Temperature and density

A hot air balloon does not fly by magic — it flies because heated air is less dense than the cool air around it, so it floats upward exactly the way a cork floats up through water. That single idea, temperature changes density, explains balloons, convection currents, and why a layer of warm water sits on top of a cold pond. But water keeps one famous secret from this pattern: cooled all the way to freezing, it does the opposite of what every other liquid does, and that quirk is the reason ice floats and ponds do not freeze solid from the bottom up. This article builds the density-temperature link from the basic formula, works it with real numbers, and pins down exactly where water breaks the rule.

The core insight

Heating a fixed mass of matter usually increases its volume and so lowers its density, cooling does the reverse, and water's unique exception between 4°C and 0°C — where it expands instead of contracting — is why ice floats.

The core idea

Density packages two simple measurements, mass and volume, into one number that tells you how tightly matter is packed:

ρ=mV\rho = \frac{m}{V}

where: ρ\rho (rho) is density, mm is mass (unchanged by heating or cooling, since no particles are added or removed), and VV is volume. Heating a substance pumps thermal energy into its particles. They vibrate and move more vigorously, and on average settle a little farther apart, so the same mass spreads over a larger volume. Since mass sits on top and volume sits on the bottom of the density fraction, a bigger volume with the same mass means a smaller density. Cooling reverses the whole story: particles slow down, drift closer together, volume shrinks, and density rises. This is the entire physical picture behind a hot air balloon — heating the trapped air lowers its density below that of the surrounding cool air, and the less-dense balloon rises exactly as a low-density object floats up through a denser fluid. Water breaks this pattern in one narrow, important range. Cooling water from room temperature down to 4°C, density rises as normal. But between 4°C and 0°C, water's particles start locking into the open, hexagonal arrangement that becomes ice, and this rearrangement makes the water expand even while it keeps losing heat. So water reaches its maximum density at 4°C, not at its freezing point, and by the time it turns to ice at 0°C it has become distinctly less dense than the liquid it froze from. That is precisely why ice floats — and why a floating ice cap insulates the liquid water beneath it, letting fish and other aquatic life survive an entire winter under the ice rather than being frozen solid. Pressure tells a related but separate story: squeezing a substance pushes its particles closer, raising density. Gas particles have huge amounts of empty space between them, so pressure changes their density a lot; liquids are already fairly close-packed, so pressure barely changes them; solids are packed tightest of all, so pressure has almost no effect on their density.

Figure
Density versus temperature for a normal liquid and for water showing waters dip to maximum density at 4 degrees before it rises again as ice forms
Density vs temperature: a normal substance compared with water temperature (rising to the right, 0 to 10 degrees C) density a normal liquid: density falls steadily as it warms max density at 4°C ice forms below 0°C, less dense than water water
Derivation AHow much density changes when volume changes a little

The goal is to connect a small, measured fractional change in volume during heating to the resulting fractional change in density, so that a percentage volume increase read off an experiment can be turned directly into a percentage density decrease without needing to know either the mass or the original volume separately.

Step 1. Start from the definition, ρ=m/V\rho = m/V, applied before heating with original volume VV, so the original density is ρ=m/V\rho = m/V.

Step 2. After heating, the mass is exactly the same (mm does not change), but the volume grows by a small fraction x=ΔV/Vx = \Delta V/V, so the new volume is V′=V(1+x)V' = V(1+x) and the new density is ρ′=mV(1+x)\rho' = \dfrac{m}{V(1+x)}.

Step 3. Divide the new density by the old one to isolate the fractional change: ρ′ρ=m/V(1+x)m/V=11+x\dfrac{\rho'}{\rho} = \dfrac{m/V(1+x)}{m/V} = \dfrac{1}{1+x}.

Step 4. For a small fractional change xx (a few per cent or less), the algebraic fact 11+x≈1−x\dfrac{1}{1+x} \approx 1-x holds closely, so ρ′≈ρ(1−x)\rho' \approx \rho(1-x). In words: the fractional decrease in density is approximately equal to the fractional increase in volume, for small changes. This is why a 2% increase in volume on heating produces a density decrease of very nearly 2%, not exactly 2%, because the true relation is the reciprocal 1/(1+x)1/(1+x) and the approximation 1−x1-x is only its first-order stand-in. Verified: for x=0.02x=0.02, exact factor 1/(1.02)=0.980391/(1.02)=0.98039, approximation 1−0.02=0.981-0.02=0.98 — the two agree to within 0.04%.

Worked example

Question: A metal ball has a volume of 50 cm350\ \text{cm}^3 and a mass of 135 g135\ \text{g} at room temperature. It is heated in a furnace and its volume grows by 2%2\%, while its mass stays unchanged. Find the ball's density before and after heating, and the percentage change in density.

Step 1. Find the original density using ρ=m/V\rho = m/V: ρ=135 g50 cm3=2.70 g/cm3\rho = \dfrac{135\ \text{g}}{50\ \text{cm}^3} = 2.70\ \text{g/cm}^3.

Step 2. Find the new volume after a 2%2\% increase: V′=50×1.02=51 cm3V' = 50 \times 1.02 = 51\ \text{cm}^3.

Step 3. Find the new density with the same mass over the new volume: ρ′=135 g51 cm3=2.647 g/cm3\rho' = \dfrac{135\ \text{g}}{51\ \text{cm}^3} = 2.647\ \text{g/cm}^3.

Step 4. Find the percentage change: ρ−ρ′ρ×100=2.70−2.6472.70×100≈1.96%\dfrac{\rho - \rho'}{\rho} \times 100 = \dfrac{2.70 - 2.647}{2.70} \times 100 \approx 1.96\%. The ball's density drops from 2.70 g/cm32.70\ \text{g/cm}^3 to about 2.65 g/cm32.65\ \text{g/cm}^3, a decrease of roughly 2%2\% — matching the derivation's rule that a small volume increase produces almost the same fractional density decrease.

Competitive stretch

Using the shortcut from the derivation directly, without dividing 135/51135/51 at all: a 2%2\% volume increase gives an approximate density decrease of 2%2\%, so ρ′≈2.70×(1−0.02)=2.70×0.98=2.646 g/cm3\rho' \approx 2.70 \times (1-0.02) = 2.70 \times 0.98 = 2.646\ \text{g/cm}^3 — within 0.001 g/cm30.001\ \text{g/cm}^3 of the exact division above. The shortcut is fast and reliable whenever the volume change is a few per cent or less; for larger changes, always divide exactly instead.

Common traps

  • Assuming density always rises as something cools, with no exceptions — water is the well-known exception between 4°C and 0°C, where further cooling actually decreases density.
  • Thinking ice sinks because it is cold and heavy — ice floats precisely because it is less dense than the liquid water it came from, the opposite of that intuition.
  • Treating the small-change rule ρ′≈ρ(1−x)\rho' \approx \rho(1-x) as exact — it is only an approximation for small fractional volume changes xx; for large changes, divide mass by the new volume directly.
  • Believing pressure affects gases, liquids, and solids equally — it strongly compresses gases, barely affects liquids, and has almost no effect on solids.

Ask the tutor: Want to work through why the small-change approximation breaks down for a big volume jump, or see more examples of the anomalous expansion of water?

Why does volume usually increase when something is heated?

Heat energy makes particles vibrate and move faster, and on average they push a little farther apart from their neighbours as they gain this extra motion. Since the number of particles and their total mass have not changed, spreading the same particles over more space simply means the substance now occupies a larger volume. That is the whole mechanism — nothing is added, the particles just take up more room.

If ice is less dense than water, why does a glass of water with ice cubes not overflow when the ice melts?

The ice cube already displaces exactly the volume of water that its weight would occupy as liquid, because floating objects displace a volume of liquid equal to their own weight, not their own volume. When the ice melts, that meltwater fits into precisely the space the floating ice was displacing, so the water level barely changes. This is a floating-object detail beyond the density-temperature rule itself, but it explains a very common everyday observation.

Does the metal-ball density calculation mean the metal itself changed into something less dense permanently?

No — the change is temporary and due only to thermal expansion. As soon as the metal ball cools back to room temperature, its particles move back closer together, its volume shrinks back toward 50 cm350\ \text{cm}^3, and its density returns to 2.70 g/cm32.70\ \text{g/cm}^3. Nothing about the metal's composition or identity has changed; only its volume flexed with temperature, exactly as the density formula predicts.

Before this

Where this leads