Daily Learning Experience · Science
You are about to weigh the whole Earth using two lead balls, a piece of wire, and the patience of a saint.
Clapham Common, London · 5 August 1797
You are Henry Cavendish
It is just after dawn and you are standing on wet grass in your own garden, forbidden to enter your own shed.
Inside it, in a sealed room ten feet high and ten feet across, a six-foot wooden arm hangs from a thin silver wire. On each end of the arm sits a lead ball the size of an orange. Beside the case, hung on a frame you can swing from outside with a rope, are two more lead balls — a foot across, three hundred and forty-eight pounds each, the weight of two grown men apiece.
You may not go in. Your body is warm. A warm body stirs the air, and stirred air in that room is stronger than gravity. So you kneel at a hole in the wall, put your eye to a telescope, and watch a strip of ivory that will never move more than the width of two hairs.
You are sixty-six years old. You are probably the richest scientist in England and you will barely speak to a living soul — your servants receive their orders as notes left on the hall table. And over the next ten months, alone, in the dark, you are going to find out what the world weighs.
“I resolved to place the apparatus in a room which should remain constantly shut, and to observe the motion of the arm from without, by means of a telescope.” — Henry Cavendish, 1798
The problem
Every object pulls on every other object. You are being pulled, right now, by your chair, by the wall, by the person next to you. You have never once noticed, because gravity is stunningly, absurdly weak.
Here is what Cavendish was hunting. A 348-pound lead ball, sitting nine inches from a 1.6-pound lead ball, pulls on it with a force of about 0.00000015 newtons. Put that on a set of kitchen scales and it would register the weight of fifteen millionths of a gram.
A grain of table salt is four times heavier than the signal he was chasing. And he had to detect it in a wooden shed, with no electricity, no computer, and no way of shutting out the weather except a locked door.
The instrument
This is your bench. On the left, looking straight down into the locked room: the mahogany case, the hanging arm, and the two great weights on their swinging frame. On the right, what your eye actually sees — the ivory scale through the telescope, lit by a lamp, magnified until a hair's breadth becomes a hand's breadth.
Pull the rope to swing the big weights to one side. Wait. The arm will begin, very slowly, to turn. Record where it settles on each side, time one full swing, and you will have the two numbers you need.
Looking down into the locked room
Through the telescope · ivory divided to 1/20 inch, vernier to 1/100
Swing the weights to one side, wait for the arm to stop swinging, then record. Do it on both sides.
The data
Cavendish did not do this once. He ran eight experiments between 5 August and late September 1797, changed the wire for a stiffer one, and ran nine more the following April and May. Out of them came twenty-nine separate determinations of the density of the Earth, and they are all slightly different — because every real measurement is.
Here they are. Every dot is a night's work.
Cavendish's 29 determinations of the density of the Earth (1798)
Each dot is one determination. The line is the mean.
Watch the third dot. In his own table Cavendish wrote down 4.88, then did his arithmetic as though it said 5.88. That single slip is why the paper announces 5.48 rather than 5.448 — and it sat there, in print, in the most careful experiment of the eighteenth century, for twenty-three years before Francis Baily caught it in 1821. Both answers are within about 1.5% of the modern value. Being careful and being right are not quite the same skill.
The road not taken
Twenty-three years earlier, the Astronomer Royal had tried something magnificently strange. Nevil Maskelyne took a team to Schiehallion, a lonely, steep, beautifully symmetrical mountain in Perthshire, and hung a plumb line beside it. If the mountain had mass, it should tug the plumb bob sideways. Measure the tug, weigh the mountain, and you can weigh the Earth.
It worked. The lean was 11.6 arcseconds — about three thousandths of a degree, roughly the width of a coin seen from four miles away. Drag the slider to make it visible, then drag it back to the truth.
To turn the lean into a number, Charles Hutton had to know the volume of an entire mountain. He covered a map in thousands of surveyed heights, then — out of sheer desperation — drew lines joining points of equal altitude so he could see the shape. He had invented the contour line, which is now on every map you have ever read. The mountain gave an Earth density of about 4.5 times water. Cavendish, with two lead balls in a shed, got 5.48. The lead balls won.
Still open
Cavendish never wrote down the number we now call G, the gravitational constant — that idea arrived about seventy-five years after him. But his measurement contains it, and when you convert his result into modern units you get G ≈ 6.74 × 10⁻¹¹, which is within about one percent of today's best value. In 1798. In a shed.
Here is the uncomfortable part. Physicists have been remeasuring G ever since with lasers, atom interferometers, and torsion balances Cavendish would recognise instantly — and the answers still refuse to agree with each other. G remains the worst-known of all the fundamental constants, roughly a hundred thousand times fuzzier than the constant that governs electricity.
Selected measurements of the gravitational constant G
Bars show each experiment's stated uncertainty. Hover a point for detail.
Cavendish's 1798 result (≈6.74) and Boys' 1895 result (≈6.66) sit off the top and bottom of this chart. In April 2026 a team at NIST led by Stephan Schlamminger published a new torsion-balance value after ten years of work — and it landed in a slightly different place again. “The results emphasize how difficult it is to measure this gravitational constant,” he said.
The thread
Tap a card.
Do this today
You will not measure the Earth with this. You will do something better: you will feel, in your hands, exactly how hard Cavendish's problem was.
You need: a metre of sewing thread, a drinking straw, two coins taped to the straw's ends, sticky tape, and a large glass bowl or a cardboard box to put over it.
Tape the thread to the middle of the straw and hang it from a doorframe so the straw floats horizontally. Cover it with the box or bowl — cut a window so you can see in. Now wait five minutes for it to stop turning, and mark where one end of the straw points. Then bring a full 2-litre bottle of water slowly up to one coin, as close as you can get without touching, and wait. Watch for ten minutes.
Almost certainly nothing will happen — the thread is far too stiff and your coins far too light. Then take the cover off and simply stand near it, and watch how violently your own warmth swings the straw around. That is the ghost Cavendish spent ten months hunting out of his shed. Now imagine finding a real signal underneath it.
Mind-blown
Examination
Seven questions. No talking — Cavendish would hate it.