A barometer is an instrument that measures atmospheric pressure: the push exerted by the surrounding air. Mercury barometers balance that pressure against a liquid column, aneroid barometers use a flexible metal capsule, and digital barometers convert the movement of a tiny sensing element into an electrical reading. Understanding how barometers work explains why their readings change with weather, height and the way the instrument is installed.

Reading a barometer properly means asking more than whether its pointer says fair or rain. You need to know the pressure unit, whether the value is measured at the instrument or adjusted to sea level, and how it has changed over time. A falling reading can be useful evidence about an approaching weather system, but it is not a promise that rain will fall at your particular window.

This guide explains barometric pressure from first principles, compares mercury, aneroid and electronic instruments, and shows how to interpret readings without confusing altitude with weather. It includes worked calculations, a practical observation routine, common faults, and clear limits on what a barometer can tell you. The same ideas connect a traditional wall instrument with pressure sensors used in weather stations, watches and navigation devices.

The 50-second explanation

Imagine the atmosphere as a moving, compressible ocean of gas. Gravity holds it around Earth, and molecules continually strike everything immersed in it. A barometer responds to this pressure through a controlled physical change: a column rises, a capsule flexes or a microscopic membrane moves.

The instrument measures pressure where its sensing opening is located. Software may then adjust that measurement to an agreed reference height or use it to estimate altitude. Those extra numbers are interpretations, not separate direct measurements.

The useful question is therefore not simply, “What number does the barometer show?” It is, “Which pressure does this number represent, where was it measured, and what else changed?” Answering those questions turns a decorative dial into a meaningful scientific observation.

Begin with the air, not the pointer

Why air can push

Air is made of matter. Its molecules move and collide with surfaces, transferring momentum. Pressure describes force distributed over area. A pressure of one pascal corresponds to one newton acting on one square metre. An atmospheric pressure reading is not a measurement of the speed of one gust; it describes the local pressure of the air around the sensor.

Gravity also matters. A lower layer of the atmosphere supports the weight of air above it. Moving upward generally leaves less air overhead, so pressure decreases. The atmosphere is not a stack of rigid shelves, however. Temperature, density, moisture and large-scale motion influence how pressure changes with height and across a region.

Think of a sealed flexible packet carried upward. The pressure outside falls while the trapped air initially retains its previous condition. The packet can expand until its material and the pressures find a new balance. A barometer uses a more carefully controlled version of a pressure-dependent physical response.

Why atmospheric pressure does not simply crush us

Pressure acts in many directions, and the fluids and gases inside the body also exert pressure. The relevant issue is usually a difference between pressures, not the atmospheric value considered alone. A closed container can collapse when the pressure inside becomes much lower than the pressure outside, even though the surrounding atmosphere has not suddenly become unusually heavy.

This distinction helps explain the instrument itself. A sensing membrane needs a reference on its other side. A barometer designed to measure absolute pressure commonly compares surrounding air with a sealed low-pressure reference. The resulting force difference produces a small movement that can be measured.

For the wider atmosphere, continue to How weather works. This article concentrates on the measuring instrument and the interpretation of its output.

The units on a barometer

Weather reports commonly express pressure in hectopascals, abbreviated hPa. One hectopascal equals 100 pascals. A millibar has the same numerical value as a hectopascal, so 1008 hPa and 1008 millibars describe the same pressure. The standard atmosphere is 1013.25 hPa; it is a reference value, not the pressure that every seaside location must display. These relationships are explained in NOAA’s air-pressure guide.

Older instruments may use inches of mercury or millimetres of mercury. Those scales connect pressure to the height of a reference mercury column. Digital instruments sometimes offer kilopascals: 1000 hPa equals 100 kPa. A reading that looks ten times or one hundred times different may be a unit change rather than a dramatic atmospheric event.

Always record the unit beside the number. Writing only “1010” makes a notebook ambiguous. Also record whether the display is showing pressure or calculated altitude. Devices with several screens can make an innocent mode change look like a sensor fault.

Precision of display is a separate issue. A screen that shows 1008.37 hPa has more displayed digits than one showing 1008 hPa. It does not automatically have better accuracy. Some of those digits may reflect electronic resolution rather than dependable knowledge about the atmosphere.

How a mercury barometer works

A traditional mercury barometer uses a vertical tube closed at the top, with its lower end immersed in a reservoir. Atmospheric pressure acts on the exposed reservoir surface. The pressure supports a column of mercury inside the tube, with a low-pressure space above the column.

The basic balance is expressed as pressure difference equals liquid density multiplied by gravitational acceleration and vertical height. Using symbols, this is delta P = rho g h. A denser liquid needs a shorter column to balance the same pressure. That is why mercury historically allowed a practical instrument height, whereas a water column would need to be much taller.

The height must be measured vertically. Tilting the tube changes the length along the tube without changing the vertical height in the same way. Temperature also affects the liquid and scale. Precision use therefore requires attention to conditions, reference points and instrument construction, not merely looking at the top of the shiny column.

A mercury barometer is scientifically elegant because the measured pressure is linked to a visible mechanical balance. Its disadvantages are equally concrete: glass can break, mercury is hazardous, and transport and maintenance require care. It is best discussed as an important instrument design, not as an invitation to make one at home.

Do not open, refill or experiment with an old mercury instrument. Keep a damaged instrument away from handling and follow local hazardous-material guidance. Modern learning activities can demonstrate pressure changes without introducing mercury or fragile liquid columns.

How an aneroid barometer works

An aneroid barometer uses no liquid column. A sealed, flexible metal capsule changes shape slightly as external pressure changes. Springs and the capsule’s own stiffness resist that motion. Levers then magnify the tiny movement and transmit it to a pointer moving across a scale.

The pressure-sensitive element is the essential part. The dial is only the display. A beautifully engraved face cannot compensate for a damaged capsule, a sticking linkage or poor calibration. The Met Office’s instrument archive describes an aneroid instrument in which changing pressure alters a spring force and a lever moves the pointer.

Mechanical magnification introduces compromises. A tiny amount of friction at a pivot can delay movement. Spring properties can change with temperature or age. The same pressure may produce slightly different readings depending on whether the pressure was approaching from above or below; this history-dependent behaviour is called hysteresis.

Some aneroid instruments include a manually movable reference hand. It is not a second sensor. Set it to an earlier reading, then compare the active pointer later. The separation gives an immediate visual indication of pressure change. Confusing these two hands is a common reason people think their barometer is broken.

A dial marked stormy, change, fair or very dry is an interpretive overlay. Those words are not physical measurements. Their usefulness depends on climate, location and the particular pressure pattern. Read the numerical scale first, then consider the weather context.

How a digital barometer works

A digital barometer starts with a pressure-sensitive structure, often a small diaphragm. Pressure changes its shape by a tiny amount. Electronics detect an associated change in capacitance, resistance or another measurable property. A processor converts the signal into pressure units using calibration information.

In a capacitive design, movement changes the relationship between conductive surfaces. In a piezoresistive design, mechanical strain changes electrical resistance. These are different measurement principles, not merely alternative names for the same circuit. Vaisala documents a silicon capacitive barometer with temperature and linearity compensation, while Bosch describes a piezoresistive membrane used for barometric sensing.

The device must also know how temperature affects its own sensing element. Compensation corrects predictable instrument behaviour; it does not remove every uncertainty. Mechanical stress from mounting, contamination at the pressure opening and changes after manufacture can still affect the result.

The sensing opening needs access to the intended air pressure. A protective enclosure may keep rain and dust out while allowing pressure to equalise. An airtight box, blocked port or unsuitable protective coating can make the sensor measure the wrong environment or respond too slowly.

Digital output makes logging easier. A device can store readings, plot trends and combine pressure with other sensors. That convenience introduces new questions: how often is it sampling, how much filtering is applied, and has software changed the reference? A smooth graph is not automatically a more truthful graph.

Three different numbers that can all be called pressure

Pressure at the instrument

Station pressure is the pressure actually measured at the observing station’s height. A sensor on a hill normally reads less than one near sea level, even when both experience a similar weather system. This is a physical consequence of elevation, not evidence that the hill always has stormier weather.

For experiments comparing instruments, put them at the same height and allow them to settle under similar conditions. Comparing a ground-floor sensor with a rooftop sensor without accounting for height mixes instrument differences with real atmospheric differences.

Pressure adjusted to sea level

Weather maps often use pressure reduced to a sea-level reference so stations at different elevations can be compared more meaningfully. Without an adjustment, the pressure pattern would be dominated by mountains and valleys. The correction depends on a model of the atmosphere between the station and the reference level.

An adjusted value is useful, but it remains a derived quantity. Different assumptions can produce different adjusted pressures from the same sensor reading. Record the reference setting rather than assuming every phone application and weather website displays the same kind of number.

Pressure used to infer altitude

A barometric altimeter turns the pressure-height relationship around. Instead of asking what pressure exists at a known height, it asks what height would correspond to a measured pressure under chosen reference conditions. Change those conditions and the calculated altitude changes, even if the instrument remains on the same desk.

This is why pressure alone cannot reliably separate “I climbed a hill” from “the weather changed while I stood still.” Outdoor and wearable devices may combine barometric information with satellite positioning or other measurements. Bosch lists altitude tracking among the applications of its miniature absolute pressure sensors.

Reading trends without turning them into promises

A single number describes one moment. A trend describes change over an interval. Record the present pressure, the earlier pressure and the elapsed time. “Falling by 3 hPa in three hours” is more informative than “falling,” because the latter could refer to a tiny overnight change or a much faster movement.

Pressure patterns are connected with atmospheric circulation, but a barometer is not a rain detector. Moisture, rising motion, temperature structure, local heating and the position of weather systems also matter. A high reading does not guarantee a dry afternoon, and a falling reading does not say exactly when or where precipitation will begin.

The distinction is especially important near the equator. NOAA describes a twice-daily atmospheric pressure cycle whose amplitude is strongest in tropical regions. It also warns that traditional weather rules developed for England do not transfer directly to the tropics. In Singapore, a small routine fall should not be turned into a confident storm forecast from the barometer alone.

An honest statement separates observation from interpretation: “Pressure at this fixed sensor fell during the observation period. That is consistent with several possible atmospheric changes; it does not establish a local rain forecast.” Additional weather observations can then strengthen or weaken a particular interpretation.

This approach is more useful than memorising a dial slogan. It teaches you to ask what evidence would distinguish competing explanations. A dark cloud, changing wind, radar image and official forecast can each contribute information that the pressure instrument does not contain.

Worked examples: make the number understandable

Example 1: converting units

Suppose a sensor reports 100650 Pa. Divide by 100 to express the same measurement in hectopascals: 1006.5 hPa. Divide by 1000 to express it in kilopascals: 100.65 kPa. Nothing happened to the air during these calculations. Only the numerical representation changed.

Now suppose another display reads 1006.5 millibars. It agrees numerically with the hectopascal value. A student who mistakes millibars for pascals could report a hundredfold discrepancy. Writing units at every step prevents a spectacular but completely artificial weather event.

Example 2: calculating a pressure tendency

At 09:00, a fixed sensor shows 1009.6 hPa. At 12:00, it shows 1007.8 hPa. The change is 1007.8 minus 1009.6, or minus 1.8 hPa over three hours. The mean change rate across that interval is minus 0.6 hPa per hour.

That average does not prove the fall was steady. The pressure might have fallen rapidly during the first hour and then stabilised. More frequent readings reveal the shape of the change. These invented observations demonstrate arithmetic only; they are not a forecast for any date or place.

Example 3: estimating a small height difference

For a short vertical interval, a simplified hydrostatic model gives delta P approximately equal to minus rho g delta z. Assume an air density of 1.2 kilograms per cubic metre and gravitational acceleration of 9.8 metres per second squared. For a 10-metre rise, the estimated pressure reduction is about 118 Pa, or 1.18 hPa.

The calculation is a local approximation using an assumed constant density. It is not a universal conversion for an entire mountain climb. Density changes with atmospheric conditions and altitude. Its educational value is to show that a movement of a few floors can produce a pressure change large enough to confuse a weather comparison.

Example 4: two displays disagree

A desktop station reports 1004.2 hPa while a nearby application shows 1012.1 hPa. Before adjusting anything, check whether the desktop value is station pressure and the application value is adjusted to sea level. Next check the observation time, location, units and elevation entered into the application.

Only after those checks does a calibration comparison become meaningful. Setting the desktop display to match an unrelated number can hide a correct measurement. Agreement is useful only when both instruments are supposed to be measuring the same quantity under comparable conditions.

Calibration, resolution and repeatability

Calibration establishes a relationship between an instrument’s output and reference values under stated conditions. It is not the same as pressing a zero button or moving a decorative dial hand. A one-point adjustment can remove an offset near that point while leaving other errors elsewhere in the measuring range.

Resolution is the smallest displayed or detectable increment. Repeatability describes how closely repeated measurements agree under similar conditions. Accuracy concerns closeness to the intended reference value. A device can repeat the same wrong number beautifully, and a display can show many digits without supporting an equally small uncertainty.

For practical observation, ask whether the uncertainty is small enough for the intended question. A student watching day-to-day tendencies does not need the same performance as a professional reference station. Conversely, a sensor adequate for noticing a broad trend may be inadequate for claiming that two nearby rooms differ by a very small pressure.

Temperature compensation, long-term drift, measurement range and response time should be considered together. Campbell Scientific’s description of a barometric sensor distinguishes pressure equalisation, response and power-up settling. Those distinctions show why reading immediately after switching on is not always equivalent to reading after the instrument has stabilised.

Do not confuse sensor noise with environmental variability. Rapidly changing last digits may be electrical noise, actual pressure fluctuations or both. Filtering can make a display easier to read, but excessive filtering can hide short-lived events. Choose the observation interval to match the phenomenon you are investigating.

Installing a barometer so it measures the intended air

Place the instrument where it can sense the air of interest without unnecessary disturbance. Avoid direct heating from strong sunlight, heaters or electronics. Protect it from water and follow its mounting instructions. An instrument designed for indoor use should not be left outside merely because atmospheric pressure is an outdoor phenomenon.

A pressure port near moving air can respond to local flow as well as the wider atmospheric condition. Building ventilation, opening doors and fans may create small pressure differences. A sudden change when a door shuts need not represent a distant weather system. The instrument has measured a local event, but the interpretation was wrong.

Keep the sensor at a fixed height for a weather diary. Record changes in its location or enclosure. When comparing two devices, give them a shared environment and time to settle. Moving one device between an air-conditioned room and a hot balcony creates temperature and installation changes alongside any atmospheric difference.

Water resistance deserves careful interpretation. A protective membrane or housing may allow air pressure to reach the sensor while excluding droplets. Blocking that path can slow response. Conversely, leaving an unprotected opening exposed can damage the instrument. Follow the device’s intended protection arrangement rather than improvising an airtight cover.

A safe observation activity for learners

Start with a working household weather station or barometer whose pressure display can be identified. Leave it in one location. Prepare a notebook with columns for date, time, pressure, unit, reference setting and brief observations. Choose two or three consistent times each day rather than checking only when the weather looks dramatic.

For several days, record what the instrument actually says before writing an explanation. Keep cloud observations separate from pressure values. “Dark cloud overhead” is an observation; “a low-pressure system caused that cloud” is an interpretation requiring more evidence. This simple separation prevents the diary from becoming a collection of guesses.

Plot time on the horizontal axis and pressure on the vertical axis. Choose a clear scale and avoid implying that the pressure became zero simply because the graph begins at a convenient lower bound. Label missing observations as missing rather than drawing invented measurements between them.

Next, inspect a pattern. Does a change repeat at similar times? Was the sensor moved? Did an air-conditioning schedule change? Do several observations support the same explanation? Compare with an official weather account without expecting one sensor to reproduce an entire observing network.

A second activity can compare pressure at two safely accessible levels of a building using the same portable instrument. Record the times and allow settling at each level. Return to the starting level and measure again. That return reading helps reveal whether the apparent height effect was mixed with drift or a changing environment.

Do not use this activity to guide aviation, climbing decisions or emergency weather action. Its purpose is to understand measurement. The strongest result may be a carefully explained limitation rather than an impressive-looking altitude estimate.

Diagnosing readings that seem wrong

The value differs from a weather website

First check reference pressure, elevation, observation time and station distance. A website may show a nearby airport’s sea-level value, while your sensor measures pressure inside a building above ground level. Those are not identical comparisons. Do not recalibrate simply to remove an unexplained difference.

The reading changes when the instrument moves

Separate vertical movement from temperature changes and enclosure effects. A sensor moved upstairs should generally see lower pressure, but carrying it inside a tightly closed case may delay equalisation. A warm hand against the instrument can also create a thermal transient. Repeat the route and include a return measurement.

The pointer barely moves

Check that you are watching the sensing pointer rather than the reference hand. Atmospheric changes may genuinely be small over the chosen period. Mechanical friction or a damaged capsule is another possibility, but diagnosis should follow the manufacturer’s instructions rather than forceful shaking or turning adjustment screws without a reference.

The display jumps or becomes implausible

Check power, units, selected mode and pressure-port condition. Consider whether the device was exposed to water or a temperature outside its specified range. A number that looks plausible after an unexplained reset is not proof of restored accuracy. Verification against a suitable reference is stronger than visual reassurance.

Common misconceptions worth correcting

A barometer does not measure humidity. A hygrometer does that. Combined weather stations can display both quantities because they contain different sensing functions. Pressure and humidity may interact in the atmosphere, but one does not serve as a direct substitute for the other.

A barometer does not directly measure wind speed either. Wind is air motion, while pressure is force per area. Differences in pressure help influence atmospheric motion, but a single stationary pressure reading does not specify the speed or direction of the wind outside.

A pressure change does not prove that the total atmosphere has gained or lost mass globally. Air can redistribute across regions and through vertical motion. Local pressure is a local observation within a much larger, moving system.

Nor is a digital instrument automatically superior to a mechanical one for every purpose. Digital logging and compactness are useful; a readable mechanical dial can be useful too. The relevant comparison is performance for the task, together with calibration, maintenance and interpretation.

What a pressure graph can hide

A graph is not the atmosphere itself. It is a selection of observations shown through choices about sampling, averaging and scale. A device that records every second collects a different description from one that stores an hourly average. Both can be appropriate, but they answer different questions.

Suppose a door closing creates a brief pressure pulse. An hourly average may barely show it, while a second-by-second graph makes it look prominent. Neither display must be wrong. The difference arises because the pulse is short compared with one recording interval and long enough to appear in the other.

A useful log states whether each point is an instantaneous sample, a moving average or an average over a completed interval. It also identifies gaps. When a sensor loses power for an hour, joining the surrounding points with a straight line is a graphical convention, not evidence that the pressure followed that line.

The vertical axis matters just as much. A change of 0.2 hPa can look enormous when a chart displays only a narrow band, and almost invisible when the axis spans hundreds of hectopascals. Read the numbers before reacting to the steepness. Visual drama depends partly on plotting choices.

For a fair comparison, align timestamps, units and pressure references. Comparing the top of one graph with the bottom of another can manufacture a disagreement. Comparing two curves after a software update can also be misleading if the app silently changed its averaging interval or sea-level correction.

Keep a copy of raw observations when possible. Correcting an interpretation should not require rewriting the original record. A note saying that an elevation setting was wrong is more useful than silently replacing old readings and losing the history of what happened.

Absolute pressure is not the same as pressure difference

A barometer normally reports absolute atmospheric pressure relative to a low-pressure reference. A different instrument may measure the pressure difference between two rooms, across a filter or between a pipe and the surrounding air. Both instruments use pressure-sensitive elements, but their connections and reference conditions are different.

Imagine two rooms at almost the same absolute pressure, with a small difference caused by a ventilation system. A general weather barometer may be poorly suited to resolving that small difference. A differential-pressure instrument connects to both rooms and measures their difference directly. Choosing the right reference can matter more than adding display digits.

Similarly, a tyre gauge commonly reports pressure above the surrounding atmosphere. Its zero does not mean the tyre contains a perfect vacuum. It means the relevant difference between tyre and ambient pressure is zero. Confusing absolute and gauge pressure can therefore create large errors even when both instruments work correctly.

These examples explain why a sensor should be selected from the question backward. First identify the quantity: absolute pressure, pressure difference, tendency or inferred altitude. Then choose an instrument, reference and procedure capable of answering that question. Starting with an attractive display and inventing a question afterward is less reliable.

Turning a reading into a defensible conclusion

A good conclusion carries its conditions with it. Instead of saying, “The weather is getting worse,” write, “The fixed indoor instrument showed a fall of 1.8 hPa between 09:00 and 12:00, using the same pressure reference.” That statement is narrow, but another observer can understand and check it.

A second sentence may offer an interpretation: “This could accompany a changing weather pattern, but the reading alone cannot locate rainfall.” A third may identify the next observation needed. The reasoning remains open to correction because the claim has not outrun the evidence.

This is not excessive caution. It is how a measurement becomes reusable. Someone revisiting the record later can distinguish the number, the conditions, the calculation and the explanation instead of having to disentangle them from one confident sentence.

Frequently asked questions

What exactly does a barometer measure?

It measures atmospheric pressure at its sensing location. Some displays then convert that measurement to sea-level pressure, altitude or a weather symbol. Those derived outputs depend on settings and models. Establish which quantity is being displayed before comparing values or drawing conclusions.

Why does pressure normally fall when I go higher?

There is generally less air above a higher position, so the weight supported by the air below is smaller. The relationship is not perfectly linear over large height changes because air density varies. For small differences, a simplified local model can provide a useful estimate with clearly stated assumptions.

Does low pressure always mean rain?

No. Pressure is one part of the atmospheric state. Rain also depends on moisture, cooling, uplift and the movement of weather systems. A low reading may be relevant evidence, but it cannot guarantee precipitation at a particular place. Local tropical weather especially resists simplistic dial-based rules.

Should I set my instrument to 1013.25 hPa?

Not simply because that is the standard atmosphere. It is a reference quantity, not a universal target for everyday weather. An adjustment should use an appropriate reference measurement and the correct pressure type. Forcing every reading toward a textbook value destroys the variation the instrument is meant to observe.

Why can two good barometers disagree?

They may be at different heights, have different pressure references, sample at different times or respond differently to temperature and airflow. Instrument uncertainty also matters. Before concluding that one is defective, make the comparison conditions genuinely equivalent and check the size of the disagreement against realistic specifications.

Can a phone be used as a barometer?

Some devices contain a physical pressure sensor, while applications on other devices may display downloaded weather data. These are different arrangements. Check the device and application documentation. A number on a phone screen does not by itself establish that pressure was measured at the phone’s location.

Why does a pressure-based altitude change while I stay still?

The atmospheric pressure can change because of weather, and the altitude calculation interprets pressure using a reference. If that reference is not updated appropriately, a stationary sensor can appear to move vertically. This is an interpretation limitation, not necessarily a fault in the pressure measurement itself.

Is a barometer useful without predicting the weather?

Yes. It teaches measurement, units, uncertainty, calibration and the distinction between direct observation and calculated output. Pressure measurements also support environmental monitoring and altitude estimation. Understanding what the instrument cannot establish makes its valid uses more dependable, not less interesting.

The bigger picture: an instrument measures less than a story claims

A barometer turns an invisible property into a readable signal. That achievement depends on a chain: air pressure produces a physical response, the response becomes a scale or electrical value, and a person or program interprets the value. Errors can enter at every stage.

Learning the chain is more valuable than memorising fair-weather numbers. You begin to distinguish a sensor from a forecast, resolution from accuracy, and altitude change from atmospheric change. The same habits improve the interpretation of thermometers, scales, speed sensors and scientific graphs.

The careful reader finishes with a better question: “What would have to remain unchanged for this comparison to be valid?” That question is the beginning of reliable measurement.

Useful routes and source notes

Continue with Weather: air pressure, clouds and forecasting for the wider atmosphere, or The drinking straw and atmospheric pressure for a familiar pressure-difference example.

For technical checks, use the linked NOAA explanation, Met Office instrument record, Vaisala pressure-measurement documentation and Bosch sensor descriptions. Manufacturer examples explain particular measurement principles; their specifications should not be assumed to describe every barometer. The numerical exercises in this guide are illustrative models, not observed weather data.