The Cold Comes First

Dew does not fall, and it does not rise. It is the air's water settling on a surface that has grown colder than the air, and the cold comes from above: a clear night sky radiates like a surface far colder than anything around you. Here is a year of that, minute by minute, from two NOAA towers, and the experiments of the London physician who worked it out in 1814, re-run.

1. The sky has a temperature

Everything warm glows in the infrared, and so does the air above you: its water vapour, carbon dioxide and clouds send heat back down all night. An instrument called a pyrgeometer, looking straight up, measures that downward glow, and you can express it as the temperature a black surface would need to glow as brightly. That is the sky's radiating temperature. On a clear night it is far below the air temperature, because across a band of infrared wavelengths a clear, dry atmosphere is nearly transparent: much of the heat a surface gives off in that band passes straight out to space, and nothing equally warm sends it back.

NOAA's SURFRAD network runs stations that measure this every minute, together with a second pyrgeometer looking down at the ground. From the downward-looking one you can read the ground's own radiating temperature: the temperature of the actual grass tops, not of the air above them. The two stations here are opposites. Bondville is flat farmland in central Illinois, humid in summer. Desert Rock is in the Nevada desert, where the air is usually very dry.

Look at the clear nights first. The violet line, the sky, sits 13 to 29 °C below the orange air. On the clear, calm, humid night of 25 August at Bondville the field was already a degree below the air at civil dusk and reached the dew point 83 minutes later; from then on water could settle on it for most of the night. Now choose overcast and calm: the violet line climbs up to meet the air, because a cloud deck is itself a warm, dense radiator, and the field stays exactly as warm as the air. No cold, so no dew, however wet the air is.

Then choose the coldest grass, and no dew. It is the night in 2025 on which Bondville's field ran furthest below its air, 7.8 °C on average and 9.4 °C at the coldest minute, and there was no dew at all, because the air was so dry that its dew point was below freezing. The cold was there. The water was not.

2. Where it forms, and where it doesn't

Everyone has seen the pattern: dew thick on the lawn and the car roof, less on the car's doors, none under the tree. Nothing about the air is different in those places. What differs is how much of the cold sky each surface can see. A car roof faces all of it. A door faces half the sky and half the street. Grass under a tree sees mostly leaves, and leaves are at about the temperature of the air.

The instrument below runs a simple energy balance, on the night you chose above, driven minute by minute by what the tower actually measured: the sky's infrared, the air's temperature and humidity, the wind. Each surface is a thin sheet with no heat store (a leaf, a grass blade, a car's steel roof) that radiates, receives the sky over its share of view and the warmer surroundings over the rest, is warmed by the air it touches, and is warmed again by any water that condenses on it. Nothing in it is fitted to the data. Its fourth surface is the experiment from section 5: a handkerchief on four sticks, six inches above the grass.

Push the wind up and every surface warms toward the air, because moving air keeps delivering heat faster than the sky can take it away. Dry the air by a few degrees of dew point and the dew stops, although the cold does not change at all. Pull a low cloud across and the cold itself goes. On the measured nights the model's open surface tracks the real field closely: across every tenth night-minute of 2025 at Bondville, it runs 0.6 °C colder than the field on average, with a root-mean-square difference of 1.25 to 1.41 °C (odd and even days). That is good enough to say which surface gets wet, not to promise a gram.

3. "Much, little, or no dew": a whole year

The instruments above show single nights, chosen by rules written into the pipeline (each rule is printed under its chart). The claim that matters is about all of them. Here is every night of 2025 at both towers:

The desert's field got just as cold as the farm's, night after night, under a sky that was colder still. It almost never got wet, because the desert air's dew point sat a median 21.3 °C below its temperature (Bondville's: 3.6 °C). Cold makes dew possible. The air decides whether there is any. That is the second of three conclusions the physician William Charles Wells drew in 1814, in words the data now sits under almost exactly:

“the same degree of cold, in the precipitating body, may be attended with much, with little, or with no dew, according to the existing state of the air, in regard to moisture”W. C. Wells, An Essay on Dew (1814), p. 59

Three of the five nights of 2025 on which Bondville's field ran furthest below its air had no possible deposition at all; the coldest of all had a dew point 18.3 °C below the air. Wells again, on his own coldest nights: “very little dew appeared upon the two nights of the greatest cold I have ever observed on the surface of the earth, relatively to the temperature of the air, both of which nights having occurred after a long tract of dry weather” (p. 60).

4. Clear and calm

Wells also wrote down which weather gives dew, and which gives none: “I have frequently found a small quantity of dew on grass, both on windy nights, if the sky was clear, or nearly so, and on cloudy nights, if there was no wind; but none has ever been seen by me on nights, that were both cloudy and windy” (p. 5). Split the year into thirds by how clear the sky was and by how hard the wind blew, and count:

The cold follows Wells to the letter: at Bondville the field ran 4.6 °C below the air on the clearest, calmest third of nights and 0.6 °C on the cloudiest, windiest third, and 29 of 39 clear, calm nights reached an hour of possible deposition. The count in the cloudy corners needs care, and the sensor-doubt box shows why. On those nights the air was already near saturation, so a field a fraction of a degree below the air crosses the dew point, and those tiny margins are within what a humidity sensor near 100% can honestly resolve. Assume the field half a degree warmer and the cloudy-calm cell falls from 8 nights to 2, and the cloudy-windy one from 17 to 8. Some of those remaining nights may be dew; some may be rain or drizzle on a wet field, which this instrument cannot tell apart. What the data can say firmly is Wells's cause: on cloudy, windy nights the cold that makes dew is almost absent.

5. Dew makes heat, which is why there is never much

Before Wells, the best observers had noticed that dewy grass is cold and concluded that the dew made it cold. Wells turned it round. He had long held the same view, he says, “but I soon saw reason, after my regular course of experiments was begun, to doubt its truth, as I found that bodies would sometimes become colder than the air, without being dewed” (p. 54), and he concluded “that dew is the production of a preceding cold in the substances, upon which it appears” (p. 54). His third conclusion follows:

“The formation of dew, indeed, not only does not produce cold, but, like every other precipitation of water from the atmosphere, produces heat.”Wells (1814), pp. 59 to 60

The order of events is in the tower data too. On 186 nights of 2025 at Bondville the field both went a degree below the air and reached the point where water could settle. On 143 of them the cold came first, and on the median such night the field was already a degree below the air at civil dusk while water could not settle until 3 hours 46 minutes later. On 14 nights the order was reversed; on those the air was close to saturation from the start.

That heat puts a hard ceiling on dew. Every gram that condenses releases about 2.45 kJ, and a surface already colder than the air and the soil beneath it can only get rid of that heat one way: by radiating it to the sky. So a night's dew can never exceed the night's net infrared loss divided by the latent heat. At Bondville that ceiling, computed from the tower's two pyrgeometers, was a median of 0.64 mm on the nights with an hour or more of possible deposition, and never above 1.6 mm on any night of 2025, even the longest winter night. A millimetre of water is a litre on each square metre. Dew is a thin thing, and it is thin for a thermodynamic reason. The bars in section 2 show the ceiling for the open surface against what the model gathers: the model stays well under it, because the air and the ground keep supplying some of the heat the sky is taking.

Wells saw the brake working. On the dewiest night he ever watched, the grass began 7½° below the air, “But after the dew had become very abundant, the difference of those temperatures never exceeded 4° and was frequently only 3°” (p. 60; Wells's degrees are Fahrenheit).

6. The handkerchief, the pasteboard roof, and the cloud

William Charles Wells (1757 to 1817) was a London physician, born in Charleston, South Carolina, who laid small thermometers on a grass plot at night and weighed tufts of wool left out on it. He compared grass with the air four feet up, and on calm clear nights “very frequently found the grass, at the ordinary place of my observations, 7, 8, or 9 degrees colder than the air. Several times it was 10° and 11° colder than the air, and once 12°” (p. 28). In Celsius that is 3.9 to 6.7 °C. Bondville's field ran 4.6 °C below its (higher, 10 m) air on the average clear, calm night of 2025.

He sheltered wool from the sky. Under a sheet of pasteboard bent into a little roof, ten grains of wool gained “only 2 grains, but that, which had been exposed to the sky, 16 grains” (p. 15). Then the experiment the model above re-runs, on p. 121: four sticks rising six inches above a grass plot, forming a square with sides two feet long, and over them “a fine cambrick handkerchief, rendered by long wear still thinner than it had been originally, and having here and there a slight rent.” Air could pass freely; only the view of the sky was taken away. The grass beneath was always warmer than the grass outside, and on p. 122 he gives numbers: on one night “when fully exposed grass was 11° colder than the air, the latter was 3° warmer than the sheltered grass”.

Wells measured; the Bondville rows are the model of section 2, run through each measured night, with the grass under the cloth seeing it over 0.83 of its sky. They are not the same nights, the same place or the same air height (his air was 4 feet up, the tower's 10 metres), so the comparison is of size and direction, not a test. The model leaves out a drawback Wells himself named: cold air draining in from the exposed grass.

And he watched clouds do what the instrument's "Wells's cloud" night shows. On p. 32:

“During a fourth night, the temperature of the grass at half past 9 o'clock was 32°. In 20 minutes afterwards, it was found to be 39°, the sky having in the mean time become cloudy. At the end of 20 minutes more, the sky being clear, the temperature of the grass was again 32°.”Wells (1814), p. 32

Seven Fahrenheit degrees in twenty minutes is 3.9 °C. In the whole of 2025 at Bondville, the largest 20-minute warming of the field under an arriving cloud, with the air steady to within 1.5 °C, was 3.15 °C, on a January evening; the largest with the air above 5 °C was 2.56 °C, at 21:04 on 24 August, which is the night the picker calls "Wells's cloud". A field of grass seen by a pyrgeometer from above is an average over many square metres; Wells's thermometer lay on one tuft. The two are the same event at different grain.

A last and stranger thing about this essay. In 1818, the year after Wells died, it was reissued in one volume with a paper he had read to the Royal Society in 1813, about a white woman part of whose skin was dark, and Charles Darwin, in the historical sketch added to later editions of On the Origin of Species, credited that paper with the first statement of natural selection: “In this paper he distinctly recognises the principle of natural selection, and this is the first recognition which has been indicated; but he applies it only to the races of man, and to certain characters alone.” The book that settled where dew comes from also carried, bound in with it, an idea its author did not follow.