INTRODUCTION
There are no sources about the vow, construction and dedication of the Temple of Castor and Pollux in the Circus Flaminius. Three calendars (Fasti Antiates Maiores, Allifani and Amiternini) attest that its dies natalis was the 13 August. Vitruvius (De arch. 4.8.4) includes it among the temples with a transverse cella (that is, with the cella larger than the pronaos). It was probably built after the triumph of either P. Servilius Vatia Isauricus in 74 bc, or Q. Caecilius Metellus Macedonicus in 71 bc.Footnote 1 Antonio Maria Colini believed that it was literally inserted into the circus. A possible location was suggested by Carlo Pietrangeli in 1952, eight years before the actual position of the circus was identified by Guglielmo Gatti, in an article dealing with the Dioscuri placed at the top of the Capitoline cordonata: indeed, these statues were recovered between the church of San Tommaso ai Cenci and the nearby synagogues.Footnote 2
This area of the circus (Fig. 1) was partially excavated in the early twentieth century after the demolition of the Jewish ghetto, and is depicted on fragments 32 h–i of the Severan Forma Urbis, that were identified by Emilio Rodríguez-Almeida in 1977.Footnote 3 In 1983, a marble plan different from the Forma Urbis came to light near the Via Anicia, in Trastevere.Footnote 4 It shows the Temple of Castor and Pollux, the area of the circus, some warehouses and the left bank of the Tiber, on which is depicted a building that might be identified with the shed housing the ‘ship of Aeneas’ described by Procopius (Goth. 4.22).Footnote 5 The topography visible on the Via Anicia plan partially overlaps fragment 32 i of the Forma Urbis — the temple would have appeared in the missing slab corner (cf. below, Fig. 15). Indeed, the two plans share the same scale (1:240) but have a different orientation; otherwise the inscriptions on the Via Anicia plan, with southeast at the top, would appear nearly upside down. Other plans incised on marble are known, though most are isolated fragments depicting unidentified areas and, like the Via Anicia plan, unlikely to have belonged to complete plans of Rome.Footnote 6 The Via Anicia plan is surely older than the Forma Urbis and is more detailed: the thickness of the walls is indicated by double lines, the names of the proprietors are inscribed in the genitive inside the buildings (exactly as ‘Castoris et Pollucis’ is inscribed inside the temple), and the lengths of the porticoes facing the river Tiber are given in Roman feet. Apparently, this plan was a cadastral document linked to the collection of taxes, either for street repairs or for the safe upkeep of the bank in front of private properties.Footnote 7
Fig. 1. Rome, the area of the Circus Flaminius. In the inset, the trench of the 1996 excavation (in grey in the plan) looking north. STC = San Tommaso ai Cenci. (Drawing and photo: author.)
The archaeologist who first published the Via Anicia plan suggested that the church of San Tommaso ai Cenci was built over the temple, and identified a few Imperial walls visible beneath that church with the substructure of the pronaos.Footnote 8 If this were the case, however, the pronaos floor would have been about 9 m above the circus, although the staircase of the temple in the Via Anicia plan has just eight steps. I have argued elsewhere that these walls belonged to a warehouse with a gabled entrance.Footnote 9 A slightly different location for the temple, east of the church of San Tommaso, was also suggested, in accordance with the slab edges of the Forma Urbis.Footnote 10 However, this can not be right, as this area contained the warehouses found in 1996. Precisely because of the slab edges, I proposed an alternative location further east, beneath the modern block of the Jewish ghetto occupied by four apartment buildings (the so-called villini) (Fig. 1). Indeed, due to the ‘wrong’ position of the Theatre of Marcellus in the Forma Urbis, it is necessary to shift fragments 32 h–i, and consequently the Via Anicia plan, c. 36 m toward the east.Footnote 11
Finally, when in 1996 an underground service tunnel was built in Piazza delle Cinque Scole, the remains came to light of a ‘monumental building’ and some warehouses that stood between the Circus Flaminius and the Tiber (Fig. 1, inset).Footnote 12 The excavation was directed by Paola Ciancio Rossetto (Soprintendenza Comunale) and supervised by Massimo Vitti, who immediately claimed they had discovered the Temple of Castor and Pollux in circo.Footnote 13 Fourteen years later a description of this excavation was published by Vitti alone, followed by an article including the analysis of the walls discovered in the early twentieth century during the construction of villini.Footnote 14 Although I had solicited the publication of at least one section of the profile of the podium, in his report of 2010 Vitti presented once again a preliminary analysis of the data with the promise of a response to my remarks, which is still missing in his latest article.Footnote 15 My main concern was, and still is, the absence of a podium, a typical feature of Roman temple architecture, which is implied by the eight steps before the pronaos of the Temple of Castor and Pollux on the Via Anicia plan.Footnote 16
The identification of the Temple of Castor and Pollux is not a minor topographical matter, but concerns broader issues, such as Roman Hellenistic architecture and the typology of Roman temples. Moreover, if the Via Anicia plan could be related correctly to the walls excavated in and around Piazza delle Cinque Scole, a terminus ante quem for the plan itself would follow (indeed, the Via Anicia plan cannot be later than the early second century ad, as will be explained below).Footnote 17 The ultimate objectives of my article are to explain why my alternative location of the temple fits better the evidence from the marble plans and the urban context, and to clarify why the exact location of the Via Anicia plan does matter: I am not just moving a temple a few metres to the east! Vitti remarks that I am the only one who disagrees with his identification, but I am afraid that no one else has examined minutely his archaeological reports, which may ultimately seem persuasive thanks to the use of archival drawings, brightly-coloured plans, and a mass of details about levels, measurements and building techniques.Footnote 18 In fact, these reports are rife with mistakes of both an archaeological and a technical nature (cf. Fig. 2), which may lead to a skewed vision of the ancient remains excavated in Piazza delle Cinque Scole.
Fig. 2. Plan of the building discovered in Piazza delle Cinque Scole. Note the incorrect metric scale (the correct numbers added in parentheses by the present author). (Drawing M. Vitti and P. Vitti, from Ciancio Rossetto, ‘Rinvenimenti e restauri’ (n. 12), fig. 11.)
THE DRAWINGS
I have already stressed that the temple depicted on the Via Anicia plan is not compatible with the remains of the monumental building of Piazza delle Cinque Scole because the presumed base of the cult statues found in the 1996 excavation is eccentric, compared with the axis of the actual temple.Footnote 19 Vitti solves this problem by overlaying the plan of the excavation on a much-reduced drawing of the Via Anicia plan, so that the base depicted on the marble plan is perfectly aligned with the presumed base (Fig. 3). However, in his drawing the length of the transverse cella is about 4.3 m shorter than the Via Anicia plan's cella at the scale of 1:240. Another ‘side effect’ is that the presumed rear wall of the temple and the warehouses behind it appear to have been displaced by at least 2.5 m — the result of an early second-century restoration according to Vitti. In fact, these modifications are just the consequence of the incorrect metric scale of Vitti's plans. I am not talking of a difference of just a few centimetres: Vitti reduces the Via Anicia plan by approximately 17%, as if it were at a scale of 1:290 instead of 1:240.Footnote 20 It is worth noting that the extension of the warehouses behind the temple over the Tiber bank and the disappearance of the road along the river are not discussed at all, although such changes do not appear on the Severan Forma Urbis (cf. below, Fig. 15).Footnote 21
Fig. 3. The complete plan of the 1996 excavation overlaid on the Via Anicia plan, according to Vitti's identification of the temple. Left, Vitti's drawing, with the Via Anicia plan reproduced at the wrong scale. (From Vitti, ‘Aedes Castoris et Pollucis’ (n. 14), fig. 17.) Right, the correct overlay of the plans. (Drawing author, showing Vitti's location of the temple.) The grey line passing along the temple entrance (in the Via Anicia plan) can be used as a reference to assess the reduction of size of the marble plan itself in Vitti's drawing. The metric scale in the bottom right-hand corner does not apply to the Via Anicia plan on the left because of its reduced scale.
Fig. 4. Above: the Capitolium of Ostia (from Albo, ‘Il Capitolium di Ostia’ (n. 30), fig. 1) and, at the same scale, the building of Piazza delle Cinque Scole (from Vitti's survey: see Fig. 2). Below: the threshold of the Capitolium of Ostia (from P. Pensabene, Ostiensium marmorum decus et decor. Studi architettonici, decorativi e archeometrici (Rome, 2007), fig. 142) and, at the same scale, the entrance to the building of Piazza delle Cinque Scole (from Vitti's survey: see Fig. 2).
In 2007 I published the overlaid plans of the Via Anicia and of the excavation according to Vitti's identification. Now that his version is finally available (Fig. 3, left), I expected a few words on the discrepancy with my drawing (Fig. 3, right).Footnote 22 Vitti just admits that the comparison between the monumental building and the Via Anicia plan highlights some metric differences.Footnote 23 As a matter of fact, in his drawings the temple is too small, and the reduced size of the cella makes it compatible with the position of the presumed axial base. We are told that the cella (in fact, the excavated building) is 10.9 m deep, instead of 9.6 m as in the Via Anicia plan at the right scale (Vitti uses the precise depth measured by Conticello De' Spagnolis on the marble plan): a difference of just 1.3 m.Footnote 24 However, in Vitti's plan the cella is less deep and this difference is of at least 2.5 m.Footnote 25
To sum up, Vitti's identification is based on the use of the Via Anicia plan at the wrong scale. With a smaller cella the presumed base of the cult statues can be compatible with the axis of the temple depicted on the marble plan, but it is necessary to pretend — without the support of archaeological evidence — that the rear side of the temple was rebuilt further back. As a mere matter of simple logic, the plan drawn at the right scale cannot support Vitti's point of view: for example, at the scale of 1:240 it is impossible to justify the misaligned temple's base and axis. Moreover, at the right scale other problems call for an explanation, such as the excessive width of the entrance to the cella, the missing podium, and the different plan and location of the warehouses behind the temple. Although the Via Anicia plan is very carefully carved and one can be absolutely certain of its accuracy, in the next sections I shall also consider the archaeological evidence independently from the topography depicted on this plan.
THE ENTRANCE TO THE PRESUMED CELLA
A block of white marble that originally supported a door-jamb was still in situ next to the concrete foundation of the presumed west side of the pronaos, and belonged to the entrance to the monumental building (Fig. 4, right).Footnote 26 In Vitti's plan the entrance outlined on the Via Anicia plan is just 3 m wide, whereas it is 3.85 m wide on the marble plan at the right scale of 1:240 — further proof that he has the scale wrong. If Vitti were right to correlate the excavated building with the Via Anicia plan's temple, he should consider that in his plan the door-jamb block implies a width of the actual entrance of 7.125 m (Fig. 3, left). He just tells us that the actual entrance seems to be larger than on the Via Anica plan.Footnote 27 Moreover, if Vitti had drawn the marble plan at the right metric scale, the width of the entrance to the monumental building as he reconstructs it would be approximately 10 m because of the different position of the axis of the temple (Fig. 3, right). This widening from 3.85 to 10 m implies a radical reconstruction of the wall between the pronaos and the cella, and such an improbably wide doorway would occupy the whole width and height of the same wall. This is an event that Vitti himself does not consider, since he does not (and can not) argue that the door-jamb block was moved to widen the original entrance. Not only its size — just imagine an architrave more than 10 m long, not to mention the doors — but the layout itself is unheard of in Roman temple architecture. The plans of other temples with transverse cellas show that there was an obvious proportional ratio: larger temples had wider entrances, and only a small temple such as that of Veiovis had an entrance as large as the pronaos.
Vitti highlights the similar depth of the thresholds of his proposed temple and that of the Capitolium at Ostia (115 and 121 cm, respectively — in fact, 106 and 127 cm), but does not consider the width of the entrances.Footnote 28 The Capitolium, the pronaos of which is as large as that of the Temple of Castor and Pollux on the Via Anicia plan, has an entrance approximately 4.5 m wide (Fig. 4) instead of the c. 3.85 m for the entrance on the marble plan, and less than the 10 m of Vitti's reconstruction.Footnote 29 In addition, at Ostia the block of marble with a square recess for the door hinge is inserted into the floor inside the cella (Fig. 4, below) — the doors opened inwards.Footnote 30 The similar block found in situ in the monumental building completely reverses Vitti's identification of the cella and pronaos of the Temple of Castor and Pollux, because it would be inserted into the floor of the latter.Footnote 31 It is likely that the threshold of the monumental building was just 3–3.5 m wide, and that the entrance stood on the left of the axis marked in Figure 3, right.
A break visible on the presumed foundation of the northwest side of the pronaos is highlighted by Vitti with a line (Fig. 5, above). According to him, this break is the imprint of the temple anta depicted on the Via Anicia plan.Footnote 32 This line, however, does not correspond to any real trace on the foundation (Fig. 5, below) and in the plan of the excavation (cf. the broken line in the inset of Figure 5, corresponding to Vitti's red line). It is not even clear why there should be a break there. The concrete foundation appears to be partially damaged closer to the blocks of tuff, as marked in Vitti's plan with an actual line that has nothing to do with the size of the presumed anta block.Footnote 33 Vitti also states that this concrete foundation has the same width (1.52 m) as the plinths of the columns of the temple, but in fact the foundation should be larger.Footnote 34 According to the Via Anicia plan, the anta was approximately 1.7 m wide, which means that it was slightly larger than the concrete foundation (which means that the red outline of the anta in Vitti's photo is ‘compressed’ within the edges of the foundation: see Fig. 5). The Temple of Castor and Pollux can be compared to the extant Temple of Portunus, where plinths of 1.45 m required a foundation nearly 2 m thick. The concrete foundation in Piazza delle Cinque Scole must have supported a thinner wall. Also, the wall of the cella visible on the Via Anicia plan would be approximately 1.7 m thick, whereas the surviving wall is just 1.06 m thick; even considering the marble veneer, the difference in thickness is remarkable.
THE BRICK FACINGS
The brick facings of the warehouses next to the monumental building, including the walls beneath the church of San Tommaso ai Cenci (Fig. 6), date to the late first–early second centuries ad. Their characteristics are summarized in Table 1. The modulus (five bricks and five joints of mortar) is around 27.5 cm, with bricks 3.6 cm thick.Footnote 35 When there is an exception, as in the case of the taberna next to the Tiber, the higher modulus (30 cm) is easily explained with the use of thicker bricks (4–4.5 cm). In general, the courses of bricks are higher than the joints of mortar (Fig. 6).
Fig. 6. The brick facings of: (a) the taberna facing the Tiber (lower sector); (b) the taberna behind the monumental building of Piazza delle Cinque Scole (in the inset, the brick facing of the latter at the same scale); (c) the outer wall of the taberna, just behind the building; (d) the warehouse beneath the church of San Tommaso ai Cenci. (Photos: author.) (Plate 1 in colour section at the back of the issue)
Table 1. Characteristics of the brick facings of the warehouses next to the monumental building.
1 Tucci, ‘L'entrata di un magazzino’ (above, n. 9), 763.
2 Vitti, ‘Aedes Castoris et Pollucis’ (above, n. 14), 76 n. 6; Vitti, ‘Il tempio dei Castori’ (above, n. 14), 110–11.
3 Vitti, ‘Aedes Castoris et Pollucis’ (above, n. 14), 75 n. 5; Vitti, ‘Il tempio dei Castori’ (above, n. 14), 112 (also a modulus of 27 cm).
4 Vitti, ‘Aedes Castoris et Pollucis’ (above, n. 14), 75 n. 5; Vitti, ‘Il tempio dei Castori’ (above, n. 14), 111.
According to Vitti, the brick facing of the monumental building, visible in two sides of the presumed axial base (Fig. 7), is identical to that of the warehouses, and therefore Domitianic or Trajanic in date. He claims that the late antique dating I suggested surely can be excluded because the bricks are all homogeneous and the courses are perfectly regular (Table 2).Footnote 36
Fig. 7. Left: the modulus of the brick facing of the building of Piazza delle Cinque Scole. Right: the wall of the building (above) and detail with the different lengths of the bricks (below). (Photos: author.) (Plate 2 in colour section at the back of the issue)
Table 2. Characteristics of the brick facings of the monumental building.
1 Vitti, ‘Aedes Castoris et Pollucis’ (above, n. 14), 77 n. 9.
In this base, the modulus is higher than the average 27.5 cm of the warehouses (Fig. 7), but the difference is not due to the use of thicker bricks. Indeed, the bricks are just 3.2 cm thick (not 3.8 cm), but the joints of mortar are about 3 cm high (different from the 2.2 cm of Vitti's article, which can be determined easily since we are told that the brick facing is homogeneous). Vitti, discussing the front and short sides of the presumed base, claims that both brick facings belong to the same structure, with yellow bricks, 3.8 cm thick and 27 cm long, and a modulus of 29–31 cm. Yet, Domitianic and, for example, Constantinian brickworks may have the same modulus but be quite different as to the thickness of the bricks and the height of the joints of mortar. The modulus alone is meaningless. Also, in the monumental building the bricks are not 27 cm long, but range between 23 and 28 cm, and some are just 14 cm long (Fig. 7). For these reasons, in my 2007 article I concluded that the wall is likely to be late antique. It is true that towards the end of the first century ad bricks were ‘tutti omogenei per dimensioni e fattura … e i ricorsi sono perfettamente regolari’,Footnote 37 but, although the brick courses of the front side of the presumed axial base are regular (horizontal), they consist of heterogeneous bricks: indeed, there is a similarity with the brick facings of the Basilica of Maxentius and of the Temple of Venus and Rome.Footnote 38 Even quick comparison of the ratio bricks/mortar in the monumental building and in the warehouses reveals a striking difference (cf. the inset in Fig. 6, b).
The ‘few courses’ of yellow bricks on the presumed short side of the axial base also warrant attention, because no photos have been published and the actual corner with the front side was not seen at all.Footnote 39 The plan of the excavation (Fig. 2) shows that these bricks were preserved up to 13.87 m above sea level (asl), therefore below the marble floor of the building (14.20 m asl) and the preparation of the lower floor (13.95 m asl). It is not impossible that these bricks belonged to the same phase as the first marble floor, but it would be important to know the number of these ‘few courses’. Anyway, since Vitti gives detailed information on its brick facing, which appears to be different from that on the front side, it is likely that his Domitianic or Trajanic modulus refers precisely to the side facing. That said, one wonders if, instead of being the remainder of the presumed base, these bricks might have belonged to a pre-existing wall razed to the ground and belonging to the warehouses next to the temple: indeed a wall may have stood there, according to my location of the Via Anicia plan (cf. below, Fig. 9).Footnote 40
My late antique dating of the brick facing of the monumental building corresponds to Vitti's dating of its actual marble floor. As for the presumed ashlar wall of the cella, which in 1999 Ciancio Rossetto dated to the late Republican period, Vitti now dates it to the late first or early second century ad. It consists of reused blocks, as suggested by a hole for lewis irons, their different heights, and their dimensions (although in the plan of the excavation the hole is missing and the joints look regular: Fig. 8).Footnote 41 No metal clamps were used to join the blocks of tuff and the only block of marble — quite unusual in Imperial squared-stone masonry. Vitti, who claims that the width of the blocks is regular and downplays the evidence of the lewis hole, justifies the absence of metal clamps by mentioning the Temple of Veiovis, the walls of which, however, were built with blocks of Grotta Oscura tuff and in the first half of the first century bc, not at the time of Domitian or Trajan.Footnote 42
THE FLOORS
Inside the monumental building a thick layer of cocciopesto with a smoothed top surface was found, and identified with the original, late Republican floor of the cella of the Temple of Castor and Pollux.Footnote 43 It was noticed only in a very small area, and (in Vitti's view) would be the only surviving element of the original temple.Footnote 44 This floor eventually was replaced by a marble floor and by a subsequent marble floor at a higher level, the slabs of which were partially preserved. However, the layer of cocciopesto might have been the preparation for the missing marble floor above or, judging from its thickness, might have belonged to another building, such as the portico depicted on the Via Anicia plan west of the temple (Fig. 9).Footnote 45
Fig. 8. The squared-stone masonry, showing the misaligned joints (note the arrows) and the only hole for lewis irons (above), not recorded by Vitti (inset: cf. Fig. 2), and the foundation of the ashlar wall (below). (Photos: author.)
Fig. 9. The plan of the 1996 excavation overlaid on the Via Anicia plan at the same scale, according to the author's location of the temple. (Drawing: author.) A = first-century ad phase (walls indicated in outline: Via Anicia plan); B = possible modification of the circus floor (steps and gutter discovered in 1910: Archivio Centrale dello Stato, Rome, Archivio Gatti, carta 3562); C = early second-century ad phase (in grey: 1996 excavation); D = fourth-century ad phase (monumental building: 1996 excavation).
In 2001 Vitti dated the uppermost marble floor, on stylistic grounds, to between the first and fourth centuries ad, but he preferred the early second century ad (with a fourth-century restoration) because of the presence, in both preparations, of fragments of ardesia, a material used mostly in the first two centuries ad.Footnote 46 However, he must admit that marble floors ‘a rettangoli listellati’, like the floor at the higher level, are not attested in temples.Footnote 47 Ten years ago, he claimed that this floor belonged to the same phase as the brick facing of the axial base. Now the same floor is considered to be late antique — an indirect confirmation of my dating of the brick facing.Footnote 48 It is apparent that the excavators have no dating elements (they mention the lack of stratigraphical units associated with the building), and so the late Republican and Imperial phases are based just on their identification of the monumental building with the temple.Footnote 49
Cocciopesto was missing from beneath the floor of the presumed pronaos, which slanted and was exposed to the elements.Footnote 50 In 2010 Vitti claimed that the slanting floor is a peculiarity attested in other temples,Footnote 51 but the relevant footnote is a reference to his article with Ciancio Rossetto of 2001, with no page number (and, indeed, in that article there is no list of temples with slanting floors). In fact, the temple in question would be the only one with a cocciopesto floor in the cella, a slanting floor in the pronaos, a unique pattern in its later marble floor, reversed (and gigantic) doors, and no podium.Footnote 52 Considering the right position of the axis, it would even be smaller than in the Via Anicia plan phase. Apparently, expectations of what might be found seem to have determined the interpretation of what was actually found.
For instance, the ‘vaulted room’ beneath the concrete core shown in the excavation plan (next to the concrete foundation), which is dated to the fourth century ad (Fig. 2), in my view was just a thin layer of concrete laid below the travertine floor at 14.1 m asl (cf. below, Fig. 11) and above a layer of debris and earth that was eroded by rainwater during the course of the excavation, thus giving the impression of a cavity that was mistaken for a barrel vault.Footnote 53 Otherwise, did this presumed vault spring directly from the concrete foundation? Just imagine the thickness of the barrel vault at the impost: it would have been impossible to see its intrados, given the narrowness of the 1996 trench. Vitti has never shown a section of this vault that, moreover, would have covered an underground room. He even mentions the tiny spaces on either side of the podium that in late antiquity would have been turned into vaulted rooms, although in 1910 no traces were found on the opposite side of the presumed pronaos, as described below (cf. Fig. 3, right).Footnote 54
THREE IMPORTANT STEPS
An excavation undertaken in 1910 during the construction of the villino Lupi, next to the site excavated in 1996, brought to light two travertine floors of the Circus Flaminius located at different levels (Figs 10 d and 11). The lower level, characterized by a semicircular gutter and at least three travertine steps (each 15 cm high), was discovered at a depth of 6 m.Footnote 55 One of the sections shows that the higher floor was 1.80 m above the other (Fig. 11).Footnote 56 In other notes, however, the depth of the lowermost floor with the gutter is of 6.5 m, whereas the upper floor is 1.2 m above it.Footnote 57 In that area, the modern street level is at circa 17.54 m asl; consequently, the lowermost floor of the circus (with a difference of 6.5 m) is at 11.04 m asl (Fig. 12), as I pointed out in 2007.Footnote 58 In 2007 I also reconstructed the hypothetical profile of the podium of the Temple of Castor and Pollux, considering eight steps at 25 cm high, giving a total height of the podium of 2 m (Fig. 12). With the floor of the circus at 11.04 m asl (and possibly at a lower level when the temple was first built) and the presumed original floor of the temple (Vitti's layer of cocciopesto) at c. 13.57 m asl — a difference of 2.53 m — each of the eight steps would be circa 31.6 cm high. I cannot provide a complete list of heights of temple steps, which rarely survive, but the most common height ranges between 21 and 26 cm: therefore 31.6 cm is too high for a convenient step.Footnote 59
If one also considers the actual position of the three travertine steps, Vitti's identification should be definitively ruled out. In his latest article, Vitti examines them for the first time and claims that they do not interfere with the temple (Fig. 13). In fact, with the right metric scales and a correct location, they overlap the southeast side of the presumed pronaos (Fig. 3), that eventually would have been covered by the higher travertine floor. In his figure 6 (cf. Fig. 13, left) Vitti redraws a plan made in 1910, which shows the walls that were discovered when the foundations of the villini were being built. In the original plan — just a sketch (Fig. 13, right) — the alignments of the four villini are not correct and other details are approximate: for instance, lines of the same length correspond to either 22.63 or 32.16 m. Vitti overlays this schematic sketch on a modern plan and, because of the misalignments mentioned above, most of the ancient walls end up outside the villini (Fig. 13, left), where no excavations were made, and wrongly occupy my temple location.Footnote 60 In his drawing, in which the elevations are not considered at all, the metric scale is missing but it is possible to determine that the plan of the modern city and the Via Anicia plan are drawn at a different scale (1:840 and 1:940, respectively). More importantly, the steps and the gutter are misplaced and even upside down, so that the staircase leads far from the temple. Vitti remarks that the only remains that can be referred to a monumental building are precisely the steps with the semicircular gutter and the nearby floor: in his view, they stood far from the temple, but at the same scale and in the right position there is a conflict with the east corner of the pronaos.Footnote 61
Fig. 10. The plan of the 1996 excavation overlaid (at the same scale) on the Via Anicia plan, according to the author's location of the temple, with the drawings of the early twentieth-century excavations. (Drawings and photo: author.) In the insets, Giuseppe Gatti's notes, from Archivio Centrale dello Stato, Rome, Archivio Gatti: a) taccuino 12, 777; b) taccuino 13, 819; c) taccuino 12, 783; d) taccuino 12, 777; e) taccuino 13, 823; f) taccuino 12, 784).
Vitti has clearly misunderstood the description in the Notizie degli Scavi according to which the direction of the steps was from the east to the west: he notes that this orientation is not attested by the archival drawing, in which there is a north–south alignment.Footnote 62 However, the author of the 1910 report was not addressing the alignment of the steps but rather the direction of the staircase, which could be ascended from east to west. Vitti also fails to see that in the sketchy plan of the structures found in 1910 the steps and the gutter are not oriented as in his own plan; their orientation is clearly visible in two more drawings — one, in particular, with a large arrow pointing toward the north (Fig. 11).Footnote 63 Although the steps and the gutter were approximately in the middle of the north side of the villino Lupi — building and surrounding property included — and they are redrawn by Vitti as such (Fig. 11, top left-hand corner), in his general plan they appear shifted toward the east, far from the site of the presumed temple (Fig. 13).Footnote 64 In conclusion, a correct version of Vitti's plan requires an enlargement of the temple depicted on the Via Anicia plan toward the southeast, and a shift of the steps and gutter toward the northwest, with the conflict mentioned above (Fig. 13).Footnote 65 Ignoring that these steps overlapped his presumed temple, Vitti remarked that their height of c. 15 cm is not compatible with the Temple of Castor and Pollux, thus sparing me the discussion of whether they possibly belonged to a later phase of the temple itself (but for the sake of completeness, see below).Footnote 66 Moreover, in the 1996 trench these two phases (steps and gutter — higher travertine floor) were clearly missing (cf. Fig. 3, right).
Fig. 11. Top left corner: the walls found in 1910 beneath the villino Lupi, redrawn by Vitti (from Vitti, ‘Il tempio dei Castori ‘in Circo’ (n. 14), fig. 6); bottom left corner: the original drawing (from Archivio Centrale dello Stato, Rome, Archivio Gatti, carta 3564, with detail in the inset; cf. the same structures in the inset below, from Archivio Centrale dello Stato, Rome, Archivio Gatti, taccuino 12, 777). Note that in Vitti's plan the staircase is upside down. Right: plan and section of the same structures (adapted from Archivio Centrale dello Stato, Rome, Archivio Gatti, carta 3562).
At this point, it is worth considering the two stripes visible around the cella and the pronaos in the Via Anicia plan (Fig. 3). They might represent the cornice of the podium (overlapping the base of the podium itself) and an offset at the foot of the wall of the cella, or indicate a travertine step at the bottom of the podium and the cornice of the podium (the cornice overlapping the base below).Footnote 67 On the Via Anicia plan the two stripes end in correspondence with the corner columns and do not continue around the pedestals of the frontal staircase (the southeast pedestal is missing on the marble plan, but we can assume that the temple was symmetrical). It is precisely in this sector that the three steps found in 1910 stood: do they represent a modification of the original plan of the temple? We can even assume that the marble plan was not accurate in this case and that the two stripes continued along the pedestal placed on both sides of the staircase, but the problem is that not only the reports and drawings attest that the steps belonged to an actual staircase (they were 15 cm high and 28–30 cm deep), but also that the third element was an actual step.Footnote 68
We might even consider a modification of the original staircase after the Via Anicia plan was carved, and imagine two side staircases (Fig. 14).Footnote 69 However, they would have left the two sides of the original podium of the pronaos exposed from beneath the corner columns to the wall of the cella, so that the tall concrete foundation would be inexplicable.Footnote 70 The Temple of Castor and Pollux in the Roman Forum was provided with a speakers' platform approached by lateral staircases, but the popular assemblies (contiones) that were put up in the Circus Flaminius (and possibly near the Temple of Bellona) date to the late Republican age, a period in which the plan of the Temple of Castor and Pollux in circo is attested by the Via Anicia plan, where there was not a platform anyway. In conclusion, it is likely that the three steps belonged to a raised area of the circus, as suggested by the direction of the gutter. They are not compatible with the podium of a temple. Since they are not depicted on the Via Anicia plan, they might date to the second century ad (cf. Fig. 9).Footnote 71
THE PODIUM
According to Vitti, the rear side of the presumed temple, with its late first- or early second-century brick facing, was rebuilt toward the Tiber (see above), and consequently the Via Anicia plan depicts an earlier phase of the Temple of Castor and Pollux. The dating of this marble plan to before the end of the first century ad is correct (not because of the presumed shifted wall, though) but, thanks to this chronological difference, Vitti interprets the archaeological evidence freely, even implying that the original podium of the temple disappeared — an unprecedented event, considering the reconstructions and restorations of other Republican temples. To discuss this part of the structure, I have drawn an axonometric view of the Temple of Castor and Pollux as depicted on the Via Anicia plan, and I have associated with it the walls found in Piazza delle Cinque Scole according to Vitti's reconstruction (Fig. 14). First of all, however, it is necessary to go back to the sections (Fig. 12) and consider that the floor in cocciopesto (at approximately 13.57 m asl) was just 15 cm beneath the top of the concrete foundation (13.72 m asl) dating to the Imperial age, the level and the technical characteristics of which (I refer to the traces of the wooden formwork discussed above (n. 53)) imply that it was an actual foundation and that, when it was built, the floor of the circus was higher than in the late Republican age. In Vitti's reconstruction the concrete foundation would have replaced the external facing of the podium on the northwest side of the pronaos. Even assuming that this part of the temple was so damaged as to require a total reconstruction (different from the nearby squared-stone masonry of the presumed cella?), the restored temple would have been built without a podium, despite the efforts made to raise the ground level around it. According to Vitti, the concrete foundation was the actual foundation of the northwest side of the pronaos. Vitti adds that eventually two rooms, or (underground) vaulted chambers, were added to the pronaos sides, apparently below the top of the concrete foundation.
This scenario is problematic. In fact, there is no space for the profile of a podium between 13.72 m asl (the top of the concrete foundation) and the second-century ad slanting floor of the presumed pronaos. The alternative is even worse: it would be necessary to assume that the concrete foundation was built well above ground level (c. 2.18 m if the circus floor was at 11.54 m asl). In Roman temples, the vertical slabs between the base and the cornice were thick enough to allow the pouring of concrete inside the podium (see the Temple of Veiovis in Figure 12), unless an actual ashlar wall was built, as customary in large temples (see the Temple of Concordia Augusta in Figure 12). It would have been even more reasonable to use a brick facing, as in several temples built or restored in the Imperial age, but the use of a wooden formwork for a podium, above ground level, can be excluded.Footnote 72 Vitti himself claims that the blocks of tuff were used exclusively for the elevations and the exterior of the core, but there is no trace of and space for such blocks on the northwest side of the monumental building.Footnote 73 The concrete foundation with the imprints of the wooden posts, which according to him was built together with the squared-stone masonry nearby between the end of the first and the beginning of the second centuries ad, requires a totally different dating and reconstruction.Footnote 74 The concrete foundation was not a shapeless mass of concrete; its edges were straight and vertical, and the traces of wooden posts make it clear that there were no slots for stone elements belonging to the external profile of a podium.Footnote 75
THE BUILDINGS DEPICTED ON THE VIA ANICIA PLAN
According to Vitti, the walls found in the 1910s beneath the four villini did not belong to the Temple of Castor and Pollux.Footnote 76 According to my location, however, the temple covered just the southwest corner of the villino Serventi and the northeast sector of the villino Arcieri, where nothing at all was recorded (Fig. 10). Similarly, no foundations, floors, or architectural elements were recorded beneath the villino Lupi, where the east half of the monumental building stood. On the basis of the early twentieth-century drawings, neither ‘my temple’ nor the monumental building would exist! Yet, the latter occupied the site of a rectangular square, known in the Middle Ages as the ‘mercatello’ (little market) and eventually as the ‘Piazza delle Scole’ (square of the synagogues). There is no reason to assume that its state of preservation was not identical in the sector excavated in 1996 and in the area occupied by the villino Lupi, which cut the monumental building diagonally into two halves: it is unlikely that its eastern half had been spoliated in the Middle Ages envisioning the plan of a twentieth-century building. Only the rear wall of the monumental building, on which the outer wall of the synagogues was built, appeared to have been completely spoliated: ‘my temple’, the site of which was occupied by several houses, might have shared the same fate.
Vitti stresses that only commercial buildings were found during the previous excavations, but if the temple was dismantled and no structures were recorded in the two small sectors reached in 1910, this does not rule my identification out. Otherwise, we should assume that there was an empty space on this side of the circus or, as I have already stressed, that even the monumental building did not exist at all. In short, Vitti's dismissal of my location is based on the argument that absence of evidence is evidence of absence. However, the area between the four villini was not excavated at all (it seems so only in Vitti's drawing, in which the ancient walls are misplaced: see Fig. 13), and the excavation of the few trenches overlapping the temple (on the north side of the villino Arcieri and the west corner of the villino Serventi) brought to light nothing at all in the first case, and an ashlar wall at a depth of 8.5 m in the second. In any case, although there is no ‘smoking gun’ for my alternative location of the temple, an indirect confirmation is provided by the other buildings discovered in the area.
Fig. 12. Sections (at the same scale) of: the Temple of Veiovis; the Temple of Castor and Pollux in Circo Flaminio (according to the Via Anicia plan); the building of Piazza delle Cinque Scole; and the Temple of Concordia Augusta. (Drawings author.) Below: section of the presumed west side of the pronaos according to Vitti. (Adapted from Vitti, ‘Il tempio dei Castori ‘in Circo’' (n. 14), fig. 16.)
Fig. 13. Left: plan of the excavations in the area of the four villini according to Vitti. (Adapted from Vitti, ‘Il tempio dei Castori ‘in Circo’' (n. 14), figs 9 and 6.) Different metric scales are used at the same time, and the walls are misplaced, as in the original sketch on the right-hand side (Archivio Centrale dello Stato, Rome, Archivio Gatti, carta 3564). I have indicated the necessary modifications; note also the position of the bank of the Tiber.
Fig. 14. Axonometric view of the Temple of Castor and Pollux according to Vitti's reconstruction, and with a hypothetical double staircase. (Drawing: author.) The question marks highlight some of the major problems in Vitti's identification (for example, the overlapping with the staircase found in 1910 and the concrete foundation in the west side of the presumed pronaos).
The walls of the warehouses found in 1996 have an alignment that is compatible with my location of the Via Anicia plan, being inserted into the pre-existing structures in a logical way (Fig. 9), as opposed to Vitti's scenario (Figs 3, right, and 13). Vitti is just concerned for the fact that in his reconstruction the walls beneath the church of San Tommaso ai Cenci stood inside the courtyard northwest of the temple, but he highlights that, according to my location, there would be bigger inconsistencies.Footnote 77 On the contrary, in my view the walls found in the 1996 excavation stood in that courtyard with precisely the same orientation.Footnote 78 Vitti assigns these walls to the tabernae between the temple and the Tiber depicted on the Via Anicia plan, and claims that there is a perfect correspondence with his location of the marble plan (cf. Fig. 3, right).Footnote 79 The wall behind the presumed temple is shifted, and the walls of the tabernae are not orthogonal anymore: they even block the road along the Tiber and invade the river-bank.Footnote 80 However, the Forma Urbis shows no modifications: in the Severan age the partition walls of these tabernae were still perpendicular to the rear wall (they even appear to be aligned opposite to how they are depicted in Vitti's plan) and the road along the Tiber still existed (Fig. 15). Vitti notes that the excavated warehouses opened toward the river, although no entrances were found.Footnote 81 It is worth noting that the only difference between the Via Anicia plan and the Forma Urbis is the greater depth of the tabernae toward the Tiber in the block northwest of the temple in the latter plan (that, indeed, is not a mere copy of the former). This is consistent with my reconstruction of that block, in which the rear wall of the four small tabernae opening toward the Tiber is shifted backwards (Figs 9 and 15). More important, in Vitti's plan there is a widening of the left bank of the river toward the circus (Figs 1 and 13). Also the levels confirm that the warehouses discovered in 1996 must have been built on the site of the courtyard visible northwest of the temple.Footnote 82
Fig. 15. Left: the marble plan of the Via Anicia (photo: author). Right: fragments 32 g–i of the Forma Urbis, showing the position of the (missing) temple of Castor and Pollux, which would have appeared in the corner of the slab (adapted from forma.urbis.stanford.edu). Note the different depth of the taberna marked with an asterisk on the Via Anicia plan and on the Severan Forma Urbis. Also compare the tabernae depicted on the Forma Urbis behind the temple (the latter not preserved) with Vitti's plan (in the inset: from Vitti, ‘Il tempio dei Castori ‘in Circo’' (n. 14), fig. 23).
Last but not least, with my overlaying of the plans (Fig. 9), it appears that the concrete foundation seen in 1996 was built in the (originally) empty space before the portico on the northwest side of the Temple of Castor and Pollux, when the floor of the circus had already covered the steps and the gutter. In my reconstruction there is also a perfect alignment of the ashlar wall of the monumental building with the edge of the portico (Figs 9 and 10). It is not unlikely that the few courses of bricks that Vitti thought belonged to the side wall of the cult base were instead the remainder of a pre-existing Trajanic wall incorporated beneath the late antique wall, which would further dismiss his identification of the temple. According to my reconstruction, the layer of cocciopesto seen in 1996 belonged to the portico facing the Circus Flaminius, and a new building was built over that portico in late antiquity.
As for the walls discovered and recorded one century ago, these are sometimes too high, and sometimes too deep, in the stratigraphic sequence, compared to the actual levels of the travertine floors of the Circus Flaminius. Indeed, these walls were not recorded in archaeological excavations, but were seen in the bottom of trenches (c. 1.3 m wide and 8.5 m deep, well below the Imperial levels of the Circus Flaminius) made for the foundations of the outer walls of the four villini. Vitti, on the basis of misplaced archival drawings (Fig. 13), claims that the ancient walls found in 1910 do not support my identification, but on the contrary the walls that were contemporary with the Imperial floors of the circus do correspond to the walls depicted on the Via Anicia plan.Footnote 83 Moreover, the easternmost wall found under the villino Lupi is already misplaced in the general plan that was drawn after the excavations for the villini combining the different surveys (Fig. 13): it is located under the outer wall of the villino, whereas it was found inside the area of the villino itself (Fig. 10, b).Footnote 84 It is worth noting its correlation with the southeast end of the portico depicted west of the temple on the Via Anicia plan.
In February 1910, an ashlar wall 6.6 m long was recorded at the southwest corner of the villino Serventi, at a depth of 8.5 m (or perhaps 7.5 m, according to a note in a drawing). It was made of blocks 60 cm high and thick. What did the excavators find before reaching the depth of 8.5 m? This lack of information testifies to the gaps in the archaeological reports. The ashlar wall might have belonged to a structure older than the Temple of Castor and Pollux, the pronaos of which, according to my identification, partially occupied the southwest corner of this villino. Next to this ashlar wall, but at a depth of just 3 m (c. 14.5 m asl), a level much higher than the late antique floor of the circus and even higher than the floor of the temple, a series of brick-faced walls was found, but they stood southeast of my proposed pronaos (Fig. 10, c: the villino Serventi is on the right).Footnote 85
Beneath the southeast corner of the villino Arcieri (cf. Fig. 10, e), in December 1909 or January 1910, at a depth of 8 m the remains of an ashlar wall (‘costruzione ad opera quadrata’) were discovered, which might have belonged to the foundations of the front wall of the tabernae behind the temple. When the other ashlar wall mentioned above was discovered in February 1910, it was noticed that it could actually have belonged ‘alla medesima costruzione scoperta nell'adiacente villino Arcieri, giacchè l'orientamento è il medesimo, cioè da nord-est a sud-ovest, come pure la profondità è uguale’. It is more likely that they both belonged to structures older than the temple. The ‘construction in square-stone masonry’ mentioned above was actually a sort of wide foundation, and indeed it is marked as such in the general plan of 1910 (‘platea’) and in the original sketch (‘platea di tufi’). A sector of this structure at least 2.95 m wide was recorded: it was built with blocks 60 cm thick, precisely like those found beneath the other villino.Footnote 86
Nearly in the middle of the west side of the villino Arcieri another structure made of square-stone masonry was found in December 1909 at a depth of 8 m (cf. Fig. 10, e). It consisted of three courses of blocks of tuff measuring 75 cm in width and 68 cm in height, that apparently did not belong to the ‘platea’ found on the opposite side of the same villino because of the different size of the blocks. According to my location of the Via Anicia plan, this ashlar wall corresponds to the foundation of the southeast wall of the courtyard northwest of the temple (cf. Fig. 10). Indeed, one of the brick-faced walls discovered in this area in 1996 rested on at least one course of blocks of tuff.Footnote 87 The different ashlar walls seen in 1910 might have belonged to two different properties, as implied by the Via Anicia plan, although they seem to share the same alignment.Footnote 88
CONCLUSIONS
Is it possible to assume that the Temple of Castor and Pollux stood on the actual site of the monumental building, and that it was radically rebuilt and altered in late antiquity?Footnote 89 Unfortunately, there are no comparable examples of such a restoration. In particular, this strange temple would have no podium, differently from all the Republican temples of the Campus Martius restored in the Imperial age and from other temples located next to the Tiber (in the Forum Holitorium and in the Forum Boarium). If in the area of the 1996 excavation the temple has to be smaller to match its depiction on the Via Anicia plan, as Vitti implies with his reduced plans, it should be observed that the walls of the nearby warehouses dismiss such a radical restoration of the temple.
Perhaps a late antique building is considered not as appealing as a late Republican temple, but the walls and floors found in 1996 appear to belong to a building smaller than and different from the temple. The travertine floor of the Circus Flaminius was raised on at least two occasions during the Imperial age, as attested elsewhere in the Campus Martius (for example, at the Area Sacra of Largo Argentina and in the area of the Horologium Augusti), and the monumental building is clearly in phase with the highest floor (14.1 m asl). After the construction of the Temple of Castor and Pollux, other monuments occupied the area of the circus, including the Theatre of Marcellus, the Arch of Germanicus (and possibly the Arch of Drusus the Younger), and the Flavian building at Via de' Calderari. Is it conceivable that between the second and fourth centuries ad nothing else was built there? My alternative interpretation, considering the archaeological evidence as well as the Via Anicia plan vis-à-vis the Severan version, shows that towards the end of the first century ad there was not the dramatic urban change implied by Vitti's reconstruction, but just the saturation of the courtyard west of the temple. The monumental building would attest that the area of the circus was still important in the fourth century ad.
The use of blocks of tuff for its ashlar wall can be compared with the travertine blocks (re)used at the bottom of the walls of the Diocletianic Senate House. The monumental building might have been planned with a transverse hall considering the limited depth of the portico facing the circus (and it is possible that its plan was quite different: see above, pp. 98–102). An entrance like a pronaos can be found in a series of Roman non-religious buildings dating to different periods, such as the Odeon in the Agora of Athens and the ‘Library of Hadrian’ in the same city, both with tetrastyle porches. In 2007 I had made a few suggestions already, including a synagogue similar to that at Dura Europos, which would also explain the construction of the main medieval synagogue in Rome in the area of the courtyard west of the temple.Footnote 90 Of course, there are other possible alternatives.
Be that as it may, at least one can confidently exclude the identification of the monumental building with the Temple of Castor and Pollux. According to my alternative location (which fits better with the archaeological evidence, the Via Anicia plan, and the shift of the Forma Urbis fragments of the circus), the temple stood on the same north–south alignment as the oldest temples in the central Campus Martius (the Pantheon, the temples at Largo Argentina, and so on). In addition, it stood in front of Neptune, probably because the Dioscuri were held to be patrons of travellers (and of sailors in particular). If so, these ‘prerequisites’ would explain the location of the temple as close as possible to the east sector of the circus, where, however, the space between the circus itself and the Tiber was extremely narrow, with the consequent adoption of a transverse cella to avoid building the rear side of the cella on the river-bank and on huge substructures.Footnote 91 For Vitti, the lack of space behind the temple is not a concern (his warehouses even occupy the bank of the Tiber), and indeed he maintains that the choice of the transverse cella did not originate from this problem.Footnote 92 One can exclude that its cella was planned considering its insertion in the circus (it is possible that they were not even aligned), or that it was located on that side of the circus because the last temple to be built in the area (in fact, there was a lot of space available).Footnote 93 Religious reasons can be excluded, too, considering that the Temple of Castor and Pollux in the Roman Forum was a peripteros built on a tall podium. In the light of the discussion on the phenomenon of transverse cella temples as part of the ‘Hellenization’ of Roman religious architecture, the Temple of Castor and Pollux in circo should be considered as a mere architectural experimentation: other innovative buildings had appeared in the area of the Circus Flaminius, such as the Temple of Hercules and the Muses (with its circular cella and a front porch — a forerunner of transverse cella temples?), the Temple of Jupiter Stator (the first marble temple in the city), and the Porticoes of Octavius and Metellus. The Temple of Castor and Pollux in circo was just the elegant solution to a topographical problem, inspired by the similar case of the Temple of Veiovis. Apparently, in our case-study a difference of a few metres does matter, in that it may explain this temple's significance and layout.Footnote 94
