Frame and Skin

For five thousand years the grammar of building was the grammar of the wall. Stone, brick and earth are strong when pressed and weak when pulled, so every structure in them had to be arranged to keep its materials in compression. A wall carried its own weight and the weight of the floors and roof, and so it grew thicker toward the ground; an opening in it had to be bridged by a lintel, an arch or a vault; a wide room needed buttresses to hold the thrust of its vault. The earlier chapters of this part are, in a sense, a catalogue of the ingenious ways builders found to make heavy materials stand up.

In the nineteenth century three materials that are strong in tension as well as compression became cheap enough to build with: iron, then steel, then reinforced concrete. Their effect was to divide the old wall into two separate things. A frame of columns and beams took over the work of carrying; a skin of glass, brick, stone or metal panels was left only to keep out the weather. Once the two were separated, each could be designed for its own job, and almost every rule of the wall could be revised. This chapter follows that division from the iron sheds of the 1850s to the shells and tents of the twentieth century.

Iron and glass

Iron entered architecture by the back door, through bridges, mills and glasshouses. The cast-iron bridge at Coalbrookdale (1779) still copied the form of a masonry arch. Cast iron is excellent in compression but brittle in tension, so it made good columns; wrought iron, and later steel, could be rolled into beams and drawn into ties. Textile mills of the 1790s already had iron columns and beams inside their brick walls to reduce the risk of fire. The great conservatories (the Palm House at Kew, 1844–48, is the finest survivor) showed that iron ribs and small panes of glass could enclose a large space with almost no solid wall at all.

The decisive building was Joseph Paxton’s Crystal Palace for the Great Exhibition in London (1851). Paxton was a gardener who had built glasshouses for the Duke of Devonshire. He designed the exhibition hall as a kit: every column, girder, gutter, sash bar and pane belonged to a small set of standard parts, made in factories, delivered by rail and bolted together on a grid: an 8-foot unit set by the length of the glass panes, multiplied into bays of 24 feet (7.32 m). The building, some 560 m long, was erected in a few months and dismantled afterward to be rebuilt at Sydenham. Its roofs were of Paxton’s ridge-and-furrow glazing, and its transept was covered by a timber-ribbed barrel vault high enough to enclose a row of elm trees.

The system came from Paxton’s glasshouses at Chatsworth, where he had used ridge-and-furrow glazing on the Great Conservatory (1836–41) and on the Lily House (1849–50), and the whole building was set by the size of a pane: panes 49 inches long gave roof ridges 8 feet apart, and three ridges made the 24-foot bay (The Crystal Palace grid of 24 feet). Because it was a kit it could be moved. In 1852–54 it was re-erected, larger and with a barrel-vaulted nave and three transepts, on Sydenham Hill in south London, above terraces and fountains that Paxton laid out in the Italian manner; it stood there until it burned down in 1936. Its gardens are described in Gardens and Landscape.

The Crystal Palace was admired by engineers and deplored by many architects, who saw in it no architecture at all. Viollet-le-Duc drew the opposite conclusion: a new material demanded new forms, and iron should be used as iron rather than disguised as stone (Use iron as iron). In Paris, Henri Labrouste had already exposed a slender iron vault on cast-iron columns in the reading room of the Bibliothèque Sainte-Geneviève (1843–50). By 1889 the Eiffel Tower and the Galerie des Machines (a hall spanning about 110 m on steel arches hinged at their feet and crown) showed that the engineers had outrun the architects entirely.

℟Use iron as iron

A new material calls for new forms: iron should be used in slender members that show its strength in tension and bending, not cast in imitation of stone columns and vaults.

The steel skeleton

Two inventions turned iron and glass into the skyscraper. The Bessemer process (1856) made steel cheap enough for building, and Elisha Otis’s safety elevator (demonstrated in 1854) made upper floors as valuable as lower ones. In the centre of Chicago, rebuilt after the fire of 1871, land prices did the rest. The Monadnock Building (1889–91) shows the limit of the old way: sixteen storeys of load-bearing brick, with walls about 1.8 m thick at the base. The Home Insurance Building (1884–85) by William Le Baron Jenney had shown the new one: an iron and steel frame that carried the floors and much of the outer walls. Within a decade the steel frame carried everything, fireproofed with hollow terracotta, and the outer wall was reduced to cladding hung on the frame storey by storey (The skeleton carries; the wall only encloses).

The consequences are easy to state. A frame wall does not thicken as it rises, so a building can grow as tall as its elevators, its wind bracing and its foundations allow. The outer wall can be opened almost entirely: Chicago offices had wide Chicago windows of three panes, and the Reliance Building (1890–95) is little more than glass in a thin terracotta-clad cage. And because every office floor is the same, the façade becomes a repetition. Louis Sullivan’s answer to the question of how such a building should look (a base, a shaft of identical floors and a crowning attic) is the subject of The tall building has three parts and of the chapter on the modern movement.

Reinforced concrete

Concrete is an artificial stone: excellent in compression, useless in tension. Embed steel bars where the tension falls and the combination behaves like a material of both kinds. The idea was patented piecemeal (the Parisian gardener Joseph Monier reinforced flowerpots with wire from 1867) and systematised in the 1890s by François Hennebique, whose monolithic system of columns, beams and slabs cast together was licensed all over Europe. Auguste Perret’s block of flats at 25 bis rue Franklin in Paris (1903–04) showed its concrete frame on the façade, filled in with panels of ceramic tile.

Reinforced concrete has one property that steel lacks: it is poured, so it is naturally continuous. Beams, slabs and columns are one piece, and a slab can run past its last column as a cantilever. In 1914 the young Le Corbusier drew the consequence in a diagram he called the Dom-ino: three flat slabs, six columns set back from their edges, a stair, and nothing else (The Dom-ino frame). Everything that makes a house (walls, windows, partitions) becomes infill that can be placed freely. The five points of 1926 (pilotis, free plan, free façade, ribbon window and roof garden) are nothing but the Dom-ino’s implications spelled out.

Compare the load-bearing house. Its walls must stand on the walls below, so its rooms repeat from floor to floor; its windows must leave solid piers between them, so they are tall and narrow; its roof must shed water, so it is pitched. In the concrete frame all three constraints disappear. The ribbon window in particular is a demonstration of the new freedom: it can run unbroken from one end of the façade to the other only because the façade carries nothing (The ribbon window).

The curtain wall

If the skin carries nothing, it need not be masonry at all. The curtain wall is an outer wall of glass and metal hung from the edges of the floor slabs, carrying only its own weight and the wind. Gropius and Meyer’s Fagus factory (1911–13) wrapped its corners in glass with no corner pier; the workshop wing of the Bauhaus at Dessau (1925–26) did the same over three storeys; Mies van der Rohe drew glass skyscrapers in 1921–22 that were nothing but a frame of slabs wrapped in a crystalline skin. After 1945, extruded aluminium, float glass and sealants turned the dream into an industry. Lever House (1952) and the Seagram Building (1958) in New York set the pattern for the corporate tower everywhere.

A curtain wall is organised by a module, typically around 1.5 m, fixed by the width of a sheet of glass and the planning grid of the offices behind. Vertical mullions run the full height; at each floor an opaque spandrel panel hides the slab edge and the ceiling void; between them runs the vision glass. At Seagram, bronze I-sections are applied to the mullions: they are not structural, but Mies added them to express, on the outside, the steel frame that fire regulations required him to encase in concrete.

The glass box brought its own problems. A wall of glass admits the sun’s heat in summer and loses heat in winter, and the sealed curtain-wall tower became dependent on air-conditioning. Le Corbusier’s answer was the brise-soleil, a deep concrete grid in front of the glass; later answers include tinted and coated glass, external blinds and louvres, and double-skin façades whose cavity is ventilated. The skin, freed from structure, turned out to have a job of its own (the management of light, heat and air), which is the subject of the last chapter of Part III.

℟The curtain wall

Hang the outer wall from the edges of the floor slabs as a continuous skin of glass and metal, independent of the frame behind it, and express its grid of mullions.

skin = vision glass + spandrel panel per floor, on a module of ≈ 1.5 m

Form-active structures: arches, shells and tents

The frame solves the problem of height; it is less efficient for wide spans, where beams and slabs must be deep to resist bending. The alternative is to let the form of the structure do the work: to shape it so that it carries its load in pure compression or pure tension, with no bending at all. The oldest statement of the principle is Robert Hooke’s anagram of 1676: as a flexible chain hangs, so, inverted, will a rigid arch stand (As hangs the chain, so stands the arch). A hanging chain can only pull, so the curve it takes, the catenary, is the path of its weight in pure tension; turned over, it is the ideal shape of an arch.

Every masonry arch has a thrust line, the path of compression through it, and the arch stands as long as that line stays within its thickness. The figure computes the line for three arches of equal span carrying their own weight. In the catenary it coincides with the centre line; in the parabola it wanders a little; in the semicircle it escapes near the haunches unless the ring is made thick (The line of thrust must stay inside the arch). Antoni Gaudí designed with this principle, building upside-down models of strings and small bags of shot and reading them in a mirror.

Reinforced concrete allowed the same logic in two dimensions. A shell is a thin curved slab that carries load by forces within its surface, like an eggshell (Strength from form). The engineers of the mid-century built shells of astonishing thinness: Eduardo Torroja’s cantilevered canopies at the Zarzuela racecourse in Madrid (1935–41), Pier Luigi Nervi’s ribbed domes in Rome, and above all the work of Félix Candela in Mexico, who built hundreds of shells from the hyperbolic paraboloid. This saddle-shaped surface is doubly curved, which makes it stiff, yet it contains two families of straight lines, so its formwork can be made of straight boards (The hyperbolic paraboloid is made of straight lines). Candela combined hypars into umbrellas, groined vaults, and the eight lobes of the restaurant Los Manantiales at Xochimilco (1958), about 42 m across and some 4 cm thick.

At the other extreme are structures in pure tension. A cable or a fabric uses its material with complete efficiency, but it has no stiffness of its own: it will flap in the wind unless it is prestressed into an anticlastic shape, curved upward in one direction and downward in the other (A membrane must be a saddle). Frei Otto, who founded the Institute for Lightweight Structures at Stuttgart in 1964, found such forms with soap films and stretched nets before calculating them, a method now called form-finding. His cable-net roofs for the Munich Olympic Games (1972, with Behnisch & Partner) join dozens of saddles into a single transparent landscape.

Buckminster Fuller’s geodesic dome reaches the same economy by a third route: a sphere, the shape that encloses the most volume for the least surface, framed by a net of short struts in triangles, so that loads spread through the whole shell and no member works much harder than its neighbours (Enclose the most with the least: the geodesic dome). The frequency of the dome (the number of divisions of each edge of the original icosahedron) trades rounder form and shorter struts against a larger number of different strut lengths.

The new grammar

The materials of the industrial age did not abolish the older rules; they changed which rules apply where. A masonry wall still thickens toward its foot, an arch still needs its thrust line inside it, and a roof still has to shed water. But the builder now chooses among several structural principles (compression, bending, tension, and form) and among materials that suit each of them. The table summarises the choice.

Above all, the division of frame and skin made the outer wall an independent subject of design. Everything that follows in the history of modern architecture (the glass tower, the brise-soleil, the colour-coded façade of services, the parametric screen and the insulated envelope of the passive house) is an answer to the question that the frame first made possible: if the wall carries nothing, what is it for?