Letters from the SkySnow crystals, grown here from water vapour

Why six?

In 1611 Johannes Kepler gave his patron a small book for the new year, On the Six-Cornered Snowflake, asking why every snowflake has six points. He thought about stacked spheres and could go no further. By 1929 X-ray studies had shown how water molecules lock together in ice: in puckered hexagonal rings, layer on layer.

Grown slowly, ice takes the shape of its own lattice, a hexagonal prism. Here a single cell of ice sits in air at −15 °C that holds a little more vapour than ice can bear, and grows into a plate.

The snowflake man

Wilson Bentley, a farm boy from Jericho, Vermont, fixed a microscope to a bellows camera and in 1885, aged 19, made the first photograph of a single snow crystal. He kept at it for 46 years and more than 5,000 crystals; the book he made with William Humphreys in 1931 holds over 2,000 of them, white on black.

This is the same model drawn in his style. The fine lines inside the arms are ridges where ice piled up a little thicker as the edge swept past.

Why it branches

A crystal feeds on vapour that diffuses in from the air around it, and it eats faster than the air can refill. The vapour thins out around the crystal. Any corner that sticks out reaches into richer air, grows faster and sticks out further. W. W. Mullins and R. F. Sekerka worked out this runaway in 1964; on a snow crystal it turns the six corners of a plate into six arms, and every bump on an arm into a side branch.

This view shows the vapour: dark where the crystal has drawn it down, bright where the cloud is still full.

Why it has flat faces

Branching has an opponent. A molecule landing on a flat face of ice has nowhere to hold on, so flat faces grow slowly and stay flat. When vapour is scarce the facets win and you get a hexagonal plate; with more vapour the corners outrun them, first as ribs and sectors, then as branches, then as ferns.

Move the humidity while the crystal grows. Each change is written into the edge it reaches.

Nakaya’s diagram

On 12 March 1936, in a cold room at Hokkaido University, Ukichiro Nakaya grew the first snow crystal in a laboratory, on a single rabbit hair. Varying temperature and humidity, he found that the shape depends sensitively on temperature. His diagram, later extended by Kobayashi (1961), runs: plates near −2 °C, columns and needles near −5, plates and stars again from −10 to −22, columns below that.

Tap the diagram to grow a crystal there. In the column bands the crystal grows along its axis; switch to 3D to see it from the side. Nobody has fully explained why the habit flips with temperature like this; it is still studied today.

Letters from the sky

A falling crystal passes through air of different temperature and humidity, and each stretch adds a different kind of growth to its edge. Nakaya saw that you can read the journey back from the shape. This view colours each part of the crystal by the kind of growth the air called for when that part froze, with a faint ring every equal stretch of time.

The six arms come out alike for the same reason: all six made the same journey at the same moment.

Twins and strangers

In the model so far every arm is identical to the last digit, because nothing distinguishes one from another. Real air is never that even. Add a little turbulence, small random flickers in the vapour from cell to cell, and the arms begin to differ in their details while keeping the same overall plan.

Two crystals that fell side by side through the same air can look much alike. Two that took different paths almost never will, because the number of possible journeys is enormous.

Your cloud

Tap or drag on the cloud chart under the crystal to change the air it grows in, at any moment. Or draw a path and let the crystal fall along it.