DANIEL KLERMAN

Settlement and the Decline of Private
Prosecution in Thirteenth-Century England

DANIEL KLERMAN

Although modern societies generally entrust enforcement of the criminal law to public prosecutors, most crimes in premodern societies were prosecuted privately. In classical Athens, ninth-century Germany, and England before the nineteenth century, there were no public prosecutors for most crimes.¹ Instead, the victim or a relative initiated and litigated the cases. This article is the first rigorously quantitative analysis of private prosecution. It focuses on thirteenth-century England and uses statistical techniques, such as regression analysis, to show that changes in the treatment of settled cases can explain the rate of private prosecution.	1
Charting and explaining the changing rate of appeals is important for both legal and social history. Appeals have always occupied an important place in the history of English law,² yet their long-term decline has never been satisfactorily analyzed. For social historians, understanding private prosecution is important because private prosecution put awesome power in the hands of ordinary individuals: the power to accuse others of crime and thus to set in motion the coercive powers of the criminal law, including the possibility of pretrial imprisonment, outlawry, fines, and hanging.	2
More generally, because statistical analysis is rare in legal historical scholarship, it is hoped that this article will show that quantitative methods can provide new insights into old puzzles. In addition, because private prosecution was common in many premodern societies and remains a subject of theoretical debate among contemporary scholars, a thorough examination of thirteenth-century private prosecutions has relevance not only to English legal historians, but also to historians of other legal systems and to modern criminal procedure scholars. Finally, although the importance of settlement to the resolution of disputes has been widely recognized in both modern and historical scholarship, this article is one of the few that focus on settlements between victim and accused in the context of criminal cases.³	3
In medieval England, private prosecutions were called "appeals." Unlike modern appeals, they were unrelated to the correction of legal errors. To "appeal" simply meant to prosecute. Although appeals continued to be brought until the early nineteenth century, their heyday was the late twelfth and early thirteenth centuries. By the end of the thirteenth century, relatively few criminals were prosecuted by appeal. This article focuses on the thirteenth century in order to understand the appeal during the period when it was most important and in order to explain why it became so marginal.	4
The substantive contributions of this article lie primarily in two areas: accurate charting of the trends in the number of appeals and a new explanation for the decline of the appeal.	5
Part 1 provides background information. Part 2 reports the results of the only systematic, quantitative study of the appeal so far attempted. It reveals large, previously unnoticed changes in the frequency of appeals. The rate of appeals fell by 50 percent between 1200 and the 1220s, climbed back to turn-of-the-century levels by the late 1240s, and then swiftly dropped by two-thirds and remained at a low level through the end of the century.	6
Parts 3 and 4 try to explain why the rate at which appeals were brought varied so much over the thirteenth century and why the overall trend was decline. The most plausible explanation for the wide fluctuations is the changing judicial treatment of private settlements. One of the victim's motives for bringing an appeal was the utility of suit in facilitating monetary settlement. Such settlements were attractive to victims because there was no routine royal remedy by which they could get monetary relief for personal injury or property damage until the mid-thirteenth century. Settlement was attractive to the accused, however, only if it protected him from further prosecution. In the late twelfth and early thirteenth centuries, settlement almost always protected the appellee, because judges let the appellee go free without trial if the appellor was unwilling to prosecute. At various times during the thirteenth century, however, judges sent appellees to jury trial even though the appellor was no longer interested in the case. The implementation and relaxation of this antisettlement policy can account for most of the changing frequency with which appeals were brought.	7
Changes in judicial respect for settlement are the most plausible explanation for changes in the rate of appeals. Section 3.A, however, discusses four other explanations that have been suggested in the literature: (1) the appeal's archaic nature, especially the use of trial by battle; (2) judicial hostility, which manifested itself in the ease with which appellees could exploit technical defects to quash appeals; (3) the introduction of presentment,⁴ which meant that crimes might be prosecuted even if the victim did not appeal; and (4) the introduction of trespass actions, which were more attractive to victims because they provided money damages. Part 4 also discusses three additional alternative explanations that have not appeared in the published literature but have been suggested to me by other scholars: (1) appeal rates may have been influenced by crime rates; (2) appeal rates may have mirrored general trends in prosecution, especially trends in presentments of crime; and (3) appeal rates may have been influenced by the possibility of settlement before initiation of an appeal.	8
While most appeals were brought by the victim or the victim's family, there was a special kind of appeal that was brought by a convicted criminal who had already been sentenced to hang. If the convicted criminal successfully appealed several of his accomplices, his life would be spared. Criminals who were appealing their accomplices were called "approvers."⁵ This article focuses exclusively on nonapprover appeals for two reasons. First, because the prosecutor was a convicted felon seeking clemency rather than a victim or relative seeking retribution or settlement, approver appeals were so different from ordinary appeals that there is little to be gained from studying the two together. Second, the majority of approver suits were heard in gaol (jail) delivery, and, as is discussed in Section 2.E, very few gaol delivery plea rolls (records) have survived. Thus, it would be very difficult to perform a meaningful quantitative analysis of approver appeals.	9
Private prosecution, mostly of minor offenses, could also take place in local, nonroyal courts, and such prosecutions may even have been called "appeals." Nevertheless, because appeals in royal courts were most important for the development of the common law, and because the records of such cases have survived in greater abundance, especially for the early thirteenth century, this article focuses exclusively on the royal courts.	10
Part One: Background
Section 1.A situates the late twelfth and thirteenth centuries in the context of the broader history of criminal prosecution. Section 1.B enumerates the offenses for which appeals were brought, while Section 1.C explains the procedure for bringing and trying an appeal. Section 1.D discusses the terms and frequency of settlements, and Section 1.E briefly describes the social context of appeals. Four cases that illustrate various aspects of the appeal appear in Sections 1.C, 1.D, and 1.E.	11
A. A Very Short History of Criminal Prosecution in England
A brief glance at the broader history of criminal prosecution may help to put this article in its proper context. For the purposes of this section, it is useful to divide English history into four periods.	12
1. The first age of private prosecution (seventh to tenth centuries). During this period criminal prosecutions were almost entirely private. Prosecution was at least partially motivated by the possibility of monetary compensation. Until at least the late tenth century, those convicted of crime were not ordinarily hanged, incarcerated, or otherwise punished, but instead owed the victim compensation (bot) or, in homicide cases, owed the victim's family the deceased's wergild, a monetary payment that varied with the deceased's social status.⁶	13
2. The rise of presentment (tenth to fourteenth centuries). Starting in the late tenth century, Anglo-Saxon kings began to change the nature of criminal prosecution. Aethelred's third code, promulgated around 1000, required the twelve leading thanes (nobles) of a wapentake (district) to accuse and arrest those suspected of crime in their locality.⁷ This procedure seems to foreshadow presentment, which, according to some historians, did not became a routine part of judicial administration until almost two centuries later, during the reign of Henry II. Under the presentment procedure, leading men were chosen from each locality and were required to present (that is, report) on oath crimes committed in their neighborhoods. These leading men were known as the presenting jury, which is the ancestor of the grand jury. Like the medieval trial (petit) jury, the presenting jury was self-informing.⁸ Little or no evidence was presented in court. The jurors were expected to gather information informally before they came to court and to present their conclusions to the judges.	14
The nature of criminal penalties also began to change during this period. As early as the late tenth century, bot seems to have been payable to church, king, or community at large rather than to the injured kin.⁹ There is also archaeological evidence that the death penalty was frequently imposed in the eleventh century.¹⁰ By the late twelfth century, these changes were firmly entrenched and are regularly attested to by the surviving records. Hanging and fines payable to the king were the only criminal penalties regularly imposed in royal courts. In addition, hanging was usually accompanied by forfeiture of land and chattels.	15
Although presentment and noncompensatory punishments were becoming increasingly important, no English king even attempted to abolish private prosecutions, which by the late eleventh century were called "appeals." In fact, until the turn of the fourteenth century, presentments were confined almost exclusively to homicide and theft,¹¹ and nearly all accusations of rape, mayhem,¹² wounding, false imprisonment, assault and battery were brought by way of appeal, as were large numbers of homicide and theft cases. Although the legal sanction for crime was death or fines payable to the king, victims (and their families) could appeal and use the threat of legally imposed hanging or fines to induce compensatory monetary settlements. By the end of the thirteenth century, however, the appeal was becoming much less common, and presentment had become the way nearly all crimes were prosecuted.	16
3. The return of private prosecution (fourteenth to nineteenth centuries). As noted above, twelfth- and thirteenth-century juries (both presenting juries and trial juries) were largely self-informing. During the fourteenth and fifteenth centuries, however, for reasons that have yet to be fully explained, juries became more passive.¹³ Trial juries began to rely on evidence that parties presented in court, and the presenting jury (now called the grand jury) less frequently made accusations based on its own knowledge. Instead, the grand jury primarily screened accusations made by others, declaring "true bill" of accusations ("indictments") it approved.¹⁴ Although these prosecutions were formally brought in the name of the Crown, the predominance of victim initiative suggests that they are properly classified as private prosecutions.¹⁵ Nevertheless, royal officials did provide investigative assistance. From the late twelfth century, the coroner had been gathering evidence in homicide cases.¹⁶ Justices of the peace performed a similar function for other crimes from, at latest, the sixteenth century, and possibly as early as the fourteenth.¹⁷	17
4. The age of public prosecution (nineteenth century to present). In the nineteenth century, partly in response to the growing problem of urban crime, pressure began to mount for public prosecution. Victims frequently did not prosecute because it was expensive, time consuming, and brought few benefits other than the satisfaction of revenge or justice.¹⁸ As a result, by the mid-nineteenth century, most prosecutions were private in name only, as the "private" prosecutor was in most instances a policeman. Nevertheless, public prosecution was perceived as a threat to liberty, and Parliament did not pass legislation to set up a national system of public prosecutors until 1879.¹⁹ Even this statute did not fundamentally undermine private prosecution, because public prosecutors had very limited authority.²⁰ It was only with the passage of the 1985 Prosecution of Offenses Act that England established an effective system of public prosecution, and even this legislation preserved a limited right of private prosecution.²¹ In America, public prosecution seems to have become common somewhat earlier.²²	18
As this outline suggests, the thirteenth century was a crucial transition period, the time when self-informing presentment replaced private prosecution. But the thirteenth century was only one of several important periods of transition. Private prosecution regained its dominant role in early modern times and in turn gave way to public prosecution in the last two centuries.	19
B. Offenses
The appeal could be used to prosecute a wide range of crimes, from simple assaults to rape and homicide. Table 1 lists the most important crimes in the order of their relative frequency of prosecution.	20

As the table suggests, the appeal was most commonly used for assaults, including beatings, woundings, and mayhems. Next most common was homicide (27 percent), then theft of various kinds, including larceny, robbery, and burglary, which accounted for 12 percent of all appeals. This figure, however, understates the rate at which appeals were brought to prosecute property crimes. About a third of the assault appeals also complained of the wrongful taking of property, as did a few appeals of rape and other crimes. If these accusations were added to thefts, property crimes would have constituted 26 percent of all appeals. The next most common crime prosecuted by appeal was rape. During the twelfth century and most of the thirteenth century, rapes could be prosecuted only by appeal.²³ Although one might think that in such a patriarchal society rape would be seen primarily as a wrong to the woman's father or husband, appeals of rape were brought exclusively by the victim herself.²⁴ Finally, 4 percent of all appeals were brought for a wide array of other offenses, from abduction, arson, and attempted burglary to false imprisonment, malicious prosecution, receiving outlaws, and selling the king's hawks. It is difficult to define the outer limits of offenses that could be prosecuted by appeal. An appeal required an allegation of breach of the king's peace, but (as later with trespass actions) the allegation seems to have been purely formal and without content. For 7 percent of all cases, the crime appealed is not mentioned or is specified merely as a breach of the king's peace.	21
C. Procedure
Prosecuting an appeal involved a long and complicated process that often took several years. Immediately after the crime, the victim (or the first finder in the case of homicide) was required to "raise the hue and cry," that is, to notify his neighbors of the crime by yelling out. The hue and cry brought people to the scene of the crime while the evidence was fresh and could lead to hot pursuit of the criminal. The victim (or prospective appellor) was then required to make "fresh suit" by publicizing the alleged crime in the neighboring villages and notifying the coroner.²⁵	22
The victim (or family member in homicide and some other cases) was required to initiate suit at the next county court, which met every four weeks.²⁶ Appellors could be either male or female, and appeals by women were common. More than a third of all appeals were brought by women, including almost two-thirds of homicide appeals.²⁷ Suit had to be in person. No attorneys were allowed unless the victim was incapacitated.²⁸ The appellee was then summoned to appear at the next county court. If he did not appear, he was given three more chances. If he still did not show up, he was outlawed.²⁹ An outlaw forfeited all his property, and it was a crime to feed, shelter, or communicate with him. If he resisted arrest, he could be killed without further legal process.³⁰ Eighteen percent of all appeals ended in outlawry.	23
The appellee, however, was not the only party required to show up at subsequent county courts. The appellor was expected to appear and affirm her³¹ prior accusation. If she no longer believed the accusation was true, if she had settled with the appellee, if the appellee had intimidated her into dropping the accusation, or if she simply had lost interest in the case, she might not show up or, upon showing up, might retract her accusation.	24
If the appellor remained steadfast in her accusation and if the appellee appeared in county court before outlawry was pronounced, the appellee would be "attached," that is, he would be required to find sureties that he would appear at trial. If he could not find sureties, he could be jailed pending trial. In cases of homicide, all appellees were supposed to be jailed pending trial, although this harsh rule was not always enforced. All procedural steps in county court were recorded by the coroners, the royal officials charged with preserving the king's fiscal rights and supervising the local administration of criminal justice.	25
Trial, however, could not take place in county court. The sheriff presided over the county court and, according to custom and Magna Carta, lacked the power to try appeals because they involved an allegation of breach of the king's peace. Trial was postponed until royal justices arrived to handle criminal cases awaiting trial in the countryside. As discussed more fully in Section 2.E, delegations of royal justices took many forms, but, for appeals, the most important were called "eyres." Eyres occurred approximately every four years at the turn of the thirteenth century. The intervals between eyres lengthened as the century progressed, averaging every five to eight years at mid-century and as long as twelve to twenty years at the century's end. At the eyre, the presenting jury reported all appeals to the itinerant justices. Their presentments were compared with the coroners' written records of county court proceedings to ensure that the jury was not concealing appeals. If the appellor was present and wanted to continue her prosecution, she would repeat her accusation. A female appellor would offer to prove the appeal "as the court adjudges." A male appellor, unless he was aged or maimed, had to offer to prove his appeal "by his body," that is, by battle. About 18 percent of appeals reached this stage.	26
The appellee, if present, then pled. His options were to deny commission of the crime or to put forward a technical defense, such as failure to raise the hue and cry, failure to sue at the first county court, or a divergence between the accusation in the county court (as recorded by the coroners) and the appellor's repetition of the accusation in the eyre. If the technical defense was accepted, the appeal was null. This happened in about 10 percent of appeals. If the defense was rejected or if the appellee offered no technical defense but merely denied the accusation, he would offer to prove his innocence by battle or, after jury trial became routine around 1220,³² he could "put himself on the country." Battle, however, was only an option if the appellor was a healthy, nonminor male, and even then appellees almost always chose jury trial. If accused by a woman or an aged or maimed male, the appellee was required to accept trial by jury. Jury trial in this period did not involve the presentation of evidence in court. Instead, the jury was expected to know about and perhaps to have investigated the case before trial. Before the abolition of the ordeal in 1215, appellees accused by women and nonbattleworthy males were put to the ordeals of cold water or hot iron to prove their innocence. Appellees seldom underwent ordeals, however, unless the presenting jury had previously rendered a "medial verdict" that the accusation had merit.³³ These medial verdicts, and the fact that those put to the ordeal were acquitted more than 80 percent of the time, reduced the danger that the threat of trial by ordeal might lead to extortion.³⁴	27
Appellees convicted of the most serious crimes (homicide and sometimes theft) were hanged, while those convicted of other crimes were usually ordered to be taken into custody until they offered to pay a fine or "amercement" in an amount determined individually (but probably loosely) according to the offender's wealth and the severity of the offense. Convicted offenders could also be castrated or blinded,³⁵ but such punishments were extremely uncommon.	28
It was relatively rare, however, for appeals to proceed through pleading to proof, that is, to battle, jury trial, or the ordeal. In a majority of cases (57 percent), appellors dropped their prosecution before the case reached the eyre.³⁶ One of the key legal issues, therefore, was the treatment of nonprosecuted appeals. As is discussed in depth in Section 3.B, the treatment of such cases changed several times during the thirteenth century. The judges basically had two options. Either they could acquit the appellee, or they could require the appellee to submit to trial in spite of the fact that the appeal was not prosecuted.³⁷ In the late twelfth and early thirteenth centuries, appellees were usually acquitted when the appeal was not prosecuted. By the 1250s, however, judges routinely put appellees to trial when appellors did not prosecute. When a nonprosecuted appellee was put to trial, he was sometimes said to have been tried "at the king's suit." The case below is typical of those in which nonprosecution led to acquittal.	29
Case 1 (Staffordshire 1199). Nicholas of Salt appeals Reginald son of Thomas and Richard, his brother, of [breach of the] king's peace and robbery. And Nicholas swore an oath to prosecute.³⁸ And he retracted [his appeal] and so is in the king's mercy [that is, must pay a fine]. And the appellees are acquitted. Nicholas's amercement [fine] is half a mark, by surety of Thomas of Erdington.³⁹
Nicholas appealed Reginald and Richard of robbery and then decided not to prosecute (retracted). As a result, the court acquitted the defendants. As was typical in such cases, the nonprosecuting appellor paid a small fine or amercement.
The following, particularly vivid case illustrates the practice of sending nonprosecuted appellees to jury trial. The procedurally important sections have been emphasized.	30
Case 2 (Bedfordshire 1247). John son of Benedict appealed Ivo Quarel, Osbert Cokel and Henry Wyncard in county court of [breach of the] king's peace, wounds and imprisonment, etc. And he [John] now comes and does not want to prosecute them. Therefore let him be committed to jail and his sureties, Ayltrop Balliol and Walter son of Odo, are in mercy [Fined]. And Ivo and the others come [to court]. And the jurors testify that they [John, Ivo, Osbert and Henry] have settled and they say that, in truth, the aforesaid Ivo and the others came to the property of Matthew of Leyham in Barford and fished there without Matthew's permission and contrary to his wishes. The aforesaid John came along and asked them for a pledge,⁴⁰ and the aforesaid Ivo would not give him one, but instead struck the aforesaid John in the head with a hatchet and made two wounds each three inches long down to the crest of the head. And they [Ivo and the others] beat him badly. And afterwards they took him and bound him and put him in a boat and took him from this county [Bedfordshire] to the county of Huntingdonshire to Ivo's house at Buckden. There they dragged him with a rope to a window of Ivo's solarium and forced him to break the window with an ax. And they painted the wall near the window with the blood flowing from the wounds the aforesaid Ivo had given the aforesaid John, and they dragged him through the window and set upon him a blanket and some linen saying that he had stolen them. And they raised the hue [and cry] and caused the men who responded to the hue [and cry] to understand that eighteen thieves had come to his house, and that all except the aforesaid John had gotten away. So they put the blanket and the linen on him and took him to Huntingdon and gave him to the sheriff to be incarcerated. And he remained in prison until his tithing delivered him.⁴¹Therefore let the aforesaid Ivo and the others be taken into custody. Later Ivo Quarel came and made fine for forty marks [i.e., promised to pay the king forty marks to be released from custody] by sureties Ralf Ridel [and eleven others].⁴²
In this case, John appealed Ivo and others of wounding and imprisoning him, but then told the eyre justices that he did not want to prosecute the case. The jury provides the motive for nonprosecution: settlement. Unlike Case 1 above, however, nonprosecution did not end the matter. The jurors, presumably at the prompting of the judges, reported fully what they thought happened. The jury's narrative to the judges constituted "trial" in the era of the self-informing jury. As a result of the jury's verdict, Ivo and the others were ordered to be jailed. Nevertheless, Ivo redeemed himself, and perhaps the other defendants as well, by paying a very large fine. The case is thus illustrative of those in which nonprosecuted appellees were tried, found guilty, and punished in spite of settlement.
The treatment of nonprosecuted appeals was especially important because it determined the extent to which an appellor could settle with the appellee. If the appellor's failure to prosecute resulted in the appellee's acquittal, an appellee would find it quite advantageous to settle with the appellor in return for nonprosecution. On the other hand, if appellees were put to proof even when appellors did not want to prosecute, settlement would offer appellees little benefit.	31
D. Settlement
One of the more surprising aspects of appeals is that they were often settled. The appellor simply stopped prosecuting the case if the appellee offered some compensation. The records are usually silent about the terms of settlements. In this respect, Case 2 is typical. Occasionally, the records are more forthcoming. For example, in the case reported below, a rape appeal was settled when the rapist gave the victim two acres of land. The sentences describing the settlement are emphasized.	32
Case 3 (Kent 1241). Gunora, daughter of John Gronge, appealed Geoffrey, son of William Broketherl, that he forcibly lay with her and deflowered her, etc. And Geoffrey comes and denies everything and puts himself on the country [that is, pleads "not guilty" and submits to jury trial]. And the jurors say that, in fact, the aforesaid Geoffrey lay forcibly with the aforesaid Gunora and deflowered her, because immediately afterwards she was seen by the headborough and by respectable men and women who saw that she was sticky with blood and had been mistreated. Therefore let Geoffrey be taken into custody. Later, the aforesaid Geoffrey comes and with permission [of the court] gives the aforesaid Gunora two acres of land in Mundham with their appurtenances. Therefore the sheriff is ordered to cause her to have seisin. And she retracts her appeal. She is poor [and is therefore not fined for retracting her appeal]. And Geoffrey made fine for his amercement by four marks [that is, promised to pay the king four marks] by sureties [names of sureties omitted].⁴³
After the jury returned its guilty verdict, the defendant gave the victim two acres of land. In what was clearly a quid pro quo, the appellor then retracted her appeal. In spite of the settlement, however, the appellee still paid a fairly large fine.
In Case 3 the appellor settled for land, but cash settlements were probably more common.⁴⁴ Monetary settlements reflect some continuity with the early medieval criminal law, in which, as described above in Section 1.A, monetary payments were the most common official penalty for crime. In rape cases, the appellee sometimes "settled" the case by marrying the victim.⁴⁵ Such settlements probably reflect the victim's reduced chances of finding a suitable husband. In some instances, however, settlement by marriage may be explained by the fact that, at least in some cases, it is clear that the man and woman had consensual sex, but that she thought he was going to marry her. When it became clear that he would not, she brought a rape appeal.⁴⁶ In such a context, termination of the case in exchange for marriage is not quite so jarring.	33
Sometimes settlements were explicitly endorsed by the judges. In the late twelfth and early thirteenth centuries, parties might come to court and ask for a "license to concord," that is, for judicial approval, which the judges would usually grant in exchange for a monetary payment.⁴⁷ This practice became much less common after 1218, probably because judges became more hostile to settlement. This is discussed more extensively in Section 3.B. Case 3, however, shows some continuation of this practice later in the century, in that the defendant gave the victim land as compensation "with permission" of the court. More often, as in Case 2, the jurors reported that the parties had settled without judicial approval. Such settlements often resulted in a small fine and, in some periods, led to trial of the appellee. The terms of settlements were sometimes written down,⁴⁸ although this seems to have been rare.	34
It is difficult to estimate how common settlement was. In a quarter of the cases, the rolls explicitly record whether the parties settled. Cases 2 and 3 are typical examples. More often, as in Case 1, nothing is recorded about settlement. Of the cases in which the rolls explicitly record whether the parties settled, two-thirds were settled. One could therefore plausibly estimate that anywhere between 17 and 67 percent of cases were settled. The low figure would assume that the only settled cases were those in which settlement was explicitly recorded, while the high figure extrapolates from the quarter of the cases in which the rolls record whether settlement occurred.⁴⁹ The true figure is probably close to 40 percent. Settlement and recording whether the parties settled usually occurred in cases in which the appellor did not prosecute at the eyre. It is therefore reasonable to assume that a little more than two-thirds of all nonprosecuted appeals were settled, rather than two-thirds of all appeals. Since nonprosecuted appeals constituted 57 percent of all appeals, if two-thirds of all nonprosecuted appeals were settled, then 36 percent of all appeals would have settled. Since about 9 percent of settled cases cannot be classified as nonprosecuted cases, it is appropriate to round up to 40 percent.⁵⁰	35
Appellees seem to have been sensible about which cases they settled. For about 14 percent of cases, including Cases 2 and 3, the records indicate both whether the parties settled and whether the jury thought the defendant was guilty. In these cases, guilty appellees settled 80 percent of the time, and innocent appellees settled only 26 percent. This suggests that appellees could usually predict jury verdicts and settled when they thought they would be found guilty. High settlement rates for guilty appellees might also indicate social pressure to settle when the appellee was in the wrong. Of course, since data on both settlement and guilt is available only for a small fraction of the cases, these figures should be treated with caution.⁵¹	36
The appellor's ability to extract a settlement from the appellee rested on the credibility of her threat to prosecute if no settlement was agreed upon and on the credibility of her promise not to prosecute if settlement was successfully negotiated. If appellees did not believe these threats and promises, they would see little advantage in settling.	37
The appellor's threat to prosecute was clearly credible because failure to prosecute after initiation of the case in county court resulted in the imposition of fines on the appellor. The appellor thus had a monetary incentive to go forward with the prosecution, if no settlement was negotiated. Of course, by this reasoning, the victim's threat was not credible in the up-to-four-week period between the offense and the first county court, when the appellor was obliged to initiate her appeal. Nevertheless, as discussed in Section 4.C, it is unlikely that many cases were settled before initiation in county court.	38
The credibility of the appellor's promise not to prosecute (or, more precisely, not to continue to prosecute) if settlement were agreed upon is more problematic. Even during the periods when judges generally respected settlement by not sending nonprosecuted appellees to trial, there is no case that squarely holds that out-of-court settlement protected the appellee from further prosecution by an appellor who changed her mind.⁵² While judges tolerated settlements, they may not have enforced them. As discussed in Section 3.C, their tolerance for settlement probably reflected lack of reasonable alternatives rather than positive endorsement of settlement. Because there is little evidence of judicial enforcement, the credibility of the promise not to prosecute would have depended on the appellor's reputation, peer pressure, public opinion, possible threats of vengeance or self-help, and the intervention of third parties. There is some evidence that third parties assisted in the negotiation of settlements,⁵³ and it seems likely that these people would have helped enforce the settlement if a party later reneged.	39
E. Social Context
It is difficult to ascertain the social context of appeals. The plea rolls are the almost exclusive source of evidence, and they are frustratingly laconic. Cases 1 and 3 are typical in this respect. Occasionally, however, the plea rolls provide more background. Many of these cases conform to a common pattern. The appellor did something that violated what the appellee perceived to be his legal rights. The appellee then used self-help to enforce his rights. Often, the appellee seems to have been relatively powerful, with armed men at his command to assist him in using violence to enforce his claimed rights. The appellor, perhaps because he lacked the wealth and power to respond in kind, turned to the law for redress and brought an appeal. The appeal of a wounding on the next page is typical.	40
Case 4 (Shropshire 1203). Robert Trainel has appealed William the reeve of Hencott that with his accessories he took him and beat him and made him bloody and held him until he was delivered by the sheriff's clerk. And this he offers [to prove], etc. And William comes and denies the wounding and felony, but says that this is the truth, that Robert came into the fishpond of his lord the abbot, where he had no right of fishing, and fished there. And Robert says that he fished in that fishpond as in that in which he ought to have right of fishing. Afterwards Robert came and withdrew and put himself in [the king's] mercy. It is adjudged that for hunger and folly he fished in that pond and not for wickedness. Judgment is for the shire court, and Robert remits to the abbot his right of fishing.⁵⁴
In this appeal, Robert and the abbot had a disagreement about fishing rights. William, the reeve of one of the abbot's villages, used force to prevent Robert from fishing. Robert, who is described as "hungry" and therefore probably poor, could not use force to defend his claim, but he could bring an appeal against the abbot's reeve. The appeal, however, was unsuccessful, and Robert renounced his claim to fish in the abbot's pond.
Other cases reveal a similar pattern. These include appeals against a lord who used violence to enter into land after the death of a tenant,⁵⁵ against a landowner who imprisoned and tortured a suspected thief,⁵⁶ and against a lord who ransacked a tenant's house in retaliation for the tenant's suit in royal court over customs and services.⁵⁷ In these cases, the appellor was clearly suing a person of much higher status.⁵⁸ In other cases, such as Case 2, the appeal seems to have arisen out of one party's attempt to enforce his rights with violence, although the relative status of appellor and appellee is less clear. The right to impound animals often occasioned such appeals. One person would try to impound another's pigs or other animals, perhaps because they were trespassing or as security for some other dispute. The owner of the animals would try to retake them by force and a violent altercation would ensue. The party wounded in the fight would bring an appeal.⁵⁹ Although the relative status of the parties in such cases is not clear, the appellee usually seems to have been at least a modest property holder, who, for example, possessed land upon which another's animals could trespass. In some such cases, the party impounding the animals may have been a lord distraining his tenant to make him attend the lord's court.⁶⁰	41
Of course, there were appeals that did not fit this pattern. Some arose out of violent retaliation for insult,⁶¹ and others involved simple theft.⁶² In addition, few rape appeals fit this pattern. And in the vast majority of cases, there is no information on the causes of the dispute. Nevertheless, it is remarkable that when more information is available, the violence that gave rise to appeals seems usually to have been sparked by a prior dispute between the parties over land, chattels, or rights, and that the appellee was often a person of at least modest wealth. These characteristics lend plausibility to the idea that many appeals were brought in order to be settled and were, in fact, settled. Many appellees seem to have had sufficient wealth to pay money or to convey land as compensation, and the violence that underlay appeals was closely related to property disputes, which themselves were frequently the subject of settlement.	42
Part Two. Trends in the Rate of Appeals
Legal historians have long known that there were many appeals at the turn of the thirteenth century and very few in the sixteenth, but no attempt has been made to determine when this decline occurred. Maitland, the great turn-of-the-century legal historian, opined that the appeal was "but slowly supplanted by indictment,"⁶³ and later historians have either accepted this view with only slight modification or remained silent on the issue.⁶⁴ Sections 2.A through 2.E describe the trends in the number of appeals brought per year from the late twelfth century through the end of the thirteenth century. They show that the appeal declined dramatically during that century, but that the decline was in no way gradual, and that periods of swift decline alternated with periods of increase and stasis. Section 2.F examines published data on late medieval rates of appeal and shows that the thirteenth-century decline of the appeal was permanent. Knowledge of the trends in the rate of appeals is useful primarily because it lays the groundwork for Part 3, which tries to explain why the appeal declined.	43
A. The Data Set
In order to chart the patterns in the frequency with which appeals were brought, I examined eyre records from fourteen English counties from 1194 to 1294.⁶⁵ These records contain 1249 appeals. The period 1194–1294 was examined because, before 1194, there are no records from which reliable figures can be drawn and because, after 1294, eyres were no longer a regular part of English justice,⁶⁶ and the organization of the courts changed so drastically that figures derived from the records of the reorganized courts would not be comparable. With the exception of Kent, the fourteen counties were chosen because they are the only ones for which eyre records have survived for both the periods 1194–1209 and 1218–63. Kent was examined because its surviving records are unusually ample for the period 1226–44. Thus, these fourteen counties are those that shed the most light on changes in the rate of appeals in the early thirteenth century. While these counties were chosen based on the survival of their records, they are fairly representative of England as a whole, ranging from Kent and Wiltshire in the south, to Shropshire on the Welsh border, Norfolk and Essex in the east, and Yorkshire in the north. The area closest to London, however, is overrepresented. For these fourteen counties, all surviving eyre records before 1263 were examined. For five counties, the records for the rest of the thirteenth century were also examined.⁶⁷ Unfortunately, many of the records are damaged and fragmentary. As a result, as discussed below, to ensure comparability over time, the data set contains cases only from districts for which records are consistently complete.	44
B. Analysis without Regression
Table 2 shows the number of appeals per year for almost all districts in the database. Each cell of the table records the number of appeals per year for the relevant districts in a given county as reported in an eyre taking place in the time period indicated at the top of the column. The blank cells indicate the extent to which records have been lost or are so damaged as to be unusable. Light shading indicates that no eyre was held in that county during the relevant period. Dark shading indicates records that were not examined, because, as noted above, those of only five counties were examined after 1263.	45
Most of the surviving records are damaged or incomplete. Fortunately, they are organized by district. To ensure comparability over time, each row in Table 2 includes appeals only for those districts for which records are consistently available and complete. If the row for a given county simply recorded all surviving information for that county, it would be impossible to tell whether an increase between two periods recorded a true increase in the number of appeals or simply the fact that the later figure was drawn from a less fragmentary source. Table 2 avoids that problem because all cells in a given row record information for the same set of districts. Thus, all Yorkshire cells exclude appeals from Harthill and Buckrose because the 1208 eyre roll lacks complete sections for these districts. Even though the eyre rolls for 1218–19, 1231, and later eyres survive for these districts, the appeals for these districts in these eyres were not counted because doing so would render meaningless any comparison to rates derived from the 1208 eyre. The column labeled "Districts" indicates both how many districts were analyzed for each row in the table and the total number of districts in the relevant county, thus providing a rough measure of the extent to which the numbers in Table 2 represent all appeals in the county or just a small fraction of them. The Appendix, Part B, lists the districts included in each row, and, the Appendix, Part C describes the criteria for inclusion in greater detail. Table 12, in the Appendix, Part A, shows the number of appeals per year for a small number of additional districts with odd survival patterns. Their inclusion in Table 2 would have cluttered the table without altering the analysis.	46

It is important to recognize that, although the figures in the table appear small, the number of appeals examined, 1249, is reasonably large. The figures seem low because they are rates: the number of appeals divided by the number of years covered by a given eyre. Since an eyre heard cases initiated in county court over the previous several years, the number of appeals is much higher than the rate. For example, the rate of 3.3 in the 1227–28 Bedfordshire eyre reflects the fact that judges in that eyre heard twenty-nine appeals, which had been initiated since the previous eyre had ended eight years, ten months, and one day earlier. Table 3 shows how the rates in the first row of Table 2 (Bedfordshire) were calculated.	47

An additional reason that the rates in Table 2 are relatively low is that, as noted in the "Districts" column, for many counties the table counts appeals only from a few districts because the records of the other districts have not survived intact. The records for Bedfordshire, Shropshire, Staffordshire, and Wiltshire are nearly complete, so their rates fairly represent the number of appeals per year for the entire county.⁶⁸ The rates in the table for the other eight counties, however, significantly underestimate the rates for the whole counties because complete records for many districts do not survive.	48
Inspection of Table 2 shows that most counties conform to the pattern graphed below:	49
Figure 1. Rate of Appeals, 1194–1294 Note: To facilitate comparison across counties, the y-axis has been numbered from zero to 100, rather than with the rates for any single county. The rate for the first decade of the thirteenth century was set arbitrarily at 100. The precise numbers on the y-axis are irrelevant, because the purpose of the graph is to illustrate relative increases and decreases in the rate of appeals over time. As is explained below (28), the graph plots regression coefficients with two modifications.
Bedfordshire conforms almost exactly to the pattern depicted in Figure 1. It shows a large decline from 1201–3 to 1226–29, a rebound to 1246–49, and then an even larger decline to the end of the century. The other eleven counties also show similar trends. All five counties with records in the periods 1194–1209 and 1218–1229--Bedfordshire, Buckinghamshire, Essex, Shropshire, Staffordshire, and Yorkshire--show declines between these two periods. Similarly, the five counties with records in the periods 1218–1229 and 1231–1249--Bedfordshire, Buckinghamshire, Essex, Kent, and Yorkshire--all show increases. All counties for which data were gathered after 1260 show rates dramatically lower during the period 1261–94 than in 1194–1209 or 1231–52.
Of course, a few counties do not fit the pattern. For example, the rate of appeals continued to rise in Essex between 1246 and 1258, while Figure 1 shows mostly decline. In addition, while Buckinghamshire and Essex show declines from 1194–1203 to 1226–1229, these are much smaller than those experienced in other counties. This difference almost certainly reflects bad record keeping before 1200 because it was not until the 1201–3 eyres that justices used the coroners' rolls to check the presenting jury's report of appeals initiated in county court.⁶⁹ Nevertheless, even taking into account these divergences, examination of Table 2 shows that most counties fit the pattern rather well.	50
C. Regression Analysis
The generally good fit withstands not only informal inspection but also a more rigorous statistical analysis. Although computationally complicated, the idea of regression is simple. It is a mathematical tool for measuring the relationship between variables, in this section between eyre dates and rates of appeals.⁷⁰ Regression is helpful for three principal reasons. First, it can take into account all of the data. The analysis in the previous section focused on the most salient eyres and counties but failed to mention any data from six counties (Hertfordshire, Lincolnshire, Middlesex, Norfolk, Northamptonshire, and Wiltshire), not to mention individual eyres (such as Shropshire 1256) and all the data in Table 12. With so many data points, informal analysis is inherently selective. Only regression analysis can synthesize and integrate the mass of data. Second, regression analysis can produce numbers (such as the coefficients discussed below) that help to produce tables and graphs to summarize and communicate complex data. Third, regression analysis can help distinguish patterns that reflect real change from those that are more likely to reflect mere chance. When used improperly, regression results can produce a false sense of precision, but regression analysis also produces statistics (such as confidence intervals and p-values, discussed below) that help to assess the appropriate degree of precision to be accorded the results and the confidence with which results can be relied upon.	51
A simple regression, which attempts to explain the rate of appeal by a variable representing eyre dates, controlling only for county, explains most of the variance and yields statistically significant results. In such a regression each eyre visitation is assigned a variable (called a dummy variable) that is one if the data point is from that group of eyres and zero if it is not. Thus, for each data point, there is one eyre-date dummy variable that is one, and the rest are zero. Similarly, each county is assigned a dummy variable. Table 4 displays the most important results: the coefficients and associated statistics for the eyre-date dummy variables.	52

Each row of Table 4 corresponds to a column in Table 2 and reports the statistics for the dummy variable for those eyres. The second column, labeled "Coefficient" reports the regression's estimate of the degree to which the rate of appeals differed from that in the 1201–3 eyres. Thus, the fact that the coefficient for the 1218–22 eyres is 0.41 indicates that the rate reported in those eyres was only 41 percent of the rate reported in the 1201–3 eyres. Similarly, the fact that the coefficient for the 1246–49 eyres is 0.94 indicates that by that time the rate of appeals had rebounded almost to the levels attained in the 1201–3 eyres. In the 1250s, however, the rate of appeals began to plummet, so that by the 1260s it had fallen to between a quarter and a third of the levels attained at the turn of the century.	53
The graph in Figure 1 essentially plots the regression coefficients, with two deviations. The scale on the graph multiplies the coefficients by 100 and thus ranges from zero to one hundred rather than from zero to one. In addition, the graph plots a steady rate from 1194 to 1203, even though the coefficients for 1194–95 and 1198–99 are less than one. As explained above (26), the figures for these years almost certainly underreport the true rate. The graph has been adjusted to take this into account.	54
The third column of Table 4, the p-values, measures the statistical significance of the results. P-values of less than 0.05 generally indicate statistically significant results, and p-values of between 0.05 and 0.10 are considered marginally significant. It is thus important to note that the p-values for the most important of the eyres are easily significant at even the 0.05 level. The p-values for the 1218–22, 1226–29, 1252–58, 1261–63, 1268–77, 1278–89, and 1292–94 eyres are all much below 0.05, and all but the 1252–58 eyres are below 0.01. We can thus be confident (although, of course, not absolutely sure) that the declines from 1201–3 to 1218–29 and from 1246–49 to the end of the century were not merely the result of the lucky survival of records. The fact that the p-values for the 1231–33, 1234–38, 1239–44, and 1245 eyres are so high, however, means that we cannot be confident that the appeal had not already completely rebounded to turn-of-the-century levels by the 1230s.	55
The fact that the p-value for the 1246–49 eyres is almost one does not suggest that we cannot be confident that the rate of appeals had not fully rebounded by the late 1240s. P-values are useful only in testing the hypothesis of difference from the base (here the rate revealed by the 1201–3 eyres), not in testing the hypothesis of similarity. The last column, however, is helpful for that purpose. It gives the 95 percent confidence intervals for the coefficients and indicates that we can be 95 percent confident that the rate of appeals for the 1246–49 eyres was between 68 and 130 percent of the 1201–3 rate. While this confidence interval allows for substantial deviation from the turn-of-the-century rate, even the lower bound is higher than the 1226–29 rate, which was 51 percent of the turn-of-the-century level. The significance of the rebound from 1226–29 to 1246–49 can also be measured by rerunning the regression using the 1226–29 eyres as the base instead of the 1201–3 eyres. By doing so, the p-values test the hypothesis of difference from 1226–29 rather than 1201–3. If the regression is rerun in this way, the p-value for 1246–49 is 0.000, indicating that the rebound from 1226–29 to 1246–49 is very statistically significant.	56
In Section 2.B above, I argued that, although Bedfordshire fit the pattern depicted in Figure 1 almost exactly, the other eleven counties also show similar trends. This conclusion is buttressed by regression analysis. If the regression described above is repeated excluding Bedfordshire, the results are nearly identical. Only three coefficients change by more than 0.05: the coefficient for the 1226–29 eyres increases from 0.51 to 0.66, the coefficient for the 1239–44 eyres increases from 0.83 to 0.94, and the coefficient for the 1246–49 eyres decreases from 0.94 to 0.88. These changes do not substantially change the overall trends. In addition, the p-values generally increase, although only two cross the 0.05 significance threshold: the p-value for the 1226–29 eyres, which increases to 0.057, and the p-value for 1252–58, which increases to 0.069. Even these p-values are close to being statistically significant. Taken together, the changes in the coefficient and p-value for 1226–29 suggest that without Bedfordshire, the rate of appeal in the 1226–29 eyres might not have been much lower than in 1201–3. On the other hand, by excluding Kent, the rate of appeal could be made to appear much lower and more statistically significant.⁷¹ Nevertheless, since there is no more reason to drop Bedfordshire than to drop Kent, the regression results for the 1226–29 eyres reported in Table 4, which include all twelve counties in the data set, are the best guide to the overall trends in appeals.	57
D. Analysis by Crime
The previous section analyzed appeals for all crimes together. This section disaggregates those results. Table 5 shows regression coefficients for each crime category. These regressions are identical to those reported in Table 4, except the dependent variable is the number of appeals of a particular crime, rather than the total number of appeals. To save space, only the coefficients are reported. Statistical significance at the 0.05 level is indicated by an asterisk (*). The last row of the table reproduces the coefficients from Table 4 for comparison.	58

Although there are some differences from crime to crime, the similarities are more pronounced. All crime categories, except rape and homicide, show large declines from 1201–3 to 1218–22 or 1226–29, and most are statistically significant. Similarly, with the exception of the miscellaneous "other" category, all crimes show 1246–49 rates near their 1201–3 levels. And finally, all crime categories show low rates (coefficients well below one) toward the end of the century.⁷² In fact, with the exception of homicide, the rate of appeals for all eyres after 1265 was less than 50 percent of the 1201–3 rate for all crimes and often statistically significant at the 0.05 level. Even the homicide rate was down more than 40 percent, although its decline is not statistically significant. Thus, most crime categories, with the exception of homicide, show patterns similar to the overall trend. Section 3.D provides some explanation for why homicide rates were different.	59
E. Appeals in Gaol Delivery, the Bench, and Coram Rege
So far, this article has described the trends in the number of appeals by examining eyre records from fourteen counties. This section justifies the reliance on eyre rolls by showing that relatively few appeals were heard in other courts. In addition, the records for these other courts do not support the hypothesis that declines in the number of appeals heard in the eyre were offset by increases in the number of appeals heard elsewhere. This analysis of sources other than eyre rolls is extremely important: if most appeals were heard in other courts, or if decreases in the eyre were offset by increases elsewhere, then the trends identified above would be almost meaningless.⁷³	60
Other than the eyre, the principal places where appeals could be tried were gaol (jail) delivery sessions, the court coram rege (later known as King's Bench), and the Bench (later known as Common Pleas or Common Bench).⁷⁴ Gaol delivery rolls record cases heard by judges acting on commissions that empowered them to try only those persons being held in particular jails. Such judges may also have tried those released on bail. In contrast, eyre judges had commissions that empowered them to hear all sorts of matters, including trials of those not sufficiently dangerous to have been imprisoned or bailed and reports of felonies committed by those who had fled and could not be caught. By the fourteenth century, gaol delivery had become the most important forum for the trial of criminal cases. The relative importance of eyre and gaol delivery in the thirteenth century has not been systematically studied, but it is probable that by mid-century, if not earlier, more criminal cases were tried in gaol delivery than in eyre. Unfortunately, only a handful of gaol delivery plea rolls survive from before 1270. The surviving evidence, however, is remarkably consistent. Gaol delivery rolls from the first part of the century record appeals at rates of up to three per county per year,⁷⁵ while those from the latter part of the century record only one or two per county per year.⁷⁶ The number of appeals heard at gaol delivery was thus relatively low in comparison to the number heard in the eyre. Since gaol delivery was restricted to persons jailed or bailed, while most appellees were simply attached to appear, the relatively small number of appeals heard in gaol delivery is not surprising. In addition, the fact that there were generally more appeals heard at gaol delivery in the early thirteenth century than later suggests that the dramatic declines in the number of appeals discussed above do not merely reflect a shift of cases from eyre to gaol delivery. Rather, both eyre and gaol delivery records show a decline over the thirteenth century.	61
The principal courts of the common law were the Bench and court coram rege. The former was generally held at Westminster, while the latter traveled with the king, wherever he went. In the fourteenth century, their jurisdictions would be sharply distinguished, but this was not yet the case in the thirteenth.⁷⁷ Each heard about one appeal per county per year.⁷⁸ As with gaol delivery, this number is much lower than the number of appeals heard in the eyre. In addition, like gaol delivery, the number heard in the Bench and coram rege did not rise through the century (and may even have been falling), so the reduction in the number of appeals heard in the eyre cannot be attributed to a shift in cases to these courts.	62
F. Rates of Appeal in the Later Middle Ages
Recent research on the later Middle Ages has suggested that the appeal "enjoyed a vigorous old age."⁷⁹ Some have even tentatively questioned whether appeals were any less common in the fourteenth and fifteenth centuries than in the thirteenth.⁸⁰ Table 6 summarizes data gathered by other scholars on appeals in later medieval gaol delivery rolls.	63

Although there is considerable variation, the average number of appeals in late medieval gaol delivery rolls was well within the one to two appeal per county per year rate observed in the late thirteenth-century gaol delivery rolls discussed in the previous section. This low rate is somewhat surprising because all of the figures in the table, except Musson's, seem to include approver appeals. As discussed in the introduction, such appeals, which were brought by convicted criminals, were systematically excluded from this article. Since such appeals often constituted a substantial fraction of appeals on gaol delivery rolls, the average rate of nonapprover appeals in the later Middle Ages was probably closer to one per county per year.	64
Only one scholar, Whittick, has counted appeals in the central common law courts. He found 398 appeals in King's Bench in the period 1485–1495.⁸¹ This yields, on average, one appeal per county per year, exactly the thirteenth-century rate.	65
The preceding paragraphs are sufficient to show that there was no resurgence of the appeal in the later Middle Ages and that the mid-thirteenth-century decline of the appeal was permanent. In fact, the figures suggest a continued decline. The general eyre, the forum in which most thirteenth-century appeals had been brought, was no longer in existence in the late fourteenth and fifteenth centuries. Therefore, if the overall rate of appeals had remained constant, there would have been a substantial increase in the number of appeals heard in gaol delivery and/or King's Bench. Similarly, because the Court of Common Pleas (the Bench) stopped hearing appeals in the fourteenth century, the rate of appeals heard in King's Bench and/or gaol delivery should have increased. In fact, rates of appeal in gaol delivery and King's Bench did not rise to compensate for the constriction in fora in which appeals could be brought. Instead, they remained at rates comparable to those in the late thirteenth century. This suggests that the number of appeals per year probably declined further from the already reduced late thirteenth-century rates.	66
Part Three. Respect for Settlement and the Changing Rate of Appeals
Knowledge of the changing rate of appeals is useful primarily because it helps explain why the appeal declined. This part addresses that question. It first surveys the reasons others have put forward for the decline of the appeal and shows why they are unpersuasive. It then argues that changes in judicial attitudes toward settlement provide the best explanation for the changing rates of appeal.	67
A. Previous Explanations for the Decline of the Appeal
Although the general decline of the appeal during the Middle Ages is well known, relatively few historians have attempted to explain it.⁸² Those who have ventured explanations have suggested four reasons: (1) the appeal's archaic nature, especially the use of trial by battle; (2) judicial hostility, which manifested itself in the ease with which appellees could exploit technical defects to quash appeals; (3) the introduction of presentment, which meant that crimes might be prosecuted even if the victim did not appeal; and (4) the introduction of trespass actions, which were more attractive to victims because they provided money damages.⁸³	68
The complex pattern of changing rates of appeals outlined in Part 2 shows that these explanations are at best only partially correct. None of them can explain why the number of appeals increased from 1226 to 1249. Nor can they explain why the rapid decline in the 1210s and 1250s.	69
Fear of trial by battle and the ease with which appeals could be quashed cannot explain the changes in the rate of appeals. Battle and technicality had been part of the appeal procedure well before the declines observed in the thirteenth century. In fact, if fear of battle were a serious impediment to bringing appeals, the rate should have increased in the latter part of the thirteenth century because, as discussed in the next section, by the second half of the century, an appellor could avoid battle, while ensuring a jury verdict on the appellee, by dropping or not prosecuting the case. Similarly, if potential appellors were deterred by the ease with which technical errors could be used to quash appeals, they should have brought more appeals in the later part of the thirteenth century because judges in that period forced appellees to submit to jury trial when appeals had been quashed.⁸⁴	70
Nor can the introduction of presentment wholly explain the decline of the appeal. Presentment became a routine part of criminal procedure at the latest under Henry II in the 1160s and 1170s, far too early to have caused the precipitous declines in the 1210s and 1250s. It is, of course, possible, even probable, that the introduction of presentment caused declines in the appeal in the period 1166–1194 or even earlier, but there are no data with which to test that hypothesis. In addition, presentment of assaults and rapes was extremely rare, so the introduction of presentment cannot explain the thirteenth-century declines in the number of these appeals.	71
The availability of trespass actions, which allowed victims of most assaults and property crimes to bring a civil tort action for damages, also cannot explain the declines in the 1210s and 1250s. In addition, if trespass had directly caused the decline of the appeal, the decline should have been confined only to offenses that could give rise to trespass actions. Trespass actions for rape did not exist until after the 1285 Statute of Westminster II, and yet the number of rape appeals fell well before that time.⁸⁵ In addition, trespass was never available for homicide, yet, as discussed in Section 2.D, the number of such appeals fell along with appeals of assaults and theft, albeit somewhat less dramatically. Nevertheless, as is discussed more fully below, the availability of trespass did play a role in the decline that occurred in the 1240s and 1250s.	72
B. Settlement Policy and the Changing Rate of Appeals
The best explanation of the decline of the appeal lies in changing judicial policy toward private settlement. In order to understand the importance of settlement policy, it is necessary to consider why people brought appeals in the first place. Some brought appeals because they wanted the appellee to be punished for harm done to the appellor or to a family member. One might characterize this motive as justice or revenge. Others brought appeals because they wanted compensation for harm done to them.⁸⁶ In the late twelfth and early thirteenth centuries, there was no routine royal remedy by which victims could obtain damages for personal injury or property damage.⁸⁷ Nevertheless, depending on the judicial policy toward settlement, victims could use the appeal to induce compensation. If the appellor was victorious at trial, she would receive no compensation,⁸⁸ and the appellee would be punished either with death or a fine. Fear of hanging or fines, however, gave appellees powerful reasons to negotiate with their accusers, and money or other consideration might induce an appellor to drop the case. Case 3 is a particularly vivid illustration of the process. The appellor claimed she had been raped and brought an appeal. When the case came up for trial, however, she withdrew her appeal in exchange for two acres of land.	73
The appellor could use an appeal to procure a settlement, however, only if the appellee thought that settlement would protect him from further prosecution. This was not always the case. Sometimes judges disregarded settlements and tried the defendant "at the king's suit." Trial without the cooperation of the victim-prosecutor was possible because the jurors were self-informing and did not need the victim's testimony in order to convict.⁸⁹ Juries seemed to have been quite willing to convict nonprosecuted appellees. In fact, the conviction rate at the king's suit was roughly the same as the conviction rate of those prosecuted by the appellor.⁹⁰ Case 2 is illustrative of the many cases in which judges took a jury verdict and punished the appellee despite settlement. Such disregard of settlements, however, severely undercut the victim's bargaining position. If settlement with the appellor did not protect the appellee from trial, why settle?⁹¹ And if appellees would not settle, victims, to the extent that they were motivated by the desire for compensation, might not bring appeals at all.	74

Table 7 charts judicial respect for settlements by recording the percentage of nonprosecuted appeals in which judges let the appellee go free without trial. Section 3.G discusses some alternative ways of measuring respect for settlement. Table 7 shows that judicial respect for settlement varied considerably. In the late twelfth and early thirteenth centuries, settlements were almost always respected. In 90 percent or more of nonprosecuted appeals, the appellee went free without trial, as in Case 1. In the 1218–22 eyres, however, the judges began disregarding settlements, letting appellees go free without trial in barely a third of nonprosecuted appeals.⁹² But this disrespect for settlement was short-lived, and in the late 1220s and 1230s the judges again let appellees go free without trial when the appellor had decided not to prosecute. Then, in the 1239–44 eyres, the judges began to return to the antisettlement policy. By the 1260s, nearly all appellees in nonprosecuted appeals were required to submit to jury trial.	75
The figures in Table 7 provide a powerful predictor of the number of appeals. When judges in one set of eyres respected settlements (that is, when the percentage of nonprosecuted appeals in which the appellee went free without trial was high), the number of appeals recorded in the next eyres tended to be high. So, for example, the appellee went free without trial in at least 90 percent of nonprosecuted appeals in the 1198–99 and 1234–38 eyres, and the rate of appeals in the subsequent eyres (1201–9 and 1239–44) was relatively high (coefficients of 1 and 0.83 in Table 4). Conversely, when judges ignored settlement (that is, when the percentage of nonprosecuted appeals in which the appellee went free without trial was below 40 percent), as in 1218–22 and 1268–77, the rate of appeals in the subsequent eyres (1226–29 and 1278–89) tended to be low (coefficients of 0.51 and 0.26 in Table 4). The relationship between respect for settlement and rates of appeal is easiest to see if the two are graphed together, as in Figure 2.⁹³	76
Figure 2 shows that judicial respect for settlements is a very good predictor of the number of appeals brought. The percent of nonprosecuted appeals in which the appellee went free without a jury verdict and the number of appeals tend to go up and down together. Unlike the four explanations for the decline of the appeal mentioned above, judicial policy toward settlement helps to explain both when the sharp declines occurred and the fact that the rate of appeals rose in the 1230s. The close relationship between respect for settlement and the number of appeals in the subsequent eyres is confirmed by regression analysis.⁹⁴	77
Figure 2. Respect for Settlement and Rate of Appeal, 1194–1294
Of course, the correspondence between respect for settlement and the number of appeals is not perfect. There are two major divergences. First, the rate of appeals rose to a high level in 1246–49, even though respect for settlement in 1245 was only moderate. One possible explanation is that, as discussed below in Section 4.A, crime may have spiked in this period. The second anomaly is that the rate of appeals was low in the 1208–9 and 1218–22 eyres, even though judges in the preceding eyres (1201–3 and 1208–9) showed a high degree of respect for settlement. One would have expected that the rate of appeal would have been high in the period 1208–22 and would only have declined in 1226–29. This discrepancy does not, however, refute the relationship between respect for settlement and the number of appeals. The 1208–9 rate is unreliable because it is based on records from only five districts in a single county (Yorkshire). This unreliability is confirmed by the regression p-value, which, at 0.201, suggests that the apparent decline from 1201–3 to 1208–9 is not statistically significant. The unexpectedly low rate for the 1218–22 eyres is best explained by the fact that judges in these eyres heard cases initiated between 1208 and 1222. This was a very turbulent period that included the interdict, civil war, and other major disruptions of ordinary judicial processes.⁹⁵ The fact that there were relatively few appeals in the 1218–22 eyres almost certainly reflects the special circumstances of this period.	78
C. Explaining Changes in Settlement Policy
Using judicial respect for settlement to explain the rate of appeals only pushes the inquiry back one step. Why did judicial policy toward settlement change? That there was a policy seems clear because judges on different circuits coordinated their treatment of nonprosecuted and settled cases. Nevertheless, in the absence of contemporary accounts of the issue, it is difficult to ascertain the motives for this change.⁹⁶ The following account seems most plausible.	79
In the late twelfth and early thirteenth centuries, judges faced a tough choice. Crimes prosecuted by appeal were considered serious. Because they were offenses not only against the victim but also against the king's peace, out-of-court settlement was not officially condoned.⁹⁷ Yet the judges had no good way of determining guilt or innocence if the appellor refused to prosecute. Jury trial was not yet an accepted mode of proof in criminal cases, so if judges wanted to try criminals in spite of settlement, they would have had to send appellees to the ordeal. But this was an unacceptable option because ordeals were controversial. Some were skeptical about the accuracy of ordeals.⁹⁸ Others doubted whether there was adequate justification in the Bible and patristic sources for their use.⁹⁹ So, faced with the choice between respecting settlements and putting appellees to ordeals, judges almost always respected settlement.	80
In a few cases, however, the justices began to experiment with an antisettlement policy. They asked the presenting jury whether it suspected the appellee, and in one of the two instances in which the jury responded that it did, the justices put the appellee to the ordeal.¹⁰⁰ Most of these early cases in which the justices asked the presenting jury its opinion of the defendant were homicide appeals.	81
In 1215, the Fourth Lateran Council forbade clerics to participate in ordeals, thus effectively banning ordeals. As a result, trial by jury became routine in criminal cases,¹⁰¹ and judges no longer faced such a difficult choice. Now they could ascertain guilt or innocence in the absence of a prosecuting appellor by referring the question to the jury "at the king's suit." They did so in a majority of cases (64 percent) in the 1218–22 eyres, the first eyres after the abolition of the ordeal.	82
Disrespect for settlements, however, caused people to bring fewer appeals. After the restoration of order and ordinary judicial processes in the 1220s, judges expected the number of appeals to return to turn-of-the-century levels, but the rates remained depressed at levels barely higher than during the turbulent 1210s. The judges realized that their disrespect for settlement had taken away one of potential appellors' primary motives for bringing appeals. By punishing nonprosecuted appellees, they had discouraged potential appellors from bringing prosecutions, because appellees were now much more reluctant to settle. The judges thus faced another tough choice: either continue the antisettlement policy and let much crime go unprosecuted¹⁰² or tolerate settlements in order to induce more prosecution. The judges chose the latter alternative and again began routinely respecting settlements. In the 1226–29 eyres, they let appellees go free without trial in 67 percent of nonprosecuted appeals. By the 1234–38 eyres, they had completely reversed the policy and let appellees go free without trial in 93 percent of nonprosecuted appeals. The policy reversal had the desired effect, and the number of appeals increased by more than 50 percent. In the 1226–29 eyres, appeals were brought at barely half (51 percent) the rate they had been brought at the turn of the century. By the 1239–44 eyres, the rate had rebounded to 83 percent of the turn-of-the-century rate, up 63 percent in less than fifteen years.	83
In the 1230s, however, the royal courts began to develop an alternative to the appeal, which would eventually be known as trespass.¹⁰³ This new action could be brought for most of the same offenses as appeals, including assaults and thefts, but did not give the defendant the option of trial by battle or require formalities such as initiation in county court. Eventually, trespass would become a general tort action by which plaintiffs could garner monetary damages. Whether the cases from the 1230s can be classified as tort, or even trespass, is open to debate. Nevertheless, by 1239, there was clearly something other than an appeal that the victim could bring.¹⁰⁴ Once this alternative was available, judges no longer feared that disrespect of settlements in appeal cases would let wrongdoers go unpunished. So they resumed their antisettlement policy. Starting in 1239, they let fewer and fewer appellees go free without trial in nonprosecuted appeals. By the 1250s this policy began showing its effect. The appeal was down more than thirty percent from its 1246–49 peak, and by the 1261–63 eyres appeals were being brought at only a third of their rate at the turn of the century or at their 1240s peak. The policy of disrespect for settlements did not, however, completely eliminate appeals. Some appeals had always been brought in order to punish (or outlaw) the appellee, and these were unaffected by the change in policy toward settled cases. In fact, the knowledge that even a nonprosecuted appeal would subject the appellee to trial and possible punishment could have encouraged such appeals.	84
Other scholars have invoked the introduction of trespass actions to explain the decline of the appeal, but here its role is different. In my explanation, the rise of trespass did not directly cause the decline of the appeal. Instead it induced the judges to reassert their antisettlement policy, and it was that policy that caused the decline. This explanation accords better with the chronology. If the introduction of trespass had caused the decline in the appeal, the decline should have been apparent in the 1239–44 eyres. Instead, the number of appeals kept rising. In addition, if trespass instigated the decline of the appeal, homicide and rape appeals should not have fallen because there were no trespass actions for these crimes. If judges respected settlements in appeals, people seem to have preferred appeals to trespass actions because they were cheaper and provided more bargaining leverage. Unlike trespass suits, appeals could be prosecuted locally and so did not require a costly trip to Westminster. In addition, appellors may have been able to extract higher settlements when appellees feared the criminal sanctions imposed after successful appeals.¹⁰⁵	85
D. Analysis by Crime
The discussion in Part 3 so far has analyzed respect for settlement for all crimes together. Table 8 charts respect for settlement for each crime separately. It was compiled in the same way as Table 7 above. Unfortunately, for some years and crimes, there were few nonprosecuted appeals so the numbers in the table may not be very reliable. Percentages based on more than five nonprosecuted appeals, which are more likely to be accurate, are marked in the table with an asterisk.	86
Table 8 indicates that settlement policy was applied uniformly to all crime categories except homicide. Before 1218, nearly all nonprosecuted appellees went free without trial, except those accused of homicide. Similarly, nearly all nonprosecuted appellees underwent trial in 1218–22. The only exception was assault, and its unexpectedly high percentage (80 percent) is probably unreliable because it is based on very few nonprosecuted assault appeals. All crime categories, except homicide, show a return to respect for settlement (high percentages) in the 1226–29, 1231–33, and 1234–38 eyres, and then all, except homicide, show precipitous declines in respect for settlement for the rest of the century.	87

Appeals of homicide defy the patterns both in the rates of appeal and in respect for settlement. As discussed in Section 2.C, rates of homicide appeals dipped only slightly in the period 1208 to 1229 and fell less dramatically in the late thirteenth century. In addition, judges began implementing the antisettlement policy against homicide as early as the 1198–99 eyres and then did not relax it in the late 1220s and 1230s.¹⁰⁶ The close relationship between settlement policy and the number of appeals suggests that these two anomalies were related. The rate of homicide appeals fell only moderately between the 1201–3 and 1226–29 eyres because the judges had begun implementing the settlement policy even before the 1201–3 eyres. Similarly, there was no rebound in the 1230s, because the antisettlement policy was not relaxed in the late 1220s and 1230s.	88
The fact that both the antisettlement policy and the rate of appeals changed less dramatically for homicide than for other crimes supports the link between settlement policy and rates of appeal, but it also raises the issue of why homicide was treated differently. The seriousness of the crime probably explains why the antisettlement policy was applied first to homicide. Before 1218, application of the antisettlement policy risked sending nonprosecuted appellees to the ordeal, and the royal judges may have been willing to do that only for homicide, the most grave offense.¹⁰⁷ The coroners' duty to investigate suspicious deaths probably explains why the antisettlement policy was not relaxed in homicide cases in the late 1220s. For other crimes, judges realized that if there was no appeal, presentment was unlikely and crimes would go unprosecuted. But by cross-checking the coroners' rolls with eyre presentments, judges could be confident that most homicides would be prosecuted by presentment, even if the antisettlement policy discouraged appeals. As a result, judges did not face the same dilemma regarding homicide as with other crimes and could keep the antisettlement policy in place. The statistics presented in Section 4.A confirm their reliance on presentment of homicide. Even as the number of homicide appeals declined, the number of homicide presentments rose more than enough to ensure that the total number of homicide prosecutions increased.	89
E. Canon Law Influence
English judges may have borrowed the idea of sending nonprosecuted appellees to trial from the canon law. As in secular law, the primary mode of prosecution in twelfth-century canon law was individual accusation, usually by fellow clerics. Like English appeals, ecclesiastical accusations were sometimes settled. The canon law on settlements, however, was ambivalent. On the one hand, as reflected in Gratian's Decretum, it encouraged settlement because the litigious spirit was thought inappropriate for clerics.¹⁰⁸ On the other hand, Gratian also collected texts condemning settlement of criminal accusations.¹⁰⁹ Nevertheless, like contemporary English law, the Decretum contained no effective way of detecting or deterring settlements.	90
Because most ecclesiastical offenses lacked clearly identified victims, settlement created serious problems for the administration of canon law. For example, suppose someone accused a priest of purchasing his ordination, and suppose the prosecutor and priest reached a settlement in which the priest paid the accuser ten pounds. This would hardly be a satisfactory resolution of the problem. When someone is accused of assault or theft, settlement can be justified as compensating the victim, but when the crime is victimless, settlement is more likely to aggravate the offense than compensate the victim.	91
Late twelfth-century canon lawyers found a solution to this problem in the actions of Pope Gregory the Great. Having heard that certain grave accusations against a bishop had been settled, Pope Gregory instructed the bishops of Corinth to investigate the matter, notwithstanding the settlement.¹¹⁰ Although Gratian and earlier canon law writers did not refer to this incident,¹¹¹ late twelfth-century collectors of papal letters (decretals) included two texts that did.¹¹² By including these texts, decretal collectors transformed Pope Gregory's instructions for a particular controversy into a precedent of general applicability. Decretal collections were meant to be used by canon law judges. Therefore, the texts would have been interpreted as instructing the ecclesiastical judge to investigate crimes even when the parties had reached a settlement. This instruction is very similar to the antisettlement policy that English judges began experimenting with in the 1190s and made routine in 1218. In both, the judge inquired into the guilt or innocence of defendants, even when the accuser was no longer prosecuting. This similarity suggests that English judges might have borrowed the idea from the canon law.	92
This borrowing cannot be directly proven, but its plausibility is enhanced by the fact that the canonical encouragement to investigate nonprosecuted accusations was disseminated widely in the 1190s with the publication of the Compilatio Prima.¹¹³ Other decretal collections, including the Appendix Concilii Lateranensis, which was probably written in England,¹¹⁴ also spread knowledge of the canonical approach to nonprosecuted accusations. A prolific and learned group of canonists flourished in England in the late twelfth and early thirteenth centuries. They would have been familiar with this new canon law approach to settlements.¹¹⁵ Their knowledge could easily have spread to the shapers of the common law because there was much interaction in this period between the canon law and the common law. Several royal judges active in the late twelfth and early thirteenth centuries had canon law training. Richard Barre studied at Bologna, where he was friends with the distinguished canonist Stephen of Tournai.¹¹⁶ Hubert Walter had been a papal judge delegate,¹¹⁷ and there is some evidence that several other royal judges were familiar with the canon law.¹¹⁸ In addition, a substantial number of eyre judges were bishops or archdeacons, who might have acquired knowledge of the canon law through their judicial responsibilities within the church.¹¹⁹ The idea that some people working in the royal courts had a thorough knowledge of canon law is also supported by the treatise attributed to Bracton, which is peppered with quotations from the canon law. One such quotation is found in the very passage in which he discusses the antisettlement policy.¹²⁰	93
F. Legal Knowledge of Potential Appellors and Appellees
The idea that changes in settlement policy can explain the number of appeals assumes that potential appellors and appellees knew about settlement policy. It assumes that appellees knew whether judges at the last eyre respected settlements and used that knowledge to predict whether settlement of their own case was likely to protect them from further prosecution. Similarly, it assumes that potential appellors knew the settlement policy enforced at the previous eyre and thus knew whether they were likely to be able to extract a settlement. Such legal knowledge among nonlawyers might seem implausible, but given thirteenth-century institutional arrangements, it is not.	94
Numerous men from every village would have attended the eyre. Every village sent four men and its reeve to the eyre to assist the presenting and trial jurors.¹²¹ In addition, anyone with a case at the eyre would have attended, as well as those summoned as jurors. Thus, at least five men from each village would have heard how judges decided criminal cases, and they could have reported back to their fellow villagers about judicial respect for settlement.	95
The very technicality of the appeal supports the idea that potential appellors would have been legally sophisticated enough to know about and respond to changes in settlement policy. It was not easy to bring an appeal, and appellors were not usually represented by counsel.¹²² The fact that appellors had sufficiently mastered the legal technicalities to bring an appeal confirms that they would have had a fair amount of legal knowledge. It may also point to the existence of nonprofessional legal experts and advisors at the local level.	96
G. Measuring Respect for Settlement
To measure respect for settlement, Section 3.B analyzed the percentage of nonprosecuted appeals in which judges let the appellee go free without trial. That analysis, however, is somewhat problematic because some nonprosecuted appeals were not settled and because some settling parties were penalized without trial. Despite these shortcomings, the measure used in Section 3.B is probably the best. To support that conclusion, this section examines three alternative measures of settlement policy.	97
In 207 cases in the data set, the eyre rolls record that the parties settled. For these cases, settlement policy can be measured directly by determining whether the appellee was put to trial. The second row of Table 9 below, labeled Measure 2, shows respect for settlement as measured by the percentage of settled cases in which the defendant went free without trial. For comparison, the first row (Measure 1) redisplays the percentages used in Table 7, that is, the percentage of nonprosecuted appeals in which judges let the appellee go free without trial.	98
Some historians, most notably Roger Groot, have suggested that settled cases were likely to have been formally retracted rather than simply nonprosecuted.¹²³ Measure 3 calculates respect for settlement by the percent of retracted appeals in which the defendant went free without trial. In Measure 1, retracted appeals were counted as nonprosecuted. The difference between Measures 1 and 3 is that Measure 1 looks at both nonprosecuted and retracted appeals, whereas Measure 3 looks only at retracted appeals.	99
Sometimes judges, instead of or in addition to asking the jury about the guilt of a nonprosecuted appellee, also asked whether the parties had settled. Since the settlement question was usually asked in order to fine those who had settled, another way of measuring settlement policy is to ask how often judges let the appellee go with neither trial nor inquiry about settlement. The fourth row of Table 9 (Measure 4) measures settlement policy by recording the percentage of nonprosecuted appeals in which judges let the appellee go free without trial or inquiry into settlement.	100
Comparison of the three measures shows broad similarities. All indicate high respect for settlement before 1218. All show a drop in respect for settlement in the 1218–22 eyres and increasing respect from then until the 1234–38 eyres. Starting in 1239, respect begins to drop again, so that by the late 1260s all measures show that less than 10 percent of settlements were respected. Of course, there are some differences. Measure 2 exhibits less respect for settlement in 1218–22 than Measure 1, while Measure 3 shows more. Both statistics, however, are somewhat doubtful since they are based on relatively few observations, as indicated by the absence of asterisks in these cells. The rebound in respect for settlement is larger in Measure 2 than Measure 1 but much more modest in Measure 4. On the other hand, the drop in respect for settlement in the 1240s and 1250s is much smaller under Measure 2 but much sharper under Measure 4. A few cells (Measure 2 between 1246–49 and 1261–63 and Measure 3 in 1250–52) defy the general trends. Nevertheless, for explaining changes in rates of appeal, the similarities are much more important than the differences.	101
Table 9 shows that the results presented in Part 3 are not dependent on a particular way of measuring respect for settlement. Nevertheless, Measure 1, the method used in Part 3.B, is the most plausible way of gauging judicial attitude toward settlement. Measure 2 is less reliable because in more than three-quarters of the cases the rolls do not record whether the parties settled. As a result, Measure 2 captures only part (and potentially an unrepresentative part) of the changes in settlement policy. In addition, the number of cases in the data set for which settlement is recorded is sometimes so small as to make inference unreliable, as indicted by the fact that most cells in this row lack an asterisk. Measure 3 would be a more accurate gauge of settlement policy only if retracted cases were more likely to have been settled than those that were simply nonprosecuted. This is not born out by the data. When jurors reported on whether the parties had settled, they reported that exactly the same percentage (64 percent) of nonprosecuted and retracted cases had settled. In addition, like Measure 2, Measure 3 suffers from the scarcity of relevant cases. Measure 4 is less reliable because inquiry into settlement seldom led to more than minimal fines (half a mark) unless a trial was also held and the appellee was found guilty of the crime for which he was appealed. As a result, inquiry into settlement is not a good measure of a serious antisettlement policy.	102

Part 4. Alternative Explanations for the Changing Rate of Appeals
Part 3 argued that changes in judicial treatment of settled cases caused changes in the number of appeals brought. Other explanations are, of course, possible. Section 3.A analyzed four published theories. Part 4 discusses three others, which have been proposed to me orally and which, in my opinion, deserve careful analysis.	103
A. Crime Trends
One potential explanation is changing crime rates. Perhaps fewer appeals were brought around 1220 or in the later thirteenth century because there were fewer crimes committed. Unfortunately, for most kinds of crime, there are simply no data on the incidence (as opposed to prosecution) of crime. For homicide, however, there are rough incidence figures because the coroner was supposed to investigate every unnatural death and because presenting jurors were fined for not reporting deaths mentioned in the coroners' rolls. Recent scholarship has shown that the coroners' rolls are themselves far from complete,¹²⁴ and the Appendix, Part G, demonstrates that eyre rolls often omit crimes mentioned on the coroners' rolls. Nevertheless, counting homicides on eyre rolls provides the best data on thirteenth-century crime rates.	104
James Given counted homicides (those presented as well as those appealed) from five counties and two cities for much of the thirteenth century.¹²⁵ Table 10 presents his data on the number of homicides per year per county or city.	105

Although Given examined fewer counties and did not cover as wide a time span as considered in this article, the trend in homicide rates is relatively clear. Bedfordshire, Bristol, Kent, London, and Norfolk all show consistently rising homicide rates. The other two counties, Oxfordshire and Warwickshire, show large increases and then smaller declines. These casual observations are confirmed by regression analysis. Table 11 displays the results of a regression very similar to that which generated Table 4. As with Table 5, only the coefficients are reported, and statistical significance at the 0.05 level is indicated by an asterisk.	106

The coefficients show a moderate upward trend. All but one of the coefficients before 1245 is lower than 1.50, whereas all but one after 1245 is above 1.50. Even more significant, the coefficients do not match the pattern described above for appeals. There is no decline from 1201–3 to 1218–22, no increase from 1226–29 to 1239–44, and no decline from 1246–49 to the end of the century. Only the spike in the 1246–49 eyres is at all suggestive of patterns in the rate of appeals. Since the incidence of homicide bears almost no resemblance to the rate of appeal, changes in the homicide rate cannot explain the changes in the number of homicide appeals. The absence of any relationship between homicide rates and rates of appeal is confirmed by regression analysis.¹²⁶	107
In modern times, homicide rates and rates of other crimes generally go up and down together.¹²⁷ There is some evidence that this correlation also held in the early fourteenth century.¹²⁸ If homicide rates and other crime rates were correlated in the thirteenth century as well, then one could infer that the changes in appeals more generally were not caused by changes in the incidence of crime.	108
B. Presentment Trends
Another possibility is that the rate of appeals simply mirrored more general trends in prosecution. Since presentment was the principal alternative method of prosecution, to test this hypothesis, one would need good data on rates of prosecution by presentment. Unfortunately, it is impossible to measure accurately the number of cases brought by presentment because, unlike appeals, a large proportion of presentments were heard in gaol (jail) delivery, and relatively few gaol delivery rolls have survived. As a result, reliance on figures derived from the eyre rolls would almost certainly severely underestimate the number of presentments and generate unreliable trends.	109
Although the precise number of criminal cases brought by presentment cannot be calculated, it is possible to make some rough inferences about the relationship between appeals and presentments. Presentments of assault and rape were extremely rare,¹²⁹ so it is safe to conclude that the rate of appeals of these crimes did not simply mirror trends in presentment. Appeals of theft were never very common and became rare by the end of the thirteenth century.¹³⁰ Presentment of theft, however, was extremely common, especially in the gaol delivery rolls of the late thirteenth century, so it is implausible that trends in the appeal of theft merely tracked more general trends in prosecution.	110
The data gathered by James Given allow a somewhat more precise calculation of trends in presentment of homicide. Although the previous section analyzed Given's data as indicative of crime rates, the data is, nevertheless, prosecution data and could be used to estimate crime incidence rates only because coroners' inquests ensured that most homicides resulted in prosecution. Although some of the homicides counted by Given were prosecuted by appeal, the overwhelming majority were prosecuted by presentment. Therefore, Given's data can be taken as a rough indicator of changing rates of presentment of homicide. There is practically no correlation between Given's rates and rates of appeal.	111
Taken together, these rough analyses of assault, rape, theft, and homicide suggest that trends in appeals did not track more general trends in the prosecution of crime. More generally, the fact that trends in appeals seem to have been independent of trends in prosecution supports the idea that the explanation for changing rates of appeal lies in factors specific to the appeal.	112
The data presented in this section also permit some rough measurement of the relative importance of appeal and presentment. The prevalence of appeals has been measured by calculating the number of appeals per county per year. One might also want to measure the relative importance of the appeal by calculating the percentage of criminal accusations brought by appeal. The rarity of presentments of rape and assault suggests that the appeal was the dominant way in which these crimes were prosecuted, even at the end of the thirteenth century. Conversely, the fact that presentments of theft were extremely common, especially at gaol delivery, suggests that appeals of theft constituted a relatively small proportion of all prosecutions for theft. Since both Given and I analyzed the Bedfordshire 1202, 1227–28, and 1247 eyres, the percentage of homicide cases prosecuted by appeal can be calculated directly. Thirty-six percent (8/22) of the homicide cases reported in the 1202 eyre were brought by appeal, 17 percent (10/58) in the 1227–28 eyre, and 28 percent (19/69) in the 1247 eyre. For the later thirteenth century, no direct comparison can be made because Given did not examine the 1287 Bedfordshire eyre, and I did not look at the 1276–77 Bedfordshire eyre.¹³¹ Nevertheless, if one assumes that the rates of appeal of homicide were similar at the two eyres (as is suggested by the similar coefficients in Table 5), one could estimate that 11 percent of the homicide cases in the 1276–77 Bedfordshire eyre were brought by appeal. These percentages suggest that trends in the rate of appeals (discussed in Part 2) and trends in the percentage of appeals were similar. Both the rates and percentages were high in 1202 and in 1247 and low in 1227–28 and in the late thirteenth century. More generally, Given's data shows an overall upward trend in homicide prosecution, while the rate of appeals shows fluctuations but overall decline. This suggests that the proportion of appeals was significantly lower at the end of the thirteenth century than at the beginning.	113
C. Preappeal Settlement
Part 3 demonstrated that when judges put nonprosecuted appellees to jury trial, the number of appeals declined. It is possible, however, that the appeal did not decline in importance because crime victims did not need to initiate an appeal in order to settle. All they had to do was threaten to appeal. If such threats resulted in settlement before appeal was initiated at county court, the king's suit procedure would not be invoked because it was only triggered if an appeal was initiated. In addition, such threats to appeal, even if followed by settlement, would not be mentioned in legal records. It is thus possible that although the number of recorded appeals dropped, the number of preappeal settlements rose, so that the overall social impact of appeals and settlements induced by the threat of appeals remained constant.	114
Of course, such preappeal settlements would have provided no protection against prosecution by presentment. The danger of presentment would have been quite high in homicide cases and for some kinds of theft. Presentment of homicide was especially likely because dead bodies are hard to conceal, and it was one of the coroner's responsibilities to investigate suspicious deaths. But for the 60 percent of appeals that involved assault, rape, and other crimes, presentment was very infrequent. Therefore, a settlement that prevented an appeal would likely have protected the offender from all punishment. In fact, if the victim and offender were discrete, the presenting jury might never be aware that the crime had occurred.	115
Although it is possible that the number of preappeal settlements rose to offset the decline of actual appeals, this seems unlikely. Victims were required to initiate their cases in the first county court after the offense. Since the county court met every four weeks, victim and offender would have had only a few weeks, and possibly only a few days, in which to settle their dispute. Given the serious nature of the offenses appealed, it seems unlikely that the parties could have come to an agreement so quickly. Physical assaults and rapes may have been caused by long-standing conflicts or quick tempers. But, whatever their cause, people probably required substantial time to put aside their differences and anger in order to settle. The few weeks before the next county court probably did not allow sufficient time for the resolution of such serious matters.	116
The lack of a credible threat provides another reason why preappeal settlement was unlikely. As discussed in Section 1.D, initiation of an appeal provided credibility to the appellor's threat to continue prosecuting the appeal because the appellor was fined if she dropped her suit. On the other hand, before initiation of an appeal, the prospective appellor faced no penalty for failure to appeal and thus may have lacked a credible threat to appeal.	117
V. Conclusion
This article makes two contributions to legal history. The preceding sections have emphasized the substantive results. They chart the changing frequency with which appeals were brought and try to show how the complex pattern can be explained as the result of changes in judicial policy toward settlement. The article also contributes through its method. Although legal historians frequently try to infer patterns from incomplete records, use of formal statistics is rare. It is hoped that this article shows that use of regression analysis can help historians gather new insights from fragmentary evidence.	118
APPENDIX A. Rates of Appeal for Some Additional Districts
To ensure comparability, Table 2 reports the rate of appeals only for those districts for which records are consistently available. Table 12 shows the rate of appeals for districts with odd patterns of survival. As can be seen, the rates are quite low and thus have little effect on the general analysis. To reduce clutter, they were excluded from Table 2, but for completeness they were included in the other tables and in the regressions. For the districts corresponding to each row, see the next part of the Appendix.	119

B. Presenting Districts
The following catalog lists the districts corresponding to the rows of Tables 2 and 12. For some counties, it has been easier to list the presenting districts excluded rather than to enumerate all of those included. To construct the list of those included, consult the tables in David Crook, Records of the General Eyre, 196–252.	120
Bedfordshire: All presenting districts except Dunstable, Houghton Regis. Bedfordshire Ho: Houghton Regis. Buckinghamshire: Ashendon, Cottesloe, Desborough, Mursley, Risborough, Stoke, Stone, Waddesdon, Yardley. Essex: Chelmsford, Dengie, Dunmow, Harlow, Lexden, Ongar, Thurstable. Essex Ro: Rochford. Essex TW: Tendring, Waltham, Witham. Hertfordshire: Baldock, Braughing, Broadwater, Dacorum, Edwinstree, Hertford, Hitchin (half hundred and vill), Odsey, Bishop's Stortford. Kent: All presenting districts except Aylesford Lathe: Brenchley, Chatham and Gillingham, Eyehorne, Littlefield, Maidstone, Twyford, Wrotham; St. Augustine's Lathe: Downhamford, Ringslow, Whitstable; Scray Lathe: Calehill, Felborough, Milton, Teynham; Shipway Lathe: Aloesbridge, Street; Sutton at Home Lathe: Bromley, Codsheath, Dartford, Ruxley; Lowy of Tonbridge: Hildenborough, Tonbridge, Washlingstone. Kent BL: Aylesford Lathe: Brenchley, Littlefield. Lincolnshire: Lindsey, West Riding: Manley, Torksey, Well. Middlesex: Edmonton, Isleworth. Norfolk: "Dodeman" part: Clavering, Freebridge, Humbleyard, Launditch, Smithdon. "Haudry" part: Tunstead. Northamptonshire: Brackley, Cleyley, Corby, Fawsley,¹³² Hamfordshoe, Huxloe, North "Naveslund," South "Naveslund," Nobottle Grove, Spelhoe, Stodfold, Albany Stoke, Towcester, Chipping Warden,¹³³ Wymersley. Northamptonshire Hi: Higham Ferrers. Northamptonshire Wi: Willybrook. Rutland: East, Martinsley. Shropshire: All presenting districts except Alveley, Bradford, Cheswardine, Chirbury, Church Stretton, Clun, Corfham, Ellesmere, Great Ness, Newport, Nordley, Stanton Lacy, Wenlock (borough and liberty), Worthen and Wrockwardine. Shropshire BS: Bradford, Church Stretton. Shropshire Ch: Chirbury. Shropshire NN: Newport, Nordley. Staffordshire: Cuttlestone, Lichfield, Offlow, Penkridge, Pirehill, Seisdon, Stafford, Tamworth, Totmonslow, Tutbury, Wolverhampton. Staffordshire AB: Alrewas, Kings Bromley, Burton upon Trent, Kingswinford, Kinver, Newcastle under Lyme (borough only). Staffordshire Al: Alton. Staffordshire BE: Bradley, Eccleshall, Tettenhall, Wigginton. Staffordshire BH: Brewood, Haywood, Longdon. Staffordshire CP: Cannock and Rugeley, Penkhull. Staffordshire Ma: Maer. Wiltshire: All presenting districts except Corsham, Downton, Knoyle, Marlborough (borough), Ramsbury, Rowde, Old Sarum. Yorkshire: East Riding: Between Ouse and Derwent; North Riding: Pickering (vill and wapentake¹³⁴). Yorkshire HR: North Riding: Hang, Richmond.	121
C. Criteria for Inclusion of District in Data Set
A district was included in the analysis (that is, in either Table 2 or 12) if the earliest surviving eyre roll is complete (not lost or damaged) for that presenting district and if records for that district are complete on some other eyre roll before 1265. Thus, Edmonton and Isleworth in Middlesex were included because records for them are complete for both the 1198 and the 1235 eyres, whereas Spelthorne was excluded because the relevant part of the 1198 eyre roll is damaged, and Uxbridge was excluded because no records survive for 1235.	122
D. Criteria for Classification of a Case as an Appeal
While there is occasionally some ambiguity about whether a given eyre plea roll entry is an appeal, appeals are generally easy to identify because they either contain some form of the Latin verb appellare or the Latin noun appellum, or because they describe outlawry at the suit of (per sectam) a particular person. I have excluded approver appeals. I have also excluded plaints, which are usually identifiable by the use of a form of the Latin verb queror. Inclusion of plaints would not have substantially affected the analysis because there were very few of them, usually less than one per county per year.	123
It is somewhat more difficult to identify appeals in gaol delivery records because the words appellare and appellum are seldom used. When analyzing gaol delivery rolls, I have counted as appeals all cases that use the phrase "captured at the suit" (captus ad sectam) of an individual who is not an approver.	124
E. Definition of a "Case"
Throughout this article, I have used the "case" as a unit of analysis. I have counted all prosecutions for the same allegedly criminal incident as a single appeal, even though medieval clerks and modern editors sometimes recorded separately (a) multiple prosecutions against a single individual for the same allegedly criminal incident, and (b) the prosecutions of a single person against multiple offenders for the same incident.	125
F. Sources Used in the Database
Bedfordshire 1202. JUST 1/1, mm. 5–6; G. Herbert Fowler, ed., "Roll of the Justices in Eyre at Bedford, 1202," in the Publications of the Bedfordshire Historical Record Society (Bedfordshire Historical Record Society, vol. 1, 1913), 214–47. Bedfordshire 1227–28. JUST 1/2, mm. 13–15d; G. Herbert Fowler, ed., "Roll of the Justices in Eyre at Bedford, 1227" in the Publications of the Bedfordshire Historical Record Society (Bedfordshire Historical Record Society, vol. 3, 1916), 141–74. Bedfordshire 1247. JUST 1/4, mm. 26–34. Bedfordshire 1287. JUST 1/13, mm. 19–30d. Buckinghamshire 1195. KB 26/4, mm. 5–5d; Frederic William Maitland, ed., Three Rolls of the King's Court in the Reign of King Richard the First, A.D. 1194–1195 (London: Pipe Roll Society, vol. 14, 1891), 137–48. Buckinghamshire 1227. JUST 1/54, mm. 15–19d; J. G. Jenkins, ed., Calendar of the Roll of the Justices on Eyre, 1227 (Buckinghamshire Archaeological Society, Record Branch, vol. 6, 1945), 46–64. Buckinghamshire 1232. JUST 1/62, mm. 1–8d. Buckinghamshire 1241. JUST 1/55, mm. 20–28d. Buckinghamshire 1247. JUST 1/56, mm. 35–46d. Buckinghamshire 1262. JUST 1/58, mm. 20–29. Buckinghamshire 1272. JUST 1/60, mm. 22–32d. Buckinghamshire 1286. JUST 1/66, mm. 1–13. Essex 1198. KB 26/9, mm. 8–8d; Sir Francis Palgrave, ed., Rotuli Curiae Regis (London: G. Eyre and H. Spottiswoode, 1835), 1:202–11. Essex 1227–29. JUST 1/229, mm. 14–18. Essex 1235. JUST 1/230, mm. 1–10d. Essex 1248. JUST 1/232. Essex 1254. JUST 1/233, mm. 41–58d. Essex 1262. JUST 1/236B. Essex 1272. JUST 1/238, mm. 46–60d. Essex 1285. JUST 1/247, mm. 1–40d. Hertfordshire 1198. KB26/9, mm. 1–1d; Palgrave, Rotuli Curiae Regis, 1:159–65. Hertfordshire 1248. JUST 1/318, mm. 19–27. Kent 1227. JUST 1/358, mm. 16–27d. Kent 1241. JUST 1/359, mm. 27–37d. Kent 1255. JUST 1/361, mm. 35–62d. Kent 1262–63. JUST 1/1582. Lincolnshire 1202. JUST 1/479; Doris Mary Stenton, ed., The Earliest Lincolnshire Assize Rolls, A.D. 1202–1209 (Lincoln Record Society, vol. 22, 1926), 93–168. Lincolnshire 1245. JUST 1/1581. Middlesex 1198. KB 26/9, mm. 9–9d; Palgrave, Rotuli Curiae Regis, 1:214–18. Middlesex 1235. JUST 1/536, mm. 6–8d. Norfolk 1198. JUST 1/559; Doris Mary Stenton, ed., Pleas Before the King or His Justices, 1198–1202 (London: Selden Society, vol. 68, 1952), 2:1–15. Norfolk 1250. JUST 1/565, mm. 2–36d. Norfolk 1257. JUST 1/568. Northamptonshire 1202. JUST 1/613, mm. 1–3; Doris Mary Stenton, ed., The Earliest Northamptonshire Assize Rolls: A.D. 1202 and 1203 (London: Northamptonshire Record Society, vol. 5, 1930), 1–18. Northamptonshire 1232. JUST 1/614A. Northamptonshire 1247. JUST 1/614B, mm. 36–49. Northamptonshire 1253. JUST 1/615, mm. 1–14d. Rutland 1202. JUST 1/613, m. 3d; Stenton, The Earliest Northamptonshire Assize Rolls, 18–21. Rutland 1263. JUST 1/721, mm. 9–13d. Shropshire 1203. JUST 1/732, mm. 2–3; Doris Mary Stenton, ed., Pleas Before the King or His Justices, 1198–1212 (London: Selden Society, vol. 83, 1967), 3:69–87. Shropshire 1221. JUST 1/733A, mm. 9–11d; Doris Mary Stenton, ed., Rolls of the Justices in Eyre Being the Rolls of Pleas and Assizes for Gloucestershire, Warwickshire and Staffordshire, 1221, 1222 (London: Selden Society, vol. 59, 1940), 533–68. Shropshire 1248. JUST 1/733B, mm. 2–5d. Shropshire 1256. JUST 1/734, mm. 21–32d; Alan Harding, ed., The Roll of the Shropshire Eyre of 1256 (London: Selden Society, vol. 96, 1981), 196–309. Staffordshire 1199. JUST 1/800, mm. 2–3; G. Wrottesley, ed., Collections for a History of Staffordshire (London: William Salt Archaeological Society, later Staffordshire Record Society, vol. 3, 1882), 38–46. Staffordshire 1203. JUST 1/799, mm. 2–3; Wrottesley, Collections for a History of Staffordshire, 91–98. Staffordshire 1227. JUST 1/801, mm. 10–14d. Staffordshire 1272. JUST 1/802, mm. 41–58; JUST 1/803, mm. 26–40. Staffordshire 1293. JUST 1/806, mm. 1–29. Wiltshire 1194. KB26/3, mm. 3–7d; Maitland, Three Rolls of the King's Court in the Reign of King Richard the First, 77–115. Wiltshire 1249. JUST 1/996, mm. 23–40d; C. A. F. Meekings, ed., Crown Pleas of the Wiltshire Eyre, 1249 (Devizes: Wiltshire Archaeological and Natural History Society, vol. 16, 1961), 152–258. Yorkshire 1208. JUST 1/1039, mm. 6–8; Doris Mary Stenton, ed., Pleas Before the King or His Justices, 1198–1212 (London: Selden Society, vol. 84, 1967), 4:94–117. Yorkshire 1218–19. JUST 1/1053; Doris Mary Stenton, ed., Rolls of the Justices in Eyre Being Rolls of Pleas and Assizes for Yorkshire in 3 Henry III (1218–19), (London: Selden Society, vol. 56, 1937), 183–389. Yorkshire 1231. JUST 1/1043. Yorkshire 1257. JUST 1/1109, mm. 2–37d. Yorkshire 1268–69. JUST 1/1051, mm. 1–45d. Yorkshire 1279–81. JUST 1/1070, mm. 1–55d.	126
G. Reliability of Eyre Records
Because eyre rolls provide the figures for most of the quantitative analysis in this article, it is important to examine their reliability.¹³⁵ Coroners' rolls present the best source for this purpose. Nearly all appeals were initiated in county court.¹³⁶ Since the coroner kept records of criminal matters raised in the county court, one can calculate the completeness of eyre records by comparing them to coroner records. Unfortunately, relatively few coroners' rolls survive from the relevant period, and all but one date from the late thirteenth century.¹³⁷ These rolls show that just over 60 percent of all appeals initiated in county court were recorded in eyre rolls.¹³⁸ This underreporting can be attributed to several causes. Some appeals initiated in county court were heard at gaol delivery, in the Bench, coram rege, or before special commissions. While the disposition of such cases was sometimes recorded in eyre rolls,¹³⁹ many were not. In addition, some cases recorded as appeals in the coroners' rolls were mentioned in the eyre rolls but as simple presentments without any mention of an appeal.¹⁴⁰ These reasons may account for all of the missing appeals. Nevertheless, because of incomplete record survival, especially the disappearance of all gaol delivery rolls for the counties and years corresponding to the surviving coroners' rolls, we cannot be sure. Maybe some appeals were simply not recorded on eyre rolls. This is unlikely because the coroners handed in their rolls at the beginning of each eyre and eyre justices consulted the coroners' rolls to ensure that presenting jurors did not conceal criminal cases. Since eyre rolls are meticulous about financial matters, and since nearly every appeal would result in some sort of revenue,¹⁴¹ one would think that almost all appeals would be recorded.¹⁴² Nevertheless, because there may have been underreporting of appeals in eyre rolls, it is important to consider whether this could render invalid the trends identified in Part 2. There are several situations to consider.	127
First, suppose that eyre rolls from the early and mid-thirteenth century were about as complete as rolls from the later part of the century. That is, suppose each eyre roll compiled between 1194 and 1294 recorded roughly two-thirds of all appeals initiated in county court. Then all of the rates in Part 2 should be increased by about 50 percent. Nevertheless, since all eyres would be equally affected, the trends identified in Part 2 would remain exactly the same.	128
Next, suppose that reporting improved over time. In general, that would reinforce the trends identified in Part 2. The sharp declines around 1220 and after 1250 would be even bigger, because the number of reported appeals would have been declining even though a greater fraction of all appeals was being reported. On the other hand, the rebound from the 1220s to the 1240s would not have been as large, and the rate of appeal might not have attained the same level in the 1240s as at the turn of the century. Nevertheless, it is nearly inconceivable that changes in the quality of reporting could completely eliminate the rebound. For example, if the 1247 Bedfordshire eyre roll recorded 50 percent of all appeals, then the 1227–28 eyre would have had to record less than 16 percent of all appeals in order to erase the apparent increase from 1227–28 to 1247.	129
Similar reasoning suggests that the trends outlined in Part 2 remain valid even if reporting worsened over the thirteenth century and even if the quality of reporting remained the same during some periods, worsened during some periods, and improved in others.	130
Daniel Klerman is an associate professor at the University of Southern California Law School. He is especially grateful to Richard Helmholz, who guided this project as the chairman of his dissertation committee, and to Paul Brand, who answered numerous questions and provided copious and extremely helpful comments on a draft of this article. He also thanks Albert Alschuler, John Baker, Robert Bartlett, John Bellamy, Lisa Bernstein, Julie Bush, Emily Buss, Isaiah Cox, Richard Craswell, Charles Donahue, Jr., Mary Dudziak, Charles Duggan, George Fisher, David Friedman, Thomas Gallanis, Jason Glenn, Jack Goldsmith, Charles Gray, Laura Greco, Thomas Green, Roger Groot, Alon Harel, John Hudson, Margaret Kerr, Julius Kirshner, Jacob Klerman, Lisa Klerman, Lorraine Klerman, John Langbein, Jonathan Lehrich, Bentley MacLeod, Kyle Mayer, Tracey Meares, Peter McCullagh, Robert Palmer, Richard Posner, Eric Rasmusen, Chris Sanchirico, Stephen Schulhofer, Steven Shavell, Matthew Spitzer, Robin Stacey, Henry Summerson, Eric Talley, Eugene Volokh and Charles Whitebread for their advice, assistance, and critical comments at various stages of this project. Three anonymous referees for Law and History Review also provided much helpful criticism. This research was supported by a Fulbright scholarship, the Herbert and Marjorie Fried Faculty Research Fund at the University of Chicago Law School, the National Science Foundation (Law and Social Science Program, grant # SBR-9412044), the Robert B. Roesing Faculty Fund at the University of Chicago Law School, the Social Science Research Council, and a USC Law School Summer Research grant Notes ¹ Here and elsewhere I use the term "crime" somewhat informally to refer to the type of offenses that were privately prosecuted in thirteenth-century England, including homicide, rape, robbery, larceny, burglary, and assault. Some legal systems, including England's perhaps into the thirteenth century, did not distinguish (or did not distinguish sharply) between civil and criminal cases. ² Appeals have, for example, merited discussion in nearly every comprehensive work on the history of English law. See, e.g., J. H. Baker, An Introduction to English Legal History, 3d ed. (London: Butterworth Legal Publishers, 1990), 574–76; Sir William Holdsworth, A History of English Law, 4th ed. (London: Methuen, 1936), 2:256–57, 361–64; S. F. C. Milsom, Historical Foundations of the Common Law, 2d ed. (Toronto: Butterworths, 1981), 406–10; Theodore F. T. Plucknett, A Concise History of the Common Law, 5th ed. (Boston: Little, Brown, 1956), 428. Several recent articles also analyze appeals. Daniel R. Ernst, "The Moribund Appeal of Death: Compensating Survivors and Controlling Jurors in Early Modern England," American Journal of Legal History 28 (1984): 164–88; Roger D. Groot, "The Jury in Private Criminal Prosecutions before 1215," American Journal of Legal History 27 (1983): 113–41; Margaret H. Kerr, "Angevin Reform of the Appeal of Felony," Law and History Review 13 (1995): 351–92; Christopher Whittick, "The Role of the Criminal Appeal in the Fifteenth Century," in Law and Social Change in British History: Papers Presented to the Bristol Legal History Conference, 14–17 July 1981, ed. J. A. Guy and H. G. Beale (London: Royal Historical Society, 1984), 55–72. ³ There is, however, a growing literature on such settlements. Much of it focuses on societies in which, unlike thirteenth-century England, feud flourished. Christopher Boehm, Blood Revenge: The Enactment and Management of Conflict in Montenegro and Other Tribal Societies (Lawrence: University Press of Kansas, 1984), 121–42; William Ian Miller, Bloodtaking and Peacemaking: Feud, Law, and Society in Saga Iceland (Chicago: University of Chicago Press, 1990), 259–99; J. M. Wallace-Hadrill, "The Bloodfeud of the Franks," in The Long-Haired Kings and Other Studies in Frankish History (New York: Barnes and Noble, 1962), 121–47. There is also some discussion of settlement of criminal disputes by English legal historians. See, e.g., Douglas Hay, "Property, Authority, and the Criminal Law," in Albion's Fatal Tree: Crime and Society in Eighteenth-Century England, ed. Douglas Hay et al. (London: Allen Lane, 1975), 41–42; Norma Landau, "Indictment for Fun and Profit: A Prosecutor's Reward at Eighteenth-Century Quarter Sessions," Law and History Review 17 (1999): 507–36. See also the articles cited above, note 2. ⁴ Presentment was accusation by a jury, which could be considered a form of public prosecution. See below, 5–7. ⁵ Frederick C. Hamil, "The King's Approvers: A Chapter in the History of English Criminal Law," Speculum 11 (1936): 238–58; A. J. Musson, "Turning King's Evidence: The Prosecution of Crime in Late Medieval England," Oxford Journal of Legal Studies 19 (1999): 468–79; Jens Röhrkasten, Die Englischen Kronzeugen, 1130–1330 (Berlin: Duncker and Humblot, 1990). ⁶ Frederick Pollock and Frederic William Maitland, The History of English Law before the Time of Edward I, 2d ed. (Cambridge: Cambridge University Press, 1968), 2:450–51; Stanley Rubin, "Bot Compensation in Anglo-Saxon Law: A Reassessment," Journal of Legal History 17 (1996): 144–54. ⁷ Naomi D. Hurnard, "The Jury of Presentment and the Assize of Clarendon," English Historical Review 56 (1941): 374, 376–96; Patrick Wormald, "Frederic William Maitland and the Earliest English Law," Law and History Review 16 (1998): 11–12. ⁸ For more about presentment, see Raoul C. Van Caenegem, "Public Prosecution of Crime in Twelfth-Century England," in Church and Government in the Middle Ages: Essays Presented to C. R. Cheney, ed. C. N. L. Brooke et al. (Cambridge: Cambridge University Press, 1976), 44–49. That juries during the thirteenth century were self-informing represents the consensus of legal historians. Thomas Andrew Green, Verdict According to Conscience: Perspectives on the English Criminal Trial Jury, 1200–1800 (Chicago: University of Chicago Press, 1985), 16–17. Recently, some scholars have challenged this consensus by compiling evidence that fourteenth- and fifteen-century juries seldom lived close to the defendant and thus were unlikely to have had knowledge of the case before trial. See chapters by Bernard William Lane, Edward Powell, and J. B. Post in Twelve Good Men and True: The Criminal Trial Jury in England, 1200–1800, ed. J. S. Cockburn and Thomas A. Green (Princeton: Princeton University Press, 1988). Nevertheless, most scholars remain convinced that the thirteenth-century jury, and probably the fourteenth-century jury as well, remained heavily self-informing and only occasionally relied on in-court testimony by witnesses other than officials, such as reeves, coroners, and justices of the peace. See Thomas A. Green, "A Retrospective on the Criminal Trial Jury, 1200–1800," in Twelve Good Men and True, 370–77; John H. Langbein, "Historical Foundations of the Law of Evidence: A View from the Ryder Sources," Columbia Law Review 96 (1996): 1168, 1170 n. 6; Anthony Musson, Public Order and Law Enforcement: The Local Administration of Criminal Justice, 1294–1350 (Woodbridge, Suffolk, UK: Boydell Press, 1996), 201, 22–21; J. G. Bellamy, The Criminal Trial in Later Medieval England: Felony before the Courts from Edward I to the Sixteenth Century (Toronto: University of Toronto Press, 1998), 98, 101–5. ⁹ Wormald, "Frederic William Maitland and the Earliest English Law," 17. Maitland, however, thought that the system of compensatory payments survived until the twelfth century. Pollock and Maitland, History of English Law, 2:458. ¹⁰ Wormald, "Frederic William Maitland and the Earliest English Law," 17–18. ¹¹ C. A. F. Meekings, ed., "Introduction," The 1235 Surrey Eyre (Castle Arch, Great Britain: Surrey Record Society, vol. 31, 1979), 1:106, 114; Barbara A. Hanawalt, Crime and Conflict in English Communities, 1300–1348 (Cambridge: Harvard University Press, 1979), 66. Presentment may have been used for other offenses at sheriff's tourn or view of frankpledge. ¹² Mayhem was the infliction of a disabling but nonlethal injury. ¹³ Green, Verdict According to Conscience, 108–13, and "A Retrospective on the Criminal Trial Jury, 1200–1800," 367–75. ¹⁴ Bellamy, The Criminal Trial in Later Medieval England, 19–27. Bellamy suggests that the bill procedure started as early as the late thirteenth century. ¹⁵ John H. Langbein, "Understanding the Short History of Plea Bargaining," Law and Society 13 (1979): 266–67; Douglas Hay, "Controlling the English Prosecutor," Osgoode Hall Law Journal 21 (1983): 167–74. ¹⁶ R. F. Hunnisett, The Medieval Coroner (Cambridge: Cambridge University Press, 1961), 1–36. ¹⁷ Bellamy, The Criminal Trial in Later Medieval England, 26, 51 n. 40; Green, Verdict According to Conscience, 109–13; John H. Langbein, Prosecuting Crime in the Renaissance: England, Germany, France (Cambridge: Harvard University Press, 1974), 63–97. ¹⁸ Hay, "Controlling the English Prosecutor," 174–80; Patrick Devlin, The Criminal Prosecution in England (New Haven: Yale University Press, 1960), 20. ¹⁹ Prosecution of Offenses Act, 1879, 42 & 43 Vict., c. 22; Philip B. Kurland and D. W. M. Waters, "Public Prosecutions in England, 1854–79: An Essay in English Legislative History," Duke Law Journal (1959): 493–562. ²⁰ "Proposed Independent Prosecuting Service: The Prosecutor's Viewpoint," Journal of Criminal Law 48 (1984): 302–3. The public prosecutors' power was somewhat increased by the Prosecution of Offenses Act, 1908, 8 Edw. 7, c. 3, but even so the power of public prosecutors remained small. Ibid.; Hay, "Controlling the English Prosecutor," 179–80. ²¹ Prosecution of Offenses Act, 1985, c. 23, sec. 6(1); Alec Samuels, "Non-Crown Prosecutions: Prosecutions by Non-Police Agencies and by Private Individuals," Criminal Law Review (1986): 33–36. ²² Lawrence M. Friedman, Crime and Punishment in American History (New York: Basic Books, 1993), 29–30. ²³ Rape was probably not presentable until the 1275 enactment of the first Statute of Westminster. See J. B. Post, "Ravishment of Women and the Statutes of Westminster," in Legal Records and the Historian, ed. J. H. Baker (London: Royal Historical Society, 1978), 154–55; Henry Ansgar Kelly, "Statutes of Rapes and Alleged Ravishers of Wives: A Context for the Charges against Thomas Malory, Knight," Viator 28 (1997): 364–66, 382–83, 387–88; Roger D. Groot, "The Crime of Rape temp. Richard I and John," Journal of Legal History 9 (1988): 325–26. ²⁴ But see Ruth Kittel, "Rape in Thirteenth-Century England: A Study of the Common-Law Courts," in Women and the Law: A Social Historical Perspective, ed. D. Kelly Weisberg (Cambridge, Mass.: Schenkman, 1982) 2:102 n. 9. ²⁵ George E. Woodbine, ed., Bracton on the Laws and Customs of England, trans. Samuel. E. Thorne (Cambridge: Belknap Press of Harvard University Press, 1968), 2:394, fols. 139b–40. ²⁶ In several northern and eastern counties, including Yorkshire, the court met every six weeks. Robert C. Palmer, The County Courts of Medieval England: 1150–1350 (Princeton: Princeton University Press, 1982), 4. ²⁷ Daniel Klerman, "Female Private Prosecutors in Thirteenth-Century England" (unpublished manuscript). ²⁸ Bracton, 2:353, fol. 125b. For a more thorough discussion of representation in appeals, see Klerman, "Female Private Prosecutors." ²⁹ Bracton, 2:354, fol. 125–125b. ³⁰ Ibid., 2:354, 361–62, fols. 125b, 128b; J. M. Kaye, ed. and trans., Placita Corone or La Corone Pledee Devant Justices (London: Selden Society, Selden Society Supplementary Series, vol. 4, 1966), 25. Bracton argues that an outlaw could be killed only if he fled or resisted arrest, although he acknowledges contrary custom and authority. Bracton, 2:354, 362, 378, fols. 125b, 128b, 134. ³¹ In general, I use feminine pronouns for appellors and male pronouns for appellees. This helps to distinguish appellors and appellees and is historically plausible, because a substantial fraction of appellors were women. See above, 10. ³² Roger D. Groot, "The Early Thirteenth-Century Criminal Jury" in Twelve Good Men and True, ed. Cockburn and Green, 3–35. ³³ Groot, "The Jury in Private Criminal Prosecutions before 1215," 113–41. Medial verdicts could be procured by the writ de odio et atia, but the writ was not necessary. Ibid. ³⁴ Margaret H. Kerr, Richard D. Forsyth, and Michael L. Plyley, "Cold Water and Hot Iron: Trial by Ordeal in England," Journal of Interdisciplinary History 22 (1992): 578–79. ³⁵ Paul Hyams, "The Strange Case of Thomas of Eldersfield," History Today 36 (1986): 9–15. ³⁶ The attentive reader will note that only 93 percent of appeals have been accounted for: 18 percent outlawed, 18 percent pled by the plaintiff, 57 percent nonprosecuted or retracted. In the remaining 7 percent, either the prosecutor died or the case assumed an unusual procedural posture that cannot be simply classified. Here and elsewhere I group nonprosecuted and retracted appeals together. See below, 17 n. 50, 39, 51, 52. ³⁷ In addition to formally acquitting the defendant, the judges could let the defendant go "without day" (sine die) or without any judgment at all. While not technically acquittals, such judgments (or nonjudgments) effectively freed the defendant. In general I treat them as equivalent to acquittals, because I have seen no cases in which a defendant so released was subsequently reprosecuted by appeal or otherwise punished. ³⁸ Ordinarily, appellors had to find people willing to assure that the appellor would prosecute the case and pay fines if she did not. When the appellor was poor, that requirement was waived and a simple oath to prosecute was deemed sufficient. ³⁹ Collections for a History of Staffordshire, ed. W. Wrottesley (London: The William Salt Archaeological Society, vol. 3, 1882), 41. Translation by the author. ⁴⁰ John probably asked Ivo and the others for a pledge that they would show up at court, if sued for fishing without permission. ⁴¹ The tithing probably secured his release, pending trial, upon a promise that they would ensure his presence at trial. Every adult male was required to be in a tithing, a group whose most important function was producing its members' attendance in court when necessary. ⁴² JUST 1/4, m. 30. All citations to manuscript sources refer to documents in the Public Record Office, Kew, England. ⁴³ JUST 1/359, m. 35d. ⁴⁴ See, e.g., Doris Mary Stenton, ed., Pleas before the King or His Justices, 1198–1212 (London: Selden Society, vol. 83, 1967), 3: pl. 746 (Shropshire 1203) (10 marks to settle mayhem appeal); Alan Harding, ed., The Roll of the Shropshire Eyre of 1256 (London: Selden Society, vol. 96, 1981), pl. 577 (40 shillings to settle false imprisonment and robbery appeal). ⁴⁵ See, e.g., Doris Mary Stenton, ed., Rolls of the Justices in Eyre Being Rolls of Pleas and Assizes for Yorkshire in 3 Henry III (1218–19) (London: Selden Society, vol. 56, 1937), pls. 959, 1086. ⁴⁶ For an example of a rape appeal following consensual sex, see ibid., pl. 669 (the jurors say that "he had her with her good will for a year and that he took another to wife and for this reason she has appealed him"). For a more thorough discussion of settlement by marriage, see Daniel Klerman, "Female Private Prosecutors." ⁴⁷ Stenton, Pleas before the King or His Justices, 1198–1212, 3: pls. 671, 746 (Shropshire 1203). ⁴⁸ CP 25(1)/212/6 no. 39. I am grateful to Paul Brand for finding this final concord and sharing his notes on it with me. ⁴⁹ Of the 1249 cases in the data set described in Section 2.A, information on settlement is recorded for 308 or 25 percent. Of these, 207 (67 percent) settled. So at least 17 percent (207/1249) of all appeals settled. ⁵⁰ There were 677 nonprosecuted cases in the data set described in Section 2.A. If two thirds of the nonprosecuted cases settled, there would be 452 settled cases, which is 36 percent (452/1249) of all appeals. Here, as elsewhere, the count of nonprosecuted appeals include retracted ones. See above, 12 n. 36, below, 39, 51, 52. Settled cases that cannot be classified as nonprosecuted include cases that the appellor prosecuted in the eyre in spite of settlement and cases in which the appellor died after having settled but before trial in the eyre. ⁵¹ For a more thorough analysis of which cases settled and why, see Daniel Klerman, "The Selection of Thirteenth-Century Criminal Disputes for Litigation" (unpublished manuscript). ⁵² When the parties received a "license to concord" (judicial permission to settle), which was rare, judges would quash later prosecutions. See Stenton, Pleas before the King or His Justices, 1198–1212, 3: pl. 746 (Shropshire 1203). Informal settlements would appear on the records as nonprosecuted or retracted appeals. The fact of a prior nonprosecuted or retracted appeal was sometimes raised as a defense to a subsequent prosecution, and that defense seems to have been accepted. Ibid., pl. 726; G. Herbert Fowler, ed., "Roll of the Justices in Eyre at Bedford, 1227," in Publications of the Bedfordshire Historical Record Society (Bedfordshire Historical Record Society, vol. 3, 1916), 1: pl. 397. Nevertheless, these cases do not prove the enforceability of out-of-court settlements, because they involve an appellor who brought a second appeal, rather than an appellor who decided simply to continue her original appeal. The former situation presented the judge with additional reasons to protect the appellee, because the second appeal was brought too late (not at the first county court) and because the judgment on the first appeal was seen as barring subsequent appeals, not unlike modern res judicata. ⁵³ See JUST 1/1043, m. 4d (Yorkshire 1231) (appellor's brother was present at making of settlement); Stenton, Pleas before the King or His Justices, 1198–1212, 3: pl. 690 (Shropshire 1203) (compensation determined "by the view and judgment of lawful men"). ⁵⁴ Stenton, Pleas before the King or His Justices, 1198–1212, 3: pl. 743. ⁵⁵ C. A. F. Meekings, ed., Crown Pleas of the Wiltshire Eyre, 1249 (Devizes: Wiltshire Archaeological and Natural History Society, Records Branch, vol. 16, 1961), pls. 44–45. ⁵⁶ JUST 1/536, m. 8 (Middlesex 1235). ⁵⁷ JUST 1/565, m. 21 (Norfolk 1250). ⁵⁸ See S. F. C. Milsom, The Legal Framework of English Feudalism: The Maitland Lectures Given in 1972 (Cambridge: Cambridge University Press, 1976), 168 (observing that "in many early appeals" the appellee was "a lord enforcing his rights"). ⁵⁹ See, e.g., Meekings, Crown Pleas of the Wiltshire Eyre, 1249, pl. 169; Harding, The Roll of the Shropshire Eyre of 1256, pl. 747. ⁶⁰ On distraint, see Baker, An Introduction to English Legal History, 271–72. ⁶¹ JUST 1/361, m. 60d (Kent 1255). ⁶² JUST 1/4, m. 34 (Bedfordshire 1247). ⁶³ Frederic William Maitland, The Forms of Action at Common Law: A Course of Lectures, ed. A. H. Chaytor and W. J. Whittaker (Cambridge: Cambridge University Press, 1936; First published with Maitland, Equity, 1909; published separately, 1936), 48–49. See also Pollock and Maitland, A History of English Law, 2:485 ("[t]o the end of our period [1307] an appeal rather than an indictment is the normal procedure against criminals"). ⁶⁴ Holdsworth, A History of English Law, 2:257 (in the thirteenth century, the appeal was "gradually decaying as a mode of criminal prosecution"); Hunnisett, The Medieval Coroner, 55 ("during the thirteenth century the number of appeals rapidly declined"); Meekings, Crown Pleas of the Wiltshire Eyre, 1249, 35 ("The ordinary appeal was declining in importance throughout the latter half of the thirteenth century"); Ernst, "The Moribund Appeal of Death," 164, 165 (opining decline 1215–1500). See also sources in notes 79–80 below. Baker, Milsom, and Plucknett pass over in silence the issue of when the appeal declined. Baker, Introduction to English Legal History, 574–76; Plucknett, A Concise History of the Common Law, 428; Milsom, Historical Foundations of the Common Law, 406–10. ⁶⁵ The sources used for this database are listed in Appendix F. The reliability of these records is discussed in Appendix G. The data set itself can be downloaded from the Inter-university Consortium for Political and Social Research (ICPSR) website <www.icpsr.umich.edu> or the University of Southern California Center for Law, Economics, and Organization (USC CLEO) website <www.usc.edu/dept/law/centers>. ⁶⁶ David Crook, "The Later Eyres," English Historical Review 97 (1982): 241–68. ⁶⁷ Although post-1263 Bedfordshire records were examined, the 1276–77 Bedfordshire eyre was excluded, because it followed the 1272 eyre, which was abandoned on Henry III's death. See David Crook, Records of the General Eyre, Public Record Office Handbooks no. 20 (London: Her Majesty's Stationery Office, 1982), 134. As a result, it is unclear whether cases arising between 1262 and 1272 were consistently reported in the 1276–77 eyre. ⁶⁸ Although appeals from only eighteen of thirty-four Shropshire districts and eleven of thirty-six Staffordshire districts were included in Table 2, the excluded districts generally reported very few or no appeals when they did report. As a result, Table 2 is nearly complete for these counties. Table 12 reports on most of the excluded districts. The vast majority of the others returned no appeals at each eyre at which they reported. ⁶⁹ M. T. Clanchy, From Memory to Written Record: England 1066–1307, 2d ed. (Oxford: Blackwell Publishers, 1993), 71. ⁷⁰ For an introduction to regression analysis, see David S. Moore and George P. McCabe, Introduction to the Practice of Statistics (New York: W. H. Freeman, 1989), chap. 10. ⁷¹ If one excludes Kent, the coefficient for 1226–29 drops to 0.42 and the p-value drops to 0.000. ⁷² It is possible that the introduction of rape presentments (1275 Statute of Westminster I) contributed to the post-1275 decline in rape appeals. The provision in Westminster II (1285) for trespass writs for rape/ravishment almost certainly had no effect on rape appeals during the period studied, because they were ordinarily used to punish "ravishment of wife," rather than rape of an unmarried women (as was typical in appeals). In addition, such trespass writs did not become common until the turn of the fourteenth century. See J. B. Post, "Ravishment of Women and the Statutes of Westminster," in Legal Records and the Historian, ed. J. H. Baker (London: Royal Historical Society, 1978), 159. ⁷³ For a more thorough discussion of the issues discussed in this section, see chapter 3 of my dissertation: Daniel Klerman, Private Prosecution of Crime in Thirteenth-Century England (Ann Arbor, Mich.: UMI Dissertation Services, 1998). ⁷⁴ Although these were the principal places other than the eyre where appeals could be tried, appeals were sometimes heard elsewhere. Theft appeals could be heard in nonroyal courts that had the franchise of infangthief. In addition, in the late thirteenth century, commissions were sometimes issued to a particular group of justices to hear and determine a particular appeal. Perusal of the Calendars of Patent Rolls revealed no such commissions in 1245 or 1246, three in 1275, sixteen in 1280, and thirty-three in 1285. Thus, although the number of such commissions was increasing, even in 1285 they averaged less than one per county per year. In addition, the increase came too late to explain the decline of the appeal, which started no later than the 1250s. For appeals not heard in eyre, see also JUST 1/1179, m. 4 (appeal heard at 1252 assize at Greenwich, Kent); JUST 1/13, mm. 19, 21d (two appeals heard before justices with oyer and terminer commissions). ⁷⁵ Doris Mary Stenton, ed., The Earliest Northamptonshire Assize Rolls: a.d. 1202 and 1203 (London: Northamptonshire Record Society, 1930), 5:99–131, 153–63 (Northamptonshire, two sessions in autumn and summer 1203, three appeals total, which is three per year); ibid., 131–53 (Suffolk 1203, two sessions at St. Edmunds and Ipswich, two appeals total, which is two per year); Doris Mary Stenton, ed., The Earliest Lincolnshire Assize Rolls, a.d. 1202–1209 (Lincoln: Lincoln Record Society, vol. 22, 1926), 266–71 (Lincolnshire 1206, Five appeals, which is two and a half per year). To calculate these rates, the number of appeals recorded in the gaol delivery rolls was divided by the time between the gaol delivery and the previous time the pipe rolls record that royal justices had heard criminal cases in the county. Other gaol delivery records from the early thirteenth century indicate similar or smaller numbers of appeals. Doris Mary Stenton, ed., Pleas before the King or His Justices, 1198–1202 (London: Selden Society, vol. 68, 1952), 2:176–78 (Cornwall 1201, two appeals); Curia Regis Rolls (London: Her Majesty's Stationery Office, 1952), 9:198–201 (Herefordshire 1220, two appeals); ibid., 11:118 (Oxfordshire 1223, no appeals); ibid., 381 (Herefordshire 1224, two appeals); ibid., 382–83 (Worcestershire 1224, no appeals); JUST 1/36, mm. 2d–7 (Berkshire 1225, one appeal); C. E. H. Chadwyck-Healey, ed., Somersetshire Pleas (Civil and Criminal) from the Rolls of the Itinerant Justices (London: Somerset Record Society, vol. 11, 1897), 1:28–85 (Somerset 1225, four appeals); JUST 1/863, mm. 3d–4d (Surrey 1225, three appeals); JUST 1/1172, m. 5 (Shropshire 1226, one appeal); JUST 1/801, m. 10 (Staffordshire 1227, no appeals). It is important to note the small number of appeals heard in the 1225 sessions. In that year, most English counties were visited by royal judges who heard assizes and delivered jails. If they heard all of the appeals pending in the county, however, that could significantly undermine the figures presented in Section 2.B for the 1226–29 eyres, because I assumed that the 1226–29 eyres heard appeals initiated since the 1218–22 eyres. The small number of appeals heard in the 1225 sessions argues strongly that they did not hear all appeals that had arisen since the 1218–22 eyres. Thus, the figures presented in Section 2.B are substantially accurate. This conclusion is reinforced by the fact that, even though royal justices did not visit Staffordshire in their 1225 sessions (see C. A. F. Meekings, "Introduction," Curia Regis Rolls, 12:xi), the 1227 Staffordshire eyre reveals a substantially reduced rate of appeal. ⁷⁶ Ralph B. Pugh, ed., Wiltshire Gaol Delivery and Trailbaston Trials, 1275–1306 (Devizes: Wiltshire Record Society, vol. 33, 1978), 34–58 (Wiltshire 1275–80, eleven appeals, which is two per year); JUST 3/18/1, mm. 6–9, 10–15 and JUST 3/18/2 (Essex 1280–85, six appeals, which is one per year); JUST 1/1177A, m. 4d and JUST 1/1179, mm. 14, 19, 25d (Suffolk 1250, 1254, 1258, 1259, two appeals, which is one per year if one assumes each gaol delivery heard appeals from the prior six months); JUST 1/1179, mm. 25, 25d (Norfolk 1259, no appeals). ⁷⁷ Pollock and Maitland, The History of English Law, 1:199; Baker, An Introduction to English Legal History, 45. ⁷⁸ Curia Regis Rolls, vols. 1, 2, 12, 16, 17 (Bench and coram rege 1201, 1225, 1242); KB 26/168 (Michaelmas 1260 coram rege); KB 26/169 (Michaelmas 1260 Bench); Pollock and Maitland, The History of English Law, 2:565, 567 (analysis of Easter 1271 Bench); W. P. W. Phillimore, ed., Placita Coram Domino Rege. . . . The Pleas of the Court of King's Bench, Trinity Term, 25 Edward I, 1297 (London: British Record Society, 1898). Since there was no reason to think that the Bench and court coram rege heard significant numbers of appeals, I examined only a small fraction of the surviving records. These records examined were chosen because they were approximately twenty years apart, and the surviving records were reasonably ample. ⁷⁹ Edward Powell, Kingship, Law, and Society: Criminal Justice in the Reign of Henry V (Oxford: Clarendon Press, 1989), 71. ⁸⁰ Whittick, "The Role of the Criminal Appeal in the Fifteenth Century," 56; Bellamy, The Criminal Trial in Later Medieval England, 36. ⁸¹ Whittick, "The Role of the Criminal Appeal in the Fifteenth Century," 55. ⁸² For example, Baker devotes two pages of his introductory text to the appeal, but provides no explanation for its decline. Baker, Introduction to English Legal History, 574–76. ⁸³ Holdsworth, A History of English Law, 2:360; Hunnisett, The Medieval Coroner, 55. ⁸⁴ As discussed below, 38–40, during certain periods nonprosecuted appeals were sent to jury trial. A similar policy was introduced for quashed appeals temporarily and tentatively in the 1218–22 eyres and then permanently in the 1231–33 eyres. Before the 1231–33 eyres, nearly all quashed appeals (95 percent in my data set) resulted in the acquittal of the defendant. Starting in the 1231–33 eyres, nearly all quashed appeals (98 percent in my data set) resulted in trial on the king's suit. Kerr, who analyzed all surviving pre-1222 appeals, found that in the 1218–22 eyres, 45 percent (10/22) of quashed appeals were sent to jury trial. Margaret H. Kerr, "Angevin Reform of the Appeal of Felony," Law and History Review 13 (1995): 388. See also Bracton, 2:402, fol. 142b; Britton, ed. and trans., Francis Morgan Nichols (1865; reprint, Holmes Beach, Fla: William W. Gaunt and Sons, 1983), 1:103–4. ⁸⁵ For a discussion of trends in rape appeals, see Section 2.D. For a discussion of the Statute of Westminster II and trespass actions for rape, see Post, "Ravishment of Women," 158–59. In addition, such trespass actions were ordinarily brought for the ravishment of wives, not the rape of unmarried women (ibid.), while appeals of trespass almost always concerned unmarried women. ⁸⁶ Baker, An Introduction to English Legal History, 575; Groot, "The Jury in Private Criminal Prosecutions before 1215," 132–33; Stenton, "Introduction," The Earliest Lincolnshire Assize Rolls, lxi. ⁸⁷ Some remedies were available in local courts. Plaints might also be used to get redress, but they were uncommon. Those who had influence with the king might pursue exceptional remedies. In the mid-thirteenth century, trespass began to provide money damages for personal injury and property damage. See below, 44. ⁸⁸ On rare occasions, judges would order the appellee to pay compensation. See, e.g., JUST 1/359, m. 30 (Kent 1241); JUST 1/614B, m. 47d (Northamptonshire 1247). ⁸⁹ See above, 6, 14. ⁹⁰ Klerman, "The Selection of Thirteenth-Century Criminal Disputes for Litigation." ⁹¹ Roger Groot similarly argued that public prosecution of concorded appeals would discourage settlement. See Groot, "The Jury in Private Criminal Prosecutions before 1215," 133. ⁹² This change has been noticed by several previous scholars. Groot, "The Early Thirteenth-Century Criminal Jury," 12–13, 21–22; Kerr, "Angevin Reform of the Appeal of Felony," 369–73 (examining all 1218–22 eyres and finding 50 percent of nonprosecuted and retracted appeals sent to jury trial). ⁹³ This graph, like Figure 1, plots a steady rate from 1194 to 1209, even though the coefficients for 1194–95 and 1198–99 are less than one. As explained above, 28, the figures for these years almost certainly underreport the true rate, and the graph has been adjusted to take this into account. ⁹⁴ In a regression similar to that described in the note to Table 4 above, except that log(lagged respect for settlement) was substituted for the eyre date effects, the coefficient was positive (0.36), and its p-value was highly statistically significant (0.000). ⁹⁵ See Meekings, "Introduction," Crown Pleas of the Wiltshire Eyre, 1249, 4 (disruption of eyres); James Clarke Holt, Magna Carta, 2d ed. (Cambridge: Cambridge University Press, 1992), 325, n. 135 (sheriffs heard criminal cases normally heard in eyre). The fact that assault shows one of the more dramatic drops (79 percent lower than 1201–03) while homicide is almost stable (only 16 percent lower than 1201–03) might imply that litigants brought their cases elsewhere during this turbulent period. Litigants often had a choice of fora for assault cases (including county and manorial courts), whereas the royal monopoly on homicide cases was relatively strict. ⁹⁶ All major twelfth and thirteenth-century treatises discuss settlement policy, but none note that the policy changed, much less explain why. L. J. Downer, ed. and trans., Leges Henrici Primi (Oxford: Clarendon Press, 1972), sec. 59, 27; G. D. C. Hall, ed. and trans., The Treatise on the Laws and Customs of the Realm of England Commonly Called Glanvill (London: Thomas Nelson and Sons, 1965), 21; Bracton, 2:402, fol. 142b; Placita Corone, 9; Britton, 1:103–4. For analysis of these sources, see Daniel Klerman, Private Prosecution of Crime in Thirteenth-Century England (Ann Arbor, Mich.: UMI Dissertation Services, 1998), 86–97. ⁹⁷ Glanvill, 21; Bracton, 2:402, fol. 142b; See also Leges Henrici Primi, sec. 59, 27 (forbidding settlement without judicial consent). ⁹⁸ John W. Baldwin, "The Intellectual Preparation for the Canon of 1215 against Ordeals," Speculum 36 (1961): 615–36. See also Stenton, "Introduction," The Earliest Lincolnshire Assize Rolls, lx (seeing judicial hostility to ordeals in early thirteenth-century cases). ⁹⁹ Robert Bartlett, Trial by Fire and Water: The Medieval Judicial Ordeal (Oxford: Clarendon Press, 1986), 82–90. ¹⁰⁰ Doris Mary Stenton, ed., Pleas before the King or His Justices, 1198–1202 (London: Selden Society, vol. 68, 1952), 2:9, pl. 44 (Norfolk 1198). ¹⁰¹ Groot, "The Early Thirteenth-Century Criminal Jury," 3. ¹⁰² On judicial concern that crime not go unpunished, see Bracton, 2:402, fol. 142b. See also Britton, 1:103–4 (prosecution at king's suit, because nonprosecuted appellee might be guilty). ¹⁰³ Harding, "Introduction," The Roll of the Shropshire Eyre of 1256, xxxvi; G. D. G. Hall, "Some Early Writs of 'Trespass,'" Law Quarterly Review 73 (1957): 65–66. ¹⁰⁴ The early history of trespass remains unclear. Most historians agree that the first trespass writs were issued in the 1220s. S. F. C. Milsom, "Trespass from Henry III to Edward III," Law Quarterly Review 74 (1958): 201; Harding, "Introduction," The Roll of the Shropshire Eyre of 1256, xxxv–vi. Trespass cases became common in the plea rolls of the Westminster courts in the mid-1230s, although many of these cases may have been initiated by plaint rather than writ. G. O. Sayles, ed., "Introduction," Select Cases in the Court of King's Bench under Edward II (London: Selden Society, vol. 74, 1957), 4:xxxvi–vii. Trespass cases from the 1230s and early 1240s are sometimes difficult to distinguish from appeals, but generally differ in that (a) plaintiffs do not allege, and defendants do not deny "felony," (b) the plaintiff puts a monetary value on the harm with a phrase such as "whence he is injured in the amount of 100 s.," thus implicitly asking for damages, (c) neither plaintiff nor defendant suggests trial by battle, (d) the rolls sometimes mention that the plaintiff produced suit witnesses (producit sectam), and (e) the formalities of appeals, such as suit in county court, are not required. See Curia Regis Rolls, vol. 15, cases 867 and 960; vol. 16, cases 143 and 1195. Contrast these cases to appeals, such as Curia Regis Rolls, vol. 15, cases 1128 and 1304; vol. 16, cases 1272 and 1744. ¹⁰⁵ Baker, An Introduction to English Legal History, 575. ¹⁰⁶ The 100 percent figure for 1239–44 can be ignored. It is based on only a single appeal. ¹⁰⁷ In no pre-1218 case in the data set, however, did the judges actually send a nonprosecuted homicide appellee to the ordeal, because the presenting jury always gave a medial verdict of nonsuspicion. Nevertheless, the judges were presumably prepared to send such appellees to ordeal if the presenting jury did suspect them. They did send one theft appellee to the ordeal. See above, 43, n. 100. ¹⁰⁸ D. 90 c. 1, 7. ¹⁰⁹ d. p. C. 2 q. 3 c. 8. ¹¹⁰ John the Deacon, Sancti Gregorii Magni Vita, in J.-P. Migne, ed., Patrologiae cursus completus: Series Latina (Paris: Garnier Fratres, 1902), 75:195. ¹¹¹ For their absence from the collections compiled by Burchard of Worms and Ivo of Chartres, see Aemilius Friedberg, ed., Quinque Compilationes Antiquae (1882; reprint Graz: Akademische Druck- und Verlagsanstalt, 1956), xxi. ¹¹² Joannes Dominicus Mansi, ed., Sacrorum Conciliorum (Venice, 1778), 417–18 (Appendix Concilii Lateranensis c. 2); 1 Com. 5.18.1 and 2; Emil Friedberg, Die Canones-Sammlungen zwischen Gratian und Bernhard von Pavia (1897; reprint, Graz: Akademische Druck- und Verlagsanstalt, 1958), 187 (table showing canonical collections including the two relevant decretals, 1 Com. 5.18.1 and 2). These texts were also included in Gregory IX's thirteenth-century collection. X.5.22.1 and 2. ¹¹³ 1 Com. 5.18.1 and 2. ¹¹⁴ Charles Duggan, Twelfth-Century Decretal Collections and Their Importance in English History (London: Athlone Press, 1963), 53, 66–117, 135–39; Stephan Kuttner and Eleanor Rathbone, "Anglo-Norman Canonists of the Twelfth Century, An Introductory Study," Traditio 7 (1949–51): 280–84. ¹¹⁵ Kuttner and Rathbone, "Anglo-Norman Canonists of the Twelfth Century," 279; Ralph V. Turner, The English Judiciary in the Age of Glanvill and Bracton, c. 1176–1239 (Cambridge: Cambridge University Press, 1985), 36–37, 226. ¹¹⁶ Turner, The English Judiciary, 95–96. ¹¹⁷ Ibid., 97–98. ¹¹⁸ Those likely to have learned canon law include Master Jocelin, archdeacon of Chichester, Richard fitz Neal, Godfrey de Lucy, Master Eustace of Fauconberg, and Master Godfrey de Insula. Ibid., 37–38, 95–99, 144, 150–51, 226, 232, 236. ¹¹⁹ Ibid., 98. In the 1194–95 eyres, two of the judges were archdeacons, and four were bishops or archbishops. Two archdeacons and a bishop served as judges in the 1198–99 eyres. Six bishops but only one archdeacon served in the 1201–3, 1208–9, or 1218–22 eyres. Crook, Records of the General Eyre, 56, 57, 58, 59, 61, 62, 64, 69, 72, 73, 74, 75. ¹²⁰ Bracton, 2:402, n. 3, fol. 142b; Kerr, "Angevin Reform of the Appeal of Felony," 369, n. 52. ¹²¹ Meekings, The 1235 Surrey Eyre, 1:21, 97; Thomas Duffus Hardy, ed., Rotuli Litterarum Clausarum (London: G. Eyre and A. Spottiswoode, 1833), 1:403; Bracton, 2:405, fol. 143b; Britton, 1:19; Paul R. Hyams, "What Did Edwardian Villagers Understand by Law," in Medieval Society and the Manor Court, ed. Zvi Razi and Richard Smith (Oxford: Clarendon Press, 1996), 76–77. ¹²² For a discussion of representation in appeals, see Daniel Klerman, "Female Private Prosecutors." ¹²³ Groot, "The Jury in Private Criminal Prosecutions before 1215," 135–36. ¹²⁴ Carrie Smith, "Medieval Coroners' Rolls: Legal Fiction or Historical Fact?" in Courts, Counties, and the Capital in the Later Middle Ages, ed. Diana E. S. Dunn (New York: St. Martin's Press, 1996), 96–101. ¹²⁵ James Buchanan Given, Society and Homicide in Thirteenth-Century England (Stanford: Stanford University Press, 1977), 36. ¹²⁶ In a regression similar to that described in note 94 above, except that ßlog (homicide rate) was substituted for alphalog (lagged respect for settlement), the coefficient ß was negative (-0.26), and its p-value was not statistically significant (0.319). In a regression with both alphalog (lagged respect for settlement) and ßlog (homicide rate), alpha remains positive and actually increases in magnitude (from 0.36 to 0.48) and remains highly statistically significant (p-value of 0.005), while ß becomes indistinguishable from zero (–0.01, p-value of 0.981). Similar results obtain when the number of homicide appeals is substituted for the number of all appeals as the dependent variable and when the respect for settlement in homicide cases is substituted for respect for settlement in all cases as an explanatory variable. ¹²⁷ John J. Donohue, "Understanding the Time Path of Crime," Journal of Criminal Law and Criminology 88 (1998): 1425–26. ¹²⁸ Using data from Hanawalt, Crime and Conflict in English Communities, 237, 241 (tables 9 and 10), I calculated correlation coefficients between homicide and other crimes for the period 1300–48. The correlation between homicide and other crimes is uniformly positive and moderately strong. The coefficients are: 0.55 for larceny, 0.40 for burglary, 0.34 for robbery, 0.19 for receiving, and 0.30 for arson. Of course, these figures measure the correlation between indictments for, rather than incidence of, various crimes. Unfortunately, Hanawalt's data set did not include enough assaults and rapes to permit statistical analysis. ¹²⁹ Presentment of rape was probably not even possible until the 1275 enactment of the first Statute of Westminster. See above, 9, n. 23. ¹³⁰ Approver appeals of theft were relatively common, but, as noted above, 4, they are excluded from this analysis. ¹³¹ For the reason, see above, 21 n. 67. ¹³² In the printed edition, the presentments for this district are incorrectly recorded as being made by the hundred of Mawesley. ¹³³ Although the rubric for the cases numbered 77–85 in the printed edition is damaged beyond recognition, it is nearly certain that these cases were presented by Chipping Warden. Case 77 records a killing by unknown persons at Eydon. Such presentments were nearly always made by the district where the killing took place. Since Eydon is in Chipping Warden, Chipping Warden was almost certainly the presenting district. The fact that other place names mentioned in cases 77–85 are nearly all from or near Chipping Warden supports this conclusion. ¹³⁴ There is no rubric for Pickering wapentake in the 1208 Yorkshire eyre. Nevertheless, it is evident that cases 3475–3483 in the printed edition are the Pickering wapentake presentments. The rubric for these cases is no longer visible, because the top of the relevant membrane has been damaged. Nevertheless, two pieces of evidence conclusively establish these cases as being from Pickering wapentake. First, case 3484 is the presentment of Pickering vill. In every other surviving eyre, the presentments of Pickering vill follow immediately after the presentments of Pickering wapentake. Second, nearly all the place names mentioned in cases 3475–3483 are from or near Pickering wapentake. ¹³⁵ For a more thorough discussion of the issues discussed in this section, see chapter 3 of my dissertation, Klerman, Private Prosecution of Crime in Thirteenth-Century England. ¹³⁶ The ordinary procedure is described in Section 1.C. Some appeals, however, were initiated by writ and would not appear on coroners' rolls. Appeals initiated by writ would most likely have been heard in the Bench or coram rege, although some were heard in the eyre. Some appeals were initiated by plaint or bill. ¹³⁷ Crook, Records of the General Eyre, 36–37. ¹³⁸ R. F. Hunnisett, "An Early Coroner's Roll," Bulletin of the Institute of Historical Research 30 (1957): 225–31 (1229 Devon coroners' roll, containing one appeal, which also appears in the 1238 eyre roll); JUST 2/261 (1268–71 Oxfordshire coroners' roll, containing one appeal, which also appears in the 1285 Oxfordshire eyre roll, JUST 1/710); R. F. Hunnisett, ed., Bedfordshire Coroners' Rolls (Bedfordshire Historical Record Society, vol. 41, 1961) (1268–71 Bedfordshire coroners' rolls, containing eighteen appeals, of which nine appear in the 1276 eyre roll and one appears in the 1272 eyre roll, JUST 1/7, m. 39); JUST 2/263, 2/264, 2/266, 2/277 (1269–85 Norfolk coroners' rolls, containing two appeals, of which both appear in the 1286 Norfolk eyre roll, JUST 1/579); JUST 2/262, 2/278 (1272–74 Hampshire coroners' rolls containing five appeals, of which three appear in the 1280–81 Hampshire eyre roll, JUST 1/789); JUST 2/260 (1285–86 Hertfordshire coroners' roll, containing four appeals, of which two appear in the 1287 Hertfordshire eyre roll, JUST 1/328). ¹³⁹ See, e.g., Doris Mary Stenton, ed., Pleas before the King or His Justices, 1198–1212 (London: Selden Society, vol. 84, 1967), 4: pl. 3509 (Yorkshire 1208 eyre roll mentions appeal of robbery removed to Westminster); Meekings, Crown Pleas of the Wiltshire Eyre, 1249, 211 (mentioning appeal of homicide that resulted in hanging at gaol delivery). ¹⁴⁰ Hunnisett, Bedfordshire Coroners' Rolls, pls. 69, 129. ¹⁴¹ Appeals that resulted in conviction or outlawry would produce forfeited chattels, if the appellee had any. Appeals that resulted in acquittal would produce amercement (Fining) of appellors, as would nonprosecuted or quashed appeals. If the appellee did not show up, his sureties would be amerced. The only circumstances that would result in no revenue would be conviction of a chattel-less appellee, appeal of a cleric who claimed privilege, an appeal in which the appellee died before trial, or cases in which amercements were forgiven. Such cases surely occurred, but it is hard to believe that they account for all the unrecorded appeals. In addition, such appeals were often recorded. See Harding, The Roll of the Shropshire Eyre of 1256, pl. 792 (defendant acquitted, appellor's fine pardoned on account of poverty); Stenton, Pleas before the King or His Justices, 1198–1212, 4: pl. 3500 (1208 Yorkshire, appellee dead); Three Rolls of the King's Court in the Reign of King Richard the First, a.d. 1194–1195 (London: Pipe Roll Society, vol. 14, 1891), 147 (Buckinghamshire 1195, appellee dead). In addition, the way the plea rolls were put together would have made it difficult to exclude nonrevenue producing cases. It appears that the clerks wrote the first few lines of each enrollment by examining the coroners' rolls and jurors' written veredicta and then filled in the rest later when the jurors presented the cases orally and responded to the judges' questions. Thus, at the time the enrollments were started, the clerk would not have known whether the case would produce revenue. Since many cases were enrolled on a single piece of parchment, those not producing revenue could not have been excluded after the cases were heard. ¹⁴² This argument for the completeness of the eyre rolls does not apply to the 1194–95 and 1198–99 eyres. As discussed above, 26, the system of checking jurors' answers against coroners' rolls does not yet seem to have been used during these eyres. Law and History Review, Spring 2001

Content in the History Cooperative database is intended for personal, noncommercial use only. You may not reproduce, publish, distribute, transmit, participate in the transfer or sale of, modify, create derivative works from, display, or in any way exploit the History Cooperative database in whole or in part without the written permission of the copyright holder.

Settlement and the Decline of Private Prosecution in Thirteenth-Century England

Part One: Background

Part Two. Trends in the Rate of Appeals

Part Three. Respect for Settlement and the Changing Rate of Appeals

Part 4. Alternative Explanations for the Changing Rate of Appeals

V. Conclusion

APPENDIX

A. Rates of Appeal for Some Additional Districts

B. Presenting Districts

C. Criteria for Inclusion of District in Data Set

D. Criteria for Classification of a Case as an Appeal

E. Definition of a "Case"

F. Sources Used in the Database

G. Reliability of Eyre Records

Notes

Settlement and the Decline of Private
Prosecution in Thirteenth-Century England