1. A QUESTION OF EVIDENCE
The city wall of Republican Rome is not well understood. Its date of construction is disputed (the alternatives two centuries apart), and in many places even its course is not known. But Rome's walls were what defined her, and not just spatially.Footnote 1 For Sallust (Cat. 6.2, trans. A.J. Woodman), the origin of Rome was when wandering Trojans and primeval Aborigines ‘had come together behind a single wall’.Footnote 2 ‘So perish all henceforth who cross my walls!’ says Romulus in the opening scene of Livy's foundation narrative (1.7.2), and in Virgil's Aeneid ‘the walls of lofty Rome’ (1.7, altae moenia Romae) are a constantly repeated synecdoche for the outcome of the hero's quest.Footnote 3 The walls were sacred, subject to divine law and could not be owned by any individual.Footnote 4 When and how they were built and rebuilt are questions central to the history of the republic.
The evidence for early Rome, however, is notoriously controversial.Footnote 5 Much new archaeological information has become available in the last twenty or thirty years (Fulminante, Reference Fulminante2014: 66–104; Hopkins, Reference Hopkins2016), but it is quite unclear how far, if at all, it is compatible with the narrative offered by the literary sources.Footnote 6 Archaeologically, for instance, the origin of Rome should be either (a) the first sign of continuous occupation of the site, which is about 1400 BC on the Capitol (Lugli and Rosa, Reference Lugli and Rosa2001; Fulminante, Reference Fulminante2014: 67–71), or (b) the first sign of the institutions of a city-state, which is about 650 BC, with the creation of the great landfill that made space for the Roman Forum (Ammerman, Reference Ammerman1990; Hopkins, Reference Hopkins2016: 27–38). Neither coincides with the eighth-century foundation date variously calculated at the start of the Roman historiographical tradition.Footnote 7 That was a wholly artificial construction,Footnote 8 resulting from the chronological framework recently established by Timaeus and Eratosthenes, which made it both possible and essential to provide dates for the regal period (Feeney, Reference Feeney2007: 86–100, esp. 90–2). Traditional stories of early Rome had ignored chronology, making Romulus the grandson of Aeneas and Numa a disciple of Pythagoras.Footnote 9
The canonical list of seven kings presupposes this construction, and may even have been invented to make it possible (Wiseman, Reference Wiseman2008b: 314–15).Footnote 10 The treatment of the city wall in the historiographical tradition takes the list for granted, attributing its construction to Servius Tullius or one of the other kings (Ziółkowski, Reference Ziółkowski2019: 69–92). Strabo provides a convenient summary of what was generally believed in the late first century BC:Footnote 11
The first [Romans] walled the Capitol and the Palatine and the Quirinal hill, the last of which was so easily accessible to outsiders that Titus Tatius took it at the first assault when he came to avenge the outrage of the abducted girls.Footnote 12 Ancus Marcius too, adding the Caelian mount and the Aventine mount, and also the flat area between them, did so by necessity, because they were distant both from each other and from the parts that had been previously walled. It was not a good idea to leave such strong hill-sites outside the walls for anyone who wanted hostile fortresses, but he was unable to complete the whole circuit as far as the Quirinal. Servius, correcting the deficiency, completed it by adding the Esquiline and Viminal hills.
Though some authors attributed it to Tarquinius Priscus rather than Servius,Footnote 13 the complete city wall was universally assumed to be a work of the regal period.Footnote 14
The surviving remains of the circuit are built largely of Grotta Oscura tuff from the territory of Veii, a resource accessible to the Romans on such a scale only after 396 BC;Footnote 15 some parts, however, evidently exploited earlier construction in cappellaccio. Two explanations are possible: either the great fourth-century defences replaced an earlier sixth-century circuit wall on much the same line (Cifani, Reference Cifani2008: 237–64), or they were a new creation, but used at certain points pre-existing walls on individual hills (Bernard, Reference Bernard2012). Despite repeated and forceful assertions of the former solution,Footnote 16 the archaeological evidence is compatible with either. Simple faith in the historicity of Ancus Marcius and Servius Tullius is not enough to justify the first alternative.Footnote 17
As for the course of the wall, we know from Dionysius that in the late first century BC it was hard to trace it in some places because of buildings surrounding it;Footnote 18 and if it was hard for him, it is all the more so for modern investigators faced with the debris of another 2,000 years of building, destruction and rebuilding.Footnote 19 Most difficult of all is the part of the circuit closest to the river, between the Aventine and the Capitol. Filippo Coarelli gives a succinct explanation of the problem (Fig. 1) (Coarelli, Reference Coarelli2007: 18; square-bracketed numerals added):
[1] According to one theory, the side of the city that faced the river would not have been enclosed in the fortification; that is, the walls are believed to have run directly from each hill to the river, leaving the area between exposed. Thus, the Porta Trigemina would have served the wall that stretched from the Aventine to the Tiber, while the section extending from the Capitoline to the Tiber would have contained the Porta Flumentana and the Porta Carmentalis. [2] Another theory envisages a wall running parallel to the Tiber that descended the Aventine and made its way towards the southwestern corner of the Palatine, from which point it ultimately reached the Capitoline. [3] Recent investigations suggest that a course parallel to the Tiber would appear to be the more likely, although it would have had to run closer to the river than previously thought.
Option 2 had authoritative support at one time (von Gerkan, Reference von Gerkan1931; Säflund, Reference Säflund1932: 138–9, 176–85),Footnote 20 but it now seems clear that the supposed evidence from the Palatine has nothing to do with a circuit wall of the whole city.Footnote 21 The choice lies between options 1 and 3, both of which have recently featured (without argument) in major works of reference.Footnote 22
Fig. 1. The three possible lines for the circuit wall between Aventine and Capitol.
Coarelli himself has consistently championed option 3, and in 1988 his book on the Forum Boarium presented the arguments at length.Footnote 23 Reservations were expressed at the time,Footnote 24 but he has not changed his mind. In his new book on administrative buildings (Coarelli, Reference Coarelli2019: 135–55), he returns to the subject, resuming and updating his arguments of 30 years ago (Fig. 2). There is also some new archaeological evidence to be taken into account: we now know that in the sixth century BC the riverbank between the Aventine and the Capitol stood about 100 m east of its present position (Ammerman, Reference Ammerman, Haselberger and Humphrey2006: 305–7). So perhaps a fresh assessment is required.
Fig. 2. ‘Pianta del Foro Boario’ (Coarelli, Reference Coarelli2019: 136, fig. 66, reproduced by permission). Items 9, 11 and 22 are the proposed sites of the Porta Carmentalis, the Porta Flumentana and the Porta Trigemina, respectively.
On each of these questions — the date of the city wall and its course between Aventine and Capitol — I believe progress can be made by paying proper attention to what is presupposed in certain episodes narrated by Livy, Dionysius and Plutarch. The question is not whether the stories are true, but what was taken for granted about the walls and gates when they were first composed.
2. THE DATE OF THE CITY WALL
Let's look at an event that took place in what we call 460 BC, about a century after King Servius Tullius supposedly built the city wall round all the hills of Rome.
As Livy (3.15.4) tells it, ‘exiles and slaves, about 4,500 men under the leadership of the Sabine Appius Herdonius, took control by night of the Capitol and arx’.Footnote 25 How did they do that? Dionysius’ version (Ant. Rom. 10.14.1–2, Loeb trans. Earnest Cary) gives more detail:Footnote 26
A man of the Sabine race, of no obscure birth and powerful because of his wealth, Appius Herdonius by name, attempted to overthrow the supremacy of the Romans, with a view either of making himself tyrant or of winning dominion and power for the Sabine nation or else of gaining a great name for himself. Having revealed his purpose to many of his friends and explained to them his plan for executing it, and having received their approval, he assembled his clients and the most daring of his servants and in a short time got together a force of about four thousand men. Then, after supplying them with arms, provisions and everything else that is needed for war, he embarked them on river-boats and, sailing down the river Tiber, landed at that part of Rome where the Capitol stands, not a full stade distant from the river. It was then midnight, and there was profound quiet throughout the entire city; with this to help him he disembarked his men in haste, and passing through the gate which was open (for there is a certain sacred gate of the Capitol, called the Porta Carmentalis, which by the direction of some oracle is always left open), he ascended the hill with his troops and captured the fortress. From there he pushed on to the citadel, which adjoins the Capitol, and took possession of that also.
The event itself is perfectly credible: independent warlords with their own armies were a familiar phenomenon in central Italy in the sixth and fifth centuries BC.Footnote 27 But the idea that such a force could just walk through a gate in the city wall, left open ‘by the direction of some oracle’, is an absurdity that demands explanation.
The same is true of the capture of Rome by the Gauls 70 years later. According to Livy (5.38.10), after the rout of their army at the river Allia the Romans were so panic-stricken that they left the gates in the city wall open; the Gauls eventually walked in through the Porta Collina just as Herdonius’ men had done through the Porta Carmentalis.Footnote 28 Diodorus’ version (14.115.5–6), equally implausible, is that the Romans closed the gates but did not man the walls, and the Gauls broke in at their leisure.Footnote 29
There can be only one reason for these absurdities: each narrative assumes the existence of the city wall, and has to find a reason why it did not do its job. Without it, the events are perfectly intelligible. Herdonius’ men seized the Capitol in a surprise attack and could only be dislodged with great difficulty and the help of an allied army from Tusculum (Livy 3.18; Dion. Hal., Ant. Rom. 10.16.3); after the Allia rout the Romans retreated into the Capitol because it was the only defensible place left to them.Footnote 30 The natural inference is that (a) the city wall did not exist before the 370s BC, and (b) it was already wrongly attributed to the regal period by the time the Romans started writing history at the end of the third century BC.
Where did Livy and Dionysius get these stories from? Who was it, in the long history of creative historiography from Fabius Pictor to Aelius Tubero (say, 210 to 30 BC), who first wrote these narratives down? No one knows, but whoever it was clearly had to accommodate them to the wall system of their own time. The events of 460 and 390 BC only make sense if the city wall did not yet exist; on the other hand, the source(s) of Livy and Dionysius believed that the city wall did exist at that time, and had to invent special circumstances (the oracle, the panic) to explain its ineffectiveness.
3. RIVER AND WALLS
Assuming then that the city wall was actually constructed in the 370s (Livy 6.32.1), we may turn to the area between the Aventine and the Capitol, and the choice that has to be made between Coarelli's first and third options on the course of the wall (Fig. 1).
The most direct evidence is a passage of Dionysius, reporting a threatened attack by the Aequi and Volsci in 463 BC. The historian tells his readers that Rome at that time had a circuit wall as extensive as that of Athens (Ant. Rom. 9.68.2),Footnote 31 and he goes on to describe it (Ant. Rom. 9.68.2, Loeb trans. Earnest Cary, slightly adjusted):Footnote 32
Some parts, on hills and sheer cliffs, have been fortified by nature herself and require but a small garrison; others are protected by the river Tiber, the breadth of which is about four hundred feet and the depth capable of carrying large ships, while its current is as rapid as that of any river and forms great eddies. There is no crossing it on foot except by means of a bridge, and there was at that time only one bridge, constructed of timber, and this they removed in time of war.
That seems to describe option 1. So too does the heroic exploit of Horatius Cocles in the second year of the republic, which depends on the assumption that once Lars Porsenna's Etruscans had seized the Janiculum, Rome's only line of defence was the bridge (Livy 2.10.1–2, trans. T.J. Luce):Footnote 33
Some parts seemed adequately protected by walls, others by the barrier of the Tiber. The wooden pile bridge, however, almost gave the enemy entrance into the city, but a single man, Horatius Cocles, stopped them.
He stopped them until the Romans had time to cut the bridge down. Dionysius spells out the situation explicitly (Ant. Rom. 5.23.2, Loeb trans. Earnest Cary): ‘the city came very near being taken by storm, since it had no walls on the sides next the river’.Footnote 34
Like the story of Appius Herdonius and the story of the Gauls after the Allia rout (see Section 2), the story of Horatius Cocles would make good historical sense if Rome did not yet have a circuit wall. As Livy's Horatius tells the consul (2.10.4), if the enemy crossed the bridge there would soon be more of them on the Palatine and Capitol than on the Janiculum.Footnote 35 There was nothing else to stop them. But both Livy (1.44.3) and Dionysius (Ant. Rom. 4.13.3, 4.14.1) took it for granted that Servius Tullius had built the city wall two generations earlier.Footnote 36 For their Horatius story to be intelligible, the wall as they and their sources understood it cannot have extended along the riverbank. It seems, then, that their narratives at this point offer conclusive evidence for the first of the three options, that the circuit wall did not run parallel to the Tiber.
A necessary condition of option 1 is that ‘cross-walls’, perpendicular to the riverbank, must have run to the Tiber from the Capitol and the Aventine, and for that too the evidence of historical stories may be helpful. When Appius Herdonius brought his four and a half thousand men down the Tiber to take the Capitol, he must have disembarked them in the Campus Martius; nowhere else near the city offered enough riverbank space. As we saw, they supposedly got into the city through the Porta Carmentalis, wide open ‘because of some oracle’ (Dion. Hal., Ant. Rom. 10.14.2). Was it imagined as a gate in a wall parallel to the Tiber, or in a ‘cross-wall’ perpendicular to it? Another story may give us the answer.
The Gauls hold the city; the Romans are besieged on the Capitol; the army that fled from the river Allia is at Veii, and wants the exiled Camillus to command it; in the version followed by Livy (5.46.4–10) and Plutarch (Vit. Cam. 24–5), the young Pontius Cominius volunteers to go and get authorization from the Senate.Footnote 37 According to Plutarch he reached the city as darkness fell, and seeing that the bridge was guarded, crossed the Tiber with the assistance of cork floats and kept away from the Gauls’ watch-fires (Vit. Cam. 25.2):Footnote 38
He made his way to the Porta Carmentalis, where it was quietest; at the gate the Capitoline hill rose most sheer, surrounded by a huge and jagged cliff which he climbed up unnoticed.
Having got the Senate's approval, he went back the way he had come ‘with the same good fortune, unnoticed by the enemy’ (Vit. Cam. 25.4).Footnote 39
Obviously Cominius did not pass through the gate, which the Gauls would have kept shut. Livy, telling the same story, does not even mention the gate, only the shrine of Carmentis adjacent to it (5.47.2).Footnote 40 Presumably the common source of Livy and Plutarch referred to the gate and shrine merely as a topographical marker, identifying the exact point where Cominius climbed the cliff. We may also assume that Cominius did not cross the river where the guards on the bridge might see him; in Livy (5.46.9) he floats downstream and comes ashore where the Capitol cliff was closest to the riverbank, which must be by the island.Footnote 41
What is at issue here is not the reality of 390 BC. As we saw in Section 2, it is more likely than not that at that time no circuit wall yet existed. We are dealing not with authentic information from the early fourth century BC, but with the elaboration of a famous episode by someone writing in the historiographical tradition from Fabius Pictor to Aelius Tubero that provided Livy with his sources. Those writers, who assumed that the city wall was a work of the regal period (see n. 14), knew very well where it ran and where the gates were. So the question is, what layout of the wall did the author of this story take for granted?
The answer is given by Plutarch's mention of the bridge and its guards (Vit. Cam. 25.1).Footnote 42 The imagined situation is exactly as in the story of Horatius Cocles: the walls defended the city everywhere except where the river defended it;Footnote 43 at that point the bridge was the equivalent of a city gate. The Pontius Cominius story gives us what the Appius Herdonius story did not: a clear indication that the city wall ran across from the Capitol to the riverbank. The Porta Carmentalis was evidently in that ‘cross-wall’, and so option 1 (Fig. 1) seems to be confirmed.
4. PORTA FLUMENTANA
The most powerful argument in favour of Coarelli's preferred option 3 (Fig. 2) is the existence of the Porta Flumentana, since a ‘river gate’ implies a wall that ran parallel to the Tiber. This gate is not often mentioned, and the references that do exist are not straightforward.
One might think at first of a wall on the riverbank itself, with the gate leading straight on to a bridge or jetty. But that is ruled out by a passage in Varro's third dialogue on farming, of which the dramatic date is 54 BC. The scene is set in the Villa Publica, outside the city walls at the edge of the Campus Martius,Footnote 44 and the point at issue is whether it counts as a uilla:Footnote 45
Just because a building is outside the city, it does not make it a villa any more than the buildings of those who live outside the Porta Flumentana or in the Aemiliana.
We may leave aside the problems presented by the Aemiliana;Footnote 46 it is enough for the present argument that there was an area outside the Porta Flumentana where people lived.Footnote 47
That is confirmed by two references in Livy, reporting portentous events in 193 and 192 BC (35.9.2–3, 35.21.5, trans. Roberts, Everyman Library):Footnote 48
There was an enormous rainfall that year and the low-lying parts of the city were inundated by the Tiber. Near the Porta Flumentana some buildings collapsed and fell in ruins. The Porta Caelimontana was struck by lightning and the wall adjacent was struck in several places.
The flooded Tiber made a more serious attack upon the city than in the previous year and destroyed two bridges and numerous buildings, most of them in the neighbourhood of the Porta Flumentana.
Note the different circumstances of the two gates mentioned: the wall (murus) next to the Porta Caelimontana, buildings (aedificia) around the Porta Flumentana. The latter was in a built-up area, the wall perhaps concealed by the buildings as in Dionysius’ time (Ant. Rom. 4.13.5).
The next reference requires some background information. On 9 December 50 BC, on his way back from his province, Cicero wrote to Atticus from Trebula in Campania. He was responding to several letters, one of which was on financial matters, much on his mind given the likelihood of civil war (Att. 7.3.6, trans. Shackleton Bailey):Footnote 49
I come to private matters. One thing more, about Caelius: so far from letting him influence my views, I think it a great pity for himself that he has changed his [and gone over to Caesar's side]. But what is this about Lucceius’ properties being knocked down to him? I am surprised you didn't mention it.
In fact Cicero is more specific about the property Caelius had acquired than Shackleton Bailey's translation suggests. As Robert Palmer pointed out, the phrase uici Luccei (‘rows of town houses’ in Shackleton Bailey's annotation) could be a toponym, ‘the Lucceius Streets’, with Luccei as nominative plural, not genitive singular.Footnote 50 Where the property was appears later in the letter, where Cicero is writing about inheritances (Att. 7.3.9, trans. Shackleton Bailey):Footnote 51
I note Hortensius’ legacies. Now I am longing to know what the heir [?] gets and what items he's putting up to auction. Since Caelius has occupied the Flumentane Gate, I don't see why I shouldn't acquire Puteoli.
It is a reasonable assumption that one of the streets Caelius had acquired ran through the Porta Flumentana.Footnote 52
Roadworks in 1939 at the piazza Bocca della Verità brought to light a dedication, dated 26 April AD 161, to the protecting deity of the ‘Lucceian warehouse’ (cella Lucceiana).Footnote 53 The find-spot was only a few metres from the riverbank, so if the cella Lucceiana took its name from the uici Luccei mentioned by Cicero, it seems to follow that the Porta Flumentana was just where Coarelli puts it (Fig. 2), in a wall parallel to the Tiber, over a street leading to a bridge.
Only two other sources mention the river gate. According to Paulus’ epitome of Festus (79L), ‘The Porta Flumentana is so called because they say part of the Tiber flowed that way,’Footnote 54 evidently an allusion to the story of Vertumnus, who ‘turned the river’ back when it flowed along the line of the Vicus Tuscus to the Forum.Footnote 55
In the final reference a much more complex story is involved, Livy's account of the trial of Marcus Manlius in 384 BC. Manlius was the hero who had foiled the Gauls’ attempt to scale the Capitol six years before. Now he was on trial for seeking royal power (regnum) (6.20.10–12, trans. Betty Radice, Penguin Classics, slightly adjusted):Footnote 56
As the People were being called by centuries in the Campus Martius and the accused by stretching out his hands to the Capitol had directed his prayers from men to the gods, it became clear to the tribunes that unless men could have their eyes diverted from that reminder of so glorious a deed, they would remain preoccupied with the service done them and never open their minds to the reality of the charge. They therefore adjourned the day of trial and summoned an assembly of the People in the lucus Petelinus outside the Porta Flumentana, from where the Capitol could not be seen. There the charge was proved, and the People steeled their hearts to pass sentence of such severity that it was painful even to those who pronounced it … The tribunes threw him from the Tarpeian Rock: so the same place commemorated one man's greatest hour of glory and the supreme penalty he paid.
The site of the lucus Petelinus is unknown,Footnote 57 but if it was outside the river gate it cannot have been out of sight of the Capitol.Footnote 58 It seems extraordinary that Livy should have made such an error. Stephen Oakley's careful analysis (Reference Oakley1997: 476–93) of the seditio Manliana episode, and of the competing and mutually incompatible versions from which Livy tried vainly to create a coherent narrative,Footnote 59 provides the only plausible explanation.
Where exactly did Manlius defend the Capitol against the Gauls? One version of his exploit (Livy 5.47.2–3: see Section 3) placed it at the top of the cliff above the shrine of Carmentis, because the Gauls had found Pontius Cominius’ tracks and went up the same way he had. But Livy (6.17.4) also knew a rival version, that they went up the Tarpeian Rock,Footnote 60 and that is the version presupposed here: ‘the same place commemorated one man's greatest hour of glory and the supreme penalty he paid’. The cliff above Carmentis was at the southern end of the Capitol, overlooking the river; the Tarpeian Rock was on the east side, overlooking the Forum,Footnote 61 not visible from the riverbank.
That completes the dossier of information about the Porta Flumentana. It clearly confirms Coarelli's hypothesis of a wall parallel to the river. But the stories studied in Section 3 presuppose a quite different layout, with ‘cross-walls’ from the Capitol and the Aventine and the space between protected by the river alone. The problem now is to see how options 1 and 3 (Fig. 1) can both be true.
5. CHANGING THE LAYOUT
Since all situations change over time, a likely answer is that options 1 and 3 are both valid, but in sequence. I propose the following hypothesis: that the original circuit left the riverbank unwalled, as presupposed by the Horatius Cocles and Pontius Cominius stories (Section 3), and that at some point in the third century BC a new wall was constructed parallel to the Tiber, with a ‘river-gate’ in it (Section 4) to combine access with security.
It is a necessary precondition of option 1 that ‘there was only one bridge, and they removed it in time of war’ (Dion. Hal., Ant. Rom. 9.68.2). Equally, the existence of a permanent stone bridge presupposes option 3. The building of the Pons Aemilius would therefore provide a terminus ante quem for a new defensive wall parallel to the Tiber, but unfortunately the date of its construction is disputed.Footnote 62
A critical moment in the Hannibalic war provides the most likely context, and an urgent motivation, for the new defensive layout. After the catastrophe at Cannae and the defection of their allies, the Romans knew they might have Hannibal at their gates at any time.Footnote 63 Even four years later, in 212 BC, the rural population was still crowded inside the walls for protection.Footnote 64 The previous year there had been a disaster within the city itself (Livy 24.47.15–16, trans. A. de Selincourt, Penguin Classics):Footnote 65
In Rome there was a terrible fire which raged for two nights and a day: everything between the Salinae and the Porta Carmentalis was burnt to the ground, including the Aequimaelium, the Vicus Iugarius and the temples of Fortune and Mater Matuta. The fire also spread to a great distance beyond the gate and destroyed many houses and sacred buildings.
Now remedial action was taken (Livy 25.7.5–6, trans. A. de Selincourt, Penguin Classics):Footnote 66
In accordance with a decision of the Senate and the expressed will of the People, an assembly was now called by the City praetor for the appointment of five commissioners for the repair of the walls and defence-towers, and of two other commissions, each of three members, one for examining the sacred vessels and making a record of temple gifts, the other for repairing the temples of Fortune and Mater Matuta inside the Porta Carmentalis, and also of the temple of Hope outside the gate, all of which had been destroyed by fire in the previous year.
Fortunately, all the buildings and places mentioned in these two passages can be confidently identified (Fig. 3).
Fig. 3. The fire of 213 BC: 1. Salinae; 2. Porta Carmentalis; 3. Aequimaelium; 4. Vicus Iugarius; 5. temples of Fortuna and Mater Matuta; 6. temple of Spes (Hope).
Since they are listed in a south-to-north sequence, from the Salinae at the foot of the Aventine to the Porta Carmentalis at the foot of the Capitol,Footnote 67 the distinction between the temples inside and outside the gate is more natural if the Porta Carmentalis was in a ‘cross-wall’ from the Capitol to the river (option 1) than if it was part of a wall parallel to the river from the Aventine to the Capitol (option 3). It might be argued that the walls and towers to be repaired were precisely that riverside stretch, damaged in the fire; but given the circumstances of the time, it seems unlikely that the commissioners’ remit was so limited.Footnote 68
Their job was probably to strengthen the entire circuit: who knew which part of it might come under attack? Nevertheless, the extensive area destroyed by the fire offered them a particular opportunity for radical restructuring, if they thought the riverside defences required it. So it was probably they who built a new wall parallel to the Tiber, and a new gate, the Porta Flumentana.
Livy's story of the trial of Manlius is not necessarily an obstacle to this idea. The details of his narrative are not, of course, authentic evidence from the fourth century BC. On the contrary, Livy's sources here were evidently very late: ‘even in a tradition replete with motifs from post-Gracchan politics, the story of the seditio Manliana stands out for the extent to which it is told in late Republican, and particularly in Catilinarian, terms’ (Oakley, Reference Oakley1997: 481, with details at 481–4). It is perfectly possible that some late-Republican author, used by Livy, placed the lucus Petelinus ‘outside the Porta Flumentana’ without realizing that the gate was only about 150 years old. By then it may have been assumed that the revised third-century layout was the original.
Yet again, a famous story of early Rome may provide significant information. In 479 BC the Fabian gens volunteered to man a permanent fort protecting Rome's frontier with Veii; two years later the fort was taken and the Fabii wiped out (Livy 2.48–50; Dion. Hal., Ant. Rom. 9.15–22; Richardson, Reference Richardson2012: 81–3, 140–2). The event itself may well be historical,Footnote 69 but the elaborate treatment of it in Livy and Dionysius is all later invention. One particular detail affects the present argument (Livy 2.49.7–8):Footnote 70
As they passed the Capitol and the citadel and the other temples, the crowd prayed to whichever of the gods was in their sight or their thoughts to send that marching column out with good omens and soon bring them safely home. But the prayers were in vain. They left by an ill-omened route, the right-hand arch of the Porta Carmentalis, and proceeded to the river Cremera.
Other sources (Festus 450–1L; De vir. ill. 14.4) say that the arch of the Porta Carmentalis through which they went was named ‘Wicked Gate’, porta scelerata. Ovid (Fast. 2.201–4) confirms the tradition:Footnote 71
The nearest way is by the right-hand arch of the gate of Carmentis. Don't go through it, whoever you are; it carries an omen. Tradition tells that the three hundred Fabii went out by that way. The gate is blameless, but yet it carries the omen.
The story is self-evidently anachronistic: in a city wall, a gate with two arches side by side makes no defensive sense. The point of it must be aetiological, explaining an arch at the Porta Carmentalis that nobody now used, perhaps walled up like the ‘Porta Chiusa’ in the wall of Aurelian (Nash, Reference Nash1968: 2.208–9). But if so, why and when?
Reorganization of the riverside defences provides a plausible context. Ex hypothesi (option 1), the original layout was a ‘cross-wall’ from the southernmost point of the Capitol to the Tiber. From the Vicus Iugarius a right turn — ‘the nearest way’, as Ovid says — would have brought the Fabii to the gate immediately under the Capitol cliff (Fig. 4a). Also ex hypothesi (option 3), a new wall was subsequently built at 90 degrees to the old, running parallel to the river from this ‘cross-wall’ to the equivalent one at the Aventine. If the junction of the two walls was immediately west of the old gate, we might imagine a more secure layout involving two successive gates, one in the new wall and the other in the old (Fig. 4b). And if the original gate was walled up with masonry but still identifiable, in due course a story would be needed to account for why it was no longer used.
Fig. 4. Hypothetical layouts for the Porta Carmentalis.
Where exactly did the walls join? Immediately south and east of the site of the Porta Carmentalis was the great square platform at S. Omobono that supported the twin temples of Fortuna and Mater Matuta. Constructed probably in the early fifth century BC,Footnote 72 it was a dominant feature standing 5 m high from ground level.Footnote 73 One might expect the commissioners of 212 BC, working urgently at a time of crisis, to have incorporated this ready-made barrier into their new defensive system, especially as it was simultaneously under repair after the fire.Footnote 74 (Coarelli's different line for the wall (Fig. 2), which has been very influential,Footnote 75 does not affect the argument here, because he takes it to be part of the original circuit, supposedly of the sixth century BC and therefore already in existence before the platform was constructed. Also irrelevant is his identification of the Porta Carmentalis as the ‘triumphal gate’, another influential notion, which is discussed in the Appendix below.)
If the platform was indeed part of the new defences, then the surviving archaeological record, however fragmentary (Fig. 5), is compatible with just the sort of layout we have imagined. People going down the Vicus Iugarius towards the river harbour, with the back of the twin temples on their left, now had a gate in front of them at the northwest corner of the platform; to their right was the ‘cross-wall’ from the Capitol cliff to the river, with the old gate blocked off but still identifiable (the ‘right arch’ of the Porta Carmentalis, no longer used); once through the gate in front of them (the ‘left arch’), a right turn would bring them to a new gate in the ‘cross-wall’. The hypothesis cannot be proved, but it does at least account for all the evidence.
Fig. 5. ‘Area sacra di S. Omobono e pendici meridionali del Campidoglio’ (Coarelli, Reference Coarelli1988: 454, fig. 112, detail, reproduced by permission). Item 10 is an archaic wall in cappellaccio: on the hypothesis presented in Figure 4, it could be part of an original Porta Carmentalis later blocked off and replaced by item 7, the late-Republican Porta Carmentalis identified by the surviving portico (item 8: see Coarelli, Reference Coarelli1988: 394–6).
6. PORTA TRIGEMINA, PORTA MINUCIA
At the other end of the wall parallel to the river was the Porta Trigemina, the approximate site of which is well attested.Footnote 76 Frontinus’ account of the Aqua Appia (Aq. 5.5, 5.9) supplies the essential information:Footnote 77
The channel [of the aqueduct] has a length of 11,190 passus from its origin as far as Salinae, which is at the Porta Trigemina … The distribution begins at the bottom of the Clivus Publicius at the Porta Trigemina, at the place called Salinae.
The Clivus Publicius was the ascent from the Forum Boarium to the northern corner of the Aventine.Footnote 78 Poggio Bracciolini refers to an arch, still visible in 1430 ‘next to the Tiber beyond the schola Graeca’ (i.e. south of S. Maria in Cosmedin), which was very probably the Augustan Porta Trigemina.Footnote 79 The question is, was the gate in a ‘cross-wall’ (option 1) or in a wall parallel to the Tiber (option 3)?
Coarelli (Reference Coarelli1988: 31–4; Reference Coarelli2019: 138–42) provides a powerful argument for the latter solution, based on the story of the flight and death of Gaius Gracchus in 121 BC. Gracchus and his supporters had occupied the temple of Diana on the Aventine (Plut., Vit. C. Gracch. 16.4; App., B Civ. 1.26.114–15; Oros. 5.12.6); when the consul's forces attacked from the Clivus Publicius, they withdrew first to the nearby temple of Minerva and then to that of Luna (Oros. 5.12.7–8; De vir. ill. 65.5); driven out again, Gracchus and a few friends fled down the Clivus Publicius to the Porta Trigemina, where one of his companions tried to hold off the pursuers; another did the same at the Pons Sublicius,Footnote 80 over which Gracchus fled to the grove of Furrina, and there killed himself.Footnote 81 That route, down the Clivus and through the gate to the bridge (Fig. 6), seems to be proof of option 3;Footnote 82 it would not be possible if the gate were in a ‘cross-wall’.
Fig. 6. ‘Itinerario della fuga di Gaio Gracco’ (Coarelli, Reference Coarelli2019: 141, fig. 69, reproduced by permission). Items 1, 2 and 3 are the approximate sites of the temples of Diana, Minerva and Luna; items 9 and 10 are the Porta Trigemina and the Pons Sublicius.
But there is also evidence that points the other way. The Augustan arch seen by Poggio in 1430 was ‘over the road between mount Aventine and the bank of the Tiber’.Footnote 83 The narrow strip of land between the steep Aventine slope and the river was progressively developed as Rome's commercial port, and the early stages of that process are reported by Livy with the unvarying description extra portam Trigeminam.Footnote 84 That is natural and inevitable if the gate was in a ‘cross-wall’ (option 1), but awkward if not (option 3), as the phrase would have to mean ‘outside the Porta Trigemina and turn left’. However, Coarelli has now brought another argument to bear.Footnote 85
One of the more conspicuous episodes of Livy's fourth book (4.12.6–16.4) is the story of the famine of 440 BC and its consequences.Footnote 86 Lucius Minucius was elected praefectus annonae,Footnote 87 but was unable to improve the situation; a wealthy equestrian called Spurius Maelius then sourced corn supplies on his own initiative, and used his popularity to aim at tyranny; Minucius informed on him to the Senate, and a dictator was appointed (Cincinnatus) at whose order Maelius was killed. Minucius was then rewarded with a gilded ox ‘outside the Porta Trigemina’.Footnote 88 In Dionysius’ version (Ant. Rom. 12.4.6) his reward was a statue;Footnote 89 the elder Pliny specifies that it was a statue on a column, and that it stood outside the Porta Trigemina.Footnote 90 Coarelli draws particular attention to the gilded ox, which he identifies as the bronze bull that gave the Forum Boarium its name.Footnote 91
If you believe that the Porta Trigemina was in a wall parallel to the Tiber, it is natural to suppose that ‘outside the Porta Trigemina’ and ‘in the Forum Boarium’ meant the same thing.Footnote 92 But that is not the case. The two toponyms never coincide in the ancient sources, and we have explicit evidence that they were topographically distinct: Macrobius (Sat. 3.6.10) tells us that there were two temples of Hercules Victor in Rome, ‘one at the Porta Trigemina, the other in the Forum Boarium’.Footnote 93 And in any case, as Livy's own usage makes clear, the gilded ox given to Minucius was not a bronze statue, or a monument of any kind, but a living beast with its horns gilded for sacrifice.Footnote 94 If the Senate granted it to Minucius in the version of the story followed by Livy, it must have been so that he could sacrifice it to some deity.
In that context, two items from Festus, surviving only in Paulus’ abridged version, are clearly significant (Paul. Fest. 109L, 131L):Footnote 95
The Porta Minucia at Rome was called after the altar of Minucius, who they thought was a god.
The Porta Minucia was so called for this reason, that it was next to the shrine of Minucius.
This altar and shrine are generally, and convincingly, identified with the column monument illustrated on the coins produced about 135/134 BC by C. (Minucius) Aug(urinus) and Ti. Minucius C.f. Augurinus.Footnote 96 It was probably an archaic family cult centre with the divine Minucius portrayed on the column, reinterpreted in the later historiographical tradition as an honorific statue of Lucius Minucius the supposed praefectus annonae. Coarelli argues that the prefecture is historically authentic, attested on the monument itself.Footnote 97 But that is not at all what Livy (4.16.3–4) says:Footnote 98
I find it stated in some authors that this Minucius passed from the patricians to the plebs, was co-opted as an eleventh tribune, and calmed a riot that had broken out as a result of the killing of Maelius … [Livy gives some reasons for finding this incredible.] … But what above all proves the inscription of the imago false is the fact that a few years earlier a law had been passed forbidding the tribunes to co-opt a colleague.
It is clear that the authorities rejected by Livy had cited the inscription attached to Minucius’ portrait in the family atrium; that is what imago always means in Livy,Footnote 99 and he knew very well that their inscriptions (tituli) were not to be relied on.Footnote 100
Whatever the historicity of the archaic annona, the story of Minucius is topographically important as the only evidence for the existence of the Porta Minucia. Since it was close to the monument that gave it its name, and that monument was situated ‘outside the Porta Trigemina’ (see nn. 88, 95), Mario Torelli's view (Reference Torelli and Steinby1993: 306) that the Minucia was an archaic gate replaced by the Trigemina is very attractive.Footnote 101 But rather than accept his suggested date, ‘after the burning of Rome by the Gauls’, it would surely be preferable to associate the change with the attested reconstruction of walls and towers in 212 BC.
If the Porta Minucia was the original gate in a ‘cross-wall’ from the Aventine to the river (option 1), it could have been replaced by the Porta Trigemina when a new wall was built parallel to the Tiber. But in that case we have to ask why a new gate was needed, and which way it faced. The argument from Gaius Gracchus’ flight to the bridge (option 3) remains as powerful as ever.
7. THE TRIPLE GATE
To find an answer to those questions, we must first ask another. Why was the Porta Trigemina so named? The natural assumption is that it had three arches:Footnote 102
The name is best explained by supposing that the gate had three openings, to accommodate the heavy traffic of this district.
It is probably not necessary to posit a meeting of three streets here to account for the name. More likely it had a central passage for wheeled traffic flanked by side passages for pedestrians.
But that is highly improbable. Whether the gate was created in the fourth (or sixth) century BC as part of the original circuit wall, or in the late third century as part of a reconstruction under the threat of Hannibal, it was designed for security, not convenience of traffic. Säflund, in his classic study of the Republican walls, was well aware of that, but his alternative proposal, that the name alluded to the three heads of the monster Cacus, who supposedly lived nearby, is quite absurd.Footnote 103
One piece of information is decisive. Gaius Gracchus twisted his ankle jumping down from the temple of Luna to escape the consul's forces (Fig. 6), so it was essential for his companions to hold back the pursuit for as long as they could. Pomponius did so at the Porta Trigemina, Laetorius at the Pons Sublicius; inevitably, both were killed (Val. Max. 4.7.2; De vir. ill. 65.5). Whether it really happened quite so dramatically is impossible to say, but the story must at least make sense, and Pomponius’ solo rearguard action would obviously have been impossible if the Porta Trigemina had had three openings side by side. And if that explanation is unavailable, the ‘triple gate’ must have been named for some other reason.
Section 5 proposed the hypothesis that the original circuit left the riverbank unwalled and that a new wall, parallel to the Tiber from the Aventine to the Capitol, was built in 212 BC. Trying to imagine how the new layout was managed, especially the junction of the new wall with the existing ‘cross-walls’ to the river, helped to explain a puzzle about the supposed two arches of the Porta Carmentalis. I suggest it can do the same for the three arches of the Porta Trigemina.
A useful starting-point is Giuseppe Lugli's passing comment that ‘the origin of the name seems to refer to three passages, either next to each other or in succession’.Footnote 104 The latter alternative does at least recognize that the city walls were up to 4 m thick, and that a city gate normally consists of two barriers, outer and inner, with a space between them. Lugli offered no explanation for a third barrier, but an explanation is available if we assume that the Porta Trigemina was a new gate constructed exactly where the hypothesized wall of 212 BC joined the old ‘cross-wall’. In that case, the ‘triple gate’ could be three gates in three walls, with a triangular space between them (Fig. 7).
Fig. 7. Hypothetical layout for the Porta Trigemina.
For Gracchus, limping down the Clivus Publicius with his two companions, the Porta Trigemina would present a single arch. Through that arch, he would be faced with a choice: go left through the second arch, outside the walls to the commercial port area that was called extra portam Trigeminam, or go right through the third arch, inside the old ‘cross-wall’ to the Pons Sublicius. He went right, leaving Pomponius to hold the pursuers as long as he could. I know of no parallels for a city gate of this type,Footnote 105 and it is of course open to anyone to dispute the suggestion on a priori grounds; but the name of the gate requires explanation, and the hypothesis of a wall-junction provides it.
8. CONCLUSION
This long and complex investigation has detected the following stages in the early history of Rome's defences:
(1) For the first three centuries of the city's existence,Footnote 106 there were probably only walls round individual hills (see n. 12).Footnote 107 Although it is formally possible that the full circuit wall was built during the regal period, the obvious special pleading required for the stories of Appius Herdonius and the Gallic sack (gates left open for implausible reasons) makes it very unlikely.
(2) The city wall was built probably in the 370s BC, with ‘cross-walls’ to the Tiber from the Capitol and the Aventine and no wall parallel to the riverbank (Fig. 8a). The stories of Horatius Cocles and Pontius Cominius presuppose this layout. The gate in the northern ‘cross-wall’ was the Porta Carmentalis, that in the southern probably the Porta Minucia.
(3) The wall parallel to the Tiber was built probably in 212 BC, with a new gate, the Porta Flumentana, at the centre of it (Fig. 8b). The story of the trial of Marcus Manlius presupposes this layout. The old ‘cross-walls’ remained, and the Porta Carmentalis (restructured) and the Porta Trigemina (replacing the Porta Minucia) were sited where the new wall met the old. The story of the departure of the Fabii presupposes this layout, as does the historical narrative of the flight of Gaius Gracchus in 121 BC.
These findings are of more than merely antiquarian interest, for two quite separate reasons. Firstly: because the city's walls were sacred and symbolic, to understand their history is to understand something fundamental about Rome itself. And secondly: if, with properly careful attention, quasi-historical stories in authors writing centuries after the events can be made to yield useful historical inferences, the perennial question of how we can ever know the history of early Rome becomes more than just a crudely schematic dispute between sceptics and believers.
Fig. 8. The proposed sequence of walls and gates: 1. from 378 BC (Livy 6.32.1); 2. from 212 BC (Livy 25.7.5).
APPENDIX: ‘THE PORTA TRIUMPHALIS’
For half a century, following a brilliantly conjectural article by Coarelli (Reference Coarelli1968), it has been generally believed that the Porta Carmentalis was the ‘triumphal gate’, the ritual point of entry for triumphal processions.Footnote 108 But the evidence for the supposed Porta Triumphalis was always controversial,Footnote 109 and after a sceptical deconstruction of the orthodoxy by Mary Beard, subsequent accounts became rather more cautious.Footnote 110
Now, however, Tonio Hölscher (Reference Hölscher2018: 66, 75) has endorsed Coarelli's hypothesis, with a confusing new twist:
There was a double exit from the city to the west: the Porta Carmentalis leading through the city wall, the Porta Triumphalis crossing the pomerium … The city wall ran parallel to the Tiber, traversed by the Porta Carmentalis. The sacred boundary of the pomerium must have run somewhat inside, with the Porta Triumphalis as its ritual passageway.
Despite its importance for Hölscher's argument,Footnote 111 this is demonstrably wrong: the pomerium was outside the walls, in the area augurally defined as ager effatus.Footnote 112 But in any case, the notion of two successive gates on the same street — one in the wall, the other to mark the pomerium — depends entirely on the unattested concept of a ‘ritual passageway’.
It is important to remember that the association of the Porta Carmentalis with triumphs, and indeed the whole idea of a fixed ritual route for the triumphal procession,Footnote 113 is a modern assumption with no support in the ancient sources. In the whole of ancient literature there are precisely five references to a ‘triumphal gate’, and only one of them is from the Republican period.
In his speech in the Senate attacking Lucius Piso in 55 BC, Cicero (Pis. 55) refers rhetorically to a porta triumphalis ‘that has always stood open for previous consuls returning from Macedonia’, in derisive contrast with Piso's inglorious entry at the Porta Esquilina.Footnote 114 Subsequently, Tacitus, Suetonius and Dio all report the Senate's decree in AD 14 that Augustus’ funeral procession should go by the (or a) porta triumphalis; but by this time, thanks to the pax Augusta, the city gates seem to have been transformed into ornamental arches anyway.Footnote 115 Finally, Josephus’ description of the triumph of Vespasian and Titus in AD 71 specifically mentions ‘the gate that took its name from the fact that triumphs always pass through it’;Footnote 116 but since Vespasian went back to it after greeting the Senate at the Porticus Octavia, it cannot have been anywhere near the Porta Carmentalis.Footnote 117
What exactly was the porta triumphalis referred to by Cicero (Pis. 55, trans. Beard, Reference Beard2007: 96)? When he said quasi … ad rem pertineat qua tu porta introieris, modo ne triumphali, did he mean ‘as if it mattered which gate you entered by, so long as it wasn't the triumphal one’, or ‘so long as it wasn't a triumphal one’? Since ‘the’ Porta Triumphalis is never referred to before 55 BC, it is worth entertaining the possibility that a prospective triumphator could make his own choice where he would enter the city, and therefore which gate would be decorated for the occasion and thus become ‘triumphal’ (Wiseman, Reference Wiseman2008a: 391).
The question must be left open. Coarelli's theory has to be mentioned for the sake of completeness, but it is not usable evidence for historical topography, and it offers no help for understanding the development of the riverside defences.
