Why Pressure Keeps Killing People: A Field Guide to Six Units That Won't Go Away

1. The Pascal: A Unit So Small It's Almost Useless

A single pascal is the pressure exerted by a one-newton force spread over one square meter. How much is that? Take a standard sheet of A4 paper �?80 gsm, 210 × 297 mm. Its weight is about 0.05 N. Lay it flat on a table and it presses down at roughly 0.8 pascals. That's a pascal: less than the weight of a sheet of paper distributed across a square meter.

This is the only SI unit of pressure. Everything else �?bar, psi, atm, torr, mmHg �?is tolerated for historical or practical reasons. The pascal is defined cleanly from the SI base units: 1 Pa = 1 kg·m⁻¹·s⁻�? No fudge factors, no calibration artifacts, no "at a specific temperature." It is as exact as the kilogram, meter, and second are exact.

The unit is named after Blaise Pascal (1623�?662), the French mathematician who, at age 19, built a mechanical calculator to help his tax-collector father. Pascal's contribution to pressure measurement came in 1648 when he convinced his brother-in-law, Florin Périer, to carry a mercury barometer up the Puy-de-Dôme volcano in central France. Périer took readings at the base, the summit (1,465 m), and several points in between. The mercury column dropped as altitude increased �?proving, for the first time, that air pressure decreases with height. It also proved Torricelli right: the mercury wasn't held up by some mysterious vacuum-abhorring force. It was the literal weight of the atmosphere pressing down on the dish.

Pascal published the results in Récit de la grande expérience de l'équilibre des liqueurs (1648). The experiment cost him his relationship with Descartes, who claimed to have suggested the idea first. It made Pascal's reputation across Europe. And in 1971, the 14th CGPM named the SI unit of pressure after him �?a unit that Pascal himself never used, because he worked entirely in pieds d'eau and pouces de mercure (feet of water and inches of mercury).

In semiconductor fabrication, though, pascals are everything. A plasma etch chamber operates at 0.1�?0 Pa. A molecular beam epitaxy system runs at 10⁻⁸ Pa �?one hundred-trillionth of atmospheric pressure. At those levels, a fingerprint on a chamber wall outgasses enough water vapor to ruin a wafer batch. Vacuum engineers count pascals because, at that end of the scale, a pascal is enormous. A 1 Pa rise in a UHV chamber means a leak somewhere, and the difference between 10⁻⁸ and 10⁻⁷ Pa is the difference between epitaxial growth and contaminated junk.

2. The Weight of the Sky: Atmospheres, Bars, and What Torricelli Actually Did

In 1643, Evangelista Torricelli filled a glass tube �?about a meter long, sealed at one end �?with mercury. He stoppered the open end with his thumb, inverted the tube into a dish of mercury, and removed his thumb. The mercury column dropped. It didn't empty. It settled at about 760 mm above the dish surface, leaving a vacuum above it.

This was the first barometer and the first measurement of atmospheric pressure. Torricelli correctly guessed that the column wasn't held up by nature's supposed horror of a vacuum �?a doctrine that had stood unchallenged since Aristotle �?but by the weight of the air pressing on the mercury in the dish. He wrote to his friend Michelangelo Ricci in 1644: "We live submerged at the bottom of an ocean of air."

That single observation defined almost every pressure unit we use today.

The atmosphere (atm)

One standard atmosphere is 101,325 pascals exactly. This was defined by the 10th CGPM in 1954, not by measuring the atmosphere �?which varies with weather, altitude, and latitude �?but by convention. It approximates the average sea-level pressure at 45° latitude. In practice, actual sea-level pressure wanders between about 98 and 104 kPa depending on the weather. The "standard atmosphere" is a legal fiction, useful precisely because it doesn't change.

Chemists love atmospheres. STP �?Standard Temperature and Pressure �?is 0°C and 1 atm. Every gas law calculation starts here. A mole of ideal gas occupies 22.414 liters at STP. Change the pressure unit and all the tables need rewriting. That's why atm survives: not because it's metric or tidy, but because too much chemical engineering paper is printed with it.

The bar

The bar equals 100,000 Pa exactly. It was introduced by the Norwegian meteorologist Vilhelm Bjerknes, one of the founders of modern weather forecasting. Bjerknes wanted a unit that was close to atmospheric pressure (so weather maps would show familiar numbers) but metric in spirit (a clean power of ten from the pascal). 1 bar = 100 kPa = 0.1 MPa. One millibar �?a thousandth of a bar �?is 100 Pa, or almost exactly 0.75 mmHg. Weather maps worldwide report pressure in millibars (or the numerically identical hectopascals) for this reason.

The bar is not an SI unit. The BIPM tolerates it but doesn't endorse it. And yet the bar is everywhere: European tire pressures, SCUBA tank ratings, espresso machine pumps, natural gas pipeline pressures, CO�?cylinder fills for home brewing. It's the metric unit that metric purists hate and working engineers reach for.

The bar's clean decimal relationship to the pascal �?100,000, no fractions �?makes conversions within the metric family painless. 1 bar = 100 kPa = 0.1 MPa = 1,000 mbar. You can do them in your head. The problems start when bar meets psi.

The 14.5038 factor is ugly. There's no way around it. The bar derives from the pascal, which derives from SI base units. The psi derives from the pound-force and the inch, which derive from an avoirdupois pound under standard gravity and the 1959 international inch. The conversion accumulates three layers of definitional cruft. You can't simplify it. You either use the exact number or you're wrong.

Quick mental shortcut that does work: multiply bar by 14.5. For anything under 10 bar, the error is under 0.04 psi per bar �?negligible for tire gauges, weather, home espresso. For anything over 100 bar, use a calculator. At 200 bar (a full SCUBA cylinder), 14.5 gives 2,900 psi instead of 2,900.76 psi �?a 0.76 psi error that doesn't matter for a recreational dive plan. At 345 bar (5,000 psi wellhead equipment), the shortcut gives 5,002.5 psi instead of 5,003.811 psi �?still only 1.3 psi off. But here's the thing: at those pressures, the steel cares about 1.3 psi. The bar to psi converter uses the exact factor. Use it.

3. PSI: The American Outlier

The pound per square inch is a purely mechanical unit. 1 psi is exactly the pressure exerted by one pound-force distributed over one square inch. A pound-force is the gravitational force on one avoirdupois pound at standard gravity (g₀ = 9.80665 m/s² exactly). An inch is 25.4 mm exactly (since 1959). Chase the chain:

That's twelve significant figures. In practice, 6,894.76 Pa is more than anyone needs. But the exact number exists because every component �?the pound-mass, standard gravity, the inch �?has an exact SI definition now. PSI is as precisely defined as the pascal. It's just inconveniently linked to it.

PSI dominates the United States, Canada (alongside kPa), and partly the UK. Car tires: psi. Natural gas distribution: psig (gauge). Hydraulic systems: psi. SCUBA tanks in North America: psi. The American Society of Mechanical Engineers writes its Boiler and Pressure Vessel Code in psi. Every ASME flange class �?150, 300, 600, 900, 1500, 2500 �?is a psi rating at a specified temperature.

The unit has two personalities. Below about 30 psi, it's consumer territory: tires, basketballs, bike shocks, air mattresses. The average car tire at 35 psi is something you can feel �?thumb into the sidewall, it gives but doesn't yield. Above about 1,000 psi, psi becomes an industrial unit. A 2,200 psi hydraulic system can sever a finger cleanly at the knuckle. A 4,500 psi breathing air cascade can turn a loose whip into a steel snake that breaks ankles. At 15,000 psi, waterjet cutters slice through titanium plate. At 60,000 psi, the fluid itself compresses by about 15% �?you're no longer dealing with an incompressible liquid at all.

If you're converting between psi and anything metric, use psi to bar or psi to kPa. The mental shortcuts �?divide by 14.5, multiply by 7 �?work until they don't.

4. mmHg and Torr: The Body's Own Pressure Scale

Blood pressure is measured in millimeters of mercury. Your 120/80 reading is 120 mmHg systolic over 80 mmHg diastolic �?the pressure your heart generates to push blood through your arteries, expressed as the height of a mercury column it could support. A sphygmomanometer �?the inflatable cuff your doctor wraps around your arm �?is literally a mercury manometer with the scale marked in mm. The unit survived digitalization. Even a modern oscillometric blood pressure monitor, which contains no mercury anywhere, reports in mmHg. It's too embedded in clinical practice to change.

One millimeter of mercury is defined as 133.322 387 415 Pa �?or, more practically, as exactly 1/760 of a standard atmosphere at 0°C, where the density of mercury is 13,595.1 kg/m³. The temperature specification matters because mercury expands. A column that reads 760 mm at 0°C reads 760.8 mm at 20°C. In a laboratory, this correction is mandatory. In a doctor's office, nobody adjusts for room temperature, and nobody has died from the error.

The torr is, for most practical purposes, identical to the mmHg. 1 Torr = 1/760 atm �?133.322 Pa. It's named after Torricelli and used almost exclusively in vacuum engineering. American vacuum pump manufacturers quote ultimate vacuum in torr. European ones use mbar. A two-stage rotary vane pump might pull 0.1 mtorr (1 × 10⁻⁴ torr). A turbomolecular pump goes to 10⁻¹⁰ torr. A cryopump can hit 10⁻¹�?torr �?the pressure regime where individual molecules are countable rather than statistical.

The mmHg/torr near-equivalence is both convenient and dangerous. 1 mmHg = 1.000 000 Torr at 0°C, but 1 mmHg = 1.000 015 Torr under standard conditions with realistic mercury density corrections. For vacuum gauging, ignore the difference. For primary pressure metrology, the difference is the entire error budget. Pick one and stay with it.

At UHV pressures, you stop thinking about fluid mechanics and start thinking about molecular ballistics. A molecule desorbing from the chamber wall travels line-of-sight until it hits another surface. The concept of "pressure" as a continuous field breaks down. The vacuum gauge is counting individual impacts, not measuring a bulk property. Yet the units persist: torr in the US and Japan, mbar in Europe, Pa in China. Three different numbers on three different gauges, all on the same chamber.

5. Megapascals and Thousands of PSI: Where Things Get Dangerous

At about 1 MPa (145 psi), pressure stops being an inconvenience and starts being stored energy. A 1-liter volume at 1 MPa contains roughly 1 kJ of potential energy �?about the same as a 1 kg mass dropped from 100 meters. Release it suddenly and you get shrapnel, not a hiss.

Hydraulic systems

Mobile hydraulic systems run at 21�?5 MPa (3,000�?,000 psi). Industrial presses go to 70 MPa (10,000 psi). The failure mode of a pinhole leak at these pressures isn't a drip �?it's an invisible jet that can inject hydraulic fluid through skin, into muscle, and along fascial planes. An injection injury at 3,000 psi requires immediate surgical debridement. The entry wound looks like a paper cut. Six hours later the arm is being amputated.

This is a pressure story, not just a safety story: hydraulic injection injuries happen most often when a mechanic uses a fitting rated in bar on a system spec'd in psi, or vice versa. A JIC 37° flare fitting rated for 5,000 psi looks identical to a metric 24° cone fitting rated for 350 bar �?and 350 bar is 5,076 psi, close enough to "fit" but not close enough to be safe if the actual system pressure is 5,000 psi on a metric-rated fitting. The fitting fails. The jet finds the mechanic's hand. Two weeks later the occupational health report reads "pressure unit confusion."

Gas cylinders and cascade systems

A standard SCUBA cylinder is a steel or aluminum bottle at 200 bar (Europe) or 3,000 psi (US). A carbon-fiber wrapped cylinder goes to 300 bar / 4,500 psi. A breathing air cascade �?the bank of large cylinders a dive shop uses to fill individual tanks �?sits at 350�?50 bar (5,000�?,500 psi).

The fill operator's nightmare: a diver hands over a cylinder marked "2,400 psi service pressure" but the shop's compressor fills to 232 bar (European standard). 232 bar = 3,365 psi. Into a 2,400 psi cylinder. The burst disc blows �?that's the good outcome. The bad outcome is a burst disc that's been painted over during a sloppy hydro test, doesn't rupture, and the cylinder fails at the neck thread. The hydraulic energy in a single 80 ft³ cylinder at 3,365 psi is about 1.3 MJ. Enough to go through a concrete block wall.

The ASME flange problem

ASME B16.5 pipe flanges are designated by pressure class: Class 150, 300, 600, 900, 1500, 2500. Those numbers look like psi but they aren't. A Class 300 flange is not rated for 300 psi at all temperatures. At room temperature with carbon steel, Class 300 is good for 740 psig. At 400°F, it drops to 655 psig. At 800°F, 500 psig.

The number is a pressure-temperature rating class, not a pressure. It derives from the flange's bolt circle geometry and allowable material stress at temperature, not from a simple conversion. But in the field, people say "it's a 300-pound flange." And someone who doesn't know the code hears "rated for 300 psi" and bolts it into a 285 psi steam line. It won't fail today. It'll fail in five years, after thermal cycling has crept the bolts, on the coldest night of the winter, when the steam load is highest. The investigation will blame the gasket or the bolt torque. The root cause was someone confusing a class designation with a pressure rating.

European codes (EN 1092-1) use PN ratings: PN 10, 16, 25, 40, 63, 100, 160, 250, 320, 400. The number is the maximum pressure in bar at room temperature. A PN 40 flange is rated for 40 bar at ambient. Simple. But a PN 40 flange is not interchangeable with a Class 150 flange �?PN 40 / 40 bar = 580 psi, while Class 150 at room temp = 285 psig max. If you bolt a PN 40 flange to a Class 150 flange because "40 bar is about 600 psi, way above 150, should be fine" �?the bolt circles don't match, the gasket surfaces don't match, and you've created a leak path that will find you at 3 AM during the first cold start.

Use the kPa to psi and psi to kPa converters for these cross-code comparisons. Better yet, never convert flange ratings. Use the matching flange.

6. The Master Conversion Table

This is every pressure unit relationship covered by the converters on this site. All factors are given to six significant figures except where the conversion is exact by definition. Print this or bookmark it �?it's the reference you'll reach for when the gauge doesn't match the spec sheet.

7. What Goes Wrong and Who Died

Every unit conversion disaster is a pressure disaster. Length errors produce scrap parts. Mass errors produce overweight aircraft. Pressure errors produce explosions. Here are the ones that changed how engineers think about units.

Reading these investigations side by side reveals a pattern that isn't about unit conversion tables. It's about pressure as stored energy. A length conversion error produces a gap. A mass conversion error produces an overload. A pressure conversion error produces a bomb. The unit is always, in the final accounting, a proxy for how much energy is waiting to be released.

8. Every Pressure Converter on This Site

Six converters covering the five pressure units that working engineers deal with daily. Every one uses the exact factors from Section 6. No rounding, no shortcuts. Drop in a number and get the right answer.

Bar ↔ PSI �?the automotive and diving pair

The one you'll use every time a European car's door placard meets an American tire gauge, or a European dive computer meets an American fill station. 1 bar = 14.5038 psi.

kPa ↔ PSI �?the engineering spec crossover

ISO specs are in kPa. ANSI specs are in psi. The piping spec, the relief valve datasheet, the turbocharger compressor map �?all of them need this pair. 1 psi = 6.89476 kPa.

atm ↔ Pa �?the chemistry and physics pair

STP calculations, gas law problems, vacuum chamber specifications. The atm is 101,325 Pa by international convention; the pascal is 1 N/m² by SI definition. The converter handles both directions exactly. 1 atm = 101,325 Pa.

Frequently Asked Questions About Pressure Conversion