sideways. Anyone who has ever watched a warm bottle uncapped too fast recognizes the sequence, at a scale that flattened forest as far as thirty kilometres from the vent.

## Where does magma come from in the first place?

Not from a molten layer. The mantle is hot enough to flow over geological time, creeping centimetres a year like a glacier of rock, but it is solid, because the crushing pressure at depth raises rock’s melting point faster than the heat down there can meet it. Melting happens only where something specific changes, and there are three somethings.

**Decompression.** Take hot mantle rock and move it upward without cooling it, and its melting point falls as the pressure falls. Somewhere on the way up it crosses its own melting temperature and begins to melt, without anyone adding a single joule of heat. This is what happens beneath mid-ocean ridges, where two plates pull apart and mantle rises to fill the gap along a ridge system that runs for tens of thousands of kilometres, almost all of it underwater and almost none of it ever watched.

**Water.** Where one plate dives beneath another at a subduction zone, it carries seawater down with it, locked into hydrated minerals in the crust and the sediment on top. [Around 100 kilometres down](https://www.nature.com/articles/s41598-024-78193-w), those minerals break down and release their water into the hot mantle above. Water lowers the melting point of rock, sharply, the way road salt lowers the freezing point of water. (The analogy is honest about direction and dishonest about mechanism, so retire it once it has done its job.) Add water to the mantle wedge above a subducting slab and it melts at a temperature it had been sitting at comfortably for millions of years, with no rise in heat required, and it melts at a fairly consistent depth because that is where the hydrated minerals give up their water. That consistency is why the volcanoes of the Ring of Fire sit in a neat line a fixed distance inland from each trench.

**Heat from below.** Hotspots, like the one under Hawaii, appear to be fed by unusually hot mantle rising from far deeper than ordinary ridge circulation, delivering heat directly. And once magma of any origin parks in the crust, it can cook the surrounding rock enough to melt it too, dissolving continental crust into the melt and changing its chemistry in ways that matter enormously later.

In every case the melting is partial. A few percent of the rock turns to liquid and the rest stays solid, and those droplets gather along mineral grain boundaries into films thin enough to see only under a microscope. Every volcano on Earth starts as a wet-looking sheen on the edges of crystals.

## Why does magma rise instead of just sitting there?

Because it is lighter than its surroundings, and rock, given a few thousand years, has no more ability to hold a buoyant fluid down than syrup has to hold an air bubble.

The numbers are the argument. Mantle peridotite has a density of roughly 3.3 grams per cubic centimetre. Melt it and you get a liquid closer to 2.6 or 2.8. That difference is small in absolute terms and decisive in effect: the melt is squeezed out of its host rock and pushed upward, gathering into ever larger volumes as it goes.

The syrup analogy has a hard limit, though, and it is the interesting part. A bubble rises through syrup by pushing the syrup aside. Magma cannot push crust aside, because crust in the upper ten or so kilometres is brittle. So the magma cracks it. Pressurized melt wedges open a fracture, flows into it, pressurizes the tip, and cracks it further, propagating a vertical sheet of magma called a dike upward like a splitting maul driven by its own weight. Dikes leave scars: drive through eroded volcanic country and you will see them standing as dark walls where the surrounding softer rock has worn away.

Eventually the magma reaches a depth where its density matches the rock around it and the free ride ends. That is where reservoirs form, and here the classroom diagram has been lying to all of us for a century. A magma chamber is not a cavern of red liquid. It is a zone of hot rock that is mostly crystals with melt in the pore spaces between them, a mush with the consistency, at its stiffest, of very wet concrete. Eruptible magma is the minority phase, extracted from the mush and collected into pockets. That is one reason a volcano can sit quietly on top of a large, well-imaged reservoir for centuries: most of what is down there cannot flow anywhere.

## What’s the real difference between magma and lava?

Molten rock underground is magma. The same molten rock, once it has reached the surface, is lava. One substance, two addresses: nothing about the chemistry has to change for the name to change, and the ground surface is the entire boundary condition. A geophysicist imaging a reservoir six kilometres down and a photographer standing beside a fountain an hour later can be describing the same batch of liquid rock under two different words.

Something about the chemistry does change, though, and this is the detail worth carrying away. Magma contains dissolved gas. Lava, by the time you can see it, has lost most of that gas to the atmosphere. So lava is degassed magma, which is why cooled lava rock is full of frozen bubble holes (vesicles) and why the pumice from an explosive eruption is so full of them it floats. I have watched a rafted pumice sample bob in a beaker for an entire class period and never met a student who was bored by it.

A few neighbouring words get tangled with these two, and untangling them costs one paragraph:

-   **Tephra:** any fragment of any size thrown into the air by an eruption, from dust to boulders. It is a size-blind catch-all.
-   **Pyroclastic material:** the same stuff, named for how it [formed](https://sciencestruck.com/geology/how-are-underwater-volcanoes-formed) (the Greek roots mean fire-broken). Volcanic ash is the fraction under 2 mm, lapilli run from 2 to 64 mm, and anything bigger is a block or a bomb depending on whether it was solid or still molten when it flew.
-   **Volcanic ash:** not ash in the campfire sense. Nothing burned. These are jagged shards of volcanic glass and pulverized crystal, which is exactly why they behave so badly, as the last section explains.
-   **Basalt, andesite, rhyolite:** the rocks lava freezes into, named by chemistry. Their slow-cooled underground twins have different names again (rhyolite’s is granite), which is how a kitchen countertop can be honestly described as magma that never made it out.

## Why do some volcanoes erupt gently and others explode?

One variable does most of the work: how much silicon dioxide the magma contains. Silica controls viscosity, viscosity controls whether gas bubbles can escape, and gas that cannot escape is what turns an eruption from a flow into a blast.

The chemistry is unusually satisfying here. Silica in a melt exists as silicon-oxygen tetrahedra, and those units link up by sharing oxygen atoms into chains and sheets. More silica means more linking, and more linking means a stiffer liquid. Basaltic magma runs around 50 percent silica and pours; rhyolitic magma runs around 70 percent and behaves more like cold tar than anything most people would call liquid. The viscosity gap between them spans several orders of magnitude, not a factor of two or three, and if a comparison you read anywhere ever looks reasonable, check your units before you believe it. Temperature compounds the difference, because low-silica magmas also erupt hotter, often above 1,100 °C, while silica-rich magmas erupt several hundred degrees cooler.

Now put bubbles into each. In runny basalt, gas bubbles rise, coalesce, and pop out of the top continuously. The eruption is effusive: lava fountains, lava flows, the glowing rivers of Hawaii that people photograph from a respectful distance. In stiff rhyolite, bubbles nucleate and cannot go anywhere. They expand in place against a liquid that refuses to yield, pressure builds until the magma itself shatters into a spray of glass fragments and gas, and the whole column is blasted out of the vent at speeds that carry ash into the stratosphere. That shattering has a name, fragmentation, and it is the actual moment an eruption becomes explosive.

Two other factors ride along. How much water the magma dissolved in the first place sets how much gas is available (subduction magmas, fed by that wet slab, carry the most). And water on the outside counts too: magma meeting groundwater, seawater, or meltwater flashes it to steam and fragments explosively regardless of silica content.

Geologists score the result on the Volcanic Explosivity Index, a scale from 0 to 8 based mainly on how much material an eruption threw out. It is logarithmic, so each step up is roughly a tenfold increase, which means a VEI 5 like Mount St. Helens in 1980 and a VEI 6 are not neighbours in any everyday sense. Readers after the fuller account of that day will find it in the facts about Mount St. Helens, which go well beyond the mechanism.

## What are the parts of a volcano, and what does each one do?

The parts of a volcano are mostly plumbing, and the diagram on the classroom wall labels that plumbing without ever saying what any of it does. Here it is from the bottom up, with the job each piece performs rather than just the place it sits.

-   **Magma reservoir (magma chamber):** the storage zone in the crust, commonly a few kilometres to a dozen or more below the summit. Mostly crystal mush with pockets of eruptible melt. Its inflation and deflation is what tilts the ground surface and gives monitoring instruments something to measure.
-   **Conduit:** the pipe from reservoir to surface. It may be a genuine cylindrical throat kept open by repeated use, or a dike, a vertical crack the magma opens on its way up. Its width and roughness control how fast magma can move, which feeds directly back into how explosively it arrives.
-   **Main vent:** the opening where magma reaches the air. Everything above it is scenery built by everything that came out of it.
-   **Crater:** the bowl at the summit, formed by explosions clearing the throat and by minor collapse around the vent. Usually a few hundred metres to a kilometre across.
-   **Flank vents and fissures:** side openings where magma finds an easier path than the main conduit. They build the small cones scattered on a big volcano’s slopes, and they can put lava on the ground kilometres from the summit anyone is watching.
-   **Sills and dikes:** magma frozen in the cracks it made, horizontal and vertical respectively. They are the fossil record of past attempts that never reached the surface, and most attempts never do.
-   **Caldera:** not a big crater. A caldera forms when an eruption empties enough of the reservoir that the roof above it founders and drops in, which is why calderas are measured in kilometres and craters are not.
-   **Layers of the cone:** alternating lava flows and tephra beds on a stratovolcano, essentially a stack of past eruptions in date order. Read a road cut through one and you are reading the mountain’s own history, youngest on top.

Two corrections to the poster on the wall, since we are here. The chamber is drawn as a cavern, and it isn’t one. And the vertical scale is a fiction: a reservoir five kilometres down under a cone two kilometres tall gets drawn as a balloon the size of the mountain, which quietly teaches every student the wrong proportions. Sketch it to scale once and the whole thing becomes a very thin straw over a very deep, very diffuse source.

## What are the main types of volcanoes?

Shape follows chemistry. Each of the four classic forms is the direct consequence of what came out of the vent and how fast.

**Shield volcanoes** are built by low-silica basalt that runs a long way before it freezes, so successive flows spread wide instead of piling up. The result is a broad dome with slopes of only a few degrees, deceptively gentle and genuinely enormous. Mauna Loa is the standard example, and standing on it feels like standing on a slightly tilted plain.

**Stratovolcanoes** (also called composite volcanoes) are the postcard shape: steep, symmetrical, snow-capped. They are built at subduction zones from viscous, gas-rich magma that alternates between explosive eruptions laying down tephra and slower flows that armour it in place. That alternation is what lets the slopes stay steep, and the layered internal structure is also what makes them prone to collapse. Vesuvius, Fuji and St. Helens are all this type. Vesuvius is the one whose deposits let archaeologists reconstruct an eruption hour by hour, and the [facts about Mount Vesuvius](https://sciencestruck.com/science-facts/facts-about-mount-vesuvius) are worth reading for that record alone.

**Cinder cones** are the small ones, often built in a single eruptive episode when gassy lava fountains fall back around the vent as loose scoria. They rarely exceed a few hundred metres and they are frequently one-and-done. Parícutin famously grew out of a Mexican cornfield in 1943 while the farmer watched.

**Calderas** are what remains when a very large eruption empties its reservoir and the ground above collapses into the space. Crater Lake in Oregon is one filled with water; Yellowstone is one large enough that most visitors never realize they are standing inside it.

Cutting across all four, the actual mountain-building process (why one becomes a peak and another a lake) is its own subject, covered in how volcanic mountains are formed.

## Where on Earth do volcanoes actually form?

Where the melting conditions from the second section exist, which turns out to mean three kinds of places, and it maps onto plate tectonics almost perfectly.

**Divergent boundaries**, where plates pull apart and mantle decompresses on the way up. Most of this is the mid-ocean ridge system, out of sight under two or three kilometres of seawater. Iceland is the exception that made the whole idea legible, because there the ridge runs above sea level. The East African Rift is a continent doing the same thing more slowly.

**Convergent boundaries**, where a plate subducts and its water triggers melting above. This is the Ring of Fire, the horseshoe of subduction zones around the Pacific that is commonly credited with about three-quarters of the [world](https://sciencestruck.com/geology/active-volcanoes-in-world)’s active volcanoes, running from the Andes up through the Cascades, across the Aleutians, and down through Kamchatka, [Japan](https://sciencestruck.com/geology/famous-volcanoes-in-japan), the Philippines and Indonesia. The arc is continuous in mechanism but not in behaviour, and if you want to see how unevenly that plays out country by country, the survey of active volcanoes in the world is the place to start.

**Hotspots**, which ignore plate boundaries entirely and sit in the middle of plates, apparently fed from deep in the mantle. Hawaii is the classic: as the Pacific plate slides northwest over a stationary heat source, the plate carries each finished island away and a new one starts, leaving a dated chain of volcanoes that reads like a tape measure of plate motion.

The useful negative result: transform boundaries, where plates slide past each other rather than converging or separating, get earthquakes but no volcanoes. Nothing melts along the San Andreas because nothing there decompresses or gets wet. The same three conditions that predict where volcanoes appear also predict, correctly, the thousand-kilometre stretch of California where they do not.

## Can volcanoes form underwater or under ice?

They can, and by volume of erupted rock, underwater volcanism is the main event on this planet. The buoyancy story is unchanged. What changes is what waits at the vent.

Under a couple of kilometres of seawater, the pressure of the overlying water column is high enough to keep magmatic gases largely dissolved, which suppresses the bubble expansion that drives explosions. Deep eruptions are therefore usually quiet: lava extrudes, chills instantly against cold water into a glassy skin, and inflates like a toothpaste tube behind that skin to form pillow basalt, the lumpy, tube-shaped rock that paves most of the ocean floor. Bring the same eruption into shallow water and the physics flips, because seawater flashing to steam against hot magma fragments it violently. That is why shallow submarine eruptions build ash cones and deep ones do not, and the full sequence of how underwater volcanoes are formed is a subject in its own right.

Ice adds a different complication. A subglacial eruption melts a cavity in the glacier above it, filling it with meltwater, so the vent finds itself under water it manufactured. The magma quenches into glassy fragments called hyaloclastite, and if the eruption persists long enough to build above the ice and the water, it can finish with ordinary lava flows on top, leaving the flat-topped, steep-sided mountain called a tuya. Meanwhile the meltwater has to go somewhere, and when it goes it goes all at once, as an outburst flood that can carry car-sized ice blocks across an outwash plain. Iceland’s [subglacial volcanoes](https://sciencestruck.com/geology/subglacial-volcanoes) have supplied both the vocabulary and most of the modern examples, including the 2010 eruption whose ice-fragmented ash grounded European air travel for days.

## How are volcanoes and earthquakes connected?

Two ways, and keeping them separate prevents most of the confusion.

The first is shared geography. Subduction zones make earthquakes because plates grinding past each other store and release elastic strain, and they make volcanoes because subducting slabs release water. Both effects come from the same plate motion, so both appear along the same arcs. That correlation is real and it is not causal in either direction: the earthquake did not cause the volcano and the volcano did not cause the earthquake.

The second connection is causal and local. Magma forcing its way through brittle crust breaks that crust, and breaking rock radiates seismic waves. Those are volcano-tectonic earthquakes, generally small, and when a swarm of them migrates steadily upward over hours or days, geologists are watching a dike propagate in something close to real time. There are subtler signals too: long-period events generated by pressurized fluid moving through cracks, and harmonic tremor, a continuous rhythmic vibration that often means magma or gas is flowing sustainedly rather than in fits.

The asymmetry matters. Nearly every erupting volcano produces earthquakes, while the overwhelming majority of earthquakes have nothing to do with magma. A magnitude 7 on a subduction thrust and a magnitude 1.5 under a summit crater are different phenomena that happen to share a name. The fuller version of that story, including the cases where a large tectonic quake does appear to have nudged a nearby volcano, sits in the [relationship between earthquakes](https://sciencestruck.com/geology/relationship-between-earthquakes-volcanoes) and volcanoes.

## How do scientists know an eruption is coming?

Because volcanoes are noisy before they erupt, and the noise is measurable. The idea that eruptions arrive without warning is largely a media artifact of eruptions at unmonitored volcanoes. A well-instrumented volcano almost always shows unrest first.

Four instrument families do most of the work. Seismometers catch the fracturing and tremor described above and, crucially, catch where the events are and whether they are climbing. GPS receivers and tiltmeters measure the ground deforming as a reservoir inflates, at precisions that resolve millimetres; satellite radar interferometry does the same over whole mountains without anyone hiking up them. Gas instruments sample or remotely measure the plume, watching sulfur dioxide output and the ratio of carbon dioxide to sulfur, which shifts as magma rises and degasses at different depths. And thermal cameras and simple webcams watch the surface for new hot ground, melting snow, and dome growth. In the [United States](https://sciencestruck.com/geology/active-volcanoes-in-united-states), the USGS Volcano Hazards Program runs the observatories that do this work and publishes both a ground alert level and an aviation colour code so that airlines and residents get the warning in the form each one needs.

St. Helens is still the teaching case. Earthquakes began in mid-March 1980, steam explosions followed, and by April the volcano’s north flank was pushing outward at roughly a metre and a half a day, a bulge measurable with survey instruments from a safe distance. The area was closed and evacuated, and that decision is why the death toll was measured in dozens rather than thousands. What the science did not anticipate was the direction: the blast went sideways.

Which is the honest limit to state plainly. Monitoring reliably answers “is this volcano restless” and often answers “is magma moving and where”. It is far weaker at “what day”, and unrest frequently fades without an eruption at all. Forecasts are probabilistic, they get revised, and they are worth exactly as much as the response to them. If you live near a monitored volcano, the single most useful thing in this entire article is that evacuation orders from the responsible agency are not advisory suggestions to be weighed against your own read of the mountain.

## What does a volcano actually produce, besides lava?

Lava is the photogenic hazard and, in most eruptions, the least dangerous one, because it moves slowly enough to walk away from. The things that hurt people move faster.

**Gas.** Overwhelmingly water vapour, then carbon dioxide, then sulfur dioxide. The sulfur is disproportionately important, because in the stratosphere it converts to sulfate aerosols that reflect sunlight, and a large enough eruption measurably cools the whole planet for a year or two afterward. Carbon dioxide is heavier than air and can pool invisibly in low ground near vents, which is why volcanologists carry gas monitors into hollows.

**Ash.** Not soot, not soft. Volcanic ash is angular fragments of volcanic glass, hard enough to abrade metal and insoluble in water, so it does not wash away or dissolve. It sandblasts turbine blades and melts in jet engines, which is why a single ash cloud can close continental airspace. It conducts electricity when damp and shorts out power infrastructure. And because dry ash weighs a lot and wet ash weighs far more, ashfall roof collapse kills people who were nowhere near the eruption. During ashfall the standing public-health guidance is to stay inside, keep it out of your lungs and eyes, and follow the instructions of local authorities. On the other side of the ledger, ash weathers into some of the most fertile agricultural soil on Earth and has a long industrial history in cement, which is the subject of the uses of volcanic ash.

**Pyroclastic flows.** Ground-hugging avalanches of hot gas, ash and rock fragments that travel at highway speeds and hundreds of degrees. They are the deadliest volcanic phenomenon by a wide margin, they follow valleys but can overtop ridges, and they are what buried Pompeii and Herculaneum.

**Lahars.** Volcanic mudflows: ash and debris mixed with water from melting snow, a crater lake, or heavy rain, moving down river valleys with the consistency of wet concrete and the density to carry bridges. They can occur years after an eruption, whenever rain remobilizes loose deposits. The 1985 eruption of Nevado del Ruiz in Colombia was small, and the lahar it triggered killed more than 20,000 people in the town of Armero, tens of kilometres away.

Every one of those is the same pressure system from the first section, expressed differently once it reaches air.

## Where to look next

Start with the shape of the nearest hill. A broad low dome and a steep symmetrical cone were built by different magmas from different sources under different plates, and the parts list above will let you guess which before you look anything up. Then go find a road cut through volcanic country and count the layers; each one is a separate eruption, in order, with the youngest on top.

And if there is a kid in the house who wants to build one, the standard instructions for how to make a volcano for the science fair run on baking soda and vinegar, which models the gas-driven part honestly and safely. One hard line, worth stating without hedging: nothing sealed, nothing pressurized, and nothing lit belongs in a model volcano. The real ones already demonstrated what happens when gas cannot get out.