PSLE Science Reality Lab Vol No.151 | “Accuracy ±1°C” — Is That the Uncertainty of This One Reading?

PSLE-SCI-REALITY-0151

Wait, What? A Number Printed on the Box Is Not Automatically the ± Number for Every Measurement

A digital thermometer advertises:

Accuracy: ±1°C

You place the probe in a beaker and the display reads 27.4°C.

A student writes in a report:

Temperature = 27.4 ± 1.0°C.

Is that automatically correct?

Not necessarily. The datasheet statement and the measurement-result uncertainty are related, but they are not automatically the same scientific object.

A manufacturer may use wording such as “accuracy ±1°C” to describe an allowed error, expected performance, tolerance-like specification or test condition over a stated range. A measurement made today also depends on the actual instrument, its calibration, resolution, environment, placement, response time, stability, method and other sources of uncertainty. The printed specification can be an important input to evaluating the measurement, but copying it directly beside one reading may pretend that all other uncertainty sources are already known and combined.

Reality Lab asks the more careful question: what exactly does the datasheet number describe, and what exactly would we need to know about this particular measurement result?

Quick Answer

A datasheet “accuracy” statement is usually a performance specification under stated conditions. Read the manufacturer’s definition and conditions.
In formal metrology, measurement uncertainty belongs to a measurement result and reflects the information used to characterise the dispersion of values reasonably attributable to the measured quantity.
A specification such as “±1°C” may constrain how large instrument error is expected or permitted to be under defined conditions, but it is not automatically the complete uncertainty of every reading.
Check the stated range, environmental conditions, calibration status, resolution, stability, probe placement, response time and method.
Do not assume that a display reading of 27.4°C means the true value is definitely between 26.4°C and 28.4°C unless the documentation defines the statement that way and the measurement evaluation supports that interpretation.
Do not erase the specification either. It is useful evidence about instrument performance; it simply has a narrower job than many readers give it.

The Exact Learner Job This Article Owns

This page owns one real-world evidence-transfer problem: how to read an instrument datasheet, product listing or package that gives an “accuracy”, permissible-error or performance specification and decide what that number does—and does not—tell you about one actual measurement result.

It does not replace the broader PSLE Science owners for choosing instruments, range, resolution, precision, accuracy, reference checks or repeated measurements. It also does not replace Reality Lab Vol No.057, which owns the case where a scientific result itself is explicitly reported with a plus-or-minus uncertainty. Vol No.151 starts one step earlier: the ± number is printed in the instrument specification, not in the result.

Original Reality Lab Case: The Classroom Temperature Probe

This is an original constructed case. The instrument and specifications are fictional.

A classroom owns a digital temperature probe with this datasheet:

Datasheet item	Fictional value
Measurement range	−10°C to 50°C
Display resolution	0.1°C
“Accuracy”	±1.0°C from 10°C to 40°C
“Accuracy” outside that band	±2.0°C
Specified operating conditions	Indoor use, probe fully immersed to marked line where applicable

A learner measures warm water. The display stabilises at 27.4°C.

Which of these statements is strongest?

A: “The temperature is exactly 27.4°C.”
B: “The temperature is definitely 27.4 ± 1.0°C.”
C: “The instrument reports 27.4°C. Its datasheet gives a ±1.0°C performance specification in this range, but the uncertainty of this measurement would also depend on how the probe was calibrated and used, its resolution and stability, the sample and method, and what the specification actually means.”

Statement C does the best scientific job. It neither worships the display nor throws away useful manufacturer information.

Five Different Numbers That Learners Often Collapse Into One

Number	Its job	Example
Displayed reading	What the instrument currently reports	27.4°C
Resolution	The smallest display increment or distinguishable change	0.1°C
Datasheet performance specification	A stated limit or expected performance under defined conditions	±1.0°C
Calibration correction or observed error	What comparison with references found for this instrument	For example, reading was 0.3°C high at a check point
Measurement uncertainty	A result-level evaluation of remaining dispersion/doubt using relevant information	Depends on the actual measurement model and uncertainty contributors

All five can matter. None should be silently renamed as another.

Why the Word “Accuracy” Needs Care

In everyday product language, “accuracy” is often followed by a numerical ± value. In formal measurement vocabulary, however, measurement accuracy is a qualitative idea about closeness of agreement between a measured value and a true quantity value; it is not itself a numerical quantity. Manufacturers may use “accuracy ±…” as convenient shorthand for a maximum error, tolerance-like limit or specified performance.

You do not need to correct every datasheet’s wording. You do need to read what the manufacturer means. The label “accuracy” alone is not enough. Look for:

the temperature or measurement range;
whether the value is a fixed amount, a percentage of reading, a percentage of full scale or a combination;
the environmental conditions;
warm-up time or response requirements;
calibration conditions;
probe, sensor or accessory assumptions;
time since calibration or adjustment;
whether the specification is typical, guaranteed, maximum or otherwise defined.

Maximum Permissible Error Is a Specification Boundary

The International Vocabulary of Metrology defines a maximum permissible measurement error as an extreme value of measurement error permitted by specifications or regulations for a given measurement, instrument or system.

That wording reveals the key learner distinction: an MPE is a limit on error permitted by a specification. It is not, simply by existing, a complete description of the probability distribution or uncertainty of a particular measurement result.

If an instrument is specified to stay within ±1°C under defined conditions, that can be useful evidence when estimating how wrong a reading might be. But a scientist still asks whether the actual measurement was made within those conditions and what other uncertainty sources contribute.

Measurement Uncertainty Belongs to the Measurement Result

Measurement uncertainty is not a sticker attached to the universe. It is evaluated for a measurement result using the information available about the measurement process.

For a temperature reading, contributors might include:

instrument calibration;
resolution;
repeatability and short-term noise;
sensor response time;
how deeply the probe was inserted;
temperature gradients within the sample;
heat exchange with surroundings;
self-heating or loading effects for some sensors;
drift since calibration;
the datasheet performance limit itself, depending on how the evaluation is constructed.

Not every school measurement needs a formal uncertainty budget. The learner job is conceptual: the whole measurement system can matter, not only the instrument’s catalogue line.

The Range Check: Which Part of the Datasheet Applies?

Our fictional probe specifies ±1.0°C only from 10°C to 40°C. At 47°C, a different specification applies.

If a website quotes only “accuracy ±1°C” without the range, it may make the instrument look stronger than the full datasheet supports. The missing range is part of the scientific claim.

This is a general pattern. Many instruments have specifications that vary with:

measurement range;
frequency;
temperature;
humidity;
signal level;
time after calibration;
sensor type or accessory;
operating mode.

A specification without its conditions is like an experiment result without its method.

The Resolution Check: 0.1°C Display Does Not Mean 0.1°C Accuracy

Our probe displays tenths of a degree. That tells us what the display can show, not how close each displayed value is to the quantity being measured.

An instrument can have:

fine resolution but poorer overall accuracy;
coarse resolution but excellent calibration within its scale;
stable repeated readings that are all shifted by a systematic error;
a good datasheet specification but poor measurement technique.

Extra decimal places should therefore never be treated as proof of extra truth.

The Calibration Check: A Specification Describes a Product Class; Calibration Describes an Actual Instrument

A manufacturer’s specification may apply to instruments of a model when they are manufactured, calibrated or maintained according to stated conditions. But the probe in your hand has its own history.

A calibration comparison can reveal how this specific instrument behaves at selected points. One probe might read 0.1°C high at a reference point; another of the same model might read 0.7°C low while both remain within a ±1°C specification.

That does not make the specification useless. It shows why class-level specifications and instrument-specific evidence are complementary.

The Method Check: A Good Instrument Can Still Be Used Badly

A thermometer can meet its specification and still produce a poor estimate of the quantity you care about if the method is wrong.

Suppose the learner measures “water temperature” but:

the water is warmer at the bottom than the top;
the probe touches a hot container wall;
the reading is recorded before the sensor stabilises;
only the probe tip is immersed when the instructions require deeper immersion;
the sample cools rapidly while being measured.

The instrument can be functioning perfectly while the measurement result fails to represent the intended quantity well.

The Representation Check: “±1” Looks Like an Interval Even When It Is Not One

A plus-or-minus symbol has a strong visual effect. It looks like it is telling you, “The answer lives inside this interval.” Sometimes that is approximately how a report intends the notation to be read. Sometimes it is not.

For a datasheet, the ± number might be:

a maximum permissible error;
a guaranteed specification under stated conditions;
a typical performance figure;
a percentage of reading;
a percentage of full scale;
a combined form such as ±(percentage of reading + fixed digits);
something else defined in the manufacturer’s notes.

Always read the definition before converting the printed line into a measurement interval.

The Comparison Check: Two Thermometers Both Say ±1°C

Thermometer A and Thermometer B both advertise “accuracy ±1°C”. Does that prove they are scientifically equivalent?

No. Their specifications may apply over different ranges, temperatures or conditions. Their response times, resolution, stability and calibration histories can differ. One may be intended for air temperature and the other for immersion. The same short catalogue number does not erase the rest of the measurement system.

Worked Case 1: The Aquarium Probe

A fictional aquarium probe reads 24.7°C and advertises ±0.5°C accuracy between 15°C and 30°C. The probe has not been checked for three years and its protective sheath is damaged.

Weak reasoning: “24.7 ±0.5°C, because the box says so.”

Better reasoning: The datasheet gives useful model-level performance information in this range, but the condition and calibration history of the actual probe create additional questions. A current reference check would strengthen confidence in the present result.

Worked Case 2: The Balance With ±0.1 g Specification

A balance is specified as ±0.1 g under certain conditions and displays 50.00 g. The student concludes that the mass is known to the nearest 0.01 g because the screen shows two decimal places.

The display resolution is finer than the quoted performance specification. The last displayed digit may be useful for observing small changes or repeatability, but it does not prove the absolute mass is correct to 0.01 g.

Worked Case 3: The Light Sensor Outside Its Range

A light sensor advertises ±3% performance from 100 to 10,000 units. A student uses it at 20 units and still applies ±3%.

The quoted specification does not cover the measurement region. The learner should not extend the specification beyond the range without evidence. This is the instrument version of extrapolating beyond tested data.

Worked Case 4: The Percentage-of-Reading Trap

An instrument states “±2% of reading”. One learner treats that as ±2 units for every measurement.

At a reading of 50 units, 2% is 1 unit. At 500 units, 2% is 10 units. The same written percentage produces different absolute limits. The learner must read both the number and its mathematical basis.

Worked Case 5: “±1 Digit” Is Not “±1 Unit”

A datasheet contains a formula such as “±(1% of reading + 2 digits)”. The “digits” term refers to counts in the least significant displayed digit, not two whole measurement units.

The lesson is not to memorise instrument formulas. It is to recognise that specifications may be structured expressions. Copying only the large visible “±” value can change the meaning.

Worked Case 6: Same Reading, Different Method Uncertainty

Two groups use the same model thermometer and both obtain 30.0°C. Group A mixes the water and waits for a stable reading. Group B places the probe against a heated wall and records immediately.

The datasheet specification is identical for both instruments, but the measurement quality is not. Method effects can dominate the question of what temperature was actually being represented.

Worked Case 7: Better Than the Specification?

A thermometer with a ±1°C datasheet specification is compared with a trustworthy reference at several nearby points and repeatedly agrees within 0.1°C. Can we say every future reading is now ±0.1°C?

No. The checks provide encouraging evidence for those points, conditions and times. A broader uncertainty claim needs an appropriate evaluation of calibration, interpolation, drift and use conditions. Good evidence should narrow uncertainty carefully, not magically erase it.

What Evidence Would Strengthen a Measurement Based on a Datasheet Specification?

The exact instrument model and sensor are identified.
The measurement lies within the stated range.
The environmental and operating conditions match the specification.
The specification is defined clearly: maximum, typical, percentage of reading, fixed term, or another form.
The instrument has an appropriate calibration or reference-check history.
The measurement method avoids obvious placement, timing or sampling errors.
Repeatability is checked when the quantity should be stable.
Relevant resolution and drift are considered.
The result-level uncertainty statement, if needed, explains how contributors were combined rather than copying one catalogue line without justification.

What Would Weaken the Claim?

The quoted ± value is used outside its stated range.
The product listing removes important conditions from the specification.
Display resolution is mistaken for measurement accuracy.
A damaged or drifting instrument is assumed to perform exactly like a new one.
A class-level specification is presented as an instrument-specific calibration result.
The method introduces large sample or placement variation.
The printed ± number is copied directly into a result without checking what it represents.
Different instruments’ specifications are compared even though the manufacturers define them differently.

Tempting Reasoning That Fails

“The box says ±1, so every result is automatically value ±1.” The datasheet number’s meaning and conditions must be checked first.
“±1 means the instrument will always be wrong by exactly 1.” A limit or specification is not a prediction of the exact error on each measurement.
“The display has 0.1 resolution, so its accuracy is 0.1.” Resolution and accuracy are different.
“A specification is useless because it is not a full uncertainty budget.” Too cynical. Specifications are important performance evidence; they simply do not answer every result-level question by themselves.
“Calibration means the datasheet no longer matters.” Calibration and specification can provide different, complementary information.
“One reference check proves the instrument is equally good across its whole range.” A check at one point has limited reach.
“If two instruments advertise the same ± value, they are identical.” Range, conditions, response, resolution and use can differ.

How Far Can the Conclusion Travel?

If the manufacturer states ±1°C performance over a specified range and the instrument is being used within those stated conditions, we may reasonably use that specification as evidence about expected or permitted instrument error according to the manufacturer’s definition.

That does not automatically establish:

the exact error of the current reading;
the complete measurement uncertainty;
the true value lying inside a simple symmetric interval with guaranteed certainty;
correct performance outside the specified range;
correct probe placement or sample representativeness;
zero drift since calibration;
equivalence to another instrument with a similarly worded specification.

PSLE-Style Transfer Case

A student measures water temperature with a digital probe. The probe displays 31.6°C. Its specification states: “Resolution 0.1°C; performance limit ±0.8°C from 0°C to 40°C under stated laboratory conditions.”

The student writes: “The water temperature is exactly 31.6°C because the thermometer is digital.” Explain one problem.

Answer: The display resolution only shows the instrument reports to 0.1°C. The datasheet itself indicates that the instrument’s measurement error can be larger than the last displayed digit under the specified conditions, and the actual measurement also depends on calibration and method. The reading should therefore not be treated as exact.

A second student writes: “The water temperature is definitely 31.6 ± 0.8°C.” What should be checked before accepting that statement?

Answer: Check what ±0.8°C means in the datasheet, whether the conditions and range apply, the instrument’s calibration/status, and whether other method and measurement effects need to be included in the result-level uncertainty.

Explained Practice

Practice A: A balance reads 100.00 g and has 0.01 g resolution but ±0.10 g specified performance. Which number tells you the screen increment? 0.01 g. Which gives broader evidence about specified measurement error? ±0.10 g, under its stated conditions.

Practice B: A sensor says ±2% from 20 to 80 units. At 10 units, can you confidently use ±2%? No. That would extend the specification outside its declared range.

Practice C: Two identical probes disagree by 1.5°C. Both advertise ±1°C. What should happen next? Investigate calibration, placement, conditions, stability and method rather than automatically averaging them.

Practice D: A probe is specified at ±1°C when new, but a current reference check shows a 2.5°C error. Which evidence matters for today? The current check is direct evidence that the present instrument may no longer meet the old performance expectation.

Practice E: A retailer removes “at 20–25°C ambient” from a manufacturer’s accuracy line. What was lost? A condition that defines when the specification applies.

Delayed Independent Return: The S-C-A-L-E Check

S — Specification: What exactly does the ± number mean?
C — Conditions: Over what range and environment does it apply?
A — Actual instrument: What do calibration, condition and history say about this unit?
L — Local method: Could placement, timing, sampling or technique alter the result?
E — Evidence statement: What uncertainty claim is actually justified for this measurement?

The mnemonic is not an exam template. It is a memory aid for real-world reading. Use the reasoning even when you forget the letters.

Parent and Tutor Teaching Guide

Put three cards on the table: DISPLAY, DATASHEET, MEASUREMENT. Write “27.4°C” under DISPLAY, “±1°C from 10–40°C” under DATASHEET, and leave MEASUREMENT partly blank. Ask the learner what other facts are needed before writing a careful result.

Then create an instrument with a very fine display but a broad performance specification—for example 12.347 units on screen with ±0.8 units specified performance. Ask which digits should make the learner cautious. The point is not to ban extra digits; it is to teach that digits and evidence strength are different things.

Finally, give two measurement situations using the same instrument: one with careful stabilisation and one with rushed placement against a hot wall. Ask why the datasheet is identical but the measurement evidence differs. That question moves the learner from “instrument worship” to whole-system reasoning.

Authoritative Sources

Formal metrology distinguishes performance specifications, measurement error and measurement uncertainty for a reason. A specification can set or describe allowed instrument behaviour, while an uncertainty statement belongs to the measurement result and the information used to evaluate it. For a Primary learner, the transferable habit is simple: read the label, read the conditions, then inspect the actual measurement.

The Quiet Return

The thermometer says 27.4°C.

The datasheet says ±1°C under stated conditions.

Those two facts belong in the same investigation, but they are not the same fact.

Do not turn a catalogue specification into a certainty statement. Use it as one piece of evidence in the measurement story.

Small Group Tutorials