TAPA

the evidence

We must start from one central fact: "ancient authors provide no hard figures for voter turnout or go beyond even the vaguest indication of scale."¹¹ The [End Page 129] evidence we do have is for individual elections and is strongly impressionistic, and hence not particularly useful in discovering how many Roman citizens normally attended. I show below that it cannot be used to argue for low turnout, either regularly or even in the isolated cases for which it has been so used (e.g., in 55 and 45). By the same token, we do have evidence which suggests large crowds at individual elections (e.g., Plut. C. Gracch. 3.1; Cic. Verr. 1.54), and occasions when either all centuries voted (Livy 22.35, 37.47), or almost all (Cic. Brut. 237; Asc. 94C). But these may have been exceptions. The evidence, as a whole, cannot be used to prove (or disprove) either high or low turnout as the normal state of affairs.

Since MacMullen and Mouritsen argue for low turnout largely based on the size of the Saepta, let us consider the evidence. The Saepta, the voting enclosure for elections on the Campus Martius, was a structure approximately 310m long and 120m wide, very probably on the site of the existing Ovile; its northern end lay immediately east of the Pantheon.¹² It was planned by Caesar but only completed by Agrippa in 26 (Cic. Att. 4.16.8; Cass. Dio 53.23.2). Voters entered at the northern end, lined up in their tribal pens (from which the joking name ovile or "sheepfold" was derived), and voted at the southern end, after which they departed. Votes were counted in the Diribitorium which adjoined at the southern end. Macmullen argues that the actual voting area could hold at most 55,000 people, which seems reasonable.¹³ But that does not mean that only 55,000 people could vote in an election held there. As Phillips counters, in a centuriate assembly the Classes voted successively, so the 55,000 figure may only apply to the maximum number of voters in any one class.¹⁴ But even this assumes that all voters within a Class would need to be contained within the structure at the same time, which is not the case.

Mouritsen, however, believes the maximum number was much lower, as there would need to be a large forecourt at the northern end to hold those voters waiting to vote. He thinks the segregation of voters was essential to prevent fraud: that is, to stop "those who had already voted from rejoining those still waiting."¹⁵ But there is no reason to think so. It might perhaps have been necessary to prevent voters from voting twice in their own Class (although even this seems doubtful and could have been achieved by simply directing voters to leave the area after voting). But Rome entirely lacked the [End Page 130] bureaucratic machinery needed for voter identification; what machinery there was relied on declarations by citizens, which the authorities were rarely in a position to dispute.¹⁶ Moreover, the Classes voted sequentially, from richest to poorest, and it seems very unlikely that any Roman would humiliate himself by voting with those beneath him, simply to add his single vote to the hundreds (if not thousands) being cast at the same time. He was especially unlikely to do this in front of people who knew him. It seems similarly unlikely that, for example, voters of the Fourth Class waited in a forecourt, often in midsummer (which is when elections were normally held after Sulla), for the several hours it might take for them to be called on. There is thus no reason to think the size of the voting enclosure was a meaningful restriction on the number of people who voted, either in centuriate or tribal elections. Rather, its great size points to the need to direct and funnel large numbers in a complex procedure over several hours; it is best understood as the response to a problem of movements in space.

There is some literary evidence which has been summoned to argue for low turnout. Mouritsen points to two Ciceronian passages in particular, but these do not have the force he ascribes to them.¹⁷ In Pro Plancio 54, Cicero mentions to Laterensis that, in gathering evidence for an ambitus prosecution, he had more witnesses in the Voltinian tribe than he received votes there. Even if we do not read this as straightforward rhetorical exaggeration, it does not actually tell us anything. If, for instance, Laterensis received five votes in this tribe, does that suggest twenty people voted? Or one hundred? Or five thousand? There must have been candidates who performed poorly, especially in particular tribes where they (and their friends) had little influence, or where their competitors had great influence. Cicero's testimony about the election of a suffect consul in 45 is more useful (Fam. 7.30.1):¹⁸

In campo certe non fuisti cum hora secunda comitiis quaestori<i>s institutis sella Q. Maximi, quem illi consulem esse dicebant, posita esset; quo mortuo nuntiato sella sublata est. ille autem, qui comitiis tributis esset auspicatus, centuriata habuit, consulem hora septima renuntiavit, qui usque ad Kal. Ian. esset quae erant futurae mane postridie.

At least you were not in the Campus when the elections to the Quaestorship began at nine o'clock in the morning. A chair of state had been placed for Q. Maximus, whom these people used to call Consul. His death was announced, [End Page 131] and the chair removed. Whereupon he, having taken auspices for an assembly of the Tribes, held an assembly of the Centuries, and at one o'clock of the afternoon declared a Consul elected, to remain in office until the Kalends of January, the next morning that was to be.

Nicolet calculates, on the basis of two voters per minute over the five hours Cicero mentions, a voting crowd of 16,800, all from the first two Classes.¹⁹ Mouritsen, however, notes problems with this calculation: due to the short mid-winter days, Cicero's five hours were really only three hours and forty-five minutes, while the reorganization into centuries of a crowd originally convening for a tribal assembly would have taken additional time.²⁰ So Mouritsen reasons a quarter of the time was lost and, with the uneven size of the voting tribes, calculates 6,000 voters in the first two Classes. This is perhaps an underestimate but, far from pointing to low turnout, actually suggests numbers were normally large. The circumstances of this election were, after all, highly unusual. It was originally meant to be a quaestorian election in the tribal assembly (i.e., not organized by Classes), so presumably upper-Class voters would not have turned up in disproportionately large numbers. So the 6,000 voters of the first two Classes might have been, on a conservative estimate, one-third of all who attended. If we can estimate that 18,000 voters showed up for a quaestorian election, at the end of December (when it was probably very cold) under Caesar's dictatorship, that suggests the turnout for a consular election, in better weather, and in the days of the libera res publica would have been much higher.

We also need to consider the Commentariolum Petitionis and what it suggests about the scale and nature of elections.²¹ Following most recent scholars, I accept it as genuine, or at least as giving an authentic and well-informed picture of electoral campaigns in the 60s.²² The work gives no indication of voter turnout but does suggest a very complex electoral landscape. Moreover, while in this article I am presenting the electoral system from the propertied rural voter's perspective, the Commentariolum Petitionis shows it from the candidate's perspective. Strength in both friends and public reputation was essential for victory (section 16), although Tatum suggests cogently this is a conventional distinction (compare Cic. Att. 1.1.2) because it is in practice [End Page 132] muddied in the rest of the treatise.²³ Overall the treatise suggests candidates saw only radical uncertainty: the electorate was too large, complex, and unpredictable to manage. The treatise definitely does not suggest, in MacMullen's words, that "politics was ordinarily a cozy business."²⁴ More relevant to this article, the Commentariolum Petitionis indicates candidates needed to win over country voters, and the way to do this was through men influential in the municipia (24, 30–31).²⁵ The political scene was littered with groups and organizations with electoral importance, such as the four sodalities Cicero had won to his side by defending their members (19); these groups were evidently important structures for channeling votes. Beyond this, there were also "men eager for advancement [who] have worked very hard, with unreserved eagerness and energy, to be able to get what they want from their fellow tribesmen" (18); this was clearly a development which was both recent and not (yet) comprehensive. Still, it points to the central importance of the tribes as electoral units. Moreover, a candidate needed to create a broad coalition of relatively small, independent pockets of support in many different tribes to win.²⁶

voting procedure and the census classes

Consuls and praetors were elected in the centuriate assembly; curule aediles, quaestors and minor magistrates in the tribal assembly. In both cases the thirty-five tribes were the basis of the voting units.²⁷ In the tribal assembly there were only the thirty-five tribal voting units. The centuriate assembly was much more complex, as voters were split horizontally (by census Class) as well as vertically (by tribe).²⁸ This meant the tribe was the key unit: to win the election, candidates had to win whole tribes, not just groups of voters within [End Page 133] them. So far, the Roman system is intelligible to voters in a parliamentary system, if we think of the candidates as equivalent to modern parties and the voting units as constituencies.²⁹ The order in which the Classes were called to vote was not strictly hierarchical; equites voted perhaps with, or more probably after, the First Class.

The procedure for voting is also reasonably well known, although it is complex.³⁰ For the sake of clarity, let us imagine a hypothetical consular election in which there are two consuls to be elected. The same principles and methods applied to praetorian elections (with six or, after Sulla, eight vacancies). Each voter could vote for two candidates (six or eight in a praetorian election); this is the principle of the multiple vote which is of central importance in this article.³¹ Within each voting unit, the two candidates who attracted the most votes were that unit's choice, without weighting. That is, in a hypothetical century of 100 voters (= 200 votes) who gave 100 votes to Cicero, 51 to Antonius and 49 to Catiline, the century's votes were simply for Cicero and Antonius.³² Once a candidate received the votes of a majority of the centuries (97 out of 193), he was declared elected. When two candidates reached this point and thus both places were filled, the election was at an end, and no further Classes were called to vote.

With only the Second Class being needed for a majority, it is uncertain how often the lower Classes would be required to vote. This is important in calculating the balance of incentives for such voters to make the journey and for others to help them to do so. We do know of elections in which one candidate won by only a few centuries and thus in which the lower Classes certainly voted. In 64 Antonius pauculis centuriis Catilinam superavit (Asc. 94C), while (possibly) the previous year L. Turius lost by only a few centuries (Cic. Brut. 237) and Q. Metellus Macedonicus won only narrowly in 143, after [End Page 134] two repulsae (De vir. ill. 61.3). There are also examples in Livy of not all the places being filled in one day and a second round of voting being required (22.35, 37.47); in such cases all Classes must have voted. But we face here the same problem as elsewhere: were such elections the rule or the exception? Arguments based on the time voting must have taken do not help us much, because we have little detailed knowledge of the Roman procedure (which was on such a larger scale than that envisaged by the Flavian municipal laws that the evidence of the latter is of little help).³³ We might interpret the move of elections to July (tentatively dated to Sulla) as a result of the need for the fourteen hours of daylight Rome offers at that time of year. Yakobson argues, with suitable diffidence in view of the lack of evidence, that the lower Classes probably were often called to vote.³⁴ Elections were, after all, contests between elite candidates, with resources of similar kind and magnitude; bribery was targeted to win whole centuries. These factors were likely to split the votes of centuries in the upper census Classes, meaning the lower Classes would be decisive.³⁵ Of course, the Third Class was more likely to vote than the Fourth, and the Fourth more likely than the Fifth. And, mathematically, praetorian elections were more likely to split the vote, with more candidates to be elected and without a centuria praerogativa to guide the voters. So, since we are concerned here with the balance of incentives for rural voters to travel to Rome, the confidence that the journey would not be wasted was stronger for those in the Third Class than in the Fifth. But it seems to me likely that a Fourth Class voter, whatever the strength of other incentives, could be reasonably confident of voting in at least one of the consular or praetorian elections. Hence, for the purposes of this article, I suggest the active centuriate electorate comprised the first four census Classes. This gives us something to work with, as long as we remember it is approximate: we are here in the realm of a balance of incentives and disincentives. [End Page 135]

Yet such considerations only applied in the centuriate assembly. In the tribal assembly, citizens voted by tribe only, so census Class did not matter. And it seems that, normally, elections for curule aedile, quaestor, and the minor magistracies in the tribal assembly were held in the days immediately after elections for consul and praetor in the centuriate assembly. Therefore, voters in rural tribes from the Third or Fourth Classes knew they would be called to vote in tribal elections and would quite likely (though not certainly) be called to vote in at least one of the consular or praetorian elections. A journey to Rome to vote would not be wasted.

When we speak of men of the Fourth Class as against the First, what type of socio-economic distinction do we mean? And how many of each type of voter were there? This steers us into the question of Roman demography, a contentious topic.³⁶ Clearly, the Social War and the enfranchisement of Italy deeply altered the Roman electorate: it became much larger and probably poorer on average, while the tribal map of Italy was massively overhauled. But the qualifications for each Class do not seem to have changed within our period. For the First Class, the amount was property worth either HS40,000 or HS25,000; I will use the larger figure here as the more conservative for my purposes.³⁷ The figures for the Second, Third, and Fourth Classes were, respectively, 75%, 50%, and 25% of the First Class qualification. What size farm do such amounts describe, understanding that the census rating included all property, including farm buildings, equipment, livestock, and slaves? Rathbone suggests the First Class census represented (notionally) 100 iugera of land (about 25 ha), while Rees (following Duncan-Jones) estimates an average land value of around HS475–660 per iugerum.³⁸ However, as Rees notes himself, this might be too low. We should expect land that was fertile, close to Rome, or close to the sea, navigable rivers or major roads to be more expensive. This description covers a great deal of the ager Romanus before the Social War. Rees estimates that a farmer with "50 iugera, a house and possibly an ox would have more than HS25,000 at the lowest price estimate."³⁹ Adapting this to a HS40,000 qualification and a higher cost of land, we can say that an 80-iugera farm, with accompanying equipment, would place its owner comfortably in the First Class. A farm of 20 iugera would be a Fourth Class rating. The income from such farms might be estimated in the hundreds to low thousands of sesterces per year; this should be remembered when we come to consider [End Page 136] the incentives to go to Rome for the elections. In any case, to a senator with estates worth a million sesterces, a First Class voter worth HS50,000 and a Third Class voter worth HS25,000 would have seemed equally poor.

The significant improvement in survey archaeology in recent decades has caused scholars to rethink the economy of ancient Italy.⁴⁰ While there is significant disagreement, the overall picture for the last two centuries b.c.e. is one of (modest) economic growth, diversity of agriculture driven by urban markets (Rome especially, but also smaller regional cities), and the empire relieving Italy of the need to concentrate on grain. There was some growth in villas, but probably not at the expense of small farms.⁴¹ Small farms could still be owned by large landowners and run by tenants; these would not appear as voters and cannot be shown clearly in the archaeological record.⁴² Moreover, the presence of small farmhouses with ornate mosaic floors does not by itself suggest ownership rather than tenancy.⁴³ While (as noted above) Rees argues for land values in the range HS475–660 per iugerum, this would only be true for arable land. But Italy was clearly a highly diverse agricultural economy, which included the raising of cash crops (and livestock such as game) for the market. Such agricultural practices could generate large incomes (and correspondingly high land values) on small plots of land. Kron points to the Villa Regina at Boscoreale, where a vineyard of 7–10 iugera, able to be worked by a single family, could possibly generate revenues of HS7,000.⁴⁴ We might compare Varro's friend's farm of 200 iugera at Reate, which generated an annual revenue of HS30,000 (RR 3.2.15). This should be understood in conjunction with a probable rise in the price of land from the second century onwards, and with the lex agraria of 111 which made more land private and so both able to be sold and able to be counted as personal wealth for census purposes.⁴⁵

This leads us to ask, how many rural citizens fitted into the active centuriate electorate? Here we are on very contentious ground and can give no certain answers. The equestrian census in the Augustan period was HS400,000. We do not know for sure whether it was already at that figure before Augustus, but Davenport is suitably conservative in saying the figure must have been higher than it was for the First Class rating, which in turns means "the first [End Page 137] property-class must in any case have been several times larger than the equestrian order in the broad sense of the term."⁴⁶ And we know there were thousands of equites in the early first century, because at least that many were proscribed by Sulla (App. BCiv. 1.95). At the other end of the socio-economic spectrum, Rosenstein argues that there were few proletarii at the time of the Second Punic War.⁴⁷ This all points to the active centuriate electorate of the first four census Classes being, at least, a large minority of the citizen body, some hundreds of thousands of men. Most of these were surely rural citizens. The comitia centuriata was not controlled by a tiny elite. Thus the question of whether "more than 10% of the adult male citizenry voted in any consular election" misses the more interesting question of who voted.⁴⁸

Rural voters of property knew their vote, and thus their presence in Rome, was valuable. As we shall see below, this fact shaped not only their decision to come, but whether it was worth others' efforts to facilitate that journey. It also shaped the behavior of candidates. It was likely but not certain that the middle and lower Classes would vote in any one year, so a prudent candidate would need to hedge his bets and attract middle and lower Class voters.⁴⁹ As attested close elections show (such as those of 143, 65, and 64 noted above), such voters could very easily decide the result. These voters were not, man for man, as valuable as those in the equestrian centuries or the First Class: there were more people in the middle and lower Classes, concentrated in fewer voting units. But that did not make them unimportant.⁵⁰ This vast size and complexity of the electorate, together with uncertainty about whose vote would count in any one year and how much, all contributed to the unpredictability which our sources depict: the electorate was often compared to a billowing and unpredictable sea which created anxiety in candidates (e.g., Cic. Mur. 35–36, Planc. 11, 15).

incentives and obstacles

While the number of potential voters gives us some idea of the scale and nature of an election, it can only be a vague guide. Yet I have already argued [End Page 138] that the evidence, such as we have it, cannot be used to prove either high or low turnout; it is suggestive at best, and no individual election for which we have evidence can be shown to be typical. A different approach is needed, one which examines the incentives operating on rural voters. Therefore, for the remainder of this article, I concentrate on the incentives and disincentives which governed whether rural citizens with property, especially those who lived at a distance from Rome, would come to the elections as well as how those incentives influenced the way they would vote. These were the incentives generated by the electoral system itself, in its interaction with the structures of Roman society. I argue that potential disincentives to travel could often be quite easily overcome. By their nature, my arguments are general. They apply primarily to what I have called the active centuriate electorate, although most strongly to voters of the First Class and least to those of the Fourth. Apart from ordinary citizens, I consider men of influence: influence over the tribes when they came to Rome and influence over voters in the rural towns of Italy. These are the men who Quintus Cicero refers to in a specifically electoral context as the gratiosi (Comment. Pet. 19, 24; cf. Cic. Mur. 47). At the lower end, these gratiosi in rural Italy were the local focus of loyalty for their humbler neighbors and the vital mediators between senators in Rome and voters on the land.⁵¹ When those rural voters weighed whether or not to go to Rome, what did they consider? What might dissuade them from the journey?

Most obviously, Rome might be a long way away: by the later second century, Roman citizens lived as far away as Parma in the north. Nevertheless, they were in the minority. Before Italian enfranchisement, most Roman territory lay along the two great routes out of the City, either north on the via Flaminia or south-east on the via Appia and via Latina. Laurence argues that groups of travelers on foot could manage 50–60km per day on the major roads, although I prefer to be more conservative and adopt ORBIS's "private" speed of 36km per day.⁵² Few people in the middle Tiber valley lived more than a couple of kilometres from a paved road, while most of the pre-Social War ager Romanus was less than two hours' walk from a road or river.⁵³ Those new citizens added after the Social War were not as well connected to Rome by road and were probably, on average, poorer. They also lacked the habit [End Page 139] of voting in Rome (and, at first, the social connections encouraging them to do so). Some rebel areas and societies had been badly devastated during the fighting of the 80s.⁵⁴ All this suggests they were somewhat less likely to vote than the older citizens. Therefore, even if the overall citizen population doubled (as the census figures suggest), the number of regular voters probably increased by a smaller amount. All this meant that, according to ORBIS, almost all who lived in the pre-Social War citizen territory closer than the ager Gallicus could reach Rome in less than four days. A journey by road from Ariminum would take eight days, compared to fifteen from Brundisium and sixteen from Mediolanum. But we know that, in the sixties and fifties at least, Cisalpine Gaul was electorally very important.⁵⁵ Cicero made a special point of joining the governor's staff in 65 to help his consular campaign because videtur in suffragiis multum posse Gallia ("Gaul looks like counting heavily in the voting"; Cic. Att. 1.1.2).⁵⁶ Cisalpine voters were of great importance to Caesar in the later fifties, supporting both Antony's augural campaign and (he hoped) Labienus in his prospective consular campaign ([Caes.] BGall. 8.50, 52). And a few years later, Cicero looked back on the days when "we" consular candidates used to canvass the province's coloniae and municipia (Cic. Phil. 2.76); he presents this as routine practice.⁵⁷ So if Cisalpine voters would normally make it to Rome in such numbers that they were worth the time and effort of consular candidates, then presumably this would also be true for those areas much closer to the City.

Moreover, people in Roman Italy were used to traveling; we should not regard a lengthy journey as itself a major disincentive unless it prevented necessary work. Farm produce had to be brought to market.⁵⁸ Military service and attendance at the census required journeys. Travel was part of the life experience of most Roman citizens.⁵⁹ Our perception of the inconvenience of multi-day travel is skewed by our own experience of airplanes and fast trains which allow us to travel across the world in barely a day (a journey with which southern hemisphere academics are particularly familiar). But compared to [End Page 140] the journey to Spain or Cilicia (which could take months), spending two to three weeks traveling would not in itself be regarded as a major inconvenience. This was particularly true when travel could bring financial benefit.

Life as a farmer required a lot of work, so one might object that the need for agricultural labor would prevent voters from leaving home. It might, but only in some circumstances. Roman Italy was characterized by mixed farming, with cereals much the most important crop. However, while farming was certainly labor-intensive, there was little work that had to be done at a specific time; most could be brought forward, delayed, or spread out if less labor was available.⁶⁰ In such circumstances, it would not be difficult for the head of the household to leave his wife, children, and slaves to mind the farm, although here again the obstacles were lowest for those further up the Class hierarchy, with their greater means allowing more slaves to be kept and laborers to be hired.⁶¹ The exception was the harvest, which, for grain, began at or after midsummer and might continue into August. The harvest needed all available labor in order to get in the crop before bad weather struck.⁶² This corresponded with the date of the consular elections after Sulla, which were frequently held in July.⁶³ The effect (and perhaps the intention) would have been to make it difficult for small farmers to attend—at least when the civil and solar calendars were synchronized, as they frequently were not. This was also the case for Ti. Gracchus's attempt to be re-elected tribune in the summer of 133, when the harvest prevented his rural supporters from attending (App. BCiv. 1.14.1). Any elections held at harvest-time would therefore be much more poorly attended. However, if harvests were gathered before the elections, then the need to bring grain to market might be a powerful additional spur to come to Rome. But harvests were not a factor in the later second century when (it appears) elections were still normally held towards the end of the year, in November or December. Overall, then, the timing of elections in relation to the harvest could affect voter turnout. But as timing was inconsistent, the effects varied.

Were there politicians who exercised such control over a voting tribe as to turn it into a "pocket borough"—that is, to simply designate a handful of [End Page 141] voters to implement their will? Or, alternatively, were any powerful forces able to successfully discourage or prevent voters from attending elections, as with "voter suppression" in the contemporary United States? These are two separate problems but amount to the same outcome: the number of voters being controlled (or at least strongly curtailed) by pauci potentes.⁶⁴ We have no positive evidence for this happening. Moreover, if it did happen, we should expect references in the sources to some tribes being predictable: winning those tribes would be a matter of winning over their boss, rather than appealing to the voters themselves. But while we do see men of influence in certain tribes, there is no indication in our evidence of anyone with that type of power; indeed, the whole tenor of our evidence is against it, at least in the Ciceronian period.⁶⁵ Carrying any tribe was a difficult and uncertain business, which could be explained in retrospect (as Cicero claims he can do at Planc. 48) but not easily arranged in advance. Tribes were too big and too amorphous for any one political boss to control them. As Taylor shows, by the second century, the majority of tribes had multiple discontinuous areas, and most of the remainder were very large.⁶⁶ Everything we know of Roman society argues against local magnates having this sort of control over their freeborn neighbors. Moreover, there were substantial incentives for voters to go to Rome (see below). Any influential man depriving them of these incentives would need powerful levers of his own, whether carrot or stick. We see this lack of control in 53–52: Clodius opposed Milo's campaign by delay, bribery, and supporting his opponents rather than by trying to exclude his supporters (Cic. Mil. 25; Asc. 30–31C). Similarly, Verres tried to prevent Cicero's election to the aedileship by bribery rather than any other means (Cic. Verr. 1.22–23, 25). Compare also Planc. 54, in which the possibility is raised of voters being bribed not to vote for a certain candidate; in practice, this was achieved by having them vote for another. If any candidate, trying to shore up his own support in a tribe, wanted to keep out certain voters, then his competitors (not to mention candidates in elections for other offices) would correspondingly want them to come. The multiple vote and lack of party tickets meant that Roman elections were not a zero-sum game. We must reject the idea of powerful political actors keeping voters away.

One might object to this picture by pointing to Comment. Pet. 18, where Quintus refers to men who have, or expect to have, control over a tribe or [End Page 142] century, or to Cic. Fam. 11.16.3 where Dec. Brutus is supposed to have this power over some centuries of the equites. This is answered by the multiple vote, discussed in more detail below. Influential men might well have enough clout to ensure a particular candidate was one of the two, or eight, chosen by a particular tribal unit, but this is a long way from completely controlling the unit. Victory for one candidate did not automatically mean defeat for another.

The elections were potentially a lucrative event for many voters. Most obvious is the prospect of cash bribery by candidates, although we have very few indications of the scale of bribery. Those we do have come from the later fifties and are clearly much higher than normal: in 54 the centuria praerogativa was allegedly promised ten million sesterces to elect Memmius and Domitius (Cic. QFr. 2.15.4), while in 52 Milo tried to counteract his unpopularity by distributing HS250 per head tributim (Asc. 35C; this was in addition to money distributed the previous year). Clearly, the amounts on offer were not ordinarily large. But if a voter remained for the aedilician elections he could vote for, and so honorably accept money from, twelve different candidates: small bribes from each could accumulate into something quite substantial. And there were more benefits a voter might receive from all the candidates, not only those he intended to vote for: banquets, shows, seats at the shows, distributions of oil or food.⁶⁷ Other nobles might distribute largesse too, either as friends of the candidates or with an eye to their own future prospects (see e.g., Cic. Mur. 73). Election campaigns were one of the primary ways by which the wealth of the empire, mostly seized by the ruling class, trickled down to the mass of Roman citizens. Furthermore, for men whose annual income might only be measured in the hundreds of sesterces, the money on offer was likely to at the very least offset the costs of the journey and might go well beyond that.

We also need to understand the elections as a social occasion, bringing people together from across Italy (Cic. Verr. 1.54). Voters could renew relationships with people from all over the peninsula. Not only could this be a pleasure in itself but it also facilitated the transaction of all sorts of business: property deals, arranging marriages, clearing up disputed inheritances, and more. At a general level, the elections would themselves contribute to the economic integration of Italy by bringing so many of the principal wealth holders together in one place each year. While elections may not have directly been responsible for the movement of many goods, they were an excellent occasion for exchanging information and doing deals. Because the elections were a major social event, they were a major economic event too. These travelers' experience of Rome was not of a place "bigger, noisier, more jostling and in [End Page 143] every way more strange and hostile than any other town in the experience of the ordinary Italian," because the average rural voter would not encounter the City alone.⁶⁸ He was probably traveling with a group from home: the incentives which brought voters to Rome operated on a large scale and would draw in many men from every community. Finally, there were the psychological benefits, encountering what Martin Jehne calls aristocratic Jovialität.⁶⁹ In no other sphere of life could ordinary citizens expect nobiles to solicit their affection.

social network incentives

The most important incentives were those arising from social networks. The basic importance of reciprocity in Roman social relations is well known: it was a positive social value at all levels. As argued below, voting was an essential favor that citizens could do for others in their social networks, and the fact that they disposed of many votes meant they could do many favors. Moreover, this social fact applied to everyone (within the active centuriate electorate) in a given voter's social network, meaning that not only did incentives operate on voters as individuals, but operated on entire networks, and so acted more strongly. Simply put, a voter's relatives and neighbors all wanted him to vote as well. The act of voting in the elections was the essential act in what we may rightly, following Wallace-Hadrill, call a "patronage network" understood by the Romans using the language of friendship and fides.⁷⁰ These incentives were not one-offs but recurred each year, with each new election cycle. The individual candidates may have changed, but the fact that there were candidates remained the same, and these new candidates enjoyed much the same place in society, and operated within the same social networks, as the previous ones. That meant they sought votes from these same networks, and in the same way: the social network incentives to vote were therefore permanent, structural features of the electoral system.⁷¹ The incentive for voters to come to Rome in groups and vote was always present.

Scholarship on Roman elections normally adopts the point of view of candidates. This is to be expected: our evidence generally and the Commentariolum [End Page 144] Petitionis in particular are written from their point of view. So far I have tried to give the voters' perspective. But the direct connection between candidates and voters, whether understood as a social relationship or a projected image, was not the only relationship. There were also intermediaries, who greatly complicated the picture. Following the Commentariolum Petitionis (19, 24, 29), I refer to these intermediaries collectively as homines gratiosi, "influential men" and I schematize them as being at two levels: the lower, working locally in direct connection to voters, and the upper, working centrally with direct connection to candidates. Their function (at the upper end of the spectrum) was to deliver to candidates the votes of centuries or tribes. These upper gratiosi included senators and their sons and politically active equites (Comment. Pet. 33).⁷² This is, I believe, the meaning of an obscure passage at Comment. Pet. 18:⁷³

Ad conficiendas centurias, homines excellenti gratia. Qui abs te tribum aut centuriam aut aliquod beneficium aut habeant aut ut habeant sperent, eos prorsus magno opere et compara et confirma; nam per hos annos homines ambitiosi vehementer omni studio atque opera elaborarunt ut possent a tribulibus suis ea quae peterent impetrare.

For delivering the votes of the centuries, men of exceptional influence. It is imperative that you recruit and retain those men who, on account of you, have, or expect to have, control over a tribe or a century, or some other advantage. For, in recent years, men eager for advancement have worked very hard, with unreserved eagerness and energy, to be able to get what they want from their fellow tribesmen.

Most senators had been or would be candidates for office and so had recently devoted considerable efforts to cultivating the electorate. The relationships and up-to-date political knowledge so acquired were not for their own use alone; they could be made available to friends (Cic. Planc. 45). A good, if oblique, example is Cic. Att. 2.1.9: Favonius meam tribum tulit honestius quam suam, Luccei perdidit ("Favonius carried my tribe with greater credit than his own, but lost Lucceius'"). Shackleton Bailey plausibly suggests this refers to Lucceius's unsuccessful support of Favonius's candidature (for tribune?): in this interpretation, Lucceius was supporting Favonius, and so the failure of his tribe to vote for Favonius reflected badly on both men. Lucceius would himself contest the consular elections only a few weeks later, so his visibility and influence among the tribes was at its height.⁷⁴ Mutual electoral support was [End Page 145] thus much of the currency of senatorial politics. This support, surely, is what Quintus means by "favors done, the expectation of future favors" (Comment. Pet. 21), at least for those senators who did not have tremendous oratorical gifts to offer.⁷⁵ This is equivalent to the role the "economy of friends" played in financing senatorial careers, with one exception: in giving and repaying electoral support, what was "spent" was not necessarily lost, but perhaps even augmented.⁷⁶

Also relevant are the coitiones (electoral compacts between candidates) mentioned at Cic. Planc. 53–54.⁷⁷ These deal with specific tribes: Plancius allegedly "conceded" his own Teretina to Laterensis, and Plotius the Aniensis tribe to Pedius. While this was an extension of the idea of cultivating support for a friend it was quite different in effect: these men were rival candidates. Yet we can reconstruct the reasoning behind such arrangements. Perhaps Laterensis similarly "conceded" to Plancius another tribe in which he was influential, in which case each had picked up a tribe, or perhaps Plancius hoped by this manoeuvre to squeeze out a stronger candidate. In any event, the multiple vote (on which see more below) allowed this to be a mutually beneficial arrangement. The closest modern analogy is political parties with a common enemy agreeing not to run candidates in constituencies where they might split the vote, as was allegedly discussed between Labour and the Liberal Democrats in the UK's 2019 General Election.

Beyond those influential in Rome, there was a lower stratum of intermediaries, the locally influential men in the municipia across Italy. Comment. Pet. 32 nicely differentiates these two levels of gratiosi, between those who can exert influence over whole tribes and those who merely wield influence over some part of a tribe. The role of these local grandees can, I think, be understood most clearly if we conceive of votes in economic terms, with reciprocal favors being the currency.⁷⁸ Votes were a commodity, an asset owned by the voters, and very much in demand by candidates; this demand was a constant, recurring every year. We would, therefore, expect a mechanism to develop to match demand to supply. Votes flowed up this mechanism, from ordinary rural voters at the bottom to candidates at the top; favors flowed down. Starting at the local level, by voting as asked voters could win or repay favors from locally [End Page 146] important connections, those men influential in their district or municipium. By mobilizing votes, these local grandees could, in turn, win favors from those higher up the social scale, and so forth up the ladder. Of course, this schematizes what must have been an incredibly complex network of influence, with every man of influence probably working for multiple candidates with a greater or lesser degree of zeal. The Commentariolum Petitionis shows us the difficulty and importance of motivating such men (and women). But in all cases, the end result must have been the delivery of the votes in question to the elections in Rome. Only then were the favors actually worth anything. This in turn meant that it was in everyone's interest—the candidates, the gratiosi, and the voters themselves—to make it as easy as possible to deliver the vote.

I define the gratiosi here as a function rather than a social group: that function was brokering votes, or blocs of votes, in exchange for favors. That this function was a structural feature of the electoral system needs emphasis; there were, of course, no political parties in Rome, but what I describe here is similar to a modern party machine. It was a permanent structure of political life. We see some glimpses of this structure in the evidence, most notably in the Commentariolum Petitionis and the Pro Plancio (cf. Cic. Har. Resp. 42). The terms involved are familiar in scholarship: divisores, sodalitates, sequestres, and the like.⁷⁹ We know these held electoral importance, and it is clear that the terms used carried a heavy moral load, differing depending on who was speaking. It is likely, for instance, that the "men eager for advancement [who] have worked very hard, with unreserved eagerness and energy, to be able to get what they want from their fellow tribesmen" (Comment. Pet. 18: homines ambitiosi vehementer omni studio atque opera elaborarunt ut possent a tribulibus suis ea quae peterent impetrare—and note the positive or at least neutral tone used to characterize such men) are the same as, or at least heavily overlap with, the (clearly disreputable) sequestres and divisores mentioned at section 57, who are to be intimidated and made afraid. It is therefore fruitless to search for the "legitimate" and "illegitimate" functions of sodalitates or divisores. We would do better to understand that every important social relationship in Rome could be and was turned to electoral ends.

the multiple vote

All this talk of the influence that gratiosi exercised over ordinary voters is reminiscent of what John North memorably called the "frozen waste" theory of Roman politics.⁸⁰ A reader would be justified in asking how it can be reconciled [End Page 147] with the picture of an open, competitive electoral market painted by Yakobson (a picture I agree with). The answer is by the multiple vote, a sadly understudied aspect of the Roman electoral system. The multiple vote is the final piece of the puzzle.

If my picture of the incentives operating on voters is accepted, then these voters would have arrived in Rome with committed votes. But not all their votes would already be strongly committed; in the post-Sullan period, each voter could vote for, and thus honor, two consular candidates, eight praetorian candidates, and two curule aediles, not to mention quaestors and (if plebeian elections happened to be coordinated with those for magistracies of the populus) tribunes and plebeian aediles. Unlike in most modern electoral systems, Roman voters did not face an either/or choice. Instead, they were choosing (perhaps) twelve individual candidates from a larger pool to give their votes; in this way, the Roman system recalls modern multi-member constituencies, but with the crucial difference of the absence of parties.⁸¹ Therefore, even if rural voters were responding to social network pressures, as discussed above, they still retained some freedom in disposing of their remaining votes.

This is best shown by examining a hypothetical voter, Q. Vibius (a name attested at Venafrum in imperial inscriptions). Vibius is a farmer of the Second Class, from Venafrum, in the Volturnus valley near where Latium runs into Campania, and of the tribe Teretina. He has come to Rome at the behest of a group of local equites, major landowners in the area, to vote for two particular praetorian candidates and one consular candidate. But he has nine more votes to bestow in the consular, praetorian, and aedilician elections, and he only decides these votes in Rome. He is more committed to some than others, and so the lead offered by the centuria praerogativa sways him to an extent (on which, see more below). But a man from nearby Atina is standing for the aedileship, and the claims of vicinitas are strong (Cic. Planc. 21). Vibius's cousin recently served in Spain, and spoke highly of a praetorian candidate who had commanded troops there as a legatus (note how Cicero emphasizes the persuasiveness of soldiers' recommendations at Mur. 38). The uncle of Vibius's wife is canvassing widely for another praetorian candidate, while the agent of yet another candidate has promised ten denarii to everyone in Vibius's century if it should choose him; this candidate also staged famously lavish games as an aedile. One of the consular candidates impressed Vibius greatly by shaking his hand on the via Sacra and calling him by name. Vibius has recently received a timely favor from an acquaintance in Puteoli, and so [End Page 148] he is anxious to repay the debt by voting for the candidate that man supports. These are just hypotheticals, but they suggest how complex was the social web in which rural Roman citizens lived their lives, and the manifold ways in which the strings of that web could be tugged for electoral purposes. And if this complexity enmeshed rural voters, it also enmeshed the gratiosi who sought to influence them. In choosing who to vote for, a voter was deciding which voices to listen to, out of the many speaking into his ear.

This perhaps explains the role of the centuria praerogativa, a single century of the First Class which was chosen by lot to vote (and have its vote announced) before the rest of the assembly. According to Cicero, the candidate the centuria praerogativa chose first was always elected either that year or the following year (Planc. 49). Other examples from republican history attest to its importance (e.g., Cic. Mur. 38; Livy 24.7–9). It has variously been interpreted as ominous, showing Jupiter's will, or alternatively as a conscious mechanism to produce more random and decisive electoral outcomes, in order to minimize the importance of campaigning.⁸² But in view of the model I have put forward in this article, it is best interpreted as having a bandwagon effect, by giving a guide to uncommitted or weakly committed voters which way to direct their vote. In any case, we should not overstate its importance. As Mouritsen says, "towards the end of the republic the praerogativa may have given little more than a general boost to the authority of its chosen candidate, improving his chances now or in the future"; it was no longer true, if it had ever been, that "the outcome of the election could be predicted as soon as its vote had been declared."⁸³ And in consular elections with only three or four serious candidates, of various strengths, we should expect a candidate who received the most votes in a particular century to usually be elected. So the centuria praerogativa should be understood as yet another voice speaking to a voter, but in this case telling him what his fellow citizens were thinking.

The end result of all this was an electoral landscape that was exceedingly complex, certainly more complex than any known in the modern world, where the existence of political parties has made for much more structured elections. The secret ballot, introduced in the 130s, both speeded up the process of voting and made campaigning much more opaque. Candidates and gratiosi could not know for sure to what extent their efforts with particular voters had borne fruit. But this complexity also allows us to reconcile the two theories which have dominated modern scholarship on Roman elections: that they were a "frozen wasteland" dominated by private social ties between [End Page 149] candidates and voters, and that the electoral market was free and voters had to be won over by reputation, magnificence, and cash. While it is not fruitful to use the terms "patron" and "client," social ties between candidates and voters did exist and did shape voter behavior—as long as we recognize that these social ties were immensely complex and passed through many intermediaries. Rather than being bound in chains of dependence, a more useful metaphor is of citizens enmeshed in webs of social ties, of all sorts. Because voters were subject to social pressures from many directions, and because they had multiple votes at their disposal, they had to choose which pressures to respond to. This meant that candidates could not simply rely on their social connections: such connections had to be worked. Other resources had to be deployed, too: magnificent games, military reputations, direct or conditional bribes. But even these are best understood as additional threads in the web, which cumulatively decided which way voters in the Campus Martius would turn on election day.