The medieval mason was no formalist. He felt free to alter the elements of the fan vault in order to accommodate the design requirements or his own taste. For example, in the chapel of King's College the vertical ribs appear to be equally spaced, but actually there are small discrepancies resulting from the fact that the decision to fan-vault the ceiling was made after the building had been partially erected. Sherborne Abbey in Dorsetshire has beautiful examples of fan vaulting that were quite influential among 15th-century designers. The conoids in the chancel (the area that includes the altar) have horizontal sections that are polygonal rather than circular. Moreover, the conoids are constructed of separate ribs and small flat panels rather than jointed masonry. nevertheless. In both King's College and Sherborne Abbey the ceilings must be described as fan vaulting because of their overall appearance, in spite of the deviations from the definition of an idealized fan vault.
The construction of fan vaults can be divided into three periods beginning with the construction of toy fan vaults in tomb canopies. The first period ended with the completion of the cloister of Gloucester Cathedral in 1412. Between 1412 and 1430 no major fan vaults were built. This was a time of labor shortage, high taxes and economic depression; fan vaulting was costly and it is likely that no individual or institution could muster the capital needed to undertake construction. The design of the Sherborne Abbey chancel vaulting in the late 1430's marks the beginning of the second period, which lasted until about 1475. In the third perlod, from 1475 to about 1540, many of the largest and most important fan-vaulted ceilings were constructed.
The overwhelming majority of fan vaults were built in ecclesiastical buildings, and a considerable number of these were constructed in chantry chapels. A chantry was a fund established to pay for masses for the soul of the founder; the chantry chapel was where the masses were said. Most of the early chantry chapels were endowed by noblemen and were intended as much to call attention to the greatness of the founder as to ensure the safety of his soul. The fan vault with its striking visual form and intricately carved tracery was well suited to the purpose.
In the second half of the 15th century the taste for fan vaulting spread to the middle class and to the King. St. George's Chapel at Windsor Castle, which was a royal institution, has fan vaults dating from the 1480's. Shortly after 1500 work began on the most magnificent of all fan vaulting: the ceilings in the Chapel of Henry VII at Westminster Abbey, which Henry conceived as a huge chantry chapel for the Tudor dynasty. Hence the church, the nobility, the middle class and the crown all contributed to the diffusion of the new architectural form.
The spread of fan vaulting was accompanied by an increase in the scale of the projects. The vaulting completed about 1380 in Trinity Chapel at Tewkesbury Abbey spans 1.7 meters across the hall. The main vault of the King's College chapel, which was completed in 1515, spans 12.7 meters. This is the longest span of any fan vaulting. One reason it took English builders more than 150 years to increase the span of the vaulting to its maximum is that they had little conception of how a fan vault works.
Twentieth-century engineers utilizing advanced mathematical tools have shown that for the conoid of a fan vault to be in equilibrium it must be supported along all its edges. A substantial weight must apply a compressive load at the upper edge along the horizontal bounding rib that separates the conoid from the spandrel. The load is provided by the spandrel itself, which is a very heavy stone plate. The large bosses, or raised decorative stones, that constitute the center of the spandrels in the King's College chapel weigh some 1,400 kilograms each.
The spandrel serves as the keystone of the vertical arches in the conoid. As in the ribbed arch, the thrusts are directed outward and downward. The downward thrusts are conveyed to the bottom of the conoid, where the structure rests against the wall. At this point the conoid is supported by the tas-de-charge, a projection built out from the wall in the corner of the bay. As noted above, the conoid is composed of concentric courses radiating upward from the springing, with each course having the form of an arch whose alignment is approximately horizontal. The outward thrust is conveyed through the horizontal arches to the wall, which provides an opposing thrust. Thus the conoid is compressed between the spandrel, the walls and the tas-de-charge.
The stresses in the conoid tend to be distributed fairly evenly rather than concentrated in the ribs. The specific way the stresses are distributed has important implications for the stability of the conoid, which, as I have noted, is a shell structure. An architectural shell is defined as a stressed structure much thinner than it is broad. The thrust acting on each point of the shell can be represented by a vector with components directed outward and downward: the set of all such vectors forms an imaginary surface called the thrust surface. Mathematical analysis shows that for the conoid to be in equilibrium the thrust surface must lie within the physical surfaces of the conoid shell. If the thrust surface emerges from the surfaces of the conoid, the structure is likely to collapse.
The builders of fan vaults had no way of knowing this fundamental fact. The analysis of stresses in a shell was far beyond their theoretical capacities. On the basis of long experience, however, the masons developed several techniques that greatly increased the stability of the vaulting. The narrow volume at the base of the conoid above the tas-decharge is termed the vaulting pocket. In many fan-vaulted ceilings the pocket was filled to a height of about a meter with a solid rubble made up of stone chips and cement. In the chapel of King's College, the rubble reaches precisely the same height in each conoid, indicating that it was a planned element of the structure rather than a casual afterthought.
The presence of the rubble fill has three significant consequences. First, the thrust surface extends into the rubble fill, which conveys the thrust to the wall. In this way the thrust is distributed over a much larger area of the wall than it would be if the shell alone transmitted it to the wall. Second, the rubble acts as a weight cantilevered over the floor. The weight opposes the outward thrust of the conoid reaching the walls in the corner of the bay and thereby further reduces the stress put on the walls.
Most significant for the stability of the conoid, the rubble fill reduces the span of the vaulting that functions as a shell structure. The rubble cannot act as a shell because it is a substantial three-dimensional form rather than a thin sheet. Since the vaulting thrust passes into the rubble, the lower part of the conoid does not act as a shell.
CHAPEL OF HENRY VII in Westminster Abbey has the most magnificent of all fan vaulting. Work on the vaulting was begun in about 1500. The ingenious design of the ceiling combines frame structures and shell structures. Substantial transverse arcbes support large pendants near the wall of the bay. Tbe conoids are built over the pendants. At the pendants the transverse arches run up tbrough the vaulting and along its upper surface, so that the arches are not visible in the central part of the ceiling. Most of the weight is taken up by the arch where it passes through the conoid at the pendant. The arch functions as a frame member and the conoid as a shell structure.
PLACEMENT MARKS on the stones for the vaulting of King's College chapel enabled the stone setters on the building site to complete the ceiling by merely following the marks. The symbols were carved in the worksbops where the vaulting was prefabricated. The large circular mark at the center of the stone indicates the quadrant of the bay in which the stone was placed. Tbe marks at the ends of the stone next to the joints are the Arabic numerals 3 and 4. They indicate the horizontal level in the conoid where the stone was to go, the position in that level and the timing of insertion. The placement marks can still be seen on the upper surface of the vaulting stones, as is indicated by the plan view of one quadrant of a hay in the main vault.
First Stage of Vaulting
Second Stage of Vaulting
Third Stage of Vaulting
Fourth Stage of Vaulting
Stresses in Vaulting