1941: freezing water
Siple and Passel timed water freezing in small plastic cylinders during the 1939 to 1941 US Antarctic expedition. Published in 1945, their index described the cylinder's heat loss per unit area.
groundtruth / an equivalent, not a thermometer
A Rate Dressed as a Temperature
“Feels like −19” sounds like a second, truer temperature. It is not. Wind chill names the calm-air temperature that would produce the same modeled heat loss from exposed facial skin.
The thermometer stays put. The wind changes the rate.
Make −19 fall out
These controls stay inside the NWS formula's domain: air at or below 50°F and wind above 3 mph.
Wind chill−19°Fsame modeled loss as calm air at −19°F
Index gap19°Fbelow the thermometer, not below-air matter
NWS chart band30 minpublished example, not a personal timer
Bands come from the NWS chart at its tabulated air temperatures (40°F to −45°F in 5°F steps) and winds (5 to 60 mph in 5 mph steps). Other inputs show “not tabulated”; no frostbite time is inferred between cells.
35.74 + 0.6215(0) − 35.75(150.16) + 0.4275(0)(150.16) = −19.40°F → −19°F
This is a standardized dry, exposed-face model, not a forecast of your sensation. It assumes a 5-foot face height converted from a 33-foot anemometer, an adult walking into the wind, fixed tissue resistance, and no sun under a clear night sky. NWS says bright sun can make the index read 10°F to 18°F warmer.
The floor that wind cannot break
A dry object loses heat faster in moving air, but its passive-cooling destination is still the air temperature. This illustrative lumped model starts both dry surfaces at 68°F and advances them for 20 minutes. Its time constants are declared choices, not NWS physiology.
air 0°F
dry pipe
skin proxy
wet surface
Dry pipe after 20 min17.5°Fapproaches air, never crosses
Skin proxy after 20 min1.7°Fillustrative passive model only
Wet-bulb limit−1.8°Fwet toggle is off
T(t) = Ttarget + (68°F − Ttarget)e−t/τ. Dry targets equal air. The wet target is the psychrometric wet-bulb temperature, solved iteratively from saturation vapor pressure at the selected humidity.
Evaporation, not wind alone, permits sub-air cooling. Wind can speed and sustain evaporation. Actual pipe and skin temperatures depend on geometry, material, internal heat, radiation, moisture supply, and time, none of which wind chill specifies.
The index began as a bottle
2001: an exposed face
The US and Canada replaced that calibration with a steady-state face model and a regression formula. At 0°F and 15 mph, the older US formula reads −31°F, while today's formula reads −19°F.
12°F colder
That gap is not a deception. It is what changes when a small freezing cylinder, high-level wind readings, and an older equivalent-temperature conversion are replaced by a face-level model. The older formula is shown only at this published modern example, not extended as a second safety scale.
The check
The live substitution computes WCT(0,15) = −19.397954...°F, which NWS publishes as −19°F with exposed skin freezing possible in 30 minutes. The companion preset solves the rounded NWS radiator example backward: WCT(−5, 25.55) = −31°F. NWS does not state the wind speed in that example. Its dry-object floor remains −5°F, exactly as NWS says.
Free choices and uncertainties: the cooling race uses a 68°F start, 20-minute elapsed time, and illustrative time constants τpipe = 50/(1 + 0.08V) minutes and τskin = 25/(1 + 0.12V) minutes. They demonstrate an exponential floor, not a measured pipe or a living face. Wet bulb uses a standard sea-level psychrometric relation at 101.325 kPa; pressure, radiation, moisture supply, and geometry would move a real surface. NWS frostbite colors are estimates from a facial-cooling model and are not personal exposure guidance. Equal WCT does not guarantee equal freezing time.
Above freezing, wind chill cannot create frostbite. NWS is equally clear that hypothermia can still occur. This page informs; for cold safety, use current guidance from the National Weather Service and public-health authorities.
Corrections
Corrected 2026-10-14: The summary said evaporation pulls anything wet below air temperature. Evaporation can cool a wet surface below air temperature in unsaturated air; at 100% relative humidity, wet-bulb and air temperatures are equal (found by the assay line, 2026-10-14).
Corrected 2026-10-14: The summary dated the Antarctic calibration to 1945. Siple and Passel's measurements were made in 1941; their results were published in 1945 (found by the assay line, 2026-10-14).
Corrected 2026-10-14: The initial cooling diagram labeled the air −5°F. Its default input and displayed cooling results use 0°F; the initial label now reads 0°F (found by the assay line, 2026-10-14).
Corrected 2026-10-14: The readout assigned NWS frostbite bands from wind chill alone, giving 30 minutes at 0°F and 55 mph. The NWS chart puts that combination in its 10-minute band. The readout now uses the chart's air-temperature and wind-speed cells and marks other inputs as not tabulated (found by the assay line, 2026-10-14).