One, Two, Three! Can Everybody Hear Me? Acoustics of Roman Contiones. Case Studies of the Capitoline Hill and the Temple of Bellona in Rome

1 Introduction

In 63 BCE, in the city of Rome, there was such a violent political demonstration in the theatre^[1] that the consul of that year, Marcus Cicero, himself decided to address the people and calm their hateful reaction to the appearance of Lucius Roscius Otho^[2] in the venue (Plut. Vit.Cic.13.2-4). Otho had infuriated the people when, while holding the office of plebeian tribune 4 years earlier, he had proposed and successfully voted down a law granting 14 rows of seats in the theatre to the members of the equestrian order (cf. Cic.Corn.1, ft. 52, Att.2.19.3, Mur.19.40, Phil.2.18.44, Liv.per.99, Hor.Epod.4.15-16, Epist.1.1.62, Vell.Pat.2.32.3, Ascon.78-79C, Tac.Ann.15.32, Iuven.3.159,^[3] Maillard, 2018; Pociña Pérez, 1976; Rawson, 1987; Scamuzzi, 1969, 1970). The situation in the theatre was further exacerbated by the fact that the equites naturally applauded Roscius, which led to an exchange of insults between members of the two orders. This threatened to escalate the conflict further. Hence the intervention of the consul.

Unfortunately, the speech he gave to appease the people has not been preserved. However, we know the place from which he spoke: the podium of the Temple of Bellona in the Campus Martius. This choice was probably dictated by the proximity of the theatre, although we do not know exactly where the performances were held. This is the only time when the written sources specifically point us to this place as the speaking venue. However, given that the senate often met in the temple (cf. Bonnefond-Coudry, 1989, pp. 151–160; 47, Table II, pp. 144–145, Table IV, pp. 146–147, Table V) and there was a free space in front of it, we can guess that it was used more often. Especially, the location of the site outside the pomerium (the sacred boundary of the city) allowed politicians who were not formally allowed to be in Rome, such as provincial governors waiting for a triumph, to attend Senate meetings and speak in front of the people (Bonnefond-Coudry, 1989, pp. 143–149, 269–274).

The speaking platform on the podium of the Temple of Bellona was one of several used to address the people. In addition to it, the Republican speaking platform (rostra vetera) at the Comitium^[4] and the podium of the Temple of Castores in the Forum Romanum were used during the Republican period, as was the podium of the Capitoline temple^[5] (Döbler, 1999, pp. 139–140; Pina Polo, 1989, pp. 182–198; esp. 183–184; 1995, pp. 212–213). During the period of civil wars, the Forum Romanum underwent changes that greatly affected the rhetorical landscape of Rome’s main square (Coarelli, 1977; 1985, pp. 125–324; Frischer & Massey, 2022; Fuchs, 2021; Sumi, 2005, 2009, 2011). The old Republican platforms were replaced by the rostra Caesaris (and later rostra Augusti) and the rostra aedis divi Iulii (associated with the Temple of Caesar) (Coarelli, 1985, pp. 308–310; Fuchs, 2021, pp. 534–547, Figures 11–13; Pina Polo, 1995, pp. 212–213). During the Imperial period, the number of public speeches to the people declined, but for a long time speaking to the people of Rome remained among the important duties of the emperor and members of the imperial house (Pina Polo, 1989, pp. 171–181).

Many aspects of these speeches, such as their purpose, composition, and crowd constitution, have received the attention of researchers (cf. Bennett, 1995; May, 2002; Vasaly, 1993). However, little has been said so far about another aspect of public speaking in Rome that seems relevant: how many people were able to hear intelligibly what the speaker had to say, how large a proportion of the crowd was likely to hear the speaker (given the total capacity of the venue), and perhaps even – indirectly – how large were the average crowds that attended the contiones. Many movies and TV series set in the past present us with visions of venues filled to the brim. This is what the Forum Romanum resembles, for example, in the classic screen adaptations of Shakespeare’s Julius Caesar from 1953 to 1970, or in more recent years, in the TV series Domina (2021). Usually, we do not see anything strange in such scenes; after all, all kinds of public gatherings today look similar. They gather hundreds, thousands, and sometimes even tens of thousands of people. However, there is an important difference between the two situations of the modern and the ancient. Today, we have artificial sound systems at our disposal. Perhaps the reality of such speeches was best captured by the makers of the Life of Brian (1979), who parodied the Sermon on the Mount. On screen, we can see how the people standing at a distance from the speaker not only misheard and distorted his words, but also drowned him out with their constant chatter, or even by asking to repeat what he had said. In academia, the question of potential crowd size has only been addressed so far by Holter, Muth, and Schwesinger (2019) and Kopij and Pilch (2019), in the case of public assemblies held in the Forum Romanum, Fron, Stappmanns, Zhou, and Leistner (2019) in the case of Senate meetings in Curia, and by Boren (2018) in the case of speeches of a military nature taking place just before a battle (in this particular case, before the battles of Dyrrachium and Pharsalos, between armies led by Caesar and Pompey). We will return to a critical discussion of the results of this research and the methodology used and variables considered in Section 4.

The aim of our study is to answer the following questions: how many people could actually hear the speaker intelligibly and what percentage of the crowd had the opportunity to hear the speaker, given the maximum filling of the available space? We base our answers on the results of acoustic simulations for three case studies: the Temple of Bellona (in two periods: ca. 54 BCE and ca. 14 CE) and the Capitoline temple (Figure 1). Furthermore, we intend to compare the acoustic properties of both venues and to consider whether practical factors influenced the choice of speaking platform. We then compare the results with those of other similar studies carried out for rostra in the Forum Romanum.

Figure 1

Plan of Rome with marked venues discussed in the article (black dot for Capitolium and red for the Temple of Bellona).

2 Methods

2.1 3D Reconstructions

In order to run the simulations, we built virtual 3D reconstructions of all venues. Doing so, we took into account the state of archaeological and architectural research on both spaces as well as reconstructions previously suggested in the literature. The reconstructions were built with an accuracy of 0.1 m, because in order to achieve realistic acoustic analysis results, higher accuracy is not needed, and the process would only increase the computational time. Objects close to the sound source were reconstructed with smaller details (where possible). Objects far away from the sound source were reconstructed more roughly. The models were constructed using Blender 2.91, according to the current knowledge of the possible appearance and construction of each structure during the Late Republican and Early Augustan periods. Export of the models to CATT-acoustic was made by plug-in created in authors team.

The state of research and the later centuries of transformation of the spaces of interest make our proposed reconstructions hypothetical in some details. This particularly concerns the reconstruction of the appearance of the Capitoline Hill, where we can only be fairly certain of the look and geometry of the Temple of Jupiter Optimus Maximus. The remaining buildings are debatable and their reconstructions, and even their spatial interrelations, differ considerably amongst the reconstructions proposed so far. After getting acquainted with the literature and contacting Alberto Danti, who was conducting the excavation from 2000 to 2018, it was decided to reconstruct the Capitoline Hill as it had already been reconstructed in model form in the Museum of Roman Civilisation (Museo della civiltà romana), with slight changes. There is very little certain information, so the reconstructions are mostly based on the experiences and intuitions of the researchers. For this reason, it was concluded that the best solution is to reconstruct the Capitoline Hill as the best-known structure and not to create a completely new version (Figure 2a and b. Bibliography for the reconstruction, see Appendix 1, 3D Virtual Reconstruction, see Appendix 2).^[6]

Figure 2

(a) Screenshot of the 3D model of the Capitoline Hill. (b) Screenshot of the 3D model of the Capitoline Hill, alternate view. (c) Screenshot of the 3D model of the Republican period Temple of Bellona with surroundings. (d) Screenshot of the 3D model of the Early Imperial period Temple of Bellona with surroundings.

For the Temple of Bellona at Campus Martius, together with the surrounding buildings, we made two reconstructions corresponding to two chronologically close periods. The first corresponds to the state of the Republican period, before the construction of the Theatre of Marcellus began during Caesar’s dictatorship in 45 BCE^[7] (cf. Suet.Iul.44, Cass. Dio 43.49.3) (Figure 2c). The second reflects the geometry of the space after the erection of the aforementioned theatre, which was finally commissioned in 13–11 BCE (Plin.HN.8.6 Cass. Dio 54.26.1, Ciancio Rossetto & Pisani Sartorio, 1999, pp. 31–35; Richardson, 1992, pp. 382–383) (Figure 2d). Again, the reconstruction, at least in part, is hypothetical. For, while the dimensions and spatial relationships of the main buildings (the Temple of Bellona itself, the adjacent Temple of Apollo Sosianus, or the Theatre of Marcellus) are not much in doubt, the state of research on the others does not always allow for an accurate reconstruction. In creating ours, we have relied on the results of archaeological research and on previous proposals to reconstruct this location in the period of interest, i.e. ca. 54 BCE and ca. 14 CE (Bibliography for the reconstruction, see Appendix 3, 3D Virtual Reconstruction, see Appendices 4 and 5).

3 Results

In the case of the Capitoline Hill, we unfortunately do not have more detailed information on the location of the speaker and his orientation. For, in the case of this temple, we know nothing about the existence of a clearly marked rostra. Biliński (1961, p. 269) postulated that during Tiberius Gracchus’ attempt for re-election as tribune, he was positioned atop the podium of the Capitoline temple, potentially on a platform specifically arranged for the occasion. However, Ulrich speculated that the temporary platform could have been situated in front of the temple or integrated into the podium itself (Ulrich, 1994, pp. 64–66). Similarly, Taylor considered the latter possibility as the more likely scenario (Taylor, 1966, p. 20). Consequently, we run simulations for two scenarios. First, we assumed that the speaker was standing at the top of the stairs leading to the temple’s podium (“Podium scenario”). He was speaking facing a potential crowd gathered at the foot of the temple (and partly on its steps). It is possible, however, that the speaker may have been speaking from the monumental altar located in front of the temple facing the Forum Romanum (“Altar scenario”).

Our simulations for the “Podium scenario” show that for the RAN background noise level of the (55 dBA), the group of people who could hear the speaker intelligibly does not exceed 200 participants. This is approximately 180 people for the average crowd density factor, and 192 people for the density moderated by STI zones. The number of individuals who could hear the speaker intelligibly increases to approximately 1,044 and 1,104 in the TAN setting (49 dBA). In the least-probable LLAN setting (36 dBA), the speaker’s words could be understood by roughly 8,676 and 9,832 individuals. The results are summarised in Table 5 and Figures 3–5 (cf. Appendices 6–8).

Figure 3

STI map for the “Podium scenario” of the Capitolium at RAN.

Figure 4

STI map for the “Podium scenario” of the Capitolium at TAN.

Figure 5

STI map for the “Podium scenario” of the Capitolium at LLAN.

The simulations for the “Altar scenario” instead indicate that for the background noise level of the RAN (55 dBA), the group of people who could hear the speaker intelligibly was much greater reaching approximately 500 participants. This is approximately 480 people for the average crowd density factor, and 512 people for the density factor moderated by the STI zones. In the TAN setup (49 dBA), these figures rise to approximately 2,052 and 2,260 people, respectively. In the least-realistic LLAN setting (36 dBA), the speaker could be intelligibly heard by approximately 8,976 and 10,564 people, respectively. The results are summarised in Table 6 and Figures 6–8 (cf. Appendices 9–11).

Figure 6

STI map for the “Altar scenario” of the Capitolium at RAN.

Figure 7

STI map for the “Altar scenario” of the Capitolium at TAN.

Figure 8

STI map for the “Altar scenario” of the Capitolium at LLAN.

In the case of the Temple of Bellona, we can assume that the speaker was standing on its podium, speaking towards the square in front of the temple. The area of this square was significantly reduced with the construction of the Theatre of Marcellus. Our simulations for the Republican period for the RAN setup indicate that a speaker delivering from that location was heard intelligibly by approximately 384 people, when we assume the average crowd density, and 428 people for the moderated density. In the TAN setting, these numbers rise to approximately 1,500 and 1,668 people, respectively. And for the LLAN setup, there are approximately 13,344 and 14,580 potential listeners. The results are summarised in Table 7 and Figures 9–11 (cf. Appendices 12–14).

Figure 9

STI map for the Temple of Bellona in Late Republican period at RAN.

Figure 10

STI map for the Temple of Bellona in Late Republican period at TAN.

Figure 11

STI map for the Temple of Bellona in Late Republican period at LLAN.

The change in the geometry of the space associated with the construction of the Theatre of Marcellus significantly affected the size of the potential crowd that could gather in front of the temple and the acoustic properties of the space. For the RAN setup, the number of people who could hear the speaker intelligibly was approximately 264 for the average density and 288 for moderated density. In the TAN scenario, these numbers increase to approximately 1,344 and 1,432 people, respectively. And for the LLAN setting, they were approximately 7,092 and 7,688 people, respectively. The results are summarised in Table 8 and Figures 12–14 (cf. Appendices 15–17).

Figure 12

STI map for the Temple of Bellona in Late Augustan period at RAN.

Figure 13

STI map for the Temple of Bellona in Late Augustan period at TAN.

Figure 14

STI map for the Temple of Bellona in Late Augustan period at LLAN.

4 Discussion

First, comparing the simulation outcomes for the two scenarios involving the placement of the speaker at the Capitoline temple area, it was observed that speaking from the altar located in front of the temple had a significant advantage in terms of reaching a larger audience who could clearly comprehend the speech, as opposed to speaking from the temple stairs.^[10] In addition to the acoustic properties of the space, this is related to the fact that people can gather around the speaker and not just in front of him. When considering the RAN setup, the estimated audience size in front of the temple was only around 37% of the potential attendees who could have gathered around the altar and heard the speaker’s words distinctly (approximately 180–192 vs 480–512 attendees, respectively). In the case of the TAN setting, the percentage difference is smaller, as it turns out that the crowd is approximately 1,044–1,104 people for a speaker delivering his speech from the temple steps and 2,052–2,260 for a speaker delivering the speech from the altar. Although speaking from the altar was possible (after all, it was a location for other public rituals), the results imply that future reconstructions of the Capitoline temple from this era should consider the option of a speaking platform incorporated into the temple steps which has already been considered by Biliński (1961, p. 269), Taylor (1966, p. 20), and Ulrich (1994, pp. 64–66), but has not been included in any reconstruction to date. It is important to mention, nonetheless, that no iconographic source provides evidence of the existence of such a platform at that site.

Before comparing the results of the acoustic simulations for the two phases of the Temple of Bellona, it is worth looking at how the total available space in front of the temple in both periods – which could theoretically be occupied by assembly participants – has changed. The visibility analysis^[11] shows that the area available during the Republican period was approximately 12,500 m², whereas the construction of the Theatre of Marcellus reduced it to less than a third of this value, i.e., approximately 4,000 m² (cf. Figures 15 and 16). Although the areas where speech was intelligibly heard cover much smaller areas, the changes in the built environment – most notably the rebuilding of the temple itself and the construction of the Theatre of Marcellus – also negatively affected the size of the potential crowds.

Figure 15

Viewshed analysis for a speaker standing at the podium of Late Republican Temple of Bellona.

Figure 16

Viewshed analysis for a speaker standing at the podium of Late Augustan Temple of Bellona.

In the RAN scenario during the Republican period, intelligible speech could only be heard by approximately between 384 and 428 people, depending on how the crowd size was estimated. In the early Imperial period, these potential crowd sizes reduced to approximately between 264 and 288. Although these figures are generally not impressively high, they represent a sizable reduction of more than 30%.

When we consider the TAN scenario in the Republican period, the ancient rhetors’ speeches would have been heard intelligibly by approximately between 1,500 and 1,668 people, depending on how the crowd size is estimated, and in the Imperial period approximately between 1,344 and 1,432 people. This represents a reduction of approximately 10–15% between the Republican and Imperial periods.

In the least-likely scenario of the LLAN, during the Republican period, speeches delivered from the podium of the Temple of Bellona could be heard by approximately between 13,344 and 14,580 people, depending on how the crowd size is estimated. In the Imperial period, this figure drops to approximately between 7,092 and 7,688 people. Only in this case does the construction of the Theatre of Marcellus directly affects the size of the potential crowd, rather than by changing the acoustic properties of the space in front of the temple. The reduction in crowd size during the Imperial period relative to the Republican period in this scenario is almost 50%.

We can now compare these results with the acoustic simulation outcomes for the Capitolium. To begin with, it should be highlighted that the “Podium scenario” for the Capitoline temple has a lower potential maximum crowd size compared to the temple of Bellona in its both phases. According to the simulation results, the estimated maximum crowd in front of the Capitoline temple is over 50% smaller than that in front of the temple of Bellona for the RAN setup in the Republican period (approximately 215 people less), and over 30% smaller in the Imperial period (approximately 90 people less). At the TAN setting, the disparities are less pronounced, with the maximum crowd in front of the Capitoline temple being around 33% smaller than in front of the Bellona temple during the Republican period (approximately 500 participants less) and less than 25% smaller during the Imperial period (approximately 315 people less). When comparing the results for the Bellona temple with the “Altar scenario” of the Capitoline temple, an inverse relationship between maximum crowd sizes can be observed. In the Republican period, at the RAN setting, the crowd in front of the Bellona temple was approximately 20% smaller (approximately 100 people less), while in the Imperial period it was approximately 45% smaller (approximately 225 people less). On the other hand, for the TAN setup, the crowd in front of the Bellona shrine was slightly more than 25% smaller (approximately 560 persons less) during the Republican period and one third smaller (approximately 700 persons less) during the Imperial period.

We can also cross-reference our results with other studies of this type for other venues in ancient Rome. To date, the results of two such studies have been published. The first investigated the maximum number of people who could hear intelligible speeches delivered from the rostra vetera towards the Comitium and towards the open space of the Forum Romanum as well as the Temple of Castores in the Late Republican period (Holter et al., 2019). Although the study was carried out for a similar period, the results are difficult to compare directly, as the authors of the study did not express variables such as the SPL of the source and, above all, background noise by numbers. For the latter variable, it was only indicated that the source was a recording from St Peter’s Square in the Vatican during the Sunday Angelus prayer of the Pope (which is a good idea; it is unfortunate that the authors did not express this numerically). The only variable expressed numerically was a crowd density of four people per square metre. The results are summarised in Table 9.

The differences in results are striking and can in no way be explained by variations in space geometry alone. They must result primarily from the use of different values of variables, above all SPL and background noise level. A comparison with our results (Table 10) shows that Holter et al. used a very low background noise value. Even with its value at the LLAN setup – which we believe is unrealistic – only the results for the Republican phase of the venue in front of the Temple of Bellona fall within a range containing similar values. Using the same crowd density factor, its “capacity” could have reached approximately 17,792 people. In that case, the Capitolium with a “capacity” of approximately 11,024 people would, in turn, fall within the lower limits of the ranges proposed by Holter et al. All this indicates, in our opinion, that these results need to be verified, as they are overestimated due to the assumption of unrealistic background noise levels.

It is easier to compare our results with those of our preliminary study of the Forum Romanum (Kopij & Pilch, 2019) because everything was expressed in numbers in the same way (cf. Table 11). This study included the rostra Augusti and the podium of the Temple of Castores after the removal of the Temple of Deified Caesar (aedes divi Iulii). The biggest methodological problem with the study, which was primarily to test the method, is that a slightly modified model of the Forum from the Late Roman period (ca. 320 CE) was used. This, in turn, differed in many details from the periods of the Late Republic and Early Empire that interested us most.

It is particularly notable that in the Late Republican Period both the venue at the altar of the Capitoline temple and the venue at the temple of Bellona performed better in terms of maximising the size of the crowd than the Temple of Castores, a popular speaking venue located in the Forum Romanum (cf. Table 12(a)). The worst performer in this respect is the podium of the Capitolium. Unfortunately, at the moment, we do not yet have our own simulation results for the Republican rostra vetera, and a direct comparison of the results with those of Holter et al. (2019) is not possible due to methodological differences. In the future, we intend to fill in these gaps in order to get a complete picture of the rhetorical landscape of the ancient city of Rome.

With regards to the Early Imperial period, the results so far show that the “Altar scenario” of the Capitolium had the greatest potential, followed by the rostra Augusti and the Temple of Bellona. Again, the podium of the Capitolium is the worst performer in this comparison (cf. Table 12(b)).

For the Late Republican period, the same ranking also holds for the lower level of background noise, i.e., TAN (cf. Table 13(a)). Again, it can be concluded that the Altar area in front of the Capitoline temple provided the greatest “capacity.” Interestingly, despite the erection of the Theatre of Marcellus, in the Early Imperial period the “capacity” of the venue in front of the temple of Bellona slightly exceeded the rostra Augusti, which must be considered the “nominal” rostra of this period in Rome (cf. Table 13(b)). The difference is, however, very small, leading to the conclusion that both venues had a similar capacity to maximise the people who could intelligibly hear the speeches delivered there.

Only for what we consider to be a least-realistic LLAN setup does the platforms located on the Capitoline Hill perform better. It ranks in terms of potential “capacity” behind the Temple of Bellona in the Republican period, and in the early Imperial period, it accommodates the largest number of people able to intelligibly hear the speakers of all the cases analysed. Interestingly, the worst performer in this respect is the rostra Augusti (cf. Table 14).

Arguably, only greater differences in the ability to maximise a crowd that could intelligibly hear public speeches were noticeable to the ancient Romans. The question is whether the difference of approximately 200 such people, as in the case of the Late Republican temple of Bellona and the temple of Castores for the TAN setup, was enough for them to notice the difference. If so, the speakers were free to consciously decide where they wanted to speak so as to achieve their objectives. This does not necessarily mean that they always chose venues that maximised the number of people who could intelligibly hear them. It is possible that sometimes the total capacity of the venue may have been more important to them, even if the vast majority of those gathered could not hear clearly what they were saying. In other cases, the venue may have been determined by other, less practical reasons, such as the nature of the issue raised by the speech, the proximity of other events (as in the case of the theatrical performance cited in the introduction), or yet other issues, such as the level of security in the case of Brutus’ speech after Caesar’s assassination (cf. Lendon, 2022, pp. 29–50). Only a full reconstruction of the rhetorical landscape of the ancient city of Rome will perhaps allow us to say how important the issue of audibility of speech was.

5 Summary

Speaking before the people was an important part of ancient Roman public life, especially during the Republican period. Although its importance declined during the Imperial period, public speaking remained an essential part of the duties of emperors and members of the imperial house. The aim of this article was to answer the question of how many people could intelligibly hear speeches delivered in two venues: the Capitoline Hill (for two scenarios for the positioning of the speaker) and in front of the Temple of Bellona in the Campus Martius in Rome (in this case, in two close historical periods: at the end of the Republic and in the late Augustan period). Carrying out acoustic simulations led us to answers that allowed us to compare venues in terms of acoustic properties, looking for the one that maximised the number of people who could intelligibly hear speeches being given. Our findings indicate that, except for the LLAN (36 dBA), the speaking from the stairs in front of the Capitoline temple provided the worst acoustic conditions for maximising the audience size that can intelligibly hear the speaker. While historical sources do not explicitly mention the speaker’s position, it is likely that speakers used a platform that was integrated into the temple’s podium (as was the case with other templa rostrata), rather than the altar in front (which compares best according to our results). Hence, our results support the researchers who postulated the existence of such a platform, which has so far not been considered in reconstructions.

The location in front of the temple of Bellona is superior in terms of maximising the crowd to the steps of the Capitoline temple, but inferior to the altar situated in front of it. Later developments, particularly the erection of the Theatre of Marcellus in front of the temple, which reduced the total area available in front of it, worsened the acoustic properties, thus reducing the number of people who could potentially hear the speeches. Also, comparisons with studies of other speaking venues – the Republican rostra vetera and the Temple of Castores, both at the Forum Romanum – indicate that Temple of Bellona as well as the altar in front of the Capitolium had better acoustic properties. In this case, however, the results presented here should be considered preliminary as there are methodological limitations in the studies used for comparison. Properly positioning all the rostra by the maximum number of people who could intelligibly hear speakers can only be achieved by conducting simulations for historically accurate reconstructions of the Forum Romanum and the use of a comparable methodology.