The chemist Irving Langmuir had already won a Nobel Prize, but he’d never screamed in delight during an experiment before. It was November 13, 1946. He was standing in a control tower at the Schenectady, New York, airport, watching a small prop plane go buzzing overhead. Fourteen thousand feet above him, his assistant was leaning out the plane’s window, tossing pellets of dry ice into a cloud. Seconds later, the cloud “began to writhe as if in torment,” one witness recalled. Within five minutes, the cloud had disappeared, transformed into rain. Even before the plane landed, he raced off to telephone a reporter. Mankind, he shouted into the receiver, had finally learned to control the weather.

In his day job, Langmuir studied surface chemistry at General Electric Labs in upstate New York. In contrast to most corporate labs, at GE he had a free hand to research whatever he fancied, and during World War II, he began studying the buildup of ice on airplane wings. This led to a series of field studies at nearby Mount Washington, in New Hampshire. The mountain often produced mists of “supercooled” water that, despite registering far below 32 degrees Fahrenheit, refused to freeze into ice. This Schrödinger’s cat–like indeterminacy—how could water not freeze below its freezing point?—intrigued Langmuir, and he wanted to know more.

To help with the work, he engaged an assistant named Vincent Schaefer. Schaefer started his experiments by commandeering a $240 open-top GE freezer ($3,000 today). He lined it with black velvet so he could see any ice crystals that formed, then huffed into the cold air to introduce moisture, which became supercooled. Yet week after week, no matter how he varied the conditions in the freezer, the water in his breath never condensed into ice.

One sweltering July day, when the freezer was struggling to keep cold, Schaefer popped over to the lab next door and borrowed a block of dry ice (frozen CO2) to tuck into one corner. It changed everything. The instant he lowered the cube into the freezer, millions of ice crystals began twinkling in the mist. They then wafted down onto the black velvet, glittering like microscopic diamonds. Schaefer thought at first that the dry ice had induced a chemical change in the mist, but further experiments ruled that out. Rather, the temperature of the dry ice seemed to be the key. Whereas the temperature in the GE freezer bottomed out at -9 degrees Fahrenheit, the frozen CO2 was below -100 degrees Fahrenheit. When exposed to such brutal, unnatural cold, even supercooled water said uncle and formed ice.

The discovery got Langmuir thinking. Scientists at the time knew that clouds in the sky were basically loose bags of supercooled water. They also knew that most rain actually begins falling from the sky as ice crystals, which melt on their way down. Langmuir reasoned that if he peppered clouds with dry ice, perhaps he could shock the supercooled water and create rain artificially. This led to his renting a plane that November and sending Schaefer up with six pounds of dry-ice pellets, just to see what happened. Twenty minutes later Langmuir was roaring about making history.

Langmuir’s team continued to toil in the lab and he soon sketched out an idea so revolutionary that he abandoned every other project on his slate to pursue it. It promised not only to improve rainmaking but to give Langmuir the superhuman power to control hurricanes.

The idea built on James Espy’s general theory of cloud formation. Espy said that clouds form when pockets of warm, less dense air rise into the sky. At some point the water vapor in them cools and condenses into droplets of liquid water. We on the ground see these collections of droplets as clouds, and for many years meteorologists assumed that rain followed automatically, whenever these droplets got around to falling. Turns out it’s not so simple. Most droplets that form within clouds don’t automatically sprinkle down as rain. They’re too small. As early balloonists knew, air provides a buoyant upward force on anything suspended within it, water droplets included. And when droplets form at high elevations, most of them are so tiny—one ten-millionth of a gram—that gravity can’t overcome the buoyant force and drag them down. Gravity keeps losing this battle, in fact, unless the droplets grow a million times larger, to a tenth of a gram. Clearly, then, for actual rain to fall, a million teeny droplets have to glom together into a larger unit. Otherwise they just sit there.

The obvious question, of course, is what makes the tiny droplets glom together. Intuitively, you might think that droplets simply collide at random and stick to one another. This process isn’t very efficient, though, and drops that form this way rarely grow large enough to precipitate out. A better way involves “seeds,” solid surfaces for water droplets to latch onto. For various reasons, once a few droplets latch onto a seed, many more follow in quick succession. As a result, droplets can finally grow heavy enough to fall out of clouds as precipitation. If you want to transform a cloud into rain, seeds are vital.

Ice crystals make the best seeds: Specks of ice within clouds tend to vacuum up every other water droplet in the vicinity. Foreign particles like dust form great seeds as well. Even airborne bacteria can serve as seeds. But Langmuir realized that artificial chemicals might do an even better job. Specifically, he wanted a chemical whose molecular structure mimicked ice—something whose shape would fool supercooled water droplets into latching on. So he ordered another GE assistant, Bernard Vonnegut, to find such a chemical. He eventually found three candidates, including silver iodide, which forms hexagonal crystals like ice does. And tests in that workhorse GE freezer showed that it did indeed fool supercooled water, causing a veritable chain reaction of ice formation. If this worked in real clouds, it would give them a literal silver lining—and allow Langmuir to manipulate them at will.

Langmuir found his mark for this endeavor with the U.S. military, which teamed up with him for Project Cirrus in 1947. Project Cirrus had several aims, including relieving droughts. But above all its sponsors wanted to neuter hurricanes, nature’s most destructive storms. This pursuit would show Langmuir at his best and worst.

The plan to neuter hurricanes involved a series of observations and deductions, with each step built upon the last. The observations began with the structure of hurricanes, which are basically swirling windstorms centered around an eye. Per the cliché, the eye is calm, but the boundary between the eye and the rest of the storm—called the eyewall—is the most destructive part, with wind speeds often topping 150 miles per hour. That’s awfully fast, and you’d think a vortex that violent would tear itself apart. After all, anything that spins should “feel” a centrifugal force pulling it outward, the same force that makes you fly off a merry-go-round if you aren’t holding on tight. Hurricanes are no exception: This outward-tugging force threatens to wrench them apart. There’s another force to consider, however, one based on pressure differences. You see, the eyewall consists of air at high pressure, the eye of air at low pressure. And because air always tries to flow from high pressure to low pressure, this second force acts inward, countering the centrifugal force and keeping the hurricane compact.

For Project Cirrus, then, the task was simple: disrupt this balance. Langmuir’s plan was this. Pilots in sturdy airplanes would plunge into the eyewall and seed it with dry ice or silver iodide. Per the discussion above, these chemicals would act as seeds and force any supercooled water there to transform into ice crystals.

As ice forms in the eyewall, the surrounding air absorbs the heat that gets released. Absorbing heat should make this air expand according to the gas laws. Once it expands, the pressure should drop, because the air molecules are farther apart. This in turn should reduce the pressure difference between the eye and eyewall. As a result, the inward-facing force that knits the storm together should decrease. The outward centrifugal force should therefore get the upper hand and widen the eye.

Almost there. Widening the eye wouldn’t just make a hurricane go poof and disappear. Storms are too big for that. But widening the eye would decrease the wind speed. That’s because the speed of a spinning object depends on its width. That might sound a little obscure, but we’ve all seen this happen in figure skating. Whenever (insert your favorite skater here) spins with his arms tucked in (small radius), he spins fast. When he widens his arms, he slows down. Same with hurricanes: When the eye widens, they slow down. And widening the eye is critical, because the destructive force of a hurricane depends on the square of the wind speed.

To summarize this chain of deductions, Langmuir proposed that seeding hurricanes would create ice and release latent heat; this in turn would widen the eye and reduce the destructive power of the storm. The idea just needed testing—and here’s where the trouble started.

On October 13, 1947, a mild hurricane named King sliced through Miami and began drifting northeast, out into the Atlantic Ocean. Because King seemed to be dying anyway, Cirrus officials decided to seed it the next day. A B-17 puttered out to meet it and scattered 180 pounds of dry-ice pellets into the eyewall. Everyone sat back and waited for the eye to widen and for King to collapse. Instead, the storm grew stronger, fiercer. To everyone’s horror, it then pivoted—taking an impossible 135-degree turn—and began racing into Savannah, Georgia, causing $3 million in damage ($32 million today) and killing one person.

Newspapers were soon denouncing this “low Yankee trick” and calling for Langmuir’s head. Meanwhile, actual meteorologists were taking a good long look at Langmuir’s results—and what they found made them suspicious. For one thing every epic rainstorm he took credit for could plausibly be traced to another cause. Even more damning, Langmuir’s experiments lacked controls, and he seemed to select clouds that were already ripe and likely to rain anyway. When meteorologists ran their own, independent tests with proper controls and random cloud seeding, they found virtually no extra rain.

The same general criticisms applied to hurricane hunting. Hurricanes changed size and direction on their own all the time, constantly waxing and waning in strength. So it was impossible to tell whether seeding actually caused the changes Langmuir took credit for or whether it was all coincidence. In fact, Project Cirrus escaped prosecution for the Savannah debacle largely because a meteorologist with a good memory tracked down reports of a 1906 hurricane that had taken the exact same hairpin turn in the Atlantic.

Although Project Cirrus ran its course in 1952 and Langmuir passed away in 1957, he managed to stir up enough enthusiasm for weather control to extend and expand upon Project Cirrus. They called it Project Stormfury.

Stormfury commanded a multimillion-dollar budget and several planes, and to go with this kickass new name, Stormfury scientists developed new approaches for attacking clouds. Rather than fling seeds by hand, pilots now packed silver iodide into aluminum canisters and fired it out of Gatling guns. Rat-a-tat-tat-tat. And rather than rely on a single drive-by shooting, 10 planes in formation might circle a hurricane for an hour, pumping the storm full of silver.

Piggybacking on this work, the U.S. military decided to invest in a separate weather-control project—and, more ominously, to convert this research into new weapons. Each year between May and September, monsoon rains sweep through Vietnam, sometimes dumping 20 inches per month. These downpours turned most of the mud roads there into veritable slip ‘n slides—including the Ho Chi Minh Trail, a vital supply route for the Vietcong that wound through several countries. The American officers figured that if they could make the monsoons even worse, they could paralyze the enemy for several months each year. They called the scheme Project Popeye.

When presented with the plan, President Lyndon Johnson practically whooped in delight, and Popeye kicked off in 1967 with seeding runs over North Vietnam and Laos. Newly elected president Richard Nixon expanded the program in 1969, making it truly bipartisan.

Overall, the Popeye crews flew 2,602 sorties and discharged 47,409 rounds of silver iodide. Officials also congratulated themselves for conducting a more humane form of warfare, since soaking the enemy with water was surely better than dumping napalm. (Popeye’s motto was “Make mud, not war.”) What’s more, cloud seeding was covert. Unless the Vietcong took samples of rainwater and tested for silver iodide, they’d never know.

But the wall of secrecy surrounding Popeye started crumbling in 1971, when the Washington Post obtained a classified memo that mentioned it. A dogged Rhode Island senator named Claiborne Pell finally hauled several Pentagon officials into a hearing in 1974. Pell agreed that drenching people with rain was more humane than bombing them. But cloud seeding was such a crude tool, he argued, that any floods or landslides would hit civilians just as hard as soldiers. Indeed, Pell grilled the officials about a series of floods that had devastated North Vietnam in 1971. Had cloud seeding made them worse? Pentagon officials denied it, saying that they never generated enough rain to cause a flood. But that admission only brought up other, more awkward questions. How much extra rain had Project Popeye produced, then? A few inches per month, tops, officials said. In fact, they doubted that the Vietcong had even noticed their efforts, since it’s kind of hard to tell the difference between, say, 20 inches per month and 22. So why, Pell asked, did the military spend $21.6 million ($130 million today) if the program wasn’t really working? We had to try something, sir.

Politics aside, support for weather control also dwindled because it just didn’t work very well. Sure, scientists believe they can do modest things like clear fog at airports or goose certain types of clouds to squeeze out a little more rain. And scientifically, the work wasn’t a waste: Meteorologists learned an awful lot from the data they gathered. Unfortunately, much of what they learned undermined the rationale of their experiments. With hurricanes, for instance, scientists had counted on the silver iodide interacting with supercooled water to release heat. But their data revealed that hurricanes actually contain very little supercooled water. This meant that (unlike with clouds) the silver iodide had no raw material to start the chemical chain reactions. The dramatic behavior they sometimes observed after seeding hurricanes, then, was probably just a coincidence.

During his lifetime Irving Langmuir spoke with great eloquence about our duty to take control of the weather, and his charisma won thousands to his cause. But for me, the most profound truth to emerge from several decades of weather-control research came from a lowly Stormfury pilot: “It was disappointing to concede we couldn’t really do it,” he said years later. “But the storms were so big, and we were so small.”

This post is adapted from Kean’s recent book, Caesar's Last Breath: Decoding the Secrets of the Air Around Us.