The Forgotten History of Plate Glass: How Flat Transparent Walls Reshaped Modern Architecture
The window in your office wall is one of the most overlooked technological achievements of the 20th century. A single sheet of glass several meters across, optically clear, mechanically flat to within micrometers, manufactured by the millions of square meters and installed at every scale from elevator to skyscraper. For most of human history this object did not exist, and the architectures that depend on it (modernist curtain walls, retail storefronts, automobile windshields, solar panels, smartphone screens) were not possible. The achievement is so complete that the underlying technology is no longer remarked on, but the historical arc that produced it is one of the strangest in industrial chemistry.
The pre-flat-glass world
Glass has been made for roughly 5500 years (the earliest known glass beads come from Mesopotamia around 3500 BCE), but for the first 5000 of those years it was made small. The available techniques were core-forming (winding molten glass around a removable clay core to make small vessels), casting (pouring molten glass into shaped molds), and blowing (inflating a gob of molten glass on the end of a pipe). All three techniques produced small objects: beads, vessels, decorative items, eventually small windowpanes.
Roman glassmakers around the 1st century CE developed the first practical window glass via the crown method: a blown glass globe was opened and spun rapidly on a rod, centrifugal force flattening the molten glass into a roughly circular disk. The crown technique produced flat glass up to about 50 cm in diameter, with characteristic concentric ripples and a thick lump in the center where the rod had attached. The technique persisted for over 1500 years and is the source of the small leaded-pane windows visible in medieval European buildings.
The cylinder method (also Roman, refined in medieval and Renaissance Europe) involved blowing a long cylinder of glass, cutting off the ends, slitting it lengthwise, and flattening it in a kiln. The technique produced larger flat pieces than crown glass (up to about 1 meter long) at the cost of more handling and lower optical quality. Cylinder glass dominated European architecture from roughly 1400 to 1850 and is the source of the wavy distortions visible in old building windows.
The 19th-century plate process
The first attempt at industrially-scaled flat glass was the plate-casting process developed at Saint-Gobain in France in the late 17th century and refined through the 19th. Molten glass was poured onto a flat iron table and rolled flat with an iron roller. The resulting plate was 1-2 cm thick, had a rough surface from contact with the iron, and required grinding and polishing to become optically usable. The grinding and polishing took weeks and produced enormous amounts of waste glass and abrasive sludge.
The economics were viable only for high-end applications: aristocratic mirrors (the Hall of Mirrors at Versailles, completed in 1684, was a wholesale demonstration of French plate-glass capacity), late-19th-century retail storefronts, and a small market for high-quality architectural windows. The process scaled poorly: a plate-glass works employed hundreds of workers across grinding and polishing stations, and the cost per square meter of finished glass was roughly equivalent to skilled labor for a week. Plate glass remained a luxury material into the 20th century.
The drawn-sheet process
The drawn-sheet processes developed in the early 20th century (Fourcault in 1904, Libbey-Owens in 1917, Pittsburgh in 1928) were the first to produce flat glass continuously. Molten glass was drawn upward (Fourcault) or horizontally (Libbey-Owens) through forming machinery that produced a continuous ribbon of glass at the desired thickness. The ribbon was annealed to relieve internal stresses and cut into sheets. The drawn-sheet processes eliminated most of the grinding and polishing, dramatically reducing cost.
The drawn-sheet glass was significantly cheaper than ground-and-polished plate, but the optical quality was lower. The drawing process produced characteristic vertical striations in the glass (visible as faint streaks) and surface irregularities that made the glass usable for windows but not for high-quality mirrors or precision optics. The 20th-century compromise was that buildings used drawn-sheet for windows and the expensive plate process continued for premium applications.
The Pilkington float process
The breakthrough that produced modern plate glass came from Alastair Pilkington at the British glassmaker Pilkington Brothers in 1952. The idea was elegant in retrospect: instead of forming the glass mechanically, let it float on a bath of molten tin. The molten glass would naturally spread to a uniform thickness on the tin surface (because the glass and tin would each find their own equilibrium level), the glass-tin interface would be optically smooth (because the tin is essentially flat at the atomic scale), and the upper glass surface would also be smooth (because the glass cools from the top down and surface tension produces a smooth surface). The result would be flat glass with both surfaces optically smooth, requiring no grinding or polishing.
The engineering took seven years to make work commercially. The molten glass and molten tin had to be kept at compatible temperatures; the atmosphere over the tin bath had to be a reducing mixture of nitrogen and hydrogen to prevent tin oxidation; the glass composition had to be tuned so it would not dissolve significant tin into its lower surface; the ribbon had to be cooled progressively from 1000C glass-on-tin to 600C glass-on-rollers to room temperature without inducing internal stresses; the production line had to run continuously for months at a time because starting and stopping wasted enormous amounts of expensive specialty material.
The Pilkington float plant at St Helens, Lancashire achieved commercial production in 1959. The process produced glass at roughly one-tenth the cost per square meter of the plate-casting process, with quality equal to or better than the best ground-and-polished plate. The economic impact was immediate: within a decade, the float process had displaced both plate-casting and drawn-sheet for nearly all architectural glass applications globally. By 1980 nearly all flat glass produced anywhere in the world was float glass, made under Pilkington licenses or under the various national variants developed in the 1960s-1970s.
The architectural consequences
The architectural consequences took longer to fully emerge. Modernist architects had been envisioning all-glass buildings since the 1920s; Mies van der Rohe's 1921-1922 Friedrichstrasse skyscraper proposal showed a fully glazed tower, and Bruno Taut's Glass Pavilion at the 1914 Cologne exhibition demonstrated the aesthetic. But the buildings of the early modernist period were limited by the available glass: small panes set in mullion grids, or large panes at high cost that limited deployment to flagship buildings.
The float-glass era made fully glazed facades economically routine. The post-1965 generation of modernist skyscrapers (the Lever House in New York was 1952, but the canonical Sears Tower was 1973 and the World Trade Center 1973) used float glass as standard curtain-wall material at scales that would have been impossible a decade earlier. The aesthetic of the late 20th century corporate skyscraper (the smooth glass facade, the seamless transitions between panes, the dark mirrored finish) was substantively float-glass-dependent.
The retail storefront transformation paralleled the architectural one. The plate-glass storefronts of the 19th century had been a sign of high-end commerce; the float-glass storefronts of the 1970s onward democratized the same effect down to gas stations and convenience stores. The window onto the street as a routine commercial feature is a post-1960 development.
The automotive industry shifted to curved float glass for windshields in the 1960s-1970s, replacing the smaller flat panels and external frames that had been standard. The aviation industry adopted float glass for cockpit and cabin windows. The solar industry of the 2000s-onward depends on float glass for module covers and is the largest single buyer of new float-glass capacity globally. The smartphone industry uses chemically-strengthened float glass (Corning's Gorilla Glass and competitors) for device covers.
The current production scale
Global flat-glass production in the mid-2020s is roughly 80 million tons per year, almost entirely from float plants. A modern float plant runs continuously for 12-15 years between rebuilds, producing roughly 600 tons of glass per day, with the molten-tin bath maintained at temperature for the entire plant lifetime (because cooling and restarting the tin bath would waste months of production). The continuous-production model that float requires is structurally different from the batch-production model of every glass-making technique that preceded it; the economic model and the engineering model are inseparable.
The major float producers (Saint-Gobain, NSG/Pilkington, Asahi/AGC, Guardian, CSG, Xinyi) operate roughly 200 float lines globally with substantial geographic concentration in China (about half) and the rest split across Europe, North America, India, and Southeast Asia. The product is essentially commoditized; the differentiation is in coatings (low-emissivity coatings for energy efficiency, anti-reflective coatings for solar applications, electrochromic coatings for switchable windows) rather than in the float glass itself.
Three observations
The first observation is that the modern flat glass industry is essentially the result of one engineering achievement in 1959. The plate-glass tradition that preceded it was distinct technology (mechanical forming followed by grinding and polishing); the drawn-sheet processes were distinct technology (continuous mechanical drawing). Float glass is a different physical principle (liquid-on-liquid forming) that happens to produce a product superficially similar to its predecessors. The displacement was total: there are essentially no plate-casting lines or drawn-sheet lines running anywhere in the world in 2026.
The second observation is the speed of the displacement. From the 1959 commercial Pilkington plant to global float dominance was approximately two decades. The pattern of one engineering breakthrough producing total displacement of the prior technology within a generation recurs across many industrial chemistries (Haber-Bosch ammonia synthesis, the Frasch sulfur process, the Solvay soda process) but the pattern is not visible from outside the industry because the public sees only the resulting product.
The third observation is that the architecture that depends on float glass is so much a part of the visual landscape of modern cities that it is essentially invisible as a technological achievement. The 1950s photograph of a streetscape and the 2020s photograph of the same streetscape differ enormously in the amount of glass visible; the change is one of the most dramatic in 20th-century material culture, and almost nobody notices because the underlying material is itself transparent.
The deeper observation is that some industrial chemistries have such a tight match to a particular fabrication problem that, once discovered, the chemistry collapses an entire prior industry into nothing. The float process was not an improvement on plate-casting; it was a completely different way to make glass that happened to be cheaper and better at the same time. The history of materials science contains a relatively small catalog of such collapses (Bessemer steel, Haber-Bosch ammonia, Pilkington float glass, polymer thermosets, semiconductor wafers) and each one quietly reshaped a piece of human material life within a generation or two.