by Unknown
7.1. Armature of relieving arches embedded in the Pantheon. (Drawing Mark Wilson Jones and Robert Grover)
The Pantheon stands today, triumphant, yet the grandness of its ambition tested the Roman building machine almost to the point of disaster. It is impossible to say exactly how close failure came; all we can do is witness signs of structural distress and constructional difficulty. These are particularly notable in three respects: the long cracks that fracture the rotunda at intervals; the building complex that butts up against the rotunda to the south, the very existence of which speaks of emergency; and the curious mismarriage between rotunda and portico. While not in itself a structural issue, this, it seems, was bound up with problems in obtaining the massive column shafts originally intended. This essay looks into these matters and reflects on the drama of the Pantheon worksite. As I see it, the project was balanced on a knife edge between success and failure. It was success that prevailed – but at the cost of perfection compromised.
Cracking, Concrete, and Centering
The rotunda displays an array of more or less vertical cracks. Typically, they run from about halfway up the rotunda to halfway up the dome. The mapping of these cracks in the 1930s by Alberto Terenzio during restoration works gives us the best idea of their scale and frequency (Fig. 7.2).4 Many can be seen in old photographs, though today most are obscured by surface finishes or by modern repairs. Nonetheless, parts of some can still be seen on the outside of the rotunda, while one can be traced in the staircase on the east side of the entrance. The largest crack coincides with the main axis on the south side, measuring up to 7 centimeters in width where it can be accessed from behind the rotunda.
7.2. Interior elevation of the rotunda, projected flat, showing the principal cracks in the structure. (Wilson Jones 2000, Fig. 9.21a; drawing by Ippolita D’Ayala Valva, after A. Terenzio)
Cracking is important for the way the dome behaves in terms of statics. It performs less like a modern monolithic shell reinforced with steel and more like an array of tapering sections of masonry comparable with segments of an orange. This is consistent with the effect of outward lateral thrust and hoop tension, both being characteristic of unreinforced domes with profiles based on arcs of circles (as opposed to catenary curves, which offer a reduction in tensile forces).5 Deformation of the section resulted, with the interior of the dome no longer matching an ideal hemisphere. Recent survey work with laser scanners conducted by the Karman Center reveals that the crown has slumped by around 1½ feet (45 centimeters) with respect to its presumed original hemispherical form. In relative terms, this equates to only about 1 percent of the total height, but in real terms, it still represents a significant shifting of stress and mass. A further cause of structural distress was settlement, for the Pantheon rises not on rock but on clay.6 The floor level slopes down by almost 40 centimeters from the front of the portico to the far side of the rotunda (see Plate XXI).7 Most of the cracking would have occurred early in the life of the building, though not necessarily all at once when the centering used to shape the dome was dismantled. Some cracking seems to have arisen even before the dome was put in place, as we shall see.8
XXI. Section combining information from excavations under portico and rotunda, with sloping floor of Agrippan Pantheon shown in dashed line and XXXX shown in solid line. (Pier Olinto Armanini in Beltrami 1898, Fig. XV)
It would be a mistake to make more than a casual analogy between Roman lime-based concrete and its modern cement-based equivalent, which is typically poured in one go in a relatively liquid state (with formwork initially supporting its entire weight).9 Modern concrete includes steel to provide tensile strength and combat cracking. By contrast, the Romans addressed performance by varying the density of aggregate, by incorporating relieving arches, and by manipulating the cross section. In fact, in understanding the dome of the Pantheon, it is crucial to distinguish between the upper half, which is a relatively thin shell, and the lower half, which is much thicker and has a quite different profile (see Figs. 1.12 and 6.3). The lower part of the dome is as thick as the drum and contiguous with it; at the crowning cornice of the drum, the section reduces (making space for a perimeter walkway), thereafter diminishing in a series of large steps until the start of the upper shell. None of this can be perceived inside the building, where the visual effect is largely determined by the coffering that follows its own independent geometry.
Construction of the lower part of the dome proceeded by stages, as for the rotunda wall. Ring followed ring, each diminishing in diameter, typically in lifts of 5 feet or so. The concrete was laid relatively dry, in more or less horizontal strata of mortar and aggregate in predominantly fist-sized pieces. Each stage would have been allowed to cure substantially before the next was added. By virtue of closing in on itself, each ring, once complete, could support not only itself but also the next ring slightly smaller in diameter, and so on, creating in effect a kind of corbeling. Higher up, after the top of the step-rings where the section is thinner, the vault was flatter, therefore demanding some kind of support up until the time when the concrete set (or “went off”). Finally, at the very top, the device of an oculus represented a wonderful solution: avoiding construction, lightening the dome while lighting the finished space, besides contributing to its symbolic mystique.
It is theoretically possible to build a lime-based concrete dome without any temporary support.10 But as regards Roman practice, there is plenty of evidence for the use of formwork; witness the imprints of wooden boarding on the vaulting of many a ruin, including Nero’s Golden House (the Domus Aurea), Trajan’s Markets, and Hadrian’s Villa. It is, unfortunately, impossible to obtain this kind of information for the Pantheon, for the rendered internal surfaces of the dome that we see today are the result of only the latest of a number of restorations, some of which date from a time before it had become customary to document the existing state prior to the commencement of work. The possibility of self-support can be reconciled with the use of formwork if we suppose that a temporary wooden assembly provided structural support for the upper parts alone. For the lower parts, the prime function of the formwork, just as its name suggests, was only to mold the form of the concrete. This was necessary for the geometrical precision of the Pantheon coffering, which creates such a magical dance of chiseled planes of light and shadow.
How was the wooden formwork itself erected? Some authorities opt for a system supported from the ground for the full width of the interior. William MacDonald visualizes “an immense hemispherical wooden form, supported by a forest of timbers and struts.”11 Recoiling at the consumption of trees on such a scale, others have imagined centering “flying” across the entire space without vertical supports. Proposals in this vein include those put forward by Eugène Emmanuel Viollet-le-Duc (Fig. 7.3, left) and most recently Rabun Taylor (Fig. 7.3, right).12
7.3. Proposals for the centering used to construct the dome. (Left, Viollet-le-Duc 1875, p. 475; right, Taylor 2003, Fig. 120)
Any uniform system of centering, however, seems to be contradicted by the marked difference between the lower and upper halves of the Pantheon dome that has already been highlighted. Accordingly, I visualize a very substantial wooden tower rising from a doughnut-shaped plan, with a ring about 11.5 meters wide (Fig. 7.4).13 The upper portion of the dome, being relatively thin, was light enough to have been carried on such a timber structure.14 Nothing, of course, was needed underneath the void of the oculus, nor indeed for the lower portion of the dome. Since this could be raised by corbeling, all that was required here was non-load-bearing formwork. This part of the wooden assembly alone would have “flown,” spanning a gap of 5 to 6 meters between the drum and the doughnut-shaped tower.15
7.4. Schematic cross section showing extent (in gray tone) of a hypothetical doughnut-shaped centering tower for constructing the dome, 1:600. (Drawing Mark Wilson Jones)
The quantity of timber consumed must have been considerable, but not beyond the Romans’ capabilities. They had a
t their disposal extensive forests of oak and sweet chestnut not far from the capital, and are known tohave employed very large timbers, for example, for the trusses spanning the 25-meter-wide nave of Trajan’s majestic Basilica Ulpia. In a treatise on the construction of siege structures, the Poliorcetica, this emperor’s architect, Apollodorus, expounded on the assembly of giant towers using small timber members. The surviving copy, which dates to the Middle Ages, has illustrations that convey an almost naive impression, but the original versions may well have been more precise and technical.16
By virtue of experience, imperial architects must have been aware that domes pushed outward at the haunches. Three main counterstrategies were adopted in the Pantheon. The first is a very thick supporting wall (thicker than simple vertical loading would require), which, to save materials and weight, was hollowed out by voids in the form of exedras and chambers (see Chapter Four). The second is the carrying up of the drum to a higher level than the springing of the dome, thus creating a mass of weight resistant to lateral movement. The third is the most obscure in its functioning: the vaults and arches embedded within the concrete known as relieving arches, which were made using tile-shaped bricks, mostly 2-foot-square bipedales with some 1.5-foot-square sesquipedales. As already noted, the Pantheon boasts the most elaborate known arrangement of relieving arches (Fig. 7.1). In part, they served to direct loadpaths to points of greatest strength, the eight “piers” of the rotunda plan. They also facilitated constructional processes, an important consideration for Roman builders. Since brick and mortar cured faster than concrete, the use of relieving arches enabled work to proceed upward faster than would otherwise have been the case.17 Although it is hard to know the full range of the ways in which they work – or were thought to work – we can still judge them, almost two thousand years later, wonderfully efficacious.
The Grottoni
In spite of these strategies, the stability of the Pantheon was not a foregone conclusion – indeed, it was evidently a matter of great concern for the builders. This is demonstrated by the annex of structures sandwiched between the rotunda and the adjacent basilica to the south (Fig. 7.5, and see Fig. 6.5). Parallel walls and associated floors and vaulting delimit a series of spaces on two levels that are collectively known in Italian, rather suggestively, as the grottoni. Above them, on the main axis, a solid brick arch supported a kind of bridge connecting the basilica with the rotunda. This whole complex constituted, in effect, a gigantic buttress, as may be deduced from the lack of any obvious ceremonial or utilitarian purpose, along with the crude fashion in which it butts up against the rotunda.18 Indeed, it is plain to see that the lower parts of the grottoni are not bonded with the rotunda.
7.5. Rotunda viewed from the south, above the grottoni. Note the scarring (particularly evident at and above the level of the three openings visible in the middle of the photograph), which testifies to the presence of a lost connection or “bridge” with the basilica to the rear (south) of the grottoni. (Photo Gene Waddell)
It is generally assumed that the grottoni were created after the completion of the rotunda, as an improvised post facto countermeasure to resist its outward pressure. However, my own observations suggest that work on the grottoni began relatively early. The key here is the connection between the rotunda and the structure overhead. Instead of casually butting up to the rotunda as occurs at low level, the “bridge” has a cornice that meets the middle cornice of the rotunda at a bonded miter, or, in other words, in a premeditated relationship.19 The springing of the arches of bipedales is integral with the rotunda, as shown by photographs taken at the time when parts of the grottoni were rebuilt, and as is still observable at high level (Fig. 7.5). Where fabric associated with the bridge has been lost, the drum is not smooth, as we should expect had an extraneous construction simply fallen away. Instead, material integral to the rotunda has been pulled away.20 Relieving arches belonging to the bridge once thrust into the rotunda, hence the faces of bipedales that are now exposed to view. Despite the lack of bond below, then, at high level construction is all of a piece.
This rather extraordinary state of affairs suggests, firstly, that the grottoni were initiated after the drum had risen to around a third of its height, and, secondly, that they were built speedily so as to catch up with the drum. This occurred before the dome was begun (or, at any rate, before it curved inward to a significant extent). All this suggests that the grottoni were built very fast. A rapid pace of work is attested at Trajan’s Baths by dates inscribed in red pigment on brick-faced concrete walls of broadly comparable width with those of the grottoni; the dates indicate that over a period of around two and a half months, one wall rose by an astonishing 15 meters. The vast substructures of the baths, comprising many other walls equally tall, were probably executed in a single season.21 We can only speculate how fast the grottoni were built, yet bearing this comparison in mind, a couple of years or less is not out of the question.
It seems, therefore, that the grottoni respond to a problem that occurred early, before the addition of the dome. The nature of the problem is suggested by the huge crack, already mentioned, that fractures the rotunda approximately on the main axis, where the wall defining the apse is at its thinnest. Unlike other cracks, which tend to peter out earlier, this one reached floor level.22 The cause could be settlement of the foundations, although this cannot be proved without a geotechnical investigation. In short, the grottoni were built so as to minimize the further movement expected when the thrust of the dome came into play. The intervention can be judged a success; despite the alarm it registers the Pantheon stands.
The Connection between Rotunda and Transitional Block
Different problems affected the north end of the Pantheon where its three main parts meet: rotunda, portico, and the structure in between. This is known in Italian as the avancorpo, and, rather less elegantly, as the “transitional block” or “intermediate block.” As noted in the Introduction (Chapter One), over the centuries the relationship among these three parts has provoked markedly contrasting interpretations. Traditionally, the explanation was thought to lie in (various different) phasing sequences, with the rotunda usually being presumed to have been built before the rest.23 Bound up as it is with perceived compositional shortcomings, the debate has an inevitable subjective component, and so it makes sense to address objective constructional realities first.
The junction between rotunda and transitional block can best be observed in the two staircases on either side of the entrance. Unlike other parts of the building, there is no marble revetment here to hinder inspection, while the stairs facilitate access for the entire height – an enormous practical advantage for the purposes of study. Achille Leclère, one of the long line of prize-winning architects awarded a period of residence at the French Academy in Rome, included a small-scale survey as part of his envoi of 1813 on the Pantheon.24 Otherwise this part of the building has been neglected, leading me to make a new survey in 2005 and 2006 of the east stair, the better preserved of the two, yielding the drawings illustrated in Figures 7.6, 7.7, and 7.8.25 These also help locate pertinent details, such as the soundings, or saggi, made by previous investigators where the rotunda wall (labeled A) meets the side walls (B and D) of the transitional block.
7.6. Pantheon, east stair, section. (Drawing Mark Wilson Jones and Robert Grover)
7.7. Pantheon, east stair. (Drawing Mark Wilson Jones and Robert Grover)
7.8. Pantheon, east stair, plan. (Drawing Mark Wilson Jones and Roberta Zaccara)
Today, the east stair is entered from one of the two great apses of the portico, the ancient doorway on the flank having been blocked up. The stairs have suffered reconfiguration at the top and bottom, but otherwise remain essentially unchanged. The trapezoidal plan makes six full turns plus an extra seventh flight against the curved wall of the rotunda. They afford access to several different parts of the building: to the semicircular chambers in the drum on three levels, to the suite of rooms fronting the tr
ansitional block occupied by the Virtuosi of the Pantheon, to the entablature of the portico, to the middle cornice of the rotunda, and finally to the roof (Figs. 7.6, 7.7, 7.8, 7.9).
7.9. Junction of the rotunda, transitional block, and portico on the east side, at high level. (Photo Mark Wilson Jones)
Inspection of the staircases shows that the rotunda and the transitional block are united at low level, but disunited at high level. It seems that both rose as one until somewhere in the region of 12 to 14 meters from the floor of the portico. From then on, work evidently proceeded on the rotunda alone, pending the completion of the transitional block.
At high level, the disjunction is obvious to the untrained eye. Wherever the rotunda is exposed to view it presents finished surfaces that can only have existed if it were built first (Plate XXII). Since the Pantheon stands intact and not exposed for study like a ruin, the unity of the lower parts is less glaringly evident, yet nonetheless inescapable. A key piece of evidence is a sounding, or saggio, located on the second short landing of the west stair, at the junction between the rotunda and the transitional block (see Plate XXIII). The ample view it offers into the “guts” of the fabric (the sounding reaches 63 centimeters deep) reveals no gap, crack, or joint, and the mortar traverses uninterrupted. In addition, there is a course of bipedales that passes unbroken from one part to the other, including a whole bipedalis right where they meet. It would have been quite impossible to insert so large and brittle an element after the original construction.26 So both the rotunda and the transitional block rose together at low level, although about halfway up, construction advanced on the rotunda while that of the transitional block was held back.