THE SUPREME COURT OF APPEAL
OF SOUTH AFRICA
Reportable
Case no: 417/03
In the matter between:
KOOS STEVENS Appellant
and
THE STATE Respondent
_______________________________________________________
Coram: Navsa, Van Heerden JJA et Ponnan AJA
Date of hearing: 26 August 2004
Date of delivery: 2 September 2004
Summary: Approach to evaluation of evidence of single witnesses • assessment of
totality of evidence • misdirections by trial and appeal courts.
_______________________________________________________
JUDGMENT
_______________________________________________________
2
NAVSA and VAN HEERDEN JJA:
[1] Courts in civil or criminal cases faced with the legitimate
complaints of persons who are victims of sexually inappropriate
behaviour are obliged in terms of the Constitution to respond in a
manner that affords the appropriate redres s and protection.
Vulnerable sections of the communi ty, who often fa ll prey to such
behaviour, are entitled to expect no le ss from the judiciary. However,
in considering whether or not claims are justified, care should be
taken to ensure that evidentiary rules and procedu ral safeguards are
properly applied and adhered to.
[2] The present appeal raises the ques tion as to whether the trial
court and, thereafter, the High Co urt to which the first appeal was
directed, followed this approach satisfactorily. The background
against which the appeal is to be decided is set out hereafter.
[3] The appellant, Koos Stevens, was convicted in the Wynberg
Magistrates’ Court on three count s of indecent assault and was
thereafter sentenced to one year’s im prisonment in terms of section
276(1)(i) of the Criminal Procedure Ac t 51 of 1977 on each count.
The appellant appealed against his conviction and the related
3
sentences to the Cape High Court, which dismissed his appeal. That
court, however, granted the appellant leave to appeal to this court.
The present appeal is directed only against conviction.
[4] At the time of the events wh ich led to his conviction, the
appellant was a detective sergeant at the Claremont police station.
He encountered the three compla inants, Rachel Vuyiswa Gxekwa
(Rachel), Samantha Lumkwana (S amantha) and N orooi Gogotya
(Norooi), for the first time on th e afternoon of 7 August 1998 at the
Claremont police station after Rachel had been arrested for
shoplifting at a clothing store.
[5] Rachel testified that shortly after her arrest, during questioning
by the appellant, he had caused h er to undress and had touched her
breasts, buttocks and thighs. According to Rachel, she was later
asked by the appellant to i dentify the two friends who had
accompanied her to the store from which the theft allegedly took
place. The result was that, ac companied by Rachel, the appellant
fetched Samantha and Norooi that night from Nyanga East, where
they lived, and transported them to the Claremont police station to be
questioned by the appellant. It is common cause that, although
4
Rachel had already been released on bail, she also returned to the
police station with the appellant and her two friends.
[6] Samantha and Norooi testified t hat they were each separately
questioned by the appellant in his office after being fetched from a
temporary holding cell in the c harge office. They were each then
taken to another room where, afte r taking their fingerprints, the
appellant caused them to undress and touched their breasts, buttocks
and thighs. According to Norooi, the appellant also touched her
vagina with his fingers.
[7] All three complainants testified that upon leaving the police
station they had informed Sam antha’s boyfriend, Bantoe Bonwana
(Bonwana) – who was waiting outside th e police station to take them
home - of what the appellant had done to them. He had immediately
confronted the appellant with thei r allegations, but the latter had
denied everything.
[8] Bonwana’s evidence was to the effect that, on the evening in
question, he had gone to the Clarem ont police station to fetch his
girlfriend, Samatha, and her two fr iends, Rachel and Norooi, all three
of whom he believed to have been a rrested. As he was walking to his
5
car with the three complainants, Samantha informed him that the
appellant had forced her to undre ss and had touched her body. The
other two complainants then told him that the appellant had done the
same thing to them. Bonwana imm ediately confronted the appellant
with these accusations. The latt er denied the allegations against him
and took Bonwana to his office whe re he purported to show him the
documents which he had completed in the course of interviewing the
complainants. Bonwana did not read these documents. Thereafter
Bonwana drove the complainants home. Upon being told by
Bonwana that the appellant ha d denied any impropriety, the
complainants became angry and in sisted that he had indeed forced
them to undress. No further eviden ce was presented on behalf of the
State.
[9] The appellant was the only witnes s for the defence. His version
was that on the afternoon in questio n, he took Rachel from the cell
where she was being held to his offi ce on the first floor of the police
station to obtain a warning statem ent and fingerprints. Whilst he was
thus engaged, a man and two wom en walked into his office and
enquired about her. It is apparent from the eviden ce as a whole that
the two women were the other two co mplainants whose identities at
6
that stage had not been revealed to the appellant. The appellant was
also not told at that time that they had been involved in the incident
which led to the charge of theft against Rachel. They stated that they
were in possession of a receipt that would prove that the item
allegedly stolen had been paid for. He advised them to produce it and
informed them that he was going to charge Rachel with theft.
[10] After obtaining a statement and fingerprints from Rachel, the
appellant permitted her to make a telephone call so as to arrange
bail. Her sister subsequently arrived at the police station and paid the
bail in an amount of R100-00. Whil e her sister was at the police
station, Rachel offered to ident ify two other pers ons who had been
involved in the incident at the clot hing store. The appellant, Rachel
and her sister thereupon travelled to Nyanga East where they found
the other two complainants who, according to Rachel, were the
persons in question. During the search for the other two
complainants, Rachel, her sister and the appellant had encountered
and spent some time with Rachel’s mother. Rachel, the appellant and
the other two complainants then travelled back to the Claremont
police station in the appellant’s motor vehicle.
7
[11] After their return to the police st ation Rachel waited in an office
on the first floor whilst Samantha and Norooi were placed in a holding
cell in the charge office on the ground floor from which each was
removed to be questioned by the appellant.
[12] Shortly after their arrival at the police station, the appellant
interviewed Norooi and Samantha in his office separately and at
different times. He questioned eac h about her involvement in the
shoplifting incident. He was satisfi ed with their explanation that they
were not present when the a lleged theft had taken place and
released them. They departed with Rachel. Shortly thereafter he was
confronted by Bonwana, who ha d apparently been waiting for the
three complainants in the parking area at the police station. Bonwana
said that the complainants had in formed him that the appellant had
touched their naked bodies. The appe llant denied this and attempted
to explain to Bonwana the proce dure he had followed in respect of
the three complainants. When Bonwana refused to listen to him, the
appellant had, via Bonwana, invited the complainants to lay a charge
against him.
8
[13] In his testimony the appell ant denied that he had committed
any impropriety in respect of any of the three complainants. Thus, as
regards the indecent assaults allegedly committed by him, his version
conflicts completely with that of t he complainants. In respect of the
alleged indecent assaults, each of the complainants was a single
witness.
[14] In convicting the appellant, t he magistrate reminded herself that
she was dealing with a single witne ss in respect of the essentials of
each of the charges and stat ed that she was approaching the
complainants’ evidence with th e necessary caution. In her
assessment of this evidence, t he magistrate found that there were
‘sekere getrouheidswaarborge in hulle getuienis met betrekking tot
die uitleg van die gebou. . . [en] die prosedure wat gevolg is. . .’. She
found it reassuring that Samantha and Norooi could describe where
their fingerprints were taken, in the face of the appellant’s denial that
he took their fingerprints. In dismissi ng criticism of the contradictions
between the complainants’ evidenc e in court and their written
statements to the police conce rning the charges against the
appellant, the magistrate reasoned that such contradictions were
understandable and excusable, as al l three complainants are Xhosa-
9
speaking and their written statem ents had been obtained without the
assistance of an interpreter. She was also dismissive of the
suggestion made on behalf of the appellant that the complainants had
conspired against him. In this regard, she stated that there had been
no opportunity for Samantha and N orooi to discuss the matter with
each other and that they had, upo n leaving the police station,
immediately reported the indecent assaults to Bonwana who then
confronted the appellant about thei r accusations. The complainants’
version of events was, in her vi ew, further corroborated by this report
to Bonwana. She concluded that t he evidence of the complainants
was satisfactory.
[15] As regards the appellant’s evid ence, the magistrate considered
certain aspects thereof as ‘nie baie duidelik nie’. She found it strange
that the appellant had placed Sa mantha and Norooi in a holding cell
when, on his version of events, th ey had not been arrested. She also
found it troubling that, although Rachel had been released on bail
earlier on the day in question, the appellant considered it necessary
to transport her to Nyanga East and then back again with the other
two complainants. She failed to understand why the appellant had
informed Norooi and Samantha that he would ‘set their bail’ at a high
10
amount. She was not convinced by t he appellant’s explanation to the
effect that he had done this because he was u nhappy that they had
wasted his time.
[16] On appeal to it, the Cape High Court considered that the period
of almost two years between the da te of the alleged offences and the
appellant’s trial explained the co mplainants’ imperfect recall of
events. In the view of the court below, the numerous contradictions in
the complainants’ evidence did not impinge on the reliability of ‘die
kern van die verhaal wat hulle vertel’. Erasmus AJ, with whom Motala
J concurred, echoed the magistrat e’s view about the language
problem leading to the differenc es between the complainants’
evidence in court and their writt en statements. In essence, the court
below, in a terse judgment, simply confirmed the convictions on the
basis as set out by the magistrate in her judgment.
[17] As indicated above, each of the complainants was a single
witness in respect of t he alleged indecent assault upon her. In terms
of s 208 of the Crimin al Procedure Act, an accused can be convicted
of any offence on the single evidence of any competent witness. It is,
however, a well-established judicial practice that the evidence of a
11
single witness should be approached with caution, his or her merits
as a witness being weighed against factors which militate against his
or her credibility (s ee, for example, S v Webber 1971 (3) SA 754 (A)
at 758G-H). The correct approach to the application of this so-called
‘cautionary rule’ was set out by Diemont JA in S v Sauls and Others
1981 (3) SA 172 (A) at 180E-G as follows:
‘There is no rule of thumb test or formula to apply when it comes to a
consideration of the credi bility of the single witness (see the remarks of Rumpff
JA in S v Webber . . .). The trial judge will weigh his evidence, will consider its
merits and demerits and, having done so, will decide whether it is trustworthy and
whether, despite the fact that there are shortcomings or defects or contradictions
in the testimony, he is sati sfied that the truth has been told. The cautionary rule
referred to by De Villiers JP in 1932 [in R v Mokoena 1932 OPD 79 at 80] may be
a guide to a right decision but it does not mean “that the appeal must succeed if
any criticism, however slender, of t he witnesses’ evidence were well-founded”
(per Schreiner JA in R v Nhlapo (AD 10 November 1952) quoted in R v
Bellingham 1955 (2) SA 566 (A) at 569.) It has been said more than once that
the exercise of caution must not be allo wed to displace the exercise of common
sense.’
(See further in this regard the usef ul discussions in respect of single
witnesses in DT Zeffert, AP Paizes and A St Q Skeen The South
12
African Law of Evidence 4ed (2003) p 799-801 an d in CWH Schmidt
and H Rademeyer Law of Evidence (2003, with looseleaf updates)
para 4.3.1.)
[18] As will be discussed in further detail below, an analysis of the
magistrate’s judgment in this case supports the conclusion that in fact
she failed to approach t he evidence of the complainants with the
necessary caution. Moreover, her judgment illustrates the dangers of
what has been called ‘a compar tmentalised approach’ to the
assessment of evidence, namely an approach which separates the
evidence before the court into compartments by examining the
‘defence case’ in isolation from the ‘State’s case’ and vice versa. In
the words of Nugent J in S v Van der Meyden 1999 (1) SACR 447
(W) at 449c-450b:
‘Purely as a matter of logic, the pr osecution evidence does not need to be
rejected in order to conclude that t here is a reasonable possibility that the
accused might be innocent. But what is required in or der to reach that
conclusion is at least the equivalent pos sibility that the in criminating evidence
might not be true. Evidence which incriminates the accused, and evidence which
exculpates him, cannot both be true – th ere is not even a po ssibility that both
might be true – the one is po ssibly true only if there is an equivalent possibility
13
that the other is untrue. There will be cases where the State evidence is so
convincing and conclusive as to excl ude the reasonable pos sibility that the
accused might be innocent, no matter that his evidence might suggest the
contrary when viewed in isolation.
. . . The proper test is that an accus ed is bound to be convicted if the evidence
establishes his guilt beyond reasonable doubt, and the logical corollary is that he
must be acquitted if it is reasonably possible that he might be innocent. The
process of reasoning which is appropriate to the application of that test in any
particular case will depend on the nature of the evidence wh ich the court has
before it. What must be borne in mind, how ever, is that the conclusion which is
reached (whether it be to conv ict or to acquit) must account for all the evidence.
Some of the evidence might be found to be false; some of it might be found to be
unreliable; and some of it might be found to be only possibly false or unreliable;
but none of it may simply be ignored.’
(See also S v Tellingen 1992 (2) SACR 104 (C) at 106a-h and the
judgments of this court in S v Van Aswegen 2001 (2) SACR 97 (SCA)
para 7-8 at 100f-101e and S v Trainor 2003 (1) SACR 35 (SCA)
para 8-9 at 40f-41c.)
[19] There were material contradictio ns between the evidence given
by the complainants during the tr ial and the contents of the police
statements made by them within a period of approximately six weeks
14
from the date of the alleged offe nces. These cont radictions related
not only to the nature of the indecent assaults to which they were
allegedly subjected by the appe llant, but also to the sequence of
events on the afternoon and even ing in question and the
circumstances and content of the report made to Bonwana.
[20] As indicated above, the magi strate purported to explain these
material contradictions on the basis that the three complainants are
Xhosa-speaking and that the polic e statements had been taken in
English without the assistance of an interpreter. In this regard, the
magistrate, in our view, overloo ked the complainants’ levels of
education and the fact that each such complainant testified that she
was proficient in English. None of the complainants was either
illiterate or unsophisticated. In deed, Rachel (21 years old) was a
student of the University of the Western Cape at the time of the
alleged offences, Norooi (24 years old) was in her Matric year and
Samantha (also 24 years old) had already passed Matric. Each
complainant confirmed that her statement to the police had been read
to her by the relevant police offi cer before she signed it under oath.
In the absence of any accept able explanation for the material
differences between the complai nants’ evidence in court and the
15
versions advanced by them in their police statements, it would appear
that the magistrate misdirected hers elf by failing to attach sufficient
weight to such differences in her assessment of the credibility of the
complainants.
[21] The magistrate’s finding of a ‘getrouheidswaarborg’ on the
basis that Samantha and Norooi were apparently able to describe the
layout of the police station buildin g, including the place where their
fingerprints were allegedly taken by the appellant, is fallacious.
Although Norooi denied visiting t he police station before she was
taken there by the appellant, it s eems clear from the evidence as a
whole that both she and Samantha had in fact gone to the police
station after Rachel had been arrested, had en tered the appellant’s
office in which Rachel was being questioned by the appellant and had
waited for a while in the passage outs ide that office. That office and
the ‘fingerprint room’ are on the same floor of the police station
building and it is common cause that both Norooi and Samantha were
later questioned by the appellant in th e said office. Neither purported
to give any description of the ro om in which their fingerprints were
allegedly taken (and in which they were also allegedly indecently
assaulted), other than stating that it contained a basin in which they
16
washed their hands. They did not provide a description of the position
of this room in relation to the appel lant’s office. Their sole testimony
in relation to the ‘procedure’ foll owed by the appellant was that he
had taken their fingerpr ints and had thereafter provided them with
soap or gel to wash their hands. Bearing in mind the fact that
Rachel’s fingerprints had been taken by the appellant that day and
that the three complain ants had, on their own versions, discussed
their experiences at the police st ation with one another on more than
one occasion, the ‘getrouheidswaarborg’ found by the magistrate to
exist in this regard carries no weight at all.
[22] The magistrate based her dismis sal of the suggestion that the
complainants may have conspired to concoct a story against the
appellant on her finding that th ere had been no oppo rtunity for the
complainants to discuss the matter with one another prior to the
report made by them to Bonwana. This finding is, however, not borne
out by the complainants’ evidence . Thus, for example, Samantha
testified that she a nd Norooi had told each other what the appellant
had done to them while they were together in the holding cell before
being allowed to leave the police st ation. Although this was denied
by Norooi, she in turn testified t hat the three complainants had
17
discussed their experiences at the hands of the appellant while they
were walking out of the police station to meet Bonwana.
[23] It must also be remembered t hat the three co mplainants were
taken by the appellant from Nyanga East to the Claremont police
station in the same car, some ti me after Rachel had allegedly been
indecently assaulted by the appellant . According to Rachel, she had
not, during the course of this jo urney, told her fr iends what the
appellant had done to her or warned them against him in any way.
The court below explained this by reference (inter alia) to the fact that
it was not clear whether or not t he complainants knew at that stage
that the appellant did not underst and Xhosa. This was not, however,
the explanation given by Rachel under cross-examination. According
to her, she had not said anything about the indecent assault on her to
her friends because she was ‘in shock’ and did not think that the
appellant would also indecently assa ult them. Her testimony in this
regard was contradictory, evasive and not at all persuasive.
[24] On a conspectus of all the ev idence the complainants did in fact
have the opportunity to discuss the matt er with one another, and
must indeed have done so, prior to making their report to Bonwana.
18
This being so, the magistrate’s find ing to the contrary indicates a
failure properly to assess the evidence in this regard.
[25] Another aspect not given du e consideration by both the
magistrate and the court below is that initially, during evidence in-
chief, the complainants’ reaction to what was clearly reprehensible
behaviour on the part of the appell ant was, hesitant (had to be
painstakingly extracted by the prosecutor by way of very leading
questions) and muted. The complai nants’ initial reaction was almost
uniformly that they found it strange that they were being ‘searched’ in
the absence of a policewoman. Later, under strenuous cross-
examination, their reactions mutated to ‘shock’.
[26] The magistrate does not appear to have given any
consideration in her judgment to the inherent probabilities of, in
particular, the appellant’s version (cf in this regard, S v Shakell 2001
(2) SACR 185 (SCA) para 30 at 194h-195a). The indecent assault on
Rachel was allegedly committed in an office on the first floor of the
Claremont police station on a Friday afternoon. By Rachel’s own
admission, other police officials (including a woman) were also
present at the station at that time. For at least part of the time which
19
Rachel spent with the appellant, Samantha and Norooi, together with
another man, were sitting in th e passage outside the appellant’s
office. While Rachel could not remember whether or not the
appellant had closed the door of t he room in which he had allegedly
assaulted her, the appellant’s evidence to the effect that he had not
closed the door of any of the rooms in which he had been with Rachel
was not disputed. Even much later that night, when the appellant had
questioned Samantha and then Norooi, other police officials were still
present on both floors of the polic e station. Two inspectors had in
fact entered the appellant’s office while he was questioning Norooi,
prior to their going off duty. Each of the complainants had been left
alone by the appellant for periods of time, with the opportunity of
telling other police staff at the st ation of what was happening, but
none of them had done so . It appears to be most unlikely that the
appellant would, in these circumstanc es, on three different occasions
have taken the risk of being in terrupted by one or more of his
colleagues while forcing young wom en to undress in front of him and
touching their naked bodies.
[27] It is also unlikely that, had the appellant indeed indecently
assaulted the complainants, his reaction upon being confronted by
20
Bonwana would have been, in a composed manner, to invite the
complainants immediately to lay charges against him. Yet Rachel and
Narooi testified that Bonwana inform ed them that the appellant had
indeed reacted in this way. F urthermore, according to Samantha,
Bonwana told all the complainants at the police station that they
should lay charges against the a ppellant. None of the complainants
was able to offer a plausible explana tion for why they had not in fact
laid charges against the appellant on the night in question.
[28] The magistrate’s concerns abo ut aspects of the appellant’s
evidence, as set out in para [15], do not take into account that his
actions, though open to criticis m, are explicable in the light of
preceding events.
[29] The contradictions betwe en the evidence of the three
complainants, and the inherent cont radictions and inconsistencies in
the evidence of each of them, a re numerous and, in many respects,
undeniably material. It is not, howev er, necessary to analyse these
contradictions in any greater detail t han that set out above. Suffice it
to say that, in our view, the magist rate failed to give sufficient weight
to these contradictions and, in a ddition, failed properly to weigh the
complainants’ evidence against that of the appellant and against the
21
probabilities. The same can be sa id of the approach adopted by the
court below in dismissing the appellant’s appeal to it. To our minds, it
cannot be said that the evidence, taken as a whole, establishes
beyond reasonable doubt t hat the appellant inde cently assaulted any
of the three complainants.
[30] The appeal succeeds. The appellant ’s convictions on the three
counts of indecent assault and the related sentences are set aside.
_______________
MS Navsa
Judge of Appeal
_______________
BJ van Heerden
Judge of Appeal
CONCUR:
Ponnan AJA