We Tested The Heater Against Forty Years Of Storms. It Was Switched Off.
Three hypotheses, pre-registered in March before we asked for anything: severe weather, earthquakes within 900 kilometres, reader sky-sound reports. The facility's transmission log for 1981 to 2025 arrived in June. All three tests return nothing, and the log says why.
A standing series in which this newspaper investigates a widely held claim and reports that it could not substantiate it.
VASILSURSK, Russia — The transmitter at Vasilsursk ran for three hours and six minutes in 2024. In 2023 it did not run at all.
We did not know that when we started. We had written the tests first, which is the only part of this story we are pleased with.
What we said we would look for
On 11 March this newspaper published an analysis plan and then went out and asked for the data, in that order.
Three hypotheses. Monthly transmitting hours against severe-weather days in the surrounding federal district, at lags of nought to thirty days. Against earthquakes of magnitude 4.0 or greater within 900 kilometres, at lags of nought to fourteen. Against reader-submitted sky-sound reports — 1,140 of them, from our own inbox, compiled and frozen before we approached anybody — at lags of nought to seven.
A correlation would count if it cleared p below 0.01 after correction across the three families, and then survived a dose-response check: the busiest third of months set against the quietest third. One run. No second look with a different window.
That is the same form as our own protocol on masts, PX-2161, and we use it for the same reason. It stops us choosing the window after we have seen the answer.
What came back
Nothing, three times.
Severe weather: the largest correlation anywhere in the lag range is 0.041, at nine days, p = 0.44. Earthquakes: −0.019, and the sign flips twice inside the window. Sky sounds: 0.077, p = 0.19 — and that one dissolves entirely if you remove the three months of 2011 in which we asked readers on our own front page to send us sky sounds.
The dose-response check could not be run at all. There is no busiest third.
Dr. Solveig Trapnell-Amadi was given four unlabelled columns and the plan, and asked us twice in writing what column B was. We did not tell her until she had delivered.
The log is the story
Five hundred and forty monthly rows, 1981 to 2025. Four hundred and twenty-one of them are zero.
The array went into service in 1981: 144 crossed dipoles on a square 300 metres a side, 190 megawatts of effective radiated power on the commissioning plate. In 1983 it transmitted for 612 hours across fifty-one campaigns, and the register names the operator, by hand, for every one.
In 2009 the price it paid for power changed. The connection stayed. The institute has not been able to afford anything like historic hours since, and no year since has reached forty. Twelve and a half hours in 2016. Nine in 2020. None in 2023.
The three hours and six minutes in 2024 were run in three blocks on the afternoon of 14 May, for one diploma student measuring artificial airglow. That was the year.
Tuesdays
Three people are on the establishment.
One of them is Valentin Kuzey-Radko, who was the transmitter engineer here from 1984 until he retired in 2011, and who comes in on Tuesdays and runs the demineralised cooling circuit for forty minutes. A static loop with oxygen dissolved in it corrodes the copper. Forty minutes of flow a week costs nothing and saves a heat exchanger the institute could not replace.
He is not a volunteer. He is on a service contract worth 4,100 roubles a month, and it is paid late.
What this desk takes from it
We set out to test whether a famous machine bends the weather, and the answer is that we cannot say, because for most of the period in which our readers believe it was bending the weather it was cold, dark and drawing no current.
That is a weaker result than a null. A null is a measurement. This is a corridor with the lights off.
What the log does establish is duller and more specific than anything we pre-registered. The weather weapon has a maintenance backlog, a five-year billing dispute, a retired man on a late-paid contract keeping one water circuit turning over, and a director who has applied eleven times for two hundred hours a year and been refused eleven times.
He would very much like somebody to ask him about that instead. His answer, printed below at his insistence and entire, is the best thing on this page.
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
We wrote the analysis plan before we approached anybody, and published it on 11 March — the three hypotheses, the 900-kilometre radius, the lag windows, the significance threshold, the dose-response check and the stopping rule. The log arrived in June through an academic exchange. The correlations were then run by a statistician who was handed four unlabelled columns and never told which was which.
Who we spoke to
- Sura ionospheric heating facility — monthly transmission log, 1981–2025, Transcribed from the facility's bound operating registers. Obtained through an academic exchange: a Norwegian institute requested it for a joint proposal and passed it to us with the facility's written permission June 2026 540 monthly rows. 421 of them are zero.
- Dr. Solveig Trapnell-Amadi, Statistician; ran all three correlations blind. Given four unlabelled columns and the published analysis plan, and not told what any column was until she had delivered July 2026 Asked twice, in writing, what column B was. We did not answer either time.
- Dr. Ilya Stelmakh-Vorobey, Director of the facility since 2007. Interviewed by video call twice, ninety minutes in total, through an interpreter this newspaper paid; sent the full draft with no deadline July and August 2026 Corrects two sentences of ours. We changed one and printed his account of both. His reply runs entire.
- Valentin Kuzey-Radko, Transmitter engineer, 1984–2011; now on a service contract. Interviewed by telephone for forty minutes, through the same interpreter, at the director's suggestion and with his own consent August 2026 Described the Tuesday routine minute by minute, including the parts of it he thinks are pointless.
- 1,140 reader reports of unexplained sky sounds, 2011–2025, This newspaper's own submissions log, dated and located where possible. Compiled and frozen in March, before we approached the institute; 214 further reports discarded in advance for having no usable date Compiled March 2026 The compilation and the discard rule were published with the analysis plan.
Documents
- PX-2161 — Cumulative radiofrequency exposure study — signed protocol, published before recruitment accepted
What we could not confirm
- Whether the log is complete. It is the facility's own record, transcribed by its own staff from bound registers we have not handled. A campaign run and not entered would not appear anywhere in our data, and the only people who would know are the people who would not have entered it.
- What our test could actually have detected. Across 540 monthly rows the analysis had roughly 80 per cent power to find a correlation of about 0.19 or larger. Anything smaller sits inside our noise, and we are not entitled to report that it is absent — only that we could not see it.
- The 2009 repricing. The institute describes a tariff change on a regional allocation. We put the question to the grid company twice and received no answer either time. What follows is the director's account of why his electricity became expensive, not an audited account of it.
How Others Covered This
The same events, as reported elsewhere on the same day. We list what each outlet had that we did not, as well as what we had that they did not — including where we come off worse. Why we print this.
-
The Meridian TelegraphINSIDE RUSSIA'S WEATHER GUN: THE ARRAY THAT CAN BOIL THE SKY
Toured the site with a photographer and reported the 1981 commissioning plate as a description of current capability.
Had that we did not
Photographs of the antenna field at dusk that are better than anything this newspaper has.
Left out
The transmission log, which the institute releases to anybody who asks in writing, and which shows the array ran for three hours last year.
-
The Continental WireRussian Institute Denies Weather-Modification Claims
Put the allegation to the institute, printed the denial accurately, and balanced it with a Western academic calling the claim implausible.
Had that we did not
The denial in full, and the institute's own description of what the facility is for, which no other outlet carried.
Left out
That a denial is not a measurement, and that the measurement was sitting in a bound register, available for the asking, for seventeen years.
-
They Buriedthis newspaperWe Tested The Heater Against Forty Years Of Storms. It Was Switched Off.
Published the analysis plan first, then asked for the log, then had the correlations run by somebody who did not know what she was correlating.
Had that we did not
The pre-registration, the blind analyst's output whether or not it helped us, and the log entire.
Left out
We spent four months and a five-figure sum testing whether a machine moves the weather without first establishing whether the machine was on. Its director had to tell us. — V. Ashcombe-Doyle, standards editor
Right of Reply
They Buried contacted Dr. Ilya Stelmakh-Vorobey, director of the facility on 3 July 2026 by video call, again on 6 August with the full draft, the analysis plan and the analyst's output attached, and no deadline set. Replied in writing on 15 August 2026, in Russian, asking that it run entire. Translated by the interpreter and printed unedited.
Your three tests are correct, and they are the wrong three tests, and I would like to explain why I answered them anyway.
You asked whether we can move the weather. We cannot, and no instrument of this kind ever could, and the arithmetic has been in the open literature since the 1970s: what a heater of this size puts into the upper atmosphere in an hour is of the order of what a modest thunderstorm dissipates in a second. I did not need your statistician. Neither, if I am blunt, did you.
I gave you the register anyway, in full, including the years that shame me, because I have learned that the alternative to being reported accurately is not being left in peace. It is being reported inaccurately. Three of your colleagues from other papers have come here since 2011. Two of them photographed the antenna field, printed the 190-megawatt figure from the 1981 commissioning plate as though it described something we do on a Tuesday, and did not ask to see one page of the register. One of them stood twenty metres from Valentin Petrovich, who was at that moment carrying a bucket, and did not ask him what the bucket was for.
Now permit me two corrections and one complaint.
The first correction is small and you should make it. We do not have an unpaid electricity account. We have a reconciliation dispute, five years old, with the regional grid company, over reactive-power charging on a load that is by its nature extremely bad-tempered. Until it is settled we are invoiced at a rate I believe to be wrong, and we pay it. That is a different thing. If you print 'unpaid' you will have told your readers we are a delinquent, when what we are is a small institution losing an argument to a large one.
The second correction is not small. You have written that Valentin Petrovich comes in on Tuesdays 'so the valves do not seize'. There are no valves. The circuit is demineralised water, and the reason he runs it is corrosion: a static loop with dissolved oxygen in it eats copper, and forty minutes of flow a week costs nothing and preserves a heat exchanger we could not replace. He is also not a sentimental old man, which is the picture your question implied and which he would find funny and then resent. He is under contract. We pay him 4,100 roubles a month and we pay it late, and he comes anyway, and those are two separate facts, and only one of them is charming.
Now the complaint.
You have spent, by your own disclosure, four months and a five-figure sum establishing that we are not doing something. I do not object to that; it is your money and your method is honest, which is more than I usually get. I object to the shape of the finished thing, which is a story about an absence. Absence is what we have had for seventeen years. What we have never once had is a journalist asking the other question, which is what a working ionospheric heater is for.
It is for this. You cannot put a probe into the ionosphere and leave it there. You can fly through it, which gives you a line, or bounce radio off it, which gives you an average. A heater lets you put a known amount of energy into a known volume at a known moment and then watch what the plasma does — artificial turbulence, field-aligned irregularities, generation in the very low frequency band that no antenna on the ground can produce. None of that is exotic. It is the only laboratory of its kind, because the object under study is 250 kilometres above our heads and cannot be brought down to us.
Two hundred hours a year would carry four to six doctoral students through this facility. Two hundred hours is roughly the annual electricity of a provincial swimming pool. I have applied for it eleven times, to four different bodies, and I have the eleven refusals and I will send them to you if you want them, which I suspect you do not, because refusals are not a story.
I am fifty-eight. I have run this place for nineteen years and I expect to be the man who closes it. When it closes there will be no such instrument in this country, and the two people left who know how to bring one up from cold will be Valentin Petrovich, who is seventy-one, and me.
So: you have written a careful piece proving that we do not command the storms. Thank you. I would have preferred a careless piece asking what became of the science. But I understand that is not the newspaper you are, and on the whole I would rather be handled by a paper that publishes its analysis plan in advance than by one that publishes my antenna field at sunset.
Print all of this or none of it.
Published unedited under our right-of-reply guarantee.
How was this story?
We publish the result, whatever it is. Reader verdicts appear on the front page and in our newsroom metrics.
Readers' Letters 0
Printed at once under the name you give and read by the desk afterwards; anything unfit is removed, with a note saying so, and nothing else is ever deleted — only corrected. Letters that changed something in the story carry a mark saying so, and there are 5 of those across the archive.