The Textbook of Architecture

Part I · Chapter 2

Walls and Openings

Masonry, bond and the hole in the wall

PlateA Roman wall of opus reticulatum with brick quoins and lacing courses, cut open at the end to show its concrete core. Choose other walls from the menu. Open in Models ↗

Before there were columns there were walls. A wall is the simplest thing an architect can make and the hardest to make well: it must carry its own weight and whatever rests on it, keep out wind, rain and enemies, divide inside from outside, and still let people, light and air pass through it. Every tradition has had to answer the same questions: what the wall is made of, how its pieces are bound together so that it acts as one body, and how a hole (a door, a window) can be cut in it without the wall above falling in.

This chapter follows those questions through the building traditions of Greece and Rome, where they were first written down, and on into the brickwork of later Europe. The answers turn out to depend on a handful of rules that have hardly changed in three thousand years.

What a wall has to do

A masonry wall works in compression. Each stone or brick presses on those below it, and the load travels down to the foundation through the contact faces. As long as the line of that pressure stays within the thickness of the wall, the wall stands, however little its mortar may hold; if the line wanders outside, because the wall is too thin for its height, or is pushed sideways by a roof or a vault, the wall overturns or bulges. Thickness is therefore the first rule of masonry, and the old rules of thumb that tie the thickness of a wall to its height (a tenth, a twelfth, more for walls that carry vaults) are rules about keeping the line of pressure safely inside.

The second rule is that the wall must behave as one piece. A heap of stones has no strength against a sideways push; a wall has, because every stone is locked to its neighbours. How they are locked (by their weight and fit alone, by overlapping in bond, by mortar, by clamps, by stones running through the whole thickness) is what distinguishes one kind of masonry from another.

Stone laid dry

The oldest great walls of the Mediterranean are made of huge, roughly shaped blocks with smaller stones packed into the gaps: the Cyclopean masonry of Mycenae and Tiryns (see Cyclopean walls). A more refined answer, used by Greek builders for terraces and fortifications and later brought to perfection by the Inca, is polygonal masonry: each block cut to an irregular many-sided outline and fitted to its neighbours so closely that no gap remains. Because the joints run in every direction, no single joint can become a line of weakness.

Fig. 1Polygonal masonry: irregular blocks fitted without mortar. Open in Models ↗
Fig. 2Isodomic ashlar: equal courses, each joint over the middle of a stone below. Open in Models ↗

For their temples the Greeks wanted something more regular. Their walls are ashlar (squared stone laid in horizontal courses), and Vitruvius distinguishes two kinds. In isodomum every course has the same height; in pseudisodomum the heights vary, usually tall courses alternating with thin ones. The joints were astonishingly fine. To achieve them with reasonable labour the masons used anathyrosis: only a narrow band round the edge of each joint face was dressed to a perfect plane, the middle being slightly hollowed so that it did not touch. The blocks were laid without mortar and tied together with iron clamps and dowels set in molten lead.

Bond: never a joint over a joint

Whatever the material, one rule governs every coursed wall: the vertical joints of one course must not line up with those of the course below. Each stone should rest on two, and each joint should be covered by the middle of a stone above. Vitruvius, describing the Greek method, speaks of binding the upright joints by interchanging the direction of the stones: some laid along the wall, some across it. In the language of later masons, those laid lengthwise are stretchers and those laid across are headers.

Thick walls raise a second problem. A wall two or three stones thick can easily become two thin walls standing side by side, with a filling between that settles away from them. The Greek remedy, which Vitruvius calls emplekton, was to lay stones across the thickness at intervals and to set single through-stones from one face right through to the other. He complains that Roman workmen, in their hurry, built only the two faces carefully and threw broken stone and mortar into the middle anyhow, making three separate walls where there should be one.

Fig. 3Emplekton, cut open at the end: headers and through-stones (red) tie the two faces across a rubble fill. Open in Models ↗

The Roman wall: facing and core

The Romans changed the nature of the wall. From the second century BC they built with opus caementicium (Roman concrete), in which fist-sized pieces of stone, tuff or broken brick were laid in thick beds of lime mortar. Made with pozzolana, the volcanic ash of the Bay of Naples, the mortar set hard even in wet ground and under the sea. A concrete wall is not a pile of units but a single artificial rock, and it can be poured into any shape: curved, stepped, pierced, vaulted over.

Concrete needs formwork, and the Romans used the faces of the wall itself as formwork that stayed in place. The facings changed with fashion. Vitruvius knew the older opus incertum, a skin of small irregular stones, and the newer opus reticulatum, in which small tufa blocks with square faces are set diagonally so that their joints form a net. He judged the first stronger and the second handsomer.

Of these, the reticulatum looks better, but its construction makes it likely to crack.

Vitruvius, II.8.1 (tr. Morgan, 1914)
Fig. 4Opus incertum: small irregular stones over a concrete core. Open in Models ↗
Fig. 5Opus reticulatum: a net of tufa pyramids, with brick at the corners. Open in Models ↗

Under the emperors the standard facing became fired brick, opus testaceum. The bricks were often triangles, cut from square tiles, laid with their long edge on the face and their points driven into the core; every metre or so a course of large square tiles, bipedales, ran right through the wall and bound facings and core together: the Roman answer to Vitruvius’s complaint. Reticulatum panels framed in brick made opus mixtum; large blocks of squared tufa laid as headers and stretchers made opus quadratum. Nearly all of it was meant to be covered with stucco or marble veneer.

Fig. 6Opus testaceum: brick facings, a bonding course of bipedales and the concrete core (red). Open in Models ↗
Facing and core

Build a Roman wall as two facings of brick or small stone with a core of concrete poured between them in layers; tie the facings into the core with bonding courses of large tiles laid through the whole thickness at intervals.

bonding courses of bipedales (2 ft tiles) every 1–1.5 m of height

Vitruvius adds two practical observations that say much about Roman building. When arbitrators valued a party wall of rubble for sale, he says, they deducted one eightieth of its cost for every year it had stood (treating such walls as good for about eighty years), whereas a sound brick wall still standing plumb was valued at its full cost. And because the law limited walls on public frontages to a foot and a half in thickness, the tall apartment blocks (insulae) of the crowded capital had to be built with stone piers and burnt brick if they were to rise storey upon storey.

Brick and its bonds

Brick is the most adaptable of masonry units: small enough to be laid by one hand, regular enough to be bonded by rule. Vitruvius describes the Roman brick of a foot and a half by a foot and two Greek sizes, of five and four palms square, and a wall laid with whole bricks on one face and half-bricks on the other in alternate courses, so that the joints always break. Later European brickwork standardised the brick so that two widths plus a joint equal one length, which makes bonding systematic.

In a wall one brick thick, stretchers show only one face; headers run through and tie the wall together. English bond alternates a whole course of headers with a whole course of stretchers; Flemish bond alternates header and stretcher within every course, each header centred on the stretcher below. English bond is slightly stronger, Flemish bond more even in appearance; both achieve the quarter-brick lap that keeps every joint covered.

Fig. 7English bond: courses of headers and of stretchers alternate. Open in Models ↗
Fig. 8Flemish bond: headers and stretchers alternate in every course. Open in Models ↗

The hole in the wall

Every opening is a wound. The masonry above it must be carried across the gap, and the load that would have gone straight down must be turned aside to the jambs. The simplest way is a lintel, a single beam of stone or timber, whose span is limited by its depth and by the weakness of stone in bending (see A stone lintel’s span is limited by its tension in the chapter on post and lintel). Builders have long relieved a lintel by leaving a triangular or arched gap above it so that the weight of the wall passes round it (Relieve the lintel with a triangle). The true answer, the arch, turns the load into compression; it is the subject of the next chapter.

Doors and windows also weaken a wall by cutting its courses. Their edges must be built of larger, well-bonded stones or bricks (jambs and quoins), and they should not be crowded close to corners, where the wall needs all its strength, nor set one above another so tightly that only thin piers remain between them. The Roman builders’ habit of framing openings with brick relieving arches embedded in the concrete, visible today wherever the facing has fallen away, is a quiet demonstration of the principle.

The Vitruvian doorway

Vitruvius gives the proportions of temple doorways in the same modular spirit as the orders. The door is sized from the room it serves: the height from floor to ceiling is divided into three and a half parts, and two and a half of them give the height of the opening. Its width then follows from its height (five and a half twelfths for a Doric door, two fifths for an Ionic), and the opening is made a little narrower at the top than at the bottom, by a third, a quarter or an eighth of the width of the jamb depending on how tall it is. Only the tallest doors keep their sides upright.

Fig. 9The Vitruvian doorway: change the height of the room and the whole door rescales; the taper disappears above 30 ft. Open in Models ↗
The door from the height of the room

Divide the height from floor to ceiling into three and a half parts and give two and a half to the opening; make a Doric door five and a half twelfths as wide as it is high (an Ionic one two-fifths), and narrow it slightly at the top.

h = 2½⁄3½ H; Doric w = 5½⁄12 h, Ionic w = h ÷ 2½; top narrowed by ⅓, ¼ or ⅛ of the jamb as h ≤ 16, ≤ 25, ≤ 30 ft

The frame follows by the same method. The jambs (antepagmenta) of an Ionic door are a fourteenth of the height of the opening and are divided into three fasciae like an Ionic architrave; the lintel is as high as the jamb is wide at the top; above it come a frieze and a projecting cornice, and an Ionic doorway adds consoles hanging at the sides. Even the folding doors are proportioned: stiles a twelfth of the width, panels three twelfths, and the rails dividing the height two parts above and three below. The taper, familiar from the great north door of the Erechtheion in Athens, gives the doorway something of the stability of a pyramid and echoes the inward lean of the temple’s columns.

Fig. 10An Ionic doorway: narrower in proportion, jambs with three fasciae, consoles at the sides of the cornice. Open in Models ↗

Windows and light

The Greek temple had almost no windows; its cella was lit through the great door and by lamps. Roman houses looked inward to atrium and peristyle and turned small, high openings to the street. Glass was known (small cast panes, or sheets of mica and selenite in the richest buildings), but a Roman window was primarily a source of light and air, placed where it did least harm to the wall.

Large buildings solved the problem of light in section rather than on the façade. The Roman and Christian basilica raised its central nave above the side aisles and pierced the upper walls with a row of windows, the clerestory, so that light fell into the middle of the hall. The Pantheon took the idea to its limit with a single oculus at the crown of its dome. Hagia Sophia ringed the base of its dome with forty windows, so that the shell seems to float on light.

Fig. 11A basilica in section: the clerestory (red) lights the nave above the lean-to roofs of the aisles. Open in Models ↗

In the medieval centuries the balance between wall and opening shifted again. Romanesque builders kept thick walls and small windows; the builders of the Gothic cathedrals concentrated the loads on piers and flying buttresses and let the wall between dissolve into glass. The modern frame completed the process by freeing the wall from load altogether. But wherever masonry still carries a building, the rules of this chapter still apply: thickness, bond, a stone over every joint, and a strong frame round every hole.

  • Keep the wall thick enough for its height and for any sideways push it receives.
  • Break joints: every stone or brick should rest on two below.
  • Tie the faces of a thick wall together with headers and through-stones.
  • In Roman concrete construction, bind facings and core with courses running through the wall.
  • Proportion openings from the room they serve, and frame them with larger, well-bonded pieces.

Cite this page Chris Paton, “Walls and Openings”, The Textbook of Architecture, https://textbookofarchitecture.com/chapter/walls-and-openings. Text, drawings and models licensed under CC BY-NC 4.0.