Glass does not flow at room temperature in any way you would ever notice, and it will not do so over any timescale that matters to human civilization. The idea that old cathedral windows are thicker at the bottom because centuries of gravity slowly pulled the glass downward is one of the most persistent myths in popular science, but the physics does not support it. Modern calculations put the viscosity of silica glass at room temperature so astronomically high that measurable flow under gravity would take longer than the age of the universe. The story is more interesting than a simple “no,” though, because glass does occupy a genuinely strange place between solid and liquid, and researchers are still uncovering subtle atomic-level changes that happen in glasses over time.
The Medieval Window Myth
Walk through a medieval European cathedral and you can often see that the stained glass panes are thicker at the bottom than at the top. This observation is real, and it spawned a widely repeated explanation: glass is actually a very slow-moving liquid, and given enough centuries, gravity pulls it downward like cold honey. The claim became a staple of science classrooms, museum placards, and pop-science books. But researchers who have studied this question closely call it an urban legend.
The thickness variation in medieval windows has a much simpler explanation rooted in how old glass was made. Before modern float-glass manufacturing, windowpanes were produced using methods like the crown process or the cylinder process, both of which yielded sheets of uneven thickness. In the crown process, a glassblower spun a blob of molten glass into a flat disc, which naturally ended up thicker near the center. In the cylinder method, a blown cylinder was cut open and flattened, again producing slight thickness irregularities. When glaziers installed these imperfect panes, they tended to place the heavier edge at the bottom for stability, though not always. Some medieval windows are actually thicker at the top or along one side, which would be impossible if gravity-driven flow were the cause.
Archaeological work on ancient window glass confirms the manufacturing explanation. Examination of over 350 panes of eighth-century Byzantine window glass from Sardis, Turkey, showed that the glass was most consistent with the cylinder method, though with some notable differences from later descriptions of the technique.1Cambridge University Press (MRS Advances). Reverse Engineering Eighth Century C.E. Window Glass Processing at Sardis, Turkey These panes showed variations in thickness, edge shape, and bubble alignment that were clearly artifacts of the forming process, not of centuries of slow deformation. The unevenness was baked in from the start.
Why the “Slow Liquid” Idea Feels Plausible
Glass is not a crystal. In a crystal, atoms sit in a neat, repeating lattice. In glass, atoms are arranged more like they are in a liquid: disordered and without a long-range repeating pattern. This is why glass is classified as an amorphous solid. Because its atomic structure resembles a frozen snapshot of a liquid, it is tempting to conclude that glass is just a very, very viscous liquid that never quite finished solidifying. That intuition is not entirely wrong in a technical sense, but the practical conclusion people draw from it, that glass will flow under gravity at room temperature, is wildly off.
The key distinction is between structure and behavior. Yes, the atomic arrangement in glass looks liquid-like. Studies have confirmed that for most glasses, the arrangement of nearest-neighbor atoms is similar to that found in at least one crystalline form of the same material, but the long-range order is missing.2Journal of Physics C: Solid State Physics. On the structure of simple inorganic amorphous solids More recently, researchers have been able to determine the full three-dimensional atomic structure of an amorphous solid for the first time, revealing that while short-range atomic packing is geometrically disordered, some of these structures connect to form crystal-like superclusters that create medium-range order.3Nature. Determining the three-dimensional atomic structure of an amorphous solid Glass has more internal organization than people often assume. And regardless of its structure, the question of whether it flows comes down to one thing: viscosity.
The Viscosity Numbers That Settle the Question
Viscosity measures a material’s resistance to flow. Water has a viscosity of about 0.001 Pascal-seconds. Honey is roughly 2 to 10 Pa·s. The threshold where a glass “transitions” from a supercooled liquid into what we functionally call a solid, called the glass transition temperature, corresponds to a viscosity in the neighborhood of 1011 Pa·s. At that point, the material resists deformation so strongly that it behaves as a rigid solid for all practical purposes.4Chemical Geology. The glass-transition, structural relaxation and shear viscosity of silicate melts
But room temperature is far below the glass transition temperature for silica glass. At room temperature, the viscosity of window glass is not 1011 Pa·s. It is not 1020. Molecular dynamics modeling has estimated the upper-bound viscosity of amorphous silica at room temperature to be around 1032 Pa·s.5Journal of the American Ceramic Society. Molecular dynamics study on the viscosity of glass‐forming systems near and below the glass transition temperature That number is so large it loses all intuitive meaning, so here is one way to think about it: if you took a pane of cathedral glass and waited for gravity to cause a measurable sagging at that viscosity, you would need to wait far longer than the roughly 14-billion-year age of the universe. The glass would not noticeably flow in a trillion years. The cathedral would crumble to dust, the stone would erode, the continent would drift and be subducted, and the glass, if somehow preserved, would still look the same.
But Doesn’t Something Happen to Glass Over Time?
This is where the story gets more nuanced, because glass is not perfectly frozen. Even well below its glass transition temperature, glass undergoes what physicists call structural relaxation, a slow process where atoms gradually shift toward a more stable arrangement. This is not the same as flowing under gravity. It is more like a material quietly settling into a slightly lower energy state over long timescales.
Atomistic simulations have accessed these long-term dynamics at room temperature and found that the energy relaxation in glass follows a stretched exponential decay pattern.6PubMed. Stretched Exponential Relaxation of Glasses at Low Temperature In plain terms, the internal rearrangements happen quickly at first and then slow down enormously as the glass settles. Extremely long molecular dynamics simulations, run to microsecond timescales that bridge part of the gap between computational models and real-world experiments, have confirmed that the principles governing this relaxation hold across a wide range of temperatures.7PubMed. Cooling rate and stress relaxation in silica melts and glasses via microsecond molecular dynamics
There is even some experimental evidence of time-dependent behavior in silica glass at room temperature. Researchers studying a pair of fused silica reference plates documented deformation that occurred over years and attributed it to a viscous relaxation process with a time constant on the order of a decade.8PubMed. Relaxation time and viscosity of fused silica glass at room temperature This finding is intriguing, but the deformations involved are extremely small, measurable only with precision instruments. They fall in the territory of internal stress relief rather than gravitational flow, and they are not the kind of change that would make a windowpane thicker at the bottom.
The way a glass was cooled also influences its long-term behavior. Glasses cooled very quickly from a melt are frozen further from equilibrium than glasses cooled slowly, so they have more “room” to relax over time. Researchers have used calorimetry to quantify the stability of glasses by measuring a temperature at which the glass and the hypothetical supercooled liquid would have the same energy, and have tracked how heating at different rates can keep a glass trapped in its initial state over a temperature range spanning up to 75 degrees.9Scientific Reports. Relaxation dynamics of glasses along a wide stability and temperature range A glass made with careful, slow cooling is more stable and less prone to any aging effects than one that was quenched rapidly.
Metallic Glasses Behave Differently
Not all glasses are silica windows. Metallic glasses, made from metal alloys cooled fast enough to prevent crystallization, have their own distinct relaxation behavior at room temperature. Research has shown that aging in various metallic glasses at room temperature follows a universal stretched exponential decay, but with a characteristic exponent that is different from what is seen in supercooled liquids or in silica glass simulations.10Nature Communications. Distinct relaxation mechanism at room temperature in metallic glass This suggests that the atomic-level mechanism driving relaxation in metallic glasses below their glass transition temperature is fundamentally different from the process that governs flow in a liquid.
Why does this matter for the “does glass move” question? Because it shows that “glass” is not one thing. Different glass compositions, formed under different conditions, relax through different mechanisms at different rates. Silica-based window glass is among the most stable glasses known. Metallic glasses and some organic polymeric glasses are considerably less stable and can show measurable changes in properties like hardness or density over months or years at room temperature. But even in these less stable systems, the changes involve subtle internal rearrangements, not macroscopic flow.
Nature’s Own Long-Term Stability Test
If you want a natural experiment testing whether silica glass holds together over geological time, tektites provide a compelling answer. Tektites are natural glasses formed from terrestrial rock melted by meteorite impacts and flung through the atmosphere, where they cool rapidly. Some tektites are tens of millions of years old. Despite their age, they remain glassy, showing remarkable long-term stability against devitrification, which is the process of a glass crystallizing back into an ordered solid.11Planetary and Space Science. Glass chemistry of tektites They also have very low water content and high chemical resistance. If silica glass were prone to significant structural change over mere centuries, tektites millions of years old would have long since crystallized or deformed. They have not.
Obsidian, a volcanic glass, tells a similar story. Archaeological obsidian tools thousands of years old retain their sharp edges and original shapes. The surface may develop a thin hydration layer where water slowly diffuses into the glass, and this process is actually used as a dating method, but the bulk material does not flow or deform.
Glass in Precision Engineering
The practical implications of glass stability matter enormously in fields where dimensional precision is critical. Telescope mirrors, interferometric reference standards, and satellite optics all rely on glass or glass-ceramic substrates that must maintain their shape to within fractions of a wavelength of light over years of service. If glass flowed, even microscopically, these instruments would quickly lose their precision.
Measurements conducted over a multi-year period by the National Bureau of Standards on four low-thermal-expansion mirror materials, including vitreous silica and glass-ceramics, found average length changes on the order of fractions of a part per million.12PubMed Central. Precision Measurements of the Dimensional Stability of Four Mirror Materials These changes are minuscule and are attributed to structural relaxation and thermal history effects rather than flow. The fact that space agencies routinely rely on glass substrates for mirrors expected to function for years in orbit is itself a strong statement about the material’s dimensional stability.
Pitch Is Not Glass
One source of confusion is the famous pitch drop experiment at the University of Queensland, set up in 1927 to demonstrate that pitch, a tar-like substance, appears solid but actually flows extremely slowly under gravity.13IOP Publishing. The pitch drop experiment A funnel of pitch has been dripping at a rate of roughly one drop per decade. People sometimes point to this experiment as evidence that glass could do the same thing, just more slowly. But pitch and window glass are entirely different materials with vastly different viscosities. Pitch has a viscosity roughly 100 billion times that of water, which is high enough to seem solid on short timescales but low enough to flow visibly over years. Window glass at room temperature has a viscosity roughly 1020 times higher than pitch. The comparison does not hold.
This conflation between pitch and glass is understandable because both are sometimes loosely described as “amorphous” or “non-crystalline.” But amorphous structure alone does not determine whether something flows at room temperature. The chemical bonds, composition, and resulting viscosity are what matter. Silica glass has strong covalent silicon-oxygen bonds arranged in a rigid network. Pitch is a complex mixture of organic molecules held together by much weaker forces. Calling both of them “glasses” and then concluding they behave similarly is like calling both a glacier and a river “water” and expecting them to move at the same speed.
Where the Science Gets Genuinely Interesting
The reason physicists keep studying glass relaxation is not because they think your windows are melting. The glass transition itself remains one of the deepest unsolved problems in condensed matter physics. When you cool a liquid fast enough, it bypasses crystallization and becomes a glass, but exactly why and how this happens, and whether there is a true thermodynamic phase transition hiding underneath the kinetic one, are questions researchers have debated for decades.
The subtle relaxation processes that glass undergoes below its transition temperature offer clues about this deeper puzzle. The fact that metallic glasses and silica glasses relax with different characteristic exponents, for example, suggests that there may be distinct classes of aging dynamics rather than one universal mechanism. The stretched exponential decay patterns observed in simulations and experiments are signatures that theorists try to connect to models of how atoms in a disordered solid explore their energy landscape over time.
Recent advances in imaging have opened new windows into this problem. The first-ever determination of the full three-dimensional atomic structure of an amorphous solid revealed something surprising: even within the disorder, crystal-like superclusters form at a medium-range scale, showing translational order but not orientational order.3Nature. Determining the three-dimensional atomic structure of an amorphous solid The coexistence of four different crystal-like medium-range order types within one amorphous sample suggests that glass is far more structurally rich than the old “frozen liquid” picture implies. Understanding this hidden internal architecture could eventually explain why some glasses are extraordinarily stable while others age more rapidly, and it may have practical consequences for designing glasses with specific long-term properties for technologies from fiber optics to pharmaceutical storage.
Common Misconceptions Worth Clearing Up
A few related misunderstandings circulate alongside the cathedral window myth:
- Glass is a liquid: Glass is an amorphous solid. It lacks the long-range periodic order of a crystal, but it has the mechanical rigidity and resistance to flow that define a solid. Some physicists describe it as a “non-equilibrium solid” to acknowledge that it is not in its lowest-energy crystalline state, but calling it a liquid is misleading in any practical sense.
- Old glass looks wavy because it flowed: The waviness in antique glass comes from the same manufacturing imperfections that cause thickness variations. Modern float glass, made by pouring molten glass onto a bed of molten tin, is optically flat. Old glass was not.
- Roman glass artifacts prove flow: Ancient Roman glass vessels and beads survive in remarkable condition precisely because glass does not flow. Surface weathering and iridescence from chemical interaction with soil moisture are common, but the shapes remain intact after two millennia.
- If glass is a solid, it should shatter like a crystal: Glass does shatter, often spectacularly. Its lack of crystal planes means cracks can propagate in any direction, which is why broken glass produces sharp, irregular shards rather than cleaving along flat faces. The brittleness is itself evidence that glass behaves as a rigid solid, not a viscous liquid.
Temperature Changes the Picture Dramatically
Everything above applies to glass at or near room temperature. Heat glass up and the situation changes rapidly. Viscosity drops exponentially as temperature rises, and near the glass transition temperature, the material becomes soft and workable. This is exactly what glassblowers exploit: by heating silica glass to around 1,000°C or higher, they reduce its viscosity enough to shape it by blowing and pulling. The relationship between temperature and viscosity in glass-forming melts is well studied, and the transition from rigid solid to workable fluid happens over a relatively narrow temperature window.4Chemical Geology. The glass-transition, structural relaxation and shear viscosity of silicate melts
This temperature sensitivity is why some industrial glasses do need to be carefully annealed, or held at a controlled temperature and cooled slowly, to relieve internal stresses locked in during manufacturing. If those stresses are not relieved, the glass can be more prone to cracking. But even a poorly annealed window sitting in a building for centuries will not flow. The residual stresses might eventually cause it to crack, but they will not cause it to sag. The energy involved is nowhere near enough to drive bulk deformation at room temperature, where the viscosity is simply too enormous for gravity to overcome.