Why Glass Is Not A Normal Solid

A popular myth claims old cathedral windows are thicker at the bottom because glass "flows" slowly over centuries, like an extremely viscous liquid oozing under gravity. It's a great story, and it's wrong: those windows are thicker at the bottom because of how medieval glassmakers manufactured flat glass panes, not because the glass moved. But the myth persists partly because it points at something real: glass genuinely doesn't solidify the way a metal or a mineral does, and its actual atomic structure has puzzled physicists for well over a century.

What makes a solid "normal" in the first place

To understand why glass is unusual, it helps to know what counts as ordinary. Crystalline solid — a solid whose atoms or molecules are arranged in a repeating, long-range ordered pattern called a lattice. Salt, quartz, diamond, and ordinary ice are all crystalline. If you could shrink down and walk through a salt crystal, every sodium and chlorine atom would sit in a predictable position relative to its neighbors, the same pattern repeating in every direction for as long as the crystal extends.

That repeating order is what a crystallographer measures with X-ray diffraction, a technique in which X-rays bounced off a material produce a pattern of sharp, well-defined spots when the atoms are arranged periodically. Crystalline materials form this way almost by default when a liquid cools slowly: as thermal energy drops, atoms settle into the lowest-energy configuration available, and for most simple substances that lowest-energy configuration is a repeating lattice. Table salt crystallizing from evaporating seawater, water freezing into hexagonal ice, or quartz crystallizing from cooling magma over geological timescales, all follow this pattern.

Crystallization is also why most solids have a sharp, single melting point. Ice melts at precisely 0 degrees Celsius (32 degrees Fahrenheit) at standard atmospheric pressure, not "somewhere between minus five and plus five." At the melting point, the ordered lattice has exactly enough thermal energy to break apart into the disordered arrangement of a liquid, and the transition happens at one well-defined temperature because every unit of the crystal shares the same bonding environment. This sharpness is a direct consequence of long-range order: because the whole crystal is built from the same repeating unit, that unit melts everywhere at once.

Glass breaks this pattern in two separate, related ways: it never forms the repeating lattice in the first place, and it doesn't melt at one sharp temperature. Both of those facts trace back to how glass is actually made, which is the subject of the next section.

Cooling too fast to crystallize

Most solids form by crystallization: as a liquid cools past its melting point, atoms or molecules lock into a repeating, long-range ordered lattice, the same principle behind salt or quartz crystals. Glass forms differently. When molten silica-based glass is cooled, it's cooled quickly enough, relative to how fast its atoms can rearrange themselves, that the disordered arrangement of atoms characteristic of a liquid gets "frozen in" before the atoms have time to organize into a crystal lattice.

The key variable here is viscosity, a measure of a fluid's resistance to flow. As molten glass cools, its viscosity increases dramatically, and the individual silicon and oxygen atoms (or, in other glass formulations, other network-forming atoms) find it progressively harder to slide past one another into the neat rows a crystal would require. By the time the material is cool enough that crystallization would normally happen, the atoms are already too sluggish to complete the rearrangement. The disordered, liquid-like arrangement simply gets stuck.

The result is a material with the mechanical rigidity of a solid, atoms are locked in place and don't flow, but with the disordered, non-repeating atomic arrangement of a liquid. Materials scientists call this an amorphous solid — a solid material that lacks the long-range, repeating atomic order of a crystal while still being mechanically rigid. Glass is the most familiar example, though amorphous structure also shows up in some plastics, certain ceramics, and in rapidly cooled metals called metallic glasses, which are quenched from the molten state so quickly (sometimes in milliseconds) that even metal atoms, which crystallize readily under normal conditions, get frozen into a disordered arrangement.

Cooling rate matters enormously here. Silica-based glass can be cooled relatively slowly compared to metallic glass and still end up amorphous, because pure silica's atomic network is naturally reluctant to crystallize; the silicon-oxygen bonding geometry makes an ordered lattice comparatively hard to assemble even given time. Metals, by contrast, crystallize so readily that producing a metallic glass typically requires cooling rates on the order of thousands of degrees per second, fast enough that the liquid metal has essentially no chance to organize before it locks solid. Both processes, slow-crystallizing silica and rapidly quenched metal, arrive at the same structural outcome: atoms fixed in place, but without the periodic order of a crystal lattice.

The glass transition: not a sharp melting point

Crystalline solids have a sharp, well-defined melting point where the material transitions abruptly from solid to liquid. Glass instead has a glass transition temperature (often abbreviated Tg) — a temperature range over which the material gradually softens from a rigid, glassy state into a viscous, flowing liquid state, without a single sharp transition temperature the way ice has a precise 0 degrees Celsius (32 degrees Fahrenheit) melting point.

Approaching the glass transition from the liquid side, the material is technically called a supercooled liquid, a liquid cooled below the temperature at which a crystal would normally form, but which has not yet crystallized. As a supercooled liquid cools further, its viscosity climbs, not gradually, but by many orders of magnitude over a relatively narrow temperature range. Right around Tg, the viscosity typically increases to roughly 10^12 pascal-seconds, a value physicists use as a practical, if somewhat arbitrary, marker for calling the material a solid rather than an extremely slow-moving liquid (the number itself is a widely used convention in glass science, not a hard physical boundary). Below that point, atomic rearrangement becomes so slow that, for any practical human timescale, the structure is effectively frozen.

Physicist Philip Anderson, in a widely cited 1995 commentary published in Science, wrote that "the deepest and most interesting unsolved problem in solid state theory is probably the theory of the nature of glass and the glass transition" (Anderson, 1995). That statement still gets quoted decades later because the underlying physics, exactly how and why a supercooled liquid's viscosity increases by twelve or more orders of magnitude over a comparatively narrow temperature window as it approaches the glass transition, remains an active area of condensed matter physics research. A 1999 poll of roughly 250 physicists conducted by Physics World ranked the nature of the glass transition among the top unsolved problems condensed matter physicists wanted to see resolved.

Why the glass transition doesn't have one exact temperature

Unlike melting, which is a true thermodynamic phase transition with a precise temperature at a given pressure, the glass transition is better described as a kinetic phenomenon: where exactly it appears to happen depends partly on how fast you cool the material. Cool a given glass slowly and Tg measures slightly lower; cool it quickly and Tg measures slightly higher, because the atoms have more or less time to keep adjusting before their motion effectively freezes. This is one reason scientists describe Tg as a "range" rather than a single fixed number the way they would describe a melting point.

A useful analogy: parking a crowd versus stacking crates

Think of a crystal as a warehouse where identical crates are stacked in perfectly aligned rows and columns, every crate's position determined by the ones around it, extending in a predictable, repeating pattern throughout the entire warehouse. If you know the position of one crate and the layout rule, you can predict the position of every other crate in the building, arbitrarily far away. That predictability, extending indefinitely through the material, is what "long-range order" means in materials science.

Glass is more like a parking lot where cars pulled in quickly during a rush and got stuck wherever they happened to stop, close together and immobile, but with no repeating grid pattern connecting one car's position to the next. Locally, a handful of neighboring cars might look roughly evenly spaced, the same way silicon and oxygen atoms in glass maintain a fairly consistent local bonding arrangement. But there's no rule that lets you predict where the car three rows over is parked just from knowing where the car next to you is. That's the difference between short-range order, which glass has, and long-range order, which it lacks.

Both arrangements are dense and rigid; neither the crates nor the cars are moving. That's the part people miss when they hear "glass is basically a liquid": rigidity and lack of long-range crystalline order are not the same axis. A material can be immobile and mechanically solid, exactly like glass, while still lacking the repeating lattice that defines a true crystal. Glass's atomic structure is the parking lot: locally packed and stable, but globally disordered.

A second analogy helps with the cooling-rate piece specifically. Imagine a large group of people trying to form a marching band's precise grid formation on a field. If the band director gives them plenty of time and calls out positions calmly, everyone can shuffle into a perfectly aligned rectangular grid, the crystal outcome. But if the director shouts "Freeze exactly where you are" thirty seconds after everyone rushes onto the field, people end up clustered in rough clumps, standing close to whoever happened to be nearby, with no overall grid pattern connecting the clumps to each other. That sudden freeze, imposed before the slow process of self-organization can finish, is structurally what happens to silica atoms as molten glass cools past its glass transition.

Glass at room temperature is estimated to have a viscosity so far beyond that of any observably flowing liquid that mathematical models built from measured glass viscosity data indicate it would take longer than the current age of the universe for room-temperature glass to visibly rearrange itself under gravity (American Ceramic Society, summarizing glass viscosity research).

Why the "cathedral glass flows" myth is false

The actual explanation for uneven old window panes traces to the cylinder or crown glass-making methods used before modern float glass processes. In the crown glass method, a glassblower spun a molten glob of glass on the end of a rod (or pontil) at high speed, and centrifugal force flattened it into a rough disc that was naturally thinner near the center, where it had been spun from, and thicker toward the outer edge. Panes were then cut from that disc, meaning individual panes routinely came out with measurably uneven thickness across their surface, a manufacturing artifact, not a sign of any ongoing physical process.

Craftsmen installing those panes commonly placed the thicker edge at the bottom of the window frame, for straightforward structural reasons: a heavier, thicker edge resting on the sill provided better support and stability than mounting a pane thick-side-up, which would have put more mass higher in the frame. Over centuries, as windows were reglazed or repaired, installers repeatedly made the same practical choice, reinforcing the pattern seen in many surviving old windows today.

This explanation has been documented by materials scientists and glass historians, including analysis published by the Corning Museum of Glass, which notes that glass at room temperature is far too viscous, with a viscosity vastly exceeding what would allow any observable flow over a human timescale, for gravity to produce measurable sagging within centuries (Corning Museum of Glass, "Does glass continue to flow even at room temperature?"). Research on the stained-glass windows of London's Westminster Abbey, reported by the American Ceramic Society and covered by outlets including Science and Futurity, reached the same conclusion using direct measurement and modeling of medieval window glass rather than relying on manufacturing history alone. Their mathematical modeling, built from measured glass viscosity, found that room-temperature glass would take dramatically longer than the age of the universe to noticeably reflow under its own weight, a timescale that makes any centuries-scale "flow" physically implausible regardless of the manufacturing explanation (American Ceramic Society coverage of Westminster Abbey glass research; Science, "Cathedral Glass Myth Shattered").

Notably, not every old window follows the thick-bottom pattern. Some historical panes were installed thick-side-up, sideways, or with no consistent orientation at all, which is itself evidence against the flow explanation: if gravity were slowly reshaping glass over centuries, every sufficiently old pane would trend toward being thicker at the bottom regardless of how it was originally installed. Instead, the orientation of thickness in surviving windows tracks how each pane happened to be cut and mounted, exactly what you'd expect from a manufacturing and installation explanation rather than a physical process acting uniformly on all old glass.

Claim	Status	Basis
Cathedral glass is thicker at the bottom	True	Observable in many, though not all, surviving windows
The unevenness is caused by glass flowing like a liquid over centuries	False	Contradicted by manufacturing history and glass viscosity measurements
Glass is technically an amorphous solid, not a true crystalline solid	True	Well established in materials science
Room-temperature glass viscosity would allow visible flow within centuries	False	Modeling shows the required timescale exceeds the age of the universe

Four common claims about cathedral glass, checked against manufacturing history and measured glass viscosity data.

Is glass technically a liquid?

No, not in any meaningful physical sense. Glass is classified as an amorphous solid: it has the rigidity and mechanical resistance to flow of a solid, its atoms are essentially fixed in position, but it lacks the long-range repeating atomic order that defines a crystalline solid. The popular claim that glass is a "very slow liquid" that flows over centuries is not supported by glass viscosity data or by the documented manufacturing history of old windowpanes.

Why the amorphous structure actually matters

Glass's lack of a crystal lattice, and the corresponding absence of the grain boundaries and crystal planes that exist in crystalline materials, is precisely what gives it useful optical properties. Grain boundary — the interface between two differently oriented crystal regions within a polycrystalline material, which can scatter light passing through it. Because glass has no crystal grains at all, it transmits visible light without the light-scattering effects that grain boundaries can cause in translucent crystalline ceramics, which is a large part of why window glass, laboratory glassware, and optical lenses can all be made highly transparent from essentially the same chemistry (silica) that, arranged into a crystal, would be opaque quartz-based ceramic.

This same disordered structure is also why glass fractures unpredictably along conchoidal fracture patterns, curved, shell-like breaks, rather than cleaving cleanly along crystal planes the way minerals like mica or salt do. Crystalline minerals tend to break along their internal lattice planes, producing flat, predictable fracture surfaces, because those planes are structurally weaker directions built into the repeating lattice. Glass has no such planes to break along, so a crack propagates however the local stress field dictates, producing the smooth, curved fracture surfaces (sometimes with faint ripple marks) that toolmakers historically prized in obsidian, a naturally occurring volcanic glass, for producing sharp cutting edges.

The disordered structure also explains some of glass's more practically important properties. Because glass lacks the well-defined slip planes that let crystalline metals deform plastically (bending without breaking), glass tends to be brittle: it resists deformation right up until it reaches its fracture stress, then breaks suddenly rather than bending. This is also why glass manufacturing techniques like tempering, a process that rapidly cools the surface of glass to put it under compressive stress, are used specifically to counteract that brittleness by making a crack much harder to initiate at the surface. And it's why amorphous solids in general, from window glass to certain polymers, behave so differently under stress than the metals and crystalline ceramics engineers are also accustomed to working with.

How glass composition changes the picture

Not all glass is chemically identical, and composition changes both how easily a material forms glass and what its glass transition temperature turns out to be. Ordinary window and container glass is soda-lime glass, a mixture of silica (silicon dioxide) with sodium and calcium compounds added to lower the melting temperature needed for manufacturing, making it cheaper and easier to produce at scale than pure silica glass. Borosilicate glass, used in laboratory glassware and oven-safe cookware, substitutes boron oxide into the network, which lowers the material's thermal expansion and makes it far more resistant to cracking from rapid temperature changes, a property called thermal shock resistance.

The underlying physics of amorphous structure and a gradual glass transition applies across all of these formulations, but the specific glass transition temperature, viscosity behavior, and mechanical properties shift with composition. Pure fused silica glass has a notably higher glass transition temperature than soda-lime glass precisely because removing the sodium and calcium "network modifiers" leaves a more rigid silicon-oxygen network that resists rearrangement to a higher temperature. This composition dependence is also why not all amorphous materials are glasses in the everyday sense: amorphous silicon used in some solar panels and amorphous metal alloys used in specialty transformers share the defining structural feature, disordered atomic arrangement without long-range order, but have entirely different chemistry and applications from window glass.

What people get wrong about glass

The single most common misconception is the cathedral window myth itself, treating an observation about historical manufacturing as evidence of an ongoing physical process. Independent viscosity measurements and manufacturing records both rule this out, and the persistence of the myth is a useful reminder that a true observation (uneven old windows) can still support a false causal explanation (glass flow) if the actual mechanism (historical manufacturing technique) isn't checked. The myth is a good example of a broader pattern worth watching for in any scientific claim: a real, verifiable observation gets bolted onto a plausible-sounding but untested causal story, and the story spreads faster than the correction.

A second, more technical misconception is thinking "amorphous" means "unstructured" or random at every scale. Glass has meaningful short-range atomic order; silicon and oxygen atoms in silica glass still bond in a fairly consistent local arrangement, typically each silicon atom bonded to four oxygen atoms in a tetrahedral geometry, the same local building block found in crystalline quartz. What glass lacks is the long-range, repeating periodicity that connects one of those local tetrahedra to the next in a predictable pattern extending through the whole material. "Amorphous" describes the absence of long-range order specifically, not the absence of all structure.

A third misconception worth flagging: some popular explanations claim glass is "somewhere between a solid and a liquid" as its permanent, steady-state condition, which overstates the ambiguity. At room temperature, well below its glass transition temperature, glass behaves entirely as a solid in every practical and most rigorous physical senses: it holds its shape, resists deformation, and shows no measurable flow on any human-relevant timescale. The genuinely ambiguous, actively studied physics is concentrated specifically around the glass transition itself, the temperature range where a supercooled liquid is becoming a rigid amorphous solid, not a permanent characteristic of glass sitting on a shelf.

The takeaway

Glass earns its "not a normal solid" reputation honestly: it's rigid and doesn't flow at room temperature, qualifying it as a solid in every practical sense, but its atoms are frozen into the disordered arrangement of a liquid rather than the repeating lattice of a crystal. That amorphous structure, and the strange, gradual glass transition instead of a sharp melting point, is exactly what physicist Philip Anderson flagged as one of solid-state physics's deepest unsolved puzzles, even as the myth about flowing cathedral windows has been thoroughly debunked by both manufacturing history and direct glass-viscosity measurement. The real story, uneven medieval manufacturing meeting a sensible installation choice, turns out to be less dramatic than the myth, but it's the version that survives contact with the physics.

---

Read next:

• Why Some Materials Remember Their Shape
• How Crystals Grow From Solutions
• Why Turbulence Is Hard to Predict

FAQ

Q: Is glass actually a slow-moving liquid?

No. This is a persistent myth. Glass is an amorphous solid: rigid, with atoms essentially fixed in place, but lacking the long-range repeating atomic order of a crystal. Viscosity measurements show glass at room temperature is far too viscous to flow noticeably over a human lifetime, let alone centuries.

Q: Why are old cathedral windows thicker at the bottom?

This results from historical glassmaking methods, particularly the crown glass process, that produced panes of naturally uneven thickness after being spun from a central point, which craftsmen typically installed thicker-side-down for stability, not from glass gradually flowing under gravity, according to analysis from glass historians including the Corning Museum of Glass and glass-viscosity modeling reported by the American Ceramic Society.

Q: What is the glass transition temperature?

The glass transition temperature (Tg) is the range over which an amorphous material like glass gradually softens from a rigid glassy state into a viscous flowing state, in contrast to the sharp, well-defined melting point of a crystalline solid like ice. Where Tg appears to fall can shift slightly depending on how fast the material is cooled.

Q: Why did physicist Philip Anderson call the glass transition an unsolved problem?

In a 1995 commentary in Science, Anderson wrote that "the deepest and most interesting unsolved problem in solid state theory is probably the theory of the nature of glass and the glass transition," because the precise physics behind how a supercooled liquid's viscosity increases by many orders of magnitude as it approaches the glass transition remains incompletely understood.

Q: What makes glass an amorphous solid rather than a crystal?

Glass cools too quickly, relative to how fast its atoms can rearrange, for its atoms to organize into the long-range, repeating lattice structure that defines a crystal. Instead, the disordered atomic arrangement typical of a liquid gets locked in place, producing a material that is mechanically rigid but structurally disordered.

Q: Does all glass have the same glass transition temperature?

No. Glass transition temperature depends on chemical composition. Pure fused silica glass has a notably higher glass transition temperature than common soda-lime window glass, because soda-lime glass contains sodium and calcium compounds that loosen the silicon-oxygen network and let it soften at a lower temperature.