THE SUPREME COURT OF APPEAL
OF SOUTH AFRICA
Reportable
CASE NO. 105/2003
In the matter between
SANTAM BEPERK Appellant
and
VINCENT BIDDULPH Respondent
____________________________________________________________
CORAM: HARMS, ZU LMAN and HEHER JJA
HEARD: 26 FEBRUARY 2004
DELIVERED: 23 MARCH 2004
____________________________________________________________
Insurance fraud – credibility - upsetting finding of credibility on appeal
2
JUDGMENT
ZULMAN JA
[1] This appeal concerns an alle ged insurance fraud. The appeal is
brought with the leave of this court. The appellant, an insurance company,
unsuccessfully sued the respondent, a farmer, for payment of R691 745,00
together with interest and costs. The cause of action was that the appellant
was induced by the respondent to pay the amount as a result of a false
claim made by the respondent on a policy of insurance issued by the
appellant. The claim related to the dest ruction of a house on a farm and its
contents in a fire.
[2] It was common cause th at in the event of it being found that the fire
was started at the instigation of th e respondent the appellant would be
entitled to a refund of the amount it paid to the respondent.
3
[3] One central issue fell for decision by the court a quo. It was the
question of whether the claim that the respondent made was false or not. In
this regard the onus rested upon the appellant. The appellant led evidence
from six witnesses in support of its claim. The main witness relied upon by
the appellant was a Mr Sigasa. The ot her witnesses called by the appellant
were the respondent’s wife; Mr P C Bezuidenhout; the respondent’s brother
in law, Mr Du Randt, a police officer; Mr J Murray, who was attached to
the appellant’s Forensic Investigation Unit; and Mr Jansen van Rensburg,
the owner of a video production business and an expert in the making of
video recordings. The respondent cl osed his case without leading any
evidence.
[4] Sigasa’s evidence wa s to the effect that he set the house and its
contents on fire at the behest of the respondent. The court a quo found that
the evidence of Sigasa was not only unli kely but that it was untruthful. In
addition it found that the respondent’s wife had given acceptable reasons
4
for retracting certain sworn statements she made in which she accused the
respondent of arranging for the fire. The court a quo made no mention in its
judgment of the other four witnesse s called by the appellant. The primary
issue on appeal is whether the court a quo erred in rejecting the
uncontradicted evidence of Sigasa.
[5] Whilst a court of appeal is gene rally reluctant to disturb findings
which depend on credibility it is trite that it will do so where such findings
are plainly wrong ( R v Dhlumayo and Another 1948 (2) SA 677 (A) 706).
This is especially so wh ere the reasons given for the finding are seriously
flawed. Over-emphasis of the advantages which a trial court enjoys is to be
avoided lest an appellant’s right of appeal ‘becomes illusory’ ( Protea
Assurance Co. Ltd. v Casey 1970 (2) SA 643 (7) 648 D-E and Munster
Estates (Pty) Ltd v Killarney Hills (Pty) Ltd 1979 (1) SA 621 (A) 623H –
624A). It is equally true that findings of credib ility cannot be judged in
isolation but require to be considered in the light of proven facts and the
5
probabilities of the matter under consideration.
[6] An analysis of the evidence as a whole, including that of Sigasa,
proper regard being had to the probabilities, leads to the conclusion that the
finding of credibilit y by the court a quo is untenable (cf Stellenbosch
Farmers’ Winery Group Ltd and another v Martell et Cie and others 2003
(1) SA 11 (SCA) para 14I–15E. Almost at the outset of its judgment the
court a quo concluded that the appellant ’s claim depended exclusively
upon the evidence of Sigasa. This wa s not a correct assessment of the
matter since the court was plainly obliged to consider the evidence of all
the other witnesses called by the appellant.
[7] Quite apart from the bare say-so of Sigasa the Court had before it as
objective facts, not dependent on th e credibility of any witness, the
following-
(a) The respondent’s wife made a vi deo recording of all the movable
property in the farm house. The da te on the film is 8 October 1998.
6
In the ordinary operation of the cam era that date would have been
generated by the person operating the camera.
(b) On 10 October 1998 the respondent , his wife and children left their
house unattended and went to spend the night with the respondent’s
brother-in-law, Bezuidenhout, at Bapsfontein.
(c) During the night of 10 Octobe r 1998 the house was destroyed by fire.
(d) Either before or after the fi re Sigasa, with the respondent’s
concurrence, fetched from the farm a defective moto r car for which
the respondent had previously as ked R2000 and whic h Sigasa had
been unable to afford.
(e) During November or December the respondent’s wife phoned
Bezuidenhout. She told him that the respondent had been responsible
for the fire and that she feared for her life.
7
(f) On 2 December 1998 Bezuidenhout phoned Crime Stop and reported
the conversation with his sister. In turn, the poli ce communicated
with the appellant which then initiated enquiries.
(g) Towards the end of January 1999, the respondent’s wife secretly
recorded a conversation with her hus band in the course of which she
attempted to entrap him into ma king incriminating admissions about
the burning of the house.
(h) Also during January the responde nt’s wife prepared a statement
(dated 4 January 1999) which she posted to Bezuidenhout. The stated
purpose was to provide evidence in the event of her untimely decease
or unexplained disappearance. In it sh e described in detail the events
surrounding the fire. Most important, she implicated Sigasa in
respects which materially accord ed with the evidence which he
eventually gave in Court including the meeting with Sigasa in the
8
township where he lived, the handi ng to him of a container of
inflammable fluid and the instructions given by the respondent.
(i) Towards the end of January the respondent’s wife left her husband
and sought refuge with the said Bezuidenhout. She delivered the two
videotapes to him for safekeepi ng. She confirmed to Bezuidenhout
the contents of the letter which sh e had sent. There is a necessary
inference that the first videotap e was not destroyed in the fire
because the respondent’s wife remove d it to safety before the fire
took place.
(j) At the beginning of February th e appellant’s representatives, aware
that the respondent’s wife was no longer within her husband’s sphere
of influence, interviewed her at Ba psfontein. She agreed to make a
statement in the presence of her attorney.
(k) On 2 February 1999 at Pretori a the respondent’s wife duly deposed
to an extensive affidavit in whic h she described the respondent’s
9
conception and development of the scheme to defraud the appellant,
how Sigasa’s services were procur ed to that end, the subsequent
arrival of Sigasa at the farm to remove the motor vehicle with the
respondent’s consent, the dealings with the proceeds of the insurance
pay-out and the disclosure ma de by her to the appellant’s
representatives. Once again ther e were material and striking
coincidences between her account and Sigasa’s later evidence.
(l) On 3 February 1999 Sigasa was taken to Sandton where he made an
affidavit setting out his version of the events before, during and after
the fire. He implicated the res pondent as the prime mover. The
statement corresponded in most (but not all) material respects with
his subsequent evidence.
(m) On 4 February 1999 the respondent and Siga sa were arrested. The
remains of the motor vehicle whic h had come from the respondent’s
10
farm were recovered from a plac e at which Sigasa had abandoned
them.
(n) In April or May 1999 the responde nt’s wife was placed in a witness
protection programme. She remained there for a year or more before
absconding. She tried to persuade Bezuidenhout to hand back the
letter and the tapes but he declined to do so.
(o) The respondent and Sigasa we re prosecuted for arson. The
respondent’s wife did not testify. At the close of the State case both
were discharged.
(p) At the time of the proceedings in the court a quo the respondent’s
wife had returned to her husband.
(q) The respondent did not give ev idence notwithstanding his direct
implication in the fraud and the ease with which he could have
rebutted the evidence of Sigasa if it were untrue.
11
[8] It was against this background th at the trial Court was required to
assess Sigasa’s credibility and reliab ility as a witness and to consider
whether it should believe or disbelieve the respondent’s wife, particularly
concerning the reason for her retraction of her previous statements, namely
that they were simply an untruthful attempt on her part to vent her anger on
her husband for his violent and abusive conduct towards her. It is important
to note that there was no basis what soever for concluding that she and
Sigasa collaborated in preparing th eir versions. On the other hand her
evidence that the content of the letter was the result of a joint fabrication by
herself and Bezuidenhout was patently untrue – but not found to be so by
the court a quo. On a balanced overall assessm ent of the probabilities, and
subject to what is said below, the tria l Court must have found that the fact
that the respondent’s wife gave the accounts of the fire that she did provide
material corroboration for the evidence of Sigasa, not because they were
true (although that must also follo w) but because she could not have
12
produced a version which so closely co incided with that of Sigasa unless it
was the product of her own experience.
[9] There was a further overwhelm ing probability in favour of the
acceptance of Sigasa’s evidence which was overlooked by the court a quo.
There was no reason to believe that th e fire was started by anybody other
than Sigasa. There was no suggest ion that he conspired with the
respondent’s wife to start the fire. Hi s confession was ag ainst his interest,
even allowing that he (together w ith the respondent) had already been
acquitted on a charge of arson arisi ng out of the same events. It was
nowhere suggested that Sigasa had a ny reason or motive to lie about the
matter or falsely to implicate th e respondent; indeed his undisputed
evidence was that he was on good terms with the respondent when he left
his employment on the farm, when he met the respondent and the
discussion took place about setting the house on fire, when he returned to
collect the vehicle and in the period preceding his arrest.
13
[10] Sigasa may not have been a sa tisfactory witness in all respects.
However, the proper test is not whethe r a witness is truthful or indeed
reliable in all that he says, but wh ether on a balance of probabilities the
essential features of the story which he tells are true (cf R v Kristusamy
1945 AD 549 at 556 and H C Nicholas Cr edibility of Witnesses (1985) 102
SALJ 32 especially at 32 – 35). This is particularly so in this case where
the trial court rejected Sigasa’s evid ence on the basis of his veracity as
opposed to the reliability of his evidence.
[11] The court a quo gave various reasons fo r rejecting the evidence of
Sigasa and for finding in favour of th e respondent. First, it concluded that
the manner in which Sigasa set the h ouse on fire was ‘niks anders as ’n
verdigsel en ’n versinsel’. I do not ag ree. Sigasa is an unsophisticated
person with a standard three education. He gave evidence in simple terms
as to how he set the house and its contents on fire. The fact that a
professional arsonist or a person of more skill or imagination would have
14
gone about the task in a more efficien t way did not indicate that Sigasa’s
evidence was either a fiction or a fabrication. Second, the court a quo took
Sigasa to task because he originally made no mention of a purchase price of
R500,00 for the engine of the Mazda vehicle. In my view nothing turned on
this since it was not disputed that th e vehicle had been removed by Sigasa
with the consent of the respondent and was found in his possession after the
fire. Third, the court a quo concluded that because of the contradictory
evidence given by Sigasa as to when he removed the vehicle that ‘hy
verdoesel dan die waarheid verder de ur geveinsd te huigel waar hy erken
dat hy vroëer getuig het dat hy die Mazda voor die brand gaan haal het en
dat hy daarvan seker is’. If regard is had to the manner in which the cross-
examination was conducted (with, regr ettably, a great deal of often
unwarranted hostile and sar castic participation by th e court) and the fact
that Sigasa’s evidence was given thr ough an interpreter, I do not believe
that anything of any consequence flowed from this contradiction. The
15
probability remained that Sigasa obt ained the vehicle from the respondent
either before setting the house on fire as an inducement to do so or received
it afterwards as a reward for the deed.
[12] The court a quo gave as a further reason for rejecting the evidence of
Sigasa the fact that in a statement to the police he made no mention that in
addition to the vehicle, money was o ffered to him to set the house on fire
and that Sigasa was unable to explai n why he omitted this. In my view
nothing is to be made of this failure es pecially if regard is had to the fact
that Sigasa made the statement through an interpreter and at a time when he
was obviously fearful of the consequen ces of his conduct. In addition the
court a quo drew attention to the fact that Sigasa alleged in the statement
that the respondent handed a can contai ning liquid to him and told him that
one of the windows of the house had been left open to enable him to gain
access so that he could pour the liqui d inside the house. In his evidence,
although he first admitted that the cont ents of the statement were correct,
16
he later denied that he said what was attributed to him in regard to the can
but was unable to say why there was this difference. Again, if regard is had
to the often unfair pressure which Siga sa was subjected to in the witness
box, his lack of a coherent explana tion is understandable. The previous
statement was given in Sesotho and tran slated into English. Sigasa stated
that he did not know English very well. As I have already pointed out he
gave his evidence in court through an interpreter (a different person from
the police officer who translated the or iginal statement, and who translated
into Afrikaans, not English). In circ umstances such as these very little
significance can be attached to rela tively minor discrepancies between
words and nuances of meaning in comparing a prior statement with viva
voce evidence. The discrepancies relied upon (i.e. pouring out liquid from a
bottle as opposed to spraying it from a can, or the vessel being discarded
when empty as opposed to when some of its contents had been used), were
minor and inconclusive. On both vers ions, the respondent gave Sigasa a
17
receptacle containing a flammable substance, which was used by him to
start the fire. Sigasa’s evidence was substantially consistent with the
contents of his prior statement. Th e fact remained that on the simple
version deposed to by Sigasa he set fire to the house using a substance
given to him by the respondent. There wa s nothing to contradict this basic
version.
[13] It cannot be fairly said that Sigasa’s evidence was so improbable or
vague and ineffectual that it could be rejected out of hand as being untrue
thereby relieving the respondent of an y obligation to contradict it (cf
Siffman v Kriel 1909 TS 538 at 543 and Minister of Justice v Seametso
1963 (3) SA 530 (A) 534 H – 535 A). Indeed he did not deviate from his
essential statement that it was th e respondent and nobody else who
instructed him to set the house on fire . He had no motive to lie and this
statement was, in all the circumstances, probably true.
18
[14] The court a quo also drew attention to what it described as ‘’n
vreemde verskynsel wat onbeantwoord ge laat is’. The f eature was that
Sigasa’s statement was given a da y after the statement made by the
respondent’s wife. It found it ‘agt erdogwekkend’ that Sigasa had been
taken to Sandton to make the statemen t and that no warning was given to
him before he made the statement. The fact that Sigasa made his statement
the day after the respondent’s wife made her statement does not advance
the respondent’s case. Indeed the pr obabilities and the evidence of Murray
indicate that the statements came to be made as a consequence of
arrangements made by him. The fact th at no warning was given to Sigasa
before he made his statement did not detract from his credibility or from
the probabilities of the matter.
[15] The court a quo regarded it as improbable that the respondent would
enlist the services of Sigasa to ca rry out the deed when he had had no
contact with Sigasa for two and a half years. If anything, however, this fact
19
supported the probability that the respon dent set about arranging for Sigasa
to set the house on fire rather than detracted from it. Sigasa knew the
respondent very well, they were on friendly terms and Sigasa knew the
farm and farm workers. In all likelihood Sigasa was chosen for the very
reason that it would be difficult, if a la ter investigation took place, to link
Sigasa to the respondent or indeed ev en to ascertain his whereabouts. This
would not have been so had the respondent chosen one of his farm workers,
for example. In addition there was an inducement readily available which
could be offered to Sigasa at little cost to the respondent namely the Mazda
vehicle.
[16] Finally the court a quo relied upon the demeanour of Sigasa in the
witness box as being such that he wa s ‘’n patetiese en wankelrige figuur
wat nie die stempel van betroubaar heid waardig is nie’. This
characterisation was unwarranted b earing in mind the record of his
evidence, the lengthy cross-examination of him, the fact that he gave
20
evidence through an interpreter and th e deplorable attitude of the learned
judge towards him to which I have already referred. In any event the
importance of demeanour as a factor in the overall assessment of evidence
should not be over-estimated. As pointed out in President of the Republic
of South Africa and Others v South African Football Union and Others
2000 (1) SA 1 (CC) para 79 p 43:
‘The truthfulness or untruthfulness of a witness can rarely be determined by demeanour
alone without regard to other factors including, especially, the probabilities . . . a finding
based on demeanour involves interpreting be haviour or conduct of the witness while
testifying. A further and closel y related danger is the implic it assumption, in deferring
to the trier of fact’s findings on demeanour, th at all triers of fact have the ability to
interpret correctly the behaviour of a witne ss, notwithstanding that the witness may be
of a different culture, class, race or gende r and someone whose life experience differs
fundamentally from that of the trier of fact.’
It is plain that Sigasa was of a differe nt ‘culture, class and race’ whose ‘life
experience differs fundamentally from th at of the trier of fact’.The learned
21
judge’s failure to have regard to th e social dynamic is quite apparent from
his questioning of Sigasa and his a ssessment of his evidence. As to the
limited value of the finding on demea nour where evidence is given through
an interpreter see S v Malepane and Another 1979 (1) SA 1009 (W)
1016H - 1017A, S v Martinez 1991 (4) SA 741 (NmHC) 758 B – D and
Nicholas (supra) 36 - 37.
[17] Counsel for the respondent contended that there was another
essential matter of probability whic h favoured the respondent and which
entitled the court a quo to reject Sigasa’s eviden ce. It was unlikely that a
person would ask someone to set fire to his own house. There are simple
answers to this contention. First, on a factual basis, the house in question
belonged to the respondent’s fath er. More importantly, however, the
respondent required finance for another project at the time and set about the
entire scheme so as to derive the benefit of an insurance policy.
22
[18] In all of the circumstances I have no hesitation in finding that the
rejection by the court a quo of the evidence of Sigasa was wrong.
Accordingly the court a quo should have found that the appellant had
discharged the onus resting upon it.
[19] The following order is made:
19.1 The appeal succeeds with costs, such costs to include costs
consequent upon the employment of two counsel.
19.2 The order of the court a quo is set aside and replaced with an
order granting judgment in favour of the defendant for:
19.2.1 payment of the sum of R10 000,00 together with
interest thereon at the ra te of 15,5% per annum from
12 February 1998 to date of payment;
19.2.2 payment of the sum of R681 745, 00 together with
interest thereon at the ra te of 15,5% per annum from
9 November 1998 to date of payment;
23
19.2.3 costs of suit including the costs of employing two
counsel.
------------- --------------------------
R H Z U L M A N
J U D G E O F A P P E A L
HARMS JA )
HEHER JA ) CONCUR