The Islamic World

In the space of a single century after the death of the Prophet Muhammad in 632, the armies and missionaries of Islam carried the new faith from Arabia to the Atlantic coast of Spain and the borders of China. The conquerors came from a land of tents, mud brick and palm-trunk shelters, and settled in the heartlands of two of the greatest architectural traditions of antiquity: the Roman and Byzantine Mediterranean, with its columns, arches, mosaics and domes, and the Sasanian empire of Iran and Iraq, with its brick vaults, great iwans and carved stucco. The architecture of the Islamic world was made from both, and for thirteen centuries it has been one of the most inventive in the world.

It is also one of the most varied. The mosques of Córdoba, Isfahan, Istanbul and Delhi look utterly unlike one another, and each belongs as much to its region as to its religion. What unites them is not a style but a small set of requirements and habits of mind: a building oriented towards Mecca; an ornament that avoids images of living beings in religious settings and turns instead to writing, plant forms and geometry; a love of courtyards, water and gardens; and a way of designing from geometric rules that could be carried in the head and taught from master to apprentice. This chapter follows those constants through five regional traditions: the early mosques, Moorish Spain and North Africa, Iran, the Ottomans and the Mughals.

A wall towards Mecca

A mosque (masjid, “place of prostration”) needs very little. Prayer can be offered anywhere that is clean, and the first mosque was simply the Prophet’s house at Medina: a square yard enclosed by mud-brick walls, with a shelter of palm trunks roofed with fronds along the side facing the direction of prayer. Everything essential to the later mosque is already there. The faithful pray in rows, shoulder to shoulder, all facing the Kaaba at Mecca; so the defining element of a mosque is not a centre but a wall, the qibla wall, and the ideal prayer hall is wide rather than deep, so that as many people as possible stand in the front rows.

℟Everything turns to the qibla

Orient the prayer hall so that its principal wall, the qibla wall, stands square to the direction of Mecca, and mark the middle of that wall with a niche, the mihrab.

qibla wall ⟂ direction of the Kaaba; mihrab on the axis

Within a few decades the essential furniture had been added. The mihrab, a concave niche in the centre of the qibla wall, marks the direction of prayer and the place of the prayer leader; beside it stands the minbar, a stepped pulpit from which the Friday sermon is preached; near it the ruler might pray in a screened enclosure, the maqsura. The minaret, a tower for the call to prayer, appears in the eighth and ninth centuries. Around the prayer hall, or haram, lies the open courtyard, the sahn, lined on its other sides with porticoes, riwaqs, and furnished with a fountain for the ritual ablutions.

The great mosques of the Umayyad caliphs made this arrangement monumental. At Damascus (706–715) al-Walid I built within the enclosure of a Roman temple a prayer hall of three long aisles running parallel to the qibla wall, cut through the middle by a taller transept leading to the mihrab, and faced it with a courtyard nearly as large, its arcades once shimmering with mosaics of trees and palaces. In North Africa the Great Mosque of Kairouan, founded in 670 and rebuilt in its present form in the ninth century, arranged the aisles the other way, running towards the qibla wall, with a wider central aisle and a transverse aisle along the wall forming a T, domes at the ends of the axis, and, opposite the mihrab across the court, a massive square minaret of three diminishing stages.

℟Court and prayer hall

Build the congregational mosque as an enclosed courtyard (sahn) with a covered prayer hall (haram) on its qibla side and shaded porticoes (riwaq) on the others.

The forest of columns: Córdoba

The hypostyle hall, a roof carried on a grid of columns, has no natural size and no single centre. Its great virtue is that it can grow. No building shows this better than the Great Mosque of Córdoba. Abd al-Rahman I, the Umayyad prince who escaped the massacre of his family and founded an independent emirate in Spain, began it in 784–85 with eleven aisles running towards the qibla. His successors extended it three times: Abd al-Rahman II lengthened the aisles in the 830s and 840s; al-Hakam II lengthened them again in 961–976 and built a new mihrab of dazzling richness; and al-Mansur, the powerful minister of the caliphate, added eight aisles along the whole east side in 987–88. Each campaign repeated the same bay, and the building remained one room: a forest of more than eight hundred columns.

℟The hypostyle grows by addition

Make the prayer hall of identical bays of columns and arcades, so that it can be enlarged by adding aisles or bays without changing its character.

hall = n aisles × m bays of one module

Córdoba also shows how a constraint can become an invention. The builders reused columns from Roman and Visigothic buildings, most of them only about three metres tall, too short for a hall of this size. They did not look for longer columns; they stood a rectangular pier on each capital, carried the roof on semicircular arches springing from the tops of the piers, and braced the tall slender supports lower down with a second row of horseshoe arches. The two tiers of arches, their voussoirs striped in red brick and pale stone, recede in every direction; walking among them is like walking in a grove of palm trees, which some visitors have supposed was the intention.

The horseshoe arch was already known in Visigothic Spain, but the Umayyads of al-Andalus made it their signature and handed it on to the Almoravids and Almohads of Morocco and to the Nasrids of Granada. They framed it in a rectangular moulding, the alfiz, filled its spandrels with carving, multiplied its curves into polylobed and interlaced forms, and in the Alhambra reduced its supports to slender columns that seem hardly able to bear the weight of the stucco above them. The style outlived Islamic rule in Spain, in the Mudéjar work of Muslim craftsmen for Christian patrons.

Ornament without images: the geometry of the girih

Islamic religious art avoids the representation of people and animals, and the builders of mosques developed instead three kinds of ornament, often combined on a single surface. The first is writing: verses of the Qur’an, the names of God, dedications and dates, in the angular Kufic and flowing cursive scripts, running round domes and framing doorways. The second is the arabesque, the endlessly branching and returning plant scroll, derived from the vine scrolls of Late Antiquity but disciplined into rhythm. The third, and the one most closely tied to architecture, is geometric pattern: interlacing bands that form stars and polygons, called in Persian girih, “knots”.

These patterns were not drawn freehand; their perfection of repetition shows that they were generated by rule. In 1905 a British scientist working in Agra, Ernest Hanbury Hankin, proposed a rule that reproduces a great many of them (Hankin, 1925). Cover the surface with a tiling of polygons (octagons and squares, for example). Through the midpoint of every edge draw two lines into the neighbouring polygons, each at the same angle to the edge, and continue them until they meet other lines. Then erase the tiling. What remains is a star pattern in which every line runs unbroken across the field.

℟Patterns from polygons in contact

To draw a star pattern, cover the surface with a tiling of polygons, draw two lines through the midpoint of every edge at the same angle to it, continue each line until it meets another, and erase the tiling.

pattern = tiling + contact angle θ at every edge midpoint

The model makes the logic of the method tangible. Each regular polygon in the grid produces a star with as many points as it has sides; the gaps between the primary polygons (squares between octagons, triangles between dodecagons, the bow-tie shapes between decagons) produce smaller secondary figures. The symmetry of the whole field is fixed by the choice of tiling, and its character by the contact angle: a small angle produces broad, calm stars and wide open polygons; a large one, a dense web of sharp points.

Hankin’s is not the only method that craftsmen used, and historians still argue about how much geometry the pattern-makers knew and how they knew it. The strongest evidence is a single extraordinary document, the Topkapı Scroll, a roll of paper some thirty metres long made in a workshop of western Iran in the late fifteenth century. It contains 114 drawings of patterns and vaults and not a word of text. Many of its patterns are drawn over faint construction lines (grids of polygons and circles from which the visible strapwork was derived), exactly as in the model above. The scroll shows pattern as a procedure, a design that could be reproduced by anyone who knew the rule.

Muqarnas: the vault as honeycomb

The same scroll contains plans of muqarnas, the most distinctive vaulting of the Islamic world. A muqarnas is built of tiers of small concave cells, like little niches, each tier stepping forward over the one below, each cell set over the junction of two cells beneath. Muqarnas probably grew out of the squinch (the small arch thrown across the corner of a square room to carry an octagon and then a dome) and appear almost at once, in the eleventh century, in north-eastern Iran, in Iraq and in North Africa. They spread to every part of the Islamic world, filling the hoods of portals and iwans, the corners of domed chambers, the capitals of columns and the balconies of minarets.

For all their apparent complexity, muqarnas are systematic. They are built from a small number of standard cell shapes, all of the same height, and the whole vault is determined by its plan (the projection of its tiers on the floor) and by that module of height. That is why the Topkapı Scroll can record a muqarnas as a flat drawing, and why a thirteenth-century plaster plate incised with such a plan could be found at the Ilkhanid palace of Takht-i Sulayman. In 1427 the Timurid astronomer Jamshid al-Kashi, in his Key of Arithmetic, gave rules for measuring the surface of muqarnas from exactly these elements, so that a builder could estimate materials and costs. In Nasrid Granada the technique reached its extreme in the domes of the Hall of the Two Sisters and the Hall of the Abencerrajes, where thousands of plaster cells dissolve the ceiling into light.

Iran: iwan, pishtaq and the double dome

Iran had its own ancient repertoire of vaulted forms, and from the eleventh century it became the most influential centre of Islamic architecture east of Egypt. Its characteristic space is the iwan, a vaulted hall walled on three sides and open on the fourth, the descendant of the great audience halls of Sasanian palaces. In the Seljuk period builders began to place an iwan in the middle of each side of a courtyard. At the Friday Mosque of Isfahan, an older hypostyle mosque was transformed in stages: a great domed chamber before the mihrab (1086–87), another dome opposite, and in the early twelfth century four iwans facing one another across the court. The four-iwan plan became the standard for mosques, madrasas and caravanserais across Iran and Central Asia.

℟The four-iwan court

Place an iwan (a vaulted hall walled on three sides and open on the fourth) at the middle of each side of the courtyard, the one on the qibla side leading to a domed chamber before the mihrab.

The iwan’s open mouth was framed by a pishtaq, a tall rectangular screen that rises above the roofline, bordered by bands of inscription and often flanked by minarets, its arch filled with muqarnas. From the court the pishtaqs announce the axes; from the city they stand like banners over the low roofs of the arcades. Under the Timurids and Safavids the whole surface of such buildings was clothed in glazed tile (turquoise, cobalt, white and yellow), first as glazed bricks, then as mosaic cut from single-coloured tiles and fitted together, then as painted tiles laid in large panels.

The dome over the qibla chamber shows the Iranian solution to a problem that every domed building faces. A dome that looks well from inside, in proportion to the room it covers, is too low to dominate the view from outside; a dome tall enough to be seen across a city is absurdly high inside. The Iranian answer was the double dome: two shells, one above the other, braced together: a low inner dome for the room and a tall outer dome, raised on a drum, for the skyline. The Mausoleum of Oljeitu at Soltaniyeh (1302–12) and the Timurid tombs of Samarkand made the outer shell ever higher; at the Masjid-i Shah in Isfahan (1611–c. 1630) there are about fourteen metres between the two.

The Ottomans: one dome for one congregation

When the Ottoman Turks took Constantinople in 1453 they inherited Hagia Sophia, built for Justinian more than nine hundred years before and still the greatest domed interior in the world. They made it a mosque, and for a century and a half the central ambition of Ottoman architecture was to equal it. The ideal was a single space under a single dome, in which the whole congregation could pray together in light; the problem, as Hagia Sophia’s own history of collapses showed, was how to carry the thrust of a great dome down to the ground without filling the space with supports.

The architect who pursued this ideal furthest was Sinan, chief of the imperial architects from 1538–39 until his death in 1588 in extreme old age. His office designed hundreds of buildings, and near the end of his life he dictated to a friend an account of his career, the Tezkiretü’l-Bünyan. In it he names three mosques as the stages of his development. The Şehzade (1543–48), his apprentice work, is a dome on four piers extended by four semidomes into a perfectly centralised square. The Süleymaniye (1550–57), his journeyman work, returns to Hagia Sophia’s scheme of two semidomes, but clarifies it: the lateral arches are filled with windowed tympana, the side aisles are covered by smaller domes, and the whole is ordered by a single logic.

The Selimiye at Edirne (1568–75), which Sinan called his masterpiece, abandons the semidomes as the principal support. Its dome, about 31 metres across, rests on eight slender piers set close to the outer walls, with small semidomes only in the corners; the space under it is almost undivided. Sinan explains in the memoirs that he built it to answer those who said no Muslim architect could raise a dome as large as Hagia Sophia’s. Outside, the weight of the dome descends through a ring of buttresses, turrets and smaller domes, and at the four corners of the prayer hall stand four minarets some seventy metres high.

Every Ottoman imperial mosque is a composition in which structure becomes image. The central dome sits on a ring of windows; around it the semidomes, exedrae and small domes step down in a cascade, all sheathed in grey lead; weight towers rise over the piers. In front lies the avlu, a courtyard with a domed portico and a fountain, and at the corners stand the needle-like minarets, their balconies (şerefes) counted out according to the rank of the founder. Around the mosque, the schools, hospital, kitchens and baths of the külliye made it the centre of a quarter of the city.

India: the Mughals and the garden tomb

Islamic rule in northern India began in the twelfth century, and the sultans of Delhi built mosques, tombs and minarets (the Qutb Minar among them) with Indian masons who translated Islamic forms into the stone-carving traditions of Hindu and Jain temples. In 1526 Babur, a prince descended from Timur, founded the Mughal dynasty, and brought to India the culture of Timurid Central Asia: its double domes, its pishtaqs and above all its gardens. The charbagh, the fourfold garden divided by water channels into quarters (an image, in Islamic tradition, of the rivers of Paradise), became the setting for the great Mughal tombs.

℟The fourfold garden

Lay out the garden as a walled square divided into four by raised walks with water channels on the two axes, and divide each quarter again.

1 → 4 → 16 beds; water on the axes

At Humayun’s Tomb in Delhi (1560s) the tomb stands at the centre of its garden, a red sandstone block inlaid with white marble under a white double dome. Akbar built in a more Indian vein, in red sandstone with brackets, beams and chhatris, at Fatehpur Sikri. Under Jahangir and Shah Jahan, white marble and inlay of semi-precious stones (pietra dura) replaced sandstone for the finest buildings, and pierced marble screens, jalis, filtered the light.

The Taj Mahal at Agra, built by Shah Jahan for his wife Mumtaz Mahal, who died in 1631, gathers the whole tradition into a single image. The tomb, a cube with chamfered corners, stands on a plain marble plinth about 95 metres square. Each of its four faces is the same: a great pishtaq opening into a deep iwan, flanked by two storeys of smaller arched recesses. Above it a tall drum lifts a bulbous outer dome, surrounded by four chhatris, far above the inner dome over the cenotaph chamber. At the corners of the plinth four free-standing minarets frame the tomb from every direction. And instead of standing at the centre of its garden, the tomb stands at the end, on the terrace above the river, so that from the gate it is seen across the length of the charbagh against the open sky, and reflected in the central tank.

One tradition, many regions

The buildings in this chapter span nine centuries and eight thousand kilometres, and they differ in almost everything that can be seen: material, structure, scale, colour, the form of the arch and the dome. What they share lies deeper. All of them are ordered by a direction outside themselves, the qibla, and by an ideal of the enclosed garden and court. All of them treat surface as a field for writing, plant scroll and geometry, and all of them were designed by rule: patterns from grids and angles, vaults from plans and modules, buildings from simple proportions set out with cord and compass. The regional styles are the result of those constant rules meeting the local traditions of Rome and Byzantium, of Sasanian Iran, of Berber North Africa, of Turkic Anatolia and of India.