In the paper that I wrote with Associate Professor Geoff Cary – ‘You own the fuel, but who owns the fire?’(2017) 26(12) International Journal of Wildland Fire 999-1008 we said:

This paper reviews Australian law to identify who is legally responsible for fire that starts on privately owned land. We argue that the correct interpretation of existing Australian law is: ‘Whoever owns the ignition owns the fire’ –

But simply being responsible for the start of the fire does not prove negligence and is not sufficient to establish liability for any damage done. In Burnie Port Authority v General Jones (1994) 179 CLR 520 the High Court of Australia confirmed that liability for damage caused by fire was to be assessed by the ordinary law of negligence. That requires an answer to the question ‘did the person who caused the fire take reasonable care in all the circumstances?’  Even with reasonable care a fire can be started or lost.

Newberry & Ors v Scammell & Anor [2020] VCC 1997, a decision of Her Honour Judge Morrish in the Victorian County Court is an example of the practical application of the law. In this case the defendant was towing a caravan. It was a very hot day and a total fire ban was in place. What happened is described by Morrish J at [3]-[8]:

Not long after leaving the caravan park, …there was a sudden jerk and bang.  The car veered slightly onto the wrong side of the road.  Mr Scammell was able to regain control of the vehicle and bring it to a halt.

When they got out of the car, Mr and Mrs Scammell noticed that the passenger wheel and tyre of the caravan had come off.  They then noticed that a grass fire had started.  The fire spread quickly and there was no prospect that they could put it out in the adverse weather conditions.

The fire spread to a nearby farm.  It destroyed property belonging to Mr and Mrs Newberry.

The caravan wheel and tyre dislodged from the caravan because the bearing holding it in place suffered a catastrophic failure.  This failure triggered a chain reaction, resulting in the fire…

It was Mr Scammell who, with assistance, installed new bearings on the caravan, including the bearing that ultimately failed.  Mr Scammell was responsible for the maintenance of the caravan.  It is common ground that Mr Scammell had vast experience in maintaining all sorts of vehicles including cars, caravans, motorboats and racing cars.  His son, a qualified light vehicle mechanic, assisted Mr Scammell from time to time in the maintenance of the caravan and its bearings.

The plaintiffs sued alleging that it was the negligence of Mr Scammell that caused the fire and their losses ([7]). The plaintiffs relied on inferences, that is the failure of the bearing must have been due to Mr Scammell’s errors in maintaining the bearing, and/or prior to the accident there must have been warning signs that should have alerted Mr Scammell to the pending failure and should have prompted him to stop the vehicle ([11]).  Her Honour continued ([13]-[16]):

There were no eyewitnesses to the accident.

The parties each called expert evidence about bearings and the circumstances in which they might fail.  As shall be explained later in this judgment, the experts agreed that in the overwhelming majority of cases, bearing failure results from human error in their installation or maintenance; however, in some cases, bearing failure can be due to other factors, such as premature metal fatigue.

Mr and Mrs Scammell gave evidence that Mr Scammell was highly experienced in installing and maintaining bearings, that he had installed new bearings on the caravan and regularly maintained them, and that the bearings were in good working order on every occasion that they towed the caravan.  Further, Mr and Mrs Scammell testified that there were no signs that the bearing was about to fail on this occasion.  They stated that the accident occurred without warning.

When the caravan was towed to Shepparton immediately after the accident, the bearing on the driver’s side of the caravan was examined and found to be intact and in perfect order.  Mr Scammell had installed both bearings at the same time, and he checked and serviced them regularly.

Although there was no question that it was the failure of the bearing that caused the accident and subsequent fire, the Scammell’s were not liable.  The plaintiffs had to do more than establish the cause of the fire (which was not in doubt), they had to prove that the Scammell’s were negligent. Mere assertion that there must have been a failure to properly install the bearing is not proof of any failure.  As Her Honour said (at [154]-[157];[159]-[160]:

The accepted research holds that when properly installed and maintained, bearings designed for use in the wheels of road vehicles, including cars or caravans, have an L10, or predicted life expectancy of 100,000 kilometres, or more. L10 assumes that the bearing has been properly installed and maintained.

The literature also holds that not all bearings will reach their L10 or life expectancy of 100,000 kilometres (that is to say, on the assumption that they have been installed and maintained properly). It is accepted that as many as 10 per cent will fail prior to achieving L10. The L10 failure rate follows a probability curve, so that the fewer the kilometres travelled, the less probable it is that the bearing will fail. At or before 5,000 kilometres, according Mr Enkelman’s interpretation of the probability curve, as many as 0.03 per cent will fail.

In the present case, it is not in controversy that the bearings on the defendants’ caravan had travelled approximately 4,640 kilometres before one of them failed. According to the L10 probability curve, the bearing had a 90 per cent probability of achieving 100,000 kilometres but a 99.97 per cent chance of achieving 5,000 kilometres if it was properly maintained and installed. To put it another way, the subject bearing had a 10 per cent chance of failing before 100,000 kilometres but a 0.03 per cent chance of failing around the 5,000 kilometre mark despite being properly installed and maintained. It is upon this probability that the plaintiffs seize to discharge the burden of proof. Put simply, the plaintiffs contend that because, according to the probability curve, only 0.03 per cent of bearings fail around the 5,000-kilometre mark, it is inherently improbable that this bearing would have failed if it was installed and maintained properly, ergo it is more likely than not that this bearing failed because Mr Scammell “made a mistake”.  This is the central piece of circumstantial evidence upon which the plaintiffs rely.

It must be noted that evidence based on statistical analysis or evidence that establishes that the occurrence of an event falls within a probability curve does not equate to actual proof, to the requisite degree, of the occurrence of that event. In other words, merely because the L10 probability curve shows that there is a greater than 50 per cent chance that the bearing would not have failed at the 5,000-kilometre mark if it had been properly installed and maintained, it does not mean that it is more probable than not that this bearing must have failed due to the negligence of either or both of the defendants. The probability curve is no more than one piece of circumstantial evidence that must be weighed, together with all of the evidence, in order to determine whether the plaintiffs have proved the case in negligence against each defendant…

In the present case, the evidence about where the defendants’ wheel bearing falls within the probability curve must be regarded in context.  The expert witnesses agreed that each year, “millions” of these types of bearings are imported for use in Australia.

Applying the probability curve to the figures referred to by the experts and accepting that they are speculative in nature, it is possible that each year in Australia, through no fault of any person in the installation or maintenance of bearings, many bearings will fail. According to Mr Hardy’s evidence, as many as 1,000,000 could fail before reaching L10. That is a staggering number of failed bearings – approximately 2,740 per day ((10,000,000 x 10%) / 365). As to those that could fail at or before the 5,000-kilometre mark, the number is also extremely high – 3,000 or 8.22 per day ((10,000,000 x 0.03%) / 365). Using Mr Enkelman’s evidence of around an approximate average of 2,500,000 imported bearings per year, the total potential number of bearings that could fail according to the L10 probability curve is as high as 250,000 or almost 685 per day ((2,500,000 x 10%) / 365). As to those that could fail at or before the 5,000-kilometre mark, the number is significant at 750, or over two per day ((2,500,000 x 0.03%) / 365).

At [188] Her Honour concluded:

This is not a case where negligence on the part of the defendants is the only possible explanation for the failure of the bearing. It is not in controversy that a bearing can fail soon after installed even if installed and maintained properly. The question is whether the plaintiffs have established that this bearing failed because of the negligent acts or omissions of the defendants. I am not satisfied that merely because the L10 probability curve predicts that the probability of bearing failure absent human error in their installation and maintenance is extremely low, that suffices to fill the gaps in the plaintiffs’ case proving the negligent acts or omissions alleged against each defendant.

As to the presence of warning signs, Her Honour said ([190]-[192]):

According to both Mr and Mrs Scammell, the accident occurred without warning.  There was nothing to suggest that the bearing was about to fail.

Neither Mr Scammell nor Mrs Scammell noticed any smoke emanating from the caravan.  No other witness gave evidence about seeing smoke coming from the caravan.  Neither Mr nor Mrs Scammell heard any noises prior to the accident which might have alerted them that something was wrong.

If I were to put the evidence of Mr and Mrs Scammell to one side, there is no satisfactory evidence to establish that: (a) there must have been smoke, and (b) that even if there was, Mr Scammell must have been in a position to notice it.  Similarly, there would be no evidence to establish that either there were other signs that the bearing was about to fail, or that they were of such magnitude that Mr Scammell ought to have noticed them.

Further (at [194]):

Even if there were such signs, there is no evidence that they would have been apparent in sufficient time to apply the brakes to bring the vehicle to a halt and thereby prevent the fire from igniting.  Moreover, Mr Hardy testified that the action of braking causes additional friction and heat to build up inside the wheel hub. There is no evidence that applying the brakes would have made a difference to the impending catastrophic failure of the bearing or that it would have prevented the fire.

Finally (at [196]) ‘on this hot, windy fire ban day there was nothing that either Mr or Mrs Scammell could have done to extinguish this rapidly spreading fire’ once it had started.

Her Honour held (at [197]) ‘The plaintiffs have failed to discharge the burden of proving that either or both of the defendants were negligent’.


The decision of the County Court is not a binding precedent and this case did not develop the law. As I suggested at the start it is a practical demonstration of the effect of the law.  In most general terms it shows that just because a person suffers losses it does not mean anyone is necessarily liable to compensate for those losses even where, as in this case, the cause is clear. It was the Scammell’s caravan that caused the fire that burned out the property owned by the Newberry’s. The Newberry’s were clearly not at fault but neither were the Scammell’s so the loss lies where it falls, that is whoever suffered the loss (the Newberry’s) has to meet the costs.

At a more focussed level, it also shows that liability for fire requires proof of negligence not merely the source of the fire.  In January 2020 a helicopter operated by the Australian Army started a fire in the Namadgi National Park in the Australian Capital Territory (Michael Inman and Craig Allen ‘A Defence chopper sparked Canberra’s Namadgi bushfire, but its crew didn’t tell authorities the location for 45 minutesABC News (Online) (30 November 2020)). The fact that the ignition source is known will not prove negligence in any claim that homeowners may chose to make. They will need to prove that the actions of the pilots when landing, or in how they responded once the fire started, were negligent. Simply proving how the fire started will not be enough.

That conclusion is true in any fire case including where people want to blame fire services for escaped backburns, that is fires lit to try and remove fuel from an oncoming fire and where conditions are clearly not ideal for lighting fires. Merely proving how the fire started does not prove negligence.


People who suffer losses due to fire have to do more than prove how the fire started, they have to prove that there was a defendant who was negligent. Merely proving how the fire started is not sufficient.