Example 1
- a = 1
- b = 2
- c = 3
0.816497
What is σ with 1, 2, and 3? ≈ 0.816497.
Enter three numbers in fields a, b, and c. The result is the population standard deviation σ with divisor 3: the square root of the mean of squared deviations from the mean. With 1, 2, and 3 the value is about 0.816497. With 2, 2, and 2 it is 0. The form has exactly three fields and does not use n−1 as the primary result.
Enter three numbers in fields a, b, and c.
Exactly three fields. The variance divisor is 3 (population), not 2 and not n−1. The two-number neighbor divides by 2.
Standard deviation of three numbers is the square root of population variance. Mean m = (a+b+c)/3, sum of squared deviations from m divided by 3, then the root. For 1, 2, 3 the mean is 2, variance is 2/3, and σ ≈ 0.816497, because √(2/3) is not an integer.
The denominator is 3, not 2 and not n−1. The neighboring two-number page divides by 2. Here there are always three fields and divisor 3 — the population convention matching the formula on this page.
The fields are a, b, and c. For 2, 2, 2 the result is 0, because nothing stands off the mean. For 0, 6, 12 the result is 4.898979. An integer result goes as a plain string; a fraction uses six decimal places.
The result panel also shows the mean, variance, SSD, and comparative s with divisor 2. The primary result remains population σ.
1, 2, 3 give 0.816497. 2, 2, 2 give 0. 0, 6, 12 give 4.898979. Three numbers, divide by 3.
m = (a + b + c) / 3, σ² = ((a−m)² + (b−m)² + (c−m)²) / 3, σ = √σ².
The divisor is n = 3 (population). A sample estimator would use n − 1 = 2; this page does not do that for the primary result.
Mathematical model: population standard deviation of a finite triple (σ, divisor n) — Karl Pearson / descriptive statistics; NIST measures of scale; Excel STDEV.P.
Standard deviation is the square root of variance. For a finite set of three numbers a, b, c the arithmetic mean is m = (a + b + c) / 3, the sum of squared deviations is SSD = (a−m)² + (b−m)² + (c−m)², the population variance is σ² = SSD / 3, and σ = √σ². This calculator implements exactly that chain: the divisor is always n = 3, never n − 1.
The population-versus-sample distinction matters. The symbol σ with divisor n describes the spread of the whole population (here: the three entered values treated as the complete set). The symbol s with divisor n − 1 (Bessel’s correction) is an unbiased estimator of an unknown population variance from a sample. This page returns population σ; the result panel also shows s with divisor 2 for comparison, but the primary result is always σ with /3.
The result unit matches the data unit: if a, b, c are in meters, σ is in meters too. Variance has a squared unit; the square root restores the original scale.
Karl Pearson popularized the term “standard deviation” around the turn of the twentieth century; earlier least-squares work by Gauss and Legendre already used sums of squared deviations from the mean. Variance as the second central moment (μ₂) and deviation as √μ₂ entered the canon of descriptive statistics and inference. The NIST/SEMATECH e-Handbook presents measures of scale, including standard deviation, as basic descriptors of spread.
Conceptually σ measures a typical distance from the mean on the data scale, unlike the range (max − min), which looks only at extremes. For three numbers the range is simple, but σ uses all three squared deviations. The neighboring two-number standard deviation page divides by 2 (also population /n); the variance-of-n-numbers page leaves σ² from a list — here there are always three fields and the square root at once.
When teaching the topic, name honestly what the software computes: spreadsheets and libraries differ by divisor (Excel STDEV.P versus STDEV.S, R sd() with /(n−1), NumPy with ddof). This page matches STDEV.P / population with n = 3.
For any a, b, c ∈ ℝ: σ ≥ 0. σ = 0 if and only if a = b = c. Shift: σ(a+c₀, b+c₀, c+c₀) = σ(a, b, c). Scaling: σ(λa, λb, λc) = |λ| σ(a, b, c). Field order does not change the result because the formula is symmetric under permutation.
Identity: SSD = a² + b² + c² − 3m² for m = (a+b+c)/3, so σ² = (a²+b²+c²)/3 − m². The calculator still computes deviations from m directly, by the formula: v = ((a−m)²+(b−m)²+(c−m)²)/3.
Relation to the sample estimator: s² = SSD / (n−1) = (3/2) σ² when n = 3, so s = σ √(3/2). For 1, 2, 3: σ ≈ 0.816497 and s ≈ 1.
Step 1. Enter three real numbers a, b, c (period or comma in a decimal).
Step 2. Compute the mean m = (a + b + c) / 3.
Step 3. Compute deviations dₐ = a − m, dᵦ = b − m, dₑ = c − m and SSD = dₐ² + dᵦ² + dₑ².
Step 4. Population variance σ² = SSD / 3; deviation σ = √σ² (nonnegative root).
Step 5. Example 1, 2, 3: m = 2; SSD = 1+0+1 = 2; σ² = 2/3; σ = √(2/3) ≈ 0.816497 (six decimals for a non-integer). Example 2, 2, 2: SSD = 0, σ = 0. Example 0, 6, 12: m = 6; SSD = 36+0+36 = 72; σ² = 24; σ = √24 ≈ 4.898979.
Formal correctness follows from the second central moment of the empirical uniform measure on three points (mass 1/3 each). The square root is uniquely defined as a nonnegative function.
Three equal values → σ = 0. Arithmetic progression 1, 2, 3 → σ = √(2/3). Symmetric values around zero −2, 0, 2: m = 0, σ² = (4+0+4)/3 = 8/3, σ = √(8/3) ≈ 1.632993. Decimals 1.5 / 2.5 / 3.5 behave like a shifted 1, 2, 3 sequence: the same σ ≈ 0.816497.
An empty field is not allowed — all three values are required. Invalid text yields an error message. Zero is a number: 0, 0, 0 gives σ = 0; 0, 6, 12 gives ≈ 4.898979.
This page does not compute standard deviation from a variable-length list, does not apply weights, does not drop outliers, and does not switch the primary result to divisor n − 1.
Computation uses IEEE-754 floating-point arithmetic in the browser (√). For typical school and engineering inputs the result matches paper arithmetic after rounding to six decimals. An integer result (for example 0) displays without trailing zeros; a non-integer uses six decimal places, as on the other statistics calculators.
With extremely large |a|, |b|, |c|, cancellation when subtracting the mean can lose precision. In educational practice three moderate-scale fields are fine. The number-line diagram marks a, b, c and m with a dashed line.
Related: two-number standard deviation (divisor 2), variance of n numbers (σ² from a list), arithmetic mean, range (max − min), median. The coefficient of variation CV = σ / |m| is not computed here.
Sources: NIST/SEMATECH e-Handbook — Measures of Scale; Wikipedia — Standard deviation (population vs sample); MathWorld — Standard Deviation; Excel STDEV.P vs STDEV.S documentation. Educational tool.
For 1, 2, and 3 the mean is 2, variance is 2/3, and σ ≈ 0.816497.
0.816497
What is σ with 1, 2, and 3? ≈ 0.816497.
0
What is σ with 2, 2, and 2? 0.
4.898979
What is σ with 0, 6, and 12? ≈ 4.898979.
0.816497
What is σ with 3, 2, and 1? ≈ 0.816497.
0.816497
What is σ with 1.5, 2.5, and 3.5? ≈ 0.816497.
1.632993
What is σ with −2, 0, and 2? ≈ 1.632993.
The population standard deviation σ of exactly three numbers a, b, and c: first the mean m = (a+b+c)/3, then variance σ² = SSD/3, then σ = √σ². For 1, 2, and 3 the result is 0.816497.
No. The variance and deviation denominator is n = 3. That is the population convention (like STDEV.P). A sample estimator with Bessel’s correction (divisor 2) appears in the panel only for comparison, not as the primary result.
σ with divisor n describes the spread of the entered triple as a population. s with divisor n−1 estimates the variance of an unknown larger population. This page returns σ; for 1, 2, 3 the sample s would be 1.
0. The mean is 2, all deviations are zero, so SSD = 0 and σ = 0. An integer result displays without six trailing zeros.
4.898979. The mean is 6, SSD = 72, population variance 24, and the square root of 24 to six decimals is 4.898979.
No. The formula is symmetric under permutation. 1, 2, 3 and 3, 2, 1 both give 0.816497.
The neighboring page has two fields and divides by 2 (also /n). Here there are always three fields and divisor 3. This is not the same calculator under a different name.
On the variance of n numbers page. Here you always use three separate fields and get the square root — standard deviation, not σ² alone.
Yes. 1,5 and 1.5 are the same number. A comma and a period in the fields are equivalent.
No. The currency and unit settings in the page header do not affect this result. 1, 2, 3 stay 0.816497 regardless of the header label.
Compute the mean m = (a+b+c)/3, the sum of squared deviations from m, divide by 3 (population), then take the square root. With 1, 2, and 3: m = 2, σ² = 2/3, σ ≈ 0.816497.
About 0.816497. That is √(2/3) — a classic school example of three consecutive numbers with population divisor 3.
About 0.816497. The same variance 2/3 as with 1, 2, 3 — only a shifted mean. Spread does not change under a constant shift.
No. The primary result is σ with divisor 3. The panel may show a comparative s with divisor 2, but that does not replace the primary result.
Not on this card. There are exactly three fields. A longer list belongs on the n-number variance page or a general standard-deviation tool — here the model is locked to a, b, c.
The σ / s definitions and n versus n−1 divisors below come from recognized sources. This page computes population σ (n = 3).
Page Updated November 2, 2026.