A shell broken into plates
Earth's outermost shell, the lithosphere, is not one continuous rind — it is roughly fifteen rigid plates, each 60 to 150 km thick, floating on the slow-flowing asthenosphere beneath. The plates are not passive rafts: they are pushed and pulled by the very heat engine that made them, and where two plates meet decides almost everything interesting about a region's geology — its mountains, its earthquakes, its volcanoes, even the shape of its coastline.
The theory took decades to accept. Alfred Wegener proposed continental drift in 1912 from the jigsaw-fit of the continents and matching fossils across oceans, but he had no mechanism and was largely dismissed. Proof arrived in the 1960s from the ocean floor: symmetric stripes of alternating magnetic polarity on either side of the mid-Atlantic ridge, laid down as new crust cooled through the Curie point and locked in whichever direction Earth's field pointed at the time. That seafloor magnetic striping is still the cleanest evidence that plates move and new crust is born at ridges.
Three kinds of boundary
At a divergent boundary plates pull apart and magma rises to fill the gap, freezing into new oceanic crust — the mid-ocean ridges, and on land the East African Rift, which is slowly tearing a new ocean basin into existence. At a convergent boundary plates collide. If one is oceanic it is denser and dives beneath the other in subduction, dragging water into the mantle, triggering melting and building a volcanic arc (the Andes, the Cascades); if both are continental, neither can subduct and the crust simply crumples upward, which is how the Himalayas got their height. At a transform boundary plates slide past each other sideways with no crust created or destroyed — the San Andreas Fault is the textbook case, and because the two sides can lock and store strain for centuries before slipping, transform faults are prolific earthquake generators.
divergent → | <--- ---> | new crust, shallow quakes, rift valleys convergent → | ---> <--- | subduction OR mountain building, deep quakes transform → | ^^^ vvv | strike-slip, shallow but often large quakes
What actually drives fifteen plates of rock
For decades textbooks credited mantle convection alone — hot rock rising, cooling, sinking in a slow-motion pot of boiling soup, dragging the plates on top like scum on a simmering pan. Modern geodynamics gives most of the credit to forces acting at the plate edges instead. Slab pull, the weight of a cold, dense subducting slab sinking under its own gravity, is now thought to supply most of the driving force — plates with long subducting edges (like the Pacific plate) move fastest. Ridge push adds a smaller shove: newly formed crust at a ridge is buoyant and elevated, and it slides gravitationally away from that topographic high as it ages, cools and thickens. Basal drag from mantle flow beneath probably contributes too, but it is now seen as a secondary effect layered on top of slab pull and ridge push rather than the primary engine.
Earthquakes as a stick-slip process
Rock at a locked plate boundary behaves elastically: stress builds for years to centuries while the fault stays stuck by friction, then releases suddenly when the stress overcomes that friction — stick-slip motion, the same mechanism that makes a violin string sing. The energy released scales with the moment magnitude, M = (2/3)·log₁₀(M₀) − 10.7, where seismic moment M₀ is rupture area times average slip times crustal rigidity; because it is a base-10 logarithm of energy, each whole step up the scale represents roughly 32 times more released energy, not a linear jump. Subduction zones produce the largest earthquakes of all — megathrust events like Chile 1960 (M9.5) — because the contact area between the two plates over which slip can occur is enormous.
Reading the boundary from the surface
Each boundary type leaves a distinctive fingerprint you can read on a map. Divergent zones show shallow, low-magnitude earthquakes strung along a linear ridge or rift, plus active volcanism with basaltic, low-silica lava. Convergent subduction zones show a defining pattern first mapped by Wadati and Benioff: earthquake depths increase steadily inland, away from the trench, tracing the plane of the descending slab down to 700 km, paired with an arc of explosive, high-silica volcanoes roughly 100–200 km behind the trench. Transform zones show shallow earthquakes clustered in a narrow band with essentially no associated volcanism, because no material is being created or subducted to melt.
Frequently asked questions
What is the actual evidence that plates move, not just theory?
The strongest single piece of evidence is symmetric magnetic striping on the sea floor either side of mid-ocean ridges, matching Earth's recorded reversals of its magnetic field, plus direct GPS measurement today showing plates drifting a few centimetres a year — roughly the rate your fingernails grow.
Why do subduction zones produce the biggest earthquakes?
The two plates lock together over an enormous contact area along a shallow-dipping fault. When that locked patch finally slips, the rupture can spread across hundreds of kilometres at once, releasing far more energy than the smaller, steeper faults typical of transform or divergent boundaries.
Is mantle convection still considered the driver of plate motion?
It contributes, but current geodynamic models credit most of the driving force to forces acting at the plate edges — slab pull, where a dense subducting slab sinks under its own weight, and ridge push, where cooling crust slides away from an elevated ridge — with mantle flow beneath playing a supporting role rather than the primary engine.
Try it live
Everything above runs in your browser — open Plate Tectonics and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.
▶ Open Plate Tectonics simulation