The storm rule, taken seriously
Revised 10 August 2026
The rule, as it is actually told
Ask ten people and you get ten versions of the same shape. A proper åskväder — not a grey afternoon of drizzle but a storm that empties itself over the hill — and then you wait. Four days, some say. A week. Nine days if the summer has been dry. Then you go, and often enough you are right.
It is worth separating which part of that is knowledge and which part is arithmetic laid over memory. The mechanism people reach for — big rain, then fruiting — has real support in the literature. The number does not. The four-to-nine-day window appears neither in the Nordic yield-modelling work nor in the long-run fruiting surveys; as far as I can establish it has never been published or tested. Folk observation is not the same thing as wrong. But hold the distinction, because everything interesting here sits inside it.
A thunderstorm is not more rain. It is deeper rain.
Start with what the canopy does. Kofroňová and colleagues (2021), measuring throughfall in a mature Norway spruce (Picea abies) stand, found the canopy had to take on 1.5–2.4 mm of rain, depending on the year, before it saturated and began passing water through in earnest. Below that, the litter takes the next few millimetres. A soft three-millimetre afternoon is largely a canopy event: needles wet, moss darkened, most of it back in the air by evening.
Depth matters because of where the fungus actually lives. Lindahl and colleagues (2007), working down the profile of a Scots pine (Pinus sylvestris) soil in Sweden, found saprotrophic fungi largely confined to litter shed within the previous four years at the surface, while the mycorrhizal fungi — the group that contains kantarell and karljohan — dominated the older humus beneath. The mycelium that decides whether you get a flush is not in the centimetre a light rain wets.
A convective cell delivers in forty minutes what a frontal system spreads across a day. That is the honest core of the storm rule: not that the storm is magic, but that its rain arrives fast enough and deep enough to reach the layer that matters.
The lag is plausible. The day-count is not established.
The best long series we have work at coarser resolution than the rule does. Straatsma, Ayer and Egli (2001) counted fruit bodies weekly on 1500 m² at La Chanéaz in Switzerland across 21 years and more than 400 species: productivity tracked precipitation from June to October, while the timing of appearance tracked July and August temperature. Two clocks, not one. Krebs and colleagues (2008), sampling thirteen areas along 210 km of the Alaska Highway from 1993 to 2007, found crops averaging 24 kg/ha but ranging from 0 to 117 kg/ha, a coefficient of variation of 143 per cent between years — and predicted 85 per cent of that variance from June rainfall of the current year plus May rainfall of the year before. Tahvanainen and colleagues (2016), across 56 Norway spruce plots in eastern Finland, found warm pre-season conditions and a wet fruiting season promoted yields.
None of these resolves to a number of days. They work in months, because that is the grain at which they sampled. That is not a refutation — a study counting once a week structurally cannot see a six-day lag — but it is a fair statement of what has been measured. Someone who has watched one hillside for thirty years has finer resolution than any of these papers. What they lack is a record of the days they found nothing.
The lightning itself, which is the interesting part
The belief that the lightning does something is not only Swedish. Japanese growers hold that a nearby strike brings a heavy harvest, and Takaki, Takahashi and Sakamoto (2018) cite that belief directly as the motivation for their laboratory programme, noting that mushrooms growing extraordinarily around the point of a strike “have been reported by some mushroom farmers”.
The laboratory work is real and it replicates. Takaki and colleagues (2014) applied 100-nanosecond pulses at 50–130 kV to sawdust substrate and to inoculated logs of shiitake (Lentinula edodes) and two other cultivated species. Fruit-body formation rose by a factor of 1.3–2.0 in total weight; accumulated shiitake yield across four cultivation seasons went from 160 g to 320 g at 50 or 100 kV. It also has an optimum — push to 130 kV and yield falls back to 240 g. The proposed mechanism is mechanical rather than mystical: the field displaces hyphae by electrostatic force, cutting and scratching them, and that damage acts as a stress cue for fruiting.
Now the gap. Every one of those results is on a deliberately inoculated substrate, with known electrode geometry, in a species bred for cultivation. Nothing in that literature tests a wild forest, and the authors do not claim it does. The geometry is also badly against it. SMHI registers on average about 150,000 discharges to ground per year across Sweden — spread over roughly 450,000 km², that is on the order of one strike per 3 km² per year, while the rain from the same storm falls across the entire cell. Whatever a storm did to your hillside, it did with water, not with current.
Nitrogen is the other proposed mechanism, and that one can more or less be settled. Lightning fixes some 5 ± 3 Tg of nitrogen globally per year (Schumann & Huntrieser, 2007) — real, but thinly spread. More to the point, the sign looks wrong. At Gårdsjön, Brandrud (1995) added roughly 35 kg N ha⁻¹ yr⁻¹ to an oligotrophic spruce forest and mycorrhizal fruit-body production dropped well below the control plots within about two years, with Cortinarius and Russula worst affected. Added nitrogen suppresses this kind of fruiting. It does not drive it.
So: not established in the wild, and the most likely reading is that the electricity is not the mechanism. Not foolish either. A folk belief that pointed a pulsed-power laboratory at a real, replicated effect on shiitake has earned better than a shrug.
What is new is that the rule can now be checked
Until recently the weak link was memory. You remembered thunder that week; you did not know whether the cell crossed your hill or passed eight kilometres north of it. That has changed. SMHI publishes located lightning openly under CC BY 4.0, with daily strike maps from 1 May 2012 and an archive behind them. The underlying NORDLIS network, shared with Norway, Finland and Estonia, has a median location accuracy of roughly 0.5–1 km and detection efficiency above 90 per cent across most of its area (Mäkelä, Enno & Haapalainen, 2014) — best, as it happens, over central Sweden.
That is enough to turn a tradition into a hypothesis. Testing it properly needs three things folk memory does not keep: the storm’s real position and date, the days you went out and found nothing, and how long you looked. With those, four-to-nine stops being a saying and becomes a distribution with a mode you can inspect — per species and per soil. It may come out at six days on sand and eleven on clay. It may come out as the fortnight’s total rainfall, with the storm being merely the part of it loud enough to remember.
Any of those outcomes would be worth having. The rule has survived several generations on the strength of the water alone, which is a decent record for something nobody has ever measured.