Crown Chickens (Pty) Ltd t/a Rocklands Poultry v Rieck
During an armed robbery at appellant’s business premises, the respondent, an employee of the appellant, was taken hostage by the robbers when making their getaway. The respondent alleged that the appellant's security personnel had fired shots at the getaway vehicle, and that one of those shots had hit her in the elbow. Alleging that the security staff had acted negligently, and that the appellant was vicariously liable for those actions, respondent successfully sued appellant for damages.
The grounds of negligence were that the security staff was aware that respondent was in the getaway vehicle, and must have realised that she could be injured if they shot at the getaway vehicle. The present appeal was noted against the trial court’s findings.
Held that the trial court was correct in its findings on the issue of negligence and the appellant’s vicarious liability. The only remaining issue was whether the claim against the appellant was excluded by section 35(1) of the Compensation for Occupational Injuries and Diseases Act 130 of 1993, which provides that no action shall lie by an employee for the recovery of damages in respect of any occupational injury resulting in the disablement of such employee against such employee’s employer. The appellant relied on this section to claim protection against a claim such as the respondent’s. However, the court noted that the respondent was in fact employed by a labour broker who contracted her services out to the appellant. Thus the appellant could not claim indemnity under section 35.
The appeal was dismissed.
 The appellant is a poultry farmer in the Uitenhage district. Attached to its poultry farm is a retail shop. On the afternoon of 13 June 2003 a group of armed men entered the shop and robbed the staff and customers. Before the robbers left the shop they became aware that security personnel had been alerted to the robbery. They seized the cashier, Ms Rieck, and, holding a gun to her head, forced her to accompany them as they fled. Outside the shop they called upon the security personnel to "back off" otherwise they would shoot Rieck. Then they bundled Rieck into the rear seat of a vehicle belonging to a customer that was parked outside the shop and, with some of the robbers on either side of Rieck in the rear seat, the vehicle sped away, swaying from side to side as the wheels spun on the gravel.
 The appellant's loss control officer was in his office at the time the robbery occurred. He was alerted to the fact that a robbery was taking place and rushed to investigate. He saw three men leave the shop with Rieck and force her into the vehicle. He ran to the main access gate. As the vehicle sped past him he fired two shots from his handgun at the departing vehicle, intending to strike one of the wheels and prevent its escape. He also heard two shots being fired by one of his colleagues. The vehicle continued on its way and security personnel clambered into vehicles and gave chase.
 Meanwhile, one of the shots had struck the rear of the departing vehicle, penetrated the rear seat, and had hit Rieck on the arm. She said that when the robbers became aware that she had been shot they appeared to panic. They sped into a nearby township, stopped the vehicle, and fled, leaving her behind. Residents of the nearby houses came to Rieck's assistance and soon the police arrived and she was taken to hospital.
 Rieck sued the appellant in the South Eastern Cape High Court for damages arising from her injury. She alleged that the person who shot her acted wrongfully and negligently and that the appellant was vicariously liable for the consequences of his conduct. The action was tried by Plasket J. With the agreement of the parties the learned Judge tried only the question, whether appellant was liable for the harm that was caused, leaving the quantification of damages for later adjudication. He held that the appellant was liable to Rieck for the damages that she suffered in consequence of being shot.
 After the conclusion of the trial, but before judgment was delivered, the appellant applied to amend its plea so as to introduce a special defence that the claim against the appellant was precluded by section 35(1) of the Compensation for Occupational Injuries & Diseases Act 130 of 1993. The application to amend the plea was considered by the court below, but refused, on the grounds that the evidence that was sought to be relied upon by the appellant did not disclose a defence.
 This appeal is against both those orders, and it is before us with the leave of this Court.
 It is not disputed that the bullet that struck Rieck was fired by either of two employees in the appellant's loss control division. Although the appellant denied throughout the trial that the employee concerned was acting within the course and scope of his employment when he fired the shot, that denial has since been abandoned. What remained in issue before us – apart from the special defence that was sought to be introduced into the plea – was only whether the employee (and hence the appellant) incurred liability for the harm that he caused.
 The evidence does not establish which of the two employees fired the material shot and only one of them gave evidence. He said that he fired at the vehicle in order to stop it because he feared that Rieck might be killed by the robbers if they managed to escape. I have assumed, in favour of the appellant, that the other employee shot at the vehicle for the same reason. All the submissions that were advanced on behalf of the appellant really came down to this: it was submitted that it was reasonable to shoot at the vehicle to avoid the risk that Rieck might be killed and accordingly, so it was submitted, the conduct of the employee concerned was neither wrongful nor negligent.
 To cause bodily injury to another by a positive act is generally wrongful and will be visited with delictual liability if the actor was negligent. The positive invasion of bodily integrity falls within what in comparative English law has been described as "the range of interests which the law sees fit to protect against negligent violation", and which our law classifies as wrongful conduct. Expressed in the idiom of one variation of the general test for wrongfulness in our law, it is conduct in relation to which "public policy considerations demand that . . . the plaintiff has to be compensated for the loss caused by [a] negligent act . . . of the defendant".
 But our law also recognises that there are circumstances in which even positive conduct that causes bodily harm will not attract liability. That is so where the harm is caused in circumstances of necessity, which have been described as occurring when the conduct is "directed against an innocent person for the purpose of protecting an interest of the actor or a third party (including the innocent person) against a dangerous situation".
It is well established that whether particular conduct falls within that category is to be determined objectively.
That the actor believed that he was justified in acting as he did is not sufficient. The question in each case is whether the conduct that caused the harm was a reasonable response to the situation that presented itself.
 But while it is clear that there is no liability for harmful conduct that occurs in circumstances of necessity, and that the standard for assessing the conduct is objective, it has yet to be authoritatively determined where necessity fits in the jurisprudential scheme of delictual liability. The weight of academic opinion is that necessity operates to justify conduct that would otherwise be wrongful, thus taking it outside the class of conduct that is susceptible to an action for damages, a view that seems largely to draw upon analogous principles that have been developed in criminal law. On the other hand it also seems at times to have been suggested that it might operate instead to avoid a finding of negligence.
 It is not necessary in the present case to question the correct jurisprudential niche that is occupied by necessity in the scheme of delictual liability. Whether it operates to justify conduct that would otherwise be wrongful, or to avoid a finding of negligence, the test for whether it operates at all calls for an objective evaluation. For the classic test for negligence, as it was articulated by Holmes JA in Kruger v Coetzee, itself requires not only that the harm was foreseeable, but also that a reasonable person would have guarded against it occurring.
 Thus whatever the correct jurisprudential approach, a person who causes bodily injury by a positive act will avoid liability for the harm that he caused, on either approach, only if a reasonable person in the position in which he found himself would have acted in the same way. Considerations that are to be brought to account in determining whether the conduct was reasonable are described by Van der Walt and Midgley as follows:
"A person may inflict harm in a situation of necessity only if the danger existed, or was imminent. . . . The means used and measures taken to avert the danger of harm must not have been excessive, having regard to all the circumstances of the case. The nature of the threat, the extent of harm, the likelihood of serious injury to persons, and the value of the interest threatened must, for example, be taken into consideration. It must have been the only reasonably possible means of averting the danger. Similarly, although any interest may be protected, the interest infringed or the harm inflicted should not be greater than the interest protected or the harm prevented."
 Essentially, what is called for is a weighing against one another of the gravity of the risk that was created by the defendant, and the utility of his conduct.
As it is expressed by Boberg:
"Proportionality, in the sense of a preponderance of avoided over inflicted harm, is a traditional postulate of necessity . . ."
In short, the greater the harm that was threatened, and the fewer options available to prevent it, the greater the risk that a reasonable person would be justified in taking, and vice versa.
 In the present case there was no outward indication that Rieck would be killed once the robbers had made their escape. What presented itself to the employees who fired the shots was that Rieck had been taken hostage by the robbers as a means of enabling them to make their escape without interference. It is true, as pointed out by counsel for the appellant, that we live in times in which robbers at times kill their victims for no apparent purpose, and that there was the potential that they would do so in this case, particularly if they feared that Rieck might be able to identify them. But as the court below observed, the wanton killing of a hostage who has served her purpose has not become the norm. While the possibility that that might have occurred is not to be discounted, it was a possibility that was founded only upon what sometimes occurs, and not on any indication that this would be such a case.
 What falls to be weighed against that risk is the more immediate risk of the harm that was brought about by firing shots at the departing vehicle. And, in my view, that immediate risk was great indeed. Quite apart from the risk that Rieck might be struck by a wayward bullet, or that she might be injured if the shots caused the driver to lose control of the car, there was the even greater risk of what might occur if the shots achieved their purpose. For if the flight of the vehicle had indeed been arrested, without harm being caused to Rieck in the process, she would have been exposed to the risk of again being held with a gun to her head, while the robbers persisted in attempting to escape, and on this occasion the risk of her being killed or injured would have been a grave one. What was to happen once the flight of the vehicle was arrested (which was the purpose for which the shots were fired) seems not to have been considered at all. In my view, it was no less than foolhardy to attempt to prevent the escape of armed robbers who were holding Rieck hostage. It exposed her to very real and immediate danger, from any of a number of causes, which far outweighed the possible risk to her safety if the robbers escaped.
 I agree with the finding of the court below that a reasonable person would not have fired at the vehicle. In the circumstances the causing of bodily harm to Rieck was wrongful (on any jurisprudential approach) in accordance with ordinary principles. The harm was clearly foreseeable, and ought reasonably to have been avoided by refraining from shooting at the vehicle, and in the circumstances it was negligent to have caused it. It follows that the court below was correct in finding that the appellant is vicariously liable for the damage that was caused.
 There remains the question whether the claim against the appellant is excluded by section 35(1) of the Compensation for Occupational Injuries & Diseases Act 130 of 1993, the material provisions of which are as follows:
"No action shall lie by an employee . . . for the recovery of damages in respect of any occupational injury . . . resulting in the disablement . . . of such employee against such employee's employer. . ."
 It is not disputed that Rieck was an "employee", and that she sustained an "occupational injury", as those terms are defined in the Act. What is in issue is only whether the appellant was her "employer". The material facts in that regard are not in dispute. What is in dispute is only what is meant by that term as it is used in the Act.
 The Act has a history that stretches back over more than a century. The pre-Union statutes were consolidated in the Workmen's Compensation Act 25 of 1914, which entitled a workman or his dependants to receive compensation from his employer, in accordance with a tariff, in the event that the workman was accidentally incapacitated or killed in the course of his work. A person was a "workman in relation to work if he has entered into, or works under, a contract of employment . . ." (subject to exceptions that are not material).
On the other hand a person "having a contract of employment with a workman to perform work" was to be "regarded for purposes of this Act as the employer of that workman".
 Those definitions make it clear that a workman could have only one "employer" at any time, which was the person with whom he was in a contractual relationship of employment, whether he performed his duties for that person or for someone else. Any doubt in that regard was removed by the following additional provision:
"If the services of a workman be temporarily lent or let on hire to another person by the person with whom such contract of employment is made, the latter shall . . . be deemed to continue to be the employer of the workman, while he is working for that other person."
 Those provisions (with the addition of others that are not material) were retained in the Workmen's Compensation Act 59 of 1934, which replaced the 1914 Act. The 1934 Act was, in turn, replaced by the Workmen's Compensation Act 30 of 1941. The 1941 Act transferred the obligation to compensate workmen for workplace injuries from the employer (who until then had effectively been an insurer against workplace injuries) to a compensation fund to which employers were required to contribute. The material part of the definition of a "workman" remained substantially unchanged.
An "employer" was redefined (in form but not in substance) to mean "a person who employs a workman" (subject to certain provisos and extensions that are not material). In ordinary language that means the person with whom he has a contract of employment. Any doubt in that regard is once more removed by the express provision that if the services of the workman were temporarily lent or let on hire to another person then the employer would "be deemed to continue to be the employer of such workman whilst [the workman] is working for that other person". That was still the position at the time the 1941 Act was replaced by the Compensation for Occupational Injuries & Diseases Act 130 of 1993.
 The long legislative history of workmen's compensation in this country (at least until 1993) has thus consistently recognised that a workman has only one employer at any time (there are exceptions that are not material), which is the person with whom the workman is in a contractual relationship of employment, and that that person remains his employer even if the workman performs his services for another.
 The 1993 Act defines an employee to mean (in so far as it is now material):
"a person who has entered into or works under a contract of service . . . with an employer . . . and includes . . . a person provided by a labour broker against payment to a client for the rendering of a service or the performance of work, and for which service or work such person is paid by the labour broker."
An "employer", in turn, is defined to mean:
"any person . . . who employs an employee, and includes
. . .
(b) if the services of an employee are lent or let or temporarily made available to some other person by his employer, such employer for such period as the employee works for that other person;
(c) a labour broker who against payment provides a person to a client for the rendering of a service or the performance of work, and for which service or work such person is paid by the labour broker."
 Rieck was a party to an employment contract with a labour broker, TMS-Shezi Industrial Services (Pty). TMS-Shezi paid her salary, deducted and remitted her income tax, and made the required contributions in relation to her employment to the unemployment insurance fund and the workmen's compensation fund. TMS-Shezi, in turn, supplied her services to the appellant in return for a fee, and Rieck performed her employment duties for, and under the direction and control of, the appellant.
 The first submission on behalf of the appellant was that the relationship that existed between Rieck, TMS-Shezi and the appellant was one that is contemplated by subsection (b) of the definition of an "employer" (a relationship that involves three people: an employer, and employee, and "some other person"). It was submitted that in such a relationship, the client (the appellant) of the labour broker (TMS-Shezi) becomes the "employer" for so long as the employee's services are made available to the client by the broker. Support for that submission was sought in Clive Thompson and Paul Benjamin
South African Labour Law, in which the following assertion is made in relation to the meaning of subsection (b):
"Where an employee's services are lent or let or temporarily made available by the employer to some other person, that person becomes the employer for the period that the employee works for them."
As the court below correctly pointed out, that assertion as to the meaning of subsection (b) is not correct. The words "such employer" in subsection (b) refer back to the word "employer" that immediately precedes it, and not to the phrase "some other person". Apart from its inconsistency with the plain language of the subsection, the construction that was advanced on behalf of the appellant would reverse the position that had prevailed for over a century, for which there is no apparent reason, and would also be inconsistent with the scheme of the Act as a whole.
 It was also submitted, as I understood it, that the rationale for extending the definition of "employer" to include labour brokers was that labour brokers are not employers as that word is used in the opening phrase of the definition. It follows, so went the submission, that where such a relationship exists, as in the present case, the person referred to as the employer in the opening phrase of the definition must be the client of the labour broker. That is a most dubious construction of the definition, which falters in logic in at least two places. But that apart, the effect of such a construction would be that, for the first time in the long history of workmen's compensation, a concept of two simultaneous employers was introduced, for no apparent reason, and then only in relation to labour brokers. It would also have the effect that a person receives the benefit of being an "employer" (the benefit of an exemption from liability for workplace injuries) but no obligation to contribute to the fund that compensates for such injuries (merely because he secures the services of the employee from a labour broker). Had it been intended to introduce these startling consequences into the Act it is most unlikely that they would have been introduced merely through a process of dubious inferential reasoning. It is far more likely that the definitions were extended to include labour brokers not because they would otherwise not be "employers" but rather to avoid any misunderstanding in that regard.
 In my view, the proper meaning of the definitions in the 1993 Act (leaving aside the various extensions and qualifications that are not material to the appeal that is before us) is one that is consistent with the pattern of the earlier legislation: the Act contemplates that an employee generally has only one employer at any time, which is the person with whom he is in a contractual relationship of employment, even when he performs his contractual obligations for some other person. The appellant was admittedly not such a person and is not immunised against actions for damages by section 35. In the circumstances the evidence relied upon by the appellant did not support the proposed special defence and the application to introduce it into the plea was correctly refused.
 The appeal is dismissed with costs, including the costs of two counsel.
(Farlam, Mthiyane, Mlambo JJA and Combrinck AJA concurred in the judgment of Nugent JA).
MA Millner Negligence in Modern Law (1967) 27, referring to the "duty element" of the English tort of negligence.
Telematrix (Pty) Ltd t/a Matrix Vehicle Tracking v Advertising Standards Authority SA 2006 (1) SA 461 (SCA) para .
JC van der Walt and JR Midgley Principles of Delict (3ed) para . See similar formulations in J Neethling, JM Potgieter and PJ Visser Law of Delict (translated and edited by JC Knobel) (4 ed) at 86–7; NJ van der Merwe and PJJ Olivier Die Onregmatige Daad in die Suid-Afrikaanse Reg (6 ed) at 81; PQR Boberg The Law of Delict Vol I 787–8.
See the writers referred to in fn 3 above.
Boberg, above, at 788.
All the writers referred to in fn 3 above subscribe to that view.
See the commentary by Boberg, above, at 795ff.
1966 (2) SA 428 (A) at 430E–H.
It seems to be suggested by Neethling et al, above, at 88 fn 269, that the belief in which the defendant acted might be relevant to whether he acted negligently, but not relevant to whether his conduct was wrongful. In my view, that cannot be correct. The law judges what is reasonable according to a single standard, that is applied in the context within which the conduct occurred.
Cf John G Fleming The Law of Torts (9ed) at 129; Clerk and Lindsell on Torts (19 ed) esp paras – and –.
Above, at 788.
S 3(1): ". . . any person who has entered into or works under a contract of service . . ."
Vol 2 H1-15 para .
Contributions to the fund are assessed on the basis of earnings that are paid by an employer to the employee (see ss 82 and 83).