Seafloor Spreading Calculator, Magnetic Stripe Explorer and Ocean Depth Model

How to use the three tools on this page

Start with the magnetic stripe explorer. The bar at the top is a slice of ocean floor with the ridge axis down the middle, dark for crust magnetised the way the field points today and white for crust magnetised the other way. Drag the rate slider and watch the whole pattern stretch or compress. That is the single most useful thing on this page, because it shows you that a fast ridge and a slow ridge record exactly the same sequence of reversals, just spread over different distances. Click any stripe and you get its chron name, its age, how wide it is at the rate you have set, and the spreading rate it would imply if you measured it in the field.

The rate calculator is where you do the actual arithmetic. Put in a distance from the axis and the age of the anomaly you measured there, and it returns a half rate and a full rate with the working shown line by line. You can also pick a chron from the dropdown and it fills the age for you. The input people get wrong is not the distance, it is which rate they think they have produced, and the tool spells that out every time rather than leaving it implied.

The depth and age model is the third tab and it is the one worth spending time on. It plots three competing models of how deep the seafloor should sit at a given age, against twenty real measured depths. For young crust all three agree. Past about eighty million years they separate, and the observations pick a side. You can toggle each model on and off and click anywhere on the chart to read all three at that age. This is not a case of one model being wrong and one being right, it is a case of a model working beautifully inside its range and failing outside it, which is a more useful thing to understand.

Two supporting tables sit behind those. The spreading centre database holds forty eight ridges with their measured rates, their rate class and the shape of their axis, and clicking a row tells you where its magnetic stripes would fall. The polarity timescale lists all one hundred and eighty eight polarity intervals back to eighty three million years, with the width each one would make on a Mid Atlantic type ridge, which turns out to be the reason some reversals are missing from slow spreading records.

What seafloor spreading actually is

Seafloor spreading is the process that makes new oceanic crust. Mantle rises beneath a mid ocean ridge, melts as the pressure on it drops, and the melt collects and erupts along a narrow line. It freezes, the two plates pull apart, the frozen rock splits down the middle, and half of it goes with each plate. Then more melt arrives and the cycle repeats.

That is the whole mechanism and it fits in a paragraph. What makes it worth a longer explanation is everything it implies.

It implies the ocean floor is young. Nothing on the seafloor is older than about one hundred and eighty million years, which is less than four percent of Earth history, while continental rocks reach beyond four billion years. It implies the ocean floor is symmetric, because the same rock is being divided between two plates. It implies the ocean floor gets systematically older and deeper away from the ridge. It implies that if the field reverses while all this is going on, the record of that reversal is written into the rock twice, once on each side, at the same distance from the axis.

Every one of those implications was checked in the nineteen fifties and sixties, and every one of them held. That is why plate tectonics went from a fringe idea to the organising theory of the entire subject in roughly a decade.

The other thing to be clear about from the start is that spreading is not pushing. The ridge is not a place where something shoves the plates apart. Most of the force comes from the far end, where cold dense oceanic lithosphere sinks into the mantle and drags the rest of the plate behind it. Ridge push is real but it is a secondary effect, a gravitational sliding force from the raised ridge, and it accounts for perhaps ten to twenty percent of the driving force. The ridge is largely a passive gap that fills because the plates are being pulled away from it, which is why ridges can migrate and why a ridge can be subducted entirely.

The evidence that settled it

The ages of the ocean floor

Drill anywhere on the ocean floor and date the oldest sediment sitting directly on the basalt. Do it repeatedly across a ridge flank and the ages come out in order, youngest at the axis, older away from it, at a rate that matches the magnetic anomalies. The Deep Sea Drilling Project did exactly this from 1968 onward and it was decisive, because there is no way to fake a systematic age progression across thousands of kilometres.

The oldest ocean floor still attached to a plate is roughly one hundred and eighty to two hundred million years, in the western Pacific and the eastern Mediterranean. Nothing older survives because it has all been subducted. Compare that with the continents, where four billion year old rocks exist, and you have the sharpest asymmetry in Earth science: the oceans are constantly recycled and the continents are essentially permanent.

Sediment thickness

Sediment accumulates on the seafloor at a slow but reasonably steady rate, of order a few metres per million years in the open ocean. If the ocean floor were ancient and static, the sediment pile would be tens of kilometres thick everywhere. It is not. It is nearly absent at the ridge axis and thickens steadily away from it, in proportion to the age of the crust beneath.

This was noticed before spreading was proposed and it was genuinely puzzling. Once you accept that the crust itself is young and gets older outward, it stops being puzzling and becomes a prediction that can be tested with a seismic profile.

Heat flow

Measure the heat coming out of the seafloor and it is highest at the ridge and falls away with distance, roughly as one over the square root of age. That is exactly what a cooling plate should do. It is also the same physics that makes the seafloor deepen with age, which means depth and heat flow are two independent measurements of the same cooling process, and they agree.

The magnetic stripes, which were the clincher

Tow a magnetometer behind a ship across a mid ocean ridge and the total field strength goes up and down in a pattern of long and short intervals. Two things about that pattern are impossible to explain any other way.

First, it is symmetric about the ridge axis. The sequence of wide and narrow bands to the east mirrors the sequence to the west, with a precision that is obvious by eye. Second, the same sequence appears on ridges in different oceans, at different rates, but always in the same order. Stretch one profile and it lies on top of the other.

Symmetry plus a shared sequence means both sides are recording the same global signal, and the only global signal available is the Earth magnetic field. The rock is a tape and the ridge is the recording head.

Who worked it out, and why it took so long

Wegener had the observation and not the mechanism

Alfred Wegener published continental drift in 1912 and assembled a genuinely strong case: the fit of the coastlines, matching fossil species across oceans, matching rock sequences, and glacial deposits in places that are now tropical. The evidence was good. The mechanism was not. He proposed that continents ploughed through oceanic crust, and geophysicists correctly pointed out that this is impossible, because oceanic crust is stronger than continental crust and the driving forces he invoked were orders of magnitude too small.

The result was that a correct observation was rejected for forty years because it came attached to an incorrect explanation. It is worth remembering that the objection was not stupid. Given what was known in 1930, Wegener's mechanism really was unworkable.

Hess supplied the mechanism

Harry Hess had spent the Second World War as a naval officer with an echo sounder running, and had accumulated a great deal of unusual bathymetry. In 1962 he published a paper he described as an essay in geopoetry, proposing that the ocean floor was not permanent at all. Mantle convection brought material up under the ridges, it spread outward, and it descended again at the trenches. The continents did not plough through the ocean floor, they rode on it.

That solved Wegener's problem completely. The continents move because the entire plate moves, and the force comes from the mantle rather than from the continent itself.

Vine, Matthews and Morley turned it into a test

Hess had a mechanism but not yet a decisive test. The test came in 1963 from Fred Vine and Drummond Matthews at Cambridge, and independently from Lawrence Morley in Canada, whose paper was rejected by two journals before the idea became accepted.

Their reasoning combined two things that were each already known but had not been put together. Seafloor spreading, from Hess. And geomagnetic reversals, which had been documented in lava flows on land since the nineteen twenties and were becoming firmly dated by the early sixties.

Put them together and you get a prediction. If new crust forms continuously at the axis and takes on the field direction of the moment it cools, and if the field reverses periodically, then the ocean floor must be striped, and the stripes must be symmetric. That is a very specific prediction, it is easy to falsify, and when the surveys came in it was correct.

The Morley story is worth knowing. One rejection reportedly said the idea was the sort of thing you would discuss at a cocktail party but not publish. The lesson is not that reviewers are foolish, it is that a correct idea arriving before its supporting data can look indistinguishable from a wrong one.

How the magnetic record is actually written

This is the part that is usually skipped, and skipping it is why people end up thinking the stripes are somehow magnets stuck to the seafloor.

Basalt erupting at a ridge contains magnetite and titanomagnetite, a few percent by volume. While the rock is hot, those grains have no stable magnetisation, because thermal energy randomises the magnetic moments faster than they can align. As the rock cools it passes through the Curie temperature, about 580 degrees for magnetite, and below that the moments can hold a direction. What direction they hold is set by the field that happens to be present at that moment, which is the Earth field.

That is thermoremanent magnetisation. It is locked in as the rock cools and it is remarkably stable, surviving for hundreds of millions of years unless the rock is reheated above the Curie point again.

Three consequences follow, and each answers a question people actually ask.

The stripes are symmetric because the rock that acquires a given magnetisation is a single body of new crust straddling the axis, and it is then split down the middle. Each side keeps half. There is no mechanism that would make the two halves different, so the pattern is a mirror.

The stripes are not literally black and white bands you could see. The crust looks the same either way. What varies is the direction of a weak remanent magnetisation carried by a small fraction of the minerals, detectable only as a change of a few hundred nanotesla in the total field, which is well under one percent of the Earth field.

The magnetometer reading goes up over normally magnetised crust and down over reversed crust because the remanent field either adds to or subtracts from the present day field. That is why the trace in the diagram above is drawn as a wiggle rather than a square wave, and why picking anomaly boundaries from a real profile involves genuine judgment rather than reading off edges.

Why the anomaly is not a sharp edge

The source of the anomaly is a layer of magnetised rock a few kilometres thick sitting several kilometres below the ship. A magnetic field from a buried body is always smoothed relative to the body itself, and the smoothing gets worse as the source gets deeper. A stripe boundary that is geologically sharp arrives at the sea surface as a gradual slope.

This matters when you are measuring distances to compute a rate. The uncertainty in picking an anomaly boundary on a real profile is typically one to a few kilometres, which on a slow ridge is a substantial fraction of a young stripe. It is the main reason published spreading rates carry uncertainties of one to three millimetres per year rather than being quoted to two decimal places.

Working out a spreading rate

Here is the calculation, done in full, using the default values in the calculator above so you can follow along and get the same numbers.

You are on a ship crossing the Mid Atlantic Ridge. You have a magnetic profile and you have identified the outer edge of the central normally magnetised band, which is the end of the Brunhes chron, the boundary where the field last flipped. On the GTS2020 timescale that boundary is 0.773 million years old. You measure its distance from the ridge axis as 19.3 kilometres.

The half rate is the distance one plate has travelled away from the axis divided by the time it took.

19.3 kilometres divided by 0.773 million years is 24.97 kilometres per million year.

Now the unit conversion, which is the part that surprises people because there is nothing to do. One kilometre per million years is one million millimetres divided by one million years, which is one millimetre per year exactly. The two units are numerically identical. So 24.97 kilometres per million years is 24.97 millimetres per year, with no arithmetic at all.

That is a half rate: how fast one plate is moving away from the axis. The rate normally published for a ridge is the full rate, the speed at which the two plates separate from each other, and both plates are moving, so

2 times 24.97 equals 49.94 millimetres per year.

Roughly fifty millimetres per year, which puts this ridge in the slow class, and that is about right for the Mid Atlantic.

Now do something useful with it. If the rate has been steady, crust ten million years old should sit 249.7 kilometres from the axis. Go and find the anomaly you expect there. If it is not where you predicted, either the rate has changed over time, which is common, or you have misidentified an anomaly, which is more common.

Half rate against full rate, which is the error that matters

If you take one thing from this page, take this one, because it is the mistake that turns up in student work, in lecture notes and occasionally in published figures.

A half rate is measured from the axis outward on one side. A full rate is the separation of the two plates and is twice the half rate. It is also what you get directly if you measure between two matching anomalies on opposite sides of the axis, because that distance already contains both plates.

The failure mode is measuring across the axis, correctly obtaining a full rate, and then doubling it out of habit. That gives an answer twice too large, and because the classes are broad it will often still sound plausible. A slow ridge becomes a fast one and the entire interpretation of the axial morphology goes with it.

Two habits protect you. Always write down which side of the axis your measurement came from. And sanity check the answer against the classes: nothing on Earth spreads faster than about 150 millimetres per year full rate, so an answer of 200 is telling you something is wrong before you have even checked the arithmetic.

The polarity timescale, and which version you are using

The stripes are only useful if you know the ages of the reversals, and that calibration is its own field of work.

How the ages are established

The young end is dated directly. Lava flows on land are dated radiometrically and their magnetisation measured, which fixes the reversals of the last few million years. Beyond that, the method changes to astronomical tuning: sediment cores record cycles driven by changes in Earth orbit, those cycles have known periods, and counting them gives ages far more precise than radiometric dating alone.

The tool on this page uses GTS2020, the current astronomically tuned standard. It runs from the present back to eighty two point eight seven five million years, which is the end of the Cretaceous Normal Superchron, a stretch of about forty million years during which the field did not reverse at all. Beyond that boundary there are no stripes to count, which is why every magnetic anomaly timescale stops there.

The older scale you will still meet

A great deal of published work, and most of the global seafloor age grids, use Cande and Kent 1995 rather than GTS2020. The differences are small but not always negligible, and they are not in a consistent direction.

The Brunhes boundary is 0.773 on GTS2020 and 0.780 on Cande and Kent, a difference under one percent. The largest divergence anywhere in the Cenozoic is in the Paleocene, where the base of chron C25r is 57.554 on GTS2020 and 58.959 on Cande and Kent, a gap of 1.405 million years. In the early Miocene the sign flips and GTS2020 is the younger of the two.

The practical consequence is that a spreading rate computed across chron C6 comes out several percent different depending on which scale you used, and if you are comparing your rate with a published one you need to know which scale that publication used. Neither is wrong. They are different calibrations of the same reversal sequence.

Reading the chron names

The naming looks arbitrary and mostly is, because it grew by accretion. Chrons are numbered outward from the present, C1, C2, C3 and so on. A trailing n means normal polarity and r means reversed. Where a chron was later found to contain shorter intervals, decimal subdivisions were added, giving names like C2An.1n, which reads as the first normal subchron within the A subdivision of chron 2.

The long intervals of the last few million years also carry traditional names from before the numbering existed, and those names are still what people say out loud: Brunhes for the current normal chron, Matuyama for the reversed interval before it, Gauss before that, Gilbert before that. The short events inside them have their own names, Jaramillo, Olduvai, Kaena, Mammoth.

One quirk worth knowing so it does not worry you: there is no C14. The numbering jumps from C13 to C15. It is an artefact of an early misidentification that was corrected without renumbering everything downstream. Nothing is missing from the record.

Why slow ridges lose reversals

Open the polarity timescale tab and sort by duration. The Jaramillo subchron lasted eighty thousand years. On a ridge spreading at twenty five millimetres per year half rate, that produces a stripe two kilometres wide.

Two kilometres is at or below what a surface magnetometer can resolve through several kilometres of water, given the smoothing described earlier. So on a slow ridge the Jaramillo may simply not appear in the profile. On the East Pacific Rise at seventy five millimetres per year the same event makes a six kilometre stripe and shows up clearly.

This is a resolution limit, not a difference in the field. The reversal happened globally. Whether your survey can see it depends on how fast the ridge you are looking at was writing at the time, which is a nice demonstration that the record and the recording medium are separate things.

How fast ridges actually spread

Rates span more than an order of magnitude, and the classes are conventional divisions of a continuum, defined on full rate.

Ultraslow is below twenty millimetres per year. Slow is twenty to fifty five. Intermediate is fifty five to eighty. Fast is eighty to one hundred and forty. Superfast is above one hundred and forty.

The southern East Pacific Rise reaches about one hundred and fifty millimetres per year full rate, the fastest on the planet, which is roughly the speed at which a fingernail grows. The Mid Atlantic Ridge runs around twenty to twenty five over most of its length. Gakkel Ridge under the Arctic ice manages about twelve, and parts of the Southwest Indian Ridge are slower still.

The database on this page holds forty eight spreading centres and you can sort it by rate. What is worth noticing is that the fast ones are almost all in the Pacific and the slow ones almost all in the Atlantic, Arctic and Indian oceans. That is not a coincidence. The Pacific is ringed by subduction zones actively pulling its plates apart from the far side, and the Atlantic is not, which returns to the earlier point that ridges are passive.

Ultraslow ridges are a different animal

Below about twenty millimetres per year the process changes rather than just slowing down. Melt supply drops far enough that the usual layered oceanic crust does not form continuously. Long segments of Gakkel and the Southwest Indian Ridge have essentially no volcanic crust at all, and mantle peridotite is exposed directly on the seafloor, altered to serpentinite by reaction with seawater.

That means the standard picture of oceanic crust, a few kilometres of basalt over gabbro over mantle, is not universal. On an ultraslow ridge the crustal thickness can approach zero and you are effectively looking at exhumed mantle with a thin discontinuous cap. Since ultraslow ridges make up a good fraction of the global ridge length, this is not an exotic corner case.

Why the shape of the axis follows the rate

Look at a cross section of a fast ridge and a slow ridge and they do not look like the same feature.

A fast ridge such as the East Pacific Rise has an axial high: a smooth broad swell a few hundred metres tall, with no valley at the crest. A slow ridge such as the Mid Atlantic has a median valley, a rift one to three kilometres deep and tens of kilometres wide, with rugged walls.

The reason is melt supply and thermal state rather than speed as such. A fast ridge is continuously supplied and holds a steady magma lens a kilometre or two below the axis, keeping the axial lithosphere thin, hot and weak. It cannot support a deep valley because it has no strength, so the topography is built by volcanic construction and it stands up.

A slow ridge is cooler. There is no steady magma chamber, only episodic bodies. The axial lithosphere is thick and strong enough to fault, so extension is taken up on large normal faults and the axis drops into a rift valley. The rough flanking topography is the record of those faults being carried away from the axis.

The transition sits near sixty to eighty millimetres per year full rate, which is why the intermediate class is genuinely transitional and contains ridges of both kinds.

The exception that proves the mechanism

Reykjanes Ridge, southwest of Iceland, spreads at about twenty millimetres per year. By rate it is slow, and it should have a median valley. It does not. For hundreds of kilometres it carries an axial high like a fast ridge, and the valley only reappears once you get far enough south.

The reason is the Iceland plume. Extra hot mantle raises the melt supply and weakens the axial lithosphere, producing the thermal state of a fast ridge at the rate of a slow one. Head north to Kolbeinsey and the effect fades along strike as the plume influence dies away.

The reverse case exists too. The Australian Antarctic Discordance spreads at about seventy millimetres per year, comfortably intermediate, yet has a deep axial valley, because the mantle there is anomalously cold and melt supply is low.

Both exceptions tell you the same thing. The rate to morphology correlation is real, but rate is a proxy for melt supply and not the cause. Where something else changes the melt supply, the morphology follows the melt and ignores the rate.

Why the seafloor gets deeper as it ages

Open the depth and age tab. The ridge crest sits around 2,500 to 2,600 metres below sea level. By fifty million years the seafloor is near 5,000 metres and by one hundred and fifty million years it is beyond 6,000. That deepening is one of the most regular relationships in Earth science and it is worth understanding rather than memorising.

The square root law and where it comes from

The plate is hot when it forms and it cools by losing heat upward. A body cooling from its surface develops a thermal boundary layer that thickens as the square root of time, which is a general result for heat diffusing into a half space and has nothing specific to do with rocks.

Cool rock is denser than hot rock. As the boundary layer thickens, more and more dense material accumulates in the column, and isostasy requires the surface to sit lower. The depth increase is proportional to the amount of extra dense material, which is proportional to the thickness of the cooled layer, which goes as the square root of age.

So depth grows as the square root of age. Double the age and you deepen by a factor of about 1.41, not 2. The usual form quoted is

depth in metres equals 2,500 plus 350 times the square root of the age in millions of years.

That coefficient of 350 is not arbitrary. It can be derived from thermal expansion, mantle temperature and thermal diffusivity, and when you do that derivation properly it comes out at 365. The gap between 350 and 365 is the difference between a coefficient fitted to real North Atlantic bathymetry and one computed from first principles, and it is a fair measure of how well the physics does: within about four percent, from a model with no fitting at all.

Where the simple model stops working

Run the square root law forward and it never stops. At two hundred million years it predicts about 7,450 metres, at four hundred million about 9,500. The seafloor does not do this. Past about eighty million years the observed depth flattens out, and the discrepancy grows steadily.

The numbers from the tool: at one hundred million years the square root law puts the seafloor 502 metres deeper than GDH1 does. At one hundred and fifty million years the gap is 1,174 metres. Over a kilometre is far too large to be measurement error, and the observations side with the flatter models.

The reason is that half space cooling assumes the plate can keep cooling and thickening indefinitely, with nothing underneath resupplying heat. In reality the base of the plate is held near a constant temperature, because once the plate is thick enough, heat arriving from the mantle below balances the heat being lost above. The plate stops thickening, so the seafloor stops deepening, and the depth approaches an asymptote.

The two plate models and how they differ

Parsons and Sclater in 1977 fitted a plate of 125 kilometres thickness with a basal temperature of 1,350 degrees. Stein and Stein in 1992 published GDH1, standing for global depth and heat flow, fitting a thinner plate of 95 kilometres with a hotter base at 1,450 degrees, using a much larger global dataset. GDH1 flattens sooner and to a shallower asymptote.

Both are on the chart and you can toggle between them. They differ from each other by less than either differs from the simple square root law, which is the honest summary: the existence of an asymptote is well established, its exact depth is still argued about.

One detail found while checking these models is worth passing on, because it is the sort of thing that quietly costs people hours. The Parsons and Sclater closed form does not exactly reproduce the Fourier series solution built from their own quoted parameters. The mismatch is about sixty two metres. It happens because their published equation was fitted with a basal temperature of 1,333 degrees while their parameter table rounds to 1,350. GDH1 has no such problem and reproduces its own series to within a metre. If you implement these models and find a small unexplained offset in the plate model, that is where it comes from.

Why the observations need correcting first

The depths plotted on the chart are not raw soundings. Sediment sits on top of old seafloor and its weight pushes the crust down, so a raw depth over thick sediment is deeper than the basement would be on its own. Before comparing with a model you unload the sediment isostatically.

The size of that correction is not trivial. It is about 28 metres at ten million years, 271 metres at one hundred, and 820 metres at one hundred and fifty. That last figure is comparable to the entire flattening signal being argued about, which means the sediment correction is not a detail, it is part of the argument. Different authors handle compaction differently and the resulting spread is tens to a couple of hundred metres.

What seafloor spreading builds

The layered structure of oceanic crust

Where melt supply is normal, spreading builds a consistent sequence about six to seven kilometres thick, and it is the same nearly everywhere because the process making it is the same.

At the top is pillow basalt, erupted underwater and quenched into rounded lobes. Below that is a sheeted dyke complex, the frozen feeder conduits, each dyke intruded into the middle of the one before it, so the whole unit is a stack of vertical slabs with no country rock between them. Below that is gabbro, the same magma cooled slowly at depth. Below that is mantle peridotite, often serpentinised near the top where seawater has reached it.

That sequence is exposed on land in ophiolites, slices of oceanic lithosphere thrust onto continents during collision, and its consistency from ophiolite to ophiolite was itself early evidence that ocean crust forms by a single repeatable process.

The thickness is remarkably constant at six to seven kilometres for any ridge above about twenty millimetres per year. Below that it thins and eventually fails altogether, which is the ultraslow behaviour described above.

Hydrothermal circulation, and the largest metamorphic factory on the planet

Seawater penetrates the fractured hot crust near the axis, gets heated to three hundred and fifty degrees and above, reacts with the basalt, and comes back out at vents. This is where black smokers live, and where the chemosynthetic ecosystems around them exist without any input from sunlight.

The chemistry matters beyond the biology. Circulating seawater strips magnesium from the ocean and adds calcium, and the entire ocean volume cycles through this system on a timescale of the order of ten million years, which makes ridges a first order control on seawater composition over geological time.

It also produces metamorphic rock in enormous volume. The primary igneous minerals react to chlorite, epidote, actinolite and albite, which is greenschist facies metamorphism, and it happens to essentially all new oceanic crust within a few million years. About three cubic kilometres of new crust forms every year and nearly all of it is altered this way. By volume that exceeds all the metamorphism happening in every mountain belt on Earth combined, and it appears on almost no rock cycle diagram.

Fracture zones and the offsets you see on any map

Ridges are not continuous lines. They are broken into segments offset by transform faults, and the fracture zones extending beyond the active transform are visible on any seafloor map as long scars running perpendicular to the ridge.

The transform is only seismically active between the two ridge segments, because that is the only stretch where the plates on either side move in opposite directions. Beyond the segments both sides belong to the same plate and travel together, so the fracture zone is an inactive topographic scar rather than a fault.

That distinction, made by Tuzo Wilson in 1965, was another decisive piece of evidence. It predicts that earthquakes occur only on the segment between the ridge offsets, and it predicts the slip direction is opposite to what you would guess from the offset shape. Both were confirmed once seismic networks improved enough to check.

Where seafloor spreading happens today

The global ridge system runs about sixty five thousand kilometres, the longest mountain range on the planet, and almost all of it is underwater.

The Mid Atlantic Ridge runs the length of the Atlantic from the Arctic to the Southern Ocean, slow spreading with a median valley for most of its length. The East Pacific Rise runs up the eastern Pacific, fast to superfast, with an axial high. The Indian Ocean carries a three armed system meeting at a triple junction. Gakkel Ridge crosses the Arctic under permanent ice, the slowest of them all.

Iceland is where the system comes above sea level, because the Iceland plume supplies enough extra melt to build the ridge above water. It is the one place you can stand on a mid ocean ridge without a submersible, which is why so much of the fieldwork behind these ideas was done there.

Two young ocean basins show the earliest stages. The Red Sea and the Gulf of Aden are narrow seas that have begun genuine seafloor spreading within the last few million years, with true oceanic crust in their axes. The East African Rift is the stage before that, continental crust being stretched and thinned but not yet parted, and it belongs in the database with a flag on it because it is not a spreading centre yet.

Back arc basins are the other family: Mariana Trough, Lau Basin, East Scotia Ridge. These spread behind subduction zones rather than between diverging continents, and they can be fast, but they are usually short lived and their rates vary strongly along strike.

The other end of the conveyor

Seafloor spreading cannot run on its own. The Earth is not expanding, so every square metre of new crust made at a ridge has to be matched by a square metre destroyed somewhere else, and that somewhere is a subduction zone.

The balance is close. About three cubic kilometres of new oceanic crust forms each year and very nearly the same volume is consumed. The two rates have to match over the long run, and the fact that they do is one of the constraints that fixes the whole plate tectonic budget.

This is also the reason ocean floor is young. Oceanic lithosphere cools, thickens and becomes denser as it ages, and past roughly one hundred and eighty million years it is dense enough that it will subduct given any opportunity. Continental crust is too buoyant to subduct, so it accumulates and survives. That single density difference explains why the ocean basins are all young and the continents are all old.

What happens at the far end, how a slab dehydrates rather than melts, and why arc volcanoes sit where they do, is a longer story and its own subject.

What is taught wrong about seafloor spreading

That the ridge pushes the plates apart. Ridge push exists but it is a minor term. Most of the driving force is slab pull at the other end of the plate, and the ridge is largely a passive gap.

That the stripes are visible bands of different rock. The crust is the same basalt throughout. What alternates is the direction of a weak remanent magnetisation, detectable only with an instrument.

That a magnetic reversal means the field switches off and comes back. It does not vanish. It weakens, becomes complicated with multiple poles for a few thousand years, and reorganises the other way round. The transition takes something of the order of one to ten thousand years, which on a spreading ridge is a stripe boundary a few hundred metres wide.

That every reversal appears in every profile. Short chrons make narrow stripes, and on a slow ridge those stripes fall below what a magnetometer can resolve. Missing anomalies are a resolution limit, not a gap in the field record.

That the seafloor keeps deepening forever as the square root of age. It flattens past about eighty million years, because the base of the plate stops cooling. The simple law is excellent inside its range and wrong outside it.

That spreading rates are constant over time. They are not. Rates change over millions of years, ridges jump and reorganise, and a rate computed across a long interval is an average that may conceal a good deal of variation.

That seafloor spreading and continental drift are the same idea. Continental drift was the observation that continents have moved. Seafloor spreading is the mechanism that explains how. Wegener had the first without the second, which is exactly why he was not believed.

Mistakes that catch people out

Doubling a full rate. Covered above and worth repeating because it is the most common error in this entire topic. Measure across the axis and you already have a full rate.

Mixing timescales. Computing a rate with a distance from one publication and an anomaly age from another, when the two used different polarity timescales. In the Paleocene that alone can shift the age by 1.4 million years and the rate by several percent.

Reading a rate off a single anomaly and calling it the rate. One anomaly gives you an average rate over the whole interval from now to that age. If you want to know whether the rate has changed, you need several anomalies and you plot distance against age and look at the slope changing.

Assuming symmetry is exact. It usually is close, but asymmetric spreading is real, with one flank receiving more crust than the other, and ridge jumps relocate the axis entirely. If your two sides disagree, that may be a genuine observation rather than a measurement error.

Comparing raw bathymetry with a depth model. The sediment correction reaches 820 metres on old crust, which is the same size as the effect being tested. Comparing uncorrected depths with a cooling model will make the model look wrong when it is not.

Treating the axial valley as the definition of a ridge. Fast ridges have no valley at all. If you are taught the median valley as a defining feature you will not recognise the East Pacific Rise when you see a profile of it.

How spreading rates are measured now

The geological method

The magnetic anomaly method described above gives you a rate averaged over whatever interval you measured. Pick a young anomaly and you get a rate averaged over the last million years. Pick an old one and you get an average over tens of millions. Global plate motion models such as MORVEL are built this way, combining anomaly picks with transform fault azimuths and earthquake slip directions across every boundary, then solving for a self consistent set of plate rotations.

The word self consistent is doing real work there. Plate motions have to close: go around a circuit of three plates and the relative motions must sum to zero. That constraint is what turns a pile of local measurements into a global model, and it is also what catches errors, because a circuit that fails to close means something in it is wrong.

The geodetic method

Since the nineteen nineties you can also just measure it. Put GNSS receivers on both sides of a boundary and watch them separate, at a precision now well under a millimetre per year. This gives a rate averaged over years rather than millions of years.

The interesting result is that the two methods agree, mostly. Geodetic rates and geological rates match within uncertainty for the great majority of plate pairs, which tells you plate motion is steady over a wide range of timescales. Where they disagree, the disagreement is informative: it usually points to a plate boundary that is not behaving as a single rigid pair, such as a diffuse deforming zone, or a genuine change in rate over the last few million years.

The places to be careful are back arc basins and continental rifts. In the Andaman Sea, geodetic and magnetic anomaly rates differ by roughly a factor of two. That is not measurement failure, it is a boundary that is deforming over a broad region rather than slipping on a line, and a single number does not describe it well.

What the rate predicts about everything else

The rate of a ridge is not just a number attached to it. It sets or strongly influences most of the other things you would want to know.

Axial morphology follows it, as described earlier, through melt supply. Crustal thickness is roughly constant at six to seven kilometres above about twenty millimetres per year and falls away below that. Hydrothermal output scales with the heat being delivered, so fast ridges host far more vent activity per kilometre of axis. Earthquake style differs, with slow ridges producing larger normal faulting events on the axial valley walls and fast ridges producing smaller, shallower events.

Segment length follows it too. Fast ridges have long segments with widely spaced offsets. Slow ridges are chopped into shorter segments by more frequent transforms and non transform discontinuities.

Even the roughness of the seafloor follows it. Abyssal hill topography, the corrugated fabric covering most of the ocean floor, is built at the axis by faulting, and slow ridges make it far rougher. You can estimate the spreading rate of ancient crust from the amplitude of its abyssal hills, long after the ridge that made it has gone.

Seafloor spreading and the height of the sea

Here is a consequence that is easy to miss and is worth the detour, because it connects a process at the bottom of the ocean to the geological record on land.

Ocean depth depends on crustal age. Young crust is shallow, old crust is deep. Now consider what happens if the global spreading rate increases. More crust is created per unit time, so the average age of the ocean floor falls, so the average depth of the ocean floor decreases. The ocean basins get shallower.

But the volume of water has not changed. Shallower basins hold less water, and the excess has to go somewhere, so it floods onto the continents.

This is a large effect. During the Cretaceous, spreading rates were substantially higher than today, the ocean floor was younger on average, and sea level stood something like a hundred to two hundred and fifty metres above the present. Shallow seas covered large parts of North America, Europe and Africa. The chalk of southern England, the same chalk that gives Dover its cliffs, was deposited in one of those seas.

So the reason there is a thick chalk sequence across northern Europe is, in part, that mid ocean ridges were spreading faster. That is a long causal chain and every link in it is quantitative.

The reverse also applies. As spreading slowed through the Cenozoic, the ocean floor aged, basins deepened, and sea level fell, which is one of several contributors to the long term regression recorded in sequence stratigraphy.

Reading a real profile, step by step

If you are handed an actual magnetic profile rather than the idealised one on this page, the workflow is roughly this.

Find the axis first. The central anomaly is the widest normal band in the young part of the profile, corresponding to the Brunhes chron, and it is usually the most obvious feature. Its centre is your ridge axis, and it should sit at or near the bathymetric axis too. If those two disagree by more than a few kilometres, note it, because it may mean the axis has jumped.

Then check symmetry before doing anything else. Fold the profile about the axis you just picked and see whether the two halves match. If they do not, either your axis pick is wrong or the spreading is asymmetric, and you need to know which before you compute anything.

Then match the sequence to the timescale by shape, not by counting. The pattern of wide and narrow bands is distinctive, and experienced interpreters recognise blocks of it the way you recognise a barcode. Counting anomalies outward from the axis and hoping is how you end up one chron out, which produces a rate error of a few percent that is large enough to matter and small enough to look plausible.

Then pick boundaries consistently. Because the anomaly is smoothed, choose a rule, such as the steepest gradient point, and apply the same rule everywhere. Consistency matters more than which rule you chose, because a systematic offset largely cancels when you take differences.

Finally, plot distance against age for every boundary you identified, rather than computing a rate from one. The slope is the half rate. If the points fall on a straight line the rate was steady. If the slope changes, you have found a real change in spreading rate, and that is a more interesting result than the rate itself.

A second worked example, this time on depth

The rate calculation earlier used the stripes. This one uses the depth model, and it is the check you can run against any published bathymetry.

You have a point in the North Atlantic where the magnetic anomalies tell you the crust is fifty million years old. How deep should the seafloor be?

Take the square root law first. The square root of fifty is 7.071. Multiply by 350 to get 2,475 metres of subsidence since the crust formed. Add the ridge crest depth of 2,500 metres and you predict 4,975 metres.

Now GDH1. Fifty million years is past its crossover at twenty, so use the exponential branch: 5,651 minus 2,473 times the exponential of minus 0.0278 times 50. The exponent is minus 1.39, its exponential is 0.2491, so the second term is 616 metres, and the predicted depth is 5,035 metres.

Sixty metres apart. At fifty million years the two models are still effectively the same answer, and sixty metres is comfortably inside the scatter of real bathymetry.

Now repeat at one hundred and fifty million years. The square root law gives 2,500 plus 350 times 12.247, which is 6,786 metres. GDH1 gives 5,651 minus 2,473 times the exponential of minus 4.17, which is 5,613 metres. The gap is 1,173 metres.

That is the whole argument in two calculations. At fifty million years you can use either model and nobody will notice. At one hundred and fifty the choice changes your answer by more than a kilometre, and the measurements say the flatter model is right.

The practical rule: below about eighty million years use the square root law, it is simple and accurate. Above that use a plate model, and be aware that the exact asymptote is still an open question.

Seafloor spreading you can walk on

Almost all of this happens under three kilometres of water, which makes it hard to study and easy to doubt. There are three places where that is not true.

Iceland sits on the Mid Atlantic Ridge and is above sea level only because the Iceland plume supplies enough extra melt to build it there. You can stand in the rift valley at Thingvellir and see the fissures. What you cannot see is a clean two sided spreading centre, because on land the deformation spreads across a wide zone of fissure swarms rather than concentrating on a single axis. Iceland is the ridge, but it is an unusually fat and complicated version of it.

The Afar region of Ethiopia is a rift on the point of becoming a spreading centre, with true oceanic crust beginning to form in places. In 2005 a sixty kilometre dyke intruded over a few days, opening the ground by up to eight metres, which is a decade or more of normal spreading delivered in one event. That episodic behaviour is a useful corrective to the smooth continuous picture: spreading averages out to a steady rate but happens in bursts.

Ophiolites are the third and most important. These are slices of oceanic lithosphere thrust up onto continental crust during collision, preserving the whole sequence from pillow basalt down through sheeted dykes and gabbro into mantle peridotite. The Troodos massif in Cyprus and the Semail ophiolite in Oman are the classic examples, and Semail exposes tens of kilometres of continuous section.

Ophiolites were mapped and argued over long before anyone knew what they were. The sheeted dyke complex in particular made no sense until spreading was proposed, because a unit made entirely of vertical dykes with no host rock between them requires continuous extension while intrusion is happening. Once you know it formed at a ridge, it is obvious. Before that, it was a puzzle.

Has the rate always been this?

No, and the variation matters.

Global spreading rates were substantially higher through much of the Cretaceous than they are now, which as described earlier is part of why sea level was so much higher. Rates then declined through the Cenozoic. Individual ridges vary far more than the global average, because ridges reorganise: they jump, they propagate, segments die and new ones open.

The Pacific plate provides the clearest case of a reorganisation, in the bend of the Hawaii Emperor seamount chain about forty seven million years ago, when the direction of plate motion changed markedly. Whether that was the plate changing course or the plume moving is still argued, and probably both contributed.

For the practical purpose of using this page, the consequence is that a rate you compute from an old anomaly is an average, and averages hide changes. If you want the modern rate, use the youngest anomalies you can resolve or use GNSS. If you want the history, plot every anomaly you can identify against distance and read the changes off the slope.

Why Earth is the only planet doing this

Seafloor spreading is not a general property of rocky planets. As far as we can tell it happens on exactly one, and asking why is a good way to check whether you have understood the mechanism.

Mars has no plate tectonics now and probably never had it in a sustained form. It shows some magnetised crustal stripes in the southern highlands, which caused excitement when Mars Global Surveyor found them, but they are not symmetric about any axis and are better explained as an ancient magnetised crust broken up by later impacts. Mars is small, so it lost its internal heat quickly, its lithosphere thickened, and it settled into a single rigid shell. Olympus Mons is the consequence: a volcano that grew to twenty two kilometres because the plate above the hotspot never moved, so every eruption piled onto the same spot instead of building a chain.

Venus is nearly Earth sized and still has no plates. It has volcanism, tectonic deformation and a young surface, but no ridges, no trenches and no linear magnetic record. The leading explanation is water. Water weakens rock substantially, and Venus lost essentially all of its water. A dry lithosphere is strong enough to resist breaking into plates, so heat escapes some other way, possibly through episodic global resurfacing.

That points at what Earth actually needs. Enough internal heat to keep convection going, a lithosphere weak enough to break rather than deform as one shell, and water to provide that weakening and to lubricate subduction once it starts. Take away the water and you likely take away the whole system, including the ridges.

It is worth noticing what that means for the rock cycle. Without plate tectonics there is no continuous creation of new crust at ridges and no destruction at trenches, so the surface is not recycled. Mars still carries its four billion year old surface because nothing has removed it. Earth has almost no ocean floor older than one hundred and eighty million years because everything has.

Common questions about seafloor spreading

What is seafloor spreading in simple terms?

It is the process that makes new ocean floor. Molten rock rises at a mid ocean ridge, erupts, and freezes into new crust. The two plates on either side move apart, the new rock splits down the middle, and half travels away on each plate. Repeat continuously and you build an ocean basin outward from its centre.

What causes seafloor spreading?

Mostly the pull of old, cold, dense oceanic plate sinking into the mantle at subduction zones at the far edge of the plate. That pull drags the whole plate along, and the gap that opens at the ridge fills with rising mantle that melts as pressure drops. Ridge push, the gravitational sliding of the plate off the raised ridge, contributes as well but is the smaller term.

Who discovered seafloor spreading?

Harry Hess proposed the mechanism in 1962, in a paper he called an essay in geopoetry. The decisive test came in 1963 from Fred Vine and Drummond Matthews, and independently from Lawrence Morley, who predicted that if spreading were real the seafloor would carry symmetric magnetic stripes recording reversals of the Earth field. Robert Dietz published a similar spreading idea to Hess in 1961 and gave the process its name.

What is the evidence for seafloor spreading?

Four independent lines. Symmetric magnetic stripes on both sides of the ridge, in the same sequence worldwide. Crustal ages from drilling that increase systematically away from the axis. Sediment that is nearly absent at the ridge and thickens outward. Heat flow that is highest at the axis and falls with age. Each would be suggestive alone, and together they are conclusive.

Where does seafloor spreading occur?

At divergent plate boundaries, which are almost all mid ocean ridges. The system runs about sixty five thousand kilometres, including the Mid Atlantic Ridge, the East Pacific Rise, the Indian Ocean ridges and Gakkel Ridge in the Arctic. Iceland is the main place it reaches above sea level. The Red Sea and Gulf of Aden are young examples, and back arc basins such as the Lau Basin spread behind subduction zones.

What type of plate boundary causes seafloor spreading?

A divergent boundary, where two plates move away from each other. Convergent boundaries destroy crust and transform boundaries slide plates past each other without creating or destroying any, so neither produces new seafloor.

How fast does the seafloor spread?

Between about ten and one hundred and fifty millimetres per year full rate, which is comparable to the growth of a fingernail. The Mid Atlantic runs around twenty to twenty five, the East Pacific Rise reaches one hundred and fifty, and Gakkel Ridge manages about twelve. The classes are ultraslow below twenty, slow to fifty five, intermediate to eighty, fast to one hundred and forty, and superfast above that.

What is the difference between half rate and full rate?

A half rate is the speed of one plate moving away from the axis, which is what you get when you measure from the axis outward on one side. A full rate is the speed at which the two plates separate, which is twice the half rate, and it is what you get if you measure between matching anomalies on opposite sides. Published ridge rates are usually full rates. Doubling a rate that was already a full rate is the most common error in this calculation.

How do magnetic stripes prove seafloor spreading?

Because they are symmetric about the ridge and identical in sequence worldwide. New basalt records the field direction as it cools through 580 degrees, then splits at the axis so each plate keeps a copy. There is no other process that would write the same global time signal into the rock on both sides of a ridge at matching distances.

Why is the ocean floor so young?

Because it is continuously destroyed. Oceanic lithosphere cools and grows denser with age, and past roughly one hundred and eighty million years it subducts readily. Nothing older survives, whereas continental crust is too buoyant to subduct and reaches beyond four billion years. The oldest ocean floor is about one hundred and eighty to two hundred million years, in the western Pacific and eastern Mediterranean.

What is a mid ocean ridge?

The underwater mountain range that marks a divergent plate boundary, where new oceanic crust forms. It stands one to three kilometres above the surrounding seafloor because the rock beneath it is hot and therefore less dense. Slow ridges carry a rift valley along the crest and fast ridges carry a smooth axial high instead.

Why does the seafloor get deeper away from the ridge?

Because the plate cools as it ages, contracts, and becomes denser, so it sits lower. Depth grows as the square root of age while the crust is young, following roughly 2,500 plus 350 times the square root of the age in millions of years. Past about eighty million years the deepening flattens off, because heat arriving from below balances the heat lost above and the plate stops thickening.

What is the relationship between seafloor spreading and continental drift?

Continental drift is the observation that continents have moved. Seafloor spreading is the mechanism that explains it. Wegener had the observation in 1912 but proposed that continents ploughed through oceanic crust, which is physically impossible, and he was rejected for it. Hess supplied the correct mechanism fifty years later: the continents do not plough through the ocean floor, they ride on plates that include it.

Is seafloor spreading still happening?

Continuously, everywhere along the sixty five thousand kilometre ridge system. About three cubic kilometres of new oceanic crust forms every year, and very nearly the same volume is subducted, which is why the Earth is not expanding. You can watch it directly with GNSS receivers, which measure ridge separation to well under a millimetre per year.

What is the Curie point and why does it matter here?

It is the temperature above which a mineral cannot hold a stable magnetisation, about 580 degrees for magnetite. Basalt erupting at a ridge is far hotter than that and carries no fixed magnetisation. As it cools past the Curie point the magnetic minerals lock in whatever direction the Earth field is pointing at that moment, and they keep it for hundreds of millions of years. That is why the seafloor is a recording of the field rather than a reflection of the present one.

Why do some magnetic reversals not show up in a profile?

Because they made a stripe too narrow to resolve. The Jaramillo subchron lasted eighty thousand years, which on a ridge spreading at twenty five millimetres per year half rate gives a stripe only two kilometres wide. Through several kilometres of water the magnetic signal from a body that narrow is smoothed to the point of disappearing. On a fast ridge the same event makes a stripe wide enough to see easily. The reversal happened globally either way.

Related Calculators and Reference

• Pangea Map and Continental Drift Reconstruction is what the spreading on this page was taking apart, and it shows why the magnetic record running out at 180 million years matters so much.
• Subduction Zone Explorer and Megathrust Calculator is the destructive half of the same system. The crust made here is consumed there, and the age it carries in sets how deep its earthquakes will reach.
• Rock Cycle Diagram and Identification Key puts the ridge in the wider system, and its cross section shows the same crust being made here and destroyed at a trench.
• Interactive Geological Time Scale is where the chron ages on this page sit relative to everything else in Earth history.
• Mohs Hardness Scale covers the minerals that make up the basalt and gabbro this process builds.
• Rock Density Calculator turns a mineral assemblage into a bulk density, which is the quantity that drives the depth versus age relationship on this page.
• Density of Basalt gives the measured values for the rock that makes up almost all new oceanic crust.
• Sediment Settling Velocity Calculator quantifies the pelagic blanket whose thickness away from the axis was one of the original arguments for spreading.
• International Union of Geological Sciences is the body behind the Geologic Time Scale that the polarity chron ages on this page come from.