Introductory Note
     Although the discussion below is a little dated (as noted at several places as part of editing for html, typographical and grammatical errors on Dec 5/6, 2014), almost all the information below is still perfectly accurate within the level of uncertainty stated in various places. In particular, the discussions of how we know what is going on inside the various Terrestrial planets is as accurate as ever, so this page should be a useful supplement to the material presented in any introductory textbook.

The Structure of the Earth
     Students: Refer to the discussion of the Earth's structure in your text before going on.
     Of all the planets, only the Earth's structure is well determined. As shown in the diagram below the Earth's interior consists of several layers: the crust, a relatively thin region of low-density silicates, the mantle, a thick region of higher-density iron-rich silicates, and the core, a central region of iron mixed with various impurities. Each of these layers is subdivided into different regions. The crust has continental regions which are made of even lower-density aluminum-rich silicates, and oceanic regions which are much richer in somewhat denser iron-rich silicates. The mantle is divided into an upper mantle within which the iron-rich silicates are gradually compressed from lower-density more open mineral structures, to higher density more compact mineral structures, and a lower mantle, where the mineral structures have been compacted to their densest forms and increase in density as one descends further into the mantle only because of the continually increasing weight compressing them. The core consists of a molten (liquid) outer core primarily consisting of iron but with substantial percentages of materials such as sulfur and silicon which lower its melting temperature, and a solid inner core which is almost certainly a crystalline mixture of iron and nickel similar to iron meteorites.

      It should be noted that it is often incorrectly stated in public media that the inner core is solid, while the outer core is liquid, because of the much greater weight compressing the inner core. This is completely wrong. For one thing, if the pressure was responsible for the change, then as you go downward the gradual increase of pressure would produce a gradual change from liquid, to a slushy mixture of liquid and solid, to solid material, which is not what is observed. The boundary between the two regions is very sudden, as though there is some intrinsic reason unrelated to the pressure which causes the change in state from liquid to solid. Secondly, as discussed below, the pressure in the inner core is only slightly greater than that in the outer core, because regions near the center of a planet, although compressed and held in place by the materials above them, have relatively little weight of their own, being pulled more or less equally in all directions by the outer parts of the planet, and as a result having their weight determined only by their own gravity.
     As can be seen in the diagram above the Earth's structure is very accurately known, with the actual depths of each layer (which vary only slightly from one place to another) being known to a fraction of one percent. This accurate determination of the Earth's structure is made possible by the study of seismic waves (earthquake waves) passing through the Earth. There are two types of waves which do so. One, called P-waves, moves about twice as fast as the other, called S-waves, and therefore arrives at a given seismic station before the other; the one that arrives first is called the Primary wave (hence P-wave), and the one that arrives second is called the Secondary wave (hence S-wave).

     The diagram above shows the velocities of P and S waves at different depths inside the Earth (it also shows the estimated densities of different parts of the Earth, but we will ignore that for the moment). Note that the P waves always move faster, as previously stated, that the velocities generally increase as you go downward, rapidly near the surface and more slowly in other regions, and that there is a nearly 50% drop in the velocity of P waves (and S wave velocities are not shown at all) in the region labeled as the Outer Core.

     How do we know this? By studying how the waves arrive at seismographs scattered around the Earth -- when they arrive, what direction they appear to be coming from, and what intensity (strength) they have (as shown in the image below).

Waves moving through the Earth move at different velocities at different depths. Where the change in velocity is slow the waves bend smoothly; where it is sudden, the waves bend more sharply. (Eugene C. Robertson, USGS)

A more detailed look at the motion of P-waves, showing the paths of many waves emanating from a specific focus (at the top), the travel times (tick marks on the wave paths represent one minute intervals), and the shadow zone formed by the core of the Earth. Waves that should pass through the shadow zone are refracted and concentrated by the core, producing greater intensity in the region opposite the earthquake than would otherwise occur. For Mercury in particular, this effect seems to be very large due to its unusually large core. (S-waves are not shown, as they are absorbed by the core and do not reach the opposite side of the Earth, save for S-waves created at the base of the mantle by P-waves emerging from the core.) (Public domain image from USGS Visual Glossary)

     As noted above, the structure of the Earth can be determined in considerable detail from the study of earthquake waves (seismology). Additional information (to be covered in the next iteration of this page) can be obtained from laboratory studies of the behavior of rocks exposed to various pressures and temperatures, and comparisons of those results with seismic studies, from the composition of melt-rock (lava) at various places, and from circumstantial evidence about the structures of other Solar System objects, in the form of meteorites.

Temperatures inside the Earth
     Laboratory studies of rock behavior show that in order to mimic the behavior of rocks inside the Earth, laboratory samples must be exposed not only to high pressure (corresponding to the weight of rocks above any point inside the Earth), but also surprisingly high temperatures as shown below.

Temperatures inside the Earth (in red) inferred from comparison of laboratory studies with seismic results, schematic representation (in black) of typical melting temperatures, and (in blue) schematic representation of temperatures of rising plumes of hot mantle material. Where melt-temperatures are lower than actual temperatures (in the outer core), the material is molten (liquid). Where melt-temperatures are higher than actual temperatures (everywhere else), the material is solid. If the melt-temperature is well above the actual temperature (near the surface and in the lower mantle), the rocks are very stiff. If the melt-temperature is close to the actual temperature (in the inner core, at the crust/mantle boundary, and for rising plumes), the rocks are "soft" or "plastic" (easily deformable).

     As the above diagram shows, near the surface of the Earth temperatures rise rapidly (as much as 100 Fahrenheit degrees per mile), so that a few tens of miles beneath the surface the temperature is close to the melting temperature of the rocks; and if that rapid temperature rise continued into the core the temperature would exceed the melting and vaporization temperatures of all known materials; but before that happens the temperature increase drops to only a few degrees per mile, so that 1800 miles down, at the base of the mantle, the temperature is only a few thousand degrees higher and well below the melting temperature of the rocks.
     The velocity of earthquake waves depends upon the stiffness (rigidity) of the bonds between the molecules that make up the materials inside the Earth. Near the surface the bonds are relatively long and "loose" and the predominant materials (olivine and similar ferromagnesian silicates) have relatively low densities (3.5 to 4.5 tims the density of water, which is much less than the average density of the Earth); but at greater depths the weight of the overlying material pushes the atoms closer together, making the bonds between them shorter and stiffer, thereby (1) increasing the density of the material (since the atoms are closer together they take up less room), (2) increasing the melting temperature (shorter bonds are stiffer and harder to break, which is a requirement for melting the minerals), and (3) increasing seismic wave velocities (shorter, stiffer bonds cause the impulse received by one atom to be transferred to the next atom at a more rapid rate).

Effects near the crust / mantle boundary
     In the diagram above, note the rapid increase in wave velocities near the surface of the Earth (in the crust) caused by the compression of the rocks by the weight of the rocks above them. Close to the surface the difference between actual and melting temperatures is very large (refer to the temperature diagram further above), and the increasing compression is the most significant factor in changing the wave velocity; but at depths of thirty to fifty miles the rapidly increasing temperature gets so close to the slowly increasing melt temperature that the bonds between the atoms, even though shorter than at lesser depths, are "softer" and the rocks, although remaining solid, become relatively "soft", or less stiff. We see the effects of this in some rocks found at the surface of the Earth in which grains of individual minerals are stretched out into linear structures. The rocks, having been heated to nearly their melting temperature, can "flow", even though in the solid state, in the same way that hot metals can be squeezed between rollers, forming thin sheets (e.g., sheet metal and foils). The reduction in the stiffness of the bonds causes the wave velocities to drop sharply, leading to the terminology shown on the right side of the diagram -- "lithosphere" (rock sphere), for the stiffer, more brittle layers above, and "asthenosphere" (plastic sphere), for the softer, more pliable ("plastic" in that sense, not in the sense of the organic materials sold as plastics in stores) region caused by the small difference between the actual and melt temperatures. The boundary between the crust and mantle is defined by this low-velocity zone, also known as the Mohorovicic Discontinuity (or for short, the Moho).

Effects near the core / mantle boundary
     At the core/mantle boundary there is an even larger, more rapid temperature increase -- three to five thousand Fahrenheit degrees in just a few miles distance. This causes large amounts of heat to flow from the core into the lower mantle. Portions of the lower mantle rock are heated to the point where, although still solid, they can slowly flow in a "plastic" way toward the surface (this is discussed below, with respect to the asthenosphere). Hot, solid but "soft" rock is forced upwards through cooler, stiffer rock at rates of 1/2 an inch to 1/2 a foot per year. In the diagram above the blue line represents temperatures in an imaginary "rising plume" of hot soft rock -- cooler than its melt temperature, but well above the temperature of surrounding rocks.

The probable distribution of radionuclides inside the Earth
     The reason for the rapid temperature rise in the crust, down to the Mohorovicic Disconuity, is a concentration of radioactive materials near the surface of the Earth when it melted and differentiated, during or very soon after its formation, 4.5 billion years ago. As the short-lived radioactive materials present at that time decayed and disappeared (that is, were transformed into other materials), the heating that caused the Earth to melt also disappeared, and the Earth began to resolidify. The crust cooled rapidly by radiating heat to space, forming a thin low-density slag similar to the highland crust of the Moon, made primarily of silica and aluminosilicates; the deeper regions below, unable to radiate heat directly to space, cooled more slowly and remained hot for a much longer period of time (in fact the core is still extremely hot, partly as a result of the 1800 miles of solid rock insulating it from space, and partly because of heavy radioactive materials such as uranium which were dragged into the core as the heavy material sank to the center of the Earth).
     As the Earth cooled minerals and rocks began to form, even in the depths of the mantle. At great depth, because of the large weights compressing the material, only very dense structures could form (as will be discussed in more detail below). Large atoms of uranium can't fit into dense structures unless in a nearly pure form (although pure uranium metal is nearly twenty times denser than water, most uranium ores are much lighter, and nearly pure carnotite, or uranium oxide, is only about the same density as rocks in the upper mantle, which are mostly less than five times the density of water), and even if they could fit into such structures the heat released by radioactive decay would soon melt them again, since when the mantle began to solidify temperatures were still very close to melt temperatures. As a result most of the uranium atoms not dragged into the core were gradually forced toward the surface, where lower density minerals where forming; and most of the uranium in the mantle is believed to lie within a hundred miles of the surface, which means most of the heating due to its radioactive decay is concentrated near the surface. This is believed to be the cause of the rapid temperature rise near the surface. In the top few tens of miles heat slowly flows through the rocks (even at 100 Fahrenheit degrees per mile temperature gradient this heat flow is slow), and is equally slowly replaced by radioactive heating. Below that depth heat flow is much slower, because the heat source is the heat of the core, which is nearly two thousand miles further down, and the temperature rise is much lower.
     Before leaving this topic it might be noted that if large amounts of uranium were pulled into the core, the core would still be generating heat through the decay of that material and would still have a very high temperature; and as shown in the temperature graph, although the exact temperatures in the core are very uncertain (because the exact composition is not known, making laboratory simulations of uncertain accuracy), they are certainly much hotter than in the mantle, and there is a bigger, faster rise in temperature at the boundary between the mantle and the core than at the boundary between the crust and the mantle. So although the uranium present in the mantle is probably concentrated near the surface of the planet, much (or even most) of the uranium in the Earth may be in the core.

Mineralogical changes within the mantle
     Referring to the diagrams above again, note that once the rate of temperature increase moderates, the greater weight compressing the material again becomes the most important determinant of the wave velocity and wave velocities begin to increase again. In the three or four hundred miles closest to the surface, however, there are two reasons for the increase in wave velocity. The first, which operates throughout the body of the Earth, is the increasing compression of the material, resulting in shorter, stiffer atomic bonds which promote faster exchange of wave impulses from one atom to the next. The other only operates near the surface, and as a result the wave velocity increases more rapidly there than in deeper parts of the mantle, as shown below.

     Note that in every part of the Earth, save for the sudden drop in wave velocity at the core/mantle boundary, wave velocities increase as you go downward; but in the uppermost part of the mantle, save for the drop in velocity at the Mohorovicic Disconuity, wave velocities increase even more rapidly than in other parts of the Earth. The reason for this is that as the weight of rocks compressing the rocks below increases as you go downward, not only can the atoms be pushed closer together, but they can be forced into completely different arrangements which are even denser and stiffer than the rocks above. This is analogous to the difference between graphite, a pure carbon compound formed at low pressures which has large inter-atom distances (particularly in some directions) and is relatively low in density; and diamond, a pure carbon compound formed at high pressures which has small inter-atom distances and is relatively high in density. In a similar way, as olivine and similar minerals are compressed more and more, for a while they remain the same minerals but in a denser state; but once the weight compressing them becomes too great they change to denser structures. This happens at two places inside the mantle. At a depth of about 250 miles olivine and similar minerals change to spinel-like structures (not true spinel, which has a different chemical composition) and other denser, stiffer, higher-melting temperature minerals; and at a depth of about 400 miles those structures change to an even denser mixture of magnesium oxide and perovskite-like structures (not true perovskite, which has a different chemical composition). The overall composition of the three regions is believed to be about the same, but the structure of the minerals is different due to the extreme weight of the layers above compressing the less-dense minerals in the upper layers into the denser structures of the layer lowers.
     In that top 400 miles of mantle material the melting temperature, wave velocity and density all increase for the two reasons mentioned above -- the greater and greater weight of material above compressing the minerals and rocks more and more, and the occasional changes from one structure to a denser, stiffer, harder to melt structure. Below that depth, however, the materials are already in the densest possible form for materials of their overall composition, and regardless of how great a weight compresses them they will not change to still denser structures (or at least that is the conclusion drawn from laboratory experiments involving such materials). In the top 400 miles, with two separate factors at work, the density, wave velocity and melting temperatures of the mantle rocks rapidly increase, but below that, with no further structural changes, the rapid increase slows (as shown in the diagrams above).

Application to other planets
     Now, what does this tell us about the other planets? After all, we do not have any samples of most of them (we do have samples from the surface of the Moon, and meteorites give us samples of asteroids and in rare cases, of surface rocks from the Moon and Mars), we have no earthquake studies of any value save for the Earth and Moon, and we have no samples of rocks from the interior of any planet (although as will be seen below we do have partial information for Earth mantle rocks as a result of studies of lavas formed from partial melting of rising plumes). So how can we know what is going on inside Venus or Mercury or Mars?
     The way we know what is going on is to ask, "What would those planets be like if they were identical in structure to the Earth, save for their smaller size"? Since they all have smaller masses, diameters and gravities than the Earth, the weight compressing various parts of their interior must be less than in similar parts of the Earth, and you would have to go downward a greater distance to get the changes that were discussed above. For Venus, with only 90% of the Earth's gravity, instead of having changes from olivine to spinel at 250 miles depth and from spinel to perovskite at 400 miles depth, such changes would occur at 275 miles and 440 miles depth. In other words, there would be a little bit more of the lower density upper mantle rocks, and even if the mantle on Venus was the same size as the Earth's, there would be a little bit less of the higher density lower mantle rocks. But since Venus is 5% smaller in radius and diameter than the Earth, if its structure were a copy of ours its mantle would be 5% smaller as well, or only 1710 miles thick, instead of 1800 miles.
     Combining these results -- a thinner mantle overall and a thicker upper mantle, Venus would have 440 miles of upper mantle material, and 1270 miles of lower mantle material, compared to 400 and 1400 miles respectively, for the Earth; and with more less-dense material and less more-dense material, Venus should be less dense than the Earth if it is otherwise similar in structure.

Comparing the inner planets
     With this in mind, let's take a look at the inner planets and see how they compare. Venus has a density only 95% the density of the Earth. Taking everything we know about the Earth's structure into account and compensating for the 10% lower surface gravity, the 20% lower central pressure (if you'd like to see how we know that number, refer to Internal Pressures of the Planets), the 5% smaller diameter and 20% lower mass, Venus should have a density of 95% of the density of the Earth, if save for differences caused by compressional forces they have identical structures. And that is the density we observe. This does not mean that Venus is a copy of the Earth, but it does mean that if there are any differences in the structure of the two planets, some of those differences must change the density in one direction and others must change them in the other direction, so that the overall density comes out as it does; and to date no structure has been suggested for Venus which is grossly different from that of the Earth, without yielding incorrect density results.
     Now as already noted, the core of the Earth is 55% of the radius/diameter of the planet. We can't be certain what Venus is like, but if somewhat similar to the Earth, it is unlikely that the size of the core differs (in comparison to the overall size of the planet) by more than a few percent, so most models of Venus have a core which is close to 55% of the size of the planet, and a mantle which is close to 45% of the size of the planet (albeit with lower densities throughout and a thicker upper mantle, compared to the Earth). When, at some distant time in the future, we have detailed studies of the internal structure of Venus we may find that these numbers are slightly "off", but any large discrepancy is considered very unlikely.
So what about Mercury and Mars? They have much lower masses than either the Earth or Venus, much smaller sizes and much lower gravities. Mars is only a little over half the size of the Earth (smaller than our core and far less massive, as will be discussed below), so its mantle would be only a little over 900 miles deep if it were a smaller version of the Earth; but since its gravity is over 2 1/2 times less than ours, the 400 mile deep boundary between the upper and lower mantle on the Earth would be over a thousand miles deep on Mars. In other words, if merely a smaller copy of the Earth, Mars would have no lower mantle; the rocks found throughout the mantle would be like those in the upper mantle of the Earth. Obviously, this means (referring to the graph of density inside the Earth, above) that Mars should be even less dense than Venus; and Mercury, which is smaller yet, should be still less dense.
     Now there are reasons for suspecting that Mars and Mercury might not be as similar to the Earth as Venus; but ignoring those for the moment, let's suppose that the above arguments are correct. How low might the density of Mars and Mercury be? Different calculations (again, refer to the Internal Pressures of the Planets for more details) give different results, but we would expect overall densities in the range of 80 to 85% the density of the Earth for Mars and a little less than that for Mercury, if they are smaller versions of the Earth. But unlike Venus, for which the calculated and observed densities are the same (within the errors produced by various assumptions necessary to the calculations), Mercury is much denser than we would expect, and Mars much less dense than we would expect. In other words, neither Mars nor Mercury can be merely a smaller version of the Earth. Still, they are both relatively dense compared to rocks found at the surface of the Earth and Moon, so they must be made of heavy materials that are in some way similar to the interior of the Earth; so how can we explain this odd result?

Review of the Earth's Structure (as discussed above)
     We have already discussed the internal structure of the Earth in detail, as far as the mantle is concerned. Near the surface, the upper mantle consists of (relatively) low density minerals such as olivine, further down the increasing compression of the rocks changes the olivine to higher-density spinel-like structures, and still further down to still higher-density perovskite-like/magnesium oxide mixtures; but according to laboratory experiments, no further change in the structure can be accomplished at any pressure which might conceivably exist inside the Earth; and most importantly, if the entire Earth consisted of such deep-mantle rocks, even though they would gradually become denser and denser as you go deeper and deeper into the Earth, the overall density of the Earth would be nearly 20% lower than observed, if that were all the Earth were made of.
     But that isn't all the Earth is made of, for the structures discussed so far only apply to the rocks in the mantle, and the core of the Earth, as it turns out, is made of much denser materials. The nature of those materials has been a matter of controversy for the best part of a century and even now there is considerable uncertainty about some of the details, but we are virtually certain that the central (inner) core is a crystalline structure consisting of pure or nearly pure nickel-iron, similar to the iron meteorites which fall on the Earth from outer space.

They Came From Space
     (Students: refer to your text and the website discussion of meteorites as background for this)
     Each day the Earth runs into hundreds of tons of small bits of debris left over from the formation of the Solar System, or "recently" lost by other objects, primarily by collisions. Most of this debris is light, fluffy stuff which is vaporized by the heat of its passage (at tens of thousands of miles per hour) through our atmosphere; but a small percentage consists of denser, tougher stuff which if in the right size range can reach the surface, and in a very small minority of cases be picked up and studied in detail. These objects are called meteorites (the -ite ending meaning a rock, and the meteor- prefix meaning that it fell out of the sky).
     The vast majority of meteorites have been established, by a number of lines of evidence, as being pieces left over from the formation of the asteroid belt, or recently broken off of asteroids. There are, as already noted, a very small minority that are pieces blasted off the surface of the Moon or Mars, but although these are of great interest in other circumstances, they are of little importance at this point in our discussion. It is the asteroidal pieces that tell us the most about the inside of the Earth and other "Terrestrial" objects, which includes, as it turns out, at least a large subset of the asteroids which are the parent bodies of the meteorites which were broken off of other objects.
     Some of the asteroidal meteorites are "primitive" objects which, based on their chemical and physical characteristics, have never been exposed to very high temperatures, so the minerals in them are more or less as they were when they first formed, 4.5 billion years ago, during the formation of the Solar System. Most of the meteorites, however, have physical and chemical properties which imply that they were broken off of objects which at some point in their history melted and differentiated just as the Earth did, so that heavy materials sank to the center and lighter materials rose to the top. The vast majority (about 90%) of these "differentiated" meteorites consist of rocky materials very similar to those believed to exist in the uppermost part of the Earth's mantle (the olivine rich portion), suggesting that they came from objects that developed a rocky mantle like the Earth, but like Mars and Mercury were so small that they did not develop the denser mineral structures found at greater depths inside the Earth. In fact one of the original reasons for suspecting this structure for the outer portion of the Earth was the example set by these rocky meteorites, or "stones".
     There is another type of meteorite, however, which is relatively rare (less than 10% of meteorites in situations where selection effects are minor) -- the "irons". These dense metallic meteorites are made of interlocking crystalline structures (see Widmanstatten structures in your text) caused by melt-materials with greater or lesser percentages of nickel crystallizing out of the molten interior of the bodies they were derived from. Most of the material is iron (close to 90%), but variations in the nickel content cause differences in the reaction of the crystals to etching by acids, which make the interlocking structures stand out when properly treated.
     To create such structures requires a molten mixture of iron and nickel slowly cooling over long periods of time. If such a mixture is cooled on times short compared to a human lifetime, the crystalline structures still form, but are microscopic in size. The structures observed in iron meteorites are easily visible to the eye, and require cooling over very long periods of time, probably in excess of a million years; and the only way we can imagine this happening is if a fairly large body (hundreds of miles in diameter) melted, allowing heavy stuff to sink to the middle and light stuff to float to the top, then cooled and solidified. But this is exactly what we think the Earth must have done, and presumably some of the larger asteroids must have done the same in order for pieces of those objects to include such irons; and if so, every Terrestrial object that was more than a few hundred miles in size during the melting episode in the early history of the Solar System must have developed a rocky mantle and an iron core. The stones are therefore interpreted as pieces from the outside of such objects, and the irons as pieces from the inside.

Application of Information From Meteorite Studies to the Earth
     But if that's the case, then the Earth and the other Terrestrial planets must have rocky mantles and iron cores; most likely, if they are solid, crystalline iron cores such as those in the iron meteorites. And the density of those cores would be much higher than that of the mantles, because the materials in the core are intrinsically about twice as dense as the rocky materials in the mantle, and with the entire weight of the planet compressing a core, its contents must be compressed in size and increased in density at least as much as the mantle. So when a calculation of the structure of a planet such as the Earth, based on a strictly rocky structure (as discussed above) gives a density which is too low to explain the overall density of the planet, the solution to the problem must be that the core, consisting primarily of much denser nickel-iron mixtures/crystals, must provide the extra mass required.
     In the case of the Earth, of course, we know exactly how big the core is, and we have a very good idea of how much mass (and deficit of mass, compared to the average density of the Earth) the mantle has. The mantle of the Earth, despite being about 85% of its volume (100% minus the 15% corresponding to cubing the radius of the core), has a mass less than 70% of the mass of the Earth. The deficit of mass must be made up by the core, which despite being only 15% of the volume, must have more than 30% of the Earth's mass -- and as a result, twice as much mass as Mars and Mercury put together. This means that whatever the core is made of can't be rare stuff like gadolinium or ruthenium, but must be common stuff like iron (one of the most abundant metals in the Universe), which agrees with the evidence from meteorites, and allows us to more or less confidently say that the core must be "just like" the iron meteorites.
     Of course, before it was possible to achieve pressures in the laboratory equal to those in the center of the Earth, it was conceivable that the higher than expected density of the Earth might be explained by yet another change in the structure of the materials, such as the olivine - spinel - perovskite changes in the mantle; but now that we can achieve such pressures, we know that no further such changes are likely to occur in the mantle, or to explain the Earth's high density. And if we test nickel-iron mixtures, we find that at the pressures which must exist in the core of the Earth, they have just the right density (and other properties such as wave velocities) to explain the density and structure of the Earth.
     However, we don't believe that the entire core of the Earth is a crystalline mixture of nickel-iron similar to the iron meteorites. As shown in the temperature diagram and in the wave-velocity diagram, the outer core of the Earth must be a liquid, or given the materials and temperatures involved, a molten material (molten meaning a liquid form of a normally solid material that requires very high temperatures to melt); and the molten outer core of the Earth cannot be made of exactly the same materials as the solid inner core, because the temperature of the Earth must be higher the further down you go (heat only goes from hot places to cold places, and since it comes from inside the Earth and escapes at the surface, the temperature of necessity must increase all the way down, regardless of any details that we wouldn't want to consider in an introductory discussion and although there is a lot covered here, so it may seem complicated, this is actually a very simple discussion compared to what a student majoring in astronomy, geology, planetary geology or geophysics would encounter).
     Now if the inner core of the Earth must be hotter than the outer core (at least 12000 Fahrenheit degrees, which is hotter than the surface of the Sun, based on the best current estimates), and despite that the inner core is solid, there must be some reason for that. One possible reason is that the greater weight compressing the inner core might raise the melting temperature; but this is unlikely to be the case. First, if that were true the boundary between the inner and outer core would be a broad transition zone, with molten material further out, partially molten and partially solid material further down, and more nearly completely solid material still further down; and the way in which earthquake waves penetrate the inner/outer core boundary tells us that there is a relatively sudden transition from liquid to solid. Second, as you near the center of the Earth, the weight of the materials drops to zero (see the afore-mentioned page on Internal Pressures of the Planets for a detailed discussion), because they are pulled equally hard in all directions by the various parts of the Earth all around them, so those forces cancel out, and the change in pressure is surprisingly low in the central regions of the Earth.

Near the surface (on the right) the weight compressing the interior rapidly increases as you go down; but near the center (on the left) the weight hardly changes, because as you near the center the weight of the materials drops to nearly zero (even though their mass is the same as if they were at the surface of the planet).

     Since the outer/inner core boundary is sharp and the pressure on either side of the boundary is nearly equal (given the nearly flat curve for pressure near the center, as shown just above), there is only one way to explain the solid nature of the inner core and the molten nature of the outer core: the inner core must be made of a material that has a higher melting temperature. And the easiest (and based on laboratory experiments and theoretical calculations, the best) way to explain this is if the inner core is a crystalline structure consisting of nearly pure nickel-iron, as for the iron meteorites, while the outer core is a mixture of similar materials with additional materials that depress the melting temperature in the same way that salt depresses the melting temperature of ice (pure ice melting at 32 Fahrenheit, and very salty ice melting at 0 Fahrenheit, so that at intermediate temperatures pure ice can exist as a solid floating in a salty mixture of water and other materials, as in polar seas). The most likely candidates for the impurities that depress the melting temperature are silicon and oxygen, because those are the basic building blocks out of which the silicate materials in the mantle are formed, and lesser amounts of other materials. Different investigators favor different ratios of impurities to nickel-iron, leading to estimates of 10 to 30% other materials and 70 to 90% nickel-iron in the molten outer core (it is the uncertainty of these values that causes the large uncertainty in the temperature of the core); but no matter what the ratio is, the large amount of dense materials such as iron in the core give it the required extra mass to make up for the deficit of mass in the mantle, compared to the overall density of the Earth.

Volcanic eruptions and the composition of the mantle
     One additional topic which should be briefly mentioned is that it is possible to directly test certain aspects of the theory that the outer portions of the Earth are at least vaguely similar to the stony meteorites. As already mentioned, the heat difference at the core/mantle boundary causes hot, relatively soft rock to rise toward the surface. As they approach the surface, the rocks cool, because at their (very slow) upward rate of motion it takes a long time for them to cover the 1800 miles from the lower mantle to the surface; but they always remain close to their melt temperatures (in the temperature graph at the top of this page, the black line represents an estimate of melt temperatures at various depths, and the blue line represents an estimate of rising plume temperatures), so when they reach the region near the Mohorovicic Discontinuity where the difference between normal and melt temperatures is small, the rising rocks are even closer to melt temperature, and lower-melting-temperature minerals within the rising plumes often do melt, and force their way to the surface. This is the source of most lavas expelled at the surface of the Earth. In some places, such as the west coast of the Americas, where continental plates ride over and push oceanic plates into the deep interior, friction between the plates can heat the downward-moving rock, partially melting it and causing volcanic eruptions. The rocks that melt in this way have very little gas or liquid trapped inside them, and the lavas are stiff and tend to produce explosive eruptions. At most places on Earth, however, fresh lava is produced from the melting of upward-moving mantle rock, and this lava always contains large amounts of dissolved gases, which make it very fluid and cause "fountain"-type eruptions such as those in the Hawaiian islands and Iceland (both being regions located above rising plumes of hotter than usual mantle rocks). The composition of these lavas and their gases provide important clues to the composition of the source material, which is of course mantle rock, and the results strongly suggest that mantle rock is indeed very similar in composition to the stony meteorites.

Summary, and Application to the Other Terrestrial Planets
     For an object like the Earth (or Venus, since it appears to be similar to the Earth, as already discussed) we have an outer rocky mantle consisting of various forms of ferromagnesian silicates, depending upon the pressure compressing the material at various depths, and an overall density 20 to 25% less than the actual density of the Earth; and an inner primarily metallic core consisting mostly of iron, but with lesser amounts of nickel and in the outer core additional materials, which has an overall density two to three times that of the Earth. The much greater excess of density in the core, compared to the small deficit of density in the mantle, may be surprising, but keep in mind that the mantle has nearly six times the volume of the core, so that every 1% deficit in the mantle requires a more than 6% excess in the core.
     At this point, we can essay a guess as to why Mars is about 20% less dense than expected and Mercury nearly 25% denser than expected if they were merely smaller copies of the Earth. Namely, Mercury's dense metallic core must be a much larger proportion of its volume and mass, and Mars' core must be a much smaller proportion of its volume and mass, than for the Earth. Whereas the Earth has a core which is 55% of its overall radius and Venus must have a core fairly close to that (certainly no less than 50% or more than 60% of the overall radius), Mars may well have a core less than a third the size of the planet, and Mercury's core is probably 70 to 80% of the size of the planet. So we can summarize by saying that the Earth and Venus have "average" size cores, Mars a "small" core, and Mercury a "large" core; but in saying that, keep in mind that the core of the Earth is bigger than all of Mars, and more than twice as heavy as Mars and Mercury put together; so although Mercury's core is large in comparison to the overall size of that planet, it is still very small in comparison to the core of the Earth.
     These numbers are of course uncertain. We cannot know, at present, the exact nature of the rocks in the mantle of either Mercury or Mars (or for that matter, Venus), and if they are significantly different from the rocks in our mantle, that might affect the mass deficit of the mantle to a significant extent, and change the size the core of the planet has to be to make things come out right. For Mars the chances that the mantle rocks are significantly different from those in the asteroids (the source of the stony meteorties) or the Earth are relatively small; but for Mercury, we might well find (when we can finally find out anything about the structure of Mercury from direct observation) that the mantle is different from ours in various ways, so the actual size of the core is a bit uncertain (although most likely, closer to the description here than not).
     Note added in December 2014 during editing of this page for typographical and grammatical errors: Thanks to the study of Mercury by the MESSENGER spacecraft in the years since this page was last updated, we now know far more about the structure and geological history of Mercury than implied in the previous paragraph; so a (separate) discussion of the structure of Mercury will add considerable detail to the discussion here. However, the overall results discussed above are, within the uncertainties stated, still correct. Similar improvements in the detailed structure of Mars have also been obtained by gravimetric studies of the planet by spacecraft orbiting the planet; and some of those details are already included in the page about the structure of Mars. However, no improvement has been made in our knowledge of Venus, because there have been no spacecraft sent to the planet in recent years.

Internal structures of the Terrestrial planets. For Venus the core must be similar to that of the Earth in comparison to its size, but for Mercury the core is relatively large in comparison to its size, and for Mars the core is either relatively small (as shown here), or made of lighter than usual materials (in particular, it appears that Mars' molten outer core may have considerably more sulfur than the Earth's). Note that although Mercury's core is large in comparison to the size of that planet, it is not large compared to the Earth's core. The Earth's core is larger than Mars and more than twice as massive as Mars and Mercury combined. (Image based on NASA image of some age; as a result, more recent models of Mars and Mercury show more detail)