Weather lore put to the test: what holds up and what is coincidence

There is this moment in early May when half the allotment association stands in front of the tomato plants and nobody dares. The nights are mild, the plants are outgrowing their pots, and then somebody says over the fence: “You are not putting anything out here before Cold Sophie.” Everyone nods. Nobody asks why.

Right now, late July · CW 31, the dog days are running. Another one of those phrases everyone knows and hardly anyone questions. Yet these rules can be checked against the numbers. Germany's national weather service and climate research have long measurement series, and they say fairly clearly which sayings rest on physics and which merely sound good.

One at a time: which rules survive scrutiny, which fail and what the warming climate is currently doing to them.

Why some rules work and others do not

Country weather lore is statistics without a spreadsheet. Somebody farmed the same land for forty years, noted what kept coming back and packed it into a rhyme so the grandchildren would remember it. That is not a bad method. It just has two weaknesses built in.

The first: you remember the hits. The year it really did freeze on Pancras gets retold for fifteen years. The twelve years when nothing happened get told by nobody. Across generations, a coincidence hardens into a certainty.

The second is more fundamental. A weather rule can only reach as far as the atmosphere itself reaches. The theoretical forecast horizon for actual weather sits at roughly two weeks. Beyond that, every scrap of information dissolves into chaos. A rule that infers February from 11 November is therefore not imprecise, it is impossible in principle.

What can work, on the other hand, are singularities. That is what meteorology calls weather patterns that show up at certain times of year more often than chance allows, because the circulation over central Europe shifts with the seasons. A singularity is not a forecast for one particular day. It says: within this window, experience shows that something specific tends to happen, and there is a physical reason for it. That dividing line is exactly where the old rules fall apart.

The Ice Saints: right phenomenon, wrong date

Five names, five days. Mamertus on the 11th, Pancras on the 12th, Servatius on the 13th, Boniface on the 14th and Sophia on the 15th of May, known in German folklore as Cold Sophie. In 2026 the window fell in CW 20. Northern Germany traditionally counts from Mamertus, while southern Germany, German-speaking Switzerland and Austria count from Pancras through to Sophia. That is no accident: the cold air arrives from the north and takes about a day to cross the Alps.

There is nothing esoteric about the physics. In May the land heats up quickly, while the North Sea and the North Atlantic lag weeks behind and stay noticeably cooler. If the flow tips round to the north, polar air pours over warm land. Behind the cold front the sky clears, the wind dies away, and on a clear night the ground radiates its heat away unhindered. The forecast says plus two degrees, and there is frost on the leaves by morning anyway.

But the rule dates from before the Gregorian calendar reform. In 1582 ten days were skipped so the calendar would line up with the sun again. Weather follows the sun, not the date in the saints' calendar. Take those ten days back out and the window originally observed would land somewhere between 21 and 25 May today. That still does not prove a cluster actually sits there: so far this is calendar arithmetic, not measurement. German series show nothing clear-cut at the shifted date. Fifty-year measurement series from Austria, on the other hand, do find a recognisable temperature drop there, sharp enough to reach frost at higher elevations.

And the hit rate at the traditional date? It is considerably worse than the reputation of the Ice Saints suggests. Long measurement series show no reliable cluster of frost nights around 11 to 15 May. Take hundred-year series for Germany and define the window generously from 8 to 18 May, and the Ice Saints failed to materialise by today's calendar in roughly 61 per cent of years. Regionally the picture pulls far apart: in the Alpine foreland, in valleys of the low mountain ranges, on the Swiss northern slope of the Alps and in frost hollows, May frost remains a genuine risk, while in the north German lowlands and the mild Valais it stays away entirely in many years. There simply is no single rule that covers the whole German-speaking region.

For next spring, when the question lands on the table again, replace the saints' calendar with five steps.

1. Look up the last frost date for your own location

   Not nationally, but for your nearest station. Between the Upper Rhine valley and the Bavarian Forest there are a good four weeks of difference in the mean last frost. That single number replaces the entire saints' calendar.

2. Read the forecast, not the date

   The large-scale switch to a northerly flow usually shows up five to seven days ahead. Whether it actually turns into a frost night is only settled two or three days out, because radiation frost depends on cloud and wind during one single night. Use the medium-range view as a warning signal and the short range as your decision. If nothing shows up, you can go out on 8 May, but the fleece stays by the bed until the end of the month.

3. Harden off young plants first

   Two weeks outside during the day, back in at night. Cold damage in tomatoes starts below roughly 10 degrees and becomes visible under five. Hardening off improves cold tolerance considerably, but not frost tolerance: at zero degrees the hardened plant dies too. How to do it is in the piece on hardening off young plants.

4. Know the cold air pockets in your own garden

   Cold air is heavy and flows downhill. The lowest corner behind the hedge freezes regularly, while it stays two degrees warmer against the south wall of the house. Put your sensitive crops in the warm spot.

5. Write the date down

   Date of the last frost, the plants affected, what the damage looked like. After three years you will have a firmer number for your garden than any inherited rule.

Sheep's cold: the rule with the clearest cause

Early to mid June, classically around 11 June, the temperature drops away. In 2026 that fell in CW 24. Five to ten degrees below whatever came before, plus wind, rain and the occasional dusting of snow on the higher ground of the low mountain ranges. That is sheep's cold. It is usually placed in the window from 4 to 20 June · CW 23 to 25, with the peak frequency around 11 June.

The reason is the same as in May, only sorted the other way round. By June the European landmass has warmed up substantially while the North Atlantic lags weeks behind. That temperature contrast shifts the frontal zone and favours a north-westerly flow. Cool, damp maritime air runs over heated land, and for a few days early summer turns back into April.

As singularities go, sheep's cold counts among the best documented in central Europe. Germany's weather service puts the probability of below-average temperature at around 80 per cent for the narrow 10 to 12 June window, and above-average rainfall at roughly 55 per cent. What is shifting, though, is how often it happens at all: averaged over the last hundred years a cold setback occurred in this window in about 61 per cent of years, between 1921 and 1990 in roughly 73 per cent, and in recent decades in only about a third. The cause persists, the event is getting rarer. The name comes from sheep shearing, which traditionally fell in May and early June. Freshly shorn animals stood out in the pasture with no protection when a setback like this arrived.

For the garden, sheep's cold matters more than the Ice Saints even though no frost is involved. Warm-season crops need soil warmth, not merely the absence of frost. Below about 10 to 12 degrees of soil temperature, cucumbers, squash, basil and peppers largely stop growing. They are not dying, they are waiting. Air temperature is a different matter: if it drops below roughly 8 degrees at night, basil and cucumber in particular take real cold injury, glassy patches and blackened leaf tissue, and that does not grow out again. So during sheep's cold, sensitive crops want fleece over them, frost or no frost. Feeding them at this stage, or watering with ice-cold tap water, only makes things worse.

The Seven Sleepers: not a day, but a switch being thrown

“As the weather behaves on Seven Sleepers’ Day, so it stays for seven weeks.” The liturgical day is 27 June · CW 26. The calendar reform struck here too, and this time we will carry the correction through properly: the date actually meant falls on or shortly before 7 July · CW 28. The name, incidentally, has nothing to do with the dormouse that shares it in German, but with the legend of the Seven Sleepers of Ephesus.

This is the most interesting rule of the lot, because it works better than it sounds. Around the turn of the month, the jet stream over the North Atlantic settles into a particular position for several weeks. If it runs far to the north, the Azores High sits over central Europe and high summer turns dry and warm. If it lies further south, one low after another comes through off the North Sea and July drowns. That switch genuinely does tend to hold for four to seven weeks.

The statistical cluster the rule rests on was demonstrated by the meteorologist Franz Baur for the window from 5 to 10 July, which is the first week of July and not 27 June. That is what the numbers refer to as well: 60 to 70 per cent is quoted for southern Germany, and up to 73 per cent for the 1991 to 2020 period. For a statement covering seven weeks, that is remarkably high. Towards the north and the coast the figure drops off sharply, because maritime influence stirs the pattern up more often there. So if you want to test the rule yourself, do not look at a single day, look at the average across that first July week.

The failures: dog days, Martinmas and a white Christmas

The dog days run from 23 July to 23 August · CW 30 to 34. The name comes from Sirius, the dog star in the constellation Canis Major, whose reappearance in the early morning sky fell in this period in antiquity. The precession of the earth's axis has since shifted that by weeks, so the calendar name is already outdated in its own derivation.

As a description the thing still half works: the window overlaps substantially with the warmest stretch of the year, but sits a little too late. The climatological maximum in central Europe falls roughly between 20 July and 5 August, because land and water reach their peak warmth a few weeks after the sun does. By the second half of August the averages are already coming back down. That, though, is not a forecast, it is a truism. The actual weather saying, the one that wants to predict autumn or winter from the dog days, survives no analysis at all.

Martinmas on 11 November · CW 46 is the classic case of the impossible rule. “If Martin has a white beard, the winter will be long and hard.” A single November day contains no information about January and February, and analyses of long series find no hit rate above chance for this rule. The same goes for “Christmas in clover, Easter in snow”.

With the white Christmas there is the added problem that the expectation itself is skewed. In the lowlands, a closed snow cover on 24 December · CW 52 is the exception rather than the rule. Only in the higher parts of the low mountain ranges and in the Alps does the ratio tip the other way. What many people remember as “we used to get it every year” is in good part the same memory bias as with the Ice Saints.

What climate change is doing to the rules

This is where it gets uncomfortable, because the rules are not changing at the same rate. Some are losing their foundation, others are unexpectedly gaining weight.

The Ice Saints lose twice over: once the date, once the phenomenon. The date was already wrong before the warming started. And now spring in central Europe is turning noticeably warmer on top of that, the mean last frost date is moving forward, and in the lowlands a May frost night increasingly fails to appear at all. It would still be wrong to read that as an all-clear. The start of the growing season has moved forward even faster than the last frost. Fruit blossom today often opens two to three weeks earlier than it did a few decades ago. So fewer frosts are meeting more vulnerable tissue, and the severe blossom frost damage of recent years in apricot, peach, sweet cherry and apple is exactly this pattern. For early-flowering trees, frost protection for early-flowering fruit trees pays off more today than it used to, not less.

The Seven Sleepers rule is likely to hold up best, because it rests on the position of the jet stream rather than on absolute temperatures. As long as the circulation makes its seasonal shift around the turn of the month, the switch stays in place. Whether it delivers a warm summer or a wet one remains open. All that moves upward is the temperature level.

With sheep's cold the picture is more ambivalent, and this is the most honest passage in the article. The driver remains, because oceans warm more slowly than continents. But the setback itself is measurably getting rarer: from roughly three quarters of years in the 1921 to 1990 period down to about a third in recent decades. So the rule with the best physics behind it is precisely the one currently losing hit rate. Anyone who has pencilled it in as a fixed date for the fleece will spend more years than before waiting for nothing.

What follows for the garden is less a new rule than a different order of worries. Frost in May stays a local matter for frost-prone sites and early-flowering woody plants. Summer drought becomes everybody's problem. How to set your garden up for that is covered in more depth in the piece on the climate-resilient garden, and the watering practice in the piece on water-wise watering in summer.

Write your own weather rule

The best thing about the old rules is not their content, it is their method. Somebody watched the same place for decades and picked out the patterns. That is entirely possible again today, and with far better tools than a rhyme.

Note down five dates every year: the last frost, the day the sweet cherry comes into flower, the date the tomatoes went out, the first ripe fruit and the first frost in autumn. That is two minutes of work per entry. After three years you will see a tendency, after five you will have a rule worth more for your garden than the entire saints' calendar. Whether your tomato ‘Berner Rose’ can go out on 10 or 25 May where you live is something no weather saying on earth knows. Your own records will know it after five seasons, and after three they already show you the direction.

That is exactly what the garden journal in Gartenkern is for. One entry per event, with a date and a photo, and the previous years stay visible alongside. At some point you will be standing in front of your young plants in May, looking at your own history instead of the calendar, and you will simply know.

Common questions about weather lore

When are the Ice Saints?

In the saints' calendar, always 11 to 15 May: Mamertus, Pancras, Servatius, Boniface and Sophia, known as Cold Sophie. In 2026 that fell in CW 20. Northern Germany counts from the 11th, southern Germany, German-speaking Switzerland and Austria from 12 May, because the cold air comes from the north. But since the rule predates the 1582 calendar reform, the window originally observed would fall between 21 and 25 May by today's calendar. Whether a real cluster sits there is an open question for Germany, though Austrian measurement series do find a recognisable temperature drop at the shifted date.

Can I safely plant tomatoes out after the Ice Saints?

Not automatically. 16 May is no guarantee, especially not in frost-prone sites, in valleys of the low mountain ranges and in the Alpine foreland, where ground frost is still possible in late May. Conversely, in mild wine-growing areas you can often plant out in early May. More reliable than the date is the mean last frost date for your location, combined with a look at the forecast: the switch to a northerly flow shows up five to seven days ahead, while whether frost actually arrives is only settled two or three days out. And keep fleece ready either way.

What exactly is sheep's cold?

A cold air outbreak in the window from 4 to 20 June, peaking around 11 June, with a temperature drop of five to ten degrees plus wind and rain. The cause is the contrast between an already warmed landmass and a still cool North Atlantic, which favours a north-westerly flow. Frost is usually not part of it, but for warm-season crops such as cucumbers, squash and basil the cold is a problem regardless: below roughly 10 to 12 degrees of soil temperature they stop growing, and below about 8 degrees of air temperature basil and cucumber take lasting cold injury. Over the last hundred years it occurred in around 61 per cent of years, in recent decades in only about a third.

Does the Seven Sleepers rule actually work?

Partly, and better than its reputation. It does not work as a statement about 27 June alone, but about the weather in the first week of July. The reason is the calendar reform: the date originally meant now falls on or shortly before 7 July, and the statistical cluster has been demonstrated for the window from 5 to 10 July. During that period the jet stream settles into position for several weeks and thereby sets the character of high summer. In southern Germany the hit rate runs at 60 to 70 per cent, with up to 73 per cent quoted for 1991 to 2020. Towards northern Germany and the coast it drops off sharply.

Why are the dog days called dog days?

After the star Sirius, the dog star in the constellation Canis Major. In antiquity its reappearance in the early morning eastern sky fell in this period. The precession of the earth's axis has since shifted that by weeks. The stretch from 23 July to 23 August does overlap with the warmest time of year, but sits a little too late: the climatological maximum in central Europe falls roughly between 20 July and 5 August. That is down to the thermal inertia of land and water after the sun's highest point, not to a star.

Has climate change made weather lore worthless?

Not across the board, but very unevenly. Rules built on a fixed calendar date lose further value, because spring and the start of the growing season are shifting. Rules built on a physical cause hold up better. The Seven Sleepers rule is one of them, because the position of the jet stream is largely independent of absolute temperatures. With sheep's cold the driver does persist, but the cold setback occurs considerably less often than it did a few decades ago. In practice: late frost stays a local matter, summer drought becomes everybody's problem.