What The Roof Does
In our environment the roof is two-dimensional.
In the inside-outside relationship this means that the roof protects an interior space against an exterior space, a space which is both over and around it. The space above is the vertical dimension, the sky. The surrounding space is the horizontal dimension, the ground’s surface (Fig. 436).
In relation to the sky, a roof may accept the sky, which means that it guides a downward motion from above. On the other hand if the roof resists the sky, the motion will be directed upwards from below. A roof may also be balanced between downward and upward motion (Figs. 437 a-c, 438, 439).
In relation to exterior space a roof may direct motion inwards toward a centre and thereby close the space or outward along a line and thus open the space. Both effects may operate simultaneously (Figs. 437 d, e, 440 a-c).
The Roof Themes
Throughout architectural history we find a series of shelter forms which each in its own way is a variation, verticality and horizontally, of these expressive components.
On closer examination we find that these variations are based on just five themes: (a) the dome, (b) the barrel vault (c) the gable roof, (d) the shed roof, (e) the flat roof (Fig. 441 a-e).
Each of these archetypes conveys specific expressions with regard to motion, weight, and substance, which in turn influence our experience of the space beneath. In describing the expression of these various forms, the terms sinking, rising and neutral will be used when referring to verticality, while the terms centralizing and directional will be used when referring to horizontality.
These impressions, however, depend upon three conditions, each of which will influence the main expression of each theme. The first concerns the spatial form created by the walls, the second pertains to the height of the walls, and the third has to do with the articulation of the walls.
When it concerns spatial form, a dome over a square will, for example, convey a far different impression than a dome over a round space. A barrel-vaulted roof over a long, narrow space will give a different effect than if placed over a short, wide space etc. Seen in this way, roof form and spatial form will affect each other mutually (Fig. 442 a-d).
When it is a question of the wall’s height, a dome above low walls will convey a close and protected feeling, whereas the same dome set on high walls will give a rising effect. A corresponding effect is conveyed by a flat roof. Placed on low walls it may seem threatening, but raised slightly it will seem light and airy (Fig. 443 a, b).
When it comes to articulation, the openings in the wall are of great importance in determining the motion effect. In a barrel-vaulted corridor the placement of the openings in either the sides or ends will influence the apparent length of the corridor (Fig. 444 a-c). Of equal consequence is the articulation of the transition between the two elements. In the case of the roof, it is a question of whether the roof form appears to bear down on the walls or to rise up from them. As for the walls, the question is whether the wall architecture is carried upward into the roof form or whether the roof architecture is drawn down along the walls (Fig. 445 a, b).
In the following we shall examine the five roof themes individually. We shall, first of all attempt to describe expressions of motion and weight in order to determine their impact. Thereafter we shall study examples showing different motifs and their variations. The use of various forms, articulation, lighting and colour will increase or lessen the expressiveness of the themes and their importance for our perception of the ‘struggle’ of interior space against the powers of the environment.
The Expression of the Dome
As a form the dome is to be understood as a roof covering based on the principle of an arch in rotation around a vertical axis (Fig. 446). The plan of the dome is circular; its form is vertical and has a continuously curved surface. The key words in describing the basic characteristics of the dome are, therefore, centrality, continuity, and rising (Fig. 447).
What then is expressed by the dome?
The dome is associated with numerous and diverse conceptions and forms.1See Baldwin Smith, The Dome…, especially the chapter entitled Domical Origins, pp. 3—10. In this chapter, he refers to how dome forms have been compared to objects such as pine cones, the cosmic egg, bee hives, etc. Common to all of them is their reference to conceptions of the cosmos. The main reason for this is that the dome is a reminder of the sky and in its very form a replica of the heavenly sphere we have above and around us. The primitive house, accordingly, was regarded as a miniature of the universe itself (Fig. 448).2See the portrayal of a house as im ago mundi, in Eliade, Dethellige…, p. 33. This association may be seen in the word ‘dome’, derived from the Latin ‘domus’ which means ‘house’.3Baldwin Smith, The Dome…, p. 5.
As a form also, the dome contains within itself the fundamental qualities which are typical both of the sky with its elevating character and of the house with its protective associations. In that the dome ascends in a circle around a vertical line, it is by inclination potentially related to the sky. But, because it also arches above and around us, it conveys a feeling of both safety and protection, which again by its very nature is related to the conception of dwelling, to the house itself. This aspect is particularly discernible in the primitive hut in which walls and roof were fused into one encompassing and continuous unit. The word cupola which is akin to the Sanskrit word ‘kupah’, which means ‘cave’, suggests the same.4A. Reichlin, Cupola, P. Portoghesi (ed.), Dizionario., II, p. 120. The dome gives an enclosing feeling of safety, which corresponds to the effect of an all-encompassing and protective universal space. But, the dome is also centralized. Its focus is the centre of the space where existence is calmed. The dome, therefore, indicates the very essence of what we mean by being inside, which in turn is the prerequisite for all security and life.
There are three main outstanding types of domes in which all these qualities are combined but in different ways and with different emphasis. These motifs may be compared to similar arch forms (see p. 227ff). As a result, we speak of the conical or elliptical dome (pointed arch), the spherical or semicircular dome (semicircular arch) and the flat or segmental dome (flat arch) (Fig. 449, a-c).
The three dome motifs represent three stages in an historical-technical development. E. Baldwin Smith, accordingly, judges the conical form to be the earliest and establishes its possible derivation from man’s first primitive round huts.5Baldwin Smith, The Dome…, p. 8. The form has its origin in the flexible or naturally bent reed and branch forms which were fastened together at the top with interwoven bands (Fig. 450). This form was subsequently transferred to stone as for example in the tholos of Mediterranean countries and in primitive folk architecture in Italy, Africa and Nubia (Fig. 448).
The spherical form is thought to be a result of mathematical and ideological speculation. As a form, the sphere corresponds to the conception of perfection in that the form illustrates complete and absolute harmony, ‘… the most perfect and the most like itself’.6Quotation Platon in Timaios, in P. E. Schramm, Sphaira, Globus, Reichsapfel, Stuttgart 1958, p. 8. In its construction too the sphere had advantages recognized by, amongst others, the Romans, who built in concrete and brick and had developed efficient and standardized building methods.
The flat dome derives from tent-shaped shelters and has been used thoughout history particularly as tabernacles and baldachins to mark important and ‘sacred’ places (Fig. 451). The four corner columns carrying the cupola are thought to have their origin in the poles of movable nomadic tents. In modern times the flat dome is used in roofs of reinforced concrete, and thanks to the rigidity of steel a much flatter span than those of high stone domes is made possible.
As we have seen, all three dome motifs are centralized and have approximately the same degree of curving continuity. The main difference between the three is first and foremost the difference in their degree of rising. The motifs, accordingly, will express the dome theme’s rising effect in the following ways: by accentuating the rising effect as the conical dome does, by neutralizing it as seen in the spherical dome, or by counteracting it as does the flat dome. Seen thus, the dissimilar degrees of rising may convey equally different effects, which vary between having a rising or ‘hovering’ effect, a balancing or ‘struggling’ effect, or one of giving and ‘protecting’.
Form of The Dome
We realize, nevertheless, that it is not the dome alone but its articulation as well which is essential in conveying its effecr. This includes not only the detailed form of the dome itself but perhaps more particularly the form given to the transition between roof and walls.
In joining a dome and a square space, the problem of transition may in principle be solved in two different ways (Fig. 452 a, b). Firstly, the dome is set directly on doubly curved pendentives, which distribute the dome surface evenly over the space beneath by way of the four corners.
In the other case, a cylindrical drum is inserted between the dome and pendentives.
Basically, the pendentive solution gives the roof a sinking character. If the dome rests on four columns, it enfolds the space within its grasp like a baldachin lowered from above. Hans Sedlmayr calls this solution ‘übergreifende’ and shows that it reached its height of development in the early Christian period (Fig. 453).7Sedlmayr, Das erste…, Epochen…, I, p. 80.
The drum solution lends a rising character to the roof. The dome is given an extra shove from below which lifts the roof up and above the lower wall zones. The rising effect, however, may be increased or decreased according to the articulation of the drum. In this context there are three alternatives in particular which stand out (Fig. 454, a-c).
In the first, the rising effect conveyed by the basic form is emphasized in that the drum unites an upper and lower space. The dome interior of St. Peter’s Basilica in Rome (1590) is an example of this (Fig. 455). Here the drum is broken up by alternating pairs of pilasters and window openings. These pilasters continue and become dome ribbing by way of projecting parts in the dividing dome ring. In this way the entire verticality of the form is accentuated and is further emphasized by the drum windows between the pilasters. They ‘free’ the dome from the walls below and give it an even greater rising effect in accord with the open oculus in the crown.
In the second alternative, the drum divides walls and dome thereby neutralizing the rising effect of the main form. St. Peter’s dome may again serve as an example, but this time as we find it in Donato Bramante’s project about one hundred years earlier (Fig. 456). The interior of the drum in this case is composed of a closely set circle of columns supporting a perfectly spherical dome with no articulation whatsoever. On the exterior we find a corresponding colonnade but here the dome is encircled at the bottom by stepped horizontal buttressing resting on the cornice.
In comparing the exterior and interior forms an obvious difference may be seen. The exterior of the dome is heavy. With its horizontal stepped buttressing it rests on the colonnade beneath. Inside, on the other hand, the dome neither weighs down nor rises. It ‘hovers’ above the colonnading like something immaterial.
In the last variation the rising effect is counteracted because the drum conveys the impression that the dome is lowered down over the interior space. One way of creating this impression is to hide the transition between dome and walls. In other words, the dome continues down behind the drum, which in turn rises into the dome. An example of this is the vaulting over Francesco Borromini’s San Carlo alle Quattro Fontane, in which the dome ring is hidden by a crown’ of foliage and flowers. Another illustration is Etienne-Louis Boullée’s project for the Church of the Madeleine (1785) (Fig. 457). Here the drum is transformed into a free-standing screen which projects upwards into the dome space. The dome is painted with heavenly visions which combined with the arcade convey a feeling of infinite space.
The painting of the vaulting extends down onto the rear wall from which the columns jut out, with the result that the expanse of the Heavens and the glory that adorns the vaulting and cupola become immense.8Quotation E.-L. Boullée, in Rosenau, Boullée…, p. 93.
In the following we shall return to individual dome motifs. We shall treat them separately and with the help of examples show how the expression is interpreted by various types of articulation.
The Conical Dome
The conical dome emphasizes the rising effect.
This applies technically also in the sense that a conical dome when its section approaches the shape of a catenary curve is the ‘lightest’ of the dome forms because its stress is led downwards through its own form with minimal horizontal stress. Giovanni Poleni demonstrated this principle in his chain experiment (1748), in which he allowed the catenary curve, illustrated by a hanging chain, to describe the silhouette of St. Peter’s dome (1590) (Fig. 458). The same principle was used in Filippo Brunelleschi’s Florence Cathedral dome (1420). In this Gothic silhouette the horizontal stress was minimal, with the result that both the dome itself and the walls beneath could be made thinner than if the form had been spherical. Corresponding ideas lie behind the construction of modern paraboloids. Great compound curved concrete shells reach the ground supported by thin arched pillars, giving the form a floating effect undisturbed by slanting horizontal buttressing.
The dynamics in these examples lies not only in the construction method but equally in the dome form itself. The contours rise continuously just as in the Gothic arch, and all lines converge at the apex in a point. This rising effect is probably one of the main reasons for the frequent use of the conical dome for exteriors whereas it is less prevalent in interiors (Fig. 459). A typical example is the double dome, in which the outer dome may well be the rising, conical type while the inner dome is either spherical or flat (Fig. 460). An important reason for this difference between the exterior and interior dome is visually determined. A spherical outer dome, when seen from below would have given a sinking impression. We know that originally Michelangelo planned a spherical dome over St. Peter’s Basilica in Rome (1569).9See R. Wittkower, Zur Peterskuppel Michelangelos, Zeitschrift fur Kunstgeschichte, II, Munich 1933, pp. 348—380, especially p. 363. Later (1590) Giacomo della Porta gave the dome a markedly conical form which resulted in a lively verticality both in the dome and in the building itself (Fig. 455).
The conical dome is seldom found as an interior vault. One reason could be that its form gives the space a diminishing and funnel-like effect as it rises. In this sense both the spherical and flat type are more ‘immediate’ and in closer harmony with the roof’s function as a canopy and as an upper completion of the space. An example of an attempt to minimize this vertical effect in a conical inner space is to be found in the Treasury of Atreus at Mycenae (c. 1350 B.C.), where the bronze covered heavenly dome is drawn down over the walls (Fig. 461).
The Spherical Dome
The spherical dome is balanced between rising and sinking.
Just as the semicircular arch, when read from the bottom up, appears to rise but read from the top downwards seems to sink, the hemisphere too is a form that balances between up and down. Basically, the spherical dome is at rest (Fig. 462). The type was very much used during the Renaissance, a period which sought after a balanced relation between elementary geometric forms (Fig. 463). Several early Renaissance examples reveal domes which were frequently without any articulation. In this way the form’s neutrality was clear and undisguised. Thus the spherical dome in itself is not decisive in determining the chief dynamics of the space below but is dependent upon articulation in order to convey an impression of rising, floating or sinking.
A classic example of the spherical dome in which all these fundamental motions are in balance is the Pantheon in Rome, built by the Emperor Hadrian (A.D. 124) (Fig. 464). This dome combined with the drum-like walls beneath creates a space famous for its perfect balance. Here all forces are equalized in a perfect calm, completely in keeping with the intention of the space as a symbol of unity between spiritual and human power in an all-encompassing universe. ‘One bows beneath a magical power emanating from the stars, from their movements and constellations, a power which is centred in this hall’.10L’Orange/Thiis-Evensen, Oldtidens…, p. 116.
The Pantheon and The Additive Dome
The dome of the Pantheon is to be understood as a coffered vault. In keeping with Classical tradition and the flat coffered ceiling, familiar to us from Greek temples, the vault too is an element which basically rests on the walls. The walls are given, thereupon the upper conclusion becomes an additional other form’.11Sedlmayr, Das erste…, Epochen., I, p. 87.
Thus, despite the fact that the roof is an addition to the walls, it is simultaneously both united to and separated from the walls. The combination of all these factors affects the spectator and creates the overall impression of the space. The impression, however, varies in magnitude according to the position of the observer beneath the dome.
The Pantheon and The Integrated Dome
The dome and its walls are united verticality, horizontally and perspectively (Fig. 465 a-c).
Besides the equality in height of dome and walls, it is particularly the interplay of the rings of coffering and the oculus which convey this effect.
The vault is divided into squares by the coffering and becomes a network of vertical and horizontal lines. In this way attention is drawn to the fundamental characteristic of all domes, which is the combination of an encompassing and a rising effect. We find the same characteristics in the division of the walls beneath. The horizontal lines are repeated in the moulding cornices, which divide a high main storey and a low attic storey, whereas the vertical lines recur in the upright columns surrounding the main level of the space.
In this way walls and dome are united verticality by means of the column lines. These again are picked up by the vertical dome ribs and led up to the oculus, which in turn opens up and expands freely towards the sky (Fig. 465, a). This, combined with the diminution of the coffers towards the apex creates an effect of walls and dome rising towards the light. The verticality, however, may be read in reverse. The oculus, true enough, when regarded as a hole is an opening which guides the inner space outward. But, seen as a light source it conducts the sunlight from outside to inside. In a compact cone the light is cast down from the dome’s crown. Now the ribs have become the sun’s own ‘accomplices’ and like a cluster of constructed rays guide the light down upon the spectator.
The upper and lower spaces are integrated horizontally also (Fig. 465, b). The wall mouldings and the roof ribs may be seen as a continuous system of rings gradually becoming more concentrated as they approach the dome opening as though the space is about to seal itself off at the crown. Regarded in this way the space may be read from the bottom up, but, the dome opening in counteracting this closure makes it possible to read these same rings from the top down as well. With their centre in the oculus they spread down over the space in ever-expanding rings just as when one casts a pebble into still water.
Perspectively, upper and lower parts are also united (Fig. 465, c). This effect is accomplished by the way the roof coffers are calculated from a starting point in the centre of the floor — thus their perspective gradually straightens the further one looks upwards into the vault. This means that the roof is connected to the centre of the space by a network of sight lines. In other words, the roof is controlled by the person in the centre. This person is the emperor, he who sits in judgement at the central point of the space and commands the universe. ‘Viewed from here the hovering vault seems to be drawn toward the observer’.12MacDonald, The Architecture…, p. 115.
The Pantheon and The Separated Dome
Vault space and wall space, as we have seen, are merged in several ways. But, the upper and lower parts are two separate elements. This is underlined by the attic storey, which acts as a dividing insertion between the main storey and vault. This applies, however, to the form of the attic before its reconstruction in the eighteenth century. The present form, with deep blind niches and intervening solid panelled areas, gives the storey a strong plastic character, causing the dome to rest heavily on the main storey beneath. This zone, originally, consisted of rows of small pilasters evenly encircling the entire space. The pilasters were unrelated both to the rib system above and to the columns beneath, thereby breaking into the continuity of the two levels. Statically as well, the attic storey was experienced as a disrupting zone. The small columns conveyed a non-supporting air of lightness in sharp contrast to the plastic realism of both the dome and main storey. The result was a hovering dome — a dome which in utter calmness and raised high above the world announced its content of ‘templum deorum omnium’.13Quotation Dio Cassius, in ibid., p. 118.
The Flat Dome
The flat dome seems weighted down in contrast to the conical dome with its rising effect and the spherical dome in its tranquility (Figs. 466).
The flattening of the form is due to a weight coming either from above, which presses the form flat, or from below, which hinders the rising effect. Both will convey an effect of tension. In the first case we sense the ability of dome and walls to withstand downward pressure, whereas in the second case we read the capacity of dome and walls to restrain expansion from below (Fig. 467, 468). The first tendency may be compared to a drop of liquid which, owing to its intrinsic weight, is flattened on top and bulges out at the sides. An example of such a form is the shell dome of the University Synagogue in Jerusalem (1958) designed by Ezra Rau (Fig. 469). The other tendency could be compared to an inflated sail anchored to the ground with posts and guy lines. This is exemplified in antiquity’s renditions of baldachins over thrones and altars (Fig. 451).
In the following we shall limit our analysis to two examples of the flat dome, specifically, the modern domes over two sports stadiums in Rome. Both are designed by Pier Luigi Nervi (1955 and 1959).
The Sports Stadiums in Rome
At first glance Nervi’s sports stadiums resemble one another in both form and structure.
Both have the same function, both are round and covered by large flat domes. In both cases the domes are constructed with visible supports and ribs and in both, the spectator benches and arena are below ground level. Even so, there are some important differences. Whereas the small sports stadium, the so-called Palazzetto, appears muscular and ‘combative’, the large Sports Stadium seems hovering and ‘ordering’. Each illustrates the expression of the flat dome, the first by its pressed down effect and the latter by its expansiveness.
The Sports Stadiums: Exteriors
The differences between these two stadiums are already apparent on their exteriors.
The Palazzetto is fundamentally associated with the ground (Fig. 470). The interior space is sunk into it while at the same time the exterior supports spring up from it. Y-shaped supports surrounding the Palazzetto give evidence of an unusual upward thrust from below. This contributing strength is not necessitated by the weight of the thin dome covering with its undulating scallops in the transition to the supports below. It is rather the dome’s form which requires this expression of power in that these supple buttresses must be there to ‘convince’ us that the flat span will hold. From this point of view the Palazzetto expresses a struggle against the vault’s own weight.
In the large Sports Stadium the dome supports are hidden by a circular glass wall (Fig. 471). In this way the visible interplay of forces between the dome and ground is interrupted so that the roof appears to hover above the glittering membrane. The glass ring itself is divorced from the ground in that the ground storey is set back slightly in relation to the glass wall above. Twelve free-standing staircaises spring directly out from the wall ring and bring the hovering dome structure into a light contact with the ground. These serve to emphasize even more the non-supporting character of the form. This is enhanced by still another factor: the location of the building. The large Sports Stadium is situated at the end of a broad boulevard and crowns a high ridge just outside Rome. In this way the ridge itself is used as a further means of elevating the structure towards the sky.
The Sports Stadiums: Interiors
The exterior differences continue in the interiors of these domed buildings.
In both cases the vault ribbing remains visible. It is questionable whether the ribs are statically necessary for the support of the dome. They do, however, have the practical advantage of hiding and conducting lighting fixtures, pipes and air-conditioning ducts. Apart from these considerations it is obvious that the ribs are of visual importance. Without them the enormous vault surfaces could seem both threatening and overdimensioned. Now the roofs are experienced as being secure’ because the ribs ‘form a pattern that suggests the isostatic lines of the principal stress’.14Nervi, Aesthetics and Technology in Building, Massachusetts 1965, p. 7.
The rib pattern, however, differs in the two buildings. The rib network in the Palazzetto is restless (Fig. 472). By means of rhombic formed coffers interspersed with large criss-crossing arches the surface is set into motion in an endless rotation around the lantern at the top. The transition to the Y-shaped supports is also questionable in that the clusters of straight ribs which carry the billowing cloaks must be joined to the network of curved lines in the dome surface above. The main impression is of a roof in a plastic ‘struggle’ — a roof which like a whirling disc must remain in motion or fall down.
In the large Sports Stadium the character of the rib pattern is one of supreme calm (Fig. 473). Effortlessly and uncompromisingly the ribs ascend in uninterrupted lines to the lantern eye at the zenith. From below, the dome is lifted up by slender V-shaped supports with the ribs radiating directly from them. Sitting inside the Palazzetto it is impossible to see how the roof meets the ground. In the large Sports Stadium the supports are brought right into the interior space and seem to spring directly from the upper tiers of the stadium benches.15This is solely the optical impression of the supportive function. The actual distribution to the ground is led along solid cross-supports which are concealed inside the outer corridor ring around the middle hall. These rows of benches form a separate gallery raised above the rest of the seats below.
This impression is consciously accented by the use of light and shadow. The roof ribs are V-shaped in section. In a broad belt at the bottom immediately above the V-shaped supports and again at the top around the lantern, the ribs are solid and unperforated. In the intermediate zone the ribs are perforated by large openings. The lighting is placed here and gives the entire section of the construction a feather-light effect, while the upper and lower areas appear darkened and heavy.
The overall effect of this articulation is strong and convincing. The dark zones ‘draw’ the dome together at both top and bottom. At the top the dome becomes more compact as it reaches its zenith around the light aperture and again at the bottom as it nears the dark tiers of benches to which it is anchored and thus hovers above a dividing belt of light. The resulting effect is one of a flat dome with an extra curvature in the illuminated area, which imparts a rising effect to the whole body and like a full sail lifts entire zones of the space beneath. The dynamics of the roof become more strongly emphasized when seen in relation to the sinking effect in the arena below. Seen in contrast, the distinctive quality of each part strengthens the other, so that the sunken floor and rising roof depend upon one another to reach a state of dramatic tension (Fig. 474).
In our survey above we saw that all dome examples were of true domes. By this is meant that they all were circular in plan, had a continuous surface, and were upright.
Throughout architectural history, however, there are many roof forms resembling domes, in other words, forms that contain one or several of the qualities characterizing the true dome.
These may be divided into two main groups. The first may be called modified dome roofs. By this is meant roof forms which are basically true domes but which have undergone changes of various kinds. The other group may be called dome-like roofs, meaning roof forms which basically are either true barrel vault, gable or flat roof and have been altered in such a way as to give them some of the dome’s characteristics.
In the following we shall examine modified dome roofs as a separate category while the second group will be covered in connection with other roof forms.
Modified domes may be divided into two sub-groups in which the main reason for the alteration will be given. The first we call the hanging or inverted dome and the other the modelled dome.
The Hanging Dome
The main difference between a hanging dome and a true dome is that the hanging dome conveys a falling impression. Such a form may impart a threatening effect, a feeling that a thin membrane is being pressed down by a powerful thrust from above. We shall give two examples of how this primary effect varies according to the dome’s articulation and juncture with the walls.
The hanging dome of St. Hallvard’s Church in Oslo, designed by architects Kjell Lund and Niels Slaatto (1966), is essentially just such a powerful ‘threat’ to the circular interior (Fig. 475). The surrounding walls are of brick, whereas the dome is cast in reinforced concrete, with the pattern of the formwork left visible. This pattern leads downward towards the dome’s lowest point and gives added motion to the sinking effect of the basic form. The feeling of weight is further intensified by the way in which the roof rests directly on the walls with no indication of joining whatsoever. In this way the walls are given accented importance in the load-bearing process, which in combination with the roof gives the total space a distinctive strength.
In the space below a hanging dome it is only beneath the lowest point that a sensation of direct ‘danger’ is felt. In contrast, from any other position it is as if the roof surface lifts the space and opens it outwards in all directions. In St. Hallvard’s Church this effect is exploited in order to give the ceremonial room a further meaning. The lowest point lies well behind the centre of the space while the altar is placed at the opposite end, giving an impression that the entire space ascends towards the choir. Thus, the threatening effect of the roof is a condition for the redeeming effect of the altar, which in addition becomes the ‘rescue station’ of the interior both optically and symbolically.
In Le Corbusier’s Notre Dame du Haut (1954), the hanging roof is both threatening and hovering at the same time. The space is not circular as in St. Hallvard’s but trapezoidal, opening fan-like as it nears the altar (Fig. 476). In section too, the space opens towards the altar. The floor sinks and the roof rises. The lowest point of the roof and its sharpest curvature is opposite the choir and immediately above the church entrance. The curve of the roof surface gradually flattens as it rises towards the choir wall at the east end. But, the roof rises transversely as well, to the south, i.e. towards a wall from which the light streams in through deep window openings (Fig. 99). The roof is separated from the walls by a narrow light slit over both the choir and south wails, while in the rest of the church it rests heavily on the walls where they are lowest. Weight and lightness are thereby combined in the same form. Heaven is both ‘near’ and ‘distant’ and as such, is a subtle reminder of Le Corbusier’s intention for this space as a place of concentration as well as meditation. Christian Norberg-Schulz says accordingly:
The hanging roof is both a heavy weight which concentrates the interior and reminds man of his precarious situation on earth, and a light ‘heavenly’ veil, which floats over the walls.16Norberg-Schulz, Meaning…, p. 408.
The Modelled Dome
The modelled dome differs from the basic form in both plan and profile. In the true dome the plan is a circle and the profile is evenly curved. We have seen the problems involved in connecting it with spaces beneath, and how these have had to be solved by means of transitional elements such as drums and pendentives.
With the modelled dome, all types of spatial forms can be directly joined to the roof, thereby uniting upper and lower areas. This may be interpreted in two ways, either that motions in the lower part are gradually gathered together and brought to rest by the dome above or that the dome little by little dissolves as it nears the walls. In this context there are two main types, the directionalized and the broken dome.
Most typical of the directionalized domes is the oval. The relation of the oval dome to exterior space has frequently been of an optical nature. A case in point is the dome of St. Maria di Montesanto in Rome by Carlo Rainaldi (1657). This building with its neighbouring church forms the gate to the city’s Corso from the north. Rainaldi designed both with the intention that they should appear as symmetrical buildings framing the entrance to the avenue. As the building lots were unequal in width, he used oval form to correct the perspective, so that from a distance the domes appeared to be equally broad and round (Figs. 477, 478).
The other important role of directionalized domes is related to oval or direction-orientated interiors. The earliest example we find of an elongated dome covering an oval space is in Giacomo da Vignola’s church, St. Anna dei Palafrenieri in Rome (1572). The oval unites a centre and a path into one form (se II Campidoglio p. 85 f). This means that the horizontally extended lines of the lower space are brought together in the centre of the roof area above. Thus, the organizing principles, point and line, are each relegated to a separate level within the same space. The lower part, which is the ‘world’, emphasizes a path, whereas the upper part, which is ‘heaven’, marks a goal (Fig. 479).
The broken dome is the other variation of the modelled dome. In this form a polygonal plan is joined to a spherical cover. The simplest variation is the plan in which the dome surmounts a square. This variation has been widely used in French Baroque to crown tower-like projecting parts of buildings. Such square domes are typical exterior motifs. This is because the form unites four separate faҫades (Figs. 480, 481). It is not formed in rotation around an axis but by two intersecting diagonal arches. The dome surfaces are stretched between these and are flush with the walls beneath. The four facades, thereby, appear to be drawn upwards into the roof, gradually becoming more curved and pointed before converging at the top. The result is a form that is both centralized and directional, making it easier to combine with other spatial forms around it.
A more complicated variation illustrating the possibilities of the modelled or shaped dome is found in Francesco Borromini’s church, St. Ivo alla Sapienza (begun 1642) (Fig. 482). The plan is developed around a hexagon with alternating concave and convex wall areas. This pulsating rhythm of expansion and contraction continuous up into the dome until finally all motion is brought to a standstill around the lantern at the top.
The space strives upwards from below towards the light (Fig. 483). Even the windows in the lower part of the dome narrow upwards to add to the vertical effect. The pilasters and ribs emphasize the unity of the upper and lower parts and flow in continuous unbroken lines from floor to lantern. The space, however, not only rises upwards but also flows downwards. But, now symbolic elements prevail. A heavenly light shining from the lantern materializes in showers of stars between the ribbing of the dome. We must imagine these symbolic rays of light streaming further down the walls to mark twelve niches between the pilasters at the bottom intended to hold statues of the apostles. Viewed in such terms, the interior of St. Ivo conveys a dynamic union of an upper and lower part, each with an essentially different meaning. The muscular and plastic lower part rises upwards, an image of man’s own pursuit of the Divine, while at the same time the Divine Spirit is directed downwards from above in the form of light and stars, an inspiration reminiscent of the tongues of fire descending upon the apostles in the Pentecostal miracle (Fig. 484).
Another variation of the broken dome is the pyramid and the spire. In these, all curvature has disappeared, but both vertical and centralized characteristics remain.
The pyramid roof is frequently used as an outer covering over an interior dome space. It is easily built of wood and has been important as a plain tower cap, particularly in Carolingian and Romanesque architecture. These tower caps were often quite low, corresponding to their width in a relation of 1:2, so that the effect was both cubic and restful, completely in keeping with the character expressed in the building’s massif as well. Seen thus, pyramid roofs became not only a capping but also a part of the tower beneath.
The pyramid roof is also used in Dutch architecture (Fig. 485). As large forms they gather all the functions of a farm under one roof, thereby avoiding the need for outbuildings on the precious land. An additional reason for this use was the necessity for protection from the gusty winds which sweep over the flat landscape. All these requirements are recognizable in the form of the pyramid roof itself. The pyramid rests heavily on the ground and defines the boundaries of the house around a collective point. But, the pyramid rises also and points directly upwards and is equally collective in its erect and upright bearing. Both aspects are equally important in an endless horizontal landscape in which it is necessary to both anchor and mark, to collect or gather around a low centre as well as upwards around a vertical.
Whereas the pyramid is in repose, the spire stretches upwards as its height is far greater than its width. Whether it consists of four or more surfaces, the broken form creates lines and angles which in combination with its height convey a distinctly linear appearance. One could imagine the same spire shaped as a round cone, in which case it would immediately convey a heavier and more plastic effect. In this sense one might say that the broken spire ‘grows’, an aspect which the neo-Gothic interpreted quite literally by allowing the ribbing to be formed as stems sprouting leaf-like projections (Fig. 486). The spire is the ‘dome’ of the Gothic and forms the natural conclusion to the verticality of the building itself.
The Barrel Vault
The Expression of the Barrel Vault
We have seen that the dome form was the result of an arch in rotation around an axis.
The form of the barrel vault is formed by a series of arches set at intervals along a line. Otherwise, it has the same qualities as the dome (Fig. 487). The primary effect, therefore, is one of emphasized horizontal motion while at the same time the eye is directed upwards. Above all, it conveys an impression of movement, by seeming to be ever in a state of new formation, so much so that given certain proportions it seems actually to rise upwards’ (Fig. 488).17Wőlfflin, Renaissance…, p. 115.
Two conditions in particular are important in conveying this directional expression. The first is contained in the roof’s cylinder effect. Made up as it is of a half cylinder, the barrel vault allows the space to open up at each end but closes it evenly on the sides. The second condition is that roof and wall are continuously joined both statically and optically. This means that the side walls are supporting elements and therefore enclosing, whereas the short walls, not having this function may be removed (Fig. 489).
Barrel Vault Motifs
We have seen that there are several factors not only in the form of the barrel vault but also in its relation to the walls beneath, all of which may influence the directional effect of a tunnelled space. We shall now examine these factors individually by dividing them into three main groups. The first deals with the proportions of the space and covers the relationships of length, width, and height. The second group deals with the modelling of the space, which means the effect conveyed by various arch forms and other plastic treatment. The third will cover the articulation of the vault, which involves the effect of certain types of openings and divisions. There after, we shall examine the groined vault, which is the result of two intersecting barrel vaults.
The directional effect of a barrel vault will be far greater in a lengthy space than in a short one (Fig. 490). The same applies if a straight barrel-vaulted space is compared to one which is curved.
A straight space will be in ‘balance’. Upon entering a barrel-vaulted corridor we find that the motions of the space and our own movements coincide. In a curved corridor, on the other hand, a conflict between them arises. This reaction is purely physical in that centrifugal force leads us out of the curve and straightforward, whereas the space with its barrel vaulting leads us around in a swing. Take, for example, the circular ambulatory of St. Costanza in Rome (c. A.D. 340). Here it is the space which assumes command, and the influence of the barrel vault in this respect is of corresponding significance (Fig. 491). This expression of being guided or led has a greater effect in a curved space than in an axial one. In the former the roof exerts an influencing power while in the latter it is primarily a part of the effect (Fig. 492).
To carry the thought further, if the width of a barrel-vaulted space is varied, the following may occur. In a wide space the transverse motion will compete with the longitudinal motion so that the feeling of being led will be weakened.18By a ‘broad’ space, it is meant that the breadth of the space is equal to or greater than the length. In a narrow space the depth effect becomes more important, while at the same time the vault comes closer to the observer. Whereas the roof in the wide space was high, swinging up and around, the vault of the narrow space is lower, thereby accenting the cylinder and with it forward motion (Fig. 493, a-c).
A somewhat similar effect occurs if two barrel-vaulted spaces are given different heights but length and width remain unchanged. If the walls are low, the lines of perspective and the depth effect are intensified. This, combined with the nearness of the barrel vault, which seems to encircle the onlooker as in a tube, accelerates the forward motion. In contrast, a high vault will seem distant and far, with walls drawing attention upward and not forward.
The same dynamic tendencies which were examined above may be obtained by other means as well. Although we were aware that height, width, and depth varied, the form of the barrel vault itself remained constant. If this too changes, the expression becomes even more complicated. Considered thus, a flat, circular, and a pointed vault will result in quite different effects (Fig. 493, c). A pointed vault will increase the height tendency and lessen the effect of depth, whereas a flat barrel vault will accent the transverse at the cost of this effect. The spherical vault causes the space to rise, encloses it and gives it direction, controlling all these tendencies in a sort of dynamic balance.
Furthermore, the form of the section is of importance. An example is the so-called hyperbolic paraboloid. The curves in such a form are created by the use of parabolas which hang between two vertical parabolas and which in section also form parabolic curves. In such a space one is always ‘on the move’ through a pulsating unrest of contractions, expansions and accelerations (Fig. 494).
In addition to proportions and modelling, articulation takes its place as the third influential factor.
This vast space is based on the cylinder (Fig. 497). In the upper part of the form, the barrel vault is the guiding element in the main direction of the space. The vault coffering accents the main directional lines. The elongated skylight also adds to the horizontal motion and at each end the space is opened by huge arches. In the lower part the rest of the cylinder is formed by surrounding galleries. These rise from the central floor like tiers of seats around the arena of a Roman circus. Along the top of these stretches a severe colonnade, which is cut off at the ends so that the space is not closed off longitudinally. All in all, a space is created in which the details intensify the main tendency of the vault. In this strongly dynamic space it might be thought difficult to ‘relax’. This, however, is managed by a distinctive transition between vault and wall. The colonnades along the top of the galleries do not support the vault above. On the contrary, these colonnades are free-standing screens in front of the vault which continues on down behind them. Thus, the same effect is gained as in Boullée’s Madeleine dome — the roof sinks downwards. In this way the cylinder effect is offset and the place dynamically defined. This, in combination with the light streaming from above, gives the entire space dramatic tension.
In the above we have considered various forms of a plain barrel vault. We shall now examine an important barrel vault combination. The most common type occurs when two barrel vaults intersect. The resulting groined vault combines the characteristics of both the dome and the barrel vault and allows for rich possibilities in directing space both horizontally and verticality (Fig. 498). As an example of how this is exploited in order to interpret definite functions, we shall examine Trajan’s Market in Rome (c. A.D. 100).
Access to the hall from the surrounding city is through the short ends. The roof rises high above the low tabernae and is in principle an elongated barrel vault, which is intersected by lesser barrel vaults lying side by side. These have the same ridge-line as the longitudinal vault. The intersecting vaults are the same width as the tabernae beneath and act as light shafts above the shop roofs.
The result is a combination of directional tendencies in the roof which is reflected in corresponding motions in the space beneath (Fig. 501,a-c). The longitudinal vault emphasizes the intention of the space as a passage from one part of the city to another. From the exterior the vault stands forths as a strong and inviting entrance motif (see p. ??). The expression of length, however, is not the sole governing influence. In moving through the hall, each crossing may be interpreted as an independent baldachin in which the intersecting lines as well as the gradual narrowing of the vaults radiate towards a centre. In this way forward motion is ‘halted’ so that at the next moment attention is diverted to the shops on either side. From this point of view, the hall unites three tendencies in all: the main longitudinal direction and transversal direction lines draw attention to space outside the hall while the intersecting points between them are the centres of baldachins which accent points inside the hall.19Thiis-Evensen, Aula Forensis, AHO-skrift, Oslo 1971, p. 2.
The Gable Roof
The Expression of The Gable Roof
The gable roof is the result of the need to rid the roof surface of rain. The form of the roof, too, was altered, for being on account of its flatness, unfit to throw off the rains … it was raised in the middle … after the form of a gable roof’. Thus simply did William Chambers explain the origin of the gable roof (Figs. 502, 503).20Quotation W. Chambers, in Rykwert, On Adam’s…, p. 71. Filarete, just as simply, asserted that it was Adam who formed the first roof by placing his hands over his head to protect himself from rain and sun (Fig. 504). This gesture took the form of a triangle, the same triangle which with the column and beam constituted the basic elements in Marc Laugier’s vision of the first architecture (Fig. 505). Together they form an aedicula, little house’, which is the very essence of all classical architecture (see ‘the frame’ pp. 271 ff). Palladio, too, considered the gable as one of the original architectural elements. The pedlment, therefore, which graced temples and public buildings was a primeval motif, a survival from the first primitive house.21Palladio, The Four…, Preface.
We have already described certain aspects of the gable roof’s open and closed qualities. These may be summarized in three points: (1) it is the gable of the roof which opens the connection between outside and inside; (2) the gable roof encloses along its sides in that the two sloping surfaces cut off the relation between inside and outside along both ground and ridge lines, (3) there is upward motion in the gable as it rises towards the ridge line. In other words, the gable roof expands verticality at its apex, it expands horisontally in the lengthwise direction of the ridge, and it sinks diagonally along its surfaces.22Typically, in Norwegian the pitched surface of the roof is referred to as the ‘takfall’ (‘the fall of the roof’). The basis for the expressions conveyed by this form is too be found within these qualities.
Gable Roof Motifs
The Rising Aspect of The Gable Roof
The ascending quality in the gable roof causes the space to rise upwards. In this connection, a Gothic sharply angled ceiling has been characterized by Andrew S. Downing as follows:
The superior effect of this ceiling arises, partly, from its carrying the eye upwards and thus recognizing the principle of perpendicular rather than horizontal support, as well as causing it to appear higher than it really is; and, also from a certain airy lightness, found in a ceiling in which the lines rise, however slightly, but never in the one entirely flat (Fig. 506).23A.J. Downing, The Architecture of Country Houses, New York 1969, (facsimile of edition from 1850), p. 385.
In the same way, the exterior gable peak will point and not ‘gather in’ as does the arch. Throughout architectural history this quality has been emphasized by the use of the most diverse ridge decorations, from figures of the gods and rosettes in Antiquity’s acroteria to spires and flower garlands in the Swiss style gable tower (Fig. 507). The former cause even the hills of Rome to rise, writes Martial in his panegyric on the fastigium atop Domitian’s palace on the Palatine: ‘It pierces heaven, and hidden amid the lustrous stars its peak echoes sunlight to the thunder in the cloud below’.24Quotation Martial, in MacDonald, The Architecture…, p. 62.
This heaven-bound quality in the gable’s triangular form is embodied in the phenomenon ‘Irminsul’. Irminsul is the universalis columna connecting earth and the heavens and is associated with the central post in the earliest versions of the gable roof (Fig. 508).25J. Trier, First, Nachrichten von derGesellschaftder Wissenschaften zu Gottungen, Phil.-Hist., IV, NF III. 4, 1940, p. 117. We find it as the motif above the royal portal (the Lion’s Gate) in Mycenae (c. 1250) (Fig. 509), as well as in decorated totem poles on the hut facades of west coast Canadian Indians (Fig. 510). The latter show fully the lengths to which man is willing to go in order to emphasize verticality in the gable roof.
The Sinking Aspect of The Gable Roof
The sinking aspect of the sloping surfaces lends a protective quality to the roof. The roof of the Greek temple has been compared to the eagle’s wings, enfolding and protecting the treasures of the inner sanctuary. ‘The eagle is the symbol of Zeus and in the temple gable, interpreted this way, it is as if the entire Olympian world descends to enfold the building within its powerful wing-span’ (Figs. 511, 512).26L’Orange, Tempelet…, p. 7.
Modern studies reveal that the pitched roof may convey a feeling of safety and security. Accordingly, Richard D. Cramer maintains that in the United Kingdom it is the gable roof and not the flat roof which psychologically is felt to be a protective form (Fig. 513).27R.D. Cramer, Images of Home, A!A Journal, September, Washington D.C. 1960, p. 40-49, p. 42. The pitched roof, therefore, has become the symbol of what is meant by a home. The same results were reached in another study in which various house forms were chosen according to the degree of comfort and ‘cosiness’ expressed by the roof.28W. Cowburn, Popular Housing, Arena-. Journal of the Architectural Association, September/October, London 1966, p. 81.
The Directional Aspect of The Gable Roof
The directional quality of the gable roof becomes apparent, on the one hand, by the frequent emphasis on the gable facade itself and, on the other hand, by the importance given to the ridgepole and the roof ridge.
The gable is used quite intentionally as a faҫade. It stands as the symbol of an opening out of or into a building, usually in connection with entrances or large window areas (Fig. 514).
The roof ridge, too, is directionally important. In the interior it is the ridgepole itself which guides a horizontal movement towards each gabled end. Figuratively also, this element is a connecting axis. As ‘Firstbaum’ it represented the axis of the earth itself, uniting the two poles of the north and south (Figs. 508, 515). On the exterior it may be the ridge itself which is formed in such a way as to accent horizontal motion. An example of this is the grimacing dragon heads at the ridge ends of Norwegian stave churches. These are attached to undulating decorative bands along the ridge itself (Fig. 516). The entire body of the dragon seems to writhe its way with great speed along the length of the ridge like a defending animal ready to spring outwards against unearthly evil spirits. The dynamic essence of the gable is contained in these dragon heads, as is its meaning, in that the vertical, horizontal and diagonal tendencies are all combined in one protective symbol.
In the following we shall study examples of the way in which various forms can emphasize the three dynamic directionalities of the gable roof. We will look at the effect of formal variations in addition to articulative variations both in the roof surface itself and in the transition between roof surface and walls. This will be treated first of all in relation to the gable roof as an interior motif and then as an exterior faҫade motif. Finally we shall examine the hip roof as an example of the most important variation of the gable roof.
The Gable Roof as an Interior Motif
As we shall demonstrate later in connection with the gable faҫade, motion in the gable roof depends on effect, the angle of the ridge peak. The following description, therefore, will be limited to showing how an almost identical form can convey either a sinking, rising or directional character depending upon its articulation or attachment to the walls. We shall, furthermore, limit ourselves to examples taken from Norwegian wood architecture.
The Sinking and Directional Interior
Two interiors may serve as examples of the sinking and the directional aspect of the gable roof.
The principle behind the first aspect may be illustrated by the console ceiling. Such a ceiling rests on consoles or brackets placed at some distance down the wall (Fig. 517). They are found frequently in Gothic halls in the form of dark brown, richly carved trusses springing from whitewashed or painted stone walls. This solution had a constructional advantage in that the exaggeratedly high wall massifs acted as supports for the diagonal stress of the roof trusses. With that, the effect of these large roof constructions was emphasized by the way the entire ceiling seemed to ‘glide’ down the walls so that wall space and ceiling space became intermingled (Fig. 518). The result was a heavy and majestic, an even threatening impression, clearly distinguishing these secular halls from the spiritual character typical of the verticality in contemporary churches.
The other example which illustrates an enclosing gable roof is found in Norwegian log-timber houses with so-called purlin roofs (åstak) (Fig. 519). The roof is carried lengthwise by large log purlins (åser) which correspond in number to every second log in the gable end walls. The principle is that the roof and long walls are joined horizontally in that the lengthwise wall logs seem to ‘roll’ up into the ceiling and on down to the walls on the other side. The continuity in this form destroys any impression of planar surfaces, thereby creating dynamics similar to that of the barrel vault.
The Rising Interior
The rising aspect of the gable form is emphasized by the pattern of the lines in the rafter ceiling (Fig. 520). In the first place, the supporting rafters are upright and in the Norwegian rural house rise straight up to meet at the ridge forming sharp angles contasting with the horizontally laid log walls beneath. Secondly, the layer of smooth planks overlaying the rafters has a neutralizing effect giving an impression of ‘open’ areas between the members.
This basic effect in the rafter ceiling may be reduced or intensified according to variations in the construction. If a ridgepole is introduced it is as though the form closes — the rising effect is brought to a standstill while at the same time the length of the roof is emphasized. If one enters through the short end, to meet such a ceiling, as is the case in the traditional Norwegian ‘three-room plan’, the log ridgepole will accentuate the act of entering. The same applies if the log ceiling beams are situated otherwise, as in the large Rauland house in Uvdal (c. 1300) (Fig. 521). Here, there is no ridgepole but rather two double beams about halfway down each side of the rafter roof. These so-called ‘bear-backs’ give the necessary support to the construction while simultaneously accenting the act of entering from the two small side rooms at the gable end. Without the ridgepole the effect is as though the other two sets of beams ‘move aside’, away from the ridgepole. This, in fact, is the case because the room originally had an open smoke vent in the roof.
The smoke vent is actually the element that most strongly emphasizes the rising tendency in the rural house rafter ceiling. The effect was as if a gable roof was transformed into a ‘dome’ concentrated around a central, vertical shaft of light with increasing darkness in the lower parts of the encircling walls (Fig. 522, a). In the imagination the smoke might be a visualization of the upward vertical surge, while the smoke blanket beneath the roof, in blotting out the constructional and horizontal lines, transformed the whole to a softly outlined vault beneath a shining crown. That this vault-like effect was comprehended may be seen clearly in the chalk-painted decorations in the ceiling area. Both the gable wall and ceiling surfaces were often covered with continuous patterns of vine foliage, the white lines standing out against the soot-darkened background. The entire upper ceiling zone down to the first horizontal wall logs was treated as an entity, with all parts given equal value, a principle we recognize in ceiling decor done with rosemaling (a style of decorative painting) (Fig. 522, b). The Viku house from Oppdal has even been given wide moulding around the walls providing added plasticity in the transition to the ceiling area.
In the rural house examples given above, we saw that although the roof gave a rising effect, the walls were horizontal and prone. This conveyed an additive effect between ceiling and wall. In other words, the upper and lower zones were disconnected, they lacked continuity.
Conversely, in stave churches the columns are led directly into the roof rafters. This is particularly noticeable in Kaupanger Church (c. 1190) (Fig. 523), but these columns extend right up to the roof in other stave churches too, springing even if they are interrupted on the way by breastsummers, St. Andrew cross struts, and capitals. The Gothic form was the inspiration and ideal of the stave church. The dominant factor, therefore, was verticality, to which the sharply pointed raftered roof was the logical conclusion (Fig. 524. The stave church’s rafter roof consists of two sets of rafters, the lower ones crossing each other diagonally like ‘scissors’. Just below this crossing is a horizontal beam frequently supported by a collar beam. As they near the roof cornice, the rafters are strengthened and unified by U-shaped frames. These resemble inverted arches and give all sides around the roof a pointed form, not only crosswise but also from the long sides and up.
From the lowest arches the entire roof seems to rise with accelerating speed in layers of increasingly pointed rafters. All in all it is as if the roof projects itself upwards level after level towards the ridge line. This was also a characteristic of the open raftered ceilings of Antiquity’s basilica. The space was not closed in by a tight lid but rose through its network of beams and rafters steadily upward in a light and airy ‘heaven’ (Fig. 525). The lighting emphasized this effect. In the stave church as well as in the basilica the light enters through small windows in the clerestory beneath the cornice. This led to the illumination of the lower space while the ceiling space remained in semi-darkness. Thus resulted an emphasized impression of a roof absorbed into a distant mystical and unattainable world.
The Gable Roof as an Exterior Motif the Gable Faҫade
We have already pointed out that the gable can have essentially different expressions, depending on the roof angle. Three variations in particular stand out: (1) the pointed gable found in the stave-church and neo-Gothic gables, (2) the shallow gable found in the Swiss chalet and the American shingle-style house, and (3) the balanced variation as expressed in the crowning pediments of the classical temple and the Roman aedicula.
The pointed gable accents the verticality of the gable roof. Steeply pitched roofs may be used for practical reasons as well. Such roofs are suitable in terminating high, narrow buildings where the site is narrow and deep as we find in Dutch buildings. And, in Christian IV’s urban code it is stated that city roofs must be pointed, ‘so that in case of fire the firebrands would not as easily remain’.29From Royal Court letter dated 4, October 1624, in A. Col left, Gamle Christiania Billeder, Christiania 1893, p. 72.
The pointed gable, however, is also used because of its expressiveness. It is a daring architectural motif in the way it actively breaks and thrusts upwards from the environment. In Dutch streets it is the gable which gives each house its individual identity. And, against the steep roof expanses covering stave church galleries, it is these pointed gables which have sufficient strength in themselves to accent the entrance and transept (Fig. 514).
The verticality of the pointed gable gives the form its own distinctive meaning. The ‘Irminsul’ ideology is accented and with it the sacred dimension. It is for this reason that the pointed gable is so frequently used in church architecture as seen not only in the Gothic cathedral and the stave church, but also in modern triangular churches such as the ‘Arctic Cathedral’ in Tromsø, northern Norway, by Inge Hovig (1965) (Fig. 526). The latter demonstrates clearly the expressive use of the pointed gable, which here rises upwards in one great frontal sweep while simultaneously being dramatically drawn right down to the ground on either side. In this way the building is directed towards Heaven and is protective at the same time — it is both ‘temple’ and ‘fortress’.
The shallow gable accents the sinking expression in the gable roof. Such flattened gables are typical of houses with extensive width if, for example, they are situated in a hilly landscape as in the case of the Swiss chalet (Fig. 527). The Norwegian rural storehouse, too, has a wide flattened gable mainly because the covered upstairs gallery is broad and outflung and also because any great roof height was considered unnecessary (Fig. 528). Besides, the sod covering required a gently pitched roof. In addition, the shallow gable was important in restraining the impression of height. The roof lent weight and width to the upstairs covered gallery so that these storehouses fitted snuggly into the landscape.
The weighty impression conveyed by the blunt gable is found also in the earliest Greek temples (Fig. 529). In these, the root pitch is markedly gentle, so much that both the gable and the high entablature seem to press down on the columns beneath. Viewed in such terms, the columns’ bulging form is a logical answer to the entire sinking aspect of the gable roof (see columns p. 201 f). This sinking aspect in the shallow gable also accentuates its protective impression. Its general character is secular and enfolding as is so clearly apparent in many American shingle-style houses (Fig. 530). Modern architecture also uses the shallow gable in the same manner and in this case as a solution to the question of putting the entire house ‘under one sheltering roof’.30Scully, The Shingle…, p. 118.
In the balanced gable, rising and sinking aspects are almost equalized. Typical examples are the pediments of later Roman temples, which, combined with slimmer and taller columns, give the building as a whole a lighter character than its precursors. The roof does not weigh down upon its walls but sits upon them as an independent unit. It is this characteristic that is renewed in the Renaissance canon in which balance was the ideal and the gable element took on a planar character, with no accentuation of either rising or sinking (Fig. 531).
We have seen that the gable motif does not act independently but must be considered in combination with the motion tendencies of the walls below. In other words, whether or not the gable in a facade composition is to show rising, sinking, or directional tendencies depends upon how it is linked to the walls. In this context there are three methods of joining them (Fig. 532, 1, 2, 3). In the first case, walls and gable are separated. Here, however, in principle, none of the main motion tendencies are accentuated in that the total effect depends upon the form given to the gable and walls individually (Fig. 532, 1). In the second method the walls are continued up into the gable which accents the rising tendency of the roof (Fig. 532, 2a, c). In the third case the walls are pulled back beneath the gable, which emphasizes its sinking aspect (Fig. 532, 3a-c).
The Addition of Gable and Wall In Exterior
A typical example of the first method of joining in which gable and wall act as two separate parts is found in the Greek temple facade. The gable frames a limited and isolated zone separated from the rows of columns by a broad entablature. As said previously, the overall effect of such a combination depends on the way in which the roof and wall ‘act together’.
In the Temple of Artemis at Corfu (sixth century B.C.) the general impression is one of heaviness and downward pressure because the flattened gable is ‘followed up’ by the underlying walls with their high entablature supported by short bulging columns (Fig. 529). On the other hand, the facade of Palladio’s temple of Mars, conveys a taller, slimmer more upright effect.
Here, the gable is more pointed, the entablature narrower, and the columns are in the slender Corinthian style (Fig. 531).
We have previously pointed out the importance of the temple roof as a symbol of the heavens. At the other extreme, the stylobate or floor has been compared to the earth’s surface. We know too from Vitruvius that the column symbolized man and the human body.31Vitruvius, The Ten Books…, p. 14. Considered in such terms, collectively the main facade of the temple embodies and represents the three levels: earth, man, and the heavens. These three worlds, although separated, nevertheless stand in a balanced relationship to one another, an attitude typical of Greek individualism. These conditions, however, may shift between a state of equal balance, as in the classical and Roman temple, and a state of imbalance in which heaviness and pressure predominate, as in archaic temples. Considered thus, the archaic temple may well reflect the uncertainties of the age which we find in the polis of the eighth and seventh centuries B.C., whereas the fully developed temple in its own way reflects the triumphant assurance of Hellenism and the Roman Empire.32T. Parsons, Societies. Evolutionary and Comparative Perspectives, New York 1966, p. 86, p. 103.
The Integration of Gable and Wall In Exterior
The gable’s rising tendency is accented by the way in which it is joined to the walls. This is mainly done in the following three ways (Fig. 532, 2a-c).
In the first method they are joined in such a way that the gable profile and wall corners form a frame, which makes a setting for the form. In the second variation they are unified in that wall and gable form an unbroken plane with the upper corners cut off to form a point. In the third case the form is created in that the upper part is concluded with a bargeboard.
The common factor in all three variations is the disappearance or disruption of the horizontal element in the triangle, so that wall and gable form one continuous surface right up to the ridge point.
The first variation is frequently seen in Scandinavian apartment houses from the 1940—50s (Fig. 533). This is connected with the increased use of prefabricated concrete framework elements and clearly limits the visual rising aspect of these concentrated exteriors.
The second type is to be found in the free-standing screen-like fronts. The screen principle opens the possibility for a free treatment of the gable profile which we find so richly exemplified precisely in the Dutch city house (Fig. 534). Typical of the Dutch example is its urban quality. It is independent of the rooms behind and may, therefore, vary in form and size according to the city space in front.
The third type is a combination of the first two in that the wall seems to lift the entire roof edge up in a sharp angle (Fig. 535). Some of the earliest examples of deviation from the classical triangular gable are the temple faҫades in Termessos and in Baalbek (Fig. 536). Here, the entablature is broken in the middle by an arch which curves up into the pediment. This means that the lower part, which is the ‘human’ zone, climbs up into the tympanum, which symbolizes ‘the heavens’ (see Palladio motif p. 237f). In some versions it is the emperor who occupies the central field beneath the arch and is thereby emphasized as the personification of the elevated and deified, a demi-god who unites earth and the heavens in ‘Imperium Sacrum’ (Figs. 333, 537 b). In the Baroque period this same motif appears again in the form of the so-called double-gable. The pediment consists of an arch framed by an outertriangle (Figs. 538, 539 a, b). This combination signifies an increasingly rising tendency as the arch initiates a motion that continues into the pointed part of the pediment above. Just the opposite may be imagined — the arch above and the triangle below, as was much practised during the Renaissance (Fig. 539 b). Now, the rising tendency is cut off by the arch above (see p. 275).
In more extreme variations of this, same form the gable point is broken as well. Here, it is as if the powerful upward thrust is given free rein, sometimes so markedly that the pediment point swings outward to either side (Fig. 540). In Lorenzo Bernini’s St. Andrea al Quirinale (1678), the open area is filled with a hovering, apotheosized St. Andrew. The result is a depiction of man transcendent, man both in this world and the next. This is found in the glorifying pediments of Antiquity, whereas in Greek temples man was only of this world, firmly attached to the columns and the earth beneath the pediment (Fig. 537 a-c).
The Contrast Between Gable and Wall in Exterior
If the wall is drawn back beneath the roof gable, the roof will project and overhang. The resulting impressions conveyed will vary greatly, depending on whether the roof overhang is: (a) only on the gable end, (b) only on the sides, (c) both places simultaneously (Fig. 532, 3a-c). These effects are determined by whether the roof seems to direct the inner space out or whether the inner space seems to withdraw in beneath the roof. Different combinations lead to an architecture in which the roof may create rich variations on the theme of the inside-outside relationship.
The first variation will convey a directional effect in as much as the roof directs the inner space out through the open gable end (Fig. 532, 3a). Such overhangs are found often as coverings over typical transitional spaces such as verandas, balconies and entrances. The primeval house in architecture, the megaron, has a projecting gable over its entrance in the short end. In recent Norwegian wood architecture, additional emphasis is frequently laid on this directional aspect of gable projection. These houses are usually long and narrow with horizontal rows of windows and horizontal panelling. In combination with large protruding roof beams, these overhanging gable ends give an accentuated horizontal character, dynamically uniting inside and outside, quite in keeping with Modernism’s principle of open space (Fig. 541).
The two following variations accentuate the gable roof’s protective character. In the first, where we find the overhang only on the long sides, the emphasis is on the sinking aspect of the space. This roof type does not direct the interior space out but ‘gathers it in’ as if under two outspread wings (Fig. 532, 3b). In the other variation the roof has an equal overhang on all sides. Here, the emphasis is not only on the outwardly directed aspect of the roof but also on the impression of walls being drawn in as under a protecting shield (Figs. 532, 3c, 542).
The typical effect of the projecting gable is the horizontal interplay between inside and outside. In contrast to an accentuation of the rising and collecting aspect, the directing quality of the gable roof will consequently cause a spreading effect between the building and its surroundings. For this reason this form is used particularly in connection with free-standing buildings requiring solutions in accordance with what Downing called ‘the local truth’, ‘To give expression of local truth to a country house, it should always show a tendency to spread out and extend itself on the ground, rather than to run up in the air’.33Downing, The Architecture of…, p. 33.
Variations of The Gable Roof
The most characteristic among gable roof variations is the hip roof (Fig. 543).
The hip roof occurs in many combinations. The earliest examples emerge in rural architecture in central and northern Europe. Originally, however, it was in French château architecture from the seventeenth and eighteenth century that this form reached its climax, with offshoots reaching into Scandinavian classicism as late as into the twentieth century.
Before looking more closely at individual examples it would be wise to seek the underlying purpose of the hip roof. If we compare a gable roof with a so-called Mansard gable, we find that the latter allows for an extra attic story with almost the same height and width as the storeys below, and this, without having to raise the entire roof height (Fig. 544, a, b). In addition, the roof angle becomes steeper than before, a particular advantage in countries with heavy snowfall. Apart from the fact that the full-hipped roof has a useful function in covering and protecting otherwise exposed wall areas, it may in certain cases have constructional advantages as well. It is, nevertheless, the visual character of the hip roof which has been the decisive factor in its use. The hip roof conveys a heavy and encompassing impression in which all sharp edges and projections have disappeared and the walls have been modelled into a unified volume. In this context it is typical that the Baroque period with its plastic conception of space evinced particular interest in the aesthetic possibilities of the hip roof. The hip roofs gave to French secular Baroque something which was lacking in the Italian, that is, a massiveness with expressive characteristics comparable to those contained in church buildings.
The hip roof must be understood as a gable roof, with dome-like characteristics (see domes p. 305 ff). Dome qualities are introduced in that the gable roof is refashioned into a closed form (Fig. 545, a-c). This restructuring, in principle, may be carried out in three ways, each of which establishes its own motif with distinctive qualities. First of all is the gambrel variation, which means that the long sides of the gable roof are broken up into several surfaces between the roof edge and the ridge. The other two motifs occur when the gables of the shortendsare pitched either in the upper part alone to form a half-hip or right down to the eaves to make a full-hip.
The Gambrel Roof
As stated above, the most typical feature of the gable roof is that it opens at the ends and closes along the sides. A gable roof will, therefore, give the space beneath direction as well as enclosure. The gambrel roof accentuates the directing aspect, because the divided roof areas guide the long-side walls up and around, thereby enclosing the sides more evenly. The space, in fact, approaches the cylinder form. Seen in these terms, the pure gambrel roof is more closely related to the barrel vault than to the dome, as the rising or sinking effect depends upon, among other things, both the number of pitched surfaces and the size relationship (Fig. 546). If there are sufficiently many sloped surfaces, the profile approaches a pure curve. In Palladio’s large Basilica in Vicenca (1549—1614) the profile is continuous and curving, rather like an extended dome. The surrounding galleries add to the expansive strength of the form (Fig. 547).
Hipping the upper part of the gable conveys a feeling that the opening in the full gable is about to be closed. This is decisive for its character in that the half-hip gives a sinking effect, as if the roof is about to be drawn down over the end walls (Fig. 548). A conflict’ arises between a closed and a directional volume in which the rising effect of the gable end is counteracted by the slanting surface of the roof above. The half-hip roof is used in Norwegian country manors, a style that emerged in the residences of government officials which were to be found all over the country. This demanded an architecture capable of adaptability, and the half-hip contained just this quality. Its drawn-down character kept a high house relatively low and anchored it to its place with closed and firm stability. At the same time, however, the half-hip roof opened the manor house out over the valley and countryside, the district governed from these residences (Fig. 549).
Whereas the half-hip roof has different facades on the gable ends and sides, the full-hip is a centralized and closed roof in which the walls have the same height all the way around and the roof has the same pitch in all directions. Seen this way, the hipped roof is related to the pyramid roof. It contains the vertical rising and sinking character of the pyramid but also the horizontal extension of the gable roof. But, in contrast to the half-hip, this extended effect is ‘tamed’ and contained within sloping surfaces on all sides.
Whereas the descending effect in the half-hip conflicted with the otherwise ascending character of the gable wall, the full-hip roof maintains a poised balance between the upper and lower parts. This applies whether interpreted as an addition of elements — horizontal walls and a closed roof — or whether roof and walls are seen as parts of one volume rounded off (Fig. 551 a, b).
This ‘neutral’ relation between rising and sinking is, accordingly, dependent upon proportions and articulation in order to express the predominating characteristic. This applies not only to the roof form alone but to the relationship of roof form and walls.
Three sets of proportions are of special importance in this context. They comprise first of all, the extent of the roof, secondly the pitch, and thirdly the roof profile.
If a hip-roof is extended, the horizontality will be emphasized and also the weight and sinking expression in the form. If the hip-roof is more pointed, the form becomes a pyramid in which the rising and sinking effect appears to be in balance (see p. 327 f).
The same conditions apply to the pitch of the roof. If it is steep, the roof will stretch upwards, if it is shallow the form will seem heavy and protective. The first condition is consciously exploited in French Baroque architecture, as in Louis Le Vau’s château, Vaux-le-Vicomte (1661) (Fig. 550). The corner pavilions have steeply pitched roofs, which in combination with the pilasters beneath, give the form an upward pointed tower-like effect. The connecting wings have the contained appearance of the Mansard roofs and lend strength to the spaces which tie the outflung pavilions together. At the same time the oval hall on the garden side is crowned by a dome, which, like an inflated volume, has an expanding effect.
The gently sloping and heavy full-hip is also utilized in certain types of Norwegian post-war architecture such as in Knut Knutsen’s summer house in Portør (1948) (Fig. 552). Low pyramid forms, linked together at different angles, seem to press the house down into the landscape. The effect is one of reassuring protectiveness in a hard climate.
The profile of the full-hip may be either convex or concave (Fig. 553, a, b). The convex profile seems to arch its back’, throwing a protective cover over the inside space in the manner of a dome. This is seen when the full-hip is combined with the divided surfaces of the gambrel motif. We see once again that these possibilities are exploited in French Baroque style, but in this case in order to gather together and integrate the many outstreched wings of the palace complex (see above).
The concave profile is often found in the curving hip-roof of the Rococo period (Fig. 554). Particularly characteristic is its use in the oriental pagoda. This concave curve gives the whole roof an air of soft compliance or ‘sweep’ and taken altogether conveys a light and hovering effect (Fig. 555). The roof seems lifted up by a thrust from below. But in another way, the downward sweep of the roof form is counteracted by the rising effect in walls, columns or brackets beneath. Thus, an explanation of the pagoda roof is of a visual nature although technical advantages have played their part.
The relation between roof and walls is just as important for the resulting effect as is the form of the roof alone. Low walls will cause the hip-roof to seem more oppressive and heavier than if the walls are high. In this context the execution of the transition beween roof and wall is particularly important. If the roof edges project far beyond the walls, the roof may seem both top-heavy and sinking. Such overhanging roofs are common in northern European rural architecture as well as in Japanese houses and the impression is always the same: an extremely sinking roof which slides down over the walls. The latter, meanwhile, seem to climb up underneath the hipping as if seeking ‘protection’. The resulting combination is an architecture conveying a strong impression of shelter and safety (Figs. 556, 557).
Another important factor in the roof’s rising or sinking effect is the form and placement of the windows, both in the roof itself and in the walls beneath. If windows are placed right up under the eaves the roof seems to sink down, if they are placed closer to the ground the roof seems lighter (Fig. 558). The same conditions apply whether the windows themselves are tall or low. Oslo Lådegard (present form from 1722) has a central attic storey rising high above the roof (Fig. 559). The result is one of overlapping — a wall that springs upwards with a simultaneous and contrasting sinking of the roof on either side. The ‘flexibility’ of the walls is emphasized by tall windows and a high foundation, while the sinking effect of the roof is en-hanced by a marked projection of the cornice. On the other hand, tall chimneys at the ridge points stress an upward-pointing tendency. The total effect, therefore, alternates between tall and stately at one moment and heavy the next, a visible reminder of the building’s function as a symbol of pride and strength.
Windows may be built into the roof area in the form of dormers. The form of the dormers is decisive. Accordingly, a dormer that lifts up only a part of the roof will convey a downward motion, while the more plastic eye-shaped dormer will give the roof a greater effect of plasticity and weight (Fig. 560).
Decisive also is whether or not the dormer windows resemble the windows below. Around Damplassen, a square in Oslo, designed by Harald Hals (1919), the form of the wall windows is repeated in the dormer windows immediately above (Fig. 561). The houses around this square appear to be two-storey buildings, whereas in reality they contain three storeys, since the dormers conceal additional attic apartments. The similarity of the windows indicates this, but at the same time announces the importance of the roof. In that the windows are drawn upwards, the roof seems to be pulled downwards. As a result, the entire roof area appears heavier and more protective. This gives the square a stronger surrounding frame, while at the same time the whole feeling of security contained in the function of dwelling is more clearly expressed.
The Shed Roof
The Expression of The Shed Roof
Characteristic of all the roof types we have examined so far is the predominant impression of balance. The dome, barrel vault, gable roof and as we shall see later, the flat roof too, create motions balanced around verticality, horizontality or both simultaneously. As spaces they also presented balanced structures. The dome enclosed on all sides and the flat roof opened on all sides, whereas both barrel vault and gable roof closed the sides equally while at the same time opening both ends.
The shed roof is one half of a gable roof and is pitched in one direction only. It creates, therefore, an asymmetric space (Fig. 562 a, b). By this is meant that in moving along the length of the slope or traversing it, the feeling in both cases will be of a transitional stage between two conditions (Fig. 563). Lengthwise the diagonality will accent the roof’s tension between rising and sinking, between vertical and horizontal. Transversely the shed roof will both open and close, both rise towards the exterior space and sink towards the ground.
Wind shelters such as those we find used by, amongst others the African Pygmy, illustrate the fundamental nature of the shed roof. These woven and leaf-covered frames are set right against the ground on one side and held up by obliquely planted poles on the other (Fig. 564). The innermost, lowest part shuts off and protects from the ground-sweeping winds, while at the same time the pitch leads the rain away. From this point of view the sinking content is important as a protection from the elements. The rising factor is equally important, but here, as a means of social contact. Pygmy society consists of small tribal units of six to eight members, each having its own hut.34See D. Fraser, Village Planning in the Primitive World, London 1962, p. 15. See also E. Camesasca (ed.), History of the House, London 1971, p. 13 (translated by I. Quigly). These huts are gathered in a half-circle, with the open sides of the shelters facing inwards towards a common fire. In this way tribal solidarity is both emphasized and made visible.
Shed Roof Motifs The Shed Roof as A Transitional Form
The shed roof’s importance as a transitional form between opening and closing made it particularly suitable as an entrance motif. In the form of baldachins or vestibules it effects the transition between inside and outside, between the building and outdoor space (see entrance motifs p. 295 f).
Above the entrance door to Le Corbusier’s Villa Stein (1927), the baldachin slants upwards towards the outside space. This means that the roof opens towards the visitor, to receive and guide him in.
Both the angle of the rise and the form of the slanting roof are important in their effect on the observer. In cross-section the form may vary by curving upwards or downwards (Fig. 565). Seen from the outside the shed roof which curves downwards in a convex arc will close the space and modify the opening. The opposite will take place if this same roof curves concavely so that the inside space is ‘pushed out’ towards the onlooker. Examples of the first solution are to be found in many modern grandstand canopies. Perhaps the best known is the roof of the grandstand in Florence by Pier Luigi Nervi (1932) (Fig. 566). The roof soars boldly upwards in a slight arc giving the spectators on the tiered bench rows beneath the feeling of sitting in an enclosed space and not just within a wide-open funnel-shaped area. At the same time it is as though the form enfolds and holds together this outdoor space, thereby establishing a closer contact between the sports event and the spectator.
Another well-known combination of opening and space-creating characteristics of the shed roof is the undulating canopy in front of Rome’s main railway station, Stazione Termini (1950), designed by L. Calini and E. Montuori (Fig. 567). Towards the square, the slanted surface flings itself audaciously aloft in a grand welcoming gesture towards the city. Nearing the building, the roof bulges upward again to create an area for the booking hall, which combined with the upward swing towards the city, gives the entrance space an air of rich and pulsating dynamism.
The Shed Roof as A Part of A Form
As a transitional space the shed roof should be understood in the broadest sense as a fragment. By this is meant that it is not only half of a gable roof but also a form capable of developing ‘out of itself’. If two shed roofs are put side by side, first with the two higher sides facing one another and then reversed, the resulting effects will differ greatly (Fig. 568).
Opening Towards One Another
In the first case the two spaces will appear to open towards one another. They belong together despite their size and the distance between them (Fig. 569, a-d). This means that the volume of each unit in any combination whatsoever may be chosen to suit function and landscape, without the composition losing its unity and falling apart (Fig. 570).
This principle is widely used in modern American architecture. ‘Sea Ranch’ (1965), designed by Charles Moore and others, is a well-known example (Fig. 571). Here, the shed roof makes the individual units settle into the hilly landscape and form groups around inner courtyards. The main volume has several offshoots’ in the form of shed-type bay-windows and projections. These attachments may be placed anywhere on larger columns without loss of cohesion. Thus, unity of form is obtained without compromising the desire for a free plan — a unity previously attained by resorting to symmetry and balance.
If a smaller shed roof is set within a larger one, as in Le Corbusier’s ‘Murondins’ project (1940), the resulting effect is an interior both closed and open (Figs. 569 a, 572). The smaller volume is encased by the larger and closes the space, while the larger rises above it to open the space. This combination creates a dynamic integration while at the same time the exterior is broken up and ‘fits naturally into the landscape, allowing picturesque grouping’ (Fig. 573).35Le Corbusier, Oeuvre…, 1938-1940, p. 97.36The vestibule into the Biblioteca Laurenziana by Michelangelo (completed 1560), is an example of how an interior can be articulated with elements that had earlier been restricted to exterior walls which faced streets or courtyards. See quotation of J. Shearman, in Sinding-Larsen, The Laurenziana…, p. 215.
If shed roofs are placed a short distance from one another, the composition seems to be split because their interpendence is otherwise so strong (Fig. 569, b). This tension is usually resolved by introducing between them a form that is either higher or lower than both sheds.
If the form is lower, the split effect itself is accented because the intermediate form conveys the impression of a ‘wedge’ pressing the two units apart (Fig. 569, c). This explains the tension in the faҫade of Robert Venturis ‘Beach House’ (1959), in which the accented motion of the entrance breaks open the form and makes way for the act of entering (Fig. 116) (see p. 93f).
When the intermediate form is higher, as also illustrated in the ‘Beach House’, the unity of the forms is underscored. The form between the two shed roofs both divides and unites the elements. It separates them by blocking the opening tendency of shed roofs. At the same time, however, it draws them together by turning their rising tendency into a straight vertical line (Fig. 569, d).
Opening Away from One Another
If shed roofs are placed with their lowest sides facing, the units are closed off from one another but open in the opposite direction. This variation creates a modelled exterior space. Whereas the first alternative primarily lent tension to the interior by excluding the space outside, this variation creates a tension-filled exterior space which either opens from the bottom or closes from the top (Fig. 574). The following is an example in which both these effects are exploited.
The Cultural Centre in Risør, southern Norway, designed by Erik Anker and Andreas Hølaas (1978), has a courtyard main entrance framed by two wings (see p. 295) (Fig. 575). The shed roofs, sloping downwards to the low walls, lend an air of intimacy to the courtyard. This modifies the size of the house so that the exterior space adapts well to the scale of the narrow streets of the town. This solution also joins the Centre to other characteristics in the local environment. A steep and dominating rock ridge rises immediately behind the building. The use of the shed roof around the courtyard gives the entire space an effect of opening towards the heights behind and drawing them into the overall impression. From this point of view, the shed roof is an agent in making the building a part of both the town and the surrounding landscape.
The Flat Roof
The Expression of The Flat Roof
The flat roof belongs to the countries of the sunny south — a roof never weighted down by snowfall.
The exterior of this roof may be compared to a raised floor on which one can walk. In Greek villages the roof is the outdoor terrace itself. For Le Corbusier the flat roof replaced the true ground on which the building stood (Fig. 576). A dome or even a shed roof is self-sufficient. A flat roof, on the contrary, must be ‘inhabited’ in order to assert itself at all. Typical of this is Le Corbusier’s roof landscape on I’Unite’ d’Habitation in Marseilles (1952) (Fig. 577). By means of large plastically formed chimneys, ‘houses’, and benches the roof becomes active as a centre for residents and children but also as a conclusion to the entire angular building below. In the same way the Baroque roof was both flat and ‘populated’, but in this case by statues and figural representations on pedestals placed along the roof edges with a surrounding balustrade as a protective ‘railing’.
Seen from the inside the flat roof or ceiling will direct the space equally in all directions. Motion is spread horizontally and in the relationship of above and below the flat roof is like a rigid lid (Fig. 442, a). Consequently, the flat roof is basically unaffected by the environment and in principle without expression. Throughout architectural history varying attempts have been made to create ‘places’ on flat roofs and thereby in the space beneath. These interventions may be divided into three main groups according to the treatment applied. The first concerns the articulation of the surface, the second concerns the transition to the walls, and the third group concerns the modelling of the roof form itself.
The Flat Roof and Surface Articulation
The white concrete roof of Functionalism illustrates the most neutral articulation of the flat roof. The intention was that roof and white walls should merge with only the angle of the junction to mark the transition. This accorded with Cubism’s demand for simple and concise volumes in which no part stood out from the whole (Fig. 578).
Other types of surface articulation aim at counteracting the lid-like tendency of the flat roof or ceiling. These are carried out mainly in two ways. The first method emphasizes the directional aspect in the ceiling with the help of accented lines. The other method is to give the impression that the ceiling rises.
Even a roof’s supporting beams may be used to convey direction (Fig. 579). In combination, therefore, a beamed ceiling will give the impression of two levels, leading in two different directions. At the same time, such a roof may also convey upward depth, particularly if the lower and upper parts are accented by using different colours.
The best example of the intentional exploitation of both these principles is the coffered ceiling (Fig. 580). The coffered ceiling is an answer to the problem of roofing large spaces and we find it, for instance, over the cella in Greek temples. The coffered ceiling may be likened to a network of equal squares framed by beams which are all laid on the same plane. Within each square is placed a gold rosette on a blue ground. We know that this motif symbolized the heavenly stars glimpsed’ through the grid of beams. The coffered ceiling, in fact, accents both the directional and the uplifting. With their quadratic form the coffers guided the motions of the space equally in all directions. Through them the ceiling seemed to open up into the endless sky. Considered thus, the coffered ceiling is symptomatic of the Greek architectural principle which balance between the horizontal and vertical reigned.
Another method of raising a flat roof has been to decorate it with illusionistic sky and heavenly motifs. The aim of such effects, which erase the borderlines between reality and the imagined, was to break up the limiting surface of space and magnify it into a world of mythological fantasy. By the use of various painting techniques the aim was to convey an impression of perspective ‘di sotto in su’.
Still another means of heightening a space has been to mirror the ceiling. In many cases it may appear as if the resulting double height causes the space to become diffused and that mirroring dissolves the entire spatial form. Two examples show an attempt made to counteract this tendency. Both are designed by the architectural firm Jan & Jon. The first is in the entrance hall of the Norwegian University Press, Oslo (1980) (Fig. 581). The space itself is a trapezoid form through which one passes diagonally. In the white ceiling an oval has been cut out. It contains deeply inset mirrors. Thus, the optical dome-like effect within the form itself is conditioned by the actual height of the space around the opening.
The other example is taken from a garden room in a house at Nesodden near Oslo (1976) (see p. 67) (Fig. 70). The entire ceiling is fragmented in mirrored surfaces which are continued down the walls. In order to combat the complete disintegration of the space, a free-standing baldachin has been placed in the centre. In this way the space becomes a bit of each: both a clearly defined space and one in the process of disintegration — all of this from a plain flat-ceilinged quadratic room.
The Flat Roof and The Articulation of Transition
In principle, the form of the actual transition between a flat roof and the walls may create four quite dissimilar impressions.
Presuming the height to be the same in all four examples, the transition in the first case will convey the feeling of an open roof, in the second of a raised roof, in the third of an expanded roof, and in the fourth case, of a sinking roof (Fig. 582, a-d).
The principles behind these four different impressions will be demonstrated by showing the use made of various combinations of dark and light surfaces. Our references, however, will be mainly to forms of classical mouldings.
A flat roof may seem open if the walls are continued without interruption part of the way into the ceiling area. The impression given is of a ceiling lying above the walls’ springing level (Fig. 582 a). If the walls are light and the roof dark, a ‘hole’ may be created, which, combined with the extended walls, gives the space an effect of opening upward. This effect may be compared to the one obtained in the classical console cornice carried by either brackets or beams projecting directly from the walls. Such cornices have their origin in palace facade architecture and served also to draw a boundary line between the street space and the open sky. Indeed, it is often more appropriate to compare Baroque rooms to covered courtyards than to enclosed interiors.
A flat roof, furthermore, may seem raised or ‘hovering’ if the roof zone appears to be detached from the walls below. If we imagine both roof and walls to be in the same dark hue but between them a paler ‘belt’ extending partly into the roof area and partly into the walls, the surface will seem to rise (Fig. 582 b). A corresponding effect is conveyed by the concave moulding enriched with ornaments and flower garlands. Against the wall the groove is often concluded with a prominent row of dentils and against the roof with a similar number of convex profiles. The richly ornamented stucco-work will, however, lend to the entire concave area a non-structural sense so that the upper and lower mouldings tend to belong to the ceiling or to the wall rather than to the cornice itself. In this way ceiling and walls are separated, and the ceiling is ‘detached’.
A flat ceiling may appear to be enlarged and ‘expanded’ if the ceiling and the belt just beneath are pale while the walls up as far as this belt are dark and low (Fig. 582 c). A contrast between ceiling and wall arises here in that the walls seem to contract around the observer while the ceiling zone appears to expand. The overall impression becomes one of a lid too large for its box (see p. 309). This phenomenon is frequently met when attempts are made to lower the height of a room and make it more intimate by extending wallpaper and panelling part way down the walls.
All the above examples have one quality in common, that of a flat ceiling appearing to open up and lighten the inside space beneath. Our final example will show how the ceiling wall transition may convey an oppressive or sinking effect. The phenomenon is the same as we found with the gable roof (see p. 341) and may occur if a dark ceiling area is drawn partly down over paler walls (Fig. 582 d). Now the walls will expand while the ceiling contracts as though the cover is too ‘small’ in relation to the space beneath. We find examples of such transitions in cases where the ceiling glides without interruption partially down the walls in a series of convex and concave mouldings. A concrete example in this connection is a ceiling in the ‘Bracketed Style’ (Fig. 583). Here, the transition to the walls is made by means of slanted areas which surround the space and are finished off at the bottom by a narrow border projecting from the walls. All of this is supported by double beams which cross one another to end in brackets fastened part way down the wall. The result is a dome-like flat roof, which is drawn down into the space, so that the entire room becomes more compact and intimate.
The Flat Roof and The Articulation of The Form
A flat ceiling may easily appear to sink in the middle, specially if the area is large and low slung. As we have previously noted in connection with the floor and straight beam, attempts have been made to counteract this tendency by gently arching the element (see p. 221 ff). With its sloping upper edges, our last example illustrated how this tendency was surmounted by plastic treatment of the form itself. Two other examples are worth mentioning, particularly because the modelling underscores the idea and meaning of the space.
The drawing-room of the house at Kruke farm, Gudbrandsdal in Norway, is a large room approximately 7 X 7 metres and has a relatively low ceiling height (Fig. 584). The log walls are broken only by small-paned casement windows. The entrance is at one corner across from the large soapstone fireplace in the corner on the other side of the room. A stout cross-beam accents the connection between them. The panelled ceiling forms a large smooth surface at a relatively low height. A ceiling of this kind would have seemed heavy and sagging if the following steps had not been taken. Along the two side walls and crossing the line of entrance, the ceiling slants down towards the walls. In this way the ceiling is raised optically; it appears to be higher and at the same time the ceiling and walls are more clearly united. Another effect is also achieved. As the ceiling is slanted on only two sides, the space is given an accented direction. This parallels the roof ridge of the house. In this way the angled ceiling becomes a part of both the inside and outside. It is a part of the exterior space by reflecting the direction of the house out towards the large valley. But, it is first of all a part of the character of the inside space. The log walls enfold and frame the room space. The form of the ceiling follows up this tendency and gives the whole an intimacy and compactness which it would otherwise have lacked. This compactness is the first thing that meets the eye upon entering, a sight which in addition to the fireplace conveys a feeling of protection and safety — the very essence of what is meant by being inside.
Our final example has a completely different function than that of the living room space just described. The space is a mausoleum housing the sarcophagi of 335 war victims, situated on the Via Appia in Rome (Figs. 585, 586). Fosse Ardeatine, (1949) is a memorial to the Italian partisans who, in 1944, were imprisoned in a cave and killed. ‘Sarcophagus’ and ‘cave’ are, therefore, the key words for an understanding of the exterior of the building. On the crest of gently sloping earthen embankments lies a high concrete wall about 40 X 60 m. The impression is one of an immovable weight pressing down towards the ground — an allusion both to the sarcophagus lid and to the imprisonment which was the victims’ fate. This feeling of primeval weight is retained as one enters. But, once inside, the atmosphere is completely transformed. One is faced with an almost endless space beneath a cosmic roof. The floor is sunk several steps down to the level of the more than three hundred sarcophagi. These black coffins are laid in rows of pairs slanting upwards towards each other to form pointed crests. The low surrounding walls slant outwards and are built of large irregular blocks of tuff. Over the whole is flung the great roof — slightly vaulted and resting on small elements which separate it from the walls by means of a narrow, surrounding light slit.
The impact is powerful and the symbolism clear. The floor and sarcophagi are like an undulating black ‘ocean’ into which one descends. The coarse massive walls appear dimly as the very earth itself into which the whole has been lowered. Above arch the detached and floating ‘heavens’. Even the sounds within the space add to this impression. The chirping of birds and the rustling of the surrounding forests are amplified by the resonance of the vaults. The entire space collectively conveys a total image of ‘the world’ in which the heavens, sea, and earth are the setting for man’s span of existence between life and death.
|1||See Baldwin Smith, The Dome…, especially the chapter entitled Domical Origins, pp. 3—10. In this chapter, he refers to how dome forms have been compared to objects such as pine cones, the cosmic egg, bee hives, etc.|
|2||See the portrayal of a house as im ago mundi, in Eliade, Dethellige…, p. 33.|
|3||Baldwin Smith, The Dome…, p. 5.|
|4||A. Reichlin, Cupola, P. Portoghesi (ed.), Dizionario., II, p. 120.|
|5||Baldwin Smith, The Dome…, p. 8.|
|6||Quotation Platon in Timaios, in P. E. Schramm, Sphaira, Globus, Reichsapfel, Stuttgart 1958, p. 8.|
|7||Sedlmayr, Das erste…, Epochen…, I, p. 80.|
|8||Quotation E.-L. Boullée, in Rosenau, Boullée…, p. 93.|
|9||See R. Wittkower, Zur Peterskuppel Michelangelos, Zeitschrift fur Kunstgeschichte, II, Munich 1933, pp. 348—380, especially p. 363.|
|10||L’Orange/Thiis-Evensen, Oldtidens…, p. 116.|
|11||Sedlmayr, Das erste…, Epochen., I, p. 87.|
|12||MacDonald, The Architecture…, p. 115.|
|13||Quotation Dio Cassius, in ibid., p. 118.|
|14||Nervi, Aesthetics and Technology in Building, Massachusetts 1965, p. 7.|
|15||This is solely the optical impression of the supportive function. The actual distribution to the ground is led along solid cross-supports which are concealed inside the outer corridor ring around the middle hall.|
|16||Norberg-Schulz, Meaning…, p. 408.|
|17||Wőlfflin, Renaissance…, p. 115.|
|18||By a ‘broad’ space, it is meant that the breadth of the space is equal to or greater than the length.|
|19||Thiis-Evensen, Aula Forensis, AHO-skrift, Oslo 1971, p. 2.|
|20||Quotation W. Chambers, in Rykwert, On Adam’s…, p. 71.|
|21||Palladio, The Four…, Preface.|
|22||Typically, in Norwegian the pitched surface of the roof is referred to as the ‘takfall’ (‘the fall of the roof’).|
|23||A.J. Downing, The Architecture of Country Houses, New York 1969, (facsimile of edition from 1850), p. 385.|
|24||Quotation Martial, in MacDonald, The Architecture…, p. 62.|
|25||J. Trier, First, Nachrichten von derGesellschaftder Wissenschaften zu Gottungen, Phil.-Hist., IV, NF III. 4, 1940, p. 117.|
|26||L’Orange, Tempelet…, p. 7.|
|27||R.D. Cramer, Images of Home, A!A Journal, September, Washington D.C. 1960, p. 40-49, p. 42.|
|28||W. Cowburn, Popular Housing, Arena-. Journal of the Architectural Association, September/October, London 1966, p. 81.|
|29||From Royal Court letter dated 4, October 1624, in A. Col left, Gamle Christiania Billeder, Christiania 1893, p. 72.|
|30||Scully, The Shingle…, p. 118.|
|31||Vitruvius, The Ten Books…, p. 14.|
|32||T. Parsons, Societies. Evolutionary and Comparative Perspectives, New York 1966, p. 86, p. 103.|
|33||Downing, The Architecture of…, p. 33.|
|34||See D. Fraser, Village Planning in the Primitive World, London 1962, p. 15. See also E. Camesasca (ed.), History of the House, London 1971, p. 13 (translated by I. Quigly).|
|35||Le Corbusier, Oeuvre…, 1938-1940, p. 97.|
|36||The vestibule into the Biblioteca Laurenziana by Michelangelo (completed 1560), is an example of how an interior can be articulated with elements that had earlier been restricted to exterior walls which faced streets or courtyards. See quotation of J. Shearman, in Sinding-Larsen, The Laurenziana…, p. 215.|