JustResearch Edition no. 13
- 4. Child Sexual Exploitation in Canada: Incidence, Sentencing and Recidivism
- 5. Legal Problems and Vulnerable Groups in Canada
4. Child Sexual Exploitation in Canada: Incidence, Sentencing and Recidivism
Jeff Latimer, Principal Statistician, Research and Statistics Division
The past two decades have given rise to an increased level of understanding and awareness concerning the sexual exploitation of children. In Canada, there are three broad categories of criminal behaviour which can be grouped under the term child sexual exploitation (CSE): child sexual abuse, child pornography, and child prostitution. These categories, however, are clearly not mutually exclusive. For example, child sexual abuse victims are often also victims of child pornography. For the purposes of this article, however, the three categories will be presented as three distinct types of criminal behaviour.
This article represents a brief quantitative summary of existing adult criminal court data on child sexual exploitation cases in Canada. The three main research questions are:
- How many cases of CSE are processed in adult criminal courts?
- How do the courts respond to the incidence of CSE?
- What is the recidivism rate of CSE offenders?
In order to answer these questions, data from the Adult Criminal Court Survey (ACCS), which is managed by the Canadian Centre for Justice Statistics, Statistics Canada, will be used. Please note that these data will undercount the true prevalence of child sexual exploitation as a large proportion of child sexual exploitation incidents are not reported to police or child welfare authorities (Latimer 1998).
It also is important to understand that most of the data contained in this article are presented according to the most serious offence in a case based upon a method developed by the Canadian Centre for Justice Statistics. Therefore, some incidents of child sexual exploitation will not be captured herein if there was another offence associated with the same case that was deemed more serious. For example, if an accused was charged with sexual interference and attempted murder, the attempted murder would be considered the most serious offence and the sexual interference charge would therefore not be recorded as a case of CSE. In the section on recidivism, however, please note that this rule does not apply. Cases of child sexual exploitation were identified even if there was a more serious offence in the case, in order to understand offending patterns for anyone who has been convicted of a sexual offence against a child.
Finally, the data in this report does not represent the entire country. New Brunswick and British Columbia only began reporting to the ACCS in 2001/2002, increasing the survey coverage from 80% to 90% of the national adult criminal court caseload. Manitoba does not yet report to the ACCS. In the section on recidivism, Manitoba, New Brunswick, and British Columbia have been excluded as a result of this limitation. The Northwest Territories has also been excluded from the recidivism analysis due to coverage issues since the creation of Nunavut.
Question 1: How many cases of CSE are processed in adult criminal courts?
In 2002/2003, there were a total of 2,854 cases of child sexual exploitation in Canada. Table 1 provides court-based data from the ACCS on the number of cases that were processed in Canadian adult courts over a five-year period. Since these data are not comparable across years due to coverage limitations, it is not possible to discuss trends. Between 2001/2002 and 2002/2003, however, the same number of jurisdictions reported to the ACCS. During this time, there was an overall increase of 12% in child sexual exploitation cases with a 12% increase in the number of sexual interference cases and a 17% increase in the number of sexual exploitation cases. The proportion of cases that proceeded through an indictment was relatively consistent (i.e., approximately 50% or more) across the time periods.
Question 2: How do the Canadian courts respond to the incidence of CSE?
Table 2 provides information on the most serious decision in a case in 2002/2003 using the following hierarchy: guilty, acquitted, stayed/withdrawn, and other. The overall conviction rate for cases of child sexual exploitation in Canada in 2002/2003 was 38.5%, which is much lower than the general conviction rate in adult court (60%), lower than the conviction rate for violent offences (50%), and slightly lower than the conviction rate for sexual assault cases (41%).
Cases that proceeded by way of an indictment were less likely to be stayed or withdrawn (41%) compared to cases that proceeded summarily (59%). Cases where the most serious offence was possessing or accessing child pornography had the highest conviction rate (60.4%), followed by communicating with a child for the purposes of prostitution (55.2%), distributing child pornography (54.6%), and exposure to a person under 14 years of age (50.8%). Cases of anal intercourse and living off the avails of prostitution were rarely convicted in adult court. The conviction rate for the remaining offences ranged from approximately 30% to 40%.
Table 3 provides information on the most serious sentence in a case from the year 2002/2003 based upon the following hierarchy: custody, conditional sentence, probation, fine, and other. For example, if an offender received both custody and probation, the most serious sentence in that case would be custody. Almost half (47.2%) of all child sexual exploitation cases resulting in a guilty decision received a custodial sentence while 29.1% resulted in probation and 21.5% resulted in a conditional sentence. In comparison, 47% of all sexual assault cases in the ACCS received custody, 32% received probation, and 15% received a conditional sentence. When looking at all convicted cases in the ACCS, we can see that the numbers are somewhat different: 35% of cases received custody, 30% received probation, and only 4% received a conditional sentence.
CSE cases that proceeded by way of an indictment were much more likely to receive custody (72%) compared to those cases that proceeded summarily (28%). When examining offences for which there were five or more cases in a cell in Table 3, we notice that incest was the most likely offence to receive a custodial sentence (87.5%), followed by invitation to sexual touching (57.5%) and sexual interference (49.7%). Exposure to a person under the age of 14 years was the least likely offence to receive custody (19.4%).
Question 3: What are the recidivism rates of CSE offenders?
Before such a question can be answered, three important methodological factors need to be discussed. In any study of recidivism, the results will vary depending on the definition of recidivism, the sample of offenders, and the follow-up periods used in the research. Generally, broader definitions of recidivism, higher risk samples, and longer follow-up periods produce higher recidivism rates. Therefore, the results of the existing research provide conflicting and varied answers to this question depending on the method utilised. That being said, however, the existing Canadian literature does provide some indication of the recidivism rates among child sexual exploitation offenders.
Before exploring the recidivism of CSE offenders, it is useful for comparative purposes to examine the general recidivism rate of all offenders. In a recent study, the general recidivism rate for federal offenders (i.e., offenders serving custodial sentences of two years or more) was reported at approximately 44% within two years of their release (Bonta et al. 2003).
Harris and Hanson (2004), in their extensive summary of existing data sources, found that within five years, the sexual recidivism rate for general sexual offenders was 14%, and by 15 years, this number had increased to 24%. Harris and Hanson (2004) further examined types of sexual offenders (e.g., rapists, incest offenders, molesters) after 15 years and found that CSE incest offenders (intra-familial) had a much lower rate of sexual recidivism (13%), while CSE offenders who molested girl victims (extra-familial) demonstrated a 16% sexual recidivism rate and CSE offenders who molested boy victims (extra-familial) demonstrated a 35% sexual recidivism rate.
Langevin et al. (2004) recently published a study of 320 sexual offenders who were seen for psychiatric assessment between 1966 and 1974, using a 25-year follow-up period. The authors reported that 88% of the offenders had sexually re-offended within the follow-up period. This is obviously much higher than Harris and Hanson's (2004) figure of 24%. As noted previously, however, there are differences in method that may account for much of the discrepancy.
First, Langevin et al. (2004) included convictions, charges, court appearances and self-reported criminal behaviour found in hospital records in their definition of recidivism while Harris and Hanson (2004) used official charges and convictions only. It is not surprising that using a broader definition of recidivism produces higher recidivism rates. Second, Langevin et al. (2004) used a sample of psychiatric patients which diminishes the generalizability of the findings to all sexual offenders. This sample consists of offenders who were assessed for a psychiatric illness and, as such, likely represents an unique sub-group of sexual offenders. Harris and Hanson's (2004) sample was primarily comprised of offenders released from custodial facilities and therefore may be more representative of the general sexual offender population.
Hanson et al. (1995), using a 15-30 year follow-up period with CSE offenders only, reported a 42% recidivism rate for any new sexual and/or violent crime. The 42% recidivism rate for CSE offenders was much lower than the 88% reported in the Langevin et al. (2004) study although the time periods were roughly similar. The difference may again be related to the definition of recidivism and the characteristics of the sample, as Hanson et al. (1995) used reconvictions to measure recidivism and sampled only CSE offenders released from prison.
In order to provide a recidivism rate that would be more generalizable to the entire CSE offender population, data from the Adult Criminal Court Survey, which covers virtually all criminal convictions across Canada, were analysed. All offenders convicted of a CSE offence in 1998/1999 were selected for this analysis. Since these were essentially all of the available cases in the eight provinces and territories reporting to the ACCS for that year, it is likely these results would be highly generalizable to child sexual exploitation offenders in general. In order to measure recidivism, the first new conviction within five years of the original conviction date was identified. This provided all offenders in the sample with the same five-year window for the conviction of a new offence following their child sexual exploitation conviction. In addition to general recidivism rates, violent non-sexual recidivism rates, sexual recidivism rates, and child sexual exploitation recidivism rates were calculated. It is important to identify the type of recidivism, as offenders may be reconvicted of rather minor crimes, such as offences against the administration of justice (e.g., probation violation) or theft, which do not pose the same level of risk to society.
The results in Table 4 indicate that 29% of CSE offenders were convicted of a new offence during the five-year follow-up period. A smaller proportion were convicted of a new violent or sexual offence – 9% of CSE offenders were convicted of a new violent non-sexual offence, 4% were convicted of a new sexual offence, and 3% were convicted of a new child sexual exploitation offence. These rates are significantly lower than those reported in the literature (e.g., Langevin et al. 2004; Hanson et al. 1995).
There are several probable explanations for the reduced recidivism rates contained in Table 4 compared to the rates found in other Canadian research. First, recidivism is defined as a new conviction, while other studies have used convictions, charges, and self-report information from mental health records. As discussed previously, since not all child sexual offences come to the attention of authorities, using convictions generally produces lower rates of recidivism. Second, the sample of offenders is essentially all child sexual exploitation offenders in 1998/1999, thus including those offences at both the low- and high-end of the seriousness spectrum and those offenders at the low- and high-end of the risk spectrum. In previous research, samples have typically been drawn from psychiatric populations or prison populations and, therefore, likely included higher risk offenders. Third, the five-year follow-up period, while relatively long and acceptable for measuring recidivism, was not nearly as long as those found in many of the published studies (e.g., 15 years). Finally, due to the data linkage method used in calculating recidivism rates, it is possible that a small number of offenders were not identified as recidivists.
A second point to consider when looking at recidivism rates is the impact of sentencing. Recent research using meta-analytic techniques, which aggregate a substantial number of previously conducted studies, has consistently demonstrated that custodial sentences are associated with slight increases in recidivism compared to non-custodial sentencing options (Gendreau et al. 1999; Smith et al. 2002). Furthermore, these same meta-analyses demonstrate that longer custodial sentences are also associated with slight increases in recidivism compared to shorter custodial sentences. Both these differences, however, diminish when the data are weighted according to sample size. Nonetheless, there is clearly no evidence to suggest that prison reduces the likelihood of recidivism. In fact, within a random sample of young offenders from youth court in Toronto and Halifax, Latimer and Dowden (2005) found that youth who received custodial sentences were twice as likely to be convicted of a new offence within three years compared to youth who received non-custodial sentences, even after controlling for criminal history, age, gender and the seriousness of the offence.
The ACCS recidivism data further supports this relationship between custody and increased recidivism. Among those CSE offenders who had no prior convictions in adult court before their CSE offence in 1998/99 (i.e., first-time offenders), 22% of those who received a custodial sentence had a new conviction in the five-year follow-up period compared to 16% of those who received a non-custodial sentence. Of those CSE offenders who had at least one prior conviction before their CSE offence in 1998/98 (i.e., repeat offenders), 56% of those who received a custodial sentence had a new conviction in the five-year follow-up period compared to 43% of those who received a non-custodial sentence. Therefore, these data suggest that those CSE offenders who receive custody tend to have higher rates of recidivism compared to those accused who receive non-custodial sentences regardless of their criminal history , which is one of the strongest predictors of recidivism. It is important to note, however, that this analysis did not control for other factors that may explain differences in recidivism, such as the seriousness of the CSE offence and characteristics of the accused (e.g., criminogenic needs, age, gender).
The following conclusions can be made based upon analysis of ACCS data and the limited literature reviewed for this article:
- Child sexual exploitation cases are less likely to result in a conviction (38.5%) compared to the general conviction rate for all offences (60%) or the conviction rate for all violent offences (50%). It is similar, however, to the conviction rate for all sexual assault cases (41%).
- Two-thirds of child sexual exploitation cases (68.7%) receive either a custody sentence (47.2%) or a conditional sentence of custody (21.5%) as the most serious sentence.
- The Crown election is directly related to the severity of the criminal sanction, in that cases that proceed by way of an indictment are more likely to receive a more serious sentence.
- Recidivism rates are directly impacted by the method employed to calculate them, in that broader definitions of recidivism, longer follow-up periods, and higher risk samples produce higher recidivism rates.
- Approximately 29% of CSE offenders were reconvicted of a new offence in adult court within five years of their original child sexual exploitation conviction.
- Only 3% of CSE offenders were convicted of a new child sexual exploitation offence within five years of their original CSE conviction.
- There is no evidence to suggest that custodial sentences are related to decreases in recidivism.
- Bonta, J., T. Rugge, and M. Dauvergne. 2003. The recidivism of federal offenders . Ottawa: Solicitor General Canada.
- Gendreau, P., C. Goggin, and F. Cullen. 1999. The effects of prison sentences on recidivism . Ottawa: Solicitor General Canada.
- Hanson, R. K., H. Scott, and R. Steffy. 1995. A comparison of child molesters and non-sexual criminals: Risk predictors and long-terms recidivism. Journal of Research in Crime and Delinquency 32:325-337.
- Harris, A., and R. K. Hanson. 2004. Sex offenders recidivism: A simple question . Ottawa: Solicitor General Canada.
- Langevin, R., S. Curnoe, P. Fedoroff, R. Bennett, M. Langevin, C. Peever, R. Pettica, and S. Sandhu. 2004. Lifetime sex offender recidivism: A 25-year follow-up study. Canadian Journal of Criminology and Criminal Justice 46:531-552.
- Latimer, J. 1998. The consequences of child maltreatment: A reference guide for health practitioners . Ottawa: Health Canada.
- Latimer, J., and C. Dowden. 2005. Sentencing in youth court: The effect of custody on recidivism. Manuscript submitted for publication.
- Smith, P., C. Goggin, and P. Gendreau. 2002. The effects of prison sentences and intermediate sanctions on recidivism: General effects and individual differences. Ottawa: Solicitor General Canada.
Dr. Ab Currie, Principal Researcher, Research and Statistics Division
This paper draws on data from a larger study of problems with legal aspects experienced by low and moderate income Canadians. The main objectives of the study were to determine the incidence of law-related problems among this segment of the population and the social and demographic groups that are most vulnerable. The present paper reports some preliminary results related to the occurrence of problems among vulnerable groups and, in particular, respondents' feelings of unfairness about the outcomes of problems.
A recent survey conducted for the Department of Justice Canada by the Environics Research Group examines the problems in civil law matters experienced by low and moderate income Canadians. The survey used a national sample of 4,501 respondents, who were interviewed by telephone in March 2004. The margin of error for a sample of this size is +/- 1.5 per cent 19 times out of 20. The survey was limited to low to moderate income Canadians. Respondents were included in the survey if they were 18 years of age or older and had incomes at or below $35,000 for individuals and below $50,000 for families.
The problem identification part of the questionnaire contained 15 problem categories: consumer, employment, money and debt, income assistance, disability pensions, housing, immigration, discrimination, treatment by the police, threat of legal action, family problems related to divorce or separation and children, other family-related problems, wills and powers of attorney, personal injury and hospitalization. Data were collected on 76 specific problems within the 15 problem categories. 
Respondents were asked to indicate if, within the past three years, they had experienced any of the specific problems included on a list read to them by the interviewers. Respondents were asked to include only those problems they considered difficult to resolve. The respondents were then asked about their seeking assistance with problems. For this part, in order to reduce the interview time to an acceptable level, up to three specifically identified problems were chosen at random for each respondent.
It is important to note that the respondents were not asked to identify "legal" problems. This was because it cannot be assumed that people will recognize in all cases that their problems have a legal aspect and a legal solution. In a statement that has now become the orthodoxy of the legal needs literature, Philip Lewis observed that saying a person has a legal problem is more a statement of one option for resolving the problem than about the nature of that problem.
"A tenant with a leaking roof may be regarded as having a legal problem. However, he may choose to get a ladder and not a lawyer" (Lewis 1973, 79). Rather, respondents were asked if they had experienced problems that were difficult to resolve, based on a pre-selected list of problems that have legal aspects and possible legal solutions. The pre-selected list of problems assured the existence of legal content. Respondents were neither asked to make judgments about the legal nature of their problems nor about any possible solution.
Almost 48 per cent (47.7 %) of the low to moderate income population in Canada experienced one or more law-related problems during the three-year reference period.  This is higher than the 34 to 37 per cent reported in the surveys in England and Wales . It is about the same as the results of the American research carried out about ten years ago and lower than the results of the research in the Netherlands.
Problems do not occur uniformly throughout the study population. A significant number of people experienced multiple problems. The table below shows the number of respondents reporting varying numbers of problems.
|Number of Problems
|Seven or More
About half of respondents reported they had no difficult problems during the study period. Among those reporting problems, one-fifth experienced only one problem. The percentage of respondents experiencing larger numbers of problems diminished with the number of problems reported.
Certain subgroups within the sample showed a fairly strong tendency to report at least some problems they considered difficult to resolve. This section summarizes the groups within the sample that were most likely to report experiencing no problems compared with one or more problems. Odds ratios are used to indicate the likelihood that respondents in certain groups are more likely to experience problems than others.
Respondents in the 29 to 45 age group were 2.0 times more likely than all others to report problems (p=.0001).
Generally speaking, the lower the educational level the less likely respondents were to report problems. People with less than high school education were 0.5 times, or half, as likely as all other educational groups to report experiencing one or more problems (p=.0001). This compares with respondents with some post-secondary education who were 1.5 times more likely to report problems (p=.0001) and respondents with a university degree who were 1.4 (p=.0001) times more likely than others to report problems. The lesser tendency for the lowest educational group may be less a reflection of fewer problems than less of a tendency to report them.
Single parents were 2.3 times more likely than all others to report problems (p=.0001) compared with singles who were only 0.8 times (p=.0001) and couples who were 0.6 times (p=.0001) as likely to report experiencing problems.
The unemployed were 2.4 times more likely than others to report at least one problem (p=.0001).
Respondents whose major source of income was a disability pension were 2.8 times more likely than all others to report experiencing problems (p =.0001).
Respondents receiving social assistance payments were 2.1 times more likely to report one or more problems (p=.0001).
Aboriginal people and members of visible minority groups were slightly more likely to report at least one problem. Respondents self-reporting as visible minorities were 1.6 times more likely than others to report at least one problem (p=.0001) and Aboriginal people were 1.4 times more likely (p=.0001).
A logistic regression procedure was used to examine which variables have the strongest predictive value for experiencing problems. Being young, a single parent, self-reporting as a member of a visible minority group, and receiving social assistance are the four best predictors of the likelihood of experiencing one or more problems.
|Best Predictor Variables
|Receiving Social Assistance
R-squared = .18, Likelihood Ratio 535.9, p = .0001
The Outcomes of Problems
The study asked respondents if the problems they had experienced during the three-year reference period had been resolved. If the problem had not been resolved, they were asked if the situation had become worse. If the problem had been resolved, they were asked if the outcome was perceived as fair. Overall, respondents indicated that 33.9 per cent of the time, the problem remained unresolved. Among unresolved problems, respondents indicated that the situation had become worse in 46.1 per cent of problems. Among resolved problems respondents reported that the outcome was unfair in 29.5 per cent of the time.
Perceptions of Fairness
The level of perceived unfairness is an important issue. Figure 1 shows the perceived level of unfairness of outcomes for specific problem types. According to Rawls (1999), the idea of fairness is fundamental to the concept of justice people have. Other authors have drawn a connection between the sense of fairness or justice and social cohesion. For example, Breton et al. (2004, 33) write:
The sense of being treated fairly, of being given a fair chance, does much to determine the degree of attachment to the institutions, the communities, and the society in which people live their lives. Fair treatment nourishes loyalty to the society and makes people more willing to contribute to its functioning. In contrast, unfairness is socially destructive.
Only a small percentage of respondents sought or received legal or other assistance with their problems. Nonetheless, it is possible that the levels of perceived unfairness reflect a sense of injustice. When asked about fairness in more general terms, a national sample of Canadians indicated that 18 per cent felt that Canadian society was unfair, and 15 per cent reported personal feelings of unfairness (Breton et al. 2004). The percentage of respondents in the present study reporting a sense that the outcomes of their law-related problems were unfair were all higher than that, and in some cases, considerably higher. The figures reported in the present survey reflect the general and widespread sense of unfairness in the justice system reported in public opinion research. A survey conducted in 1992 by the Angus Reid Group (1992) found that 64 per cent of Canadians disagreed or disagreed strongly with the statement that "everyone, no matter who they are, is treated the same by the justice system in Canada ." This directs our attention to the linkages between the provision of legal services to the poor, respect for the rule of law and confidence in the justice system by the public , and broader public policy issues relating to fairness, trust, and the maintenance of civil society.
Unfair Outcomes and Vulnerable Groups
Only three subgroups were more likely than others to perceive outcomes of problems that had been resolved as unfair; the unemployed, visible minorities, and foreign-born respondents.
- Visible minorities were 1.4 times more likely than all other respondents to perceive outcomes as unfair (p=.004).
- Foreign-born respondents were also 1.4 times more likely to perceive that the outcome of a resolved problem was unfair (p=.03).
- Respondents who were unemployed were 1.5 times more likely to report unfair outcomes (p=.001).
CONCLUSION AND DISCUSSION
The law is pervasive in modern bureaucratic societies, and in Canada , as is similar in other countries, problems with legal aspects are ubiquitous. For example, research has shown that about 34 percent (Genn 1999) to 37 percent (Pleasence et al. 2004) of the population of England and Wales has experienced one or more problems having legal implications that were difficult to resolve. A similar study carried out in Scotland estimates that about 24 percent of the population of that part of the U.K. had experienced one or more 'justiciable' problems (Genn and Paterson 2001). A study carried out in the U.S. estimates that 47 percent of low income Americans and 52 percent of moderate income Americans experienced at least one law-related problem over a three-year period (American Bar Association 1994). A more recent national study carried out in the Netherlands found that 67 percent of the sample experienced one or more 'justiciable' problems (Van Velthoven and Ter Voert 2004). Research on legal needs carried out in New Zealand in 1999 estimates that 51 % of the population experienced one or more problems within a three-year period (Maxwell et al. 1999). A study conducted in the Province of Ontario in 1987 found that about 34 % of the sample had experienced serious problems over a three-year period (Bogart and Vidman 1990).
The extent of problems with legal aspects in Canadian society makes this topic a matter for serious concern. The occurrence of problems is not spread equally among the population. The research shows that certain vulnerable subgroups within the population have a greater likelihood than others of experiencing law-related problems.
Problems tend to get resolved, most of them within a year or two of first occurrence. However, in a significant proportion of cases, respondents perceive the outcomes as being unfair. This finding has important implications for perceptions about the "justness" (i.e., fairness) of Canadian society by its citizens. Further, given the fundamental importance of justice as a dimension of all other social institutions, the perceptions of unfairness may weaken the fabric of social cohesion in Canadian society.
- American Bar Association. 1994. Legal needs and civil justice: A survey of Americans. Washington, DC: American Bar Association.
- Angus Reid Group. 1992. Canada and the world: An international perspective on Canada and Canadians. Winnipeg: Angus Reid Group.
- Bogart, W. A., and N. Vidmar. 1990. Problems and experience with the Ontario civil justice system. In Access to Civil Justice , ed. A. Hutchinson. Toronto: Carswell.
- Breton, R., N. J. Hartmann, J. L. Lennards, and P. Reed. 2004. A fragile social fabric? Fairness, trust and commitment in Canada. Kingston: McGill-Queens University Press.
- Gabrielle, M., M. Maxwell, C. Smith, P. J. Shepherd, and A. Morris. 1999. Meeting Legal Service Needs. Wellington, NZ: New Zealand Legal Services Board.
- Genn, H. 1999. Paths to Justice: What people do and think about going to law . London : Hart Publishing.
- Genn, H., and A. Paterson. 2001. Paths to Justice Scotland: What people do and think about going to law. London : Hart Publishing.
- Lewis, P. 1973. Unmet legal needs. In Social Needs and Legal Action, ed. P. Morris, R. White, and P. Lewis. London : Martin Robertson.
- Rawls, J. 1999. A theory of justice . 2nd ed. Oxford: Oxford University Press.
- Pleasence, P., A. Buck, N. Balmer, A. O'Grady, H. Genn, and M. Smith. 2004. Causes of action: Civil law and social justice. London : Legal Services Commission.
- Van Velthoven, B. C. J, and M. Ter Voert. 2004. Paths to Justice in the Netherlands. Paper presented at the 5th International LRSC Conference, Cambridge.
Report a problem on this page
- Date modified: