# The Curvature of the Lateral Chapels in San Carlo alle Quattro Fontane

Michael Hill National Art School, Sydney

Abstract Leo Steinberg exposed the complex geometry of Borromini’s San Carlo alle Quattro Fontane (1634-41) in his doctoral thesis of 1960, and despite some reservations (Connors argues that many of the drawings we see today were reworked after the church was built) his scheme has been the basis for later interpretations. Nevertheless, something is missing. Steinberg, like others before, presumed that the plan begins with the diamond shape of the two triangles. From here one can follow the formative steps, with one important exception – the rationale of the inner arcs of the lateral chapels remains unclear. They are not segments of a circle, parts of an oval, nor freehand arcs. The problem arises because the equilateral diamond is not the beginning of the plan, but stage two. The easiest way to construct such a diamond is to interlock two circles of the same radius, with the centre of one placed on the circumference of the other. The resultant cell provides not only the co-ordinates of equilateral triangles but also the dimensions of the plan. And, most significantly, its segments provide the curvature of the lateral chapels. Before any symbolic significance (touched upon at the end of the paper), the interpretative advantages of this proposition are formal, providing a key to unlocking the developmental sequence of Borromini’s spatial planning.

Introduction: The Importance of Geometry∗ In 1634 the Spanish Trinitarians commissioned Francesco Borromini to design a new convent at the four fountains intersection in Rome, dedicated to San Carlo Borromeo and the Holy Trinity. Despite a tiny site and barely secured funding, a monastic ensemble was soon raised that would be the envy of every religious order in the city.1 The ingenuity of the church in particular has been the subject of much debate, with discussion often skirting the topic of the nature of the Baroque style itself. What does it represent, a cross or an oval? How should its shifting

Steinberg himself had been suspicious of the geometrical scheme he so cogently outlined. perpendiculars erected over their sides. Steinberg highlighted the importance of Alb. yielding the foci – and the short segments – of an inscribed oval. He summarised this as follows: 1) Two triangles with shared base. 4) Semi-circular chapels in the long axis articulated by four columns.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland. 172.5 For Connors. the prior of the Trinitarians had mentioned how often Borromini was harassed to explain his plan) of the church’s recondite armature. 173 was diminished when Joseph Connors persuasively argued that it and others (including Alb. in which the constructional lines are all but effaced. even the crystalline geometry of Alb.3 The status of Alb. 173 is thus deceptive – more a public relations document than a picture of the San Carlo’s spatial meaning. arguing that Alb. 173 from the conceptual stage implies its secondary importance. 2) Two tangent circles inscribed. moreover. 3) a double-rail rectangle tangent to the oval. Borromini layered over the plan a geometrical armature.6 While Connors does not deny the existence of a geometrical armature. 175. 2-5 July 2009 th 2 curves be understood? Interpretation was clarified by Leo Steinberg’s brilliant doctoral thesis of 1960. oval.7 . Among Borromini’s drawings the true ‘ectoplasmic’ quality of the original plan is instead better represented by Alb. Casting aside the erroneous plans that had misled earlier commentators.4 At this point. and demonstrated how the church evolved from an elongated Greek Cross to the eventual synthesis of cross. 173 should be understood as a regularisation rather than ideation of the form. hidden in the Vienna archive until 1958 and which seemed to reveal once and for all the plan’s underlying geometrical structure. 168. 173. Alb. the displacement of Alb. and octagon. 171 was drawn without ruler and compass. which gained wide acceptance following its publication in 1977. when Borromini and the printmaker Domenico Barrière decided to publish engravings of the monastery. Connors argues. 5) Chamfered corners reducing the rectangle to an octagon.2 He also made an extensive exegesis of Alb. thereby exposing for the educated reader the long held secret (already in 1650. 6) Completion of the side chapels. 171. 169. the earliest surviving plan. and 176) were reworked or drawn anew around 1660.

8 Even though Connors is surely correct in arguing that Borromini prepared or revised drawings for publication in the 1660s. however.732). Moreover. as if ropes might not be required to erect a tent. bisecting lines for the small circles. it is nevertheless replete with constructional lines – ghostly traces are visible of equilateral triangles. On close examination there seems no compelling reason to reject the geometry of Alb. and so on.) The idea that the geometry is detachable from the church’s shape is itself perverse.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland. as will be discussed below. This is so even of the project-style drawing Alb. 2-5 July 2009 th 3 The point. tangents framing the oval. an idea of the plan. Borromini seems to have been one of those masters for whom intention and technique were fused. It is one thing to say that the ‘geometrical’ plans are post facto. and it is inconceivable that we are in the presence of a complete sequence of preparatory studies (Borromini destroyed many of his drawings before his suicide in 1667). and imply in the process that Borromini contrived a geometrical structure that would adhere to a shape conceived at a separate non-geometrical stage. is that it depicts an undersized church. San Carlo’s shape is difficult to achieve without geometry. 170.10 (In fact the major thing wrong with Alb. but it has not accrued redundant geometry – rather it has been tidied up.9 Alb. 171. or. as Alb. so to speak. 173 does indeed present an ideal version of the plan. indicating that Borromini wanted to make the church appear less cramped within its site than it really was. and the ratios near impossible. 173 as a good. representation of the ordered sequence by which design of the church was constructed. may be overdrawn. 173. the dating of the drawings remains an open issue. it is another to suggest that they rationalise the plan in a manner that distorts its conceptual basis. with the highlights emphasised. which involve the ratio of 1:√3 (1. which is to say that the conceptual language of his spatial planning was already geometrical. which suggests that whenever the plan was drawn the underlying geometry was too – in short. all of which can be articulated in reconstruction. 173 so memorably demonstrates. albeit simplified. If this were not the case the dimensions of the seminal schemes of Alb. . 172. would be inexplicable. a composite and artificial – but hardly false – picture.11 Thus reconstructing San Carlo’s geometry presents. to put it a better way. while there is no explicit geometrical armature in Alb.

reprising Sedlmayr. Such a method. the curvature of the chapels would be more pronounced. The problem arises because the equilateral diamond is not the beginning of the plan.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland. 172. Reconstruction of Alb. 173. 169. 2-5 July 2009 th 4 Figure 1. 175 and 176. once he had discounted the idea that they echoed the interior cupola. which in one sense is an elaboration of the means by which the longitudinal axis is formed. whose centre is the exact centre of the plan – but if that were so. but stage two. each of which will be one-third the length of a larger circle. One early commentator suggested they are both segments of a single large circle. The easiest way to construct such a diamond is to set the compass at either end of the desired base. Scholars assume that the plan’s geometrical armature. sees the lateral arcs as segments of two elongated ovals. creates a cell- .14 This view requires that the short elbow-like curves – connecting the straight diagonals to the shallow curves of the chapels – be segments of small circles.12 To Steinberg. to be drawn freehand. The rationale of the arcs of the lateral chapels remains unclear. From here one can follow the steps in the formation of the plan – with one important exception. and draw two arcs. the lateral chapels cannot be construed as half ovals.13 Portoghesi. whether retrospective or not. which answer the semi-circularity of the longitudinal chapels. something is missing. they appeared. 168. Biangolo and the Lateral Chapels Nevertheless. Whatever the case. which they quite clearly are not (it these four elbows which are in fact drawn freehand). begins with the diamond shape of the two triangles. as seems to be apparent in Albs.

Figure 2. as we shall see. Thus the first step for Borromini would have been to set the compass to the width of 45 palmi. which we can call. 175-76) that show the church in its final or near definitive state biangolo segments correspond exactly to the arc between the two innermost lateral columns. which is what many would call the shape. a biangolo. initial steps. San Carlo. chosen. is anachronistic). end on end equilateral triangles are created. But this is not the case. On each of the plans (Alb. for its numeric symbolism and its fittingness to the site. 2-5 July 2009 th 5 like shape with a width to length ratio of 1: √3. 168-170. . and construct a biangolo. If the triangles were all that Borromini wanted to inhabit the plan. for the biangolo provides the curvature of the lateral chapels. following Filippo Juvarra and for want of a better term (the term vesica pisces.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland.15 When the tips of the base line are connected to the intersecting arcs above and below. thence follows the vertical axis and equilateral triangles. then the biangolo would be no more that an invisible means to an end.

Figure 4. I hasten to point out that this line was not built. The outer track.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland. which precisely describes the outer lines of the plan’s lateral wings. . Inner line of San Carlo. is also formed by the curvature of a wider biangolo. Inner and Outer lines of San Carlo. which corresponds to the walls of the actual church. rather it is part of the inner track for the setting of the columns. 2-5 July 2009 th 6 Figure 3. at least on the ground rather than cornice level.5 palmi further along the horizontal axis (making the wall to wall width of the plan 50 palmi). Thus a second biangolo can be drawn. enveloping the first. with each segment set about 2.

8) second biangolo. at least in the lateral chapel. for there is a further recession to accommodate the side altar paintings. 2-5 July 2009 th 7 In fact this outer track built was not built either. and 7. then running parallel to the diamond and ending at the long side of the rectangle. I have set out graphically each stage in the construction of the plan. 4) rectangle tangential to oval and another parallel to it. set the compass to 45 palmi. and. 7) complete plan with freehand curves to the lateral segments of the biangolo. inscribe base to base equilateral triangles. with the centres set another 1 palmi (according to Alb. 10) final biangolo to establish altar exedrae. a summary of the scheme is necessary to highlight the theory here being advanced (needless to say. draw biangolo and vertical axis. by a width midway between broad segment of the oval and point of the diamond (3 palmi). bisect arcs for diagonal axes. making the total width of the church’s horizontal axis 52 palmi.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland. a further 2 palmi wider (a depth that is provided by the extending the inner diagonals of the plan down to the . construct an oval. San Carlo completed: altar recesses. 3) within resulting diamond. to establish the outer track of the plan. 5) establish semi-circular apses (with 10 palmi radius) on the long axis. Figure 5. 6. the numbered stages are my own. and adjacent chapels. commencing from the intersection of the apse semi-circles and rectangle. 6) rule diagonals. five palmi wider than the first. 9) repeat steps 5. columns. 175 and 176) along the horizontal axis. although it is tedious to describe. not Borromini’s): 1) establish scale and horizontal axis. 2) within the biangolo intersection. following Serlio’s method. These curved exedrae are also segments of a biangolo.

2-5 July 2009 th 8 horizontal axis). as Borromini is interested only in their dimensions and their lateral segments. take their cue from the biangolo’s wider elaboration. However. just under 3 palmi in diameter. Biangoli therefore announce three periods in the plan. needed to accommodate a sacristy. at least. 171 measures 45 palmi. for in squashing the initial lobed cross Borromini has arrived at a flat walled form in the side chapels – my eyes. in four groups. 171. can detect no lateral curvature in the any of Alb.16 Borromini. less once wall thickness is deducted. The first surviving draft of the monastery is Alb. such as chapels and corridors. which shows the plot destined for the church measuring. it is at this point that Borromini laterally compressed the cross into a shape that would sit within the biangolo. 11) sixteen columns. is a sort of elongated Greek Cross. 171’s pentimenti. The squared space that remains for the church measures 52 x 82 palmi. it is useful to keep their full. barely visible under the later revision. a ratio of exactly 1: √3. when the shape was developed in relation to the site. so he blocked off one of the lateral chapels and shifted the central vertical axis south about 6 palmi. . Biangoli and the Development of the Plan Apart from explaining the curvature of the lateral chapels.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland. once the slight angle of the Strada Felice (now via Quattro Fontane) is squared off. because as the plan is developed the secondary spaces. and an octagonal chapel in the top corner (the Cappella Barberini. dedicated to the Madonna). The inner width (which will define the cornice supporting the pendentive zone) in the second scheme of Alb. and the length. approximately 65 x 82 palmi. The initial idea for the church. however. Of course the biangoli are not drawn in full. with the long axis centred on the site. as Borromini may have done. representation in mind. highlighting biangoli in San Carlo’s plan helps us to reconstruct the earlier planning phases. 78 palmi. underscoring the cross. At this stage the biangolo supplies only the proportions and not the shape. twin-circled.

172. The solution is Alb. drawing in the four central columns so that they now stand as points on a 34 palmi sided square. while ample space is found for the high altar exedra behind and even miniature vestibule in front. This is more or less what was built. 171 is a tight fit and looks as if it has been shunted on both sides. where Borromini makes the church smaller still: he narrows the longitudinal apses from 25 to 20 palmi in diameter.67) did not prevent the high altar exedra from projecting beyond the site and into the corridor behind. . and crucially for the present argument. compared to Alb. its length 70 palmi (thus returning to the precise biangolo proportion of 1: √3). 173) he solved the problem by shrinking the church nearly 10%. 171 is reestablished: indeed. finally. but Borromini still ran out of room: even shortening the longitudinal axis. he pinches the diagonal bays (which in Alb. 172.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland. although they are differently shaped. although it is unclear whether Borromini preferred convexities or concavities). the wall length of longitudinal and lateral chapels is the same. The lobed cross of the initial scheme in Alb. or 1:1. he installs the curvature of the biangolo into the design by setting the lateral arms onto its shallow arcs.18 In other drawings (Alb. about 33 palmi. The second scheme of Alb. 2-5 July 2009 th 9 Figure 6.17 When he later drew a plan intended for publication (Alb. by 3 palmi (the width to length ratio line is 45 to 75 palmi. so that its width is 41 palmi. as is the inter-columniation. 171 were not straight. San Carlo: Development of the church in relation to site.

with the consequence that the dormitory was shifted east and the garden diminished. a different strategy of distortion is employed: Borromini maintained the actual size of the church. 171. but lengthened the plot on which it stands by 10 palmi.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland. Once the smaller biangolo is inscribed it is simply a matter of drawing the semi-circular apses on the vertical axis and the small free hand curves to the lateral chapels to arrive at the definitive shape. Figure 7. 2-5 July 2009 th 10 170. It shows that the plan is but a modification of the initial scheme in Alb. I have proceeded from the inner dimensions (that is. The 52 palmi cell yields the curves of the altar recesses and equilateral triangles. whose sides come close to providing the diagonals of the plan. . Demonstrating the genesis of the plan in the manner above raises an interesting possibility. 45 palmi) and worked outwards.19 In Borromini’s mind. The result is a far simpler geometrical construction. 175-176). However. the relationship of the size of church and the dimensions of its site was never reconciled. what if we worked from the outer dimensions (52 palmi) and then inwards? The reason for suggesting such a sequence is that the width of larger biangolo fits precisely within the boundaries between sacristy and cloister allotted to the church. San Carlo dimensions as constructed.

22 This cornice is a continuous band duplicating the inner track of the ground plan. The number 186 thus refers not to the dome. which mentions 186 palmi length of ‘Ornamento sotto il gocciolat[oia]e della cornice principe della ch[ies]a f[att]a a stampa con rose’ [ornament under the drip profile of the main cornice of the church. it supports the pendentive zone. but to the perimeter of San Carlo’s plan. is the particular numerical value of 45 palmi. is not remotely close to the required length. Borromini’s hero. hypothetical scheme. where r = 45). Against this method. it would seem likely that the coincidence was contrived by Borromini himself. 186 palmi. while the diameter of its dome. This I would suggest is the biangolo – one drawn with a radius of 45 has a length of 188 palmi (4rπ/3.5. and in favour of beginning with the smaller dimension. which even allowing for the difference between plan and construction (within the standard deviation of 5%). then creating circumferential length of the plan to exactly to fit the magic number 186 would have required some guiding method. In fact. . Steinberg noted that since the early eighteenth century St Peter’s and San Carlo were often compared: the former could contain the latter within one of its piers. 186 palmi is achieved with a radius of approximately 44. relief-moulded with rosettes]. which turns out to be almost exactly the width of the church as constructed (9920mm = 44.21 The source for the story that Steinberg cites is a stucco account for the church. However.20 Given that the earlier dome was designed by Michelangelo. is same as the circumference of San Carlo’s.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland. 2-5 July 2009 th 11 Figure 8. If Borromini did wish for this particular concordance with St Peter’s.4 palmi). San Carlo. the legend is skewed: by my calculation San Carlo’s oval dome is approximately 136 palmi in circumference. which in turn supports the dome.

is fantastically complex. every line is needed. San Carlo’s plan is a perfect example of Borromini’s belief in the power of the compass to create symmetry. The dominance of geometry is noteworthy. A drawing of the entire church precinct. Even in the tiny components. Façade analysis: Alb. but he only had to set it to scale once. for the plans of San Carlo comprise something like a self-contained graphic world. The drawings for San Carlo are unbelievably rigorous in their construction. Even in the 1660 plans. recalled by a confidant in these words: ‘The beauty of building. every line is in fact there in some form or another. or whatever was the size of the biangolo at that stage of planning – thereafter. an elaborate rationale can be detected. depends on numbers. when everything but façade had been built. Borromini opened and closed the compass many times. without ever moving it.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland. with every constructional line visible. too much so perhaps to be of any use. almost every element is scaled automatically. the graphic dialogue between geometry and detail remains vital. niche. like music.’28 Of course. when establishing the initial radius of 45 palmi. with divergence from one plan to the other at the micro-level of wall. which must be manipulated if even a single component is to change. such as the jambs giving onto the niches or the orientation of column bases. 169 (left) and Alb. and their countless pentimenti suggest a relentless search for perfection. 2-5 July 2009 th 13 Figure 9. .27 It is the plan’s organic circuitry. and column base. but not fanciful. by geometry. who would draw all night rather than sleep. everything is measured. which I freely admit has often left me floundering to reconstruct. One recalls contemporary descriptions of Borromini as living only for his art.29 To good purpose. 177 (right). so that all the parts have such a proportion that with a single opening of the compass.

it would surely relate to the Trinitarian dedication of church. BEATOQ. AN. TRINITATI. 1625) and which appears as a stucco moulding on San Carlo’s lavabo niche (near the old sacristy) and as wrought-iron portalumi installed at the base of the dome. DC. Above the inscription. SAL.33 Often cited.… And the more one regards the church the more . its imagery nevertheless suggests something of the spiritual-poetic language of artistic intention. 2-5 July 2009 th 14 If the biangolo has any significance beyond its demonstrable presence within the construction of the plan. a three into one motif that Paolo Aresi had featured as a Trinitarian symbol in his Imprese sacre (Milan.32 The inscription at the base of the dome lantern reads: SANCTISS. we see the other half. and for those who look upon it each part seems to glance at the other. the beginning of the beginning. XL [To the Most Holy Trinity and Blessed Carlo Borromeo. and in parts owing something to Procopius’ description of the St.34 … when [visitors] are inside they cannot do anything other than gaze upwards to the vault. Poetically. D. the Trinity’s emblem. At the very least. or what is implied.Cultural Crossroads: Proceedings of the 26 International SAHANZ Conference The University of Auckland. the lantern is a valve connecting here to the beyond: through it the Dove enters. The idea of San Carlo as an interior reverberating with sacredness was advanced in Fra Juan’s 1650 account of the convent. for everything is disposed in such a way that each calls to the other. BORROMEO. is an image of the Dove. for the lantern calls in response to the ground plan. seen from above. framed in an equilateral triangle and suspended within an aureole of light created by the lantern windows below. If we turn our heads upward. the passage merits quoting at length – while no doubt imbued with a fair share of self-promotion. 1640]. year of the Lord. and it spawns the equilateral triangle. Sopia (particularly how it produces in the viewer an insatiable desire to see it again). the topic Steinberg illuminated. CAROLO.31 This paper has examined the idea of the church’s footprint. its dimension √3 embodies Trinitarian irreducibility.30 The symbolic parallels are obvious: the biangolo is the creative basis of form. but not as far as I know translated. likely written with the imprimatur of Borromini himself. M. unlocking the majesty of the plan and sending a spiritual pulse throughout the room. like heaven to earth. we know that interlocking circles to form the biangolo also results in a variation on the emblem for the Trinity: adding an additional circle to the two required for a biangolo creates a trefoil with a threepointed cell in the centre. on the lantern ceiling.