Naar de inhoud
OneKitly

The Range of Five Parts Divided by 2.326 Is a Standard Deviation

Gepubliceerd op 17-11-2026 · 3 min leestijd · Zakelijke tools

Marco Bianchi

Marco Bianchi — Redacteur wonen, klussen en auto bij OneKitly

Verbouwen · Materialen

Getoetst aan 2 bronnen

Profiel bekijken →
Kort gezegd

Before calculators reached the shop floor, quality control needed a measure of spread that a machinist could work out between parts. The range — largest minus smallest — is that measure, and dividing it by a constant that depends only on the sample size converts it into an estimate of the standard deviation. For samples of five the constant is 2.326, so a batch measuring 9.8, 10.0, 10.1, 10.2 and 10.4 has a range of 0.6 and an estimated standard deviation of 0.258 against a true 0.224. One subtraction and one division replace squaring five numbers, and the answer is close enough to act on.

Advertentie

Five measurements spanning 0.6 mm give an estimated standard deviation of 0.258 with one division. The real value is 0.224, and the shop floor never needed a calculator.

Why the divisor grows with the sample

The more parts you measure, the further apart the extremes are likely to fall, even though the underlying process has not changed. Two measurements straddle about one standard deviation on average; five straddle about 2.33; ten straddle over three. The constant simply undoes that growth, and it is why a range from a sample of five cannot be compared with a range from a sample of ten without correction. Charts fix the sample size for exactly this reason: keep it constant and the ranges are comparable day after day with no arithmetic at all.

What a control chart actually watches

Two things at once: whether the average has drifted, and whether the spread has widened. A tool wearing down moves the average without touching the range; a loose fixture widens the range without moving the average. Watching only the mean misses half the failures, which is why the range chart sits underneath the average chart on the same sheet and is read first. A process can be perfectly centred and completely out of control at the same time.

Divisor
The divisor by sample size
Sample sizeDivisorUse
21.128Paired checks
42.059Short runs
52.326The classic chart

Uitgewerkt met onze eigen calculator

Calculator spreidingsbreedte

Gegeven

Getallen (met komma's gescheiden)
4, 8, 17, 18, 28, 53

Resultaat

Spreidingsbreedte
49
Minimum
4
Maximum
53

Deze cijfers komen uit de calculator hieronder, ze zijn niet met de hand ingetypt — ze worden opnieuw berekend telkens als de tool verandert.

Reken het door met je eigen cijfers →
Advertentie

Veelgestelde vragen

Is the estimate always this rough?
It is unbiased on average and noisy on any single sample, which is a different thing from rough. Over many samples of five the range divided by 2.326 lands on the true standard deviation; on one sample it can be out by a fifth in either direction, as the example here shows. Charts get around this by pooling twenty or twenty-five samples before setting limits, which is why a control chart cannot be started on the first morning of production.
Why not just use the standard deviation now?
Many plants do, and the charts are named accordingly. The range version survives because it is readable: an operator can see that today's spread is twice yesterday's without trusting a computed number, and a chart people understand gets filled in. There is also a technical argument — the range reacts faster to a sudden widening in small samples — but the honest reason is usually that the method has to be operated by the person at the machine, and simplicity is a feature of the system rather than a limitation of the maths.

Artikelen die je misschien interesseren

Alle gidsen →

Vergelijkbare tools

Bronnen

Een fout in dit artikel gezien?