Gilbert's Terrella

Use a spherical magnet 40-60 mm across, or fix two strong magnets inside a wooden ball with their axis through the centre. Gilbert used a turned sphere of natural lodestone; the geometry is what matters, not the material.

Strong magnets damage cards, drives and hearing aids, and they pinch hard when they snap together. Work well away from screens and phones, and never let two large magnets close under their own attraction.

Bring a small compass near the sphere and find the two opposite points where the needle points straight in or straight out. Mark them. Those are the magnetic poles of your little Earth.

Draw the equator halfway between the poles and a few meridians running pole to pole. Now the sphere is a globe and you can say where on it you are measuring.

Sit the sphere in a non-magnetic ring — wood, cork or plastic — so it stays put and every part of its surface is reachable. Steel anywhere near it corrupts every reading.

Magnetise a needle by stroking it 30 times in one direction with a magnet, push it through a slice of cork and float it in a shallow non-metal dish of water. This is the horizontal compass — the versorium of Gilbert's experiments.

Balance a second magnetised needle on a horizontal pivot — pushed through a cork sitting on two pin supports — so it can tip up and down in a vertical plane. This is the instrument that carries the real argument.

Get the dip needle balanced level while it is still unmagnetised. Any tilt left over is gravity, not magnetism, and it will masquerade as a reading later.

Carry the floating compass slowly across the terrella. Wherever it goes it lines up along a meridian, pointing toward the same pole. This is the behaviour of a ship's compass, reproduced on a sphere in your hand.

Hold the dip needle just above the equator and record its angle. It should sit close to horizontal — the field runs parallel to the surface there.

Take the dip at the equator, a quarter of the way up, halfway, three quarters and at the pole, writing each angle down. Five numbers are a result; one is an anecdote.

Plot your angles against position. Dip rises from about horizontal at the equator to vertical at the pole. That curve is the whole of Gilbert's case, drawn from your own measurements.

Look up the magnetic dip where you live and compare it with the reading at the matching latitude on your sphere. The agreement is the argument: the Earth behaves like this sphere, so the Earth is a magnet like this sphere.

Now put a single small magnet at the north pole of a non-magnetic ball and repeat the dip measurements. The pattern is wrong — dip changes far too abruptly near the magnet and stays near zero elsewhere. You have just falsified the rival hypothesis with an experiment rather than an argument.

De Magnete, 1600. William Gilbert, an English physician, published De Magnete in 1600 — a treatise on magnets, magnetic bodies, and "that great magnet the Earth". It contains some fifty distinct experiments and it argues from them rather than from authority, which is why it is so often named as one of the first genuinely experimental works of science.

The terrella. Gilbert turned lodestone into spheres he called terrellae, little Earths. Carrying a compass across one, he showed the needle aligns north-south everywhere on the surface; carrying a dip needle — balanced on a horizontal axis so it can tip — he showed the inclination varies with position on the sphere exactly as magnetic inclination varies with latitude on Earth. From that correspondence he concluded that the Earth is itself a great magnet.

What it replaced. The prevailing explanations were that the compass was drawn to the pole star or to a magnetic island or mountain in the north. Gilbert's model does something those cannot: it reproduces the whole pattern of compass behaviour — direction and dip together, at every latitude — rather than just the fact that needles point north. That is the strength of a model argument, and it is why the dip needle is the crucial instrument rather than the ordinary compass.

What he got wrong. Gilbert believed the Earth's magnetism was a permanent property of its solid substance. It is not: it is generated by convecting molten iron in the outer core, and it reverses at irregular intervals of hundreds of thousands of years. Being wrong about the mechanism does not damage the finding — that the Earth is a magnet is correct, and it was established by building a model and measuring it.

The word persists. Physicists still call a magnetised sphere used to model planetary magnetism a terrella, and such devices are still used to study aurorae. It is a rare case of a four-hundred-year-old piece of apparatus keeping both its name and its job.