The Prosecution Project <https://prosecutionproject.griffith.edu.au/> is a large-scale digital project that aims to provide a new way of exploring the context and impact of changes in the criminal trial during the nineteenth and twentieth centuries. It does so from an elementary platform: the digitization of the court calendars of criminal trials in the higher courts in the six main Australian jurisdictions over time periods as long as 130 years. The objective is to address questions of the criminal justice process centered on prosecution, from arrest, committal, and indictment, to verdict, sentence, and beyond. In a field of historical research that is more often characterized by the richness of discursive analysis, the Prosecution Project's comparative data sets are designed to offer a new understanding of quantitative context over long periods of time. The challenge of building the data platform is, however, considerable, requiring significant planning, collaboration and investment by a large number of researchers, working with relevant archive repositories, and, in this case, assisted by the engagement of an interested community lying outside the regular academy. This article describes the background to the project, its development as a collaborative digital initiative, and its technical and organizational requirements and possibilities, before we explore briefly some of the research outcomes that this project makes possible.
Background
Australian criminal justice institutions are a legacy of the settlement of the Australasian colonies by Britain from the late eighteenth century. As is well known, this settlement was initially shaped by the intention to establish a new destination for convicts sentenced to transportation in Britain and Ireland following the termination of North American transportation. Not all Australasian colonies, however, were established as convict depots, and even those that were, always had a mix of free and convict settlers. An important result from the earliest days of New South Wales was the adaptation of some precepts of English law to the realities of a complex jurisdiction; the relaxation of the law of attaint, which enabled convicts to litigate in local courts, is a much noticed example.Footnote 1 However, it would take some decades for the burden of military government to be lightened by civil administration, leading to important political struggles in colonial New South Wales around issues such as trial by civilian jury.
From their beginnings, a number of colonies were free of the taint of convict settlement, and their legal institutions were established on familiar models. The forms of English Quarter Sessions were adapted for the use of the settlers of the small Swan River colony in Western Australian after 1830, and in South Australia the colony of free settlers boasted of their rights under a supreme court established only a year after first settlement in 1836. As each of the colonies attained a full measure of (colonial) self-government, the statutory provision of a supreme court quickly followed. Judicature statutes typically established colonial supreme courts as single jurisdictions, including ecclesiastical and maritime jurisdictions in their remit.
Criminal jurisdiction was inevitably an important part of the colonial court's business throughout the nineteenth century, and every chief justice was regularly engaged in hearing capital cases and those for the more serious felonies. After the federation of the colonies (without New Zealand) into the Australian Commonwealth (1901), the state courts retained the majority of criminal business, with domestic criminal jurisdiction reserved to them. Unlike in Canada, there was no federal criminal code for most of the twentieth century; although Western Australia (1902) and Tasmania (1924) followed Queensland's lead in legislating a criminal code in 1899. Whether or not, and to what degree, codification of the criminal law made a difference to the prosecution of offenses and trial outcomes is, therefore, a matter on which Australian experience might shed light; a natural experiment, so to speak. Other local idiosyncrasies lend themselves to comparative inquiry: did convict origins exercise any sway in the later administration of justice; did the powerful elements of autocratic governance and the imperial administration of settler colonies have any impact on criminal law and its modes of justice; did regimes of punishment or systems of policing develop heterogeneously in these separate State jurisdictions? Alongside these macro level concerns, the possibility of comparing the experience of law by specific populations in different states also draws attention. Particularly importantly for Australian history and studies of law in colonial contexts was the question of whether or not the different stages at which colonial regimes became dominant over the many indigenous peoples of the Australian continent had any identifiable impact on the administration of justice to and for such people.
To questions such as these, the answers of historians have to date addressed themselves primarily through textual approaches, qualitative inquiry into archives where they are available, and, occasionally, quantitative inquiry into small samples of material. In spite of the strong traditions of government and administration in Australia since colonial times, the survival of legal and related records is uneven. The newspaper record is indispensable, and has been of inestimable importance in filling the gap left by the absence, for example, of published law reports for much of the colonial era. The enduring work of Bruce Kercher and colleagues in building substantial records of case law for jurisdictions that lacked any such body of work is by now well known, and continues to exercise a major influence in expanding the study of legal history in Australia and beyond.Footnote 2 What started as a work of recovery into print form quickly became a major digital resource in its own right. Alongside that initiative, Australian research has more recently benefited from the move into legal historical records available online through AUSTLII, which has established an online Australian Legal History Library to improve access to historical case law, statutes, manuals, and legal ephemera.
For Australian and other scholars working in sociolegal and criminal justice history, case law and statutes are only part of the requirements for understanding the deployment, impact, and experiences of those who are law's agents and subjects. The social history of the legal domain has been a growth area since the 1970s; in Australia it has been able to draw on as well as contribute to the expanding digital resources of the kind outlined previously, but there have been other kinds of resource innovation enabled by the emergence of digital technology and the Internet that amplify the possibilities of understanding criminal justice procedures and outcomes. Two that have directly influenced the vision of the Prosecution Project have been the Old Bailey Online (OBO) database of Old Bailey trial transcripts <http://www.oldbaileyonline.org/> and the Founders and Survivors online database of convicts transported to Tasmania <http://www.foundersandsurvivors.org/>. Through its comprehensive documentation of the criminal trial in a single court over a very long period of time, the OBO has made an immeasurable contribution to contemporary scholarship on criminal justice. Its data is proving adaptable to a very wide range of research questions, ranging from changing ideas about criminal responsibility through to macro-questions about the history of attitudes to violence during the long eighteenth century.Footnote 3 Digital tools are thus shown to facilitate conventional textual analysis while also enabling data mining of a scope that would have been once impossible. Qualitative studies of the criminal trial can now proceed with a greater appreciation of their quantitative context. Inevitably, the digitization of one set of records in one court opens up the question of whether things would be the same or different in other political and cultural contexts. The Australian Prosecution Project seeks to pursue that query.
The enterprise evident in the digitization of the Tasmanian convict registers offers another example of the possibilities of digital applications to support a range of research questions. The mode of digitization pursued there was innovative, involving a partnership between academic researchers and the wider community of family and local history researchers interested in the particular subject matter of the convict records. This partnership enabled the carefully curated transcription of the convict idents (the registers of names and other details of all the convicts who landed in Van Diemen's Land). The result was a database, again available online, searchable by general users, that would also be accessed by researchers interested in exploring the convict transportation process, the experience of convict life and labor, as well as the economic, social, and demographic histories of the populations from which the convicts were principally drawn (Great Britain and Ireland).Footnote 4
These initiatives from cognate fields helped shape our view of how the Prosecution Project might develop its approach to researching the changing course of the criminal trial. Our interest was in approaching the subject in a way that would address the development of distinctive characteristics of a system that owed everything in origin to British legal traditions, but was supplanted to new locations more than two centuries ago. The shared heritage of North American and Australian legal systems could not account for the different traditions of policing, prosecution, and trial procedure in the two countries; a particular issue for Australian studies of contemporary criminal justice that draw, like so many others, on American social science for understanding the criminal justice process.Footnote 5 Likewise, the sway exercised by English legal and political culture in Australian institutional life fell short of accounting for emergent differences in penal culture, or in accommodating cultural difference of the kind posed by the presence of indigenous peoples with persistently challenging legal cultures of their own. Much could be learned from comparison of Australia with other settler cultures, and North American, New Zealand, and South African legal and sociolegal histories have been particularly fruitful for Australian scholarship.Footnote 6 However, in many respects, productive comparison has been hampered by the simple absence of sustained, broad, durable, and replicable study of the criminal justice process in any of these countries over the long duration between foundational decades and the contemporary setting. Much is known, for example, about the history of women's experience of sexual violence, or the sorry record of criminal law's dealing with indigenous peoples in punishment of inter-racial crime, or even the history of violence, or of public order and dissent in labor and working class history, but almost nothing is known about the course of the criminal trial in its everyday operation against the majority of those accused (in Australia, young white males accused of property crimes). Likewise, most crime studies in Australia as elsewhere do not extend beyond the earliest decades of the twentieth century, a limitation also evident with respect to other digital repositories of crime data.Footnote 7 These were the kinds of gaps that we imagined might be filled by a more comprehensive approach to the criminal justice archive, to which we now turn.
A Longitudinal Database of Criminal Prosecutions
The Prosecution Project database is a record of criminal trials in Australian higher courts over more than 100 years, from as early as 1830. Potentially, the database is a record of all those trials. Because the database is a reconstruction based on the original trial registers, it represents a more authoritative data set than any provided in annual reports of agencies, allowing for greater analytical power by showing the influence of a variety of variables such as time, region, offense, judge, and defendant sex and race on prosecution process and outcomes. Because it is also comprehensive, it has the potential to test the learning of previous research based on small samples usually shaped only by a narrow set of research questions. Because it is a structured database of unique and persistent records of a wide range of attributes centered on the notion of a single trial event, it is also adaptable to a wide range of empirical investigations. In effect, this database becomes a “once-entered, use many times” data set of a kind too little used in historical research. We explore later some of the ways in which we consider that these data may generate new ways of understanding the trial in context, its antecedents, and its outcomes. In this section, we describe the process of building the database, its sources, and its curation. Later, we consider the sources; the design of the database; the mode of entering and checking data; the capacity for sampling, extending, and linking the data; and the question of archiving and access.
Sources
As described, Australian criminal law is administered primarily at the state level. Court administration is organized centrally. Smaller states have their criminal jurisdiction for indictable offenses centered mainly in the Supreme Court; larger states have devolved the handling of less serious offenses to district courts or quarter sessions. Our starting point for the project was the record of the Supreme Courts. The court registrar or similar official has responsibility for maintaining court records, which include registers of cases. Our aim has been to digitize wherever possible the register of criminal trials maintained at the Supreme Court registry. In one state, Queensland, the most continuous set of data is found in the Register of Depositions, a clerical device originally designed for managing the flow of depositions among the courts, prosecutors and relevant parties including case solicitors. Over time, the register developed as the most thorough record of cases. In Western Australia, the “Criminal Indictment Registers” date from 1830, just a year after first settlement, and continue in the same format until the 1960s. Registers vary in the types of information available. In Victoria, the data by early twentieth century included name; committal date and location; trial date and location; judge, prosecutor, and defending counsel; names of witnesses (including their titles if a police officer or medical expert); plea and outcome including sentence when convicted; and appeal outcome when that applied. Because the organization of the criminal court varies among states, for some jurisdictions we are also collecting quarter sessions records, which again are typically maintained in registers over long periods of time. For the most part, the registers are in manuscript, unlike the records of some other jurisdictions such as England, where quarter sessions records were sometimes printed from the later nineteenth century.
Design of the database
A priority issue for the researchers on the Prosecution Project was the design of a database that was collective in its construction, robust in its data management, and durable. In place of the bespoke efforts of individual researchers working with Excel spreadsheets or word processed tables, idiosyncratic in sampling and coding, unable to be shared or even checked by other researchers, we sought to provide a reconstructed criminal trial archive that could be and would be used by many researchers and even community users, according to their information needs. Previous research experience had already highlighted the advantages of a relational database, but that had been in the pre-Internet phase of mainframe computing.Footnote 8 The possibilities opened up by new technologies since the 1980s have shaped an entirely new approach. There were limits on what was possible. Machine transcription was not feasible; no level of optical character recognition (OCR) technology currently available can cope with the messiness and variation of the handwritten records we are dealing with. Instead, we could contemplate data entry of manual records into a single database on a range of devices (laptops, desktops with different operating systems, and even iPads or tablets) if we could deliver to those devices the visual images that we wished to have transcribed. We could manage data drawn from a range of jurisdictions and register formats through the flexibility offered by the relational database, with a multitude of data tables defining the different characteristics of each jurisdiction's data. We could make some data available to public inspection through a web portal, while preserving other data for research purposes. We could make use of the availability of other digitized data to verify and enrich our own data.
Doing these things required a considerable investment of time and resources in a collaborative interdisciplinary environment. The researchers worked from the beginning in regular meetings, usually every two weeks, with a team of eResearch specialists and database and web designers. The result was a database managed through a secure web portal, with the project researchers having a high degree of control over the development of the data structures, including defining attributes (e.g., trial date), creating glossaries (e.g., judges' names), and adding new jurisdictions or court levels (e.g., quarter sessions). The early data entry process was conducted largely by the project researchers with employed research assistants. A significant new addition in capacity has been the engagement of volunteers, who are able to enter data through the web portal from images delivered to them in a browser. Many of these volunteers have a background in genealogical research, although they are not often familiar with criminal justice records. The design and maintenance of the database has, therefore, also required systematic quality assurance, with records being amended after checking by project researchers.
Data are entered as they appear in the original register, organized into attributes (data fields) appropriate to the particular jurisdictions. A unique record is established for each individual appearing in a single trial (whether or not appearing on multiple charges, which are enumerated and described for any particular individual trial); co-accused are each given separate entries with outcomes as determined. Table 1 summarizes the scope of the main collections at May 2016, after approximately 30 months' progress.
Table 1 Prosecution Project: data by jurisdiction and time
Extending and linking the data
By itself, the database capturing the information entered just in the court registers would already constitute a significant new resource in exploring the changing context of the criminal trial in Australia, offering comparison with what is known of the history elsewhere. We consider later in this article some of the kinds of research analyses made possible with these data. However, it is also important to recognize the potential of the data as they are enriched by other data sets: the basic case data amplified with information about the context of an offense, the attributes of the offender and victim, the location of a crime, and the characteristics of other players in the trial process including judges and defense counsel. In the context we are dealing with, this process of enriching the data is made possible through semiautomated linking to the National Library of Australia's Trove database of digitized newspapers (more than 1,000 titles across 1803–1954)Footnote 9 ; as well as manual searching of other digital data including law reports through the AUSTLII database, and the various state police gazettes, generally available in Australia from approximately 1860–1950. The last of these (the police gazettes) offer a particularly rich record of policing practices and changing information technologies, including crime reporting, warrant processing, and criminal identification. Data linkage thus enables the enrichment of case information and the verification of names and outcomes, as well as offense context, work typically more labor intensive than the first entry of data, and, therefore, being conducted more often as the researchers work with specific samples (e.g., an offense data set for a particular jurisdiction, or a comparison of plea data for a number of jurisdictions).
Such extension of the data amplifies the possibility of contextualized analysis of the criminal justice system at case and population level. As we move outwards from the case to the web of connections that characterize that case, we build a three-dimensional view of the criminal justice process from the reporting of a crime to police through arrest, committal proceedings, trial, sentence, and appeal. By approaching the data at a population level and over long periods of time, we can understand more completely the process, with its numerous exit or diversion points (withdrawn prosecutions, trials avoided through guilty pleas, sentences suspended or commuted, appeals overturning convictions, and postconviction declarations of habitual criminal status). And as we will explore later, such amplified data also enable us to illuminate different agents of the criminal prosecution obscured in the bare record of a trial and its outcomes: the judges, juries, witnesses, prosecutors, defense counsel, police, and the co-accused.
Archiving and accessing the data
Finally, we note our long-term ambition that the project should become a resource for future users. With careful curation of the data and a plan for their archiving and retrieval in the future, the Prosecution Project database will provide a robust resource for researchers exploring a wide range of questions of legal and social history. Its inclusion of a number of jurisdictions in Australia offers particularly valuable prospects for comparative study. Importantly, the design of the project enables research in other jurisdictions, subject to whatever institutional arrangements may be necessary to secure that possibility. Already we have such a prospect in place: one of our team who is engaged on a comparative study of Australian and English criminal justice responses to offenses against children is adding records to the database from the West Yorkshire Quarter Sessions. Appropriate institutional support will be necessary to maintain the database as a “live” concern, enabling the addition of new records, beyond the funded date of 2018. With that qualification, we signal here the intention to provide research data collections through the project portal on an open access basis in the future. In the meantime, we welcome enquiries for collaboration on the existing data and on adding new data sets.
Having outlined the context and principal features of the Prosecution Project database, in the remainder of this article we explore some of the kinds of analysis that this project makes possible.
Broadening and Deepening Criminal Justice Knowledge
The data being compiled by the Prosecution Project offer a broader comparative perspective than that of the traditional studies by individual researchers, who necessarily have to limit their analysis to particular offenses, time periods, regions, or types of offenders. At the same time, the project data allow for a deeper understanding of the criminal justice process than that provided by studies that rely predominantly on official statistics. The criminal registers from which the database has been compiled include many details not recorded in such official statistics, and are not subject to the limitations imposed by the aggregated nature of such data. This reconstruction of the original data allows analysis to take into account the inter-relationship of variables over time. Through case enrichment as described, we also situate the case in its spatial as well as legal contexts. Through access to the records of a single jurisdiction over long periods of time, the Prosecution Project offers new possibilities for exploring subtle changes (e.g., sentencing, the plea process) in the trial, and their consequences.
For the purposes of this article, we offer some preliminary findings based on a point-in-time access (January 2016) of 52,495 trials entered in relation to the four jurisdictions for which either complete populations or large samples across time have so far been entered into the database. These are Western Australia (15,061 records spanning the 1830s to 1960s), Queensland (15,137 records, 1860s to 1960s), Victoria (11,157 records, 1860s to 1960s), and New South Wales (11,140 records, 1860s to 1950s). Variations in the original register collections and missing information in original trial records, as well as the ongoing process of data cleaning means that information across every category is not available for all 52,495 trials. Nevertheless, the overall large sample size means that indicative results can be provided in relation to a number of legal issues that have received limited historical attention, although they have become a regular item on the agendas of contemporary criminological and sociolegal research. These include matters such as judicial discretion in sentencing, bail, the effect of legal representation on outcomes, jury decision making, and the role of guilty pleas. Finally we will comment on the significance of these data for understanding the categorization of crimes in official statistics and sociolegal/criminal justice histories.
For four of the six jurisdictions in our data, we have consistent data on the judge presiding at trial. Therefore, the database will enable unprecedented historical examination of variations in judicial sentencing. Eliminating from the 52,945 trials those cases involving judges who presided over fewer than 300 trials (in order to provide a robust sample when analyzing by offense and sentencing type) leaves a sample of 26,148 trials presided over by forty-nine judges across the four jurisdictions. Statistical analysis of sentencing by offense for these high-volume judges suggests a highly significant effect of judicial officer in the sentencing outcomes for convictions involving sexual offenses, homicides, and the more serious forms of property crime or assault. Moreover, there was clear variability among judges as to when particular types of sentences were deemed appropriate; in the case of placing defendants on bond or suspended sentences, for example, some judges distributed such sentences equally for those convicted of property or personal crimes, but other judges favored one category of crime much more than the other in delivering those sentences. For example, whereas Patrick Real during his tenure on the Queensland judiciary (1890–1922) was more likely to institute bond or suspended sentences in cases of crimes against the person, his fellow Queensland judge Granville George Miller (1882–1910) was five times more likely to award such sentences to property offenders across an overlapping period. This process also reveals judicial outliers, such as Sir Joseph Henry Hood, a Victoria Supreme Court justice from 1890 to 1921, who was far more severe in the length of the prison sentences he dealt than other judges from his jurisdiction and period.Footnote 10 For example, Hood gave terms of more than 3 years' imprisonment to 36% of those who appeared before him on breaking and entering charges, compared with the 8% who so appeared before his contemporaries Sir Thomas à Beckett (1886–1917) and Henry Hodges (1889–1919).
Bail, a neglected area of historical study of the criminal justice process, can be analyzed from the Prosecution Project database. Details of whether defendants were admitted to bail in the lead-up to their trial are available for more than half the cases in the Queensland and Victoria records. Bail was allowed in approximately 56.4% of the 18,396 trials from these jurisdictions in which such details are known. Admission to bail rose from approximately 33.7% of trials in the 1860s to approximately 67.4% in the 1950s. There was a relationship between bail and offense type, with defendants significantly more likely to be released on bail if charged with crimes against the person than crimes against property (64.5–51.3%). This relationship was present across both jurisdictions, but was stronger in Queensland than in Victoria. Perhaps most importantly, being released on bail was very strongly associated with a higher likelihood of acquittal or the abandonment of the case by the prosecution. Such evidence prompts further inquiry into the reasons for that association: how far were doubts about the strength of a case already in play at the committal stage, thereby shaping a bail decision? Or, as contemporary criminologists who have discerned similar trends have speculated, did pretrial release better enable defendants to present themselves and their case in a more favorable light, whereas prolonged detention hampered their efforts to properly consult with legal counsel and perhaps encouraged them to simply plead guilty?Footnote 11
The evolution of the modern criminal trial has been of much interest to historians of the legal profession,Footnote 12 but what role did lawyers play in shaping outcomes for defendants, once their role was consolidated? In Victoria, whether a defendant had legal representation was also sometimes recorded in the trial registers, more consistently so from early in the twentieth century. The availability of these data will be of immense value in tracking the history of legal representation in the nineteenth and twentieth centuries. Of 5,451 cases in which this detail is known for Victoria, the accused had representation in 81.3%. Analysis of missing data suggests this may be slightly skewed, indicating that the presence of a lawyer was more likely to be noted than the absence of one. Initial analysis suggests that defendants accused of personal crimes were far more likely to be defended than those accused of property offenses, and that women were more likely to be defended than men. Representation rates rose across the twentieth century, likely influenced in part by the introduction of legal assistance schemes. This was significant, as having legal representation dramatically increased the likelihood of acquittal across both offense categories, and for both men and women. Whether legal representation determined that outcome is a question for future investigation.
Another issue that is ignored in most official statistics but which the Prosecution Project database will illuminate is that of co-offending. Little sustained historical analysis has been conducted on the large number of defendants who faced trial not alone, but with co-accused, although criminology has long recognized co-offending as an important element in understanding criminal behavior.Footnote 13 Approximately one quarter of the accused in the sample were tried alongside other defendants, with the largest trial involving thirty-seven co-accused tried for unlawful assembly after a political demonstration in Brisbane in 1939.Footnote 14 The database reveals that co-offending was significantly more common in relation to crimes against property than in relation to those against the person. However, when the data are disaggregated, considerable variation emerges in relation to individual offenses. Whereas defendants charged with embezzlement or stealing as a servant, clerk, agent, or trustee were almost always tried alone, almost half of those tried for burglary or robbery were co-defendants.
Apart from the simple failure of official statistics to record information on topics of potential interest to historians, one of the biggest drawbacks for researchers reliant on government statistics is the tendency of officials to aggregate information.Footnote 15 This is fine if the topic to be analyzed concerns only statewide or national crime trends, or only requires knowledge of the overall conviction rates. However, it becomes problematic for researchers interested in examining regional differences in criminal offending and prosecution, or the inter-relationships and co-dependencies of the many factors shaping prosecution. As the Prosecution Project data show, these nuances can be important.
A central element of the project is its comparative nature, bringing together data from different states into a format that allows for easy comparison. This shows, for example, that acquittals were more likely in New South Wales than in Queensland, and that sentencing penalties were harsher in Western Australia than in Victoria. However, it also allows nationwide comparison of metropolitan and regional variation. This is a significant advantage as well over the Old Bailey database, which Sharon Howard suggests has led to a metropolitan distortion in English criminal research by skewing scholarly attention disproportionately toward London, while the records of harder-to-access regional areas remain untouched.Footnote 16 Of the 52,495 trials from New South Wales, Victoria, Queensland, and Western Australia, 23,555, or 44.9%, took place outside the capital cities of Sydney, Melbourne, Brisbane, and Perth. The data also suggest that there was a statistically significant relationship between trial place and outcome, with regional juries more likely to acquit than metropolitan ones, although this relationship was less apparent in New South Wales than in the other three jurisdictions.
Official aggregation of verdicts into cases of conviction, acquittal, or the abandonment of prosecution likewise obscures key issues and changes in the trial process. In particular, it reduces the role of the jury to mere arbiters of guilt or innocence, whereas historically they also played a part in determining the degree of guilt through gradations of guilty verdicts, possibly affecting sentence outcomes. Sympathy for a defendant or the circumstances surrounding a crime, if it did not result in jury nullification, could result in a partial or alternate verdict. Such verdicts were returned for approximately 5.3% of the 52,495 trials; however, they were notably more common in relation to some offenses than others. For example, of the 2,521 defendants tried for murder, in 532 instances (17.9%) the defendant was found guilty of the lesser crime of manslaughter. The likelihood of such an outcome increased in the mid-twentieth century.
Jury sympathy for defendants, and their desire to play a role not only in determining the prisoner's guilt but also the prisoner's ultimate level of punishment, also led to a number of verdicts of guilty with a recommendation for mercy. Such recommendations applied to 791 (4.6%) of all 16,987 guilty verdicts. Again, it was particularly likely in cases of murder. Very occasionally, the registers also provide the jury's reason for instating the recommendation to mercy: in forty-five cases it was because of a defendant's youth; in thirty-two cases it was because of the provocation suffered; in twenty-one cases it was because of the defendant's good character; and in others, poverty, health issues, or ignorance of the law was cited. The database also reveals the extent to which juries struggled to arrive at an ultimate verdict. Currently, only 1.6% of the trial records entered into the database terminated with the jury being unable to agree, demonstrating the considerable pressure placed on juries to reach a verdict historically.Footnote 17 Nevertheless, some types of offenses evidently remained more prone to disagreements than others. In particular, juries were unable to agree in 8.7% of the 241 abortion trials, 5.1% of the 410 indecent assault trials, and 4.1% of the 751 arson trials.
As well as highlighting the historical importance of juries, the database also documents their gradual disappearance in the face of rising guilty pleas. Whereas scholarship exists on the guilty plea in other jurisdictions, there has been little research on its history in Australia.Footnote 18 Initial results suggests the guilty plea, present in 14,030 of the 52,495 trials, began its rise in the 1890s and continued to grow through the early twentieth century, before climbing dramatically in the 1950s and 1960s to account for the bulk of prosecution outcomes. (Its increasing use is also clearly associated in these data with the number of charges levelled against a defendant.) Similarly, the database can be used to analyze the type of cases in which prosecution was likely to be abandoned either before or during the trial (which occurred in approximately 5.3% of all trials), as well as other rare outcomes such as findings of not guilty by direction (0.7%), not guilty by insanity (0.4%), failure of defendant to appear (0.2%) and the rare autrefois acquit ruling (only three cases).
The aggregation of data in official statistics has fundamentally affected the way that criminal offenses themselves are understood and categorized. When officials first began to collate crime data in the early nineteenth century, they devised categories of offending that have proved remarkably durable, dividing crimes into the categories of personal, property, public order, and statutory offences. Crime historians have largely accepted these categorizations, employing them in their own analyses. Collapsing offenses into discrete categories makes sense, given the huge variability in data. Several thousand different offense descriptions are currently entered into the Prosecution Project database. Even when these are sorted into offenses of roughly the same nature, several hundred distinct offenses remain. Apart from the expected offenses, these include such obscure crimes as unlawfully obstructing the sending of a telegraph message, causing an obstruction on a railway line, personation during an election, unlawfully celebrating a marriage, and fortune telling.
Despite this variation, the naturalization of the crime categories used in official statistics has tended to result in a focus in scholarship on the crimes that fall under the two main categories of personal and property offending. In contrast, offenses that do not fit, or fit poorly into these groupings, have attracted less historical attention. The neglect of other types of offenses might also be caused by the difficulty in generating robust samples given the relatively small annual numbers constituted by such crimes. Viewed across time, however, crimes that fall under the category of statutory or miscellaneous offenses constitute a more noticeable quantum of criminal trials. Within our sample, for example, there were 539 trials for bigamy, 614 trials for attempted suicide, and 805 trials for perjury. Furthermore, it seems that the prosecution of these more unusual offenses varied among jurisdictions, with approximately 35% of the bigamy trials occurring in Queensland and almost 40% of perjury trials taking place in Victoria. Were there identifiable jurisdictional, chronological, statutory, policing, or social factors that shaped these differences? The availability of the Prosecution Project data will prompt greater investigation of such offenses.
Another problem associated with the hegemony of the crime classifications used in official statistics is the anomalies that present potential problems in terms of the usefulness of these groupings. First, they group together crimes of very different levels of severity, conflating murders with assaults, and armed robberies with shoplifting. However, the degree of aggravation of the offense would likely influence a defendant's experience of the criminal justice system, including choice of charge by the prosecution, as well as conviction and sentencing outcomes. Burglary defendants, for example, were far more likely to be acquitted than those tried for the lesser offences of breaking and entering or stealing in a dwelling. In line with existing research on historical attitudes to nonlethal violence, however, Prosecution Project data suggest that persons tried for minor assaults were significantly more likely to be acquitted than those tried for more serious acts of violence.Footnote 19
In aggregating crimes existing categorizations can obscure important differences among crimes within the major categories. The bulk of offenses designated as property crimes relate to theft, be it by burglary, larceny, or fraud. However, the category also includes crimes such as arson and property damage; the only connection between these offences and theft is that they both involve property. However, whereas the motivation of offenders who commit burglaries or larcenies may be similar, the intent of arsonists—and the issues of relevance at their trials—usually differs. The differing nature of the offenses likely also influenced attitudes toward them, as suggested by the lower conviction rate that pertained in arson cases compared with property crime in general. Again, Prosecution Project data allow these outcomes to be viewed from new perspectives; the association of arson with mental health issues also meant that it was more associated with findings of not guilty by reason of insanity than other property crimes. These factors imply that the incidence and prosecution of arson offenses more closely follow the patterns found within crimes against the person.Footnote 20 Similarly, the blurring of violence and theft within acts of robbery meant that statisticians have long struggled with whether it should be included in personal or property offending.
The ease of information access facilitated by the Prosecution Project may encourage researchers to rethink these existing classifications. Reviewing existing crime categories also has the potential to encourage scholars to develop research that examines commonalities across offense types, as well as differences among them. Researchers could group offenses according to the severity of offenses using either the maximum sentence possible for different crimes, or the actual sentence range that the data show was likely to result from such offenses. In 2001, criminologists Brian Francis, Keith Soothill, and Regina Dittrich proposed a schema for evaluating offense seriousness across time using a paired-comparison methodology; the only drawback was that it required large data sets.Footnote 21 In addition to facilitating such forms of analysis, the Prosecution Project data might enable innovation in developing meaningful categorization based on point-in-time shifts. Studies of particular groups of offenders, such as female defendants or, in Australia especially, indigenous offenders, might benefit from analysis based on categorizations of the likelihood of involvement by that type of offender at different points in time. Other modes of classification might involve groupings based on the common motivations behind the crime, such as acquisitive versus non-acquisitive offending, or the means employed to perpetrate the offenses, such as crimes of deception versus crimes of violence.
Conclusion
By creating a digital archive of longitudinal prosecution information across multiple jurisdictions, the Prosecution Project offers a radical improvement in the types of historical quantitative studies that can be undertaken in relation to criminal justice. This is a repository of potential value to criminologists and legal scholars, as well as to criminal justice historians, and even ordinary citizens interested in family or social history. The data that the Prosecution Project will make available to researchers moreover offer an opportunity to examine neglected areas of criminal justice research, be it in relation to more obscure offenses or verdicts, or to legal issues that have yet to attract significant attention. It is, therefore, not only the breadth of data being collected by the Prosecution Project that is important, but also their depth and adaptability to the needs of researchers. We anticipate that this will encourage innovation not just in terms of the results generated, but also in the questions asked of the data. By this means, the Prosecution Project offers the possibility of enriching our understanding of transitions in the criminal justice system in the nineteenth and twentieth centuries to match the achievements of scholarship on criminal justice and its social and legal contexts in the long eighteenth century, and beyond.Footnote 22
