In short
The geometric mean (GM) of two positive numbers a and b is \sqrt{ab} — the single number that, when placed between them, creates a three-term geometric progression. More generally, inserting n geometric means between a and b creates a GP of n + 2 terms, with common ratio r = \left(\dfrac{b}{a}\right)^{1/(n+1)}. The GM is always less than or equal to the arithmetic mean of the same numbers, with equality only when the numbers are equal — a relationship that underpins one of the most powerful inequalities in mathematics.
A farmer owns two square fields. One has side length 4 metres, the other has side length 9 metres. Their areas are 16 m² and 81 m². The farmer wants to buy a single square field whose area equals the geometric average of those two areas — not the sum, not the ordinary average, but a kind of multiplicative midpoint. What should its side length be?
The area of the new field should satisfy: 16, x, 81 form a geometric progression. That means the ratio from 16 to x equals the ratio from x to 81:
Cross-multiply: x^2 = 16 \times 81 = 1296. So x = 36, and the side length is \sqrt{36} = 6 metres.
Notice what happened. The side lengths were 4 and 9, and the "multiplicative midpoint" side length turned out to be \sqrt{4 \times 9} = \sqrt{36} = 6. The areas were 16 and 81, and their multiplicative midpoint is \sqrt{16 \times 81} = \sqrt{1296} = 36. In both cases, you multiplied the two numbers and took the square root. That operation is the geometric mean.
The GM of two numbers
Take two positive numbers a and b. Their geometric mean is
The name "geometric" is fitting. The arithmetic mean adds and divides; the geometric mean multiplies and roots. Where the AM is about additive balance — equal distances on a number line — the GM is about multiplicative balance — equal ratios.
The three numbers a, G, b form a geometric progression. The common ratio is
and you can verify that \dfrac{b}{G} = \dfrac{b}{\sqrt{ab}} = \sqrt{\dfrac{b}{a}} — the same ratio. So the GM is the middle term of a three-term GP, just as the AM is the middle term of a three-term AP.
Geometric Mean of two numbers
For any two positive real numbers a and b, the geometric mean is
The three numbers a, G, b form a geometric progression with common ratio r = \sqrt{b/a}.
For the farmer's fields: a = 16, b = 81, and G = \sqrt{16 \times 81} = \sqrt{1296} = 36.
A visual way to see the GM
There is a beautiful geometric construction of the GM using a semicircle. Take a segment of length a + b. Draw a semicircle with this segment as diameter. At the point that divides the diameter into lengths a and b, erect a perpendicular to the diameter. This perpendicular meets the semicircle at a height of exactly \sqrt{ab} — the geometric mean.
The proof uses the right-triangle altitude theorem. The triangle inscribed in the semicircle has a right angle at the top (by Thales' theorem). If the altitude from the right angle to the hypotenuse has length h, and the hypotenuse is split into segments of length a and b, then h^2 = ab, giving h = \sqrt{ab}.
Inserting n geometric means
Just as you can insert arithmetic means between two numbers, you can insert geometric means. Placing n geometric means between a and b creates a GP of n + 2 terms starting at a and ending at b.
Take a = 2 and b = 162 with n = 3 means. You need 3 + 2 = 5 terms: a, G_1, G_2, G_3, b. Since these form a GP, the common ratio satisfies
So the three inserted means are:
and the full GP is 2, 6, 18, 54, 162.
The general formula: to insert n geometric means between positive numbers a and b, the common ratio of the resulting GP is
and the k-th inserted mean (for k = 1, 2, \dots, n) is
The logic parallels the AM case. You have n + 2 terms forming a GP, so there are n + 1 multiplicative steps. The total multiplicative span is b/a, so each step must multiply by the (n+1)-th root of b/a.
The product of n geometric means
Here is the multiplicative counterpart of the AM sum property. The product of all n inserted geometric means equals the n-th power of the simple GM of a and b:
The proof: G_k = a \cdot r^k, so the product is a^n \cdot r^{1+2+\cdots+n} = a^n \cdot r^{n(n+1)/2}. Since r^{n+1} = b/a, you get r^{n(n+1)/2} = (b/a)^{n/2}, and thus a^n \cdot (b/a)^{n/2} = a^{n/2} \cdot b^{n/2} = (ab)^{n/2}.
For the example above: 6 \times 18 \times 54 = 5832, and (\sqrt{2 \times 162})^3 = (18)^3 = 5832. Confirmed.
Properties of the geometric mean
Property 1: The GM lies between the two numbers. For positive a \leq b, the GM satisfies a \leq \sqrt{ab} \leq b. The geometric mean never falls outside the range of its inputs.
Property 2: Scale invariance. If you multiply both a and b by the same positive constant k, the GM also gets multiplied by k: \sqrt{(ka)(kb)} = k\sqrt{ab}. This makes the GM natural for comparing quantities that differ by scale — growth rates, ratios, percentages.
Property 3: The GM of reciprocals is the reciprocal of the GM. \sqrt{\dfrac{1}{a} \cdot \dfrac{1}{b}} = \dfrac{1}{\sqrt{ab}}. This symmetry between a set and its reciprocals is a property the AM does not have.
Property 4: Logarithmic connection. The logarithm of the GM is the AM of the logarithms:
This is why the GM appears naturally in any context where logarithmic scales are appropriate — decibels, pH values, Richter magnitudes, percentage growth rates. Taking the GM of two numbers is the same as taking their AM on a logarithmic scale.
The GM is never greater than the AM
This is one of the most fundamental inequalities in mathematics. For any two positive numbers a and b:
with equality if and only if a = b.
The proof is three lines. Start from the fact that any real square is non-negative:
Expand:
Rearrange:
Equality holds precisely when \sqrt{a} = \sqrt{b}, that is, when a = b. As soon as a and b differ, the AM strictly exceeds the GM. The gap grows larger as the two numbers become more unequal.
The AM-GM inequality extends to n positive numbers:
with equality if and only if x_1 = x_2 = \cdots = x_n. The full proof and applications are in AM-GM-HM Inequality.
Two worked examples
Example 1: Insert 2 geometric means between 5 and 320
The problem asks for a sequence 5, G_1, G_2, 320 — four terms forming a GP.
Step 1. Count the ratio-steps. There are 2 + 2 = 4 terms, so 4 - 1 = 3 ratio-steps.
Why: n geometric means between two endpoints create n + 2 terms total, and a GP of n + 2 terms has n + 1 ratio-steps.
Step 2. Find the common ratio.
Why: the total multiplicative span b/a = 64 must be split into 3 equal ratio-steps, so each step multiplies by 64^{1/3} = 4.
Step 3. List the means.
Why: each mean is a \cdot r^k for k = 1, 2.
Step 4. Verify: the full GP is 5, 20, 80, 320. Each ratio is \dfrac{20}{5} = \dfrac{80}{20} = \dfrac{320}{80} = 4. Also check the product property: 20 \times 80 = 1600, and (\sqrt{5 \times 320})^2 = (\sqrt{1600})^2 = 1600. Confirmed.
Result: The two geometric means are 20 and 80.
The bar chart makes the multiplicative structure visible: each bar is exactly four times its predecessor, so the ratios are uniform even though the absolute height differences are vastly different (15 from 5 to 20, but 240 from 80 to 320).
Example 2: Prove that if $a$, $b$, $c$ are in AP and $a$, $b$, $c$ are all positive, then $b^2 \geq ac$
This connects the AM-GM inequality to progressions. The claim is that the square of the middle term of an AP is at least the product of the outer terms.
Step 1. Write the AP condition. Since a, b, c are in AP:
Why: the middle term of a three-term AP is the arithmetic mean of the outer two terms.
Step 2. Apply the AM-GM inequality. Since a and c are positive:
Why: the arithmetic mean of two positive numbers is at least their geometric mean.
Step 3. Substitute b = \dfrac{a + c}{2} into the inequality:
Step 4. Square both sides (both sides are positive, so squaring preserves the inequality):
Why: squaring is a monotone operation for positive quantities, so the direction of the inequality is preserved.
Result: b^2 \geq ac, with equality if and only if a = c (which forces a = b = c, a constant sequence that is simultaneously an AP and a GP).
The figure gives a concrete check: for the AP 4, 7, 10, you get b^2 = 49 and ac = 40. The inequality 49 \geq 40 holds, and the gap of 9 comes from the fact that a \neq c.
Common confusions
-
"The GM of two numbers is always an integer." Rarely. \sqrt{2 \times 8} = 4 is an integer, but \sqrt{3 \times 5} = \sqrt{15} \approx 3.87 is not. The GM is an integer only when ab is a perfect square.
-
"The GM can be applied to negative numbers." Not in the standard real-number sense. \sqrt{(-3)(5)} = \sqrt{-15} is not a real number. The GM is defined for positive real numbers (or for two negatives, where the convention takes the negative root).
-
"The GM is always smaller than the AM." Almost — the GM is always less than or equal to the AM. Equality holds when all the numbers are equal. The word "or equal to" matters.
-
"n geometric means between a and b uses r = (b/a)^{1/n}." The exponent should be 1/(n+1), not 1/n. There are n + 1 ratio-steps between n + 2 terms. This mirrors the AM error where people write d = (b-a)/n instead of (b-a)/(n+1).
-
"The GM and the AM of two numbers can never be equal." They can — precisely when a = b. In that case both means equal a itself.
Going deeper
If you came here to learn what the geometric mean is, how to insert geometric means, and why the GM never exceeds the AM, you have it. The rest of this section explores the GM for more than two numbers and its connection to growth rates.
The GM for n numbers
For n positive numbers x_1, x_2, \dots, x_n, the geometric mean is
All the properties carry over. The logarithm of the GM is the AM of the logarithms:
and the AM-GM inequality holds:
Average growth rates
The GM is the correct mean for averaging multiplicative rates. Suppose an investment grows by 60% in year one (r_1 = 1.6) and shrinks by 20% in year two (r_2 = 0.8). After two years, ₹100 becomes 100 \times 1.6 \times 0.8 = ₹128. The average annual growth factor is not the AM (1.6 + 0.8)/2 = 1.2 — that would predict 100 \times 1.2^2 = ₹144, which is wrong. The correct average is the GM: \sqrt{1.6 \times 0.8} = \sqrt{1.28} \approx 1.1314, and 100 \times 1.1314^2 \approx ₹128. The GM preserves the final outcome, just as the AM preserves the total.
The AM-GM inequality in optimisation
The AM-GM inequality is a powerful tool for finding maximum or minimum values. For example: among all rectangles with a fixed perimeter 2p, which one has the largest area? If the sides are x and p - x, the area is x(p - x). By AM-GM:
with equality when x = p - x, i.e., x = p/2 — a square. The maximum area is \dfrac{p^2}{4}, achieved by the square. This is one of the simplest applications, but the same technique handles far more complex optimisation problems.
Where this leads next
The geometric mean sits between the arithmetic mean and the harmonic mean, and the three together form a chain of inequalities that is one of the most important tools in algebra.
- Geometric Progression — the sequence whose consecutive ratios are constant, and whose middle term is the GM of its neighbours.
- Arithmetic Mean — the additive counterpart of the GM, always at least as large for positive numbers.
- Harmonic Mean — the third classical mean, defined via reciprocals, and always at most as large as the GM.
- AM-GM-HM Inequality — the full chain \text{HM} \leq \text{GM} \leq \text{AM}, with applications to optimisation and inequalities.
- Sum of Geometric Progression — what happens when you add up the terms of a GP, and the infinite sum formula when |r| < 1.