MERIDIAN

Nikolai Vetrov keeps his drafting table at the window because the light off the Barents Sea is, he says, the only honest light there is. It arrives flat and grey for eleven months and then, for a few weeks, turns the ice the colour of a peach.

He has been mapping this coast since 1985, first for a bureau that no longer exists, then for a university that stopped paying him in 2004, and since then for nobody at all. The maps are still made. They are stacked in flat files that reach the ceiling.

What changed is not the method. It is the coast. Permafrost that held the cliffs together has been letting go for two decades, and each spring returns a shoreline that is a few metres shorter, a few metres softer, than the one he drew the year before.

"People think a map is a photograph," he says, sharpening a pencil with a scalpel. "It is closer to a diary. It says: this is what I understood on that day."

He does not use satellite data. He has seen it, and he accepts that it is more accurate than he is. But accuracy, he argues, is not the point of the exercise, and never was.

Somewhere in the flat files is a 1991 sheet showing a spit of land that ran two kilometres into the sea. It is gone now. The only remaining evidence that it existed at all is a rectangle of paper in a room in Murmansk, and one old man who remembers walking to the end of it.