ElAnt v6n1 - Pardon and Revenge in Tacitus and Ammianus

Volume 6, Number 1
July 2001


Pardon and Revenge in Tacitus and Ammianus

Ronald Newbold
Department of Classics, University of Adelaide
ron.newbold@adelaide.edu.au

Abstract

This study is based on 258 references (Tacitus 166, Ammianus 92) generated by 12 keywords such as venia, ignoscere, ultio and vindicare in Ammianus and in Tacitus' Histories and Annals . Not every instance of pardon or revenge in these works is thereby covered but abundant material from the keywords permits analysis from a number of angles, such as the seeker and the sought from, what incidents caused the need for pardon or revenge, how it was sought or gained (if it was). In addition, the authors often supply illuminating comments of their own or report the real or imagined thoughts or comments of the participants. The paper opens with some general remarks about pardon and revenge, and deals with problems of defining the terms. Here and throughout the paper relevant observations from ancient and modern authors are used. There are a number of ways in which Tacitus and Ammianus differ in their treatment of pardon and revenge. Their different epochs, choice of material and perhaps their personalities are some of the reasons for this. The elastic way in which the authors use the terms means the references can include a wide variety of situations, from the personal to the political.

Pardon and Revenge in Tacitus and Ammianus

Pardon and revenge are important areas of human behaviour and signifiers of societal values. Both concepts can reveal, amongst other things, the role of the state in settling disputes amongst citizens that might otherwise lead to lengthy feuds and damaging cycles of self-help justice. Both have much to do with justice, honour and definition of the state's power. The range of motives for seeking pardon or revenge, and the ways in which they are sought, form a social document. This study confines itself to the evidence of Tacitus' Histories and Annals , and Ammianus' Res Gestae . Although very different in their historical techniques and methods of presentation, both authors write expansively on contemporary or near-contemporary history. And Ammianus probably did see himself as a continuator and imitator of Tacitus [ 1 ]. They lived, however, in eras that in some ways differed markedly over what was appropriate reconciliation and retribution. Christians, for example, had strong views on these issues.

Within this sample, the study is confined to the use of certain words for pardon, viz., ignosci , venia and clementia , plus clemens and clementer , and, for revenge, ultor, ultrix, ultio, ulcisci, vindex, vindicare and vindicta . Neither author uses talio . There is inevitably something arbitrary about proceeding in this admittedly empirical manner but a considerable amount of material (258 references) is generated. The concept of pardon is part of a nexus that includes forgiveness, amnesty, mercy, condonation, reprieve, indulgence, restraint, moderation and pity ( moderatio, mansuetudo, concessio, indulgentia, lenitas, misericordia ), and elements of these could be present in any formal act of pardon [ 2 ]. Frequently involved in any experience of giving or receiving pardon are guilt, shame, justice, courage and faith. As we shall see, Tacitus and Ammianus do not use the chosen terms with the exactitude of a philosopher [ 3 ]. As in our society, so in Rome, asking for pardon may be no more than a courteous request for indulgence or permission to, for example, speak. Venia and ignoscere can be used in this sense, and are so used in our two authors. Initially the first word in the title of this study was going to be Forgiveness, which commonly connotes an inner letting go of anger and hostility, a restoring of a relationship not necessarily present in a formal act of pardon. However, to infer such absolution in a certain act may be unwarranted [ 4 ]. Use of pardon makes more justifiable the inclusion of a key factor in Roman imperial life, clementia , the merciful face of power, and requires less speculation about motives for granting it. Clementia may be defined as treating a person less harshly than one could, leniency, in effect, but when no action at all is taken against a person, clemency resembles the absolution of forgiveness.

This article will proceed with a further general discussion of the concepts of pardon and revenge and some of the problems of identifying instances in our authors (I); discussion of pardon and clemency in Tacitus and Ammianus (II); discussion of revenge in Tacitus and Ammianus (III); and conclusion (IV).

I

The concept of revenge is even more problematic than that of pardon because of the way it entwines with justice, retribution and punishment, terms which have connotations of impersonality, freedom from guilt and detachment on the part of the inflictor, especially when exacted through the brokerage or on behalf of an institution or the state [ 5 ]. The language of revenge tends to colour legal actions. A recent newspaper article on the acquittal of a suspect for the murder of a 14 month old boy spoke of the toddler's death as "set to go unavenged" [ 6 ]. The proper relationship of justice and revenge has always been a major preoccupation of law, literature and religion. Aulus Gellius saw the purpose of punishment as "the preservation of honour, when the dignity and prestige of the injured party must be protected, lest, if the offence is allowed to go without punishment, he be brought into contempt and his honour be impaired" [ 7 ]. It is precisely these considerations that fuel the search for vengeance, a vengeance which is, preferably, public, even dramatic, and which lead some aggrieved people to think that, where honour is concerned, they, as revenge-seekers can operate in a criminality-free zone [ 8 ].

Another difficulty is that acts of what seem like pardon or revenge can be narrated without the use of our keywords. Much of the senatorial business of A.D.70 reported by Tacitus was driven by a desire to make informers pay for their activity under Nero (cf. Hist . 4.10.1, 40.3 ), but not all the episodes contain revenge vocabulary. Or, feelings of revenge can be lurking in behaviour where one might not at first suspect it. Even though their victims have nothing to do with causing the feelings of anger, grievance and insult that they harbour, criminals and vandals often see their crimes as an act of revenge against an indefinite, hurtful Them [ 9 ]. The Vengeful Man may spend a lifetime redressing hurts endured in childhood, and the slights in adulthood that keep impelling vindictiveness may be so trivial that they do not make the historical record, particularly in the case of non-emperors. If a vengeful person is not only grudgeful and unforgiving, but also "remorseless, ruthless, heartless, implausible and inflexible...Passionately he moves towards punitive and retaliatory action" [ 10 ], one is irresistibly reminded of Ammianus's portrayal of Valentinian. Tacitus's Tiberius has been seen in just this way [ 11 ]. Moreover, a view that a ruler was prone to vengeance can be indicated by indirect means, such as enhancing the image of a more lenient figure (Germanicus versus Tiberius) or simply by describing numerous acts of saevitia that eventually raise questions about motivation [ 12 ].

Strictly speaking, the passion for revenge stems from receipt of an unprovoked injury where instant retaliation is impossible, but in any act of aggression it is often difficult to be certain that it is not part of a cycle of blow and counter-blow where the primal provocation (or provocations) has been lost and where incidents more serious than the immediate pretext have occurred. Even a recent provocation, which has been inflicted upon a relative or friend of the revenge-seeker, may simply have vanished from the historical record, or the revenge is so long delayed that one does not spot the connection unless it is spelt out [ 13 ]. What is called justice or punishment is often (consciously or unconsciously thought of as) vengeance exercised on behalf of a person or persons (sometimes dead) by others, such as state institutions and representatives, and the vengeful residues linger in terms such as ultio and vindicare when they are used in connection with jurisdiction. Hence it could be argued that judges who order executions are little more advanced morally than a lynch mob [ 14 ]. The state, once it assumes the power to punish, is wary about too much being left to private enterprise, partly because so many communal values and standards are at stake. The self-help element in revenge does, after all, favour the strong. Where revenge is driven by the need to demonstrate courage, or to repair damaged honour or a "narcissistic wound" [ 15 ], it may be incompatible with legalism if the law is not felt to offer suitable redress. Feuds are supposed to be different from wars in that they involve fewer deaths and have the maintenance of honour as their only objective. But feuds can escalate into warfare and, anyway, warmongers often appeal to national honour. Rivalries such as those between Marius and Sullla, and Pompey and Caesar, can be seen as contests for honour that got out of hand and harmed the public interest. And so Augustus had to combat the perception that the inimicitiae he inherited from his adoptive father were pursued out of odium privatum rather than utilitas publica (cf. Ann . 1.10.3).

Motives for revenge can range from the desire to repay humiliation and regain lost honour, prestige and self-esteem, ("I retaliate, therefore I am"), to making a statement about what behaviour by others is acceptable in the eyes of powerholders, how the hierarchy of power should be maintained. The best form of revenge is usually one that is highly visible. Roman aristocrats in the Republic competed in a charged atmosphere where honour was constantly at risk, and where politics was the most charged and risky arena [ 16 ]. Pursuit of revenge shaped much political behaviour and prosecutors sought to exact vengeance from defendants. And Tacitus uses ulciscebantur in referring to three famous political prosecutions of the Second Century B.C.( Ann . 3.66.2). While aristocratic competition was more muted in the Principate, it and the sometimes ritualistic pursuit of revenge by no means disappeared. Moreover, the emperor, de facto above the law and identifying the honour of himself with that of the state, could use the charge of laesa maiestas to avenge personal injuries. In the military context, ira , odium and cupiditas ultionis together may spur troops to action ( Hist . 2.77.3, of Otho's defeated troops). Motives for pardon can range from expediency and helplessness to a genuine conviction that nursing anger, hatred and thirst for revenge do more harm to the self than to an other. There may be a belief that the beneficiaries of pardon are capable of mending their ways if given an opportunity, that a cycle of violence needs to be broken, or that resistance can thus be more quickly ended. The law, or conventional morality, may be thought too harsh in some circumstances.

On the other hand, emperors, who have the greatest means with which to pursue vengeance, may believe that this is a better means of ensuring their safety than the clemency Seneca urged upon Nero. After all, sparing enemies is risky. So is accepting offers of pardon. King Archelaus of Cappadocia was offered clemency if he came to Rome. Having fulfilled that condition, he failed to achieve it ( Ann . 2.42.3; cf. Hist . 4.70.1). Particularly if emperors have identified themselves with the state, it is hard to determine whether punitive action by powerholders is dictated by state or personal interests [ 17 ]. Cicero had said that the state's security should come before clementia . Boudicca is made to tell her assembled forces that it is not her (and their) lost kingdom but her own scourging and her daughters' rape that she is avenging ( Ann . 14.35.1; cf. Ag . 32.4). Tiberius finds it necessary to distinguish what in Piso's conduct he will avenge ( ulciscar ) personally, and what vengeance owed to Germanicus' children and to himself as a parent - indeed to any person whose kin is murdered - must be exacted by the senate ( vindicandum , Ann . 3.12.2). Given a world where the importance of defending one's honour and the expectation that no insult to it, however slight, should pass unavenged, that sometimes the restoration of esteem may require extreme measures, that there are some people in any society who take accidents as personal insults, our two historians can give us only a glimpse of what occurred in these areas of behaviour [ 18 ].

This article is a study of how certain words that indicate pardon and revenge are used, and how their elasticity enables them to be used in a wide range of situations. They occur sufficiently often in our authors to permit some conclusions about how they saw these concepts, and how Roman society may have changed from Tacitus' time to Ammianus'. Because episodes of pardon and revenge can be narrated without using one of our keywords, it is less than a full treatment of pardon and revenge in two historians. But it is also more, because some usages of the venia etc. veer towards permission and leniency, and of ultio etc. towards relatively impersonal punishment. Discussion now turns to occurrences of clemency and pardon in our authors.

II

Tacitus' Annals and Histories use ignosci 9 times, venia 38, clementia 33, and clemens twice, total 82 occuring, on average, once per 10.9 pages of Bude text. The respective figures for Ammianus are 6, 25, 10, 12, total 53, occuring, on average, once per 14.8 pages of Bude text [ 19 ]. Tacitus's references contain 30 instances of individuals or groups who seek or obtain pardon. These include senators, senators' wives, members of imperial families, rebels and foreign rulers (whose difficulty may be with the Roman government rather than with emperors personally), legionaries and commanders who have mutinied or surrendered. Senators may need pardon for usurious money lending, for leading lives of ill-repute, or for having offended the emperor in some way. There is also a mythological reference to Amazons seeking asylum at an altar of Liber. Ammianus has 18 such references and the profile of the supplicants is noticeably different. Political opponents of an emperor, a usurper, and rebel leaders figure, but it is barbarian tribes and defenders of recalcitrant cities such as Aquileia and Pirisabora who are more prominent. Also different from Tacitus's list is the case of Constantius's negligent bodyguard.

Emperors naturally figure largely as bestowers of pardon in Tacitus, but commanders of Roman or rebel forces, foreign rulers, and the senate also feature. In Ammianus it is only emperors (including the usurper Procopius) and Roman commanders who thus bestow. But pardon may not be sought, because it is felt to be a hopeless quest (the harsher penal code of the Fourth Century, in particular, did not encourage people to think mercy was in abundant supply) or because pride prevents. The first and sixteenth legions which had surrendered to Civilis in 69 were too ashamed to ask their commander Cerealis for pardon. It was Cerealis's other troops who, by their tears and silence, sought pardon for their fellows. Cerealis eventually responded by saying that neither he nor Vespasian had retained any memory of their misdeeds [ 20 ]. The defeated troops of Caesennius Paetus, less severely disgraced, could, however, bring themselves to ask Corbulo for clementia ( Ann . 15.12.2). If pardon is sought, it may not be granted, as, for example, the knight Clutorius Priscus found ( Ann . 3.50.2). Someone may intervene to ensure the failure of a request, as Vitellius's sister-in-law Triaria does to Dolabella ( Hist . 2.63.1). Or third party intervention, such as Varus's on behalf of his friend Dolabella or Tigellinus's for Vinius's daughter, may not succeed, whereas Agrippina's for Seneca and the Aorsian king Eunones's for the former Bosporan king Mithridates do ( Ann . 12.18.2, 12.19-20). Vitellius's generals Caecina and Valens burnt their boats by pouring insults upon Otho to a degree that made any prospect of pardon unlikely [ 21 ]. A shared exclusion from any hope of pardon can bind people to alliances ( Hist . 4.70.5). Civilis recognised that his behaviour precluded any hope of pardon from Vitellius ( Hist . 5.26.2). As part of the motif of vengeance that runs through the demise of Germanicus, Germanicus himself hopes that his "murderers" get no pardon.

Gallus Caesar worries over whether he will be pardoned by Constantius (14.11.7). His pessimism was to prove justified. In 358, in a battle with Constantius's forces, the Sarmatians preferred defeat in action to venia (17.11.13). When the renegade and commander of some Persian troops, Antoninus, begins his short address to Ursicinus with ignosce mihi , he is not asking for pardon but permission to explain his desertion (18.8.6). He knew that his future was within Persian jurisdiction.

The reasons for seeking pardon in Tacitus, given 30 times, focus on warfare and rebellion: resistance to Roman power, within or without the empire or client kingdoms, and defeat in battle or civil war produces suppliants whom it was often sensible to forgive. This was a sphere where clementia traditionally but not invariably operated. Crimes requiring pardon include bribery, cruelty or malversation in the provinces, tampering with the loyalty of troops, plotting against or being negligent about the emperor's security, alleged or real adultery with a member of the imperial family (Seneca - Julia, Silius - Messalina), insulting the majesty of the emperor with abusive verses, and senators lending at usurious rates. Being related to a losing protagonist or living unseemly lives also produces suppliants.

In Ammianus, the most common reason (8 times) of the 18 given for seeking pardon is because barbarians have rebelled, infringed upon Roman territory, broken an agreement. Supporting a rival claimant to the throne or opposing an emperor also figure. Julian writes to Constantius pledging cooperation on certain conditions and asking for pardon ( ignosce mihi ) on the grounds that he was forced by his troops to become an Augustus (20.8.12). Other reasons include resisting a siege, being faced with annihilation in battle, or failing to adequately guard the emperor's security. The Alamanni seek peace and pardon because unfavourable auspices warned against battle (14.10.9).

Sometimes included in a notice is reference to how the pardon was sought: 14 times in Tacitus, 12 times in Ammianus. These references illustrate ritualised forms of behaviour. Seeking pardon is often difficult because of the amount of pride that has to be swallowed and because it amounts to an admission of guilt but, unless one had a character like that of the younger Cato or one is caught up in the heat of battle, the instinct for survival and the need for security usually compel. The act of seeking can amount to a tacit apology that is often sufficient to assuage another's wounded honour. We have already noted the role of intermediaries. Other examples are Livia interceding on behalf of Piso's wife Plancina, or Messalina approaching the most senior Vestal Virgin for help. Messalina also sent her children to Claudius to intercede, and resolved to seek clementia from him in person ( Ann . 11.32.2). Narcissus's intervention to prevent these pleas reaching Claudius was decisive for the failure of her plan. Officers appeal on behalf of troops ( Hist. 2.51.1 ). Explicit apology or some form of self-abasement increase prospects for receiving pardon. Suppliants accordingly beseech, make excuses, weep, blame themselves, send a petition or delegation, or approach in a body, as the usurious senators do in 33. Some kind of deal may be offered. The besieged inhabitants of Uspe in the Crimean Bosporus thought that offering 10,000 slaves to Eunones and a Roman force would improve their prospects of survival ( Ann . 12.17.1). They were wrong.

References to body language or location, such as falling on knees or throwing down weapons or supplicating at an altar, provide further information on how pardon was sought [ 22 ]. The rebel and suppliant Mithridates of Bosporus, we are told, adjusted his dress and features to the occasion when he subsequently appeared before and fell at the feet of Eunones ( Ann . 12.18.2). By contrast, when he appeared before Claudius, he proudly refused to act the suppliant before one who, it seems, was already committed to clemency ( Ann . 12.21). Notably defiant too is the behaviour of the British king Caratacus before Claudius in Rome, not only because of the amount of circumstantial detail supplied or speech put in his mouth, but for the pride and spirit he shows while successfully asking for clemency when in the power of his conqueror. Not for Caratacus the fearful prayers and submissive look of his vassals and kin. He argues that it is his worthiness as an opponent that has given Claudius the excuse to celebrate a triumph, and that the same worthiness will serve as a perpetual memorial of granted clemency ( Ann. 12.16-17). Perhaps Claudius intended mercy anyway but Tacitus suggests Caratacus's demeanour played a part. The perceived "triumph-worthiness" of Caratacus helped persuade Claudius he could derive honour in dealing mercifully with this man. On the other hand, Claudius had already dismissed Mithridates as not "triumph-worthy" and had to endure contumacious behaviour from him.

Generally more interested in nonverbal communication and perhaps influenced by the theatricality of his age [ 23 ], Ammianus places more emphasis than Tacitus on the body language of the suppliants, such as prostration, bowing the neck, outstretched hands, or, on one occasion, the use of a supplicatory Assyrian gesture. Excuses may be offered. Ammianus has five references to the use of intermediaries - envoys, chiefs, representatives, commanders, Christian priests. Pardon may be offered without it being requested, such as when prolonged anti-guerilla war fails to subdue Tacfarinas' rebels and a different tack is tried ( Ann . 3.73.3).

Forms taken by the pardon in Tacitus (supplied 19 times) are unconditional or conditional, requiring action by a certain date [ 24 ], coming with a reprimand for some or all of the recipients, or exacting punishment from ringleaders. Surrender of hostages may be part of the deal. Lives are spared, safety guaranteed, former foes allowed to settle in Roman territory, exiles recalled, senators restored to the senate, or a lesser penalty imposed, such as exile instead of death. In the sort of quirk that Tacitus relished, pardon is granted for an offence (treason) different from the one in question (extortion) to three senators who had been expelled from the senate in 69 ( Hist . 1.73.3). Pardon at the bar of public opinion is supposedly bought by Nero giving gifts to potential accusers or condemners of his murder of Britannicus ( Ann . 13.18.1).

In Ammianus, forms of pardon (supplied 11 times) are mostly a matter of lives being spared. This may come at the price of giving hostages or witnessing one's city being burnt. Striking a peace with a foreign enemy (14.10.9) or allowing them to settle in a different place from that which they have usurped can be presented as a pardon. The elder Theodosius's diplomacy won tribal allies in Africa by a mixture of fear, rewards and pardon for past insolence (29.5.33). A generous note appears to be struck when suppliants are received with a kiss or given provisions. Ammianus says that this was in Rome's interests ( rei publicae conducebat , 29.5.16). Julian not only sends a suppliant home safe but bids him be carefree (22.14.5). This man, a former governor, had begged Constantius to send him the head of the rebel Julian. He had broken no law but assumed Julian would want to punish his behaviour. Julian's magnanimity was not a matter of undermining the workings of Roman law, simply a manifestation of the humanitas that tempered power. Forgiveness can be inspirational.

There are a number of instances where the author or a participant in the episode comments on the issue of pardon, or the reference includes a generalised, speculative, or hypothetical observation, such as the likelihood of Germanicus surpassing Alexander the Great in clemency if he had lived longer [ 25 ] or the governor Didius Gallus exaggerating disorder in Britain in 51 so that any difficulty in curbing it would be more easily forgiven ( Ann . 12.40.1). Such comments might provide clues to contemporary thinking and points of comparison with the views of people like Cicero and Seneca. Thus Tacitus, fluctuating between rigour and compassion, can suggest that offering pardon or seeking it on behalf of others might be prudent, convenient or praiseworthy, depending on the character or situation of those involved [ 26 ]; but that moral standards, military discipline or the public good may thereby be endangered ( Hist . 1.77.3); and, ever ready to attribute motive, he imputes that the seekers are driven by self-interest, vainglory, shame, cowardice or fear, or by some mixture of motive. Tacitus conveys the agonising dilemma faced by the troops besieged by Gallic rebels at Mogontiacum: choose the venia offered by Classicus, or famem ferreumque et extrema ( Hist . 4.59.3). Despite dilating upon their severe further hardships as they considered the offer, Tacitus makes it clear his view that their ultimate choice for flagitium and survival over laus and decus was shameful. In the end, their choice was really about death by starvation or death by fire and sword when they were treacherously ambushed as they left the camp. Tacitus stresses that it was not from feelings of mercifulness that Tigellinus intervened to try to save Vinius's daughter but to incur a potentially lifesaving obligation for himself. After the first battle of Bedriacum, Otho supposedly seeks a reputation for clemency, but also and more altruistically, hopes that by a swift suicide he can win clemency for his nephew ( Hist . 2.48.2). Earlier, he had sought the clementiae titulus in his treatment of the consul-elect Marius Celsus, who was widely admired, but hated by the partisans of Otho for his loyalty to Galba ( Hist . 1.71). However, Otho's failure to take action against Vitellius's family in Italy may have been due to fear, says Tacitus ( Hist . 1.75.2). Civilis and Classicus are aware of the value of a reputation for clemency as they establish a novum imperium in Gaul ( Hist . 4.63.1). The ready clemency of Gaetulicus, governor of Upper Germany, won him such popularity with his troops that, as a result but possibly as an intention, he alone of Sejanus's adfines survived unscathed to the end Tiberius's reign ( Ann . 6.30). When the knight Cominius was spared by the emperor for anti-Tiberian verses, Tacitus reports wonderment that Tiberius did not pursue further a reputation for clemency ( Ann . 4.31.2). Even the ferocious Valentinian thought it worthwhile to appear clement at times, according to Ammianus (30.8.2).

Tacitus acknowledges that a combination of pardon and firmness can be effective in quelling an uprising such as that of the Gauls ( Hist . 4.69.4). In Armenia Corbulo uses one or the other depending on the conduct of the local community ( Ann . 14.23.1). With grim irony, Tacitus records as clementia Claudius's granting Valerius Asiaticus, condemned for treason and adultery, the right to choose the manner of his death ( Ann . 11.3.1). The murder of Britannicus by Nero prompts the comment that the gods may be less forgiving than humans towards such crimes as the removal of a dangerous rival, a comment that raises the perennial issue of whether rulers should be accorded greater indulgence for certain crimes, especially when their security is at stake ( Ann . 13.7.1; cf. Sen. Clem .1.4). The role of public opinion in granting indulgence or pardon to offenders is evoked when Tiberius contemplates the prospect of criticism for allowing temples to be erected to himself ( Ann . 4.37.3), and when Piso and Germanicus' "murderers" are thought of - they can expect none. The spread of certain vicious rumours made any indulgence in the latter case very difficult to earn. Chance factors may apply. Tacitus suggests it was an old emperor's forgetfulness rather than clemency that saved Rubrius Fabatus towards the end of Tiberius's reign, i.e., pardon was involuntary ( Ann . 6.33.2). There are caustic comments about Seneca's doctor, Annaeus Proxumus, making a nonsense of pardon from Nero by the folly of his end, the Parthian king Gotarzes cutting off Meherdates's ears to advertise his clemency and Roman dishonour ( Ann . 12.14.3), and Claudius having the front to recommend to Meherdates that he practice clemency in Parthia. A wry comment that mercy for one partner in crime (Plancina) made it less likely for the other (Piso), like some other Tacitean comments on this issue, is not that helpful. The granting of pardon can be controversial because it may be seen as implied criticism of current codes of retribution. But, for Ammianus, laws had become so perverted and powerholders were so often out of control and tyrannical ( lex versus nimia potentia ) that the need to mitigate them was compelling.

As for comments or thoughts attributed to participants, the Cheruscan and Roman serviceman Flavus is reported as contradicting his brother Arminius and saying that Roman mercy was always available for those willing to submit: a heavier hand, poenae graves , awaited those who did not ( Ann . 2.10.1). When Eunones commends appeals for mercy on behalf of Mithridates and points out that pardon can be an honourable way to end a war, Claudius responds by saying that although he was able to inflict the harshest penalties, it was the custom of his ancestors to be merciful to suppliants: triumphs were to be won against undefeated foes ( Ann . 12.19.3). Gallic rebels, temporarily ascendant in 70, see the expediency and manipulative value of granting mercy to shorten the resistance of Roman troops to them ( Hist . 4.56.1). The young Nero's sense of compassion that would prefer to pardon rather than execute men is worth reporting for the grim contrast with what came later. Then cruelty combined with a facade of mercifulness [ 27 ]. When Torquatus Silanus committed suicide after being accused of imperial ambitions, Nero asserted, as usual ( ex more ) in such situations, that he would have exercised clemency had the trial proceeded ( Ann . 15.35.3). Nero supposedly allowed the revival of maiestas charges so that he could demonstrate clemency. Such a desire to appear merciful could be exploited. When the praetor Antistius Soranus was arraigned for scurrilous verses against Nero in 62, Thrasea Paetus argued that exiling Antistius would furnish a noble example of clemency while at the same time making him suffer more ( Ann . 14.48.4). He denied Nero the kudos of clemency that would have come with vetoing a harsh, capital verdict by the senate. The theme of pardon as part of a deal surfaces four years later in Antistius's comments on why he should be given a pardon: he has valuable information to divulge ( Ann . 16.14.2). Flavian troops regard the renown of clemency as an empty prize if they cannot attack and plunder Cremona.

Turning to Ammianus, the renegade Antoninus says he deserves consideration for his suffering at the hands of the greedy creditors but knows that desertion to Persian jurisdiction rules it out for him (18.8.6). The Persian king Sapor's claim to be clementissimus is belied by the harsh terms he offers to the Roman army under Jovian (25.7.6). Excuses offered by suppliants or their supposed thoughts about its likelihood could be taken as authorial comments on the issue of pardon. The African rebel Firmus simply states that he deserves clemency because he was driven to rebellion by Roman iniquity (29.5.8). Gallus was reportedly anxious about his fate because he believed Constantius did not accept excuses or forgive mistakes but was more likely to punish with death, just as he himself had done as ruler in the East (14.11.7). Plausible, but this could be Ammianus speaking.

Ammianus gives some indication of his own views on pardon by citing past sages such as Cicero (a figure regularly appealed to) and Isocrates on the topic. Cicero is supposed to have said that when in a position of power he sought reasons for pardoning ( ignoscendi ) rather than for punishing ( puniendi , 19.12.18). Ammianus approves of this sentiment and, with Valentinian's saevitia in mind, of Isocrates's view that a ruler should merit pardon more for a military defeat than for being ignorant of what is just. Ammianus agrees with Tacitus that a record and a policy of pardoning, especially when a foreign enemy has reason to fear Roman generals, can lead conveniently to readier submissions and shorter battles, sieges and insurrections, whereas lack of mercy on the part of the Roman government or its officials prolongs or encourages action by individuals or groups that harms the Roman state, an echo of Diodotus's contribution to the Mytilene debate (Thuc.3.46). Constantius's clemency towards some Sarmatians in 358 encourages others to come forward (17.12.12). When the Alamanni in 354 send their chiefs to seek peace and pardon for their offences, Ammianus inserts a speech by Constantius to his troops that puts the decision to them and concludes with a pointed observation that making peace will not be seen as craven but a mark of restraint and humanity (14.10.9-15)-Ammianus's sentiments exactly.

As Julius Caesar discovered on the Ides of March, pardon can be abused when beneficiaries repay their benefit with treachery (15.4.8). Traitors have no right to it, although Ammianus feels Procopius's supporters were harshly treated by Valens. Nor were Roman commanders Lupicinus and Maximus forgivable for their unpardonable treatment of starving Goths in Thrace, giving dogs for food in return for slaves (31.4.10-11). Those offering pardon may be planning treachery. Bloodthirsty emperors like Valentinian and Valens were so cruel that public opinion would never pardon them, and Constantius's faults would have been pardonable if he had listened more to his eunuch Eutherius and been freer with pardons (14.5.5). Instead he disfigured his reign with paranoia and cruelty, although he sought a reputation for being just and clement, and indeed showed leniency at times (22.16.10-11). The magnate Petronius Probus had qualities such as the readiness to defend a slave and client even if flagrantly guilty, but once having decided to harm someone he would stop at nothing and could not be induced to pardon errors (27.11.6). In contrast to Valens, who, in Ammianus's view took measures for his security that were pardonable but who was also guilty of unpardonable cruelty and disregard for justice (29.1.18), Julian deplored a fugitive having to hide without any hope of pardon (22.7.5). Like literary forebears who go back to Hellenistic times, Ammianus strongly approved of a ruler like Julian who sees the exercise of clemency as one of his cardinal virtues (22.14.5). Never mind that it was a substitute for freedom under the law. Julian's clemency was a facet of his concern for justice and a most admirable attribute. (22.10.5). Ammianus believes he was as clemens as Antoninus Pius [ 28 ]. Although the treacherous king of the Alamanni, Vadomarius, had no hope of pardon from Julian when he was captured; he was dealt with mercifully, being sent to live in Spain without even being reproached (21.12.20). Since he could, presumably, pose no further threat to Rome, an impressive display of clemency was in order.

To take stock: as inevitably happens in works about high politics, references to pardon in Tacitus and Ammianus depict it as a powerful instrument of policy rather than, say, a display of altruism, a manifestation of divine grace or a doorway to the redemption of an individual's character [ 29 ]. Clementia , after all, was a prerogative associated with victory in civil war. Factors that make pardon possible, or more likely, include past services, indispensability, turning state evidence, apology, intervention by another, a powerholder's forgetfulness, desire to contrast with a predecessor, or simple inability to do much else in the circumstances. The sharp distinction drawn by Seneca between venia (remission of deserved punishment) and clementia (free discretion that does not judge according to a formula, a better way to justice) is not observed by Tacitus or Ammianus. Tacitus in particular is willing to ascribe interested or shabby motives. Self-interest dictates that offering enemy troops hope of pardon can reduce one's casualties ( Hist . 4.56.1). Thus, despite the harm and humiliation Civilis had inflicted on Rome, Cerealis could yet offer pardon to him ( Hist . 5.24.1). Roman diplomacy could countenance almost anything if advantage offered. Germanicus pardoned the Cheruscan Segimerus and his son in 15 despite rumours that the latter had dishonoured the corpse of Varus ( Ann . 1.71.1; cf. 2.22.1).

Bestowing or withholding pardon can create a powerful impression or a strong controversy on the merits of such an action. Selective pardons might help to divide and rule. Withholding pardon may be the flip side of a quest for revenge. Valentinian was as niggardly with pardon as he was free with punishments (27.7.7). In some circumstances, it may be deemed disgraceful to seek pardon, and even more shameful to grant it. Reputation and honour are at stake here, and the pardon of a political offender has many more repercussions than the pardon of a common criminal. It is a paradoxical situation, to say the least, when Roman foes offer Roman troops pardon if they surrendered ( Hist . 4.59.3). Senatorial enthusiasm for imperial clemency shows how firmly autocracy was in place. Displays of mercy can aggrandise the bestower at the expense of others. Therefore the city prefect Flavius Sabinus refuses to compromise his own safety and spare Dolabella if it means winning a reputation for clemency at the expense of Vitellius ( Hist . 2.63.2). Granting pardon is a risky ploy in the competition for honour, and may thus require more courage than revenge does. It has been suggested that the reluctance of officials to exercise clemency in the Fourth Century stemmed from a reluctance to encroach upon an imperial prerogative [ 30 ]. The bestower of pardon may feel that, just then, they are not despised as they render what may be seen as a particularly measured form of humiliation. But later they might be scorned as weak - or arrogant. There is also the risk of patterns of wrongdoing being perpetuated through leniency and, if pardon is sufficiently indiscriminate, the distinction between good and bad is lost and society suffers moral damage (Sen. Clem . 1.2). Pardoning the flagrantly guilty, moreover, will often leave the injured third parties still thirsty for justice or revenge. Relatives, who may feel guilty about failure to protect the injured person, may be less willing to forgive than the victim.

Another impediment to pardon is that insults to an emperor may be thought of as involving not just his honour but the fabric of the society that he stands for and is supposed to safeguard (14.5.4). Informers and the senate may zealously take it upon themselves to repair damage to the emperor's honour/safety of the state, as they do in the case of Clutorius Priscus, Sejanus's associates and Antistius Soranus. Questions of fortune and justice arise when being a brave man in a cowardly military unit or being related to a wrongdoer or opponent puts one in need of clemency. While it is a cardinal principle of justice that the children of wrongdoers should not be directly punished, fear of revenge from such a quarter and the temptation of a pre-emptive strike put them in need of a kind of ongoing clemency. The less protection offered to the individual by the law, the more scope is provided for the exercise of executive clemency, which is why Ammianus is so concerned about conduct in this area. Cruelty, vengefulness, and self-interest often reinforced each other to undermine the exercise of clemency.

Tacitus's and Ammianus's treatment of pardon raises a number of issues, some of which have already been touched on. The status, degree of responsibility, connections or the indispensability of a person, together with the circumstances of the bestower, determine whether the offences secure indulgence. In the cases of the erring Julias, daughter and grand daughter, their relationships with Augustus worked against them ( Ann . 3.24.2). Ringleaders of mutinies in 14 are executed while the rest are pardoned ( Ann . 1.44.1). Leaders of Aquileia's resistance to Julian in 362 are beheaded or burnt alive, the remainder pardoned (22.8.49). But during civil wars even ringleaders may escape stern measures ( Hist . 2.29.2), and when all senators were guilty of usurious lending, only a blanket pardon could apply. The pretender Procopius in 365 pardoned all his opponents except Serenianus, whom he imprisoned (26.8.11). With foreign foes Rome played Hard Cop and Soft Cop roles, employing vis atque clementia in Thrace ( Ann . 4.50.1). This is a recurrent theme. Some executions, and pardon for the rest, quelled the Brigantes in 50. But this policy failed against the Silures. Only occupation of their territory worked, vis alone ( Ann . 12.32.2; cf 2.10.1). Corbulo, campaigning in Armenia in 60, calculated that much could be achieved through clemency. Nevertheless, he knew the value of fear to induce the locals to seek Roman mercy. As Tacitus tells it, they had a choice. Submit and receive venia , resist and be burnt alive or put to the sword ( Ann . 14.23).). In the midst of the Messalina - Silius crisis, Narcissus seeks pardon from Claudius for keeping quiet about past scandals ( Ann . 11.30.2). Being indispensable just then, he could be confident of receiving it. Pardon is more readily available from protagonists, desperate for support, in a civil war. At least some potential critics can be bought off by generous gifts ( Ann 13.18.1). Plancina is depicted as benefiting from the favour of Livia but pardon may not remove dishonour from troops who surrendered ( Hist . 4.72.3). Tiberius had only public disapproval to fear in saving Plancina but when Vitellius wanted to spare the commander of the German fleet, Julius Burdo, from the soldiers who demanded his blood, ignoscere non nisi fallendo licebat : Vitellius had to keep him in prison and only release him in the euphoria of victory ( Hist . 1.58.2). Victory, and the power and confidence it brings, permits the dispensation of pardon that in turn advertises one's power. The role of public opinion is important because it is where, if nowhere else, a ruler may have to seek pardon. And people who never seek pardon themselves may be mean about granting it to others. It is rulers who most disregard public opinion (ironically, showing a Stoic disregard for externals) who were the most dangerous and tyrannical.

Like Cicero, Tacitus is wary of the damage done to discipline or the social fabric if pardon amounts to excessive leniency. Some deeds or persons should not be pardoned. For Ammianus, not pardoning is what threatens law and social cohesion, because rulers and magistrates become despots. He approves of Julian for saying that clementia which violates the law is proper and praiseworthy (16.5.12; cf. 22.3.4). It is lack of clemency which violates justice and undermines natural law (21.16.11). The excessively punitive attitude of Valentinian was a menace to everybody. When the senate weighed the scales of justice and sentenced Hymetius, proconsul of Africa, to exile, Valentinian was enraged at such "clemency" (28.1.23). Merciless behaviour also jeopardises one's own prospects of enjoying clemency, something Gallus may have pondered in his last days, if Ammianus's portrait of his ferocity is justified. Defeated at the second battle of Bedriacum, Vitellians found that the clemency they showed to the vanquished after the first battle stood them in good stead, a nice example of the power of pardon to evoke reciprocal behaviour ( Hist . 3.31.1: cf. 2.51.1). This concludes the discussion of pardon and clemency.

III

Turning to revenge, Tacitus uses ultio 46 times, ultor , omitting Mars Ultor, 9, ulcisci 14, vindicare , omitting instances where it means claim, 5, vindex , omitting Julius Vindex, 3, and vindicta , omitting instances where it means rod, 7, total 84, occurring, on average, once every 10.6 pages of Bude text. Ammianus' incidence is, respectively, 1, 6, 2, 15, 3, 8, plus ultrix 4, total 39, occuring, on average, every 20.1 pages of Bude text. In Tacitus there are 53 instances of individuals or groups who seek or obtain, sometimes posthumously, revenge. As will be seen in this section, seekers include emperors, senators, Germanicus's friends, disgruntled commanders, troops (especially defeated ones), mutineers, rebels in provinces, opposing sides in a civil war, victims (and their relatives) of informers, the younger Agrippina, a murdered city prefect, foreign rulers, the populations of provincial cities, and "everyone". Those from whom revenge is sought include, emperors, governors and extortioners, cruel centurions, slanderers and assassins of emperors, conspirators, informers and former imperial favourites, mutinous troops, Germanicus's "murderers", the mob who rejoiced at the rumoured recall of Octavia, foreign rebels, Agrippina, Agrippina's accusers, Silius and Messalina, insolent slaves, people of another provincial city, enemies of foreign rulers. This heterogeneity of seekers and sought suggest that a variety of methods and reasons for seeking revenge are likely to be in evidence.

Ammianus has 16 instances of revenge sought or gained. Seekers are emperors (6 times), Roman soldiers, the elder Theodosius, the barbarian Austoriani, a woman of Smyrna, some slaveowners, Rome and Romans, and the eternal power of Justice. Targets include Persians, the barbarian Limigantes, Goths, Alamanni, Sarmatians, rebels, peasants near Lepcis, a husband and stepson, Septimius Severus's Praetorian Prefect Plautianus, tardy slaves, those guilty of capital crimes, trivial offenders, Valentinus, and those who caused disasters to Rome.

The reasons or acts that occasion search for vengeance are given 35 times by Tacitus, 17 by Ammianus, and are adumbrated in the above two paragraphs. In Tacitus they include the murder of Julius Caesar ( Ann .1.9.4), and the death of Germanicus who on his deathbed encouraged others to avenge him ( Ann . 2.71.5). Wiping out the disgrace of military defeat is commonly expressed in terms of gaining revenge. Appealing to this notion can motivate battle-weary troops. Cicero had said that vengeance and defence were the only two right reasons for undertaking a war ( De Rep . 2.35). The plotter against Nero in 65 who wanted to make him pay for his scelera had joined the plot in ultionem ( Ann . 15.61.4). The inhabitants of Lyons urge Vitellian forces to exact vengeance from the city of Vienne for supporting Vindex and Galba (and point to the prospect of booty). In Tacitus's account, the hatred and envy that spurred this exhortation stemmed from Galba's earlier discrimination in favour of Viennne: he had fiscally exploited the Lyonese and at the same time conferred great honour on the Viennese. Revenge in the sense of punishment for conspiracy, treason or insults, such as the mad indiscretion of Silius and Messalina, cannot be postponed unduly, but those seeking satisfaction from informers or imperial favourites usually have to wait until a subsequent reign. Rebellion against Roman rule also evokes punitive responses, but rebels such as Boudicca have their own good reasons for seeking vengeance from Rome. Indeed, because Roman rule was sometimes perceived as so tyrannical, one could only exact vengeance upon it. One cannot expect justice from it. Use of revenge words to refer to action against riot, extortion, mutiny and lower class insolence are reminders of how often revenge and punishment are not distinguished. More clearly situated at the revenge end of the spectrum are the pique and disappointment of the navarch Volusius Proculus, caused by insufficient reward for his contribution to the murder of Agrippina ( Ann . 15.51.2), of Lucilius Bassus, appointed prefect of the Ravenna and Misenum fleets by Vitellius but not appointed Praetorian Prefect by him ( Hist . 2.100.3), or of Junia Silana, deprived of a desired match by Agrippina's disparaging remarks about her to the young nobleman in question ( Ann . 13.19.3). Whether or not a participant thinks punishment suffered was justified, revenge may be sought. Or so Tacitus represents the Marcomannian noble Catualda, smarting over his exile and leading an attack on King Maroboduus ( Ann . 2.62.2).

Ammianus too has plots or strikes against rulers, rebellion and attacks on Rome that have to be responded to, such as the killing of Julian and the treacherous charge of the Limigantes against Constantius when on the Danube frontier. In 372, the illegitimate son of the Moorish king Nubel, Zammac, was murdered at the hands of his brother Firmus. The comes Africae , Romanus, who had close ties with Zammac, sought revenge for Zammac's murder by Firmus by vilifying the latter to Valentinian (29.5.2). Capital crimes should be a matter for the state to respond to with punishment rather than revenge but the interchangeability of the two concepts is, as we have noted, commonplace. Nevertheless, when Ammianus speaks of Roman senators as severi vindices of trivial domestic offences, such as the slave being slow to bring water (28.4.16), is he implying that these conceited and narcissistic men are responding to what they perceive as a personal slight? Less uncertain is the case of the woman who killed her husband and stepson for killing her son by an earlier marriage (29.2.19). She may well have been thinking only of revenge and have had no faith in due process. The Austoriani seek vengeance for the execution of one of their tribe by Romans. The superhuman arrogance of Plautianus, attracts the avenging sword, gladio ultore (26.6.8.), presumably at the behest of Adrastia [ 31 ].

In 40 cases Tacitus tells us how revenge was sought, Ammianus 4 times, a large disproportion. In Tacitus, the main method is prosecution in court (12 times), usually leading to a conviction in the cases described by Tacitus, but possibly leading to pre-emptive suicide (e.g. Cn. Piso). More stories of court cases, then, are one reason why Tacitus has more on this aspect of revenge. And he has twice as many revenge references anyway. But there may be an additional factor. Seeking revenge through force of arms is also common, and Ammianus has plenty of military activity where such motivation might be mentioned but is not. So, why not? The revenge story is a popular literary motif and depends on vengeance not being exacted immediately. The moral ambiguity that often informs an avenging mission lends complexity and psychological fascination to the narrative. If one had a particular interest in revenge, then, arguably, it would be the manner in which it was sought that would most attract one. Typically, in stories of vengeance sought, the seeker delays plots and disguises intent, and has to because the intended victim is on guard. Tacitus takes a keen interest in this aspect of revenge. Thus, we are told that king Artabanus of Parthia controlled his libido vindictae for a conspiracy against himself that involved the eunuch Abdus sufficiently to invite Abdus, specie amicitiae , to a banquet and then administer poison ( Ann . 6.32.1). Delay is inherent in the process of initiating a trial, plotting a coup or planning a military campaign. Waiting for an enemy to become weaker is a good reason to delay, and the culture of the Principate was redolent with dissimulation and distrust. Informers and imperial favourites often become vulnerable in subsequent reigns. Prosecution of such people then may give access to records of maiestas trials from previous reigns ( Ann . 13.42-43, Hist . 4.40.3). Under Nero, Claudius's infamous prosecutor, Suillius, was accused of various offences, tried, convicted and exiled. Silana would not have tried to avenge herself upon Agrippina in 55 in the way that she did if the latter was still Claudius's wife. Catualda waited until Maroboduus's power was broken before launching an attack. The mutineers amongst the German legions in 14 had to wait until the breakdown of order engendered by the mutiny to repay the injuries they had suffered from centurions ( Ann . 1.32.4-33.1).

Other notices of how revenge was sought place less emphasis on deviousness and delay. Agrippina demanded an interview with Nero to position herself for revenge against her accusers. Her own mother had taken measures to maximise sympathy for herself as a new widow as a means of ensuring that the "murderers" of Germanicus would pay. Similarly, Sulpicia Praetexta, wife of M.Crassus executed in 61, and her four children, appeared in the senate to encourage prosecution of her husband's destroyer ( Hist . 4.42.1). An emperor can, however, try to seek to avoid odium by ordering those from whom he wants revenge to be hounded or prosecuted by a henchman. Or he can expect it to be delivered by a compliant senate. The means and the end of vengeance are indistinguishable when one only hears of the killing or assaults on unpopular centurions during a mutiny, or when the rebellious Britons under Boudicca use a variety of cruel methods on their Roman oppressors. If vengeance is presumed to enjoy divine sanction, then supernal powers may aid one's quest. Vibius Serenus calls upon ultores deos to deal with the son who prosecuted him for treason ( Ann . 4.28.3), and the Roman commander Vocula declares that they will deal with rebellious Gauls ( Hist . 4.57.2). Another commander Cerealis states that Roman history and tradition demand vengeance from enemies and that the gods will help achieve it ( Hist . 5.24.2). Boudicca thinks the gods will help the Britons exact vengeance from the Romans ( Ann .14.35.2).

Forms of revenge in Tacitus (given 38 times), partly foreshadowed above, include successful prosecutions of informers from a previous reign or of extortionate officials by provincials, and execution of alleged plotters. Quartering a legion on Capua for the winter of 69/70 and thereby ruining many families was a form of Flavian revenge on a community that had supported Vitellius. Words for revenge and the denial of pardon cluster around the death of Germanicus and the prosecution of Piso [ 32 ]. After the trial of Piso in 20, in thanking the imperial family for their services, Valerius Messalinus sees the outcome as a success for vindicta , (not justice: Ann . 3.18.3). Tacitus concludes the episode with the words is finis fuit in ulciscenda Germanici morte ( Ann . 3.19.2). Victory in battle, plunder of a city, invasion of territory and overthrow of an abusive ruler are forms of revenge for those who have previously suffered a defeat or oppression, or who have taken on a filial duty, such as Octavian pursuing Caesar's assassins. Agrippina is able to get Silana and her clients exiled for trumping up charges against her. The Parthian king Artabanus poisons Abdus for conspiring against him. Those of the elite who suffer from the insolence of the rabble get them convicted and imprisoned. Strong-arm action by troops against pro-Octavia demonstrators is described by Poppaea as iusta ultio ( Ann . 14.61.4).

Ammianus' four references to the method by which revenge is sought are force of arms, Roman soldiers making demands, handing a plotter and his closest associates over to general Dulcitius, and "many formidable means" (29.5.2). Forms of revenge in Ammianus (given 11 times), like Tacitus, include military victory and invasion, execution and poisoning, dismissal from office and prosecution, but also transferral to remote places, looting, inflicting 300 lashes, cutting off hands or compelling to run a gauntlet. It is less often possible to distinguish the means of seeking and the actual form of revenge in Ammianus.

Something of Tacitus's worldview emerges from observations that it is easier to repay injury than kindness, and that the gods care more about revenge than human welfare. He notes that soldiers are naturally keen to avenge the hardships of a guerilla war and that people can take sides on a matter according to whether or not they fear vengeance from others. Thus, Helvidius Priscus's revenge-driven prosecution of the Neronian delator Eprius Marcellus was, says Tacitus, arguably a blow for justice but too many complicit senators feared they too might have to pay if Priscus succeeded. The case failed ( Hist . 4.3.2). In the vendettas unleashed against informers in 70, only Egnatius Celer was convicted [ 33 ]. When Tiberius declares that it was the Roman way to exact vengeance from enemies on the battlefield, not by poisoning their leaders ( Ann . 2.88.1), Tacitus sneeringly dismisses this as an effort by him to put himself on the same high moral ground as the Roman commanders who once rejected an offer to poison king Pyrrhus ( Ann . 2.88.1). In different circumstances, however, Tacitus approves of Corbulo's successful plot to assassinate the German pirate and onetime auxiliaris , Gannascus. As a deserter and traitor he deserved no better ( Ann . 11.19.2).

Moving from authorial comment to the thoughts, motives and actual behaviour of participants, only someone overwhelmingly more powerful and secure in that belief need not fear the vengeance of those they have harmed [ 34 ]. This fear of revenge from those you have injured is plausibly attributed to the people of Cologne for having harmed the Germans in their midst, to Nero for having tried, so far unsuccessfully, to kill Agrippina, to Sejanus for having tried to usurp Drusus's power, to vulnerable Romans for their crimes against Britons, and to the son of Vibius Serenus for prosecuting his father. The governor of Upper Germany, Pomponius Secundus, expects the defeated Chatti to be eager for revenge and deploys his forces in expectation of their attack ( Ann . 12.28.1). King Rhescuporis fears vengeance from Augustus for violating arrangements the latter made for the Thracian client kingdom. The prospect of vengeance from Claudius quickly brings the Messalina-Silius wedding revellers to their senses. Junius Silanus is presumed to want vengeance upon Agrippina for having his brother killed and Britannicus upon those responsible for his mother's death. Tigellinus accuses Faenius Rufus of wanting to avenge his patron and benefactor, Agrippina ( Ann . 15.50.3). The SC Silanianum of 57 threatened the vengeance of the state upon slaves and freedmen if a master was killed by a slave ( Ann . 13.32.1), something that duly happened to the slaves of Pedanius Secundus in 61. Also with the future in mind, Vitellius tried to protect his power by making clear that assassins of emperors could face what he exacted from Galba's assassins ( Hist . 1.44.2: cf. 1.40.2). We are told that those who had falsely claimed credit for Galba's death were undeterred by the prospect of revenge, or rather, posthumous retribution. They miscalculated.

Anger and resentment can lie so deep that rebel Britons pursue vengeance even though they know they will pay for it later [ 35 ]. To witness the exaction of revenge from an informer is a mixed pleasure when the prosecutor was also an informer, or when the seeker after revenge is as bad a man as the one he targets, or is thought to be aiming more at fama and gloria . At the conclusion of the trial of Piso, Tiberius vetoed the addition of a gold statue and altar of Vengeance to the temple of Mars Ultor in order to distance the trial from Augustus's pursuit of Caesar's assassins, when private motives for revenge competed strongly and perhaps too overtly with public ones [ 36 ]. The sweetness of revenge does not have to be spelt out but it surfaces when people express their satisfaction at its achievement, as Valerius Messalinus did at the end of Piso's trial, and in 24, when Asiatic cities appreciatively voted temples for the vengeance ( ultio ) gained when an extortionate governor was condemned the previous year ( vindicatum erat , Ann . 4.15.3).

Ammianus' comments and hypothetical situations feature the by now familiar complaints about the excessive readiness of emperors to punish/exact vengeance, to which the merciful Julian was a welcome contrast. However, vengeance against foreign enemies was judged by different criteria. Premature death prevented the elder Theodosius from exacting a proper vengeance from the Quadi and Sarmatians. Julian was right to seek it from Sarmatians and Persians. He showed loftiness of spirit in determining to redress the many injuries Persia had inflicted on Rome. Desire for ultio and gloria launch his Persian campaign (22.12.1-2). The emperor Aurelian was a severissimus ultor of the harm done by the invading Goths (31.5.17; cf. 30.7.10). In contrast to Tacitus's somewhat sceptical and jaundiced view of the gods, Ammianus is more sanguine about divine interest in justice, and hence, as appropriate, punishment/revenge, when he refers to its ceaselessly vigilant eye, ( iustitiae arbiter et vindex rerum , 29.2.20), judging and preparing vengeance on the crimes of Valens. It is the vigilance of the Iustitiae oculus sempiternus and the sempiternus Iustitiae vigor that avenged ( vindicavit ) Africanas clades and the events surrounding the comes Romanus [ 37 ]. It is the unsleeping curses of victims and the shades of the unavenged ( inulti ) that sustains the vigilance and long memory of Vengeful Justice, aliquotiens serus sed scrupulosus , that eventually distributed deserts of some kind to all those "trapped in a cycle of violence and retribution beyond any of their powers to control" [ 38 ]. The fall of Gallus elicits a eulogy to the workings of Adrastia/Nemesis, daughter of Iustitia and guide of the universe, vigilant, ultrix facinorum impiorum and intervening aliquotiens ( atque utinam semper ) [ 39 ]. Faith in a Just God or Divine Justice might encourage one to leave both punishment and revenge to this higher power, especially in a legally impaired world. On the other hand, rulers and judges can easily convince themselves that they are the terrestrial agents and faithful imitators of a force for good. An environment that breeds fatalism and helplessness in many is precisely that which favours coercion and exploitation by a few. And even Julian abused the maiestas process [ 40 ]

To take stock: some motifs emerge, the most basic of which is the desire for revenge for a slight to one's honour or status; the need to act on behalf of others, especially family members; the division between and mingling of private and public motives for revenge, which in turn can blur distinctions between right and wrong; the delaying or preventing of revenge until opportunity, such as a weakening of the enemy, presents; and, above all, the difficulty of distinguishing between punishment and revenge, poena and ultio , illustrated in a somewhat bizarre way by Ammianus's story of the woman of Smyrna, whose judges postponed her trial for one hundred years because they could not distinguish between scelus and ultio and did not want to acquit a murderer or punish an avenger ( ultrix ) of her kin (29.2.19).

As already noted, both authors have a number of instances where pardon or revenge is explicitly not sought or gained. These were worth including in the sample because of the circumstances, motives and authorial or participant comments they provide. To some extent they are the flip side of each other, since revenge not sought may amount to a pardon, or pardon not gained may amount to revenge, or at least, punishment. The rebellious Mazices of Mauretania, having unsuccessfully sought pardon from the elder Theodosius, are told a dire fate awaits them. Valens rejects the Goths' assertion that their error in supporting Procopius was pardonable and he marches against them (27.5.1-2). Corbulo as general does not want to win a reputation for leniency or clemency. His efforts to restore military discipline involves a zero tolerance of faults ( Ann . 13.35.4). Only towards victorious troops will he be merciful ( Ann . 15.35.3). The commander Didius Gallus used mercilessness as a way of striking dismay into Armenians ( Ann . 12.17.1). Emperors can set examples of restraint and curb others' thirst for revenge. Nero in his early days as emperor not only says he has no wish or motivation for revenge, but prevents Suillius's accusers targeting his son ( satis expleta ultione , Ann . 13.43.5). Tiberius prevents senators from exacting further vengeance on Sejanus's supporters and on the family of Piso. Otho says he has no wish to take vengeance on Vitellius's relatives. The Roman mob, however, complain that they cannot exact vengeance from Otho for his attempted coup against Galba.

As we have noted earlier, people are often expected to want revenge but if they are in no position to exact it, no great disgrace is incurred. Different is the case of the senator Titius Labeo who failed to punish/take vengeance on his wife Vistilia for advertising herself as a prostitute. The senate sought an explanation from him and exiled her ( Ann . 2.85.1). Dolor iniuriarum and libido vindictae urge Claudius to capture Mithridates by force rather than accept him as a captive from Eunones, to whom Mithridates had surrendered. He would then be able to celebrate a triumph and thus more effectively expunge the narcissistic wound that Mithridates's rebellion had delivered. Tacitus suggests that it is Claudius's rather than Rome's honour that is at stake here. When the costs and risks of such a course were pointed out to Claudius, to avoid imputations of weakness and becoming an object of scorn, he resorted to inflationary bluster about himself and to disdain and belittlement of his foe: he had the power to execute Mithridates but Rome traditionally spared suppliants, he avowed. And in sour grapes vein, he pointed out that triumphs were won over rulers or peoples who were still undefeated. Such transparent manoeuvring to salve his wounded honour deserved a comeuppance and Tacitus cannot resist supplying the episode's conclusion. Whereas Mithridates had conducted himself with due humility in asking mercy from Eunones, before Claudius he showed various forms of defiance and arrogance, including a remark that became common currency: he had not been delivered to Rome, he had returned, and if Claudius doubted it, let him go and see if he could be caught. Mithridates refused to inflate Claudius's honour and diminish his own. What Claudius feared thus came about. He became an object of scorn but, because he had already committed himself to clemency, he could do little about it ( Ann . 12.20-21).

Still on cases of failed vengeance, Ammianus reports willingness by soldiers to take vengeance on Persian troops for their fatal attack on Julian but they are simply unable to enact their wish. Goths are dissuaded from trying to avenge themselves on Romans. As a final illustration of how one concept may be shaped by the absence of the other, Julian in effect pardons pagans who murdered a Christian bishop by failing to punish them.

IV

In conclusion: incidents which one might see as simple retaliation for aggression have been brought into our sample by the use of a revenge word. Such words occur twice as frequently in Tacitus than in Ammianus. This may owe something to personality features. If, as Maranon has argued, Tacitus was indeed full of anger, frustration and resentment, his thoughts may naturally have gravitated to this theme [ 41 ]. If one accepts that Resentful Man becomes Vengeful Man, that Tiberius is a case study in resentment and delayed vengeance, and that Tacitus has a psychological affinity with Tiberius, this explanation gathers plausibility. But one reason why Tacitus has a greater incidence of revenge words is simply because, in addressing primarily a senatorial audience, he reports more legal trials, where revenge vocabulary is common. As Tacitus observes, desire for revenge could easily be wrapped up in or impossible to disentangle from the desire for renown and a principled concern for justice [ 42 ]. So can the granting of pardon. Neither author sees revenge as ipso facto undesirable, although powerholders with a vindictive streak and who are capable of disproportionate responses certainly are dangerous, particularly in the Fourth Century when privilege accustomed itself to disregard restraint [ 43 ]. The search for honour is perhaps the chief driver of revenge but, as Ammianus in particular indicates, elements of sadism and bloodlust can be present too [ 44 ]. A man with Tacitus's stern sense of honour (witness his strictures on those who dishonourably seek mercy or pardon) will find it difficult to disapprove entirely of behaviour that may restore impaired honour [ 45 ]. On one occasion he speaks of the dignitas won by achieving ultio ( Ann . 6.7.1). This, incidentally, is about as close as he comes (and it is not very close) to explicitly acknowledging that revenge can, as it commonly does, provide release from such burdens as obsession, guilt, anxiety and depression. Fervent pursuit of revenge may seem insane at times but achieving it can help restore sanity to troubled minds. While Ammianus, in particular, expresses satisfaction at the intervention of Justice/Vengeance, he, too, is reticent, in the instances thrown up by our keywords, about how revenge can transform the lives of avengers. Neither author would have been impressed to read that "resorting to vengefulness is a startling proof...of...unconscious infantile fears...(of) feelings of inferiority and smallness" [ 46 ]. Much achievement is driven by a desire to avenge oneself upon the hurts delivered by critics, scorners and doubters. Augustus did not think his pursuit of those who killed Rome's top executive in 44 B.C. was anything but a credit to his name. Fortified by a sense of justice, a punisher can act without guilt or remorse. Later, Augustus became a master of the indirect, light-handed exercise of power. The more adept of his successors remembered what an effective instrument clementia was for this gentler approach.

We have seen that granting pardon too may be impelled (or be seen as impelled) by desire for good reputation, as well as by expediency, the fears of the moment, self-preservation, and a wish to store up social credits. Granting pardon may be the cheapest and most effective way of overcoming a problem such as an insurrection. Status is a factor in judging the behaviour of vindictive or forgiving people. Winning kudos may depend on the nature of the vengeance achieved. It should be roughly proportionate to the original injury. Disproportionate revenge weakens one's claims to justice and may be no better than criminality. It increases the prospect of escalation and perpetuating a cycle. Because rulers to some extent represent the vengeful will of the community, because they have greater security problems of their own and need to help the community's security and sense of what is right by deterring crime, they may be granted more latitude in pursuing what could be seen as vengeful behaviour. Yet because of their great power, the opportunity to pardon is valued, and is often seen by rulers as another demonstration of their power. How they behave, as indicated by incidents signalled by use of words for pardon, fills out the picture of emperors conveyed by revenge words. Valentinian, who we are told sometimes valued a reputation for clemency, did not intentionally appoint harsh judges, but once having learned of their asperity, heaped praise upon them and urged them to severely punish ( vindicarent ) minor offences (30.8.13). Emperors are also in a position to win pardon at the bar of public opinion by good deeds or generosity. There is much scope for hypocrisy and charade when some form of pardon is granted. Although Tacitus is willing to ascribe motivation, his speculations are not necessarily sound and it usually impossible to know whether an injury has been consciously or unconsciously provoked in order to provide a pretext for pardon or revenge. This issue becomes acute in the case of governors or commanders (Germanicus, Corbulo) who were felt to be stirring up trouble for their own ends and thus made emperors (Tiberius, Claudius) nervous.

The world projected by Ammianus is more dependent on pardon to curb some of the excesses of power and the injustices he so frequently condemns. More than other ancient historians, Ammianus mentions the effect of power on the recipient, that is, he adopts the perspective of the object. Restraint, highly valued, depends on humanitas and aequitas . So many people were behaving like wild animals [ 47 ]. But it is difficult to be sure that feuds, vendettas, vigilantism and vengeance became more common or unrestrained in the Fourth Century compared with the early Principate, only that Maximinus's view that crimes should be vindicari suppliciis acrioribus (28.1.10) is illustrative of what so disturbed Ammianus. The authority of the law was indeed more stridently asserted but not necessarily more respected. If one could be certain of less respect for legality in the Fourth Century (and the emperors were major offenders), one would expect not only the more violent trappings of the honour code to reassert themselves but also a stronger belief in vengeful gods and in a supernatural avenging Justice and divine will which complements executive clemency and guides human affairs. Powerholders could no longer be relied upon to ensure justice. This is what we find in Ammianus, together with a nostalgia and reverence for the past where models of virtuous people who respected the old gods and resisted elite self-indulgence might be found. Too often emperors did not measure up against an ancient ideal of law and justice [ 48 ].

It was not until the Nineteenth Century in Europe that there was a decisive shift from revenge to rehabilitation and correction when jurisdiction dealt with wrongdoers, and the transition is far from complete even in judicial circles today [ 49 ]. The control of anger that is often necessary to plot revenge can also be cultivated and built in as the self-control that permits pardon to be given. Hence Seneca, while acknowledging the dangers to one's reputation of not repaying injuries, claims that those able to exact vengeance but who do not are praised ( Clem . 1.7). Self-control offered an alternative path to honour to that of immersion in cycles of revenge. The premium placed upon self-control amongst aristocrats may explain the comparative rarity of duels, vendettas and blood-feuds in our and other ancient authors writing under the empire and in an otherwise typically honour-obsessed, post-archaic, pre-modern society [ 50 ]. There is, however, more aristocratic feuding, by means of prosecutions and delation, in Tacitus than Ammianus. In seeking revenge the desire is for acknowledgment of one's power, superiority and judgement. Just such a desire may motivate pardon. Both behaviours could betoken largeness of spirit.

Comprehensive analysis and prolonged illustration of the satisfactions and drawbacks of pardon and revenge are suitable subjects for treatises, plays and novels. While not eschewing dramatic episodes, Tacitus and Ammianus work on a wide canvas and confine themselves to examples (the trial of Cn Piso is an extended instance), brief comments and implicit illustration. In their use of revenge words they do not try to follow Aristotle and show that revenge is what an injured party seeks from an equal, punishment is what he seeks from an inferior: or Plato's distinction between justice (rehabilitative and reconciliatory) and vengeance (an act of passion, hence innately unjust). Two characteristics that are often emphasised in studies and stories of revenge are the frequent irrationality and unprofitability of pursuing it. One might expect to find some mention of these characteristics in Ammianus, whose pages are so full of frenzied violence and excess. One does not find it, and the reason may lie in the fact that of 17 cases where revenge is sought, in 14 cases persons or groups of greater status and/or strength (numbers can make up for deficiencies in status) seek it from inferiors. In three cases the protagonists appear equally matched. At no time does an inferior seek it from a superior. When an emperor, say, exacts vengeance from an inferior, it may be tyrannical but it does not disadvantage him immediately, although his reputation may suffer. In Tacitus, in 49 instances where a judgement about relative strength and status can be made, 28 times superiors seek revenge from inferiors and 11 occasions they are equal. But on 10 occasions inferiors seek it from superiors. Either this is behaviour so crazy Tacitus does not need to spell it out, or else, living in a world where hierarchy was less strong and opportunities for success more promising, pursuit of vengeance from superiors did not seem that irrational. It was more likely to occur with people who plot against the emperor, or rebels who rise against Roman rule, such as Boudicca's Britons, where the pain of oppression fires vain hopes, and where Tacitus is moved to comment on their disregard for the retribution that had to come ( Ann . 14.33.2; cf. Ag . 31.3). Arguably, there is in Tacitus less emphasis on the vulnerability of those without power [ 51 ]. Another common feature of the pursuit of revenge in some societies is the willingness of the seeker to use or sacrifice others in order to harm the intended victim. The clearest example from our authors is perhaps Junia Silana's use of two clients, Iturius and Calvisius, and a freedman of Domitia, Atimetus, to destroy Agrippina for depriving her of a desired marriage. "Evidence" was provided that Agrippina was plotting to depose Nero. Failure of the scheme meant exile for Silana but it also meant relegation for her clients and death for Atimetus ( Ann .13.19-21). Rulers and commanders can be seen as sacrificing the lives of their troops in order to obtain vengeance in the field but otherwise use of others to achieve revenge is not a feature of our authors' accounts.

To sum up: there are grounds for thinking that some of the differences in our authors' treatment of pardon and revenge stem from personality factors as well as changed environments. Episodes flagged by our elastically-used keywords contain a wealth of opinions and issues on an area of human behaviour that exercised Roman minds as much as it does ours. Since it is mainly the public arena they treat, it is relations between peoples and between governors and governed that the references mainly deal with. The motivation behind acts of pardon and revenge can be difficult to determine and disentangle from displays of power or the pursuit of justice.

R.F. Newbold
Centre for European Studies and General Linguistics
University of Adelaide

Notes

[1] T. Barnes, Ammianus Marcellinus and the Representation of Historical Reality (Ithaca 1998); L. Roselle, Tacitean Elements in Ammianus ( U. Columbia Diss.1976); J. Matthews, The Roman Empire of Ammianus (London 1989) 482-483. Just as Ammianus's contemporary, the author of the Augustan Histories, was a continuator of Tacitus's contemporary, Suetonius, so does study of pardon and revenge in both their works provide insights that are useful for this study. See R. Newbold, 'Pardon and Revenge in Suetonius and the Historia Augusta ', Prudentia 33 (2001) 40-57.
[2] See L. Sebba, 'The Pardoning Power - a World Survey', Journal of Criminal Law and Criminology 68 (1977) 83-121; id., 'Clemency in Perspective', in Criminology in Perspective: Essays in Honour of Israel Drapkin edd. S. Landau and L. Sebba (Lexington 1977) 221-40. New forms of punishment, e.g., transportation, often originate in an act of clemency such as commutation of the death penalty. A commonly accepted use of clemency nowadays is for miscarriages of justice. Paradoxically, pardon is often granted to the innocent (as well as model prisoners and to those showing evidence of reformed character). For discussion of the Latin terms mentioned in the text, plus others such as abolitio and liberalitas , see H. Cotton, 'The Concept of Indulgentia under Trajan', Chiron 14 (1984) 245-266, and W. Waldstein, Untersuchungen zum römischen Begnadigungsrecht . Abolitio, Indulgentia, Venia (Innsbruck 1964). Both scholars bring out how differently authors and jurists use the terms, and how their meanings change over time. T. Adam, Clementia Principis , (Stuttgart 1970) focuses on Tacitus at 105-107. For pity and its relationship to pardon and clemency, see D. Konstan, Pity Transformed , (London 2001) 38,101. Konstan argues that pity is more an unstable, ephemeral and spontaneous emotion whereas clemency is more a disposition towards mercy and forgiveness, a virtuous character trait, an admirable form of restraint, temperantia . See too E. Aubrion, 'Tacite et le misericordia ', Latomus 48 (1999) 383-91. There is a discussion of lenitas/lenitudo in Ammianus in A. Brandt, Moralische Werte in den Res Gestae des Ammianus Marcellinus , (Göttingen 1999) 189-193. For Ammianus it is an executive virtue important for minimising possible injustices. In discussing misericordia , Brandt distinguishes between the bestowal of benefits ( beneficia erogare ) and the forgiveness of wrongs ( peccata ignosci , 193-197).
[3] Let one example suffice for the moment. Seneca De Clementia 2 carefully distinguishes clementia , an imperial virtue, from respectable synonyms for it, misericordia , venia , ignoscere . For Stoics, misericordia was an attribute of the mob and a sign of weakness, and was undesirable because it made inner well-being susceptible to external influences. At Ann . 3.50.2, the trial of Clutorius Priscus, Tacitus clearly uses misericordia for imperial clemency. Seneca's distinction between clementia (involving the use of reason) and venia (like misericordia , an irrational leniency) simply does not work for Tacitus and Ammianus. See M. Griffin, Seneca . A Philosopher in Politics (Oxford 1976) 154-71, for Seneca's discussion of severitas , clementia , and respective synonyms. Seneca moulded clementia into something more than mercy or leniency. It could mean complete exoneration. Griffin points out that the reasons for clementia most emphasised by Seneca, curability, age, dignity, and the glory of mercy, derive from rhetorical arguments used in deprecatio (164).
[4] Ability to truly forgive may require a capacity for self-forgiveness, not an idea easily found in ancient thought on the subject. Responses to hurtful behaviour can include sadness, disappointment and scorn. The last may serve as an acceptable alternative to resentment and revenge. Ignoring one's wounder may mortify him/her, forgiving them may mortify them even more. On forgiveness see L. Smedes, Forgive and Forget (San Francisco 1984); J. Murphy, 'Forgiveness and Resentment', Philosophy 7 (1982) 503-16; on some of the nuances of the term, A. Kolnai, 'Forgiveness', Proceedings of the Aristotelian Society 106 (1973-4) 91-106; and on some of the issues involved, N. Richards, 'Forgiveness' in Ethics and Personality ed. J. Deigh, (Chicago 1992) 223-43; S.Cherry, The Coherence of Forgiveness . An Essay on the Theology of Being Forgiven and Forgiving Others (U. London Diss. 1995).
[5] Hence the title of works such as E. Ayers, Vengeance and Justice . Crime and Punishment in the Nineteenth Century American South (Oxford 1984). Cicero Off . 1.88 says that magistrates resorting to stern measures against others need to put aside scorn and anger, and focus on the interests of the state. Clementia must give way to severitas when the interests of the state are involved. (Such a stance, we point out, permits guilt-free aggression). The point he makes about penalties needing to be proportionate to the crime also applies, in popular morality, to the relationship between injury and revenge. As G. Thome, 'Crime and Punishment, Guilt and Expiation: Roman Thought and Vocabulary', Acta Classica 35 (1992) 73-92, points out, the initial intention in Roman arbitration procedures was reparation and just recompense, not punishment.
[6] The Weekend Australian , 5/6th December, 1998, 1. To provide an early example of what will appear frequently, when Ammianus refers to Valens's zeal in punishing ambitious intrigues ( ambitiones , probably those directed against himself), he calls him an ultor (31.14.2).
[7] NA 7.14.3 (cf.7.14.9), cited and discussed by P. Garnsey, Social Status and Legal Privilege in the Roman Empire (Oxford 1970) 1: "Roman theories of punishment were characterised by a respect for status (of which honour and prestige were attributes)".
[8] A. Burnett, Revenge in Attic and Later Tragedy (Berkerley 1998) 1 n.1, cites the definition of Samuel Johnson in 1773: "Revenge is an act of passion, vengeance is justice". She notes that the distinction is not always observed. It is not in this article. J. Kerrigan, Revenge Tragedy (Oxford 1996) 17, neatly resorts to etymology to bring out the kinship between justice and revenge. Both need to be seen to be done. "Symptomatically, deik -, the Indo-European root of 'revenge', 'vendetta', and 'vindicate also gives rise to Greek deiknunai , 'to show', Latin dicere , 'to say' and dicare , 'to proclaim', Old English tacen , 'sign, token', tacnian, 'to signify', and Old French, tache , 'mark, stain'".
[9] T. Scheff, Bloody Revenge (Boulder 1994) 67.
[10] C. Socarides, 'On Vengeance: the Desire to Get Even', Journal of the American Psychoanalytic Association 14 (1966) 356-75, at 357. Socarides brings out how subtle and complex the causes of vengeful feelings can be. He continues, "Whether he feels and acts from the conviction he is engaged in 'just retribution' or 'malicious retaliation', the clinical picture is identical". Cf. J. Elster, 'Norms of Revenge', in Deigh [n. 3] 155-78. After discussing how revenge can be variously evaluated in different societies, Elster poses questions that can to some extent be answered from the narratives of Tacitus and Ammianus: what constitutes an affront that must be avenged? Who are allowed to exact revenge? What means can be legitimately used to take revenge? Whose death shall expiate an affront? The question, what is the fate of those who fail to exact the revenge the social norm requires, is answered more implicitly than explicitly: emperors were expected to respond to rebellion or invasion, Germanicus's friends could not have got away with not prosecuting Piso. The question, how soon is revenge allowed or required to take place, is not answerable here.
[11] Notably by G. Maranon, Tiberius . A Study of Resentment (London 1956), with discussion and examples of Tiberius's long-delayed vindictive behaviour.
[12] R. Castro, Tacitus and the "Virtues" of the Roman Emperor: the Role of Imperial Propaganda in the Historiography of Tacitus (Indiana U. Diss. 1972) 364, 371. Clementia was advertised by every First Century emperor except Otho, and was especially important in the propaganda of Tiberius, Nero and Vitellius. Tacitus was particularly anxious to rebut Tiberius's claims to be a practitioner of clementia . It emerged into the political vocabulary during the Caesarian civil wars. C. Wirszubski, Libertas as a Political Idea at Rome during the Late Republic and Early Principate (Cambridge 1960) 150-53, sees Cicero's Pro Marcello and Pro Ligario as the moment when the idea that a citizen's rights were guaranteed only by law ended. Henceforth, the will of the ruler or victor decides, as Seneca later accepts. Whereas Cicero had distinguished the good and bad ruler by the criterion of justice, Seneca uses clemency as the standard. Seneca said that great power involves great restraint ( Clem . 1.8). Subjects can now rely only on that restraint. "Seneca's De Clementia represents its ( libertas in legibus consistit ) final collapse" (151). Seneca used his tragedies to explore the dynamics of revenge.
[13] As Tacitus does in relating Tiberius's conduct of the trial of the elder Vibius Serenus in 25 ( Ann . 4.28-30). Tiberius was supposedly driven by a grudge originating in Serenus's complaints about the trial of Libo Drusus in 16 ( Ann . 4.29.3). No revenge word, however, is used here. It is worth noting that the reason Tiberius forced the prosecutor to resume a case he had abandoned was to discourage irresponsible use of legal machinery to gratify private feuds: cf. B. Walker, The Annals of Tacitus (Manchester 1952) 103.
[14] S. Jacoby, Wild Justice . The Evolution of Revenge (New York 1983) 1-9.
[15] The psychological term used to describe the blow to self-esteem that an injury may inflict. It can thus be not only a gratuitous insult to one's honour but also defeat in battle or contest.
[16] D. Epstein, Personal Enmity in Roman Politics, 218-43 B.C. (London 1987), esp. 2, 9, 84, 115. The joys of revenge upon his old enemy Pompey could outweigh the pleasure Lucullus took in his fishponds. Revenge against those responsible for the death of his brother was the driving force of Caius Gracchus's career. Insofar as Epstein's book is a study of revenge at work in the Republic, one notices differences from events reported by Tacitus and Ammianus, e.g., Metellus Celer's thirst for revenge on Pompey for the disgrace he brought to the Metellan family by divorcing Mucia in 62 is not the sort of pretext they record. M. Dowling, 'The Clemency of Sulla', Historia 49 (2000) 303-340, despite its title is more about Sulla's vengefulness and cruelty.
[17] Nero clearly sees it this way after the revelation of the Pisonian conspiracy; ultio privata versus vindicta publica , as Velleius puts it when recording criticism of the anti-Gracchan conduct of the consul Opimius, 2.7.6. Cf. Val.Max. 4.2.1.
[18] Especially if it is true that slave owning encourages impetuousness and short temper, that slavery provides a hothouse for breeding a concept of honour based on lack of self-control. See Ayers [n.5], 10, 24, and illustration of how revenge-driven killings can enhance one's standing in an honour-obsessed society. Cf. J. Lendon, Empire of Honour . The Art of Government in the Roman World (Oxford 1997).
[19] Where clemens refers to climate, speed or gradient, it was not counted. The Bude text of the Annals and Histories comes to 892 pages. The last and sixth volume of the Bude Ammianus has not yet, to my knowledge, appeared, and so a calculation had to be made to get a putative total of 783 pages for Ammianus, based on the 648 pages filled by Books 14-28 in the first five volumes.
[20] Hist . 4.72-73. Cf. the discussion by Lendon [n.18] 243.
[21] Hist . 2.30.3. Cf. Nabdates, burnt alive by Julian after the taking of Pirisabora. Having abused an unhoped-for pardon for an earlier offence, he then reviled Ormisda with every kind of abuse, 24.5.4.
[22] There was a standard visual representation at work: the emperor extending a right hand towards a kneeling figure, such as one sees on the Boscoreale cup (Augustus), on the column of Marcus Aurelius, and on coins. On the iconography of imperial clemency in the early Principate, see R. Brilliant, 'Gesture and Rank in Roman Art', Memoirs of the Connecticut Academy 14 (1963) 74-75, 90, 122, 147; for the late empire, 171 ("the bestowal of clemency...the charismatic right of victory"), 192-93.
[23] For a study of the ritual of supplication, see G. Koziol, Begging Pardon and Favour. Ritual and Political Order in Early Medieval France (Ithaca 1992).
[24] See R. Newbold, 'Nonverbal communication in Tacitus and Ammianus', Ancient Society 21 (1990) 189-200.
[25] Eighteen months to comply with the law in the case of the usurious senators, Ann . 6.16.3. Seneca would say such conditionality was granting mercy, clementia , rather than pardon, venia , Clem . 2.7. Cf. the description of P. Plass, The Game of Death in Ancient Rome (Wisconsin 1995) 95, of clementia as "suspended coercion". The conditionality of a pardon may, of course, be unspoken. ("You know what you have to do"). There were situations, such as pardoning mutinous troops during civil war, where it was pointless to think in terms of "suspended coercion", unless one confidently entertained the prospect of a day of reckoning.
[26] Not much of an achievement. See D. Quint, Montaigne and the Quality of Mercy (Princeton 1998) 35. The Sixteenth Century French essayist Montaigne's discussion of pardon and revenge has much to offer the student of Greece and Rome. He had a good deal to say about the role of demeanour in obtaining mercy and challenged the contemporary aristocratic code in pre-modern France whereby honour hindered seeking mercy. He also saw the competitive value of bestowing it.
[27] Hist . 3.31.1, Ann . 14.23.1. Tacitus is more likely to favour compassion for honestiores , rigour for freedmen, slaves, soldiers. Or else, leniency to the compliant, severity to the intransigent in time of war in time of war. See K. Gilmartin, ' Corbulo's Campaigns in the East', Historia 22 (1973) 583-626, at 600. It is relevant to note that a word which can sometimes mean pardon, indulgentia , is always used negatively in Tacitus, a granting of what is not earned or claimable as a right: M. Morgan, ' Indulgentia in Tacitus' Studies in Latin Literature and Roman History vol.ix, ed. C. Deroux, (Brussels 1998) 411-424.
[28] On pre-Neronian clemency and on Nero's initial clemency, see Griffin [n.3] 123-50. Clementia was one of the virtues on Augustus's clipeus virtutum , and a virtue Seneca sees as part of Augustus's general self-control. Whereas for Cicero clementia was a negative conception, based on the absence of anger, arbitrariness and excess, for Seneca it had the potential to reform as well as deter plotters and ensure security. On the ploy of posthumous pardons for suicides, or provoking capital charges in order to offer it, see Plass [n.22] 113. Plass suggests that clementia towards free men corresponded to the incentive part of the carrot and stick to controlling slaves.
[29] 16.1.4. In effect, Julian reasserted the humanitas and clementia of an earlier age over his age's poenae metus , even if his age still proclaimed clementia as an imperial virtue and hailed emperors as tua clementia . See Brandt [n.2] 186, 189, 198, on the clemency of Julian and Pius; R. Baumann, Crime and Punishment in Ancient Rome (London 1996) 160. It could, of course, be all three of these. Konstan [n.2] 99 asserts that there is no evidence that Julius Caesar's contemporaries found his clemency unwelcome or demeaning.
[30] J. Harries, Law and Empire in Late Antiquity (Cambridge 1999) 152.
[31] According to Harries [n.30], 136 n.7, the phrase occurs nine times in the Codex Theodosianus
[32] "This classic story of vengeance, in which, by a series of reversals, the victor is transformed into the victim, is typical of ancient historiography in general and is presented by Tacitus in his most ordered and effective manner". This sentence is part of the introductory essay ('Vengeance for Germanicus: Piso's Return and Trial') by A. Woodman and R. Martin, to their commentary on Ann . 3.7-19, in The Annals of Tacitus. Book 3 (Cambridge 1996) 113.
[33] J. Evans, 'The Trial of P. Egnatius Celer', CQ 29 (1971) 198-202; E. Keitel, 'Speech and Narrative in Histories 4', in Tacitus and the Tacitean Tradition , edd. T. Luce and A. Woodman (Princeton 1993) 39-58.
[34] As, to take an example from outside our sample, Caligula did not fear the anger of the regularly insulted Cassius Chaerea. Nero obviously thought he could insult his poetic rival Lucan with impunity. Lucan's desire for revenge led him to join the Pisonian conspiracy.
[35] Ann . 14.33.2. Cf. Sen. Clem . 1.26: slaves avenge themselves on cruel masters even knowing they will be crucified.
[36] Ann . 3.18.2. Cf. Woodman and Martin [n.27] 113.
[37] 8.6.26, 30.2.9; cf. 16.12.52, and F. Paschoud, 'Justice et Providenz chez Ammien Marcellin' Studi Tardoantichi vol I, ed. S. Calderone, (Sicania 1986) 139-161.
[38] See Matthews ([n.1] 387) on events in Africa.
[39] 14.11.24. Hence it may be possible to speak of a greater contrast between universal, just law (cf. 21.13.13) and arbitrary power in Ammianus than in Tacitus. See R. Rike, Apex Omnium . Religion in the Res Gestae of Ammianus (Berkerley 1987).
[40] On how the culture of the age may have shaped attitudes, see R. Newbold 'Authoritarianism, Autonomy and Ammianus Marcellinus', in The Imperial Muse , ed. A. Boyle (Bendigo 1990) 261-273. On Julian's legal irregularities, see A. Demandt, Zeitkritik und Geschichtsbild im Werk Ammians (Bonn 1965) 50-56.
[41] Maranon [n.11]; G. Tanner, 'Tacitus and the Principate', Greece and Rome 16 (1969) 95-99; R. Mellor, Tacitus (London 1993) 18-19; J. Lucas, Les Obsessions de Tacite (Leiden 1974) 180-182. Ag . 1-3 alone would explain some of this attitude.
[42] Hist . 4.6.1. As Socarides [n.10] 367 puts it: "The avenger attempts to masquerade as his victim's superego". Tacitus comes close to Socarides's sentence when he has Cerealis tell rebel Batavians that if they transgressed further, iniuria and culpa would be theirs, vengeance and divine backing would belong to the Romans ( Hist . 5.24.2).
[43] A major theme of Burnett's book [n.8] is that Greek mythology and Attic tragedy remembers vengeance as "an honourable imperative, essential to the preservation of order" (6; cf. 32) in the times of embryonic, rudimentary states, as are oaths that invoke divine vengeance upon transgressors. The host societies of Tacitus and Ammianus are intermediate between such primitive states, and modern states who see the public prosecutor as an important figure in trying to preserve for themselves the right to pursue vengeance for serious crimes.
[44] For factors involved in judicial savagery, see D. Garland, Punishment and Modern Society (Chicago 1990) 97.
[45] Cf. the approval of Cicero and the younger Pliny for readiness to avenge assaults upon honour ( dignitas ). See R. McMullen, 'Personal Power in the Roman Empire', in Changes in the Roman Empire (Princeton 1990), 190-97, esp. 192-93. Deliberately provoking a feud to generate occasions for winning honour is quite another matter, but societies which encourage such behaviour are not unknown. See Elster [n.10] 169, and, on the Roman Republic, where it was honourable to have inimicitiae , Epstein [n.16] 2, 65; in Greece, K. Dover, Greek Popular Morality (Oxford 1974) 180-184; in Nineteenth Century southern USA, Ayers [n.5] 16.
[46] Socarides [n.10] 367.
[47] Cf. R.Seager, Ammianus Marcellinus . Seven Studies in his Language and Thought (Columbia 1986) 131-133. The evidence for an object-perspective on Ammianus's part derives from an unpublished study of power motivation in ancient historians, referred to in Newbold [n.40], 265, nn.28, 39.
[48] Newbold [n.40] 266. On Ammianus's nostalgia, see further T. Barnes, 'Literary Convention, Nostalgia and Reality in Ammianus Marcellinus', in Reading the Past ed. G. Clarke et al . (Sydney 1990) 59-92. Ammianus's references to divine vengeance are more elaborate and less sceptical than Tacitus's ( Hist . 1.3.2, 4.3.2, Ann .3.17.1). Barnes speaks of Ammianus "fixing his emotional fervour on transcendental forces", 73. Ammianus may thus illustrate what Thome [n.4] 92 sees as the trend to intensification and moralisation in Roman jurisdiction. But Thome argues that this trend, which one sees manifesting in Christian thought and behaviour, had earlier received a decisive boost from Seneca. Tacitus gets close to the idea of vengeance as a ritual owed to divine powers when he has Germanicus's troops speaking of the need to sacrifice the treacherous Cherusci to vengeance and glory, ultioni et gloriae mactandos ( Ann . 2.13.1).
[49] See Garland [n.44] 136; M. Foucault, Discipline and Punish trans. A Sheridan (New York 1977/1995). Just as the language of revenge coloured litigation at Rome ( vindex, vindicare ), so did it in Greece, where both plaintiffs and judges could be called timõrountes . Trials for violent crimes at Athens were held on the last three days of the month, which were sacred to the Semnae, the powers of revenge. See Burnett [n.8] 54.
[50] Cf. Lendon [n.18] 41, and Tacitus's tendentious account of how law developed at Rome and operated in the early Principate, Ann . 3.25-28. But perhaps some plots against emperors should be seen as endemic, irresistible feuding. See D. McAlindon, 'Senatorial Opposition to Claudius and Nero', AJP 72 (1956) 113-32, who paints a picture of noble families pursuing inherited enmities and feuds with the Julio-Claudians. Whereas private vengeance usually dies with the individual, group vengeance is more likely to endure.
[51] R. Anderson, The Rise and Fall of Middle Class Loyalty to the Roman Empire: a Social Study of Velleius Paterculus and Ammianus Marcellinus (U. Cal. Berkerley Diss. 1962) 113, sees this vulnerability as Ammianus' chief topic.

Exculpatory thanks are due to Don Lateiner for his comments on an earlier version of this paper. Thanks too to the anonymous referees.