Installing Slate in Snow and Freeze–Thaw Climates: What Thirty Winters Taught Me About Ice Dams, Snow Load and the Cold

I've been putting slate on roofs in Norway, Sweden and the Alpine valleys for thirty years — winters, I should say, because in that part of the world a roofer is hired by the season, not the year. If you're about to install a slate roof somewhere the snow settles and the nights freeze, here is the one thing I tell every client before the first tile goes up: the roof you build will be tested every single night from December to March. Not once. Every night.

Here's what one of those nights looks like. At eleven in the evening it's minus twelve, and the slate is frozen solid. At noon the next day the sun finds the south slope and the surface of the roof warms to maybe plus six — the snow on top of it melts, but the batten underneath and the nail inside it are still at minus three. By four o'clock the shadow crosses the roof and the whole thing freezes again. That's one freeze–thaw cycle. A slate roof in a snow climate goes through something close to that most days of the winter, and the cycle lands on it twice a day in the worst months — more freeze–thaw cycles in one season than a roof in the south of France sees in a decade.

So the honest way to think about a snow-country slate roof is as four seasons in one winter. The snow falls, the roof freezes and thaws, the eaves build a wall of ice, and then the snow comes off in the spring. Each of those four acts does something different to the roof, and each one needs a different installation decision. I'll walk them in the order the winter meets them.

Short version first. A snow-country slate roof lives on four rules. One: the slate itself must have passed a freeze–thaw test with almost no weight loss — a slate that drinks water will crack in the cycle, and no installation trick fixes that. Two: the roof is built cold and ventilated — the colder the roof surface, the less snow melts and the fewer ice dams you ever see. Three: the eaves get a waterproof backup layer, because the one leak that happens in snow country is not the slate's fault, it's the ice wall's. Four: the snow is given a place — deeper headlap, higher pitch, and snow guards where the snow will slide. That's the whole job in one paragraph; the four sections below are the detail.

The first snow: the roof that carries the winter

The first snow of the season is a load test, and most people get this one wrong. Fresh, dry snow is light — around 50 to 100 kg per cubic metre. But by the end of the winter the snow on your roof is the wet, packed kind, the sort that sits at 300 to 400 kg per cubic metre, and on a 35-degree roof that's a serious weight sitting on the slates. Slate handles it easily: a good 5 mm slate laid on proper battens carries far more than any snow can throw at it. The snow load is not the threat. What the snow does is seal the roof — and that's where the trouble starts.

When the sun comes out, the surface of the snow melts; the meltwater runs down under the snow, along the headlap of the tiles, and sits in the laps. That's a different kind of water test than rain. Rain runs down the outside of the slate and off the face. Meltwater underneath snow runs under the laps, sideways, in the one direction a slate roof is designed to keep water. So in snow country I do two things that matter more than any other installation habit. I keep the roof steep — for a snow climate I don't like anything below 30 degrees where the snow settles (the minimum pitch guide gives the rain floor, snow wants more). And I add to the headlap: in a snow village I use 25 mm more than the sheltered standard, because the meltwater works the lap from inside, not outside. The calculation method is all in the headlap calculation guide — here I'll just say the snow doesn't negotiate.

The freeze–thaw cycle: how the frost works on the slate at night

Now the second stage of the winter, and the one that shortens more roofs than all the others put together. A slate roof in a snow country doesn't usually die of old age. It dies of water that froze inside a tiny crack. This is the slow work: a slate with slightly too high water absorption takes in a little water, the water sits in a hairline crack, the temperature drops below zero, the water expands by about nine percent and pushes against the crack. Every cycle opens the crack a little more, and in the winter the roof is cycled every night. That is the freeze–thaw cycle doing its work, and it is exactly why the lab test matters so much in a cold climate.

The honest check for a snow-country slate is not the brochure, it's the freeze–thaw test: the sample goes through hundreds of freezing and thawing cycles in a machine and the lab weighs it at the end. A slate with low absorption loses almost nothing — our own Jiujiang slate, for example, sits at around 0.2% water absorption and comes out of the freeze–thaw test with a weight loss under 0.1% (the full figures are on our test results page). A slate that loses 1% or 2% of its weight is losing its surface, and in a cold country that is a ten-year roof, not a hundred-year one. The deep dive on the test and how to read it lives in the freeze–thaw resistance guide — the short version here is that in a snow country, freeze–thaw performance is a ticket to enter, not a nice-to-have.

Installation can't fix a slate that drinks water. But installation can make sure it doesn't make the frost's work any easier, and there are three habits I drill into every team that works with me. First, the nail holes and all the cutting are done from the top side, with clean edges — a ragged punched hole is a little pocket where water collects and frost works on. Second, every nail head sits flat, not proud: a proud head lifts the tile by a fraction, and a fraction of a millimetre is a space where water can sit and freeze. Third, the side joints stay as tight as the material allows — a wide side joint is where the meltwater pools, and a pooled meltwater freezes hard. The geometry of that is in the joint spacing guide; here I just want you to know that in a freeze–thaw climate, the joints are a winter matter, not a looks matter.

And the fixings. A nail in a snow-country roof is doing two shifts: it holds the tile, and it holds a wet, freezing batten that swells and shifts all winter. So I only ever fix snow-country roofs with copper or stainless steel — a nail that rusts is a nail that loosens, and a loose nail is where the next frost finds its purchase. The fixings methods and metal choices are covered in the nails and hooks and copper vs stainless guides — the snow-country conclusion in one line: don't let the cheapest item in the roof decide the winter.

The ice dam: why snow-country leaks come from the eaves, not the middle

Now the third stage of the winter, and the one that sends the phone calls to the roofer. If a snow-country roof leaks in the middle of the winter, the cause is almost always the same thing, and it is not the slate: it's the ice dam. This is what happens. The house leaks heat, the attic stays warm, the roof surface above the snowline melts the snow on the roof. The eaves, hanging out in the cold air, don't melt anything — they stay at outside temperature. Meltwater runs down the roof, meets the frozen eaves, and freezes. A wall of ice builds up at the bottom of the roof. The dam grows taller. And then the meltwater pools behind the ice wall, and the pool — not the snow, not the rain — pushes the water up between the laps and back under the slates, and the leak arrives at the ceiling.

Here's the thing I say to every owner who shows me a ceiling stain: the slate did not fail. The nails did not fail. The temperature failed. An ice dam is a temperature problem, and it's fixed underneath the slate, not at the lap. Four pieces. One: the roof is ventilated cold — a counter-battened, vapour-permeable assembly so the roof surface sits at roughly outside temperature (the full construction is in deck, battens and underlayment); when the roof is cold, the snow doesn't melt on the slope and the ice dam never forms. Two: the bottom courses of the roof get a self-adhesive waterproof backup membrane — the ice-and-water shield — over the deck for at least the first metre, so even if a dam forms one spring, the backed-up water can't get in. Three: the insulation goes to the house, not the roof — the heat is the problem, so keep the heat where it belongs. Four: and this is the counterintuitive one, let the snow move — a steeper pitch and a working headlap mean the snow leaves the roof in the winter instead of sitting on it and forming the next dam. The eaves details, the metal and the flashing, live in the eaves, ridges, hips and valleys guide — but please don't go looking for the fix in the metalwork. The dam is a temperature problem, and you fix it under the slates, not at the laps.

I'll say it one more time because it saves people thousands: I've seen ice dams on a 300-year-old Welsh roof and on a brand-new Chinese slate roof, and the slate was never the difference. The difference was whether the roof was built cold. That is a construction decision made before the first slate goes on.

The spring melt: the slab that comes off the roof

The fourth act of the winter is the one that surprises everyone. In the spring, a slate roof doesn't melt its snow off. It throws it off. A slate roof is smooth and cold, and the packed snow slab separates from it in one piece on a sunny afternoon — a slab of wet snow, half a metre thick in a heavy year, comes down the slope like a glacier on rails. That is where people get hurt in snow country, and it's also where a slate roof loses a tile or two, because the sliding slab catches a verge or a bottom tile and takes it with it.

So the last of the snow-country habits is the one most installers skip: snow guards. They're the little feet, or the rails, or the pipes that sit on the roof and hold the snow in place until it melts in peace — they keep a metre or two of snow on the roof instead of dropping the whole slope's worth at once. I fit them above doors, above paths, above the garden seats, and on any roof where the snow would land where people stand. On slate they fix to the deck through the slates, they're set in snow-sealed fasteners, and they live for as long as the roof does. If the same project is in a windy, cold valley, the wind-and-snow combination is the roughest winter mix there is — I'll point you to the high-wind guide for that side of it; here the snow is the story, and the story ends with a slab of ice safely held on the roof until it drips away in place.

I keep a photograph in the van of a spring roof in northern Norway where a snow slab took a lead-gutter with it — the snow was doing a job that the roof should have done: keeping itself on. Ten minutes of snow guards and the same roof would have melted in place all spring. That's a cheap lesson, and I'd rather you learn it from me than from your own spring.

The snow-country installation table: eight decisions to make before the first tile

One line of context on that last row, because I'm asked about it every November: slate can be installed in winter, but the rules are stricter — the slate needs to be dry when it goes up, the battens need to be frost-free, and the laps need to close without an ice crystal stuck between them. Wet slates + freezing night = frost damage that shows up the following spring. In the cold months I work with dry, pre-warmed slates and I stop at the first sign of ice between the laps. Better a roof half-done in December than a roof that spent its first winter with water freezing inside its own seams.

Quick answers from the snow country

• Freeze–Thaw Resistance of Roofing Slate — the test that separates a cold-climate roof from a short one.
• Roof Deck, Battens and Underlayment — the cold, ventilated assembly under the slates.
• Installing Eaves, Ridges, Hips and Valleys — where the ice finds the roof.
• How to Calculate Roofing Slate Headlap — the +25 mm snow allowance, properly calculated.
• Installing Slate in High-Wind Areas — the winter's other test, covered separately.
• Roofing Slate Test Results — the freeze–thaw and absorption figures behind the 0.2%.