The three numbers

What the wall thermometer tells you

A thermometer reports one thing: the temperature of the air immediately surrounding the sensor. One number, one sensor, one point in space. The reading is only as good as where the instrument sits.

That siting question was settled in 1864. Stevenson 1864 described a louvered wooden enclosure — the Stevenson screen — that became the global reference instrument. The rules: 1.2 m to 2.0 m above ground, freely ventilated but shielded from direct sun, away from paved surfaces that radiate heat, away from HVAC exhausts. The WMO-No. 8 CIMO Guide codifies this as the international standard a century and a half later. Every official air-temperature reading you've ever seen came from an instrument following those rules — or should have.

Two thermometers 50 m apart can disagree by 10 °C and both be correct. Asphalt radiates stored heat into the air column above it; grass stays cooler through evapotranspiration. Still air in a sunlit courtyard is 5–8 °C warmer than air in a shaded breezeway twenty meters away. Rooftop instruments read differently from ground-level ones. "Air temperature" is always a local measurement, not a property of the sky.

The figure below shows what the Stevenson screen shields the thermometer from. Inside the louvered box, the sensor sees only dry-bulb air temperature. It does not see relative humidity. It does not see how hard the sun is working on a body standing next to it. It does not see the wind carrying heat away from exposed skin. It does not see whether the person reading it has been working for four hours or four minutes.

A wall thermometer in direct sun, mounted on a south-facing brick wall, above a parking lot, next to a dryer vent, will read 15 °C high. A screen-mounted station instrument two blocks away will read correctly. Both numbers appear in the same "current conditions" dashboard.

To start accounting for humidity, the next chapter introduces the heat index.

What it feels like

The body cools by sweating. Sweat evaporates and carries heat away — but only if the surrounding air is dry enough to absorb vapor. Humid air is already nearly saturated. The evaporative gradient collapses. Sweat stays on the skin. The body's primary cooling pathway narrows. Apparent temperature — the perceived thermal load — rises.

Steadman 1979 built the first systematic model of this effect: a lookup table combining dry-bulb temperature and dewpoint, calibrated against a clothed human body walking at a moderate pace. The table was rigorous but not broadcast-ready. In 1990, NWS SR-90 — Lans P. Rothfusz at NWS Fort Worth — fit a polynomial regression to Steadman's table. That regression is what every NWS heat index forecast has run since, and it's what this page implements.

The polynomial does something the raw table obscures: it makes the humidity gradient visible as a single number. At 35 °C air temperature and 30% relative humidity, the heat index is roughly 34 °C — barely above the thermometer reading. At 35 °C and 75% RH, the heat index reaches 41 °C. Same air temperature. Seven degrees of felt difference, driven entirely by moisture in the air.

The nomogram below makes this gradient visible. The horizontal axis is air temperature; the vertical axis is relative humidity. The umber contour lines are heat-index isopleths — the same "feels like" temperature traces diagonally across a wide range of T/RH combinations. Two annotated points sit at 35 °C air temperature: one dry day (30% RH) and one humid day (75% RH). The gap between them is not a rounding artifact. It's the difference between heat illness risk levels.

What the heat index does not see: solar radiation. A worker standing in direct midday sun absorbs 700–900 W/m² of radiant load above and beyond the air temperature. Wind speed. Convective cooling from air movement over exposed skin is substantial — a 10 km/h breeze at 35 °C provides meaningful cooling that the Rothfusz polynomial never accounts for. Work intensity. A body generating metabolic heat from carrying loads adds internal heat production that no weather measurement captures. Clothing. PPE — hard hats, Tyvek suits, gloves — traps heat between the body and the air no matter what the thermometer says.

To account for sun, wind, and the actual workload on a body, the next chapter introduces WBGT.

What it does to a working body

Wet-bulb globe temperature is not a weather metric. It's an occupational one. WBGT was designed to answer a specific question: given this environment — the air, the sun, the wind — how fast is the heat load on a working body accumulating? Air temperature and heat index don't answer that question. WBGT does.

The origin is precise. In the 1950s, recruits at Parris Island, SC were dying during summer training. Yaglou & Minard 1957 — a CSM-CDC collaboration — instrumented the training grounds and found that a three-thermometer weighted average predicted heat casualties better than any single measurement. That weighted average is WBGT. Every occupational heat standard since — NIOSH, ACGIH, ISO, OSHA — traces directly to that Parris Island work.

The three thermometers work as follows. The dry-bulb thermometer (Td) is ordinary air temperature — the same reading as the Stevenson screen. The natural wet-bulb thermometer (Tnwb) is a standard thermometer with a water-soaked wick wrapped around the bulb, exposed to natural airflow — not artificially ventilated. Evaporation from the wick cools the reading; how much it cools depends on the combined effect of humidity, wind, and radiant load. The black-globe thermometer (Tg) is a matte-black 150 mm hollow copper sphere with a thermometer at its center. It absorbs solar and reflected radiation, rises above air temperature in proportion to the radiant load, and falls back toward air temperature in wind. The outdoor WBGT formula combines all three: WBGT = 0.7·Tnwb + 0.2·Tg + 0.1·Td.

The 0.7 weight on the wet bulb is not arbitrary. It reflects the dominance of evaporative cooling in human thermoregulation. When a body is working hard in heat, sweat evaporation accounts for the majority of heat dissipation — more than convection, more than radiation. The wet bulb is the best single sensor for whether evaporative cooling is available. The 0.2 on the globe and 0.1 on the dry bulb account for radiant and convective load respectively.

What WBGT still doesn't see: clothing and PPE. A Tyvek suit insulates and blocks evaporation; NIOSH 2016-106 §6 provides WBGT correction factors by garment type that add up to 10 °C equivalent load. Acclimatization. A heat-adapted worker can sustain labor at WBGT values 5–7 °C higher than a first-week new hire — NIOSH 2016-106 §6 and ACGIH 2025 TLV both carry separate columns for acclimatized vs. unacclimatized workers. Individual fitness and health. Work rate. The NIOSH and ACGIH tables are keyed to four work-intensity levels: light, moderate, heavy, very heavy — the metabolic heat generation per hour is as important as the environment.

Measurement protocol matters. ISO 7243 specifies sphere diameter, wick material, ventilation conditions, and calibration procedure. A WBGT reading made outside ISO 7243 protocol isn't a defensible occupational measurement — it's an approximation. This page computes approximations. An inspector's reading at your jobsite, made with calibrated instruments under ISO 7243 protocol, is the defensible one.

With three numbers in hand, the next chapter compares them on the same days.

Where each metric fails

An honest accounting. Each metric was built to answer a specific question. None was built to be a universal oracle.

Each card above lists what that metric was not designed to see. The limits are not flaws — they reflect the design constraints of instruments built for specific purposes. A wall thermometer was designed for weather observation, not occupational safety. The Rothfusz heat-index polynomial was fit to forecast comfort for a pedestrian in shade, not for a construction worker in direct sun wearing PPE. WBGT was designed to predict heat casualties on a military training ground — a good proxy for occupational settings, but still missing variables that a full physiological model would need.

The right way to think about them is as a layered sensor system, not as competitors. Air temperature provides screening: if it is cool, none of the other metrics will alarm. Layer heat index for humidity-aware comfort warnings: if the air is warm and humid, the felt load rises. Escalate to WBGT for jobsite work/rest calls: if WBGT is high enough to cross NIOSH thresholds, the body's heat balance is under measurable strain regardless of what the thermometer says.

One variable is hidden across all three metrics: time of exposure. None of them captures how long a worker has been in the heat. A WBGT reading at 28 °C matters differently for a worker who has been outdoors for six hours than for one who arrived fifteen minutes ago. Acclimatization protocols — the NIOSH recommendation to ramp up heat exposure over the first seven to fourteen days — address this separately, because the numbers alone cannot.

The next two chapters show how to use this in practice: first with a sandbox calculator where you can enter your own conditions, then with a live query to the National Weather Service for current conditions at any U.S. station.

How professionals use these

The previous chapters introduced three metrics and the divergence between them. This chapter is for the practitioners — the safety officers, compliance specialists, EHS managers, and field inspectors who have to convert these numbers into specific work/rest decisions on specific days.

The short version: the wall thermometer is the worst of the three for any decision that matters. Heat index is what triggers heat-priority days under the OSHA Heat NEP — not because it's the most physiologically complete metric, but because it's the most widely reported one. WBGT is what supports specific work/rest orders on specific jobsites — and a citation defense without WBGT measurements is structurally weaker than one with.

The table below is the NIOSH 2016-106 §6 recommended work/rest schedule for an acclimatized worker. It's read by crossing WBGT (left column) with work intensity (top row); the cell value is the maximum minutes of work per hour at that combination. Above 32.2 °C WBGT, NIOSH recommends stopping work entirely.

How this plays in practice. An OSHA inspector arriving at a jobsite with a calibrated WBGT meter and a documented work-intensity classification can write a citation that references this table directly. The same inspector with only an air thermometer cannot — there is no NIOSH or ACGIH work/rest table keyed to air temperature. Heat index is the trigger for a heat-priority day, but it is not the tool for specific work/rest orders; that is WBGT's job. NIOSH 2016-106 §6 The federal NPRM, when finalized, will likely require employers to use Heat Index for the initial trigger and to maintain WBGT-equivalent measurement capability for the high-heat trigger. Federal NPRM 2024

For unacclimatized workers — new hires, returning workers after a 7+ day absence — these thresholds shift downward by 2.5–3 °C WBGT. The ACGIH TLV documentation has the adjustment tables; NIOSH §5 has the canonical acclimatization protocols. The first 14 days of new heat exposure carry the highest risk; NIOSH recommends a staged ramp where workers build from 20% of expected heat load to 100% over five days. Most heat fatalities in published OSHA case files occurred within the first three days on the job — before any physiological adaptation can take hold. ACGIH 2025 TLV

The lab ends here. The next stop, if you want to follow how this site was built and what trade-offs ran through it, is the colophon below.