Bruce, D. (1998). Towards a Strategy for Prevention: The occurrence of deaths in custody or as a result of police action in Gauteng, April - December 1997. This report was produced at the request of The Independent Complaints Directorate, 2 July.

 

by David Bruce

This report was produced, following a request by the Independent Complaints Directorate (ICD) in January 1998 for NGOs to assist the ICD in developing recommendations on the prevention of deaths in police custody or as a result of police action. This report is an edited version of the report which was subsequently produced and which was first released at a press conference at the ICD office in Pretoria on 2 July 1998.

Special Acknowledgement needs to be made to Ross Atkins, then a researcher at the Centre for Policy Studies, and Ian Liebenberg, of the Human Sciences Research Council, for their assistance in conducting the analysis of ICD dockets on which this report is based.

This email address is being protected from spambots. You need JavaScript enabled to view it. is a Senior Researcher in the Criminal Justice Programme at the Centre for the Study of Violence and Reconciliation.

Introduction

In January 1998 the Independent Complaint's Directorate (ICD) requested a group of researchers to come up with suggestions regarding the type of recommendations which the ICD should make to the SAPS regarding measures which should be implemented in order to prevent deaths in police custody or as a result of police action (henceforward referred to as police custody/action deaths).

The researchers indicated that in order to make such suggestions they would need to obtain a better understanding of the types of circumstances in which police custody/action deaths where occurring. As an initial step it was agreed that:

  • A group of the researchers would go through the ICD files dealing with police custody/action deaths which occurred in Gauteng during 1997.
  • The researchers who had access to the files would be bound by a confidentiality agreement. In particular the concern is that the research team and the report should not identify any persons who are involved in particular incidents or the police stations or units to which they belong. No details regarding the names of police officers or of stations or units linked to the deaths were therefore recorded. The confidentiality agreement was entered into in order to ensure that the ICD remained within the provisions of sub-section 53(5) of the SAPS Act, No. 68 of 1995.
  • The group of researchers would then make suggestions as to the types of issues which would need to be covered in such recommendations.
  • A report on the work would be submitted with a view to further research being conducted regarding the specific recommendations which should be made.

The following report is submitted in terms of this agreement. The report therefore provides a profile of the circumstances of occurrence of police custody/action deaths in Gauteng in the April to December 1997 period. This was the primary purpose of the research as it was initially conceptualised. The intention was to identify key areas where there may be scope for improvement in police practices in such a way that police custody/action deaths may be prevented and the present high rate of these deaths reduced.

However the report goes beyond simply looking at the circumstances of occurrence of police custody/action deaths in Gauteng and attempts to provide the framework for a strategy which would aim to assist in the prevention of these deaths, with the intention that the rate of these deaths be reduced dramatically in the medium term. The report is therefore intended to stimulate discussion around the prevention of police custody/action deaths as well as providing direction to further research.

Overview of Research

The research on the circumstances of occurrence of police custody/action deaths involved examining virtually all ICD files on police custody/action deaths which occurred in Gauteng during 1997. These files primarily cover deaths which occurred in the 9 month period from April to December 1997 though a few of the cases may have occurred in the January - March period. A standard form was used to record data from each of the files. The information on the occurrence of police custody/action deaths in this report is derived from the data collected on the standard forms.

(a) Once the data forms had been filled in the cases were divided into 8 categories as a basis for analysis. The categories were selected on the grounds that they might be useful in analysing the issue of possible preventive measures. They do not correspond exactly to the categories used by the ICD. There may also be some inconsistencies in the way that deaths have been classified. There would appear to be a number of inherent complexities associated with the classification of police custody/action deaths and with any set of categories which is used. The categories used are reflected in the left hand column of Table 1. The categories are explained and examined in detail in section 6.

(b) The report covers 158 incidents, 160 files and 168 deaths.

  • The discrepancy between these numbers relates to the fact that some incidents are covered in 2 files (2 cases are duplicated in this way), while in a number of cases more than one person was killed in the incident.
  • In 5 cases (6 deaths) the number of the file (case no) was not recorded. In 2 of these cases (covering 3 deaths) it is reasonably clear that the incidents have not been duplicated in the report. The 3 other cases (where it is less clear whether the file has been duplicated or not) occur in section D (2) and E(1). Unless there are other errors it would appear that the report therefore covers 165 to 168 deaths in 155 to 158 incidents.
  • The figures for persons killed do not include police officers killed. In one of the incidents in Section D and one of the incidents in Section F, a police officer was also killed. In the latter case the policeman killed himself after killing his wife.

The distribution of cases in terms of categories is reflected in Table 1.

Table 1: Overview of cases analysed.
Category
No. of deaths
% of total deaths
No. of Incidents
No. of Files
Possible error
A. Suicide in custody
18
10.7
17
17
 
B. Injuries prior to custody
15
8.9
15
15
 
C. Other deaths in custody
21
12.4
21
21
 
D. Death of person deemed to pose an immediate threat
59
34.9
54
54
2 (deaths, incidents, files)
E. Other shootings during the course of arrest
34
20.1
30
32
1 (death, incident, file)
F. Other shootings or use of force
15
8.9
15
15
 
G. Vehicle deaths
3
1.8
2
2
 
H. Cases excluded
4
2.4
4
4
 
TOTAL
169
100%
158
160
3 (deaths, incidents, files)

(c) It must be noted that the approach taken was, to some extent, to take the information provided at face value. Where there was a conflict of information this was taken into account though one version of events was not necessarily preferred over another unless it was clearly more convincing. In some cases, where deaths occurred in particularly suspicious circumstances, it also appeared reasonable not to take the information provided at face value. All together there appeared to be few cases where it appeared that there was a strong probability that the police had deliberately acted unlawfully. Often however the police were the only witnesses to the events and. In general autopsy and, particularly, forensic reports had not yet been completed. Table 2 indicates that in 55% of incidents statements by witnesses who were not members of the SAPS were included in the ICD files. However if one excludes the cases in relation to which this data was not recorded then 87 out of 146 cases (60%) contain statements by non-police witnesses.

However in only 41 out of the 81 cases (50%) in categories D and E in relation to which this data was recorded were statements by non-police witnesses included. The fact that in the 50% of these cases no non-police statements were provided may partly be related to the fact that, frequently in the case of deaths connected to the police there may be no non-police witnesses present, or those who were present - perhaps associates of an alleged offender - are reluctant to come forward. It should also be noted that frequently, the non-police statements had been collected by members of the SAPS. Furthermore frequently the evidence provided in the statement was not directly about the incident in which the death occurred but rather about the surrounding circumstances. For example a file might include the statements of witnesses who were in the shop where the robbery occurred but no statements from non-police witnesses who actually saw the police shoot the alleged felons outside the shop. In the end therefore this report primarily reflects the accounts which have been provided by members of the SAPS as to the circumstances in which deaths occurred.

(d) The focus of the report is on preventive measures rather than on investigative issues. The report does not evaluate issues relating to investigations or the possible improvement thereof though it would appear that rigorous investigations into police custody/action deaths, accompanied by prosecution or disciplinary action in appropriate circumstances, might in themselves contribute to discouraging unlawful or non-procedural behaviour by the police.

(e) There were some problems with the consistency of data recorded. These problems reflect the quality of information contained in some of the files which were scrutinised as well as inconsistencies in the information recorded by the different researchers. To take a random example the fact, mentioned in section 6, under category (c) that in 4 cases the place of death is not recorded may reflect the fact that this information was not in the ICD docket, or not recorded in the data sheet. The inadequacies of the data are also related to the fact that a single generic data sheet was used to record information relating to deaths which occurred in a variety of different circumstances. Thus, for instance, it would appear preferable for specific data sheets to be developed which are better orientated to the specific categories rather than a single data sheet for all police custody/action deaths. For instance data which is of specific relevance in relation to deaths which occur in custody would often be different from the data which is useful on deaths as a result of police action which occur outside of custody. In the latter category for instance issues such as whether or not there was a direct threat to the police, whether they were on or off duty, type of weapon used by deceased, number of suspects involved in incident and the offence of the accused and the location of the shooting (specifically whether it was at the scene of a crime or not), are more likely to be pertinent than in the former category.

Table 2: Files where statements by non-police witnesses were included or not included
Category
Incidents
Files with no statements by non-police witnesses
Data omitted ( data sheets don't indicate whether non-police statements included)
Files with statements by non-police witnesses
% of files per category with statements by non-police witnesses
A. Suicide in custody
17
7
1
9
53
B. Injuries prior to custody
15
2
3
10
67
C. Other deaths in custody
21
4
3
14
67
D. Death of person deemed to pose an immediate threat
54
23
2
29
54
E. Other shootings during the course of arrest
30
17
1
12
40
F. Other shootings or use of force
15
3
1
11
73
G. Vehicle deaths
2
0
0
2
100
H. Cases excluded
4
3
1
0
0
TOTAL
158
59
12
87
55

(f) Nevertheless the study is useful as a starting point for developing a fuller understanding of the types of circumstances under which police custody action deaths tend to occur and consequently the types of preventive measures which may be worthwhile as well as the types of issues which need to be addressed in terms of ss53(2)(b) investigations. Furthermore useful lessons have been learnt about the types of questions and issues which need to be addressed in further research.

(g) In general the term "suspects" has been used - though the term here may refer to anyone from a person stopped by the police and searched in the street where there is no apparent evidence of an offence to people who are directly linked to an offence by having been present while it was taking place or fleeing from the scene.

(h) The information on race needs to be approached with particular caution. Frequently this was inferred from the surname of the person (deceased or police officer). Unless there was evidence to the contrary, Afrikaans surnames where taken to indicate a white person, while African surnames where taken to indicate a black person. In particular this might mean coloured police officers, who often might have Afrikaans sounding surnames, were recorded as "white".

(i) It was assumed that the police were on duty and that no police were injured unless it was specifically stated otherwise.

Note on Race and Gender of Deceased Persons and Police Involved in Shooting Incidents

As noted above (see 2(h)) the data on race needs to be treated with particular caution. Table 3 records the occurrence of people who were not "black males" amongst the 169 deaths. Table 4 records the race of police involved in shooting incidents in sections D, E and F. A female police officer was present at one of the 84 incidents in sections D and E. In section F only the race of the shooter is recorded while in sections D and E the race of the police present at the incident is recorded.

While these issues are not evaluated here information on race was regarded as relevant in relation to the possibility that racial attitudes may be a contributing factor in relation to the actions, such as the use of lethal force, undertaken by police officers. However this data is likely to be difficult to evaluate as it will also be affected by other factors in relation to which there may also be significant differences in the distribution of police officers by race. These might include the type of areas and units in which members of the police service are posted or tasks which they are allocated to as well as the overall composition of the police service.

It is apparent that the overwhelming majority of persons who died in police custody and particularly as a result of police action were black men. Factors which might be relevant to evaluating the data would be the distribution of persons by race in the overall population, as well as the distribution of persons of particular races arrested. In looking at the virtual 100% representation of black males amongst those alleged offenders shot by the police, it might be worthwhile to look at the racial profile of perpetrators of violent crime and crimes such as vehicle theft and housebreaking as reflected for instance in arrest and conviction statistics.

The data sheet also included a question about the nationality of the deceased persons. In general the files did not contain information about the nationality of the deceased persons and therefore it was not possible to generate any data in relation to this issue.

Table 3: Distribution of persons who were not classified as black males amongst the deceased
Category
Females
% of females in category
Whites
Coloureds
Asians
% of whites, coloureds and Asians out of total deaths in category
Race or gender not classified
A
   
3
2
1
33.3
1 (race)
B
         
0
1 (race)
C
3
14.3
4
1
 
23.8
 
D
       
1
1.7
6 (race)
3 (gender)
E
         
0
 
F
4
11.8
3
   
8.8
1 (race)
G
2
(children)
66.6
     
0
1 (gender)
H
         
0
1 (race and gender)
TOTAL
9
5.3% of
total
10
3
2
8.9%
10 (race)
5 (gender)

Type of Units Involved in Incidents

Table 5 reflects recorded information on type of unit linked to particular incidents. Of the 50 cases in section D and E where the unit was recorded, 32 (64%) involved station based units, 11 (22%) involved the Flying Squad or Highway Patrol and 8 (16%) involved other units. Units recorded as "other" units include murder and robbery (3 cases), tracing unit (2 cases), firearm unit, public order police, "taakmag", robbery reaction unit, police reservists, and in one case both members of a station based unit and a public order police unit. It may have been

Table 4: Race of police involved in incidents in sections D, E and F

Note that in this table references to "Black and White" may include cases where the Black police present were, Black, Coloured or Asian or a mixture of these groups.

Race
Section D
Section E
Section F
Total
Black
18
5
7
30
White
17
18
1
36
Coloured
1
   
1
Asian
1
   
1
Black and White
11
3
 
14
Unknown
6
4
7
17
TOTAL incidents
54
30
15
99
Table 5: Types of unit linked to incidents in each category
Category
Station Based
Flying Squad/
Highway Patrol
Other
Not recorded
A
8
   
9
B
5
   
10
C
8
 
2
11
D
23
6
3
22
E
9
5
4
12
F
4
 
2
9
G
   
1
1
TOTAL
57
11
12
74

more appropriate to record for instance, tracing and firearms units as station based. In a high proportion of the incidents here the information on type of unit was not recorded.

It may be useful to evaluate this information in more detail partly in relation to the possibility of targeting possible pilot interventions. In this regard it may also be worthwhile to look at the relationship between rank, and possibly age, and involvement in police custody/action deaths. Data on rank was collected but is not reflected in this draft report.

General Note on the Occurrence of Shootings and Other Use of Force in This Report

The general approach followed by the ICD is to classify shootings - probably with the exception of suicides - as "deaths as a result of police action". However shootings are only one form of the use of force by the police and theoretically all deaths which are the result of some use of force or violence could be included in one category. In this report however:

(a) Six alleged suicides in this report involved firearms - 3 in section A, one in section D, and two in section F.

(b) Most of the cases of "injuries prior to custody" (Section B) are cases of death as a result of the use of violence. In most cases the police claim to have arrived at the scene after the violence had occurred.

(c) Certain shootings have been classified in the "other deaths in custody" category where they involved a person who had already been placed in custody. Of these 3 involved the shooting of persons attempting to escape from custody (in 2 of these cases there may have been an immediate threat to life posed by the escapees), 1 involved the shooting of a person who allegedly grabbed a police firearm during a "pointing out" of hidden firearms. In 5 other cases it is suggested that an assault by police may have precipitated the death of a person already in custody.

(d) The 59 cases in section D all appear to have involved the use of firearms by police. In most cases it is to be presumed that this was the direct cause of death. However in 1 case the immediate cause of death was vehicle accident. Here the fact that the police fired on vehicle which was being driven in a particularly reckless manner may be considered to have been a precipitating factor of the 2 fatalities which occurred. In 1 other case the police allege that a person involved in a gunfight with police allegedly committed suicide with his own pistol. If this is true then, here also, the immediate cause of death was not the use of firearms by police.

(e) All 34 deaths in section E were apparently directly caused by the use of firearms by police.

(f) Fourteen of the 15 deaths in section F were caused by firearms. This number however includes one case where the wife of a police officer allegedly committed suicide and where, therefore, the death was not the direct result of the actions of a police officer. In 1 case a death was allegedly caused by the assault by members of the police service on an alleged robber.

Incidents of shooting or other use of force therefore occur in a number of places in this report. The police are also not directly responsible for the shootings or other use of force in all cases. Overall 110 (65%) of the 169 deaths recorded here may be attributed to the use of firearms by the police while a further 6 (3.5%) may be attributed to the other use of force by the police. The breakdown for these figures is provided in Table 6.

Table 6: Distribution of cases where probability is that death caused by shooting by police or other use of force by police
Category
Column 1:
Shooting by police
Column 2:
Other use of force by police
Total:
Column 1 + Column 2
Total deaths in this category
A, B, G
0 - some deaths in categories A and B are by shooting or other violence but this is not attributed to the police. An argument may also be made that driving a vehicle amounts to a use of force and that the deaths in category G should also be recorded as being linked to the use of force by the police.
40
C. Other deaths in custody
4
5
9
21
D. Death of person deemed to pose an immediate threat
59
0
59
59
E. Shootings during the course of other arrests
34
0
34
34
F. Other shootings or use of force
13
1
14
15
TOTAL
110
6
116
169

The Circumstances of Occurrence of Police Custody/Action Deaths

Category A: Suicide in custody (Total: 18 deaths)

(a) Of the 18 suicides recorded 13 occurred in police cells (including one double suicide) and 2 in court cells while in 3 cases a person who was in the process of being arrested shot himself.

(b) Of the 14 non-firearm cases 11 were suicides by hanging, 3 were overdoses and in 1 case the method is not recorded.

(c) Of the 11 suicides by hanging 3 used (torn) blankets, 3 used belts (in 1 case supplemented by his overall top), and in 3 cases the means used was not recorded. In the two other cases the means used were "shoe laces" and "shirt sleeves".

  • According to the information recorded in the dockets the time period which had elapsed between the last cell inspection and the discovery of the suicide was: 30 minutes (2 cases), 35 minutes (2 cases including the double suicide), 40 minutes (1 case) and 60 minutes (1 case). In the remaining 6 cases of suicide in police cells this information was not recorded.

(d) The offence for which the persons who committed suicide had been arrested were: domestic violence, vehicle theft and murder (2 cases each), fraud, armed robbery and attempted murder (1 case each). In 9 cases (including the double suicide) this information was not recorded.

(e) All of the 18 were male. Eleven were black, 3 were white, 2 were coloured, and one was Indian while in one case the information is not clear.

(f) In future it might be useful also to collect data on whether or not there was evidence of behaviour sometimes associated with suicide such as suicide notes or suicide threats.

Category B: Injuries prior to custody (Total: 15 deaths)

(a) The 15 cases all involve situations where a person believed to have been responsible for an offence or about to commit an offence died of injuries sustained prior to the involvement of the police in the incident.

(b) In one of the cases - a shooting at a small-holding - the person died prior to the arrival of the police. Here 1½ hours elapsed from the case being reported to the arrival of the police. While the case strictly does not qualify as a police custody/action death the response time of the police may potentially be regarded as a factor which needs to be taken into account in looking at deaths in this category.

(c) Of the other 14 cases 8 incurred their injuries during assaults by members of the public, 2 were shot by members of the public during alleged hijackings, and 1 was shot by a security guard while in 3 cases the cause of death is unknown. The 8 assaults by members of the public included 6 cases where a person was assaulted by a group of people (sometimes described as "public" though one case refers to "angry crowd" and one to "taxi drivers"). In 4 of the cases the information recorded refers to the victim as having been connected to an offence - 3 cases of robbery and one of "throwing stones". One of the other two cases in this group refers to the person assaulted by a "kangaroo court" while the last one is referred to as a shebeen incident.

(d) For the 14 cases the location of death is recorded in 9. Five (5) of these were in hospital, 3 at a police station, and one in a police van.

(e) In the 4 cases of deaths at a police station or in a police van one had received medical treatment and in one other case it appears that medical assistance had been called by the police but had not yet arrived at the time of death.

Category C: Other deaths in custody (Total: 21 deaths)

(a) Of the 21 deaths, 6 occurred at a police station, 4 occurred at the scene of the arrest, 2 occurred in a police van, 2 occurred in hospital, 1 at court and 1 where a suspect was accompanying the police "on investigation". In 4 cases the place of death is not known.

(b) Cause of death is attributed to natural causes in 6 cases ("sick", hypothermia, unstable angina, "collapsed", headache, and urine disorder), alcohol in 3 cases, and the (possible) use of force by the police in 9 cases (one of which is the same as one of the alcohol related cases). In 1 case the person burnt to death in a cell fire. In 3 cases the cause of death is not indicated.

(c) The 9 cases attributed here to a possible use of force by the police include: 3 cases of persons being shot during attempted escapes (one while being booked in to custody at a police station, one at court and one just after he had been handed over to the police by community members), 1 case of a person who allegedly "fell and hit his head on a table" while being interrogated, 4 cases where evidence or statement to the effect that assault is suspected has been made ( one of which may also have involved traffic police and one which occurred during a "pointing out" by a suspect who allegedly grabbed a police firearm), and 1 case where a person had been arrested at the scene of a crime, handcuffed and left on his own (a body had just been discovered), following which he purportedly attempted to flee and died from a fall.

(d) In 8 of the 21 cases there was at least some indication in the information recorded that the injury or illness had been acknowledged by the police and medical attention sought while in 9 cases it appears that the police failed to recognise or acknowledge that the person was ill or, if they did recognise this and call for medical assistance, this is not recorded. In 4 cases this was not applicable as the person was apparently killed instantly.

(e) The 4 deaths which occurred at the scene of the arrest, 3 which occurred during attempted escapes and 1 which occurred during investigation, could technically be regarded as "police action" rather than "police custody" deaths. This would depend partly on a factual/legal analysis of whether or not the deceased person was in police custody at the time when the fatal injuries occurred as well as on resolving a question relating to whether it is the point at which the injury occurs or the point at which the death occurs which determines the classification as "custody" or "action" death. (If it is the point of fatal injury category B would fall away, if it is the point at which death occurs many of the shooting deaths would end up being classified as deaths in police custody). The approach taken here is that deaths which occur when a person has already been placed under the control of the police, even if they are the result of police action (such as shooting) are regarded as deaths in custody.

(f) In particular issues relating to persons who are killed while escaping from custody or at the scene of arrest subsequent to having been arrested are likely to overlap with issues relating to the use of force by the police discussed in the following sections.

Category D: Death of person deemed to pose an immediate threat (Total: 59 deaths)

(a) Incidents which have been included in this section include all those where, according to the accounts provided, the police involved had some grounds for believing that there was an immediate threat of death or serious injury to themselves or to other persons. This would include situations where a person was involved in direct aggression (such as shooting) against the police as well as other situations where the police might have had justification in believing that the person posed a threat to them such as situation where they claim that they "saw something that looked like a gun" in the hands of a person that they were pursuing, or the person had just been involved in an offence involving serious violence.

(b) While most of the cases here are situation where the police were ostensibly attempting to affect arrests this was not necessarily true in all cases. Thus, for instance, in a situation where a police officer claims that s/he was held up by people whom he believed wanted to rob and possibly kill him, the immediate purpose behind the shooting is "self defence" and not "affecting an arrest". In other cases the purported need for the officer to defend him or herself might have arisen following an attempt to arrest a person.

(c) In terms of the circumstances surrounding the 54 incidents during which the 59 deaths occurred these may be divided into: 37 incidents where there was direct aggression or a direct threat (e.g. the pointing of a firearm) against the police or another person, and 17 incidents which have been placed in this category on the basis that there were other reasons for perceiving a threat.

  • The 37 incidents where the suspect was involved in direct aggression or posed a direct threat to the police or another person included: 27 cases where this is the only fact recorded, 1 case where the suspect was killed while fleeing immediately after directly pointing his firearm at the police, 2 cases where the direct threat or aggression occurred at the scene of the crime, 2 cases where the direct threat or aggression occurred during or immediately following a vehicle chase, 1 case where the classification is merely derived from information that the person was armed and shot in the chest and face. In only 3 cases was the direct threat or aggression not directed against the police but against another person.

  • The 17 incidents where the information recorded indicates that the police perceived an immediate threat but not a direct threat include: 6 cases of a fleeing armed person, 4 cases where the person was armed and near the scene or fleeing from the scene of a violent crime, 1 case where the person was believed to be armed and fleeing from a vehicle following a car chase, and 6 cases which have been placed in this category for other reasons, such as that the person was suspected of a violent crime or was armed but that other details of the circumstances in which the death occurred have not been recorded.

(d) The weapons allegedly used or in the possession of the deceased person were recorded as follows in the 54 incidents: firearms/guns in 25 incidents, rifles/AK47s in 3 incidents, knives in 3 incidents. In 9 incidents the information is merely that the person was armed. There was 1 incident each of a person who was armed with a screw driver, stick, or rock. In one incident the suspects were armed with a rifle and knife. In one incident the immediate threat was posed by a person driving a vehicle dangerously. In 4 incidents no specific weapon is recorded. In 4 incidents it is stated variously that there "was an object in the persons hand believed to be a gun", or that the person was "believed to be armed" or that no weapon was found but that the police "said the suspect was armed". In one case the person had no weapon but allegedly leapt at the police. (Note that a question which might be worthwhile in further questionnaires would be "when was the fact that the person was armed discovered" - in some cases this might have been only after the persons death or fatal injury).

(e) In 11 of the 54 cases it is indicated that some form of warning was used. In 5 cases this was said to have been a verbal warning and in 5 cases these were said to have been warning shots while in one case "warning lights" were said to have been used during a car chase.

(f) In 11 of the incidents a member of the police service or public was also injured or killed. In 6 incidents one or more members of the police service were injured (in one incident it was three members) and in 1 incident a member of the police service was killed. In 3 incidents civilians were injured (one of these overlaps with an incident where a police officer was injured) and in 2 incidents civilians were killed. In at least one of the incidents where a civilian was injured the injury was caused by the police (a warning shot). In both incidents where a civilian was killed, the actions of the police may be regarded as having been a contributory cause of death.

(g) The breakdown of the alleged offences of persons who were deemed to pose an immediate threat is given in Table 7.

Table 7: Alleged offences committed by persons who were deemed to pose an immediate threat and who died as a result of police action.
Alleged offence
No (incidents)
%
Robbery (includes bank robbery, armed robbbery, attempted armed robbery, hijacking, heist
22
40.7
Murder and attempted murder
4
7.4
Combination (rape, murder)
1
1.9
Vehicle theft
4
7.4
Housebreaking
2
3.7
Stock theft
1
1.9
Unlicensed firearm
2
3.7
Reckless driving
1
1.9
Assault
1
1.9
Other
3
5.6
Not known
11
20.4
Mental patient
2
3.7
TOTAL
54
100.2

(h) In relation to the police-suspect ratio, in 17 of the 54 incidents there appear to have been more police than suspects present, in 7 the number of police and suspects was equal, and in 14 the number of suspects was greater. In 14 cases it is not clear what the ratio was while in at least 2 it would appear that this information is arguably not relevant.

(i) In at least 6 of the cases the police involved were off-duty while in one case it was both on and off-duty police who were involved.

Category E: Other shootings during the course of arrests (Total: 34 deaths)

(a) Incidents which have been included in this section include all deaths as a result of shooting by the police during arrests where there is no apparent immediate threat to the police or other person.

(b) In 13 of the cases the person who was killed has merely been recorded as "fleeing". In one of these cases this is indicated as having been after a struggle with the police. In a further 5 cases the person is recorded as having been fleeing from a vehicle (in 4 of these cases this is indicated to have been immediately following a vehicle chase), and in 5 cases it appears that the people were shot in the vehicle during or immediately following a vehicle chase. In 2 of the latter cases the identity of the deceased person is referred to indicating in one case that the person was a "passenger" and in the other an "accomplice". In 1 case the police claim to have shot the person during a robbery but this is disputed. In 6 cases no details regarding the circumstances of the shooting have been recorded.

(c) In 4 of the cases here there is some indication that the deceased person was armed or possessed some item that could have been used as an offensive weapon. In one case the police involved alleged that the deceased had been carrying a knife though this was disputed by the investigating officer. In one case the person was in possession of housebreaking implements and in one case the police reservist involved claimed that he "saw something" in the hand of the deceased but no weapon was found. In one case it is merely indicated that the person was armed.

(d) In 20 of the 30 cases it is indicated that some form of warning was given. In 7 cases this was said to have been a verbal warning and in 7 cases both a verbal warning and warning shots (in one case the suspect was killed by one of the warning shots). In 5 cases warning shots alone were fired, in at least one of these cases during a car chase. In 1 case "warning lights" and warning shots were used.

(e) In 5 of the cases another person was injured or killed. In 3 cases another suspect or person associated with the suspect (a passenger in a stolen vehicle) was injured. In 1 case a civilian was killed by the police (this death is recorded as one of the 34 deaths in this section) and in 1 case a civilian was injured by the police.

(f) The offence which the deceased persons was allegedly associated with is recorded in Table 8.

(g) In relation to the police-suspect ratio, in 8 of the 30 incidents there appear to have been more police than suspects present, in 8 the number of police and suspects was equal, and in 6 the number of suspects was greater. In 8 cases it is not clear what the ratio was.

(h) In 2 of the incidents the police officer involved was off duty while in one case the incident involved a police reservist using his private pistol.

Table 8: Alleged offence of person killed in other shootings during the course of arrests.
Alleged offence
No (incidents)
%
Robbery (including attempted hijacking)
4
13.3
Kidnapping
1
3.3
Vehicle theft
10
33.3
Housebreaking (including HB business)
5
16.6
Theft or attempted (including of: car stereo, clothes and gear boxes)
4
13.3
Not known
6
20
TOTAL
30
98.8
Category F: Other shootings or use of force (Total: 15 deaths)

(a) As noted in Section 3 above (see Table 4), while in general the classification of police involved in shooting incidents has indicated the race of those police who were present at the incident, in relation to those in this section it refers only to the race of the police directly involved in shooting or other use of force.

(b) In at least 2 of the incidents it appears that alcohol may have played some role.

(c) In 5 of the incidents the police officer was a constable, while in 4 he was a sergeant. In 6 incidents the rank was not recorded.

(d) Details of the circumstances in which these deaths occurred are recorded in Table 9.

Category G: Vehicle deaths (Total: 3 deaths)

One case which could have been included here has been included in category D.

Table 9: Circumstances of deaths by other shootings or use of force.
Police on or off duty
Action during which shooting occurred formed part of police duties
Description
Off-duty (?)
No
Shot person in the head while drunk
and showing off his pistol
Off-duty
No
Murder in shebeen using service pistol
Not clear
No
Accused of neglect of duty by member
of public whom he shot
Off duty
No
Sgt who had booked out rifle was allegedly seen in
a vehicle used in the perpetration of taxi killings
Not clear
No
Intentional family killing using rifle
On
Yes
Sub-machine gun went off in a charge office
Off (?)
No
Off-duty policeman kills wife and
then himself using private firearm
On
No
On duty police officer shoots landlady in apparent argument - not clear if deliberate or accidental
-
-
"Police shooting"
On
Yes
A shooting occurred during resettlement of a small community. Police officer asked for reinforcement.
A shot went off and wounded one person.
On
Yes
Shooting following scuffle after attempt at body
search after person seen smoking dagga
-
-
Recorded as "possible class one intentional shooting"
NA
NA
Suicide by policeman's wife. If, for instance, she used his pistol this might be grounds for holding him to be negligent. However this is not clear.
Off
Unclear
Allegations of the involvement of off-duty members
in the beating to death of a robber
Off duty
No
Shot girlfriend's mother after she pushed him gently
-
-
Possible intentional shooting
Category H: Cases excluded (Total: 4 deaths)

Cases were excluded either because the recorded information was not usable or because the information provided indicated that the death concerned could not be regarded as a police custody/action death. (Cases where there was no indication that a death had occurred where excluded altogether from the study).

(a) One case was excluded because the data sheet provided no useful information.

(b) Three cases were excluded on the grounds that they were not police custody/action deaths. One involved a suicide in prison, and in two cases a person was killed but there is no indication that the police were involved, or even that they arrived at the scene prior to the person's death. One case of this kind has been included in section B on the basis that there was a substantial delay between the reporting of the case and the arrival of the police. Other cases, particularly some of those in section F, such as the suicide of a policeman's wife, could also arguably fall in this category.

Unlawful Actions by the Police as a Cause of Death?

In relation to all of the eight categories listed here the potential exists that some of the deaths which "at face value" are not connected to unlawful actions by the police might in fact be connected to such unlawful actions. Thus for instance it is possible to speculate that:

  • People whose death is attributed to suicide (category A) or other causes (category C) might have been killed by the police, or died as a result of police negligence possibly where the police where aware that they could prevent a person's death;

  • Virtually any of the deaths recorded in category B, might actually have been the result of an assault by the police, or the police might have stood by while an alleged perpetrator was assaulted, or the police might have handed the person to a crowd of people for them to "dispense justice" after having already arrested the person;

  • Particularly in relation to the deaths - generally as a result of shootings - in categories D, E and F, the police might for instance have known that they did not have legal justification for their actions. Similarly in relation to any of the vehicle deaths (category G), the police might have been driving recklessly but claimed that this was not the case.

As indicated in section 2 of this report this analysis of police custody/action deaths is effectively largely based on the accounts of events provided by members of the SAPS. It can be fairly certain that, to some extent therefore, the picture which is presented here presents the actions of the police in a relatively positive light. For the police the versions of events which are presented here might in a number of cases serve the purpose of protecting them against possible incrimination in any criminal, or disciplinary, offence.

At the same time the degree to which the versions presented here represents a distortion of the true picture is not at all clear. It may be the case that at least some of the deaths recorded in this report may be the result of unlawful actions by the police even though this was not self-evident in terms of the material made available to the researchers. The following factors might be seen as motivating for this possibility to be taken into account:

  • The legacy of policing under apartheid which may be said to have been characterised inter alia by an excessive reliance on the use of force. The then South African Police, were also inculcated with values and attitudes essentially resistant to and hostile to change. Members of the present SAPS, who on the level of practice, or in terms of their values and attitudes, continues to identify with the "old style" of policing may be disposed to continuing to operate in terms of this "style".

  • Current evidence of high levels of vigilantism ("informal justice") in South Africa (reflected in part in the large number of deaths in Category B) would appear to be a manifestation of public disillusionment with and a lack of confidence in the Criminal Justice System. The police, integrally concerned with the enforcement of the law, while also part of the Criminal Justice System, may also be prone to such disillusionment.

  • The evidence provided in this report which provides reason to question the assumption that a primary cause of the high number of deaths as a result of police action is that police officers are forced to act in self defence while carrying out their duties.1 Fifty-nine of the deaths recorded in this report (those in section D) are of people who ostensibly presented some kind of immediate threat to the lives of a police officer or other person, for instance on the grounds that they were brandishing a firearm. However, altogether in the 54 incidents recorded in section D members of the police service were killed or injured in 7 incidents (13%, 1 death and 6 incidents of injury)). Altogether this means that, members of the police service were killed or injured in 6% (7 out of 116) of the incidents where a person was allegedly killed by them.2

In relation to the level of violence depicted (in 37 of the incidents in section D the suspect was recorded as having been involved in direct aggression or having posed a direct threat to the police) the number of injuries and deaths of police officers seems to be relatively small. How to account for this fact is not very clear. On the one hand it may indicate, for instance, that the police are simply far superior in the use of force and that they therefore generally gain the upper hand in violent encounters with offenders. On the other hand it may suggest that the level of violence or threat of violence by suspects against the police is sometimes exaggerated by the police, possibly as a way of justifying, and legitimating in legal terms, their own use of force. Whatever the explanation the figures indicate that the high number of police deaths which is also a feature of the current period are not necessarily strongly related situationally to the high number of police/custody action deaths.

Other Variables - Legislation, Management, Training and Police Attitudes and Values

At the end of the day therefore an open question remains as to in what proportion of the cases here the police have acted lawfully and in terms of proper procedure. Whatever the answer to this question may be it might still be the case that even if the level of adherence by the police to the law and police procedure were to be improved the rate of police custody/action deaths would not be substantially reduced. Other factors which might also be regarded as variables which contribute to the high rate of police custody/action deaths include:

  • The legal and regulatory framework - the present legal position continues to justify the use of lethal force in relation to particular property offences including vehicle theft, housebreaking and stock theft. One of the questions on which there therefore needs to be greater clarity is on the whether the existing legal and regulatory framework is appropriate. Particularly in relation to the use of force by the police it may be worthwhile to bear in mind Geller and Scott's remark that "the empirical research suggests with remarkable unanimity … that restrictive [shooting] policies seem to have worked well where they have been tried. Their adoption usually is followed by marked decreases in shootings by police, increases in the proportion of the shootings that are responsive to serious criminal activity, greater or unchanged officer safety, and no resultant adverse impact on crime levels or arrest aggressiveness".3

  • Police training and management practices - the demanding nature of policing in South Africa at the moment requires a high levels of skill from members of the SAPS. Potentially situations may occur where the police are acting technically within the law but where deaths may still be prevented if police are better skilled. Furthermore there may be considerable room for improvement in relation to management practice within the SAPS, whether this is in relation to attitudes which are inculcated or practices which are encouraged or discouraged or the way in which particular problems, such as police stress, which may impact on the overall rate of police custody/action deaths, are dealt with.

  • Police attitudes and values - as has been suggested in relation to the use of force the attitudes of police may "weigh more heavily than written rules".4 From the afore going report a picture has emerged as to the types of circumstances in which police custody/action deaths tend to occur. However it can also be assumed that certain attitudes, also feed into the overall problem of the high rate of these deaths. These would include police attitudes in relation to crime, suspects, offenders, the criminal justice system, the Bill of Rights, and their own role in. For instance an attitude which prioritises "fighting crime" over "protecting life" might feed into an attitude which sees the lives of suspects as expendable. Other attitudes might relate to how police understand suspects (for instance people in a stolen car), how police believe offenders should be dealt with by the criminal justice system (many of them for instance might support capital punishment), what police believe about the effectiveness of the criminal justice system and what needs to be done to redress problems in the functioning of the criminal justice system, and how police understand the Constitution and Bill of Rights. The types of attitudes would also be influenced by the attitudes of managers.

Thus, the suggestion is, the problem of the high rate of police custody/action deaths can potentially only be tackled effectively if the legal and regulatory framework, police training and management practices, and police values and attitudes are addressed in addition to the problem of unlawful actions by the police. Other factors relating to the overall environment of the SAPS such as levels of morale and discipline are no doubt also relevant.

The suggestion is that addressing these factors will assist in reducing the overall number of police custody/action deaths notwithstanding current circumstances in South Africa. If other factors such as high levels of (violent) crime and the high levels of possession and use of weapons amongst the offender population, were to be addressed this would no doubt also assist in contributing to reducing the overall rate of these deaths.

The Scope for Prevention - Is the Rate of Deaths Reducible?

Is it possible to reduce the overall number of police custody/action deaths? A study of civilians killed by the principal police agencies in the United States's 59 cities with populations greater than 250 000 over the period 1980-1984 shows a sharp reduction in the total number of civilians reported killed. "The peak year during that period was 1971, when 353 civilians were reported killed. The low year was 1984, with 172 reported civilian casualties".5 Examples of a particularly drastic declines in the number of people killed were Philadelphia where the number dropped from 96 during the five-year period 1970-1974 to 38 during the period 1980-1984, and Detroit which dropped from 159 to 44 during the same respective five year periods. "Los Angeles also dropped from 107 to 96, a seemingly less significant decrease, but when ..the enormous population growth and even larger growth of the city's homicide rate during this period [were factored in] the reduction in killings in Los Angeles is comparable to the reductions" in other major US cities.6

These consistent and dramatic declines in the number of civilians killed (during the ten years from 1979 to 1988 the number of police officers feloniously killed by firearms ranged from a high of 100 in 1979 to a low of 62 in 1986) preceded the 1985 U.S. Supreme Court ruling in Tennessee v. Garner that "state laws which authorised police officers to use deadly force to apprehend unarmed, non-dangerous criminal suspects where unconstitutional".7

The US evidence at least would therefore appear to indicate that there may be the potential for dramatically reducing the overall number of police custody/action deaths. While the evidence provided relates primarily to "police action" as opposed to "police custody" deaths there would appear to be no reason to assume that significant reductions in the number of deaths in both areas are not attainable. However the reduction in overall number of police/custody action deaths which was achieved in the USA appears to have been achieved over a period of roughly 10-15 years. It would appear to be desirable that reductions in the number of these deaths in South Africa be achieved in a much shorter period. How can this be done?

Factors Relevant to the Development of a Framework for Prevention

It is suggested that the framework which is developed for the implementation of preventive measures should take into account the following factors:

  • Time frames

    In discussions held following a presentation of an earlier draft of this report it was suggested that what was needed was two or three "magic bullet" type measures which could be implemented in the short-term and which could contribute to a dramatic and relatively quick fall in the overall number of these deaths. However at this stage it appears that no such measures are readily apparent. It is suggested that the strategy which needs to be developed should rather focus on the implementation of a program of measures which may have a more substantial impact in the medium term (1- 5 years). It is suggested that a measure of the success of the programme would be for the overall number of police custody/action deaths to have declined to half of the present rate within 5 years.

  • Targeting

    It is suggested that a high level of impact may be better achieved by effective targeting of certain measures which are implemented. Thus it is suggested that measures which are implemented could be targeted at areas where there is a proportionately high rate of police custody/action deaths. It may also be appropriate to target some of the measures implemented at specific ranks of the police service if targeting the measures at these ranks has the greatest potential to have an impact on the rate of these deaths. Certain measures may be intended for a more general impact, while other such measures, such as pilot programmes, could be targeted at specific areas. The question of targeting would have to be evaluated more clearly by further research.

  • Implementability

    Recommendations should be realistic in the sense of being implementable rather than idealistic. Priority should be given to identifying measures which have a direct potential to be implemented. Factors which might influence whether or not measures are implementable include resources, capacity (in terms of available skills), as well as the potential credibility of measures which are proposed.

  • Methods of implementation

    Method of implementation might include: reformulation of laws or regulations, public education programmes targeted at the police, training, pilot project/s, focus on best practices? In so far as measures involve the use of media of one kind or another special care should be given to ensuring that it is effectively executed. As will be seen it is suggested that ss53(2)(b) investigations in relation to police custody/action deaths should also be implemented in a preventive manner.

  • Integration with the prevention of other problems?

    A question in relation to which there needs to be greater clarity is the degree to which measures which are implemented should be integrated with measures which are intended to address other types of abuse. This might for instance to be seen as appropriate in relation to the issue of torture. On the one hand it may be argued that the issue of torture overlaps very directly with the issue of treatment and care for persons in custody. On the other hand it needs to be borne in mind that torture is always illegal and preventable while deaths are not necessarily the result of unlawful actions but even so, may be preventable. Tarring the two issues using the same brush might not therefore be entirely appropriate.

  • Getting "buy in" from the police

    For preventive measures to be implemented in such a way as to have the maximum impact in it necessary that their implementation be supported by the police. Any measures which are implemented should therefore be motivated for partly at least in relation to the benefits which they will have for the police, such as enhancing their image, improving their relations with the public, and winning support and respect for the police from the community. Measures which are implemented may also be motivated for on the grounds that they are part of an initiative towards improving the professional standards of the police. The idea of "greater professionalism", may also win support for measures from the police partly if they feel that such professionalism will win them respect.

  • Strategy

    These issues however need to be handled with sensitivity. On the one hand it needs to be recognised that, if the proposed measures have credibility with the police it will be easier to implement them effectively. On the other hand it needs to be recognised that winning support from the police service (management and rank and file) for the issue of police custody/action deaths to be taken seriously and for measures aimed at reducing the overall number of these deaths, may involve something of an "ideological struggle". At stake are likely to be a range of issues including: why "offenders" lives should be a matter of concern; the relative importance of the rights of the "perpetrator" and the general public or victims; the responsibilities of the police in relation to human rights; how to win the respect and support of the public - is this through more effective law enforcement or respecting human rights (or combining the two). What this implies is that the importance of the issue of values and attitudes should not be underestimated as well as that whatever approaches are developed be informed by a strong sense of strategy.

  • Political will

    It must be emphasised that the suggested programme will need to be implemented in a manner which is clearly motivated by a will and determination that the envisaged measures succeed.

A Proposed Strategy for Prevention

The framework is intended as one which will maximise the motivation and ability of the SAPS to prevent police custody/action deaths. It is a framework for measures to be implemented jointly by the SAPS and ICD and not purely a framework for measures to be implemented by the SAPS.

The framework has two major components:

  • The first set of measures are suggested which will contribute to building an environment within the police service which supports the prevention of custody/action deaths.

  • The second set of measures are focused on attitudes, problems and practices within the police service which have a direct bearing on the occurrence of police custody/action deaths.

A. Creating an environment which supports the prevention of police custody/action deaths
(a) Getting the issue acknowledged and taken seriously

This may involve measures directed at police management, politicians and government safety and security structures. These will aim to ensuring that the issue is taken seriously and that the programme of prevention is undertaken with a high degree of support.

(b) Review of laws, regulations and procedures

Such a review should aim to ensure that there is consistency between different laws and regulations which are relevant to the issue of police custody/action deaths as well as ensuring that such laws and regulations support the prevention of such deaths. Consideration should be given both to improving policies (for instance by changing regulations) and to issued of improve policy implementation.

Issues covered should include:

  • laws authorising the use of force and particularly the use of lethal force
  • regulations covering the management of persons in custody
  • laws and regulations relating to the investigation of police custody/action deaths and shooting incidents or other incidents of the use of force. In particular the issue of duplication or overlap between the ICD and police investigative systems particularly in relation to fatal "shooting incidents".
  • the legal position and regulations relating to the position of police officers off-duty, the issue service pistols and the use of force by them.
  • other laws and regulations which may be identified.
(c) Using the investigation of police custody/action deaths to improve prevention.

Presently it appears that ss53(2)(b) investigations are primarily focused on the question of whether or not police officers are in any way criminally liable for custody/action deaths. In so far as ss53(2)(b) investigations are essentially criminal investigations police officers who are the subject of the investigation are entitled to make use of their right to remain silent and therefore not to co-operate with the investigations.

It is suggested that an approach which uses investigations for the purposes of "professional review" as well may be preferable in that it may be more conducive to cooperation by police officers with investigations, and may be more effective in enabling investigations to address issues, and contribute to building up a body of understanding, as to how custody/action deaths may be prevented.8

This type of approach has been supported in relation to the use of force by Geller and Karales who "see promise not only for an adjudicatory panel but for a departmental panel whose mission would be to review shootings for the purpose of identifying preventive strategies, whether in the area of policy development, training, weapons modifications, or supervisory changes".9

An approach which emphasises prevention will be reflected in the types of findings which are made following investigations. Thus for instance "18 percent of the 2,155 Firearms Discharge Review Board Adjudications Fyfe reviewed in New York City held that, although the shootings were technically within policy guidelines, the officers should be given additional training in the use of firearms or in relevant law and policy".10

(d) Institutionalising accountability in respect of police custody/action deaths

A system which institutionalises accountability may also have the effect of encouraging an attitude whereby members of the police service are concerned to be able to explain that they acted reasonably. However for a system of accountability to operate properly the issue of police officers making use of their "right to remain silent" will have to be addressed.

Accountability also should not necessarily be limited to the "line" officers. Consideration should also be given to the issue of the accountability of supervisors and managers (in particular station commanders and the heads of special units) for implementing practices which support the prevention of custody/action deaths.

B. Building a body of knowledge, experience and understanding which supports prevention

This component of the strategy potentially encompasses all practices, attitudes, values and problems within the police service which may be seen to have an impact on the rate of police custody/action deaths. What is suggested is a process of continuous review and improvement of relevant practise and procedure with a view to building up the body of knowledge, values and practices within the SAPS which supports the prevention of police custody action deaths. This process would happen through:

(a) Building understanding around good/best practices which are relevant to preventing police custody/action deaths.

Such understanding would be drawn from existing knowledge within the police service, processes of review and evaluation of specific incidents, international experience and possible pilot projects. It would deal with practices relating to:

  • Self defence and the security of police officers
  • Arrests and the use of force in affecting arrests
  • The security and care of people in custody
  • Off duty firearms
  • Physical and mental health and stress issues
(b) Addressing issues of police attitudes and values

The object of such efforts "is to encourage a high respect for human life among officers and encourage responsible - but not foolhardy - restraint in the use of force".11 The point is also supported by Skolnick and Fyfe who say that "if reform is to last, it must change the systems and values to which officers adhere rather than just the officers themselves".12

(c) Training and education

Training and public education programmes would be developed and targeted at appropriate components of the police service relating to practices and attitudes which are relevant. Such initiatives should aim in part to support a process of change within the police service to a culture which supports appropriate practices and values.

Particularly in relation to the use of force a question arises as to the type of training which may be most appropriate (if training is judged to be one of the solutions). Should such training, for instance, focus on the use of non lethal force, dealing with situations of potential violence by non-violent means, or should it focus on better practice with regard to the use of firearms.

C. A note about vigilantism

While the issue is not examined in any detail here it is apparent that the broad societal problem of vigilantism in South Africa is a contributing factor to the overall problem of police custody action deaths (notably the deaths in category B). While the strategy suggested here focuses on the police service it is no doubt also true that addressing the broader societal problem of vigilantism could also contribute to reducing the overall number of police custody/action deaths.

Further Research

From the above it follows that further research could usefully be conducted on issues including:

  • The legal and regulatory framework in so far as it impacts on police custody/action deaths - such research should aim to identify both existing laws and regulations which can contribute to the prevention of police custody/action deaths, if they are better implemented, and amendments to laws and regulations which may enhance prevention.
  • Training and managerial practices within the police service which affect the overall problem of police custody/action deaths.
  • Further research on circumstances of occurrence of police custody/action deaths - potentially necessary particularly to the targeting of interventions. In particular it might be worthwhile to attempt to establish to what extent the problem of police custody/action deaths is linked to specific areas, stations or members of the police service. Specific measures would have to be implemented to ensure that such research falls within confidentiality parameters.
  • Issues relating to the conduct of investigations into police custody/action deaths.
  • Issues relating to police attitudes and values in so far as they impact on police custody/action deaths.

Notes:

1 In this regard see also for instance Minnaar, A (1998): "An Analysis of the Murder of Members of the South African Police Service: 1994-1997". Paper presented to a Technicon SA conference, 25-26 February 1998. Division: Management Services. SAPS. The paper indicates that 924 me4mbers of the SAPS were murdered during the period 1994-1997. Analysis of a table describing the circumstance of death of 455 SAPS members indicates that 37% of those killed (170) may be said to have been killed in the course of their duties, 33% (150) in domestic or recreational circumstances, 126 (28%) as a result of being victims of criminal acts (including 83 robberies or attempted robberies of service pistols) and 9 (2%) in miscellaneous other circumstances.

2 This figure excludes the suicide by a police officer who had just killed his wife which is recorded in section F.

3 Geller, W.A and Scott, M.S. (1991): "Deadly Force: What we know", in CB Klockars and SD Mastrofski (eds), Thinking about Policing: Contemporary Readings, McGraw-Hill, New York, 1991, p.465.

4 Scharf and Binder referred to in Geller and Scott, Ibid, p.465.

5 Geller and Scott, ibid, p.452.

6 Geller and Scott, ibid, p.542.

7 Geller and Scott, ibid, p.463.

8 This issue is discussed further in Bruce, D: "Automatic suspects? Issues in the investigation of deaths in police custody or as a result of police action" Centre for the Study of Violence and Reconciliation. Forthcoming.

9 Referred to in Geller and Scott, ibid, p.466.

10 Fyfe referred to in Geller and Scott, ibid, p.466.

11 Delattre referred to in Geller and Scott, ibid, p.470.

12 Skolnick J.H. and Fyfe J.J. (1993) Above the Law: Police and the Excessive Use of Force. The Free Press. New York, p.187.

© Centre for the Study of Violence and Reconciliation

Copyright © 2020 CSVR. All Rights Reserved.