Wait, What? 9.80643721 m s⁻² can be a worse scientific answer than 9.8 m s⁻².

If the experiment only resolves the result to about ±0.2 m s⁻², the extra digits do not contain extra knowledge. They are arithmetic residue from the calculator. Scientific reporting should preserve information, not manufacture precision.

Significant figures are an evidence convention

Significant figures communicate the meaningful precision of a number. They are not a ritual where every answer is automatically three significant figures. The appropriate reporting depends on the measurement uncertainty, instrument resolution and calculation context.

Decimal places and significant figures are different

12.30 has four significant figures and two decimal places. 0.00450 has three significant figures and five decimal places. Decimal places locate digits relative to the decimal point; significant figures count meaningful digits from the first non-zero digit.

Match the result to its uncertainty

If a result is 12.3478 ± 0.6 cm, reporting 12.3478 cm is misleading. A sensible form is approximately 12.3 ± 0.6 cm, with the central value rounded to the same decimal place as the uncertainty.

Depending on convention, uncertainty is often reported with one significant figure, or two when the first digit warrants it. The key principle is alignment between uncertainty resolution and central-value resolution.

Do not round too early

Keep extra guard digits during intermediate calculations, then round the final reported result. Repeated early rounding can accumulate numerical error.

This creates an important distinction: calculation precision can be temporarily higher than reporting precision.

Instrument display digits are not automatically significant

A digital balance displaying 12.345 g suggests a certain resolution, but calibration, drift, vibration or environmental effects may make the last digit unreliable. Conversely, an analogue scale may allow interpolation between marks under a defined reading method.

Count evidence, not screen characters.

Exact numbers are different

Counting 20 oscillations is an exact count if none were missed; the number 20 does not limit significant figures in the same way as a measured time of 15.8 s. Defined conversion factors can also be exact by convention.

Do not apply measurement significant-figure rules mechanically to exact counts and definitions.

Quantitative window

A cylinder diameter is measured as 2.40 ± 0.05 cm. A calculator gives cross-sectional area:

A = π(1.20)² = 4.523893421… cm²

The long calculator output is not the experimental result. Radius uncertainty is about 0.025/1.20 ≈ 2.1%, so area uncertainty is roughly 4.2%. The area is therefore about 4.52 cm² with uncertainty around 0.19 cm², making a report such as 4.5 ± 0.2 cm² defensible under the chosen convention.

False precision can hide weak apparatus

Writing many digits makes a result look authoritative. This is dangerous because readers may mistake numerical detail for experimental quality. A stopwatch reaction-time experiment does not become high precision because a spreadsheet prints six decimal places.

Too little precision also throws away information

If a calibrated instrument reliably distinguishes 1.24 V from 1.25 V, reporting both as 1 V destroys useful evidence. Good reporting is neither maximal digits nor minimal digits; it preserves the resolution the experiment actually supports.

Tables should preserve consistent resolution

Measurements made with the same instrument and method should usually be recorded to consistent decimal places. Writing 2.1, 2.13 and 2.134 in one column can falsely imply changing measurement resolution.

Graph axes need enough precision to reconstruct the data

Do not round transformed graph variables so aggressively that distinct measurements collapse onto the same plotted value. Keep sufficient working precision for plotting and regression, then report fitted parameters according to their uncertainty.

Observation versus inference

Observation: “The timer displayed 15.8 s for 20 oscillations.”

Transformation: “Mean period = 0.790 s using guard digits.”

Inference/report: “The period is about 0.79 s at the resolution supported by the timing method.”

The calculator’s third decimal place can be useful internally without deserving equal status in the final claim.

Failure modes

• Automatically using three significant figures for every result.
• Copying every calculator digit.
• Rounding intermediate steps too aggressively.
• Reporting central value and uncertainty to mismatched decimal places.
• Treating exact counts as uncertain measured quantities.
• Assuming digital resolution equals total measurement accuracy.
• Rounding graph data before fitting until information is lost.

Unfamiliar transfer: a regression gradient

Software reports a gradient of 2.37491862 ± 0.083715. The scientifically useful report is not the raw output. A form such as 2.37 ± 0.08, with units, communicates the fitted information much more honestly.

Secondary → JC → deeper Science

Secondary: record consistent instrument resolution and avoid meaningless calculator digits.

JC: align reported precision with uncertainty, preserve guard digits and report fitted quantities responsibly.

Deeper Science: extend to uncertainty intervals, metrological traceability, numerical conditioning and reproducible computational reporting.

Checkpoint

A calculation gives 6.2831853 N, but propagated uncertainty is ±0.4 N. Which is the better report: 6.2831853 ± 0.4 N or 6.3 ± 0.4 N?

Answer key and WHY reasoning

6.3 ± 0.4 N. The uncertainty resolves the tenths place, so digits beyond that imply precision unsupported by the experiment.

How to study this

For every final answer ask: what measurement limits the last trustworthy digit? Keep guard digits while calculating, then let the uncertainty decide the final reporting precision.

Evidence boundaries

Significant figures communicate numerical resolution; they do not replace an uncertainty analysis. Two results with the same number of significant figures can have very different reliability, bias and evidential strength.

Authoritative next steps

• NIST measurement uncertainty guidance
• BIPM/JCGM measurement publications
• Institute of Physics practical resources

Teaching Guide

Give students three calculator outputs from experiments with different uncertainties. Ban the phrase “three significant figures” until they first justify the last trustworthy digit from the measurement. This reconnects reporting to evidence.