An analytical balance will happily display a number to five decimal places regardless of whether the last two of them describe your sample. That is the central difficulty with precision weighing: the instrument gives no outward sign when it is wrong. A thermometer reading nonsense usually looks like nonsense. A balance reading nonsense looks exactly like a balance reading the truth.
Most of what corrupts sub-milligram weighing is environmental rather than mechanical. The balance is not broken. It is faithfully reporting the force it feels, and that force includes contributions from air currents pushing on the pan, charge on the sample attracting the shield, a convection cell rising from a vessel that is warmer than the room, and the buoyancy of displaced air. None of these appear as an error message.
These effects are finite in number, well understood, and controllable by a routine that costs almost no extra time once it becomes habit. What they are not is visible. Nobody notices their absence until a result refuses to reproduce and the search begins in the wrong place, usually somewhere downstream in the chemistry.
Key takeaways
- A modern balance does not weigh mass directly; it measures the electrical current needed to hold the pan in one position, which makes it sensitive to any force at all.
- Draughts, vibration and temperature gradients act on the pan exactly as extra mass would, and none of them announce themselves.
- Static charge is the largest single source of unexplained instability in dry powder work and low-humidity rooms.
- Air buoyancy matters for low-density materials and is the reason weighing standards distinguish conventional mass from true mass.
- Internal calibration corrects sensitivity drift, not linearity or eccentricity, so external weights still have a role.
How an Analytical Balance Senses Mass
Nothing on a modern analytical balance is a spring, and nothing is a set of counterweights on a beam. The mechanism is electromagnetic force restoration, and understanding it explains almost every quirk that follows.
The pan sits on an arm attached to a coil suspended in a permanent magnetic field. A position sensor watches the arm. When you place something on the pan, the arm begins to move down, the sensor detects the displacement, and a control circuit increases the current through the coil until the electromagnetic force pushes the arm back to exactly its original position. The pan never really moves. What the balance reports is the current required to hold everything still, converted into a mass reading by a calibration factor.
This gives extraordinary resolution, because the sensor only ever has to detect a null position rather than measure displacement across a range. It also means the instrument is literally a force meter, unable to distinguish the downward force of your sample’s weight from any other force applied to the pan assembly.
That distinction is the whole subject. A finger of air across the pan pushes on it. A charged weighing boat attracts the earthed metal of the shield, and the pan feels that attraction. A vessel warmer than the surrounding air sets up a convection current that lifts the pan. The balance adjusts its current accordingly and prints a number that is honest about force and wrong about mass.
Draught, Vibration and Bench Choice
The draught shield exists because air movement is the fastest way to make a five-decimal balance useless. The forces involved are tiny in absolute terms and enormous relative to the resolution being claimed.
Air currents in a laboratory come from more sources than people expect. Fume hood extraction changes the pressure in the room whenever a sash moves. Air conditioning cycles. Doors open. A person walking past creates a wake, and the largest offender in many labs is the operator, standing close, radiating body heat and breathing. Closing the shield doors solves most of this, provided the doors are closed fully and the air inside is given a moment to settle. Opening a shield door introduces a slug of room air at a slightly different temperature, and that air needs time to equilibrate before the reading means anything. Waiting for the stability indicator is not optional decoration.
Vibration acts differently with the same result. The control loop is trying to hold a null position, and a bench that oscillates continuously means the loop never settles. Balances damp this electronically, but damping trades against speed and cannot remove low-frequency building sway. Bench choice therefore matters more than most laboratories treat it. The ideal is a heavy stone or cast slab decoupled from the building structure, on damping supports, in a corner away from doors and traffic. A balance sharing a bench with a centrifuge is being asked to do something impossible.
| Disturbance | How it reaches the pan | Typical sign on the display | Practical control |
|---|---|---|---|
| Air current | Pressure on pan and vessel walls | Reading wanders, will not stabilise | Close shield fully, wait, keep away from hoods and doors |
| Vibration | Mechanical coupling through the bench | Last digits flicker constantly | Stone bench, damping feet, isolate from rotating equipment |
| Static charge | Electrostatic attraction to shield and housing | Slow one-directional drift, or a reading that shifts when the door moves | Ionise, raise humidity, use metal or coated vessels |
| Thermal gradient | Convection around a warm vessel | Steady drift in one direction that gradually slows | Equilibrate the sample in the balance room before weighing |
| Buoyancy | Displaced air volume | No visible sign at all | Correct arithmetically for low-density materials |
Note the fourth column of that table more than the others. Only the first two disturbances look like disturbances. The rest produce clean, stable, confident, wrong readings.
Static Electricity and Charged Samples
Static is the effect that most often produces the phrase “the balance is broken” when the balance is nothing of the kind. Charge is generated wherever dissimilar materials rub and separate. Tipping a dry powder into a plastic weighing boat does exactly that, thousands of times over, as does wiping a vial with a dry tissue, pulling a plastic bag from a box, or walking across a synthetic floor. The charge has nowhere to go, because the materials involved are insulators by design.
The charged vessel then sits inside a metal-framed, earthed enclosure. Electrostatic attraction develops between the charge and the earthed surfaces, and the pan experiences that attraction as an apparent change in weight. Because charge leaks away slowly through the air, the apparent weight drifts. The classic signature is a reading that creeps in one direction over tens of seconds and then slows, or a reading that changes noticeably depending on whether the shield door is open or closed, or where in the pan the vessel is placed.
Humidity is the dominant variable. Moist air conducts charge away in seconds; dry air lets it persist for minutes. This is why static problems appear suddenly in winter in heated buildings, and why a laboratory that has never had an issue develops one after the ventilation is upgraded. Relative humidity comfortably above forty per cent makes most static problems vanish on their own; humidity in the low twenties makes them near inevitable.
The controls are ordinary. An ioniser floods the weighing area with both positive and negative ions, neutralising whatever charge is present within seconds. Metal or antistatic-coated vessels conduct charge away rather than holding it. An antistatic mat under the operator helps, and where none of that is available, placing the charged vessel inside a small earthed metal container shields the pan from the field. Static is also why transferring dry powder while the vessel sits on the pan is poor practice: each transfer regenerates charge exactly where it does most damage.
Temperature Equilibration Before Weighing
A vessel that is warmer than the air in the balance chamber heats the air immediately around it. That air rises. Rising air exerts an upward drag on the vessel and the pan, and the balance reads low. A vessel colder than the chamber does the reverse and reads high.
The magnitude is not trivial. A crucible fresh from a drying oven, or a vial taken out of a fridge, can produce an apparent difference of several tenths of a milligram or more, drifting steadily as the temperature difference decays. On a four-decimal balance this is the difference between a defensible result and a fictional one.
The remedy is patience, which is the least popular solution in any laboratory. Items removed from an oven should cool in a desiccator until they reach room temperature, and the desiccator itself should be in the balance room, not the oven room. Refrigerated samples need to sit out until condensation has stopped forming and the surface is dry, which introduces a second problem: condensed moisture is real mass, and it weighs.
Handling matters for the same reason. Picking up a crucible with bare fingers deposits both heat and skin oils, one of which drifts away and one of which does not. Forceps or lint-free gloves are standard practice for the mundane reason that a fingerprint has weight. The instrument itself needs a warm-up too: the electronics and mechanical assembly reach thermal equilibrium over hours rather than minutes, which is why manufacturers specify leaving a balance permanently powered, and why standby mode exists.
Buoyancy Effects on Low Density Materials
Everything weighed in air is subject to Archimedes’ principle. The object displaces a volume of air, and that displaced air exerts an upward force equal to its own weight. The balance therefore reads slightly less than the object’s true mass, and how much less depends on the object’s volume, not its mass.
The consequence surprises people: two objects of identical true mass but different densities do not read the same. A dense metal object displaces very little air, while a bulky low-density powder or a large plastic component displaces considerably more and reads light. For routine work the effect is small enough to ignore, because calibration weights and samples have broadly similar densities and the errors partially cancel. It stops being ignorable for genuinely light materials: organic powders, foams, large plastic articles, anything weighed as a volume rather than a lump.
Metrology handles this with the concept of conventional mass: the mass an object appears to have when weighed in air of a standard density against reference weights of a standard density. Calibration weights carry conventional mass values, which is why routine weighing works without anybody thinking about air. When a correction genuinely matters it is computed from the density of the room air, the sample material and the reference weights, and air density itself varies with pressure, temperature and humidity. For everyday work the advice is narrower: expect the effect to matter for low-density materials, and be suspicious when two correctly calibrated balances disagree about a bulky sample.
Internal Calibration Versus External Weights
Most modern analytical balances contain a motorised internal weight that the instrument can place on the mechanism at the touch of a button, or automatically when it detects a temperature change or a set time interval has passed. This is convenient and genuinely useful, and it is also routinely misunderstood.
Internal calibration adjusts one thing: sensitivity, meaning the slope of the relationship between measured force and displayed mass. Sensitivity is the parameter that drifts most with temperature, so correcting it frequently is worthwhile. But a balance can have perfect sensitivity and still be wrong in several other ways.
Linearity describes whether the balance is accurate across its whole range or only near the calibration point. A single-point internal adjustment cannot detect a bow in the response curve. Eccentricity, sometimes called corner load error, describes how the reading changes when the object sits off-centre on the pan. Repeatability describes the spread of results when the same object is weighed many times. Internal calibration says nothing about any of these.
External weights address them. A set of certified weights, handled only with forceps and stored in a fitted case, lets a technician check the reading at several points across the range, place a weight at the centre and at each quadrant of the pan, and repeat one weighing many times to estimate scatter.
Certified weights are themselves consumables: they need periodic recalibration by an accredited body, and they degrade if handled badly. A weight picked up with bare fingers has gained mass. A weight that has been dropped may have lost some. A sensible schedule runs the internal adjustment daily and on any temperature change, verifies against a single external weight before critical work, and checks linearity, eccentricity and repeatability periodically or after any move.
A Weighing Routine That Avoids Drift
None of the individual controls described here take long. Assembled into a fixed sequence, they take almost no time at all, and they remove the majority of unexplained variation from small-quantity work.
Begin before the sample arrives. Leave the balance powered continuously. Confirm the levelling bubble is centred, since an unlevel balance measures a component of gravity rather than all of it, and levelling drifts whenever an instrument is moved or a bench is knocked. Run the internal adjustment and verify with an external check weight, recording the result where a trend would show.
Bring the sample and its vessel into the balance room and leave them to reach room temperature, in a desiccator if they have come from an oven. Handle vessels with forceps or gloves throughout. If the room is dry or the material is a fine powder, run the ioniser over the vessel before it goes near the pan.
Close the shield, allow the reading to settle, and tare. Then open the door as briefly as possible, place the vessel centrally on the pan, close the door, and wait for the stability indicator rather than for the number to merely look calm. Where the balance offers a stability filter setting matched to the environment, use the setting appropriate to the room rather than the fastest one available.
Record the result immediately, and prefer weighing by difference for anything critical, since taking one reading from another cancels any constant offset that persists across both, including a stubborn static contribution.
Frequently asked questions
Why does my balance give a different reading when I move the vessel on the pan?
That is eccentricity, or corner load error, and it is a property of the mechanism rather than a fault. The force restoration assembly is designed to respond identically wherever the load sits, but no mechanical linkage achieves that perfectly, and the deviation grows with distance from the centre and with the size of the load. Small deviations are normal and are one of the parameters checked with external weights during periodic servicing. The practical response is simply to place every load in the middle of the pan every time, so that whatever residual error exists is constant and cancels when you weigh by difference.
How long should I really wait for a reading to stabilise?
Longer than the display suggests and shorter than folklore claims. The stability indicator is the instrument’s own judgement that consecutive readings agree within a defined tolerance for a defined period, and it is reliable for ordinary loads in a settled environment. It becomes unreliable in exactly the cases discussed here, because a slow, smooth thermal or static drift can satisfy the stability criterion while the number continues to move. If the reading is still changing in a consistent direction after the indicator appears, the indicator is describing the smoothness of the drift, not its absence.
Does the balance need recalibrating if I move it to a different bench?
Yes, and the reason is more fundamental than settling. Local gravitational acceleration varies with latitude and altitude, so the factor converting restoring current into displayed mass is only correct for the place where it was set. Moving between floors of a building is enough to matter at analytical resolution. Beyond gravity, the levelling will have changed, the mechanism needs time to reach thermal equilibrium in its new position, and the new bench may have entirely different vibration characteristics. Level it, leave it powered, let it settle, then adjust and verify.
Is an ioniser worth buying, or is it a luxury?
It depends almost entirely on what you weigh and how dry your building is. If the work is aqueous samples in glass vials in a humid climate, an ioniser will spend its life idle. If it is fine dry powders, filter membranes, plastic weighing boats, or anything handled in a heated building in winter, it removes a category of error that is otherwise very difficult to control and easy to misattribute. The alternatives are cheaper but less convenient: conductive vessels, an antistatic mat, and enough humidity in the room.
Why do two calibrated balances disagree about the same sample?
First check the obvious: both level, both recently adjusted, both verified against traceable weights, and the sample weighed at the same temperature in both rooms. If all of that holds and the disagreement persists, buoyancy is a strong candidate, particularly when the material is bulky and light, because the two instruments may have been adjusted against weights of different density or in air of different density. Genuine differences in linearity between two instruments at the particular load in question are also possible, which is why a linearity check across the working range is more informative than a single-point verification.
Finally, treat anomalies as information. A reading that drifts steadily in one direction is a temperature or static story. A reading that flickers without settling is a vibration or draught story. A reading that is stable but disagrees with an independent measure is a calibration or buoyancy story. The instrument rarely lies about force. It simply has no way of telling you that the force it feels is not the one you meant to measure.
