Range calculator

Definition

For a finite nonempty list of real numbers x₁, …, xₙ (n ≥ 1), the minimum is the smallest value on the list, the maximum is the largest, and the statistical range is R = max{xᵢ} − min{xᵢ}. When n = 1, min = max and R = 0. An empty list has no defined minimum or maximum; with an empty field this calculator shows a placeholder instead of inventing a number.

This model belongs to classical descriptive statistics: the range is a measure of scale (dispersion), alongside the standard deviation, variance, interquartile range (IQR), and mean absolute deviation. The NIST/SEMATECH e-Handbook of Statistical Methods lists the range under Measures of Scale. MathWorld (Statistical Range) and Wikipedia (Range (statistics)) give the same definition R = max − min for a finite sample.

On this page n can be arbitrarily large within one text field: the calculator builds a list from tokens separated by commas, spaces, or semicolons, drops invalid fragments, then computes min, max, and R in the browser’s floating-point arithmetic. The computation matches min / max on the parsed array.

Do not confuse statistical range with wave amplitude: for a pure sinusoid, peak-to-peak is often 2×amplitude, but this calculator does not assume a waveform model. It only evaluates max − min on the numbers you enter.

Historical and conceptual context

Sample extremes and the range were used in empirical sciences long before the arithmetic mean and standard deviation became standard tools. Nineteenth-century astronomy and surveying reported the span of observations as a quick quality check. In the twentieth century the range entered quality control: in Shewhart X̄–R charts, R is the subgroup statistic used to estimate process variability. NIST and SPC textbooks still describe R as a simple, readable measure of spread for short samples.

Conceptually the range is an extreme order statistic: it depends only on x₍₁₎ and x₍ₙ₎. That makes it insensitive to “middle” reshuffling of the list but highly sensitive to outliers — the opposite of the robustness of the median and the IQR. In exploratory data analysis (EDA), Tukey and later writers treat the range as a first look at scale, often just before a boxplot and the standard deviation.

This calculator does not compute σ, s, variance, IQR, or percentiles. It does not weight observations or reject extremes. For the arithmetic mean, standard deviation, and min/max with three fixed fields, use the sibling pages in the Mathematics → Statistics hub.

In teaching, R usually appears in the first descriptive-measures chapter, before the variance formula. Emphasize that R is not divided by n−1 and is not an unbiased estimator of a population scale parameter — it is a pure sample functional.

Formal properties

Let S be a finite nonempty subset of ℝ (or a multiset/list with repeats — repeats do not change the extremes). Then: (1) min(S) ≤ max(S); (2) R(S) ≥ 0; (3) R(S) = 0 iff all elements are equal; (4) min and max are idempotent in the finite-union sense: min(min S, min T) = min(S ∪ T).

Translation: for a constant c ∈ ℝ, min{xᵢ + c} = min{xᵢ} + c and R{xᵢ + c} = R{xᵢ}. Scaling: for λ > 0, R{λxᵢ} = λ R{xᵢ}; for λ < 0 the extremes swap roles but R{λxᵢ} = |λ| R{xᵢ}. For two numbers a, b one has R = |a − b|.

The midrange M = (min + max)/2 is an extra quantity in the result panel. M is not the arithmetic mean of the whole list when n > 2: for 2, 5, 8, 11 the mean is 6.5 and the midrange is also 6.5 only by coincidence; for 1, 2, 100 the mean is 34⅓ while the midrange is 50.5.

Units: R has the same unit as the observations. The calculator does not normalize to [0,1] and does not compute a coefficient of variation R/μ — those are separate models.

Full derivation

Step 1. Read the raw string from the “Numbers (list)” field. If it is empty, there is no sample — show the placeholder.

Step 2. Tokenize with MathCalcCommon.parseList: if the string contains spaces or semicolons, split on /[\s;]+/ and treat a comma inside a token as the decimal separator; if there are no spaces/semicolons and at least two commas, split on commas; if there is exactly one comma with no spaces (e.g. 1,9 on a locale that uses comma decimals), treat the whole string as one decimal. Drop tokens that fail parseNum.

Step 3. Let L = [y₁, …, yₙ] be the list of valid numbers, n ≥ 1. Set min ← y₁, max ← y₁, then for k = 2…n update min and max by pairwise comparisons. Equivalently: min and max are the smallest and largest elements of the whole list L.

Step 4. R ← max − min. Midrange ← (min + max)/2. A sorted copy of the list is only for the panel display — it is not part of the definition of R.

Step 5. Existence of min/max in a finite nonempty subset of ℝ follows from total order and induction on n.

Worked example: L = {2, 5, 8, 11}. min = 2, max = 11, R = 9. The panel also shows n = 4, midrange = 6.5, and the sorted list 2, 5, 8, 11.

Special cases

Singleton: the list “5” → R = 0. That is mathematically correct, not a defect of the calculator.

All equal: “4, 4, 4, 4” → R = 0, midrange = 4.

Negative and positive: “−2, 0, 4, 8” → min = −2, max = 8, R = 10.

Decimals with spaces: “1.5 3.5 2” → min = 1.5, max = 3.5, R = 2.

Pair: “1, 9” (comma with space) → R = 8. A bare “1.9” is one number → R = 0.

Junk tokens: “2, foo, 11” → parsed 2 and 11, R = 9, with a note about skipped tokens.

Very long lists: limited by browser memory and time; extremely long lists can exceed browser limits — educational lists of tens or hundreds of values are fine.

Numerical notes

The result is a double-precision floating-point number. For typical classroom and lab data, max − min is exact relative to the input representation. Near the floating-point limit, subtracting extremes may yield Infinity; the calculator does not emulate arbitrary-precision arithmetic.

The calculator accepts scientific notation (e.g. 1e3) and a decimal comma or dot, following the list-splitting rules above. It does not interpret thousands separators.

The number-line diagram marks list points, the range segment, and min/max labels. It illustrates the same definition of R, not a different model.

Compared with the standard deviation: for 2, 5, 8, 11 the range is 9, while population σ is smaller and depends on every point relative to the mean. R does not replace σ in inference.

Related concepts and sources

Related ideas: set minimum and maximum, order statistics x₍₁₎ and x₍ₙ₎, midrange, IQR = Q₃ − Q₁, standard deviation, variance, coefficient of variation, X̄–R control charts, peak-to-peak in signals (separate context).

Sources NIST/SEMATECH e-Handbook — Measures of Scale; Wikipedia — Range (statistics); Wikipedia — Maximum and minimum; MathWorld — Statistical Range. Introductory statistics textbooks repeat R = max − min in descriptive chapters.

Educational tool only. It does not replace full-sample analysis, plant-floor control charts, or SPC software.