S v Musingadi and Others (22/95) [2004] ZASCA 128 (23 September 2004)

[Home] [Databases] [WorldLII] [Search] [Feedback]
South Africa: Supreme Court of Appeal

You are here: SAFLII >> Databases >> South Africa: Supreme Court of Appeal >> 2004 >> [2004] ZASCA 128

[Database Search] [Name Search] [Recent Decisions] [Noteup] [Help]

S v Musingadi and Others (22/95) [2004] ZASCA 128 (23 September 2004)

.RTF of original document

THE SUPREME COURT OF APPEAL OF SOUTH AFRICA

Case No 22/95

REPORTABLE

In the matter between

DANIEL MUSINGADI FIRST APPELLANT

SARAH LAMBANI SECOND APPELLANT

SAMUEL MALENGA THIRD APPELLANT

DAVID MUTHIDZA FOURTH APPELLANT

and

THE STATE RESPONDENT

Coram: Farlam, Mthiyane JJA et Comrie AJA

Heard: 31 August 2004

Delivered: 23 September 2004

Summary: Criminal law – common purpose – expansion of – whether effective disassociation from.

Criminal law – circumstantial evidence – weight of – failure to controvert by defence testimony – inferences drawable.

Criminal law – extra-curial declarations by accused – weight of. Criminal law – death sentence abolished – other sentences substituted.

______________________________________________________

JUDGMENT
______________________________________________________

COMRIE AJA

[1] In December 1994 the four appellants were convicted by a circuit court, comprising Marais J and assessors, sitting at Louis Trichardt, of:

(1) murder; and

(2) robbery with aggravating circumstances.

A fifth accused, Ms Martha Mahtshemule, was acquitted. The first and second appellants were sentenced to death for the murder and to 12 years’ imprisonment each for the robbery. In terms of s 316A of the Criminal Procedure Act 51 of 1977, as it then stood, they enjoyed an automatic right of appeal to this court, which they exercised. The third and fourth appellants were sentenced to effective terms of imprisonment of 18 and 20 years each. The trial court granted them leave to appeal against their convictions only. Now, almost ten years later, the appeal is before this court.

[2] Upon our enquiry counsel were unable to furnish a full or complete explanation for this highly regrettable delay. Mr van Heerden for the state undertook to bring the matter to the attention of the Director of Public Prosecutions, Pretoria, and to institute inquiries, all with a view to avoiding a repetition. It is often said that justice delayed is justice denied. While this may be an overstatement in some contexts, it does underline the need for reasonable expedition. The present is not a case where leave to appeal was granted late or the appeal itself noted late. In the ordinary course it should have come before this court eight or more years ago. The unexplained delay is to be deprecated.

[3] The events giving rise to the convictions occurred in early July 1993. By the end of October practically all the evidence upon which the state eventually sought to rely had been assembled. That was before the Interim Constitution (Act 200 of 1993) came into force. The trial itself, however, was heard during the last quarter of 1994, and thus under that Constitution. In S v Makwanyane and another 1995 (2) SACR 1 (CC) the Constitutional Court declared the death sentence unconstitutional. By a subsequent amendment to the Criminal Procedure Act the death sentence was removed from the statute book (see the Criminal Law Amendment Act 105 of 1997, s 34). Counsel were correctly agreed that in the event of the murder convictions of first and second appellants being sustained, their death sentences must be set aside and replaced by other proper sentences. (See s 1(10) of Act 105 of 1997.)

[4] Appellants 3 and 4 were both arrested in Randburg on 30 August 1993. They were conveyed to Louis Trichardt where, later that same day, they made statements to magistrates Gericke and Boshoff respectively. The admissibility of these statements was contested on the ground that they had not been made freely and voluntarily. After a trial within a trial the court below received the statements in evidence, although at the same time it rejected statements made to magistrates by appellants 1 and 2. All four statements complied with the requirements of s 217(1)(b) of the Criminal Procedure Act, with the consequence that the onus rested upon the appellants to prove on a balance of probabilities that their statements were not freely and voluntarily made (the so-called reverse onus). The trial court held that appellants 3 and 4 had not discharged that onus. The reverse onus was subsequently held to be unconstitutional: S v Zuma and others 1995 (2) SA 642 (CC). The declaration of invalidity does not apply to the present matter: see para [44] of the judgment of Kentridge AJ. It should be noted, moreover, that the trial court also said:

‘Dit is onnodig om op die bewyslas te steun. Ons bevind dat die bekentenisse bo redelike twyfel vrywilliglik deur beskuldigdes 3 en 4 gemaak is.’

[5] The deceased, Mrs Dercksen, was a middle-aged woman who, with her husband, lived on the farm Doornspruit in the Louis Trichardt district. Her parents, Mr and Mrs du Toit, lived in another house on the same farm. Security around the Du Toits’ house was tight: it included barbed-wire fencing and Pitbull terriers. Within the same house, in a room called the office, substantial sums of cash were accumulated from the farm’s business transactions and stored in a safe. The Du Toits’ domestic worker was the second appellant, Ms Sarah Lambani. In the absence of the Du Toits – they were away at the relevant time – she had the keys to the compound and she was under strict instructions as to who might and might not be admitted. The key to the safe was held by the deceased.

[6] On 5 July 1999, in her parents’ house, the deceased was assaulted and strangled to death. There was also an attempt to poison her by the administration of a noxious mixture containing acid. The safe was opened and cash was taken. As there were no signs of forced entry, suspicion fell on the second appellant. Within days she, her alleged aunt (accused no 5) and the first appellant were arrested. On the day of his arrest the first appellant pointed out certain places to Capt van Staden. After a second trial within a trial, the trial court admitted the contested evidence of such pointings out and accompanying declarations. On this occasion the trial court held that the state had proved voluntariness beyond reasonable doubt. In the same judgment the trial court declined to accept evidence of admissions made by accused no 5 to another police officer, and of a pointing out by her to him, on the ground that the state had not discharged the last mentioned onus of proof.

[7] The police trail led to appellants 3 and 4 who were eventually located in Randburg, where they were arrested on 30 August 1993. As I have said they made their statements to magistrates in Louis Trichardt on the same day.

[8] The appellants applied for bail in the magistrate’s court. In terms of s 235 of the Criminal Procedure Act the trial court received the evidence given by the first appellant at the bail enquiry, but not the evidence given by the other applicants for bail.

[9] At the trial all the appellants, and accused no 5, pleaded not guilty. Appellants 1, 3 and 4 (and accused 5) raised alibi defences. The second appellant’s defence, announced by her counsel at the plea explanation stage, was that she had been a victim of the robbery, which had been carried out by men wearing balaclavas over their faces. The state adduced no direct evidence implicating any of the appellants in the commission of the robbery or the murder. At the close of the state case the evidence implicating them indirectly consisted in the main of the material to which I have already referred, plus the testimony of Ms Janet Majiye who was living with the third appellant as his wife. She gave evidence against appellants 1 and 3 and against accused no 5.

[10] The first and second appellants closed their cases without giving evidence or calling witnesses. Appellants 3 and 4 testified and persisted in their alibis. Appellant 4 called his mother in support of his alibi. Accused no 5 testified in her own defence; among other things, her evidence affected appellants 1 and 2.

Admissibility of Evidence

[11] It is convenient to deal first with the attacks on the admissibility of evidence advanced by counsel for the appellants, Mr Barnard. He submitted that the evidence given by the first appellant at his application for bail should not have been received by the trial court, because it did not appear from the record that the appellant had been ‘properly warned’ before testifying in support of bail. The evidence in question was to the effect that on the morning of 5 July 1993 he visited the second appellant, with whom he claimed to have had a relationship, at the farm from about 9.00 am to 10.00 am. This prima facie destroyed his alibi, which was also undermined by the evidence of Ms Majiye, but did not link him directly with the crimes which were probably committed later in the day. At the bail hearing first appellant was represented by an attorney, Mr Hamman, who elicited this piece of evidence from his client during examination in chief. As to this, the prosecutor also put a few questions.

[12] The bail evidence was tendered, and received, by the trial court in terms of s 235 of the Criminal Procedure Act. It was not received in terms of s 60(11B)(c); that subsec – which provides for a warning by the judicial officer presiding – was not yet in existence. At the trial counsel for no 1 raised a single objection, namely that the record of the bail proceedings was not properly certificated in terms of s 235. The objection was not upheld. Counsel for other accused successfully objected to their bail evidence on other grounds, but counsel for no 1 raised no question of unfairness. In the absence of such an objection, there was no enquiry by the trial court into fairness. See S v Nomzaza 1996 (2) SACR 14 (A); S v Dlamini; S v Dladla and others; S v Joubert; S v Schietekat 1999 (2) SACR 51 (CC). It can safely be inferred from attorney Hamman’s conduct of the bail application, and from his evidence at the trial on other issues, that he was an experienced and competent attorney. We have to assume that he advised his client of his rights and about the implications thereof. Prima facie fairness was served. Absent an objection on this score, and a consequent enquiry by the trial court into the issue thus raised, there is no basis upon which we can conclude by way of conjecture that there was any unfairness when the appellant testified as he did. That such evidence was later to prove adverse to his alibi does not amount to unfairness. As Kriegler J trenchantly observed in Dlamini’s case at [95]:

‘That [constitutional] shield against compulsion does not mean, however, that an applicant for bail can choose to speak but not to be quoted. As a matter of policy the prosecution must prove its case without the accused being compelled to furnish supporting evidence. But if the accused, acting freely and in the exercise of an informed choice, elects to testify in support of a bail application, the right to silence is in no way impaired. Nor is it impaired, retrospectively as it were, if the testimony voluntarily given is subsequently held against the accused.’

[13] As I have said earlier, the bail evidence did not implicate the first appellant in the commission of the crimes. What implicated him to some extent were his pointings out to Capt van Staden and the accompanying explanations. The fairness of the procedure whereby Van Staden, who was independent of the investigation, conducted and recorded the pointings out and accompanying explanations, was not questioned at first instance or on appeal. The appellant was arrested on the morning of 9 July 1993 and later that same day accompanied Van Staden to the scene. The defence objected to this evidence on the ground that after his arrest the appellant was assaulted by other police officers, one such assault in the bush being particularly severe. This was the subject of the second trial within a trial. The appellant also testified that in small measure he was told by those policemen what to say, but that for the most part he invented the information which he conveyed to Van Staden. The police witnesses denied the assaults and attributed the wounds on the appellant’s forearms to a struggle that occurred when, wearing handcuffs, he refused to climb onto the back of the police bakkie and had to be forced to do so. Dr Botha, who examined the appellant a few days later, expressed the opinion during the first trial within a trial that the forearm injuries could not have been caused by the handcuffs. She changed her opinion during the second trial within a trial when she was shown similar handcuffs and the circumstances were explained.

[14] Of greater significance was that the appellant did not manifest the injuries which were to be expected from the assaults, and especially from the severe assault in the bush. The appellant claimed that among other things he had been repeatedly hit with the handle of a pick-axe on his body and on his head. To Dr Botha he claimed multiple blows to the head; in evidence he claimed one blow. An injury to his chest, which the appellant claimed to Dr Botha was the result of a kick, turned out to be an old, healed rib-fracture sustained, it would seem, in a bus accident. But recent injuries, other than to his forearms, were not present.

[15] Much was made during the second trial within a trial of an alleged swelling of the appellant’s face. Dr Botha did not observe it, nor did magistrate Roos. The photographs (exhs S and S1) taken of the appellant immediately prior to his departure on the pointing out expedition and immediately after his return, do not support the existence of such a swelling, especially bearing in mind that the appellant has a naturally full face. I accordingly do not think it necessary to go into the evidence of attorney Hamman and magistrate Boshoff on this point save to mention that the pointings out occurred before the appellant was advised by attorney Hamman not to make any statements.

[16] By the end of the lengthy trial within a trial the appellant was thoroughly discredited, and the trial court so found. On appeal Mr Barnard was unable to challenge that finding. He sought, however, to paint a picture of persistent police violence towards all five accused, including first appellant. Among other matters we were referred to the statements by appellants 1 and 2 which were excluded (no 1’s statement was made almost three months later, on 5 October 1993); to the exclusion of admissions allegedly made by accused no 5 on the day of arrest; and to the episode on 14 July 1993 when first appellant was taken to magistrate Roos, claimed to have been threatened and assaulted, and then decided not to make a statement. While all these matters may raise a suspicion in the mind of a court, the trial court after careful enquiry found it proven beyond reasonable doubt that the pointings out to Van Staden were made by the appellant freely and voluntarily. I see no ground for interfering with the conclusion of the court a quo.

[17] With regard to the statements made on the day of their arrest by third and fourth appellants to magistrates Gericke and Boshoff respectively, Mr Barnard submitted: (a) that the trial court applied the wrong (ie reverse) onus of proof; and (b) that applying the correct onus (ie beyond reasonable doubt) the trial court should have concluded that the statements were not made freely and voluntarily. The first submission is unsound in law. Para [44] of the judgment in S v Zuma, supra, reads:

‘[44] The application of s 217(1)(b)(ii) since 27 April 1994 may well have caused injustice to accused persons, but we cannot repair all past injustice by a simple stroke of the pen. Weighing all the relevant considerations it seems to me that the proper balance can be struck by invalidating the admission of any confession in reliance on s 217(1)(b)(ii) before the date of our declaration, but in respect only of trials begun on or after 27 April 1994, and not completed at the date of delivery of this judgment. The effect might be in those trials to require reconsideration of the admissibility of confessions already admitted, including the hearing of further evidence.’

[18] The present appeal falls precisely within the class of transient cases excluded from the declaration of invalidity. When this was pointed out to Mr Barnard during argument, he correctly conceded the point. He was then asked whether he could submit that appellants 3 and 4 had discharged the reverse onus. He conceded that he could not. That too was a correct concession on the evidence. Mindful, however, of the remarks of Kentridge AJ, I will briefly re-examine the position. It will be recalled that the trial court said:

‘Dit is onnodig om op die bewyslas te steun. Ons bevind dat die bekentenisse bo redelike twyfel vrywilliglik deur beskuldigdes 3 en 4 gemaak is.’

In truth by the end of the first trial within a trial the credibility of appellants 3 and 4 was in tatters. The high watermark of counsel’s argument was the portrait of persistent police violence to which I have adverted earlier. That is insufficient to disturb the trial court’s finding quoted above. On the contrary, the strong probability in my view is that once apprehended in Randburg, these appellants realised that the game was up, at least in part. They accordingly decided of their own volition that it was in their interests to admit the less serious charge of robbery, but to distance themselves from the more serious charge of murder. That is how their statements read. In my opinion, therefore, submission (b) above must also fail.

[19] It follows from the aforegoing that all the contentious evidence admitted against the several accused was in my judgment correctly admitted. I should mention that in the judgment on conviction Marais J was at pains to point out that the extra-curial declarations by appellants 1, 3 and 4 were admissible against their authors only, and not against their co-accused.

The Convictions

[20] It is convenient to begin with the second appellant. The evidence against her, which I shall call the general evidence, was entirely circumstantial. It was admissible against the second appellant and provided that an adequate link was established, against the other appellants as well. We should perhaps remind ourselves at this stage that there is nothing wrong in principle with circumstantial evidence. On the contrary it can sometimes be compelling. In the prelude to their discussion of R v Blom 1939 AD 188 and the rules of inferential reasoning, Zeffertt, Paizes and Skeen: The South African Law of Evidence rightly say at 94:

‘Circumstantial evidence is popularly supposed by laymen to be less cogent than direct evidence. This is, of course, not true as a general proposition. In some cases, as the courts have pointed out, circumstantial evidence may be the more convincing form of evidence. Circumstantial identification by a fingerprint will, for instance, tend to be more reliable than the direct evidence of a witness who identifies the accused as the person he or she saw. But obviously there are cases in which the inference will be less compelling and direct evidence more trustworthy. It is therefore impossible to lay down any general rule in this regard. All one can do is to keep in mind the different sources of potential error that are presented by the two forms of evidence and attempt, as far as this is possible, to evaluate and guard against the dangers they raise.’
This passage was quoted by Mthiyane JA in the judgment of this court in S v Mcasa and another (delivered 15 September 2003, unreported, Case No 638/2002) at para [8]. The substance of the passage can be traced back to Hoffmann: SA Law of Evidence (1ed, 1963) at 31.

[21] The evidence against the second appellant may be summarised as follows. She was employed at the farm as a domestic worker in the house of the Du Toits. They were away at the time. The house was heavily secured by barbed-wire fencing, said to be eight foot tall, and two dogs of savage mien. When visitors called, the dogs had to be locked away. The windows had burglar bars. The appellant had the key to the compound. The gates had to be kept locked at all times. She was under strict instructions as to who might and might not be admitted. She knew that cash was kept in the safe in the office. She knew that the deceased (the Du Toits’ daughter) had the key to the safe. She knew that there was caustic soda in the pantry. She was a short, thin woman whereas the deceased was a thickset middle-aged woman. The deceased was overpowered, tied up, and severely assaulted. There was an attempt to poison her with a noxious acidic mixture. She was eventually strangled to death by the use of what seems to have been a tie around her neck. The safe was opened and cash taken. There were no signs of forced entry on the day in question.

[22] The appellant’s defence, stated on her behalf at the beginning of the trial, was that she was a victim, not one of the perpetrators: that three men wearing balaclavas had gained access to the house while she was working, detained her and apparently committed the crimes. The appellant gave no evidence to this effect. Two matters were raised in cross-examination of Mrs du Toit by counsel for no 2. The first had to do with the sale of meat. This can be disregarded because Mrs du Toit said that such sales were rare and occurred well away from the house. There was no evidence to the contrary. The second aspect related to a possibly different arrangement with regard to the key between the appellant and the deceased in the absence of the Du Toits. Naturally, Mrs du Toit could not testify with certainty on this point. The appellant did not give evidence of such a different arrangement.

[23] None of appellants 1, 3 and 4 implicated the second appellant in oral evidence. No 1 did not testify; no’s 3 and 4 were steadfast in their alibis. The case appeared to take a turn right at the end, when accused no 5 testified. In answer to questions by the court, and then by the prosecutor, she spoke of inconsistent explanations of the robbery furnished to her by the second appellant. Her counsel was afforded the opportunity of additional cross-examination. In the judgment on conviction, weight was attached to the evidence of no 5 and to the propositions which were put and not put to her by counsel for the second appellant. Without holding that the trial court erred in his connection, I am of opinion that this was an unnecessary excursus, given the weight of the prosecution case.

[24] The uncontroverted evidence all points in one direction. The absence of the Du Toits presented a good opportunity to steal money from the safe, to which the deceased held the key. To obtain that key the deceased had to be overpowered, a feat which no 2 was physically unlikely to achieve on her own. So the three men, to whom she refers in her plea explanation, were called in to assist. They could not have got through the barbed wire, and past the dogs, unless the appellant admitted them. This is confirmed by the absence of signs of forced entry. The inference is inescapable that the second appellant was party to a conspiracy to rob the deceased and its implementation. Since grievous bodily harm was inflicted on the deceased before, during or after the robbery by one or more of the robbers, the appellant’s guilt on the charge of robbery with aggravating circumstances was in my view proven beyond reasonable doubt.

[25] However, there was also an attempt to kill the deceased by poisoning her, followed by her actual death as a result of strangulation. The explanation for these events must surely lie in the fact that no 2 was well-known to the deceased. Had the deceased survived the robbery, she would have identified no 2 as the household traitor. The deceased therefore had to be killed in order to avoid no 2’s detection. Again, the inference appears to be inescapable that the appellant was party to the murder. First, she had the prime motive to avoid detection. Second, she knew about the caustic soda in the pantry, whereas the others probably did not. It is possible that the appellant did not participate in or foresee the strangulation. If so, it matters not: the deceased’s death by whatever means was in the air, it was part of the appellant’s plan. I accordingly have no doubt that the second appellant was correctly convicted of murder with direct intent to kill.

[26] I turn to appellants 3 and 4. The evidence which specifically implicated no 3 was his statement to magistrate Gericke, made on the day of arrest. This implicated no 3 in at least the robbery. In addition Ms Majiye testified that on what seems to have been the morning of 5 July he was fetched by a woman who may have been accused no 5. The appellant had previously told his wife that there was money on the farm. Later that day, according to Ms Majiye, the appellant left for the city. The third appellant’s defence was that he had been working in Randburg since the beginning of the year as a welder. It was common cause that he was arrested in Randburg at the end of August. He attributed his wife’s adverse testimony to the improper influence of the investigating officer. The trial court found Ms Majiye to be a good witness. On a perusal of the transcript of her evidence, I can see no reason to disagree with that finding. She undermined the alibi and furnished some tentative links between the appellant and the commission of the crimes.

[27] Faced with the statement, which once admitted could not be explained away, and supported by the evidence of Ms Majiye, the trial court rightly rejected the alibi defence. In doing so, however, the court erred in one respect. It held against no 3 that he had failed to call as a witness his employer, whose identity was known, to corroborate his alibi. At the commencement of the trial counsel for no 3 informed the court that his client’s defence was an alibi: ‘Hy was in Randburg gewees op daardie stadium, op die perseel van J & J Service’. (In his evidence no 3 stated the same thing.) The state was thus aware from the beginning of the trial of the appellant’s alleged alibi and of the identity of the employer. It was not suggested by prosecution or defence that there was any difficulty in locating the employer and ascertaining whether the employment records supported the alibi for 5 July or supported no 3’s prolonged absence from home. Nor was it suggested by the prosecution that a representative from the employer would for some reason be unduly well disposed towards the appellant. In these circumstances it seems to me that it was equally open to either side to call the employer and that an inference against no 3 was not warranted. Since the onus of disproving the alibi was on the state, there is even something to be said for the view that an inference should have been drawn against the prosecution. R v Bezuidenhout 1954 (3) SA 188 (A) at 196H-197E; 226. While the error by the court below constituted a misdirection, I do not consider it to have been material in the circumstances. On the evidence the alibi was bound in any event to have been rejected.

[28] The evidence which specifically implicated appellant 4 was his statement to magistrate Boshoff, made on the day of his arrest. This implicated him in at least the robbery. His defence was an alibi, namely that he had left for Randburg on 1 June 1993 where he remained until his arrest there at the end of August. The appellant’s mother testified in support of the alibi, but she proved to be hopelessly unreliable and her evidence was rightly rejected by the trial court. The statement, once admitted, could not be explained away. No 4’s alibi was correctly rejected.

[29] In broad terms the statements of appellants 3 and 4 were similar, although inevitably some of the details varied. They admitted going to the farm in the morning (in no 3’s case, to fetch money), where the domestic worker, after locking up the dogs, allowed them in. It was only in the afternoon, however, that the deceased returned to the house unaccompanied. It was then that the appellants and third man (Daniel – accused no 1?) grabbed the deceased, tied her up and blindfolded her. Money was procured from the safe or office. It was at this juncture, the robbery completed, that the domestic worker (according to no 3, ‘Sarah’ – accused no 2?) proposed poisoning the deceased. According to no 3, the men refused and Sarah appeared to go ahead with the poisoning on her sole account. According to no 4, the men refused and Samuel (no 3?) threw the poison away; but the domestic and ‘Daniel’ (no 1?) made a second attempt using other poison. According to both appellants all three men thereupon left the premises taking with them the money which was later shared. In both statements there was mention of a ‘Martha’. If that was intended to mean accused no 5, it suffices to say that her guilt was not proved beyond reasonable doubt, which is not the same thing as saying that she was innocent.

[30] Neither appellant 3 nor appellant 4 mentioned, in their statements, the deceased’s death by strangulation. No 3 could not say whether the deceased was still alive when the men left, only that they had tied her up. No 4 could not say whether the second poisoning attempt succeeded because he did not see it – he heard about it from Daniel. On a fair interpretation he too left the deceased while she was bound. According to his statement the domestic worker said that if the deceased was not poisoned, she (the domestic) would be arrested.

[31] The trial court accepted the two statements at face value. It accordingly accepted the reasonable possibility that the deceased was still alive when the men left the house, taking the money with them, and that the deceased may have been strangled to death by no 2 after their departure. In convicting appellants 3 and 4 of murder on the basis of dolus eventualis the court pointed to a number of facts:

- that the appellants were responsible for the deceased’s captive state as part of the joint enterprise to rob;

- that when they departed, they left the deceased trussed up and helpless;

- that the appellants knew, when they departed, that no 2 was intent on killing the deceased; and

- that the appellants must have known, and therefore knew, that the deceased was powerless to resist or withstand no 2’s murderous intent.

[32] Applying R v Chimbamba and another 1977 (4) SA 803 (R AD) the trial court said that in the above circumstances appellants 3 and 4 ‘cannot in law just be allowed to wash their hands of what they now knew to be the consequence of leaving the deceased a bound, helpless captive at the mercy of a vicious would-be murderer’. The trial court said further:

‘By failing to release the deceased when they knew her death was probably imminent if she was not released, accused 3 and 4 persisted in the unlawful activity of holding deceased captive at the time when they as a fact foresaw that the continuance of that unlawful act would enable accused 2 to kill the deceased. They therefore unlawfully continued to hold deceased a captive, reckless of whether or not deceased was killed as a direct result of being held captive.’

and

‘What we are dealing with in this case is not common purpose. It is continuation in an unlawful act, to wit, the holding of the deceased as captive, when they as a fact foresaw that that was likely to result in murder.’

[33] The latter proposition assumes that an intent to kill was not part of the common purpose in the first place. It is implicit in the judgment that the trier of fact so assumed or found otherwise the reasoning to which I have referred would have been unnecessary. Perhaps the court below took that unstated view of the facts because there was no evidence to indicate that the men went armed to the farm, or that arms were used. It seems to me, however, that the facts raised two closely related issues. The first is whether the common purpose to rob was expanded, as events progressed, so as to include a common purpose to murder. If so, the second issue is whether appellants 3 and 4 effectively disassociated themselves from the expanded common purpose.

[34] The appellants purported to disassociate themselves from the murder (they refused to be part of the poisoning) but not from the robbery (they went off with the money and shared it). What had become clear to them, however, was that the robbery was developing into a murder which would be facilitated by their own prior conduct. It appears to me that by departing the scene, and leaving the helpless deceased to her probable (and actual) fate, the appellants must be taken to have acquiesced in the expansion of the common purpose unless they took steps effectively to disassociate themselves from that development. That our law recognises a defence of disassociation (in some other jurisdictions called withdrawal) is clear. S v Singo 1993 (1) SACR 226 (A); S v Nduli and others 1993 (2) SACR 501 (A); S v Lungile and another 1999 (2) SACR 597 (SCA). See too S v Nzo and another 1990 (3) SA 1 (A) at 11D-I. In the case of a conspiracy or common purpose, Gubbay CJ ventured the following dictum in S v Beahan 1992 (1) SACR 307 (ZS) at 324b-c:

‘I respectfully associate myself with what I perceive to be a shared approach, namely, that it is the actual role of the conspirator which should determine the kind of withdrawal necessary to effectively terminate his liability for the commission of the substantive crime. I would venture to state the rule this way: Where a person has merely conspired with others to commit a crime but has not commenced an overt act toward the successful completion of that crime, a withdrawal is effective upon timely and unequivocal notification to the co-conspirators of the decision to abandon the common unlawful purpose. Where, however, there has been participation in a more substantial manner something further than a communication to the co-conspirators of the intention to dissociate is necessary. A reasonable effort to nullify or frustrate the effect of his contribution is required. To the extent, therefore, that the principle enunciated in R v Chinyerere (supra at 579B and 578E) is at variance, I would with all deference, depart from it.’

[35] This court has twice expressly left open the correctness of this dictum and whether it is rule of law or a rule of thumb. See Nduli’s case, supra, at 5076; Lungile’s case, supra, at 603 para [20]. What may be gathered from our case law, however, is that not every act of apparent disengagement will constitute an effective disassociation. Compare Snyman: Strafreg (4ed) at 267-9. It appears that much will depend on the circumstances: on the manner and degree of an accused’s participation; on how far the commission of the crime has proceeded; on the manner and timing of disengagement; and, in some instances, on what steps the accused took or could have taken to prevent the commission or completion of the crime. The list of circumstances is not exhaustive. To reduce this composite of variables to a workable rule of law may be artificial, even unwise. In an article entitled ‘Accomplices and Withdrawal’ (1981) 97 LQR 575 Professor David Lanham reviewed the case law in the Commonwealth and USA. He concluded at 591:

‘While it is not possible to produce a detailed definition of withdrawal as a defence to accomplice liability, a number of principles can be extracted from the weight of authorities examined above. These principles are as follows:

1. Any withdrawal, voluntary or otherwise, which negates the actus reus of accomplice liability exculpates the accused.

2. A withdrawal which does not negate the actus reus of accomplice liability may nonetheless be defence if certain conditions are satisfied –

(a) Such a withdrawal must be a voluntary withdrawal.

(b) Whatever form the participation takes, reasonable steps to prevent the crime may exculpate the accused even if there is no countermand.

(c) Where the act of participation goes beyond encouragement, mere countermand may not be sufficient to exculpate the accused.

(d) Where the participation takes the form of encouragement (eg counsel, command or agreement) a potentially effective countermand will afford a defence even if no other steps are taken to prevent the crime. Such countermand may be expressed in words or implied by conduct.

(e) Withdrawal must be capable of being effective: a withdrawal which is untimely, uncommunicated, or misunderstood or a countermand which is not received by all principals will be no defence.’

See too SA Criminal Law and Procedure Vol 1 (3ed, Burchell) at 318-320.