domus | chiesa | libri del borsaio | Iacopo di Talento | prevasariana | Borghigiani 1556 | organo | HALL, The ponte | SMITH, Navata | HAINES | Effigi domenicane

Marcia  B. Hall, The ponte in S. Maria Novella: the problem of the rood screen in Italy,
«Journal of the Warburg and Courtauld Institutes» 37 (1974) 157-73

The renovation that Vasari carried out under orders from Duke Cosimo, beginning in 1565, in the two great Mendicant churches of Florence, S. Croce and SMN (= S. Maria Novella), was intended to modernize them both liturgically and aesthetically. To this end the rood screens were demolished and the friars' choirs levelled and transferred to their present location in the cappella maggiore in each church. The high altar was pulled forward in both cases to make room for the choir behind. In addition, chapels were placed in each bay [= campata] of the nave with tabernacles apparently designed by Vasari, though in the case of S. Croce, Francesco da San Gallo was actually responsible for the design. In SMN Vasari’s tabernacles were, in turn, demolished in the nineteenth-century regothicization of the church and replaced with substitutes thought then to be more consonant with the style of the architecture. What we see today in the aisles are the altarpieces executed by Vasari and his selected younger contemporaries enframed in neo-Gothic tabernacles, c. 1860. Had the nineteenth century really wanted to reconstruct the pre-Vasarian interior, however, it would have had to produce a totally transformed space, for the ponte in SMN, like the tramezzo in S. Croce, was a far more significant and conspicuous structure than has heretofore been recognized.

The Vasarian remodelling was the expression of a view which became increasingly widespread in the last third of the Cinquecento that the rood screens and choirs in the naves served no essential function and impeded the |158| view of the high altar. One of the chief abuses the Council of Trent undertook to correct was the remoteness of the layman from the liturgy. Again and again, the Decrees of the Council emphasized the importance of lay participation in the Mass. To facilitate the layman’s involvement, S. Carlo Borromeo, that zealous Counter-Reformer, advocated in his book on church architecture (Instructionum fabricae et supellectilis ecclesiasticae, 1577) that a church should be designed so that anyone attending Mass would have a clear view of the high altar. As Vincenzo Borghini attested (1585), this was directly contrary to existing conditions. He remarked that one had seen little of the altar even in private chapels where the gates or wrought iron screens were kept closed to separate the clergy and his ministers. Neither Trent nor Borromeo laid down an explicit rule that rood screens should in all cases be demolished. The Council’s Decrees, the fruits of many compromises, rarely stated any absolute prescription, but rather proposed the criteria for judgement. Thus it was that certain tramezzi, late in date of construction and already representing something of a compromise with the Renaissance preference for open and clearly demarcated space, were permitted to stand. Since the Anglican Church took no such position on the desirability of a clear view of the chancel the English rood screens were to a great extent preserved until modern times. In Germany and France, where the heat of reforming fervour was often dissipated in traversing the Alps, some examples escaped demolition. Italy, however, always more susceptible to Rome, went about the business of renovating old churches with far greater energy than its transalpine neighbours. In the rush to conform to the new norms virtually all the Italian rood screens and choirs were swept away. The transformation was effected with such efficiency, in fact, that by the time the urge to regothicize these churches arose, in the nineteenth century, the memory of what they had actually looked like had been obliterated. It is possible painstakingly to reconstruct the appearance of a few, such as S. Croce, where accident has preserved the crucial fragments of a record. In the case of SMN, what has survived nicely complements the visual evidence we have for S. Croce, for it is a verbal description of the church as it appeared before the Counter-Reformation purge.


Pianta di Hall, Renovation
pp. 167-68, 197 Fig. 1

This description is incorporated in the eighteenth-century chronicle of SMN compiled by Borghigiani [= Vincenzo Borghigiani, Cronica annalistica di SMN [1757-61] III, ASMN I.A.30, pp. 330-41; ed. Orlandi, Necrologio II, 397-404]. He is dependent, for the passage |159| that concerns us, upon an earlier author whose writing is apparently now lost, who gives the date of his description as 1556, just nine years before the Vasarian renovation was undertaken. The text was evidently amended in minor details, probably later in the sixteenth century, because it contains scattered references to conditions as they were after Vasari’s remodelling. Our writer begins his tour of the church in the right aisle and moves along the wall towards the high altar (Necr. II, 401). When he gets to the fourth bay he tells us that just beyond the two steps in the pavement began the support of the main wall of the vaults of the ponte. Close to it, under the ponte, was the Minerbetti chapel. A little further along, but still under the vaults, one came to the fake altar of the Gaddi family. Passing through the ponte one exited by the open arch at the rear. When our author has catalogued the decorations of the transept chapels he approaches the ponte from the back and begins his description of the left aisle wall moving towards the façade. Just beyond the tomb of Beato Giovanni da Salerno in the fifth bay he describes the stairs which were attached to the lateral wall of the church and above the side door leading to the cloister. They mounted from the back of the ponte to the organ on top of it.

From this preliminary description, amplified in succeeding paragraphs, we know that this was a deep structure, the vaults of which supported a floor above on which one could walk. In all these particulars, and even in the location of the organ, it resembles the tramezzo in S. Croce as reconstructed in my article. After surveying the decorations of the left aisle our author turns into the nave and moves towards the cappella maggiore from the façade (Necr. II, 402). At the point where the steps rise, he says, we have the façade of the ponte with three large doors corresponding to the three naves and opposite the three doors in the façade of the church. In the plan (Pl. 36d) they are drawn on line with the façade openings which are off-axis in the aisles. Behind, opposite the crossing, were the three open arches of the ponte. Our author makes a distinction between doors on the front and open arches at the back. He is consistent in this usage and the Libro di Billi confirms that there were indeed doors on the front, for it speaks of some tabernacles for relics made by Fra Angelico when he was young for SMN which were ‘tra le tre porte del tramezzo’ (Billi, p. 20, no. 3). On the façade of the ponte were the arms of many of the benefactors who contributed to the construction of the church. He then informs us that the ponte extended about fourteen braccia in depth, or more than eight metres (26½ feet). Compared with most survivin rood screens, or those whose measurements have been preserved, this is surprisingly deep. The jubé in Bourges was only about 2·3 metres deep; that of Chartres, 2·7 metres. The Westlettner in the Mainz cathedral at its furthest extension |161| was 4·16 metres. Even the large-scale tramezzo in S. Croce stretched only seven metres. We might have expected that the ponte would have extended only from the top of the steps through the piers to a depth of 4·37 metres, or about seven and a half braccia. Perhaps our writer was mistaken in his measurement? Careful cross-checking with other sources, however, has definitely elirninated such a possibility. We can be certain that the ponte reached some distance past the fourth piers and into the fifth bay.

At the end of the Cronaca (iii, ASMN I.A.30, pp. 358-65) Borghigiani includes a register of tombs to be found in the church in 1556. He lists them alphabetically by family name and designates the location of each tersely. Fortunately for our purposes, those in the upper church he describes in relationship to the ponte and choir. There exist also Sepoltuarii which describe the tombs as they were before their removal in the 1860 renovation of the church pavement: A.S.F., MS 621 (Sermartelli), and MS 625 (Rosselli). There is also a late copy of Sermartelli, MS 812, containing diagrammatic drawings of the tombs in plan which, although not entirely reliable, can be helpful when used in conjunction with the other sources. From the Sepoltuarii we can reconstruct the location of the tombs listed in Borghigiani’s register. Let its take the tombs in the fifth bay aisles. Borghigiani lists as sotto al ponte the whole line of tombs just beyond the fourth pier, left aisle, i.e. Buondelmonti, Tornaquinci, Anselmi, Vecchietti. The same is true of the tombs in the corresponding area of the right aisle. Sotto al ponte are: Saltarelli, Giambullari, Agolanti, della Scarfa. It is clear, therefore, that the ponte did not end at the fourth piers and that the measurement fourteen braccia is not altogether unreliable.

The next logical point at which to terminate such a structure would certainly be the middle of the fifth bay. Projected on to our plan, fourteen braccia falls about half a metre short of the mid-point of the bay, but we must remember that our author intended his measurement to be only an approximation (he says ‘per Br. 14 circa’). He has given us a further clue to its depth by describing it in relation to the tombs in the pavement. Although these were all removed in the nineteenth century, there seemed to be enough information in the Sepoltuarii to enable us to reconstruct their placement and from that reconstruction to make a more precise estimate of the ponte’s depth. This procedure at first failed and there seemed to be no way to make sense of this description until I discovered that an error of transcription by Orlandi had reversed the sense of the original text. Our author actually tells us that the ponte extended from the steps to the end of the first file of marble sepulchres in the floor, that is, as far as the beginning of the pavement that has no tombs.

The correct reading is:

Orlandi, Necr. II, 403

 ed. EP

e stretto quanto è dagli scalini alla fine delle prime file de sepolchri di marmo stessi sul suolo, cioè fino al principio del matonato, che non ha sepoltura, quali lapidi sepolcrali tornavano sotto al ponte (p. 161, n. 13).e stretto quanto è dagli scalini alla fine del mattonato, che non ha sepoltura, quali lapidi sepolcrali tornavano sotto al Ponte.e stretto quanto è |339| dagli scalini alla fine delle prime file de’ sepolchri di marmo stesi sul suolo, cioè fino al principio del mattonato che non ha sepolture; quali lapidi sepolcrali tornavano sotto al ponte.

Our diagram of the tombs is based upon one found in the nineteenth-century Sepoltuario, but the placement of tombs in the fifth bay of the nave and the row below it in the fourth bay has been corrected to correspond to the written descriptions in the Sepoltuarii (Fig. 1, in p. 160). Those records inform us that there was only a single row of tombs in the fifth bay of the nave and that it was in the lower part of the bay.

There was a line of five tombs in the nave: Boni, Rucellai, Strozzi, Ilarioni (Bardi), Leali. They are described thus in Sermartelli (A.S.F., MS 621, fol. 71r): ‘Nella nave del mezzo rincontro alla porta di chiostri e quella del fianco sono cinque monumenti in fila con lapidi grandi di marmo, e chiusini simili con arme e lettere. Il primo a cominciar di verso I’organo e di Boni. . . .’ The fifth tomb, that of Leali, is described as ‘rincontro alla capella o quadro di S. Pier Martire’ (fol. 71v). (The fourth pier on the right was known as the pilastro di S. Pier Martire, see Richa, iii, 1755, p. 79, and Paatz, iii, p. 737.) The location of this line of tombs is clear. Its placement in relation to the door of the cloister indicates that it was in the lower half of the fifth bay just beyond the fourth piers. In fact, it continued the line of tombs in the aisles. Rosselli (MS 625) described a line of tombs as if it were continuous from Petrucci, Leali, Giovanni (blank), Strozzi, Rucellai, Boni, Agli, Vecchietti, nos. 203-10. This we are able to establish despite the confusing error by the author of the diagram from which our Fig. I was made (MS 812, fol. 126). He misplaced the whole row Boni-Leali in the second half of the fifth bay, precisely where there should be an area of pavement that had no tombs.

The ‘pavement that has no |162| tombs’, the point to which the ponte extended, is roughly the upper half of the fifth bay. The tombs overlap, however, the mid-point of the bay, <?> hypothetical terminus for the ponte. Perhaps this was the case, and our authour statement that the ponte extended as far as the end of the first file of sepulchers was, again, an approximation. Or it could be, and this is indeed probable that the nineteenth-century diagram is only approximate, especially w<e> regard to the dimensions of the stones, and that the tombs did not reach <as> far as is indicated. It is impossible to fix with any greater precision <the> terminal point of the ponte on the basis of this kind of evidence. This <fa?> highlights the importance of exploring for the foundations which, like those <of> S. Groce, certainly lie beneath the present pavement. It is probable that with that <o?> whit that kind of new evidence will it be possible to resolve the questi<on>. However, we cannot be far from the truth in asserting that the ponte was ind<eed> a surprisingly deep structure, stretching from the steps probably to <the> middle of the fifth bay.

Our writer tells us that when one comes out from under the ponte <o?> traverses as much space as there is pavement without tombs before comi<ng> to the door of the choir (Necr. II, 403). It has always been assumed, as it was assum<ed> about S. Croce, that ‘ponte’ was an inexplicably curious designation for the choir screen, or at least for the back wall enclosing the choir, but <our> description makes it clear that there were two separate structures. The choir, our author says, was enclosed with a high wall on three sides open <on> at the high altar side. It was arranged so that the laymen could not se <the> friars in the choir without actually going up the steps into the cappella maggiore. Inside were the inlaid wooden choir stalls in two tiers as well as two lecte<??,> one for chanting the Mass and the Antiphons, the other for reading <the> lessons. These must have been what was regularly used in the liturgy, to <??> replaced by the procession to the top of the ponte only in celebration of special |163| feasts. (In a passage of Borghigiani’s Chronicle not transcribed by Orlandi some of liturgical uses of the ponte are described. This passage is quoted as part of a discussion of the function of rood screens in my article on S. Croce). There is no specific mention of an ambo above the ponte. We are told, however, that a parapet about half the height of a man ran from one side to the other. It was for this reason that it was called the ponte: its length (the whole width of the church) was much greater than its width and it therefore resembled a bridge.

As for its height, he tells us that the upper storey was about one braccio above the door that goes into the cloister. The present lintel of this door is not the original one, nor even Vasari’s, but was remade in the nineteenthcentury remodelling. We can see from the cloister side that the door today reaches the top of the arcade and could at no time, therefore, have been any higher. It is possible, however, that Vasari raised it to its present height. As we have no evidence on this point we can only take the measurement of the present door, realizing that it may not accurately reflect conditions as they were in the pre-Vasarian church. The top of the door is today 2·95 metres above the floor. Adding a braccio, we have a height of about three and a half metres, or about eleven and a half feet, for the floor of the second storey. If we assume that the parapet rose above this, the total height may have been in the vicinity of 4·5 metres. Compared with the larger church of S. Croce, where the top of the parapet was 5·8 metres high, this would appear to be in the right order of magnitude.

It has long been noted that the two bays of the nave nearest the high altar were constructed shorter than the other four, and many hypotheses have been offered to explain why this is so. With a structure of the dimensions of the ponte rising in front of those bays it is evident that the difference would not have been discernible. Our ideas have been so thoroughly reformed by the post-Vasarian concept of the space of these churches that we fail to recognize that the medieval designers regarded it as two churches: one for the friars and one for the laymen. Architectural differences between the two churches occur frequently. In S. Croce, for example, there is a change in the fenestration behind the tramezzo where the windows are doubled. We usually attribute such differences to an arbitrary change in the design, but it is more probable that there was a very pragmatic reason for it here, namely the need for more light in the friars’ choir for reading. The length of the bays in SMN may not have been so arbitrary, either, as we have heretofore believed, for an interesting relationship appears between the two bay lengths when the ponte is drawn in on the plan. The fourth bay (the first long bay), measured from the pier to the façade of the ponte, is equal in length to the shorter bays behind. The recovery of rood screens where they were an original part of the design of the church, as they were in both SMN and S. Croce, may provide rationales for variations in design which previously have been taken as evidence of a curious medieval haphazardness and willingness to change the design of a church without apparent reason.

The friars’ church formed a unit complete unto itself. The high altar, concealed behind two solid walls, could not have been visible at all from |164| the laymen’s church. The division would have been even greater in SMN than in S. Croce, for there must have been more solid wall in the ponte (as the use of the term façade to refer to its front implies), than in the tramezzo of S. Croce. Unlike the arrangement in the Franciscan church, the front was not made up of a series of privately endowed chapels but of more or less solid wall punctuated with three doors, also apparently solid, and decorated, probably in a handsomely random fashion, with coats of arms and family imprese. Its resemblance to a ponte must have been more from the back where the three open arches resembled the arches of a bridge. Even the new painting for the high altar, executed by the Ghirlandaio brothers in the 1490’s, would have been completely hidden from view from any point in the layman’s church. In fact, if we assume the ponte to have been in the range of 4·5 metres high, calculations show that in order for a man standing in the door of the church to see an object in the cappella maggiore, it would have to be elevated 6·8 metres above the level of the lower church. As we can see in the eighteenth-century engraving showing the installation of Ghirlandaio’s altarpiece, the high altar was raised seven steps. Even adding the twostep rise at the front of the ponte, this would elevate it only about two metres. One wonders, then, why the curious platform base raising the panel about two metres above the altar table (still far short of the height required to make the picture visible outside the rood screen) was constructed, unless it was to make it visible over the choir enclosure (perhaps three metres high) to someone standing on the church’s axis at a time when the central door of the ponte was open.

The arrangement of chapels on the ponte was quite unlike S. Croce. There were eight chapels in all, four on the upper storey. Though we know of few instances in Italy of chapels being placed above, this in itself is not unique. Doberer points out that there were two such chapels in the loft of the Lettner in St. Stephan’s Cathedral, Vienna. The arrangement of chapels on the lower level is not at all what we might expect. Two were against the lateral wall of the church: the Minerbetti, whose chapel is still in this location, occupied the right aisle wall under the ponte. The chapel of the Compagnia di S. Caterina was beneath the organ in the left aisle. The other two, those of the Cavalcanti and the Castiglioni, were attached to the nave piers. The former is said by Borghigiani to have been founded here at the time the pier |165| was built.

This information is given by Borghigiani, iii, p. 400, in another passage not transcribed by Orlandi. The Cavalcanti retained rights to the pier even after the ponte was demolished, for Borghigiani tells us further that in 1598 when the Benedetti family placed the Buontalenti-Cigoli shrine to St. Peter Martyr on the pier they had to ask the Cavalcanti permission to mount a plaque beneath it with their names.

One would have expected these altars to face out, as those on the tramezzo in S. Croce did, but our author tells us, rather, that they faced each other. Any doubt which might have lingered concerning their orientation is resolved when we note on our diagram that the family sepulchres which had been at the foot of each altar are laid crossways in the church (Fig. I). As we can see, the other tombs, with a few scattered exceptions, are placed longitudinally, their heads towards the high altar, whereas the Castiglioni tomb heads towards the left aisle, the Cavalcand towards the right aisle. Clearly then, the altars must have fitted into the space between the ponte’s façade and the pier. It is difficult to imagine, given this disposition of chapels, where were the laymen’s altars considered indispensable by scholars of northern rood screens. These chapels on the ponte would have been enclosed and concealed from view. Our writer tells us, in fact, that especially the chapels of the ponte, but also those throughout the church (except the cappella maggiore), had wooden gates and were closed with a key (Necr. II, 403).

Nothing here suggests the arcade across the front of the rood screen that we see in S. Croce, in Vezzolano and found indicated by the documents at the Santo in Padua. The ponte was not a loggia-type screen. In terms of Doberer’s typology it comes closest to the closet-type screen (Schrankenlettner), i.e. the second storey supported on a closed understructure, of which the thirteenth-century screen in Notre-Dame de Valére at Sion (Switzerland) is an extant example. The jubé at Sion is two solid walls in the nave pierced by the central door supporting a stage above. The closet thereby created within is filled on one side by the staircase that rises to the loft. The Sion church is much smaller than SMN and the jubé is accordingly only 2·16 metres deep. There is scarcely space, therefore, for private chapels underneath it as we have found them at SMN and, possibly for this reason, there are none. But the ponte differs from Schrankenlettner in a significant feature, for the ‘closet’ of the former was quite open at the back. Although we cannot know how wide the arches on the back of the ponte were, our writer’s description indicates something quite unlike the solid wall behind that we see at Sion. At the present time, however, we know of no rood screen |166| which more closely resembled SMN. The interesting question of the origin of the form of the ponte must await future discoveries.

(… non riprodotte pp. 166-72 …) 

(|172| ...) Rood screens today are generally dismissed as later insertions into the church, interrupting the flow of space intended in the original design. Viollet-le-Duc (1868) appears to have launched our present-day attitude. Unlike his predecessors and contemporaries, he regarded the rood screen as an unprecedented innovation, occurring first around the middle of the thirteenth century, which bore no relationship to the ambo, pulpit, chancel screen, iconostasis, or any other early Christian element. His argument, which smacks |173| of the French Revolution’s secularist polemic, is that the bishops threw up the jubé, took refuge and entrenched themselves behind it when King Louis put an end to their dreams of dominion in the temporal realm by depriving them of jurisdiction over secular cases. For Viollet the screen was a social barrier and therefore a symbol of ecclesiastical pretentiousness. It is strange that this attitude, so conspicuously the product of nineteenth-century anticlericalism, should still prevail today. Yet Paul Frankl dismisses rood screens with the statement, ‘Their purpose ... was to express a system of social strata.’ For him they were for the most part late insertions which ‘destroy the open view originally intended’. John White does not break with this tradition when he speaks of Giorgio Vasari’s clean-up campaign in S. Croce as a restoration of its architectural unity. We have shown in the case of S. Croce that, quite to the contrary, the screen was planned for at the time the nave piers to which it attached were built. (The bases of the piers through which the rear wall of the tramezzo ran were constructed at different levels front and back, indicating that they were never intended to be seen without the tramezzo wall which masked the difference in height.) Whenever steps occur in a church’s floor at the point where the rood screen rose (as in SMN) it can be assumed that the screen was intended from the start.

For generations we have taught our students that these barn-like structures were designed with open flowing space, nothing interrupting the view of altar and pulpit, because they were preaching churches and had to accommodate large crowds. While it remains true that the Mendicant churches were large, they were only half so large, from the point of view of the lay congregations they could hold, as we have thought. Nor were they so unencumbered as we have asserted. When we recognize that many of the screens in Italian Mendicant churches, and specifically the Florentine ones, were integral parts of the church, we must look again at these spaces. If we imagine a solid partition rising four or five metres in the middle of SMN, or a Gothic screen reaching up with its pinnacles almost seventeen metres in S. Croce, the whole proportion changes, to this observer decidedly for the better. The nave arcades look less as if they had been raised on stilts and more as if the design had been derived from the classical canon. The ponderous bareness of the Italian churches, for which they compare unfavourably with the aspiring northern Gothic, is relieved and lightened. With apologies to Vasari, I believe that the interiors with rood screens in place may have made an architecture of aesthetically more satisfying proportions.

Tyler School of Art, Temple University