They Do Modify The Weather. Nobody Can Prove By How Much.

GRAND JUNCTION, Colo. — The generator is a propane burner, a tank of solution, and a mast, bolted to a concrete pad on a ridge at 2,900 metres. It costs about the price of a pickup truck. There are eleven of them in this district.

It works, in the sense that the chemistry works. Silver iodide has a crystal lattice close enough to ice that supercooled droplets will freeze onto it at temperatures where they would otherwise stay liquid. You can watch it happen in a cloud chamber. Nobody disputes this and nobody has for seventy-five years.

What nobody can tell you is how much rain came out of it.

The problem is not physics

Professor Idrissa Bergqvist-Achebe spent twenty minutes making sure our correspondent understood that the argument is not about whether seeding does anything.

It is about subtraction.

To know what seeding added, you need to know what would have fallen without it. That number does not exist. It is not measured, it is modelled — and the model is built from other storms, in other years, over other ground.

"You are subtracting an estimate from a measurement and reporting the remainder to two significant figures," he said. "Every honest paper in this field is a paper about that subtraction."

What the numbers actually look like

The famous Israeli experiments give about 13 per cent. Thailand's warm-rain work is significant with an uncertain magnitude. Wyoming's winter orographic trial lands somewhere between 5 and 15.

The US Government Accountability Office's review puts the range across programmes at 0 to 20 per cent.

Zero is in the range. That is not a scandal — it is what a wide confidence interval looks like when it is reported honestly, and this desk would rather have the interval than the headline.

The question we went to ask

Marguerite Ostrowski-Fane runs this district's programme. Our correspondent asked her the only question that matters: how do you know?

Her full answer is printed at the foot of this story and it is the reason the piece exists. She does not know, in the sense the question means. She has physics that is not in doubt, radar showing precipitation in the seeded plume and not beside it, and a positive statistical signal roughly the size the physics predicts.

What she does not have, and cannot have, is the same winter twice.

She then gave this desk a diagnostic we intend to use: the honest operators tell you that; the dishonest ones give you a single number with no interval on it, "and you can identify them from the doorway."

The one nobody has answered

Two of our sources independently named the same open question and neither would guess at an answer.

If you make it rain here, does it rain less somewhere downwind?

It is physically plausible. It has been investigated repeatedly since the 1950s. It has never been demonstrated, and it has never been ruled out, and the reason is the same subtraction problem one county over.

What this desk thinks

This newspaper's weather editor has published his own error rate every Friday for twenty years and has a settled view about numbers without intervals on them.

Weather modification is real, permitted, published, budgeted in public documents, and about as secret as a snowplough. Its effect is somewhere between nothing and a fifth, and the people doing it will tell you that if you drive up the ridge and ask.

The interesting story here was never whether they are doing it. It is that they are doing it, at scale, with public money, and cannot prove it works — and have said so, in print, for decades, to an audience that was not listening because it was busy with something else.