Measuring a coastline means first choosing a ruler, which sounds like a technical footnote. A ruler 100 kilometers long gives Britain roughly 2,800 kilometers of coast. Halve the ruler and the answer climbs to roughly 3,400. So how long is the coast?
The gap is not measurement error. A shorter ruler drops into bays and inlets that a longer one steps straight across, so the measured length rises every time the ruler shrinks, and it never settles on one value. Lewis Fry Richardson noticed it first. 1967 paper, titled with exactly this question, turned it into the opening move of geometry.
The question sounds like a matter of fact and is really a trap. The length of Britain’s coast is not one number waiting to be measured. It is a function of the ruler you picked before you started.
The mechanism is plain. A 100-kilometer ruler cuts across bays and shortens the trip, landing near 2,800 kilometers. A 50-kilometer ruler drops into some of those bays and reaches about 3,400. Shorten the ruler again and the figure rises again, and it does not converge on anything.
Lewis Fry Richardson ran into this while working on the causes of war. He was comparing the lengths of shared borders and found that neighboring countries published wildly different numbers for the same border. Mandelbrot took that buried observation and turned it into a claim about shape: a coastline is a fractal, about equally rough at every scale.
What comes out of it is a measuring tool and not only a paradox. Plot ruler length against measured length on a log scale, get a straight line, and read its slope as the , a number for roughness. It also warns you off a whole class of statistic: a length that is means nothing until the resolution is published next to it. In the physical world the growth does stop: below the size of a sand grain or a single wave the question has no physical meaning left.
wat. Halve the ruler measuring Britain’s coast and it drops into more inlets, so the coast gets longer and no single length exists.
wat wat
It is why two official sources can publish very different coast lengths for the same country without either of them being wrong.
How do we know?
Plot ruler length against measured length on a log scale. The points fall on a straight line, and its slope is the fractal dimension of that coast.
What is still unknown
In the physical world the growth stops somewhere. Below the width of a grain of sand or a single wave the question stops meaning anything.
Learn something?
Where to next
Is a nautilus shell really built on the golden ratio?measurement
How is the distance to the Moon known to the millimeter?measurement
Why does lightning return again and again to the same corner of Venezuela?geography