The Unintended Consequences of Penal Reform: A Case Study of Penal Transportation in Eighteenth-Century London

Abstract

What were the consequences of penal transportation to the New World for eighteenth-century British criminal justice? Transportation has been described by scholars as either a replacement of the death penalty responsible for its decline, or a penal innovation responsible for punishing a multitude of people more severely than they would have been punished before. Using data from the Old Bailey Sessions Papers and the Parliamentary Papers, this study examines sentencing and execution trends in eighteenth-century London. It takes advantage of the natural experiment provided by the passage of the 1718 Transportation Act that made transportation available as a penal sentence, thus enabling one to assess the “effect” of transportation on penal trends. This study finds that the primary consequence of the adoption of transportation was to make the criminal justice net more dense by subjecting people to a more intense punishment. While it was also associated with a small decline in capital sentences for some types of offenders, the adoption of transportation was also associated with an increase in the rate at which condemned inmates were executed. The study closes with a discussion of the conditions that may lead to law's unintended consequences, including the mesh-thinning consequences observed here.

The law rarely behaves as its creators intended, while people and organizations respond to the law in unexpected ways. This can lead to a number of unintended consequences. First, the law has a tendency to diffuse beyond its original target. For example, in the wake of Civil Rights legislation, businesses often went beyond the letter of the law to protect against legal sanctioning (Dobbin 2009). More mundanely, disputants who do not use the courts to settle their dispute rely on the threat of going to court to achieve a better outcome (Mnookin and Kornhauser 1979). Second, while the law has a certain amount of “creep,” it also leads to results that may be contrary to the original intention. In some cases, business organizations' compliance with legal regulation is purely symbolic, without the anticipated substantive effects (Edelman 1992). In other cases, business organizations' “internalization of law” frustrates the progressive democratic and rights foci of the law (Edelman and Suchman 1999). Third, the law may be generally disregarded in decision-making processes. Individuals may disregard the law, whether out of poor knowledge or normative preferences, and resolve disputes without relying on the law (Ellickson 1986; see also Macaulay 1963). In other contexts, the effect of the law in shaping organizational practice regulated by law may be eclipsed by other powerful social institutions (Heimer 1999). Perhaps most disturbingly, significant court victories can have a minimal impact in practice on the rights these victories endorsed (e.g., Rosenberg 1988). In a large variety of contexts, then, the law may have larger-than-intended effects, somewhat sinister effects, or simply no effect on the various practices and activities it is thought to regulate.

In the criminal justice context, we see this theme of unintended consequences emerge most clearly following the adoption of new penal policies and innovations. Many penal innovations are the product of well-intentioned reformers who seek to reduce the number of individuals receiving a severe form of punishment by replacing it with a more lenient punishment. Historically, however, instead of replacing this punishment, the new punishment often leads to two unexpected consequences. First, the new punishment “widens the net” of criminal justice in that more people are punished under the new regime: given this more attractive alternative, criminal justice agents use the new punishment against many more individuals than would have previously received the more severe punishment reformers sought to avoid. Included in this number are some individuals who would not have been punished at all under the old regime or those who previously would have been punished less severely, leading to the second consequence. The new punishment “thins the mesh” of the criminal justice net by punishing some people more severely under the new regime: other offenders who would have been punished less intensely under the previous regime are subsequently diverted away from lesser punishments and subjected to this new punishment which is comparatively more intense (Cohen 1979). For example, the development of probation was intended to divert offenders from receiving a prison sentence; in reality, individuals who never would have gone to prison (or perhaps been punished at all) were suddenly subjected to a community supervision sentence complete with myriad regulations and requirements. By punishing people who previously would have gone unpunished, probation widened the net; by punishing some people more intensely with supervision than with a brief, petty fine, probation made the net of criminal justice more dense (Cohen 1979, 1985; Scull 1984). This story of net-widening and mesh thinning, first discussed by critical criminologists of the 1970s, has been the framework for understanding the development of the prison (Spierenburg 1991: 280), especially the promise of rehabilitation therein (Ignatieff 1978; Rothman 1971); juvenile justice (Platt 1977; Rothman 1980; Sutton 1990); American adult reformatories (Pisciotta 1994); parole and probation (Cohen 1979, 1985; Scull 1984; Simon 1993); problem-solving courts (e.g., Nolan 2003); modern intermediate and community-based sanctions (Blomberg, Bales, and Reed 1993; Blomberg and Lucken 1994; Tonry and Lynch 1996); and modern juvenile diversion programs (Austin and Krisberg 1981; Blomberg 1977; Klein 1979; Lemert 1981). The variety of penal contexts in which these phenomena appear is extremely varied.

However, it is not always clear from the historical record whether such untended consequences have occurred. This is the case for eighteenth-century penal transportation—the practice of sending convicts from Great Britain to the North American colonies for seven or fourteen years of hard labor. Transportation has been variously portrayed as a successful reform that reduced the severity of the criminal justice system and a reform with unintended net-widening and mesh-thinning consequences. Despite the presence of what has been called the “Bloody Code” and its many capital statutes, death sentences and executions in England declined in the seventeenth and eighteenth centuries. Historians attribute this decline to transportation, which was gradually incorporated into the British penal schema during this period (Jenkins 1990). Recently, this narrative has been challenged by Feeley (1991, 1999, 2002), who argues that transportation could not have been responsible for the decline in capital punishment. Instead, Feeley argues that transportation enabled the punishment of greater numbers of offenders (many of whom would previously have gone unpunished) and the more severe punishment of others (many of whom would previously have been released following corporal punishment). Neither Jenkins' nor Feeley's statements, however, have been rigorously tested through the appropriate use of statistics.¹

In this article, I seek to produce a more rigorous empirical account of the systematic adoption of transportation in 1718 with particular emphasis on its effects on criminal sentencing in one important locality. Did transportation cleanly replace capital punishment or did it have other, unintended consequences for the criminal justice system? To answer this question, I test hypotheses generated from these competing narratives. For the sake of brevity, I only examine the possibility of mesh-thinning consequences, the more interesting of the unintended consequences, and leave the question of net-widening effects to future research. I use data from the Old Bailey Sessions Papers to examine the sentencing practices for serious offenses tried at London's Central Criminal Court from 1684 to 1776. I also rely on data collected in a report from the Parliamentary Papers (1819) to examine the trends in executions between 1699 and 1776.² I find that transportation sentences both thinned the mesh of criminal justice and worked to replace some death sentences. Sentences to less severe punishments declined significantly after 1718, when transportation became an available sentencing outcome. At the same time, death sentences declined after 1718 for most offenders (but they increased for non-theft offenders). However, the portion of offenders actually executed significantly increased after 1718, suggesting that to the extent that transportation replaced capital punishment, it did so only at the level of sentencing and not at the level of life-saving pardons. Thus, transportation appears to have played both roles, replacement and mesh-thinning innovation. I briefly discuss some tentative explanations for these results, and close by suggesting the implications of these findings and directions for future research.

The Relationship Between Transportation and Capital Punishment

Eighteenth-Century Criminal Justice

In the late seventeenth century and into the eighteenth century, British courts had a wide array of punishments at their disposal. Most offenders were branded on the cheek or the thumb; pilloried; whipped publicly or privately; fined; pressed into military service; sentenced to hard labor; or sentenced to stay at Newgate prison, the house of correction, or another institution. Some received more severe sentences of death or (after 1717) transportation. Indeed, for much of the eighteenth century, the British Parliament expanded its list of capital offenses, forming the so-called “Bloody Code.” While estimates vary, there were “about fifty” capital offenses in 1688 (Tobias 1979: 140), well over 100 in the 1700s, and over 200 capital offenses by the nineteenth century, sanctioning anything from murder to stealing pre-picked fruit.³ Given this severity, legal and penal historians have been puzzled by the fact that actual executions in this period were in decline. Sharpe (1999) explains, “To put it at its most basic level, a person accused of felony at the assizes in Elizabeth's reign stood a one in four or five chance of being executed; for his or her counterpart under Queen Anne, the chances were more like one in ten” (93). Paradoxically, then, while British penal law on the books was growing harsher, declining rates of execution showed decreasing severity with respect to this most severe punishment.

While executions declined in this period, transportation slowly increased until undergoing a massive expansion. For the first half of the seventeenth century, offenders were transported very infrequently, and only through a conditional pardon preventing execution. Smith (1934) explains, “It is certain that no very satisfactory system for regularly reprieving and transporting convicts was evolved until the middle of the century” (235). Between 1615 and 1628, only 53 convicts were ordered to be transported from England, or almost four per year. Between 1634 and 1650, 64 convicts were ordered to be transported (Smith 1934: 236), or an average of about 9 persons per year.⁴ After 1650, reliance on transportation increased further, with about 100 individuals transported to the Americas each year (Maxwell-Stewart 2010; Smith 1934: 238). Still, the number of offenders transported was a small portion of all convicts sentenced to any punishment, and no one was transported as part of their official sentence.

The expansion truly occurred following the Transportation Act of 1718, when transportation became an official punishment in its own right. The Act had two main provisions: The first provided that an offender convicted of a low-level capital (clergyable) offense could be directly sentenced to transportation. This was new. The second provision, however, merely formalized what was already an institutionalized practice. Convicted offenders sentenced to death could instead be transported as a condition of a pardon from the king or his council. The judicially imposed sentences to transportation would last seven years, while transportation as a condition of a pardon would last fourteen years. Those who returned earlier than allowed would be executed. Judges immediately took advantage of this new sentencing option, sentencing offenders to transportation in great numbers. Between 1718, when the first official transportation sentence was issued, and 1775, the last year of uninterrupted shipments, approximately 50,000 offenders were transported to North America from Great Britain. Of these, 18,600 came from London and its environs alone (Ekirch 1985: 188).⁵ Many more were sentenced to be transported but escaped their sentence, often by a conditional or full pardon. While this period is most often known as the period of the Bloody Code and its plethora of capital crimes, the number sentenced to transportation or actually transported dwarfed the number of death sentences and executions_.⁶

Previous Research

Prior research on the effect of convict transportation on the criminal justice system generally is divided into two schools. The first views transportation as a replacement of capital punishment, while the second describes transportation as a significant innovation that tightened the mesh and widened the net of criminal justice. The replacement view of transportation holds that capital punishment declined in large part because penal officials increasingly came to rely on transportation over capital sentences and executions; in this view, transportation was simply an alternative to the death penalty and therefore took its victims from the realm of those who otherwise would have been capitally sentenced or executed. While early historians like Radzinowicz (1948), Marxist historians like Rusche and Kirchheimer (1939: 58–60), and more recent historians (Beattie 1986, 2001; Sharpe 1999: 95; Tobias 1979: 159, 183) have described transportation as an alternative to or replacement of capital punishment, this view has been most clearly expressed by historian Philip Jenkins. After examining a number of different archival and secondary sources to recreate the execution rate in Great Britain and specific localities, Jenkins (1990) estimates that execution rates began declining in the seventeenth century and continued to decline into the eighteenth century, the same period, he notes, when officials began to use transportation. He describes a decline in the execution rate from about 20 (per hundred thousand) in the 1580s, to 10–15 in the 1630s, to less than five in the late seventeenth century (133–34; see also 138). In light of this trend, he argues that the early (unofficial and ad hoc) use of transportation “replaced” capital punishment (130, 145) beginning in the seventeenth century and is therefore responsible for the decline.

However, this view has been challenged indirectly and directly, casting doubt that transportation replaced capital punishment. First, the timing is suspicious. In contrast to Jenkins (1990), Feeley notes that capital punishment began declining before transportation became an official or even common punishment (2002: 340–41, n. 5; 1999: 48). As noted above, transported offenders before 1718 represented a very small fraction of all convicted offenders, yet death sentences had begun their descent by the early seventeenth century. Thus, Feeley reasons that “transportation had at best only a marginal impact on executions” (1999: 61). Second, several scholars have suggested that the people who were ultimately transported would not have been executed in the absence of transportation, and thus could not have contributed to the decline in executions. Examining over 1,100 inmates of the London Bridewell who were transported in the early to mid-seventeenth century, Griffiths (2008: 284) notes that many of these early transportees were children, who were not likely candidates for execution. Moreover, he reports that many of the offenders had committed low-level offenses—not the kind that would render an execution likely. He notes,

in cases where offences are recorded, over six-in-ten prisoners “kept” for ships were vagrants (63.10 per cent). Others who set sail for the New World included nearly 100 thieves, forty-one people caught walking after curfew, twenty-six nightwalkers, twenty-one badly behaved servants, forty-five beggars, five cheats, one drunk, and a single ballad-singer. (2008: 286)

This profile of transported offenders remained constant even after the 1718 Act. In their study of 4,500 offenders transported from northern England between 1718 and 1776, Rushton and Morgan (2003: 63) found that their subjects had been “convicted for mostly petty offences. Indeed, only about a third of those transported from assize courts had been condemned to death and reprieved on condition of transportation.” These findings suggest that many who were transported would have received some punishment less severe than capital punishment. Finally, there is preliminary evidence that transportation in fact had no effect on executions. Beattie suggests that the Transportation Act's second major change—officially allowing a conditional pardon of transportation instead of hanging (the post-sentencing effect)—“did not induce decision-makers to grant more pardons” (2001: 457) and “hanging remained the principal resource against dangerous offenders and the principal means by which the state demonstrated the power of the law” (2001: 459).

In light of this evidence, a second view of transportation, challenging the first, has been promulgated primarily in the work of political scientist and historian Malcolm Feeley (1991, 1999, 2002). In this view, transportation is not described as a replacement, but rather as a mesh-thinning (punishing more severely) and net-widening (punishing more people) punishment. Feeley reasons that transportation could only have had a marginal impact on capital punishment after the 1718 Act because transportation “was imposed overwhelmingly on those who previously would have been punished less not more severely, or would have escaped formal punishment altogether” (2002: 328). Thus he argues, “transportation was an alternative not to the gallows but to an outright pardon, an exercise of benefit of clergy, a fine, a whipping, or a brief period in jail” (1999: 49), and consequently resulted in the punishment of more people and harsher punishments. For Feeley, the description of transportation as a replacement downplays its true significance, viz., revolutionizing criminal justice by expanding criminal justice.

[T]his new form of punishment was an innovation of gigantic proportions, one that multiplied many times over the state's capacity to punish. To characterize it as an alternative or substitute for executions is to fail to comprehend that it was an innovation that profoundly expanded the state's capacity to punish and thus radically transformed the criminal justice system of the time. (1999: 41)

In punishing lesser offenders more severely, it directly increased the reach of criminal justice; but, in increasing the size and efficiency of criminal justice, Feeley argues, “it unleashed expectations to expand the capacities of other criminal justice institutions, and thus set in motion the movement to reform the entire criminal process” (2002: 328).

While Feeley focuses more on the net-widening role, other scholars have also emphasized transportation's mesh-thinning role, pointing out that transportation was a much more intense punishment than offenders would have received otherwise. Beattie explains, “Men and women who might have left the court with a branded thumb, as most of the prisoners charged with similar offenses continued to do, found themselves returned to jail for what might turn out to be many months and then transported to America” (Beattie 1986: 475). Similarly, Ignatieff explains, “All we have is the clear fact that both Parliament, the judiciary, and the jury cooperated in extending the use of transportation as a punishment in place of both whipping and hanging” (1978: 20). Beattie (2001) concurs, finding that “Whipping sentences at the Old Bailey were immediately and significantly reduced when the Transportation Act came into effect” (446). Relying on Surrey County data from sampled years, Beattie (1986: 487) found a similar result.

Thus, the portrayal of transportation in the historical literature has been variegated. Transportation has been described as a mere alternative responsible for the decline of capital punishment in a period traditionally known for its brutally severe capital laws. Following this view, we would expect to see a dramatic decline in the rate of capital sentences when transportation became an official sentence, and a decline in the rate of executions as penal officials increasingly relied on pardons conditional on transportation. By contrast, transportation has also been described as a revolutionary punishment that extended the reach of criminal justice by punishing both more people and more severely despite its apparent lenient nature relative to an execution. Those holding this view would not anticipate an effect on the death penalty, but rather a significant decline in the rate at which convicted offenders were sentenced to lesser punishments as transportation thinned the mesh of criminal justice. However, neither account of transportation has been rigorously tested with statistical analyses. Those studies that have investigated quantitative trends in sentencing and executions have not employed a careful research design, taken into account serial correlation, or considered statistical significance. This study remedies the situation by rigorously and statistically evaluating these competing accounts. While these views of transportation also have implications for convictions, I examine only sentencing outcomes instead of pre-penal outcomes. Therefore, this study compares the accuracy of the replacement and mesh-thinning descriptions of transportation, and leaves a test of the net-widening description to future research.

Hypotheses

To determine which of the two competing narratives—replacement or mesh-thinning—is more accurate, I test three hypotheses using quasi-experimental methods. These hypotheses speak to the portion of secondary punishments, the portion of death sentences, and the portion of executions out of all death sentences.

Hypothesis 1: Secondary Punishments

How did the rate of other, lesser sentences (i.e., whipping, branding, and fines), change, if at all, after the emergence of transportation as an official punishment? If transportation merely replaced capital punishment, then the availability of transportation would have had no effect on these non-capital punishments; thus, we would expect no change in the rate of sentences to these secondary punishments. If transportation represented a mesh-thinning punishment that diverted individuals from less punitive punishments, we would expect these other punishments to decrease.

Hypothesis 2: Death Sentences

How did the rate of death sentences change, if at all, when transportation officially became an available punishment? If transportation replaced capital punishment, we would expect death sentences to have dramatically decreased once transportation became an alternative. If transportation simply made the net of criminal justice more dense, we would expect death sentences to have remained unaffected, proceeding at a constant rate.

Hypothesis 3: Execution Rate

How did the emergence of transportation as both an available sentence and as a conditional pardon affect executions?⁷ Specifically, once judges and others had a legally endorsed alternative to death, did they more readily recommend offenders for the king's mercy? If transportation in fact replaced capital punishment, we should see a decline in the rate of executions. Regardless of the number of death sentences, the portion of those actually executed would have decreased as the king distributed more pardons conditional on transportation. If transportation sentences were only or largely distributed to those who would not otherwise have been capitally punished, the Transportation Act should not have changed the portion of people sentenced to death who were executed.

Data and Method

Data

The data for this study come from two sources. The first is the Old Bailey Sessions Papers (OBSP) from 1684 to 1776, which provides sentencing data. These are early, newspaper-like “pamphlets that recount the trials at a single monthly ‘sessions’ of the Old Bailey, the court of regular jurisdiction for cases of serious crime in London and the contiguous county of Middlesex” (Langbein 1978: 268). From the 1680s onwards, the OBSP “included a substantially complete record of all the cases that had been tried, revealing for the first time in a systematic way the numbers of men and women convicted and acquitted, and the range of punishments imposed on the guilty” (Beattie 2001: 2–3). Despite format and content changes throughout the eighteenth century, the OBSP “continued to include all the trials in each session and to that extent it remained a complete record of the Old Bailey proceedings” (Beattie 2001: 374).⁸ Because of their utility, the OBSP have been used in numerous other projects examining early English criminal justice (e.g., Beattie 2001; Feeley and Little 1991; King 2000; Langbein 1978, 1983).

The Old Bailey Proceedings Online Project (Old Bailey Proceedings Online ND) has coded these trials from 1674 to 1913 and made them available online.⁹ Using their online tools, I generated a longitudinal dataset providing summary statistics for each year, including the number of defendants; defendants found guilty and type of conviction; and defendants sentenced to death, transportation, branding, whipping (both public and private), or a fine.¹⁰ Each of these categories was further stratified by crime type (non-violent theft and all other offenses), because theft was the largest category of offense in the dataset and the one most affected by the use of transportation.¹¹ Finally, I created variables representing the percentage of guilty offenders sentenced to a particular punishment.

The second data source is an 1819 “Report from the Select Committee on Criminal Laws, &c.” in the Parliamentary Papers on the criminal law. This Report has been utilized by Radzinowicz (1948), Gatrell (1994), and King (2000), among others. The several appendices to this Report include annual summary statistics of the number of people sentenced to death and executed. They report data for London, the combined jurisdiction of London and the county of Middlesex, and the Home Circuit. Though written in 1819, the appendices' information extends from 1699 at the earliest into the early nineteenth century; the periods for which data are reported vary by jurisdiction. I focus on data for London in the years 1699 to 1755 and the Home Circuit between 1689 and 1718. Because these do not share the exact jurisdiction of the Old Bailey, I do not make comparisons between the datasets. Instead, the execution data enable a further test of the effect of transportation on a related but different jurisdiction and with respect to actual executions, which cannot be evaluated by the OBSP data.

To these yearly data, I added seven additional variables. First, I included a running variable, YEAR-1718, a continuous variable equal to 0 in the year 1718 (a technique that assists the interpretation of regression coefficients). Second, I included a binary treatment variable, POST CHANGE, to indicate whether the offender was sentenced in the year 1718 or later. Third, I included an interaction term, POST*YEAR, which is the product of YEAR-1718 and POST CHANGE. Finally, I included four binary variables for each of four wars that occurred during the period examined (only two of which occurred during the period for which I have execution data). All of the variables used in the analyses are summarized in Table 1.

Table 1. Variables

Source: The Old Bailey Sessions Papers and The Parliamentary Papers (1819).

Sampling Frame

In this study, my sample consists of the full population of Old Bailey defendants (and, in the Parliamentary Papers data, the full sample of those sentenced to death). However, the data necessarily suffer from a degree of selection bias: there were many points in the path to delivering and executing a sentence at which the accused or convicted criminal could be removed and ultimately avoid this final stage (e.g., Beattie 2001; King 2000). Observations in the OBSP dataset represent only those cases of serious offenses in which a victim brought a case to a lower court, the case was not settled informally or summarily, and a grand jury found a true bill. Those who received a sentence represent those who had been found guilty or partially guilty.

The sample is further restricted by period and location. The study ends with the year 1775 because, at the outbreak of the American Revolution, the Old Bailey interrupted its use of transportation to North America before adopting Australia as the new destination several years later. Moreover, the outbreak of the war with America confounds further comparisons because of changes to the crime, indictment, and sentencing trends that occur during wartime. The choice of start year, however, is less clear. The OBSP records available for the early years are not representative and have fewer observations because records were inconsistently recorded. As Langbein (1978: 268) explains,

The earliest surviving OBSP exemplars, from 1674–1676, are still recognizably of the older chap-book format in size, appearance, content, and tone. They are quite selective, reporting only a few cases of greatest general interest; and they preserve the moralizing tone that was long characteristic of the chap-books.

It is unclear when the records began to be recorded completely, but it was sometime in the 1680s, beginning what Langbein (1978: 269) calls their “newspaper phase.” By the mid-1680s, the pamphlets “were published regularly and they recount a goodly number of cases; crude sensation-mongering dies out from the title pages, and moral instruction disappears from the accounts” (Langbein 1978: 269). While Langbein (1978) uses all of the OBSP pamphlets, Feeley and Little (1991: 722, f. 7) begin their quantitative study in 1687 to ensure reliability. I adopt 1684 as a reasonable start date. Langbein (1978: 269) refers to pamphlets from 1684 as representative of the more mature OBSP that persisted into the eighteenth century. In 1684, the OBSP adopted their formal, lasting title, following a rule that disallowed unauthorized reports of the court's proceedings (Beattie 2001: 2). Further, the OBSP were sufficiently impressive by 1684 to merit a monopoly over reporting the Sessions' proceedings announced in January 1685 (Beattie 2001: 3). Moreover, the number of defendants mentioned and found guilty in the reports stabilizes around 1684. Finally, the context of the seventeenth and eighteenth centuries prevents a more conservative start date: The late seventeenth and early eighteenth centuries experienced several wars of sizable durations that affected the crime and sentencing rates; indeed, in several years (1705, 1706) there is no data at all. Beginning this study with the late 1680s or the brief period between wars in the late 1690s would produce a skewed picture as crime rates were especially high after soldiers returned home to the post-war economy. However, reporting practices and the frequent presence of war are taken into account in the interpretation of the data.

The sample is also restricted to offenders in London, particularly those who were tried in the Old Bailey. London was undoubtedly Great Britain's most important city, and the Old Bailey its most important criminal court. Consequently, generalization to the rest of the country is unwise. As Cockburn (1994: 165) described it, the “culture of punishment” in London “was quantitatively and qualitatively distinct from that pertaining elsewhere in the country.”¹² However, because of this uniqueness, London and the Old Bailey are extremely interesting case studies. Moreover, one of London's particularities is especially useful for this study: London's judges relied on transportation far more heavily than did judges in other jurisdictions.¹³ London offers a useful setting because an investigation of the impact of transportation on capital punishment should look at where the new punishment was most heavily used.¹⁴ However, we should expect that whatever effects are shown for London would likely have been different elsewhere.¹⁵

Challenges to Analysis

Analyzing the relationship between transportation and capital punishment in this period is difficult for several reasons. The largest challenge to examining this period, regardless of jurisdiction, is the high frequency of war. The period begins with King William's War (1689–1697). Beginning only a few years after the Old Bailey Sessions Papers are considered whole and consistent, this war does not leave much of a baseline for sentencing figures. Another significant war, the War of the Spanish Succession, follows soon after in the years 1701–1714. This war only allows three full non-war years before the passage of the Transportation Act of 1718, most of which would have been characterized by an exogenous increase in the number of offenders brought before the court confounding any attempt to analyze a change in frequency after 1718. Two more wars follow in 1740–1748 (the War of the Austrian Succession) and 1756–1763 (the Seven Years' War), adding further noise to the underlying trends.

The heavy frequency of war is problematic for an analysis of conviction and punishment trends. War had a dual impact on crime rates and, by extension, indictments and sentencing. First, it transferred many young men (in the peak of their crime-committing years) from the streets into the military, thus decreasing the crime rate directly. Second, according to Beattie (2001: 42), the removal of these men raised the level of employment as more jobs were available; with something closer to full employment, crime rates decrease. The situation is reversed when a war ends. War is often followed by a period of substantially increased crime rates as soldiers disband (flooding the labor market, raising levels of unemployment) and the economy declines as the war-time boost draws to a close. This data cannot simply be ignored, as it would induce selection bias as well as remove a substantial portion of the data, especially given the long durations of these wars. Thus, this data must be taken into account when testing the hypotheses. I enable visual inspection of the data and employ separate controls for each war to absorb this fluctuation, but the frequency of war remains a limitation of this study.

Analytic Strategy and Model

To analyze the data, I use a combination of regression discontinuity design (Imbens and Lemieux 2008; Thistlewaite and Campbell 1960), accounting for serial correlation and controlling for war, and two-sample z-tests. The three hypotheses are tested by taking advantage of a natural quasi-experiment enabled by the 1718 Transportation Act. Before 1718, transportation as a formal sentence was extremely rare; it was, however, heavily relied upon thereafter. By examining whether there were any significant differences in penal outcomes before and after 1718, I can demonstrate whether the availability of transportation had an effect on penal outcomes.

I use regression discontinuity analyses to examine whether there is a statistically significant difference in various sentencing and execution trends before and after 1718, for as many years as possible. While it is extremely risky to assume little changed in the nearly forty years before and almost sixty years after the Transportation Act, analyses employing a regression discontinuity design are robust enough to establish the large trends discussed in the hypotheses. In particular, examining the coefficient on POST CHANGE can illustrate whether sentencing experienced a sudden increase or decrease in 1718. The coefficients on the interaction term POST*YEAR and the running variable YEAR-1718 (a continuous year variable centered at 1718) further illustrate whether long-term sentencing experienced an upward or downward trend after 1718. Indeed, one of the largest sources of variation in sentencing is the frequent presence of war; I control for war in most of the regression models below, but the coefficients on war for each dependent variable are not discussed. While each analysis employs a different dependent variable, determined by the hypothesis tested, the standard regression follows the form of Equation . Additionally, all models reported display Newey-West standard errors, using a lag of four, because regression discontinuity analyses initially suffered from serial correlation.¹⁶

(1) $\begin{array}{c}{Y}_i=\alpha +\beta {\text{YEAR}}_i+\gamma {\text{POST}}_{\text{1718}i}+\delta ({\text{POST}}_{\text{1718}i}*{\text{YEAR}}_i)+\zeta {\text{WAR}}_{1i}\\ +\eta {\text{WAR}}_{2i}+\theta {\text{WAR}}_{3i}+\iota {\text{WAR}}_{4i}+{ϵ}_i\end{array}$

I further employ two-sample z-tests of the difference in outcome means (or percents) to measure change on a shorter timescale—the years immediately before and after the law went into effect. These work as a kind of check on the regression models: if the 1718 law is associated with a significant change in a particular sentencing trend, we should also see this change immediately after the law went into effect. Another benefit to these analyses is that the years immediately before and after the Transportation Act should be more similar to each other than are the longer periods compared in the regression discontinuity analyses. Agreement between the regression discontinuity analyses and tests of significance would indicate a robust result.

Findings

Discontinuity in the Running Variable

Before performing the main analyses, however, I must review possible threats to the study's internal validity. In particular, it is always a concern in regression-discontinuity approaches that an observed change in the outcome of interest is due to a contemporaneous change in some exogenous factor rather than a change in the independent variable of interest. In this case, if the population of inmates coming before the court changed around 1718, this population change rather than the availability of transportation may be responsible for any changes in sentencing outcomes. I first examine whether the number of defendants coming before the court changed—increased or decreased greatly—and then whether the composition of offenders, based on the kinds of crimes of which they were accused, changed. If either of these changes occurred, further changes in court outcomes could have resulted from these population changes rather than a change in the law.

Population increase does not appear to be a threat. The number of defendants increased after 1718 in general and when stratified by offense type (see Figure 1, Table A). However, this change was not significant.¹⁷ This general increase in offenders at the Old Bailey can be explained by population growth in London. There is no strong evidence to suggest that the number of defendants entering the Old Bailey changed around the time of the 1718 law. Moreover, by conducting the analyses below in terms of the percent of offenders with particular outcomes, a change in the number of defendants should not confound the results statistically.

Figure 1. Number of Defendants at Old Bailey, 1684–1775. Reference line imposed at 1718 (Year = 0). Standard errors in the regression account for year-based serial correlation. Regression controls include year, a dummy for the period after the 1718 law, an interaction variable between year and the period after the 1718 law, and a separate dummy variable for each war.

Composition change does not appear to be a problem for the analyses. While judges may sentence all offenders differently if they see an increase in a particular kind of offender, and a compositional change could also lead to misleading statistical artifacts,¹⁸ there was no substantial change in the composition of crimes. In the roughly 20 years before and after 1718, theft-based offenses vary from year to year but generally remain in the high 70 percent range (see Figure 2). There was also no change in the short-term period, when any change would be most damaging to the study. Relying on tests of significance, I find no significant difference between the portion of offenders brought to the Old Bailey for theft offenses (and, by extension, in the portion brought for non-theft offenses) between 1717 and 1719¹⁹ or between the years 1715–1717 and 1719–1721 (see Table 2). Consequently, there is no clear evidence to suggest that something else changed around 1718 that could be responsible for any other changes observed. I now turn to the results of the various hypotheses.

Figure 2. Defendants before Old Bailey by Crime Type, 1684–1775. Reference line imposed at 1718 (Year = 0).

Table 2. Two-Sample Z-Tests: Portion of Defendants before the Old Bailey for Theft, 1717 v. 1719 and 1715–1717 v. 1719–1721

No significant results at the p < 0.05 level.

Source: Old Bailey Sessions Papers.

Results of Hypothesis Testing

Hypothesis 1: The Effect on Sentences to Secondary Punishments

In the first analysis, I determine whether we observe in 1718 a shift in the portion of sentences to what I call “secondary punishments,” viz., whipping, branding, and fines. Under the replacement thesis, we expect no change; under the mesh-thinning thesis, we expect a substantial decrease in their frequency. I find support for the mesh-thinning thesis: Though their share of total punishments sentenced had been increasing from the 1680s to the 1710s, the portion of secondary punishments immediately decreased markedly after the 1718 Act and remained low thereafter (see Figure 3, Table A). On average, such punishments dropped, significantly, by 56.7 percent, and remained around this rate over time, with little annual fluctuation. This finding suggests that offenders who once would have received a secondary punishment were suddenly much more likely to be transported, and thus that transportation did play some mesh-thinning role.

Figure 3. Secondary Punishments Distributed, 1684–1775. Reference line imposed at 1718 (Year = 0). Standard errors in the regression account for year-based serial correlation. Regression controls include year, a dummy for the period after the 1718 law, an interaction variable between year and the period after the 1718 law, and a separate dummy variable for each war.

Hypothesis 2: The Effect on Death Sentences

In the second analysis, I determine whether we observe in 1718 a shift in the portion of death sentences. Under the replacement theory, we expect a substantial decrease in their frequency; under the mesh-thinning thesis, we expect no change. The Act's effect on death sentences provides something of a mixed picture. In general, death sentences decreased after the 1718 law, both in the short term and the long term. The regression discontinuity analyses (see Figure 4, Table A) demonstrate that death sentences decreased sharply and significantly in the period after 1718. On average, death sentences were about 8.5 percent less common for all crimes in 1718 and after, and 13.3 percent less common for theft offenses. Death sentences for these offenses also continued to decline thereafter, but insignificantly so. However, death sentences for non-theft offenses increased significantly by about 18.6 percent, and they continued to increase slightly (and insignificantly) thereafter.

Figure 4. Rate of Death Sentences by Crime Type, 1684–1775. Reference line imposed at 1718 (Year = 0). Standard errors in the regression account for year-based serial correlation. Regression controls include year, a dummy for the period after the 1718 law, an interaction variable between year and the period after the 1718 law, and a separate dummy variable for each war.

The significance tests show similar findings for the short term. Death sentences significantly decreased for all offenses from 36.3 percent in 1717 to 23.6 percent in 1719 (z = 3.33), and 34.2 percent to 19.1 percent for theft offenses (z = 3.76), but the change was negligible for the handful of non-theft offenders (z = 0.03) (see Table 3).²⁰ To clarify, while death sentences generally decreased significantly after 1718, much of the overall decline was largely because the portion of theft offenders receiving the death penalty decreased, even though the smaller group of non-theft offenders received death in greater portions. Overall, these results support the suggestion that transportation helped to replace capital punishment, but only in the case of theft offenses (which represented the bulk of the caseload).

Table 3. Two-Sample Z-Tests: Death Sentences by Crime Type, 1717 v. 1719

Source: Old Bailey Sessions Papers.

Hypothesis 3: The Effect on Portion of Executions

Finally, I determine whether we observe a shift in the portion of executions in 1718. Under the replacement thesis, we expect a substantial decrease in their frequency; under the mesh-thinning thesis, we expect no change. Unexpected for either theory, but clearly in contradiction to the replacement thesis, the portion of individuals executed among those sentenced to death in the London increased following the Transportation Act. That is, in the years after the Transportation Act, the probability of escaping execution after being sentenced to death was much smaller. The rate of execution increased significantly by between 17.1 percent and 18.4 percent (See Figure 5, Table A.).²¹

Figure 5. Portion Executed Before and After 1718. Reference line imposed at 1718 (Year = 0). Standard errors in the regression account for year-based serial correlation. Regression controls include year, a dummy for the period after the 1718 law, an interaction variable between year and the period after the 1718 law, and a separate dummy variable for each war.

In the short term as well, I find that executions may have increased after the Act, instead of decreasing under the competing weight of transportation. In 1719, the portion of condemned individuals actually executed increased to 54.5 percent from 20.0 percent in 1717. When we compare 1717 and 1718, that is, the year before transition and the year of transition, we see increase but without significance.²² (See Table 4.) These additional findings suggest that the Transportation Act's effect was not immediate. In the long run, however, rather than decreasing the frequency of execution, transportation apparently made execution more likely for those capitally sentenced.²³ These findings suggest against the replacement theory.

Table 4. Two-Sample Z-Tests: Proportion Executed, 1717 v. 1719 and 1717 v. 1718

Significance denoted by ^*at the p < 0.05 level.

Source: Parliamentary Papers (1819).

Discussion

These results offer a complicated picture. On the one hand, the evidence suggests that the Transportation Act of 1718 allowed transportation to replace the death penalty in the sentencing process to some extent. Death sentences for theft offenders declined when transportation became a widely available punishment. This replacement, however, appears to have stopped at the sentencing phase: pardons appear to be significantly less common after 1718. These two findings suggest that sentences to transportation replaced death sentences, but that the availability of transportation did not replace executions once a convict was condemned.

On the other hand, the evidence also suggests that the Transportation Act had unintended consequences. Secondary punishments—branding, whipping, and fines—declined heavily with the passage of the 1718 Transportation Act, suggesting that transportation was used to punish offenders who would have received what was considered a less severe punishment. The huge diversion of offenders from more lenient corporal or financial punishments to a lengthy punishment represents significant mesh thinning of the criminal justice net. Moreover, this decline was proportionately much larger than the decline observed in capital sentences, suggesting that the unintended mesh-thinning consequences of the Transportation Act was far more significant for the criminal justice system than its role in partially replacing capital punishment.

Three puzzles remain. First, the availability of transportation, a punishment intended to reduce capital sentences, in fact caused a substantial number offenders—approximately 56 percent of convicted offenders in the post-1718 period—to receive a much more severe punishment than they previously would have received. The literature on the unintended consequences of penal reforms may provide guidance. There is currently little consensus over the causes of unintended consequences following criminal justice reforms. Following Foucault's insights, Cohen (1979: 358–360) suggests that the expansion of experts, claiming knowledge derived from community corrections, expanded community corrections programs to justify their positions (see also Dobbin 2009). Blomberg and Lucken (1994) suggest that the process of “piling up sanctions,” and thereby increasing the number of hurdles through which a person under community supervision must proceed, increases a person's probability of failing and therefore returning to prison. Austin and Krisberg (1981) suggest that reforms frequently backfire because reformers fail to take into account the social, political, and ideological context; in some cases, a net-widening or mesh-thinning result is exactly the goal of non-reformers who can hijack the reforms to achieve these ends (see also Polk 1987; Sutton 1990). While Decker (1985) suggests intentionality is a possible explanation, he also suggests that poor training and inadequate knowledge of the law may be responsible for the failure of juvenile diversion programs.

Unintended consequences in these other venues may provide guidance on possible explanatory factors in the present case. In the absence of lay professionals needing to validate their position and (presumably) judges lacking training and adequate knowledge of the law, we might suspect that the true, legislative intention of the Transportation Act was to punish petty offenders more severely or that the progressive reformers' law was essentially hijacked by judges who did intend this result. In fact, some combination of these two possibilities appears likely. On the one hand, those responsible for passing the Transportation Act sought to replace the death penalty with transportation.

[A]n overriding concern in the years following the Restoration and into the eighteenth century was the need felt for a non-capital punishment for relatively petty crimes against property. By 1660 the threat of hanging for a second conviction for a clergyable felony had lost any deterrent power it may once have had. The history of punishment over the following sixty years is very largely concerned with the search for an alternative—for a sanction that could be imposed by the courts for the myriad relatively minor offenses against property. (Beattie 2001: 470–71)

Moreover, the person most responsible for transportation viewed it as a replacement of capital punishment. William Thomson, the author of the 1718 Act, may have died believing he had established “a reliable alternative [to capital punishment] that made the royal pardon credible” according to Beattie (2001: 461). On the other hand, the text of the Transportation Act suggests its main targets were offenders who committed robbery and relatively low-level property offenses, individuals who were not likely candidates for execution. Its preface explains, “The punishments inflicted by the laws now in force against the offences of robbery, larceny and other felonious taking and stealing of money and goods, have not proved effectual to deter wicked and evil-disposed persons from being guilty of the said crimes” (cited in Beattie 1986: 503). Moreover, the Act specifically limited the sentence of transportation to those who would otherwise be “whipt or burnt in the hand” or sentenced to a workhouse (cited in Coldham 1992: 165). Because the intention of the law would have been more accessible to the various judges meting out justice, the letter of the law may in fact have guided judges in thinning the mesh. Thus, the apparently unintended consequences in this case may be attributable to conflicting intentions behind the 1718 Act.

More broadly, however, we may still wonder what accounts for the meteoric rise of transportation sentences once they became available as an official sentence, far outpacing any other available punishment. Tonry and Lynch (1996) remind us that modern diversionary punishments are often motivated not simply as a means of reducing severe punishments, but also to alleviate financial burdens. Similarly, Rusche and Kirchheimer (1939: 58–60) have suggested that transportation was adopted for financial and geographic motivations related to the system of mercantilism in place at the time. In fact, it is reasonable to suppose that, in addition to possible humanitarian incentives to reduce the severity of punishment or social control incentives to make the net of criminal justice denser, transportation also fulfilled political and economic incentives. Politically, those transported to the New World could, in theory, contribute to its founding through their number, directly helping to establish territories as decidedly British and eventually through their ability to produce new citizens who can further enhance the British claim to the land. Indeed, once the British claim to the New World ended, convicts were transferred to the Australian colony, continuing the process there. Research on this later development suggests these kinds of economic and political motivations sustained the practice of transportation to Australia (Nicholas and Shergold 1988); it would be reasonable to believe that similar motivations lay behind the technology's use in both Australia and North America.

Economically, transported offenders could enhance national wealth through their forced labor by creating taxable wealth in the colonies as well as potentially building colonial infrastructure. Admittedly, the quality of the labor actually expended is questionable, for reasons discussed above. However, the success of labor-based incarceration (imprisonment at hard labor and confinement in the house of correction) after the cessation of North American transportation suggests the labor potential of punishments was very appealing to policymakers. At the Old Bailey, sentences to imprisonment at hard labor did not become routine until 1776, the year in which transportation to the New World was interrupted. There were rarely more than ten total imprisonment sentences each year until 1776; in that year, imprisonment sentences increased, nearly by a factor of nine, from their amount in the previous year. Thereafter, with few exceptions, imprisonment sentences remained well above 100 offenders per year (Old Bailey Proceedings Online ND). Indeed, Ignatieff (1978) directly relates the rise of the prison reforms of the 1770s in England to the decline of transportation during the war (but see Willis 2005). The belief that offenders could repay society for their crimes through labor may have transferred from transportation to the use of incarceration. In sum, the popularity of transportation may be explained by the fact that its direct consequences, nation building and economic advancement, were effects no other punishment could produce at the time transportation came to dominate criminal justice.

The second puzzle is the concurrent decline in death sentences for theft offenders and increase in death sentences for non-theft offenders following the 1718 Act. Indeed, this mixture of severity and lenience across offense type is somewhat contrary to previous research. Archer (1991: 246–47) shows that the early Elizabethan air of severity aimed at serious offenders, in both legislation and conviction rates, also translated to severity for low-level offenders as well through higher conviction rates. Thus, given the growing lenience for theft offenders in the period of interest, we might also expect to see similar lenience for non-theft offenders. We might have hypothesized that since theft offenses occupied such a large portion of the docket, judges grew used to passing non-capital sentences in general. But rather than observe a decrease in death sentences for both theft and non-theft offenses, we see increases for non-theft offenders.

Why did death sentences for theft offenders decrease while death sentences for non-theft offenders increase after the 1718 law? Transportation as an official option for conditional pardon may have indirectly impacted capital sentencing. Here, I assume that the severity of judicial preferences remained relatively constant, but that judges acted strategically to maximally realize their preferences. First, death sentences no longer made sense for petty theft offenders as these may be overturned in a full or conditional pardon. Rather than taking the chance that the offender might escape punishment, judges may have been more willing to sentence petty criminals to transportation. Second, judges may have been more willing to sentence non-theft offenders to death, their true preference, knowing that a sufficiently punitive alternative to capital punishment existed in the event of a pardon. Before, judges may have decided to sentence marginal non-theft defendants to a secondary punishment to avoid the scenario in which their capital sentence is overturned and the offender escapes sufficient punishment by securing a full pardon or receives the secondary punishment anyway through a conditional pardon. After 1718, judges may have felt more at liberty to sentence these marginal offenders to death because if the defendant did secure a pardon, these defendants would likely be transported, a far better (from the judge's perspective) punishment than a mild sanction like whipping, branding, or no punishment at all. This rather convoluted calculus may explain why judges were apparently more lenient with theft offenders and less lenient with non-theft offenders after 1718.

The third puzzle is the concomitant decrease in death sentences and increase in executions following the Transportation Act. Why would the availability of transportation have contradictory effects on capital punishment? One potential explanation is that the king and his council, and the judges who recommended pardon, may have been less merciful after 1718 precisely because the pool of condemned criminals had changed. With the changes in sentencing practices, offenders receiving capital sentences may have been (perceived to be) more blameworthy, as the less blameworthy offenders were more likely to be sentenced to transportation. Indeed, the condemned population had a somewhat larger portion of non-theft offenders after the 1718 law, making the population sentenced to death potentially appear more deserving of their sentence. Moreover, commonplace reasons for pardoning—lack of offense severity, age, etc.—would be less common as offenders matching these criteria may have already been sentenced to transportation.

Conclusion

Significance and Implications

This article has added data, rigorous statistical analyses, and nuance to the replacement–mesh-thinning debate over the role of transportation. Some scholars have described transportation as a replacement of or alternative to capital punishment responsible for its declining frequency. Other scholars have suggested that the use of transportation led to unintended consequences, including increased severity in the punishments received by offenders who would previously have received lesser punishments. This study suggests that transportation did both. Transportation played something of a replacement role: the Transportation Act of 1718 apparently reduced the frequency of capital sentences overall. However, we observe more death sentences for non-theft offenders and increased rates of execution in the period after 1718. These findings suggest a limited role for transportation in replacing capital punishment. This study also finds that the official adoption of transportation tightened the mesh of the criminal justice net, a situation that has followed the development of the prison, community supervision programs, and other diversionary punishments since 1800 at least. After 1718, thousands of individuals who previously would have been released from custody with a fine, a lashing, or branded flesh instead endured seven years of hard labor in a foreign land, if they survived the hazardous sea voyage there.

These findings imply both substantive and methodological implications for the historical study of crime and punishment. First, the focus on transportation as an explanation for the decline in capital punishment may have been diverting attention from the fact that something else may have been driving changes in capital punishment. This study has shown that, at least in the London area, the rate of executions actually increased after the availability of transportation. Moreover, while death sentences did decline, transportation does not appear to have snuffed them out, as they remained around 10 percent of all sentences. Consequently, historians must revisit the puzzle transportation was thought to explain. Second, disaggregating the data by crime type shows divergent trends that have been overlooked by general discussions of declining death sentences and executions. As shown above, overall these were declining, but they increased for non-theft offenses. Because theft offenses were in the clear majority, examinations of overall trends may obscure trends for serious offenders.

This study also has implications for the study of contemporary punishment. I have suggested that the presence of mixed motivations, or the ability to appeal to different groups for different reasons, may explain why transportation, an ostensibly lenient reform, had mesh-thinning consequences—consequences that may not have been altogether unintentional. This story is consistent with some previous accounts of mesh-thinning consequences resulting from criminal justice reforms, as discussed above. Given the current climate surrounding criminal justice issues in which politicians can variously benefit from acting tough on crime, reducing the costs of criminal justice, or seeking to reduce the severity of the criminal justice system, it is possible that some reforms disguised as progressive or cost-effective advances will in fact result in more intense punishments. According to the lessons of transportation, penal reforms that enjoy the support of unlikely allies—those seeking harsher or more intense punishments as well as those seeking less intense punishments—may be most likely to have unintended consequences. Punishments that have the capacity to satisfy competing groups must contain elements that appeal to these different groups; thus, broad appeal may be a necessary factor leading to unintended consequences. Whether the punishment in fact has mesh-thinning or other severity-enhancing consequences likely depends on which party is most represented at the level of on-the-ground implementation (see Heimer 1999). Perhaps the ultimate lesson of transportation, then, is that a punishment that can ostensibly satisfy such disparate parties—those with humanitarian, economic, or social control motivations—can revolutionize criminal justice by significantly altering the penal landscape and eclipsing the role of more traditional punishments, particularly if those involved in carrying out the law support such changes.

Finally, this study advances the body of law and society scholarship that examines the law's unintended effects. Legal reforms that frequently look progressive on paper can in reality lead to less progressive outcomes, and the realm of criminal law is no exception. By relating mesh thinning to other unintended consequences of the law examined in the law and society canon, I hope to make explicit the connection between two previously disparate literatures. As with Civil Rights legislation or the broad shadow of the law, criminal law that dictates punishment often diffuses beyond its original target. As with corporations' ability to invert the intention of the law, punishments designed to be less severe can lead to greater punishment. Just as individual discretion may obscure the intentions of the law, judges can act conversely to reformers' intentions. The reasons for these various unintended consequences may differ across settings, but the phenomenon is clearly general.

Areas for Future Research

This study had a narrow focus—empirically resolving an historical puzzle over the true consequences of transportation to contribute incrementally to our knowledge of a broader phenomenon. As such, it left much unresolved. I did not explore the extent to which transportation widened the net of criminal justice by punishing many more offenders than before. Moreover, I only examined the effect of the 1718 Act in London—the extent to which the 1718 Act had similar consequences in more rural or geographically distinct jurisdictions remains unknown. Likewise, we do not know the impact of the earlier adoption of transportation as part of a conditional pardon. Further, the explanations behind these findings remain speculative. An analysis of judges' diaries or letters may explain their choices to sentence offenders to transportation rather than death or a secondary punishment, while an examination of pardon requests may explain why the rate of pardons declined so heavily after 1718. These and other puzzles remain for future research.