Is Glass a Solid, or a Liquid?
— The "Slow Flow" Story Doesn't Add Up
"Old church windows are thicker at the bottom, because glass has been slowly flowing like a liquid for centuries" — you may have heard this before. It's a charming story, but when you actually run the numbers, it falls apart. For glass to genuinely flow that much at room temperature would take not centuries, but a length of time larger by many orders of magnitude.
Some European churches built in the Middle Ages still have their original glass in place today. Look closely, and you'll sometimes see that the glass is thicker at the bottom than at the top.
From this comes the explanation: "Glass looks solid, but it's actually an incredibly slow-flowing liquid, and it has sagged downward over centuries." It's still told today as a bit of fun science trivia.
But the story has one weak point. Nobody actually calculates "if it really did flow over that much time, how far would it flow?"
Once you do the calculation, you find that a few centuries is far too short a time for glass to flow by any visible amount.
Its atoms are arranged irregularly, like in a liquid, yet mechanically it's firmly solid. This double nature is where the misunderstanding creeps in.
Calculate how resistant to flow room-temperature glass really is, and you find that a few centuries isn't even a rounding error against the time actually needed.
So why is old glass unevenly thick? We'll look at the real reason too.
Glass isn't "frozen liquid" — it's an "amorphous solid"
Ordinary ice or metal is a crystal, with atoms arranged in regular order. Cool it from a liquid, and at some temperature it snaps into a neat, ordered crystal.
Glass sets in a different way. Cool molten glass down, and its atoms never line up neatly. They carry over the disordered arrangement they had as a liquid, and simply stop being able to move. It looks like a transparent solid, yet its atomic-scale arrangement resembles a liquid — this is what's called an "amorphous solid."
This duality — "arranged like a liquid, but behaves like a solid" — is exactly what invites the misconception that it's "really a liquid." But mechanically, glass is unquestionably a solid. Hit it and it shatters; it holds its shape; it doesn't flow and spread out.
That doesn't change the fact that it's firmly solid.
Let's actually calculate the "slow flow"
This is the heart of the article. You often hear that "glass is an ultra-slow-motion liquid that flows given a few centuries." But hardly anyone puts a number on "how much it would flow, and over what time."
Glass, like anything else, has a property called viscosity. As with honey or tar (pitch), the higher the viscosity, the harder something is to flow, and the longer it takes to change shape.
At the temperature where glass softens enough to actually work with (the glass transition point, roughly 550°C for window glass), this viscosity has already dropped to a level far stiffer than honey. But at room temperature, the viscosity climbs by yet more orders of magnitude beyond that.
Run the actual formula (details in the fold-out section at the end), and the time needed for room-temperature glass to flow visibly turns out to be not centuries, but something vastly longer. The time from the Middle Ages to now is just a tiny fraction of that required time.
So why is old glass unevenly thick?
The real reason lies in how it was made.
From the Middle Ages for some time after, sheet glass was made by hand. Molten glass was blown into a balloon-like shape, then opened out or spun to flatten it into a sheet. This method doesn't produce an even thickness. It creates a manufacturing quirk where the part nearer the centre ends up thicker and the part nearer the edge thinner.
Squares were then cut from the finished large sheet and fitted into windows. Even within a single cut piece, thickness already varied.
And it's thought that glaziers, when fitting a pane into a window, often placed the thicker side at the bottom. The thinking is that a thicker bottom edge sat more stably in the frame and was easier to handle. This is a human decision, not a physical phenomenon.
If glass really did slowly flow downward under gravity and thicken at the bottom, then every piece of old glass should be thicker in the same direction (downward).
But in fact, old glass panes have been found with the thicker side mounted at the top or on the side. This is evidence that the direction of the thickness variation was random from the start, pane by pane. That's something that couldn't happen if gravity-driven flow were the cause — and it fits much better with the explanation that manufacturing variation, plus the glazier's own judgement in fitting it, decided which way the thick side faced.
Glass doesn't flow at all — not quite true either
To be clear about one thing: "it doesn't flow over centuries" is not the same as "it never flows at all."
Heat glass up, and of course it softens and can be made to flow and change shape — that's exactly how glassblowers shape it freely. It's also thought that, over an extraordinarily long span of time (the kind of order-of-magnitude timescale this article's calculation points to), some slight change might in theory be possible.
What this article is disputing is only the claim that it flows visibly over the comparatively short span of a few centuries. Get the temperature and time scales right, and the whole story changes.
Something you can check for yourself
- Get two clear glasses and fill them to the same level with ① water and ② honey (or a similar syrup)
- Tilt each glass slowly, by the same angle
- Notice that the water moves right away, while the honey moves slowly
- If you have a hard candy (like a boiled sweet — chemically similar to glass), tilt it the same way too. Notice it doesn't move at all
- Line up the order of viscosity: water → honey → hard candy (essentially motionless)
Steps 4 and 5 are the point of this experiment. You can feel for yourself that "something that doesn't seem to move at all" is just the far end of "high viscosity." Glass at room temperature is thought to have a viscosity even more dramatically higher than this candy.
Summary
Glass has an atomic arrangement resembling a liquid, but mechanically it is a genuine solid (an amorphous solid). The story that "old glass is thicker at the bottom because it flowed over centuries" doesn't hold up once you actually calculate the time involved. The real reason is that old hand-made manufacturing methods never produced an even thickness in the first place.
Glass is a solid pretending to be a liquid.
But its acting can't outlast a few centuries.
Want to know more? — Terms, numbers, and how this connects to the textbooksWe've labelled which level each part belongs to, from middle-school science to open research questions
- MSCovered in middle-school science
- HSCovered in high-school "Chemistry Basics"
- HS+High-school "Chemistry"/"Physics," or advanced/sidebar textbook content
- UnivNot covered in high school — undergraduate specialist content (materials science)
- ResearchNot even settled as "established" at university — something researchers are actively investigating
MSTerms: the vocabulary of glass
- Crystal: a solid in which atoms or molecules are arranged in regular order. Ice and table salt are examples.
- Amorphous: a state with irregular atomic arrangement. The term used for the state of glass.
- Viscosity: a measure of resistance to flow. The higher the viscosity, the longer it takes to change shape.
- Glass transition point: the rough temperature at which cooled, set glass starts to soften on heating.
- Crown glass / cylinder glass: hand-made sheet-glass manufacturing methods used from the Middle Ages into the modern era.
HSChecking with a formula: just how resistant to flow is glass at room temperature?
The main text says "it doesn't flow in a few centuries." Let's confirm this using actual physical quantities. We'll use a simple relation that gives a rough "time to change shape" from viscosity.
Rough time to change shape = Viscosity ÷ Stiffness (elastic modulus)
| Rough time to change shape | units: seconds |
| Viscosity | units: Pa·s (pascal-seconds). Larger means harder to flow |
| Stiffness (elastic modulus) | units: Pa. For glass, taken as roughly 3×10¹⁰ Pa |
This is a commonly used relation representing the rough time it takes a material to "dissipate an applied force entirely through viscosity" after being pushed. The higher the viscosity, the longer this time becomes.
At the glass transition point (roughly 550°C), where window glass starts to soften, the viscosity works out at roughly 10¹² Pa·s — this is in fact how this reference temperature is defined.
| Viscosity at the glass transition point | 10¹² Pa·s |
| Stiffness (elastic modulus) | 3 × 10¹⁰ Pa |
| Divide | 10¹² ÷ (3 × 10¹⁰) ≒ 33.3 |
| Rough time to change shape | ~33 seconds |
The calculation gives shape change starting in just over 30 seconds at the glass transition point. That matches the actual feel glassworkers get working at this temperature (it softens almost immediately).
Room temperature (~20°C) is about 530°C lower than the glass transition point (~550°C).
Amorphous substances like glass have the property that viscosity shoots up by orders of magnitude as temperature drops below the transition point. Let's make a fairly conservative assumption here.
| Conservative estimate: room-temp viscosity as a multiple of the transition-point value | 10¹² times (thought to actually be far larger) |
| Multiply the time from ② by this factor | 33 × 10¹² seconds |
| 1 year ≈ 3.15 × 10⁷ seconds | |
| Convert to years | (33 × 10¹²) ÷ (3.15 × 10⁷) ≒ 1.05 × 10⁶ |
| Rough time to change shape | at least ~1.05 million years |
Even under this "conservative assumption," the figure comes out above a million years. It's only been about 800 years since medieval churches were built.
| Time needed (conservative estimate) | ~1.05 million years |
| Time actually elapsed | ~800 years |
| Ratio of the gap | 1050000 ÷ 800 = 1312.5 times |
Even on the conservative estimate, only about 1/1300th of the required time has elapsed. And as noted in the main text, the actual room-temperature viscosity is thought to be far higher still than this calculation assumes. However you estimate it, "glass flowed and changed thickness over a few centuries" is nowhere near having enough time.
* The way viscosity changes between the glass transition point and room temperature isn't a simple constant multiplier. This is a conservative approximation meant only to show that "there simply isn't enough time."
HS+The glass transition point isn't a fixed temperature
The glass transition point isn't fixed at a single temperature. It's a reference temperature defined by reaching a chosen viscosity value (10¹² Pa·s in this article).
Because of that, the observed glass transition point shifts somewhat depending on how fast the glass is cooled. Cooling more slowly gives the atoms more time to approach an ordered arrangement, so the transition point tends to be measured slightly lower. In other words, "becoming glass" isn't an instant event — it's a gradual transition with some breadth to it.
UnivWhy does viscosity rise "abnormally" as temperature falls?
In an ordinary liquid (like water), viscosity rises gently as temperature drops. But for most glass-forming substances, the rate at which viscosity increases becomes abnormally steep as they approach the glass transition point. This is called "fragile" behaviour, and it can't be described by a simple exponential function.
A more complex relation called the Vogel-Fulcher-Tammann equation is used to describe this sharp change. The "simple multiplier" used in the ②③ calculation above is only a rough simplification to get the order of magnitude right — the real physics follows this more complex equation.
Because of this property, directly measuring accurate viscosity at temperatures far below the glass transition point is extremely difficult experimentally. The time needed for the measurement itself would stretch far beyond what's practical.
ResearchWhat's still unclear
- The nature of the glass transition itself is considered one of the great unsolved problems in physics. Debate continues over how it differs from a phase change (like water becoming ice) — whether it's simply "motion slowing down until it looks solid," or whether some structural change is also involved.
- There's a theoretical idea called the "ideal glass transition." It's the theoretical prediction (related to the Kauzmann paradox) that if you could cool infinitely slowly, there would be a temperature at which viscosity truly becomes infinite — but this has never been confirmed experimentally.
- Because accurate viscosity near room temperature can't be measured directly, estimates rely on theory or extrapolation from measurements at slightly higher temperatures. Depending on how that extrapolation is done, estimated values still carry considerable uncertainty. That's why this article describes its calculation as a "conservative estimate."
- Exactly how atoms in glass keep moving very slightly (if they do at all) is also an active research topic. Computer simulations are used to probe motion on extremely short timescales.
Behind the seemingly simple question "is glass a liquid or a solid?" lies territory physics still hasn't fully answered. A familiar material that's also, it turns out, a subject of cutting-edge research.
Connections to the textbook (by level)
| Level | Subject/Unit | Where in this article |
|---|---|---|
| MS | Science — states of matter and crystals | Difference between crystalline and amorphous |
| HS | Chemistry Basics — states of matter | Calculating time from viscosity and elastic modulus |
| HS+ | Chemistry — amorphous solids | Definition of glass transition point, relation to cooling rate |
| Univ | Materials science / rheology | Fragile behaviour, Vogel-Fulcher-Tammann equation |
| Research | Condensed matter physics (unresolved) | Ideal glass transition, room-temperature viscosity extrapolation, atomic-scale motion |
| — | History of science / craft | Crown glass method, examples of old church glass |
- Zanotto, E. D., Do cathedral glasses flow?, American Journal of Physics 66, 1998 (a well-known paper that quantitatively tested this popular myth).
- Debenedetti, P. G. & Stillinger, F. H., Supercooled liquids and the glass transition, Nature 410, 2001 (a review of the physics of the glass transition).
- Explanations of the glass transition point and viscosity definitions from Ceramic Society of Japan (日本セラミックス協会) glass-engineering textbooks.
- Angell, C. A., Formation of glasses from liquids and biopolymers, Science 267, 1995 (research on fragile behaviour).
- Historical materials on the history of sheet glass and medieval crown-glass manufacturing methods.
* Figures for viscosity, elastic modulus, and the temperature multiplier are representative approximations and vary with actual composition and conditions. Room-temperature viscosity is estimated by extrapolation, since direct measurement is impractical.
*This article is a general-audience science explainer. The figures given are approximations meant to aid understanding, and in particular the room-temperature viscosity of glass is based on a theoretical estimate. For details on the glass in historic buildings, please consult the relevant institution or specialist sources.