A convection current forms because the hot rock heats the water beside it, that water expands and becomes less dense, and colder, denser water sinks beneath it to take its place. That sinking pushes the warmed water upward and sideways, and the loop repeats for as long as a temperature difference survives.
The tank version of this is a closed loop. Heat enters at the hot rock, leaves at the cold rock, and the water in between carries the energy from one to the other. Nothing needs to be added after the start except the two rocks themselves. The current is the water's way of evening out a temperature gap it cannot conduct away fast enough.
The buoyancy engine, step by step
Water near the hot rock absorbs heat through its surface. The molecules move faster and spread slightly farther apart, so the same mass occupies more volume. Density falls. Fresh water at roughly 68°F (20°C) drops about 0.0002 to 0.0003 grams per cubic centimeter for every 1.8°F (1°C) of warming, which sounds trivial until you stack it across a tank.
That density difference creates a pressure imbalance. A column of warm, light water weighs less than a column of cool water of the same height. The heavier column wins, sliding underneath and shouldering the lighter one upward. This is the same reason a hot-air balloon rises, scaled down and slowed by water's much higher density and viscosity.
The upward-moving parcel does not rise forever. It loses heat to the surrounding water as it travels, and when it reaches the cold rock it dumps what remains. Chilled, it contracts, grows denser, and begins to sink along the far side of the tank. The result is a closed circuit: up the hot side, across the top, down the cold side, across the bottom.
The loop is self-sustaining only while the rocks stay at different temperatures. Once both reach the same temperature, the density difference vanishes and the current stalls. In a real tank this happens over minutes to hours depending on insulation, tank size, and how much thermal mass the rocks carry.
Quantified context: how much water moves, and how fast
The driving force is buoyancy, and its size scales with the temperature difference between the two rocks. A tank holding a 50°F (28°C) gap between a hot and a cold rock generates a density contrast of roughly 1 percent between the two water columns. That 1 percent is enough to drive visible circulation in a tabletop aquarium within seconds of placing the rocks.
Flow speed depends on geometry. In a roughly 12-inch (30-centimeter) tank with a modest heat source, convection currents typically creep along at fractions of an inch per second near the rock surfaces. In a larger vessel or with a stronger heat source, the same mechanism can move water at several inches per second. The current is fastest where the tank is narrowest, because the same volume of water is squeezed through a smaller channel.
Water's heat capacity sets the timescale. It takes about 1 BTU to raise one pound of water by 1°F, and a cubic foot of water weighs roughly 62 pounds. A five-gallon tank holds about 42 pounds of water, so warming the whole tank by 10°F requires on the order of 420 BTU, minus whatever the room and the rocks absorb. This is why small tanks equilibrate faster than large ones.
The counterintuitive detail: the water closest to the hot rock does not simply rise in a straight line. It tends to form a thin boundary layer that clings to the rock, warms, then peels away in a plume. Most of the tank's water stays relatively still while a narrow, fast-moving circuit does the actual transport. The bulk of the water is a passenger, not a driver.
The timeline of a convection current
| Time after placing rocks | What happens in the tank |
|---|---|
| 0–10 seconds | Heat begins crossing the hot rock's surface into the thin water layer touching it |
| 10 seconds–2 minutes | Warm water near the hot rock becomes less dense and starts to rise; cold water near the cold rock begins to sink |
| 1–5 minutes | A visible circulation loop establishes: up the warm side, across the top, down the cold side |
| 5–30 minutes | The current reaches steady speed, set by the temperature gap and tank geometry |
| 30 minutes–several hours | The two rocks drift toward the same temperature; the current slows gradually |
| Once temperatures equalize | The density difference disappears and the current stops entirely |
Exact timings shift with tank size, insulation, and the rocks' heat capacity. A well-insulated tank with large thermal masses stretches the whole sequence toward the longer end.
Why the loop is closed, and when it is not
In a sealed tank, water cannot enter or leave, so the current must be a closed loop. Whatever rises must eventually come back down somewhere else. This is why you see a coherent circuit rather than random swirls: the tank's walls and the two rocks constrain the flow into a predictable pattern.
Open systems behave differently. If the tank is heated from below with no cold rock, the water still circulates, but the pattern is driven by the bottom plate rather than a horizontal temperature contrast. This is the classic Rayleigh–Bénard setup, and it produces rising plumes and sinking sheets rather than a single loop.
The Rayleigh number predicts whether convection starts at all. It compares buoyant forces to the damping effects of viscosity and thermal diffusion. Below a critical threshold, heat moves by conduction alone and no current forms. Above it, convection kicks in. For water in a tank, that threshold is crossed easily with even a modest temperature difference, which is why the current appears so reliably in classroom demonstrations.
Common misconceptions
Hot water does not "want" to rise. It rises because cooler, denser water pushes it up. The buoyant force is really the surrounding pressure difference acting on a less-dense parcel. Remove the surrounding water and there is nothing to push anything.
The current is not driven by the rocks pushing water. The rocks are stationary. They transfer heat; the water does the moving. Thinking of the rocks as paddles leads to wrong predictions about flow direction.
Convection is not the same as conduction. Conduction moves heat through a material without moving the material. Convection moves the heated material itself. In your tank, both happen: conduction carries heat from rock surface into water, then convection carries the warmed water away.
A single temperature reading does not tell you the current is strong. You can have a large temperature difference with a weak current if the tank is wide and shallow, or a small difference with a brisk current in a narrow channel. Geometry matters as much as temperature.
Practical actions for a reliable demonstration
Use a tank at least 8 inches (20 centimeters) deep so a full loop can form. Shallow trays produce mostly surface spreading. Place the hot rock and cold rock on opposite sides, not in the center. A horizontal temperature contrast drives a cleaner loop. Start with a temperature gap of at least 18–27°F (10–15°C). Smaller gaps work but take longer to show visible motion. Add a few drops of food coloring near the hot rock. The dye traces the plume and makes the current visible without altering the physics. Insulate the tank's outer walls if you want the current to persist. An uninsulated glass tank loses heat to the room and shortens the run. Do not stir the water after placing the rocks. Stirring destroys the boundary layers that the current depends on.
Red-flag warnings as bullets: Do not use a glass container that is not rated for the temperature swing; thermal shock can crack ordinary glass. Do not seal a tank that will be heated; trapped pressure can build as water warms. Do not handle hot rocks with bare hands or place them directly on a cold countertop.
Comparison: convection versus other heat-transfer modes
| Mode | What moves | Speed in water | Needs a fluid? |
|---|---|---|---|
| Conduction | Heat only, molecule to molecule | Slow in water | No |
| Convection (free) | The water itself, driven by buoyancy | Moderate, set by temperature gap | Yes |
| Convection (forced) | Water pushed by a pump or stirrer | Fast, set by pump | Yes |
| Radiation | Infrared energy through space | Very fast, no medium | No |
The tank's current is free convection. Add a stirrer and you convert it to forced convection, which overwhelms the buoyancy-driven pattern and makes the flow direction depend on the stirrer instead of the rocks.
Limits of the simple picture
Real tanks are messier than the textbook loop. Water at the surface evaporates and cools, adding a second sinking region. Room drafts cool one wall more than another. The hot rock's surface cools unevenly, so the plume wobbles. These effects can split the single loop into multiple cells.
The buoyancy explanation also assumes the water is a simple fluid. It is close enough for a science demo, but water's density peaks near 39°F (4°C), so cold water below that point behaves oddly: cooling it further makes it less dense, not more. In a tank kept near freezing, the current can reverse direction as the cold rock's water passes the density maximum. This is a genuine exception to the "cold always sinks" rule.
Finally, the current only redistributes heat; it does not create or destroy it. The total energy in the tank stays fixed (minus losses through the walls), which is why the rocks and water all converge on one temperature in the end. The convection current is a transient phenomenon, not a perpetual one.
A practical takeaway: to keep a current going, you must keep supplying a temperature difference. That is true of a classroom tank, a lake in spring, and Earth's mantle, which convects over millions of years for the same buoyancy reason at vastly higher viscosity and temperature. The mechanism scales from a tabletop to a planet; only the numbers change.