Anomalous behaviour of water

Fill a glass bottle to the brim with water, seal it, and leave it outside on a freezing night — by morning the glass is cracked, not the water. Almost every other liquid would simply shrink as it cooled and freeze into a smaller, denser lump of solid. Water refuses to play along: it gets denser as it cools only up to a point, then does the opposite, and the ice it finally forms is puffier, not tighter, than the liquid it came from. This one stubborn exception is why ice cubes float in your drink, why frozen lakes still have swimming fish under the ice in the middle of winter, and why pipes burst in a cold snap. This article explains what water actually does as it cools, why it does it, and why that "wrong" behaviour turns out to be one of the luckiest facts in nature.

The core insight

Water is densest at 4 degrees Celsius, not at its freezing point, so as it cools further and turns to ice it expands rather than contracts — making ice float on liquid water instead of sinking.

The core idea

The normal rule, and where water breaks it. For most substances, cooling squeezes particles closer together, so density rises smoothly and steadily all the way down to the freezing point, then jumps a bit higher again on freezing into a compact solid. Water follows this normal rule only part of the way. As liquid water cools from room temperature toward 4 degrees Celsius, its density rises just like any other liquid. But below 4 degrees Celsius, instead of continuing to climb, the density turns around and falls again as the water heads toward 0 degrees Celsius and freezes. This reversal — density rising, peaking, then falling — is called the anomalous expansion of water, "anomalous" simply meaning "not following the usual pattern."

Density itself is nothing exotic — it is just how tightly mass is packed into a volume:

Density=MassVolume.(1)\text{Density} = \frac{\text{Mass}}{\text{Volume}}. \qquad (1)

where: density is measured in grams per cubic centimetre (or kilograms per cubic metre), mass is the amount of matter present (unchanged by heating or cooling), and volume is the space that mass occupies (which does change with temperature). Since mass never changes as water is heated or cooled, every change in density is really a story about volume expanding or shrinking.

The three-stage story of cooling water. Track a beaker of water as it cools from room temperature down past freezing, and volume tells the whole tale in three acts:

  • Stage 1 (above 4 degrees Celsius): normal contraction. Cooling reduces the random jostling of water molecules, they pack slightly closer, and volume shrinks — so density rises, exactly like any ordinary liquid.
  • Stage 2 (4 degrees Celsius): the density peak. At this exact temperature, water reaches its smallest volume for a given mass, so its density is at its highest — higher than at 0 degrees Celsius and higher than as ice.
  • Stage 3 (4 degrees Celsius down to 0 degrees Celsius, and freezing): anomalous expansion. Below 4 degrees Celsius, water molecules begin locking into a loose, open, cage-like hydrogen-bonded arrangement that needs more room than the molecules use when they can still tumble past one another. Volume increases even though the water is still cooling, so density decreases. When the water finally freezes at 0 degrees Celsius, this open cage structure becomes fixed — ice occupies roughly 9% more volume than the same mass of liquid water at 0 degrees Celsius, which is why ice is less dense than liquid water and floats on it.
Figure
Density of water plotted against temperature showing a peak at 4 degrees Celsius, higher than at 0 degrees and much higher than ice
Density of water versus temperature — the peak at 4 degrees Celsius temperature in degrees Celsius density of liquid water 0 4 15 25 density highest here ice block (bulkier, less dense) floats on liquid water

Why it works

There is no numerical derivation needed here — the anomaly comes from how water molecules bond, so the reasoning is structural rather than mathematical. A water molecule can form up to four hydrogen bonds with its neighbours. Above roughly 4 degrees Celsius, thermal motion is vigorous enough that molecules keep breaking and re-forming these bonds constantly, so they can still tumble and slide close together as the liquid cools — the ordinary contraction effect wins, and density rises like it would for any liquid. Below about 4 degrees Celsius, thermal jostling has calmed down enough that hydrogen bonds start to survive longer and organise the molecules into small clusters that resemble the open, hexagonal cage structure of ice — a structure that is roomier than randomly packed liquid water, precisely because each molecule now holds itself at arm's length from four neighbours instead of squeezing in close to as many as possible. This "bond-ordering" expansion effect grows stronger as the temperature keeps falling, and below 4 degrees Celsius it wins the tug-of-war against ordinary thermal contraction — so overall volume starts increasing again even though the water keeps losing heat. At exactly 4 degrees Celsius, the two competing effects are in balance: it is the last point where shrinking-by-cooling still dominates, so it is the single coolest temperature at which water is at its most closely packed. Push past it and the hydrogen-bond cage-building effect takes over completely, culminating in the fully open lattice of ice at 0 degrees Celsius, where every molecule is locked into just four bonds arranged as an airy hexagonal scaffold — emptier, and therefore less dense, than the liquid it froze from. This has a major ecological payoff. When a lake's surface cools toward freezing, the coldest, lightest surface water floats above the denser 4-degree water below, and finally the ice that forms on top floats and acts as an insulating lid. Denser water at 4 degrees Celsius stays liquid at the bottom of the lake all winter, letting fish and other aquatic life survive beneath the ice rather than being frozen solid throughout.

Worked example

A sealed glass bottle is completely filled with water, capped tightly, and left outdoors overnight where the temperature falls well below 0 degrees Celsius. By morning, the bottle is found cracked. Explain what happened, using the anomalous expansion of water.

Step 1. The bottle starts completely full of liquid water with the cap sealed, so there is no empty space inside for the water to expand into.

Step 2. As the outdoor temperature first drops toward 4 degrees Celsius, the water behaves normally: it contracts slightly, so its volume actually decreases a little — this stage alone would never crack a sealed bottle.

Step 3. As cooling continues from 4 degrees Celsius down toward 0 degrees Celsius, the anomalous expansion effect takes over: the water's volume stops shrinking and starts growing again, even though it is still losing heat, because hydrogen bonds are organising the molecules into a roomier, cage-like arrangement.

Step 4. When the water reaches 0 degrees Celsius and freezes solid, this open cage structure locks in place completely, and the ice occupies noticeably more volume — about 9% more — than the same mass of liquid water did.

Step 5. Since the rigid glass bottle was already completely full and sealed, it has nowhere to give, so the expanding ice pushes outward on the glass walls with enough force to crack them.

The cracked bottle is direct proof of the anomalous expansion of water: ice takes up more space than the liquid water it froze from, and a completely full, sealed container cannot hold that extra volume.

Competitive stretch

The same reasoning explains a subtler puzzle: why do underground water pipes in cold countries burst most often on the coldest, calmest nights, and why does damage often appear at a bend or joint rather than along a straight run? A straight pipe segment full of water can sometimes accommodate a little expansion by bulging slightly along its length, but a bend or joint is the stiffest, most constrained part of the system — exactly where the outward push of expanding ice has the least room to be absorbed, so it concentrates stress and fails first. This is the same physics as the sealed bottle, just distributed along a pipe instead of a bottle wall.

Common traps

  • Assuming water gets denser and denser all the way down to its freezing point like most liquids — remember the density peak is at 4 degrees Celsius, not at 0 degrees Celsius.
  • Thinking ice must be denser than water because it feels hard and solid — solids are usually denser than their own liquid, but ice is the exception and is actually less dense than liquid water.
  • Getting the direction of volume change backwards — freezing water expands (volume increases), the opposite of most substances, which contract when they freeze.
  • Forgetting the connection to volume — density changes because volume changes (mass stays fixed), so any explanation should trace back to why the water's volume is growing or shrinking.

Ask the tutor: Want to see how the same hydrogen-bond idea explains water's unusually high boiling point, or try more density-and-temperature reasoning problems? Ask the AI tutor to walk you through it.

Why does water behave so differently from almost every other liquid?

It comes down to hydrogen bonding. Water molecules can form up to four hydrogen bonds with their neighbours, and below about 4 degrees Celsius these bonds start locking molecules into an open, cage-like arrangement rather than letting them pack in as closely as thermal motion alone would allow. Most other liquids do not have this strong, direction-specific bonding, so they simply keep shrinking as they cool, all the way to freezing.

Is the anomalous behaviour of water true for all temperatures, or only near freezing?

Only in the narrow range between 0 and about 4 degrees Celsius. Above 4 degrees Celsius, water behaves completely normally: it expands on heating and contracts on cooling, just like other liquids. The anomaly is strictly a low-temperature effect caused by hydrogen bonds starting to dominate over ordinary thermal contraction, and it disappears once the water is warmer than 4 degrees Celsius.

If ice is less dense than water, does that mean ice weighs less than the same amount of water?

No — the mass is exactly the same, since freezing does not add or remove any matter. What changes is the volume: the same mass of water spreads out into a larger volume once it becomes ice, so its density (mass divided by volume) goes down. A one-kilogram block of ice and one kilogram of liquid water have identical mass; the ice simply takes up more space to hold it.

Before this

Where this leads