The water rises primarily because of thermal expansion and contraction of the trapped air, not because the candle “uses up” the oxygen. When you light a candle sitting in a dish of water and cover it with a glass, the flame heats the air inside. That warm air expands, and some of it escapes as bubbles around the rim. Once the flame dies, the remaining air cools rapidly, its pressure drops below atmospheric pressure, and the higher pressure outside pushes water up into the glass. This thermal mechanism is well documented in physics education literature, yet the oxygen-consumption explanation remains stubbornly popular in classrooms and online demonstrations alike.
The Explanation Most People Learn Is Wrong
The version you probably heard goes like this: the candle burns oxygen inside the glass, the oxygen disappears, and since oxygen makes up roughly a fifth of air, the water rises about a fifth of the way up. It sounds tidy, and the water level does end up somewhere in that neighborhood, which makes the story feel confirmed. But the chemistry does not actually support it. When a candle burns, it consumes oxygen molecules and produces carbon dioxide and water vapor. In simple terms, the gas molecules consumed are largely replaced by new gas molecules. The net change in the total number of gas molecules is small, nowhere near enough on its own to account for the dramatic rise of water you see. The fact that the final water level often lands near the one-fifth mark is coincidental, a quirk of how the thermal dynamics and other factors happen to shake out in a typical setup.
This misconception has been a topic of concern in science education for decades. A paper in the Journal of Chemical Education specifically examined its persistence, noting how the wrong explanation continues to circulate through textbooks, science fair guides, and classroom demonstrations despite clear evidence against it.1Journal of Chemical Education. The Persistence of the Candle-and-Cylinder Misconception The problem is partly that the incorrect explanation feels intuitive. People know fire needs oxygen, and they know oxygen is part of air, so the leap to “fire removes oxygen, gas volume shrinks, water fills the gap” seems logical. But intuition and physical reality part ways here.
What Is Actually Happening Inside the Glass
The real driver is temperature. The moment you place the glass over the burning candle, the flame heats the air inside the enclosed space. Air expands when heated, and in this setup the glass is sitting in water with its rim submerged just slightly. That means the expanding hot air has somewhere to go: you can often see bubbles escaping around the base of the glass in the first few seconds. Those escaping bubbles represent air that has been pushed out of the container. The total amount of air trapped inside the glass actually decreases while the candle is still burning.
Then the candle goes out. Without the flame, the air inside begins to cool. As it cools, it contracts and occupies less volume. But the glass is now sealed by the water around its rim, so no outside air can rush back in to replace what was lost as bubbles. The contracting air creates a partial vacuum, a zone of lower pressure compared to the atmosphere pressing down on the water surface outside the glass. The atmosphere pushes the water up and into the glass until the pressures equalize. That rising water is what everyone notices, and it is driven overwhelmingly by the cooling and contraction of air, not by the chemical removal of oxygen.
A 2025 study in Physics Education confirmed this explanation by simultaneously measuring temperature, pressure, and water height inside the glass over time. The researchers found that the water level tracked closely with the pressure changes inside the glass, which in turn tracked with the air temperature, exactly what the thermal model predicts.2Physics Education. Joint measurement of temperature, pressure and water height as a function of time in the candle covered by a glass experiment
Four Stages of the Water’s Movement
If you watch the experiment closely, the water does not simply stay still and then rise. It goes through a more complex sequence that researchers have broken into roughly four stages.
In the first stage, the candle is burning and heating the air. The air expands, pressure inside the glass temporarily rises above atmospheric pressure, and the water level actually drops slightly or stays flat as air is pushed outward. You may notice a few bubbles escaping under the rim during this phase.
In the second stage, the system reaches a rough equilibrium. The flame is still burning but the rate of heating slows as the glass itself warms up. The water level holds relatively steady, though small fluctuations can occur.
The third stage begins when the flame goes out. This is where the dramatic action happens. The air temperature drops quickly, pressure inside the glass plunges, and water rushes upward. Most of the visible rise occurs in this stage, and it happens fast, often within a few seconds of the flame dying.
The fourth stage is a gradual settling. The air inside continues to cool toward room temperature, and the water creeps up a bit more, but much more slowly. Eventually the system reaches a new equilibrium where the weight of the water column inside the glass balances the pressure difference between inside and outside. The four-stage model emerged from sensor-based measurements that tracked all the variables simultaneously, showing that each phase has a distinct pressure and temperature profile.2Physics Education. Joint measurement of temperature, pressure and water height as a function of time in the candle covered by a glass experiment
Why Does the Flame Actually Go Out
A related question people often conflate with the water rise is why the candle stops burning. The short answer is that the flame suffocates, but not quite in the way most people imagine. Oxygen levels inside the glass do drop as the candle burns, but the flame typically goes out well before oxygen is fully depleted. Candle flames extinguish when the oxygen concentration drops to somewhere around 15 to 16 percent, not zero. At that concentration, the combustion reaction can no longer sustain itself.
Carbon dioxide buildup plays a role too. COâ‚‚ is denser than the surrounding air mixture and tends to accumulate near the base of the glass and around the flame. As COâ‚‚ concentration increases and oxygen decreases, the flame weakens and eventually dies. The remaining air inside the glass still contains a substantial amount of oxygen, which is another reason the “oxygen consumed equals volume lost” story does not work. If the flame went out because all the oxygen was used, you would expect the water to rise by roughly the fraction of air that was oxygen. But plenty of oxygen remains when the flame dies, so even the chemistry-based volume change is much smaller than people assume.
Research has examined how COâ‚‚ accumulation and heat distribution inside the glass jointly contribute to flame extinction.3Journal of Chemical Education. Changes of CO2 Concentration and Heat Illustrate Why the Flame Is Extinguished in the Candle-and-Cylinder Experiment The flame dies from a combination of diluted oxygen and accumulated COâ‚‚, not from oxygen depletion alone.
Why the Myth Persists
The oxygen-consumption story is one of the more stubborn misconceptions in science education, and a few factors explain why. First, it produces a prediction that roughly matches observation. The water rises somewhere near one-fifth of the glass height, and oxygen is roughly one-fifth of air. That numerical coincidence is powerful. It feels like confirmation even though the real mechanism is unrelated to that fraction.
Second, the thermal explanation is harder to see. Bubbles escaping during the heating phase are easy to miss if you are not looking for them. The water drop before the rise is subtle. And the rapid cooling after the flame dies is invisible. By contrast, “fire consumes oxygen” is a story with clear, familiar characters. People know fire needs oxygen, and they feel they can “see” the oxygen being consumed as the flame shrinks and dies. The thermal explanation requires you to think about something you cannot observe directly: invisible air expanding, escaping, and then contracting.
Third, the wrong explanation keeps getting printed. It appears in science activity books, on educational YouTube channels, and in lesson plans shared online. Once a misconception is embedded in teaching materials, it perpetuates itself. Teachers learn it from the books they studied, then pass it on. The persistence of this particular error has been documented as an example of how incorrect explanations survive in educational settings even when the correct explanation is well established in physics literature.1Journal of Chemical Education. The Persistence of the Candle-and-Cylinder Misconception
How to See the Thermal Effect for Yourself
If you want to convince yourself that heat, not oxygen consumption, is the main driver, a few simple modifications to the experiment make the point clearly.
Try using a piece of hot metal instead of a candle. Heat a steel rod or bolt with a torch, place it on the dish inside the glass, and cover it quickly. There is no combustion at all, no oxygen being consumed, yet the water will still rise as the hot metal heats the air and then cools. The effect is the same because the mechanism is the same: thermal expansion followed by contraction.
Alternatively, try varying the number of candles. Use one candle, then repeat with two or three. More candles produce more heat before the flame goes out, which means more air expansion, more bubbles escaping, and a greater pressure drop when the flames die. The water rises higher with more candles, often well beyond the one-fifth mark that the oxygen story predicts. If the water rise were caused purely by oxygen removal, the number of candles should not matter much because the same amount of oxygen is available regardless. The fact that more candles produce a dramatically higher water rise is strong evidence for the thermal explanation.
You can also watch the rim of the glass carefully at the start. If bubbles escape in the first few seconds after you place the glass, that is air being pushed out by thermal expansion. Every bubble that escapes represents trapped air that will not be there when the system cools, which means more water will be pulled in later. Counting bubbles is admittedly tricky, but simply noticing their presence tells you that the gas volume inside the glass was reduced before the flame even went out.
If You Want to Actually Measure Oxygen Content
The candle experiment is sometimes presented as a way to measure the oxygen fraction of air, but it is a terrible tool for that purpose. The thermal effects swamp the chemical signal, and the flame goes out with plenty of oxygen still present. If your goal is genuinely to measure how much of the atmosphere is oxygen, there are better approaches.
One well-documented alternative uses steel wool. When fine steel wool is moistened and placed inside a closed container inverted over water, it slowly rusts. Rusting is an oxidation reaction that consumes oxygen and produces solid iron oxide, meaning the oxygen is converted from a gas into a solid rather than into another gas. This avoids the confounding thermal effects of a flame and the gas-for-gas swap problem of combustion. Over the course of a day or two, the water rises to fill the space vacated by the consumed oxygen, and the result lands reliably near the expected one-fifth mark. A study in the Journal of Chemical Education highlighted that two key factors make this experiment work: the closed atmosphere and the fact that the oxidation reaction converts a gaseous reactant into a solid product, producing a clean volume change that is easy to measure.4Journal of Chemical Education. A Simple Experiment To Measure the Content of Oxygen in the Air Using Heated Steel Wool The contrast is instructive: the steel wool experiment works precisely because it avoids the thermal noise and gas-replacement issues that make the candle version unreliable.
Water Vapor and Other Complicating Factors
The thermal story is the main one, but a few smaller effects are also in play. One is water vapor produced by the candle flame itself. Burning paraffin wax produces both COâ‚‚ and Hâ‚‚O as vapor. That water vapor initially adds to the gas volume inside the glass. But as the system cools after the flame dies, some of that vapor condenses on the inside surface of the glass and on the water surface. Condensation removes gas molecules from the air, further reducing the pressure inside and contributing a small additional water rise beyond what cooling alone would produce. You can sometimes see the inside of the glass fog up, which is this condensation happening in real time.
Another factor is the dissolution of COâ‚‚ into the water. Carbon dioxide is moderately soluble in water, and the thin layer of COâ‚‚-rich air near the water surface can lose some of its gas into solution. This effect is small over the short timescale of the experiment, but it does nibble away at the gas volume inside the glass. Neither water vapor condensation nor COâ‚‚ dissolution is the primary driver, but they contribute to why the total water rise can sometimes overshoot what a purely thermal calculation predicts.
The geometry of the setup matters too. A tall, narrow glass produces a more dramatic visual rise than a short, wide one, even if the volume of water entering is the same. The diameter of the glass relative to the candle flame changes how much the air heats, how quickly it cools, and how many bubbles escape. People who try the experiment with different containers often get different results and wonder what went wrong. Nothing went wrong; the physics is just sensitive to the proportions of the setup.
Using Sensors to Settle the Debate
For a long time, the argument between the thermal and oxygen-consumption explanations was somewhat qualitative, resting on logical arguments about gas replacement rather than direct measurement. Modern sensor technology has changed that. Recent studies have placed temperature probes, pressure sensors, and even oxygen and COâ‚‚ monitors inside the glass during the experiment, tracking everything simultaneously with sub-second resolution.
These sensor-based studies consistently show the same pattern. Pressure rises briefly as the flame heats the air, drops sharply when the flame goes out, and correlates tightly with temperature changes throughout.2Physics Education. Joint measurement of temperature, pressure and water height as a function of time in the candle covered by a glass experiment A separate sensor-based investigation using the same general approach confirmed the thermal mechanism.5European Journal of Physics. A sensor-based study of the candle-under-glass experiment The oxygen level does drop, but the timing does not match the water rise. The water stays mostly flat or even drops slightly while the candle is burning and oxygen is being consumed. It only surges upward after the flame dies and the air cools. If oxygen removal were the cause, you would expect the water to rise steadily as the candle burned and oxygen disappeared. Instead, the water waits for the thermal event. That timing mismatch is the clearest single piece of evidence against the oxygen story.
For anyone teaching or learning from this experiment, the sensor data is worth knowing about. It transforms the candle-under-glass demonstration from a misleading prop for a wrong explanation into a genuinely rich illustration of gas laws, thermal dynamics, and how easy it is for a plausible-sounding explanation to survive decades of being wrong.