The Snow That Fell from a Clear Sky

On 13 November 1946, a man flew into a cloud carrying six pounds of dry ice. The cloud was four miles long and 14,000 feet above western Massachusetts. Within minutes, white streamers fell from its underside.

The snow did not reach the ground. The air below the cloud was too dry, and the crystals evaporated after falling about 3,000 feet. It was still the first time anyone had made snow in a natural cloud on purpose. The result was less a winter wonderland than a brief, vanishing demonstration: a weather system changed by a passenger opening an aircraft window.

The man in the plane was Vincent Schaefer, a self-taught chemist working at General Electric’s Research Laboratory. He had not set out to invent a way of making snow. He had been trying to understand why water in clouds could remain liquid below the freezing point.

The cloud inside the freezer

A cloud is not necessarily made of ice. In the right conditions, its tiny water droplets can remain liquid below 0°C. They are supercooled: ready to freeze, but waiting for a suitable place to begin. In nature, a speck of dust or salt can provide that starting point. Once an ice crystal appears, water vapor can gather on it, and the crystal can grow large enough to fall.

Schaefer wanted a controlled version of this delicate state. In 1946 he used a small food-freezing cabinet at the GE laboratory in Schenectady, New York. He lined it with black velvet so a beam of light would make the particles visible, then breathed into the cold chamber to form a miniature cloud. The droplets stayed liquid. He tried introducing different substances to make ice crystals appear, with little success.

The breakthrough arrived because the laboratory was too warm. The chamber was not cold enough for the experiment, so Schaefer put in a piece of dry ice, solid carbon dioxide, to lower its temperature. At once, the cloud filled with ice crystals. The object he had added was not a magical snow-making ingredient. It was an extraordinarily cold surface and a source of intense local cooling. In his later account, Schaefer described the discovery as serendipitous, but not random: he had been looking for a way to nucleate a supercooled cloud and recognized the answer when it appeared.

The cold box made the invisible process visible. A tiny fragment of dry ice dropped through the cloud, leaving a white trail. The supercooled droplets froze along it, and the ice multiplied. Schaefer’s laboratory paper, published in Science on 15 November 1946, described the conversion of a cloud of liquid droplets into one containing ice crystals. The paper appeared two days after his flight, an unusually neat coincidence between discovery and demonstration.

Six pounds over the Berkshires

The laboratory result was persuasive, but a box was not the sky. Schaefer needed a real cloud, a plane, dry ice, and weather cold enough to cooperate. The summer and autumn had offered few suitable supercooled clouds. On 13 November, a cloud finally appeared.

Schaefer and pilot Curtis Talbot left Schenectady Airport in a small Fairchild aircraft. They found an isolated cloud over Pittsfield, Massachusetts, at roughly 14,000 feet, where the air was about -20°C. Schaefer began scattering crushed dry ice into the cloud. The dispenser jammed, so he opened the window and let the slipstream pull the remaining pellets out.

From the ground, Irving Langmuir and other GE scientists watched the cloud. The change came quickly. Within about two minutes of Schaefer’s radio signal, snow streamers appeared beneath the seeded section. From the aircraft, Schaefer could see glinting crystals and a halo around the sun, an optical sign that ice had replaced liquid droplets. He later remembered turning to Talbot and saying, “We did it.”

The next day’s GE release was careful about the result. The transformed cloud was about three miles long, and the snow fell for an estimated 3,000 feet before disappearing into dry air. The experiment had produced snow, but not a snowfall on the ground. It proved that the laboratory effect could be transferred to a natural cloud. It did not prove that a plane could summon precipitation from an otherwise empty sky.

That distinction mattered. Cloud seeding could rearrange water already present in a cloud; it could not manufacture moisture. Schaefer later recorded a more dramatic experiment in which seeding was followed by four or five inches of snow in the Hudson and Champlain valleys, but he explicitly noted that there was no proof the seeding caused that snowfall. The boundary between a changed cloud and a changed weather system was much harder to measure than the white streak in a photograph.

When weather became a material

The flight made cloud physics feel newly practical. If a supercooled cloud could be made to snow, perhaps fog could be cleared, aircraft icing could be reduced, or dry regions could receive more rain. The GE team began to imagine weather not only as something to forecast but as something that could be nudged.

The ambitions quickly grew beyond dry ice. On 14 November, the day after Schaefer’s flight, Bernard Vonnegut, another member of the GE group and the brother of the future novelist Kurt Vonnegut, demonstrated that silver iodide could create ice-forming nuclei. Silver iodide was easier to distribute from aircraft and generators than chunks of carbon dioxide. It became one of the principal materials in the new science of weather modification.

The language of control was bigger than the evidence. Schaefer and Langmuir discussed producing huge quantities of snow, clearing clouds, and even influencing storms. The 1946 GE release suggested that a single pellet might generate enough ice nuclei to produce several tons of snow, while also acknowledging that heavy precipitation required a continuing supply of moisture. The physics allowed a striking local effect. The atmosphere remained a far less obedient machine.

There is a useful humility in the failed landing at Mount Greylock. The snow was real, but it vanished before anyone on the ground could catch it. A cloud had been pushed across a threshold, not commanded to perform. The experiment made nature look programmable for a few minutes, then restored the terms: the temperature had to be right, the cloud had to exist, and the air below had to let the crystals survive.

The weather we can influence

Modern cloud seeding still begins with the same question Schaefer confronted in his cold box: is there enough water in the cloud, and are the conditions suitable for ice to grow? The answer is never contained in the seed alone. A chemical or a pellet can provide a nucleus, but it cannot supply the cloud’s moisture, its motion, or the timing of its precipitation.

That is why the most honest legacy of the flight is not weather control. It is a change in what could be tested. A cloud, once treated as a distant shape in the sky, became a laboratory with a beginning, a treatment, and an observable response. The experiment did not give us command of the weather. It gave us a way to discover how much command we did not have.

The first artificial snowstorm ended before it reached the mountain that gave it a name. Its consequence lasted longer: from then on, a cloud could be entered, altered, and asked to show its work.