The Cambridge Encyclopedia of Earth Science claims what generates the Earth's magnetic field is not understood. Here I suggest a simple model. FACTS: 1. The field is roughly a dipole pointed approximately North, and at previous times in Earth history has pointed approximately South instead. North and South (with respect to rotation axis) appear to be stable states, but pointing in other directions (or nonexistent field) appears to be an unstable state which quickly vanishes and is replaced by one of the two stable states. This historical record is revealed by magnetism trapped in volcanic rocks at time of formation, versus that time. You can see a graphic in http://en.wikipedia.org/wiki/Geomagnetic_reversal which shows 9 N-pointing and 9 S-pointing periods occurred during the last 5 Million years, each with apparently-random durations. There may also have been additional switchovers with short durations. The short ones are hard to detect. The best known such short-duration event was the "Laschamp event" in about the year 39000 BC, a reversal lasting only about 440 years with the actual change of polarity lasting around 250 years. The reversed field was 75% weaker and the strength dropped to only 5% of the current strength during the transition. This resulted in greater radiation reaching the Earth, causing greater production of beryllium 10 and higher levels of carbon 14. Perhaps these short events are best regarded as failed attempts to reverse -- we drop to a low field state, and then that state is attracted to the two N- or S-pointing stable states. If attracted back to the original one then we get a "failed reversal" instead of a genuine reversal. The switchover-times are each <=50,000 years. There was at least one 40 million year span containing no reversal, but overall average has been about 2.2 reversals per million years. 2. The Moon has only tiny magnetic field, mostly a few nanoTesla. (Versus: Earth's is 25-65 microTesla.) Comparably-tiny fields: Mars, Venus. Intermediate strength: Mercury. Jupiter's field is 10X Earth's. Saturn also has a strong field. Uranus and Neptune have large magnetic fields, but they are not well axis aligned. MY MODEL: The field is generated by a "dynamo" from convection currents of hot electrically conductive fluid inside the planet. Moon and Mars have no field since they have no such internal convection currents, they are geo-tectonically dead in the sense they have no volcanos (although Mars at an earlier time did have volcanos) and no plate tectonics, indicating internal convection currents are much weaker and less-well defined (i.e. tend to cancel out over time averaging). Venus still has active volcanos, but its lack of magnetic field will be explained below. Mercury's, Saturn's, and Jupiter's fields are presumed to be generated by same mechanism as Earth's. The more gravity and more internal heat a (non-dead) planet has, the more powerful its convection and hence more powerful its fields -- also axis alignment more-perpendicular to planet-sun direction helps generate more field for reason explained below -- all explaining why Mercury<Earth<Saturn<Jupiter. Now my claim is that the convection currents flow outward from core toward equator. The fluid then cools, and drops back to the core from the poles. This requires that (and is predicted by) the poles tend to exert more cooling effect than the equator. If so, then a N- or S-pointing magnetic field, will induce (via Lorentz force) an equator-circling electrical current, This in turn will reinforce the field. Result is a stable "dynamo" powered ultimately by internal heat (radioactively generated and/or left over from planet formation). For Earth, situation is complicated by the presence of oceans over thin crust, and continents over thick crust. Presumably there is more cooling in the ocean regions and/or volcanic regions. Also there are "hot spots" such as Hawaii. Also the Earth's rotation axis precesses and is not exactly perpendicular to the Earth-sun line. All this complexity, plus fluid "turbulence," presumably has something to do with why we get occasional reversals. I would presume the field continually fluctuates and whenever (rarely) a fluctuation is comparable to the field itself, then we are ready for either a reversal or failed-reversal. In the case of Venus, there are no plate tectonics (although there is volcanism) and there is a huge atmosphere which redistributes heat efficiently, meaning the equator may not have a much greater cooling effect than the poles. For Uranus and Neptune the sun is so far away that its thermal effects are pretty negligible, explaining why their fields are not axis aligned (and Uranus has a huge axial tilt too, so even if the sun's heat non-negligible, that'd still mostly cancel out its effects). SO??? That seemed pretty easy. Why wasn't this thought of ages ago? Too shed light on that, let's do some crude numerical calculations. Assume convection speed of v = 1 cm/year -- same order as continent drift speeds. Total voltage induced circling equator (25000 mile circumf.) then is of order v * (50 microTesla) * (25000 miles) = 6.4 * 10^(-7) volts. Assuming 40 ohm*meter electrical resistivity for lava (this was a Hawaii measurement) the earth going round the equator is about a 10^(-5) ohm resistor. [However it actually is found that high pressures and high temperatures both act to decrease resistivity for molten rock apparently like Arhennius law exp(const/T), plus there are stairsteps caused by temperature surpassing ionization-energy thresholds.] So this voltage would induce a current circling equator of order 0.06 Amps. That in turn would cause a magnetic dipole moment of about 8*10^12 * meter^2 * ampere which is 10^10 times smaller than the actual Earth dipole moment 8*10^22 * meter^2 * Ampere. Oops. It seems still within the reasonable realm that the resistivity might be only 1 ohm*meter and the flow speed might be 10 cm/year. This would increase our predicted dipole moment by a factor of 40*10=400. The flow speeds necessarily will be larger still nearer the center of the Earth due to volumes being smaller there, and the magnetic fields there might be much larger due to larger currents at smaller distances. So assuming both 10X greater flow speed and magnetic field down there, we could get another factor of a few hundred. But this is still well short of the required 10^10 factor. CONCLUSION: The model sounded great for a little while, but it is predicting too small fields compared to observation. What went wrong?? -- Warren D. Smith http://RangeVoting.org <-- add your endorsement (by clicking "endorse" as 1st step)