On 1 January 1993, shortly after midnight, a distress flare fired by an unknown New Year's Eve reveller landed on the roof of the first respondent's cold store at K berth, Duncan Dock, Table Bay Harbour. The flare ignited the fibreglass valley gutter on the roof, which in turn ignited expanded polystyrene insulation (Kulite panels), causing a fire that spread to the refrigeration chambers and largely destroyed the cold store and its contents. The cold store was a joint venture between Portnet (30%) and Afco Holdings Limited (70%), operated by the first respondent. The building housed refrigeration chambers operating at -30°C and -60°C for storing frozen fish products. The appellants (Sea Harvest Corporation and South Atlantic Islands Development Corporation) had entered into oral contracts of deposit with the first respondent to store fish and lobster products, which were destroyed in the fire. The cold store had been designed by consulting engineers Worthington-Smith and Brouwer. Despite the city fire officer's recommendation to classify the building as "J2 occupancy" (moderate risk storage) requiring sprinkler installation, the port engineer classified it as J3 (low risk) based on industry practice, and no sprinkler system was installed. Firing distress flares in the harbour (except for emergencies) was prohibited by regulation, but it had been a regular New Year's Eve practice for at least 20 years. There was no evidence of flares ever having caused a fire previously. Mr Bell, the first respondent's engineering manager, was at the cold store checking equipment before midnight, observed defective flares being fired, waited until the firing stopped around 12:10 am, then left at 12:18 am after checking with security. The fire was discovered shortly after 1:00 am.
The appeal was dismissed with costs, including costs of two counsel. The finding of the Cape Provincial Division (King J) in favor of the respondents on the issue of liability was upheld.
For negligence to be established in delict, it is not sufficient that harm of a general kind (such as fire) was reasonably foreseeable; a reasonable person in the defendant's position must have foreseen as a reasonable possibility the general manner in which the harm actually occurred. Reasonable foreseeability must be confined to those events which fall within the parameters of reasonable possibility, not remote or speculative possibilities. Where an event is so unusual or unprecedented that it has never occurred despite long-standing similar circumstances, and where a reasonable person in the defendant's position would not have known of the specific risk factor, the defendant cannot be held negligent for failing to guard against it, even if guarding against fire in general might have prevented the loss. The test for negligence requires flexibility and cannot be applied as a rigid formula in all cases. Courts must guard against the influence of hindsight when determining what was reasonably foreseeable at the relevant time.
Scott JA made important observations about the evolution of negligence tests in South African law, noting that while the formula in Kruger v Coetzee (the "absolute" or "abstract" theory requiring foresight of reasonable possibility of harm in general) has been widely applied, courts have recognized limitations and the need for flexibility. He noted the "relative theory" advocated by Boberg (requiring foresight of the kind of harm that actually occurred) and referenced the formulation in Mukheiber v Raath, but observed that Mukheiber did not unequivocally embrace the relative theory and in fact appeared to apply both approaches in different parts of the judgment. Scott JA cited Lord Oliver's observation from Caparo Industries that attempting to state general principles applicable in infinite variety of circumstances "serves not to clarify the law but merely to bedevil its development." He noted that rigid adherence to any formula risks injustice in unusual cases and there must be flexibility to accommodate "grey area" cases. He observed that where the inquiry into culpability ends and the inquiry into remoteness (legal causation) begins - both involving foreseeability - must depend on circumstances, and in borderline cases the distinction may not be clear. Streicher JA's separate judgment discussed the traditional approach (determining negligence broadly, then applying legal causation to determine remoteness) versus the relative approach (building foreseeability of the specific harm into the test for negligence itself), noting that in this case both approaches led to the same result, thus making it unnecessary to choose between them.
This case is significant in South African delictual law for its treatment of the test for negligence and the requirement of reasonable foreseeability. It emphasizes that: (1) No single rigid formula for determining negligence is appropriate in all cases - flexibility is required; (2) Courts must guard against hindsight bias ("even a fool is wise after the event"); (3) Foreseeability must relate not just to harm in general but to the general manner and type of harm that actually occurred; (4) The reasonably foreseeable must be confined to what falls within the parameters of reasonable possibility, not remote possibilities; (5) The distinction between the "abstract/absolute theory" and "relative theory" of negligence may in practice lead to similar results in many cases. The judgment also illustrates the interplay between foreseeability in determining negligence (culpa) and foreseeability in determining legal causation (remoteness), acknowledging that these inquiries may overlap in borderline cases. It demonstrates the court's reluctance to impose liability for highly unusual events that could not reasonably have been anticipated, even where general fire risk was foreseeable and precautions against fire in general may have been warranted.
Explore 1 related case • Click to navigate