Silverlink Trains Ltd v Collins-Williamson

This is an appeal against a judgment of His Honour Judge Reddihough, sitting as a judge of the High Court, on 7 November 2008. In an action for damages for personal injuries by Paul Phillip Collins-Williamson (“the respondent”), the judge held that Silverlink Train Services Limited (“the appellants”) were liable to pay damages to the respondent by reason of their negligence. The judge assessed contributory negligence at 50%. The trial was on liability only.

The respondent was injured late in the evening of 11 June 2003 when, having fallen from a platform, he was struck by a train operated by the appellants, who accepted that they are responsible both for the operation of the train and of the station. The respondent was considerably under the influence of drink at the time.

At paragraph 5 of his judgment, the judge set the scene:

“There are various relevant facts which, I believe, are not in dispute. At the time of the accident, Gunnersbury Station was used, as it still is today, by trains on the North London Line and by London Underground District Line trains. The type of train operated by the defendant on the North London Line was a class 313 electric multiple unit consisting of three carriages. Each carriage has two passenger entry doors located roughly one third and two thirds of the length along each carriage. These doors are power operated sliding doors. At each end of the train is a driver's cab with a guard's vestibule adjacent to it. The door to that vestibule is a single leaf sliding door, also power operated. That door is referred to as the guard's local door. The window in that door is not capable of being opened so that the guard can look out. These trains were operated by the defendant with a driver and a guard. The driver would be in a cab at the front of the train and, for a train travelling in the direction of Richmond, the driver's position would be to the left, facing forwards and, therefore, on the side away from the platform at Gunnersbury. The guard would be at the rear of the train in the vestibule where, adjacent to the local door, were controls for the operation of the passenger doors and local door and for communicating to the driver via a bell system.”

Class 313 units are, we were told, in general use on the railway network. The appellants operated a substantial network, mainly in and around London but including services to Birmingham and Bedford.

The judge made two findings against the appellants. The first was on the basis of their vicarious liability for negligence of the train guard, Mr Sanjayan, who should have observed that the behaviour of the respondent on the platform was such that the guard should not have given the signal to start the train. The second was on the basis that the appellants’ system for operating class 313 units was defective in that, because the window in the guard’s local door did not open, he could not observe the platform, and the behaviour of passengers on it, from a position inside the train before and after giving the signal to start.

The train was moving southbound towards Richmond. The respondent fell into the gap between the platform and the train at some time before the train started and was struck soon after the train started by a shoe beam near the front of the rear carriage. The shoe, which is held in place by a wooden beam, maintains contact with the third rail to obtain power.

The train proceeded out of the station the respondent having been dragged, it is agreed, a distance of a few metres. He suffered serious injury as a result of the impact, his left leg requiring amputation below the knee.

At paragraph 6 of his judgment, the judge described the gaps and the precautions taken:

“As can be seen from the photographs of it, Gunnersbury Station is constructed on a curve in the railway track. There are two platforms immediately adjacent to each other and centrally located between the railway lines to and from Richmond. By reason of the curve at the station, [to the left in the Richmond direction] when one of the defendant's class 313 trains is stationary there, there are significant gaps between the platform edge and the side of the train. At the mid point of the carriage, the gap is some 430 mm and, at the point where the carriages are coupled, it is in excess of 1m. These gaps are typical of stations with a curved platform and are within railway industry standards. The gap at the point of the passenger entry doors is 269.5 mm, which is within the allowable maximum of 275 mm. It is agreed that the gaps between the train and the platform at Gunnersbury are sufficiently wide for a person to fall down to track level. Painted along the edge of the platforms are "Mind the Gap" notices. A train guard on these trains is also required to make an announcement to passengers on approaching Gunnersbury Station to warn them of these gaps.”

There are no staff on the platforms to assist with train despatch and that is the responsibility of the guard on the train. On over three quarters of the stations on the railway network, there are no station staff to carry out train despatch duties.

It is accepted on behalf of the appellants that, if the guard did see, or should have seen, behaviour of the respondent on the platform, variously described in the proceedings as “idiotic” and “playing the fool“, while the train was stationary, he should not have given the signal to start.

At paragraph 53, the judge described the behaviour:

“I find that the claimant was shouting, doing what is described as Ali G impressions and banging or tapping on the side of the train. By reason of the manner in which he was doing that, and because as I accept he was unsteady on his feet and stumbling, it would have been apparent to anyone who saw him that he was intoxicated or drunk. Indeed that was the view formed by those three witnesses who saw him.”

The first issue turns on a question of fact: when did that behaviour occur in relation to when the guard re-entered the train and gave the signal to start? More general questions of procedures and rules will arise on the second issue and, because the first issue turns on that question of fact, I would defer consideration of them until later.

The fact finding exercise was a difficult one. The judge had the use of CCTV photographs of a part of the platform, taken at 6 second intervals while the train was entering, stationary at, and leaving the platform. There was evidence from 4 witnesses of fact: the guard, Mr M.R. Gregory, whose evidence was confined to a written statement given to British Transport Police on 17 June 2003 because he could not subsequently be traced, Mrs S. Wileman, formerly Clarke, and Ms M. Steenkamp. The two women gave police statements in June 2003 but without having seen the CCTV photographs. A considerable time later, they made further written statements. They also gave oral evidence, and were fully cross-examined. The respondent himself did not give evidence. He had claimed to remember little. Medical evidence about him was available to the judge and included evidence that the respondent’s mental state is such that his evidence may not be reliable.

The evidence of fact has been closely scrutinised at this hearing. For the appellants, Mr Norris QC submits that, on analysis of the evidence as a whole, the judge’s finding as to when the behaviour commenced is not tenable. The judge found, at paragraph 57:

“I find that there was a relatively short period during which, as Mr Gregory stated, the claimant walked away from the train before returning to it and commencing his drunken and obvious behaviour. Thus I find that he commenced that behaviour along the side of the train, over the distance I have mentioned, at the latest at 23.42.49. It may be that Mrs Wileman and Ms Steenkamp did not see the commencement of the drunken behaviour, but clearly as it progressed their attention was attracted to it.”

If the judge was entitled to make that finding, it is, subject to contributory negligence, conclusive in that it is common ground that, at 23.42.49, the guard was on the platform and in a position to observe the behaviour. Mr Norris submitted that, on analysis of the evidence, the behaviour did not on balance of probabilities begin until after 23.42.53 when the guard re-entered the train having done his checks. (When referring to times I will from now on omit the 23).

Other timings are not in dispute. The train came to a halt at 42.31 and the doors were opened at 42.35. They were closed at about 42.46 so that the claimant must by then have alighted. The train restarted at 43.07. While the expert witnesses considered that the fall was at 43.03 the judge found (paragraph 59) that it must, on the basis of Mr Gregory’s evidence about attempting to assist the respondent, have occurred 3 to 5 seconds prior to that. That was 5 to 7 seconds after the guard re-boarded, or 10 seconds on the experts’ findings.

It is not seriously challenged, and the judge was entitled to find, that the respondent alighted from the train which subsequently struck him and that he alighted from the forward of the doors on the centre carriage. From there to where blood was found on the shoe beam, where it is agreed impact between the respondent and train occurred, was 14 metres and to the gap between the middle and rear carriage about one metre less. The judge stated, at paragraph 54:

“On the balance of probabilities the claimant was behaving in this way over a distance nearing 14 metres, that being the distance between the door from which he alighted up to the rear end of the middle carriage, which I find was the point near to which he fell. In relation to this there was evidence contained in Mr Peterson's report that blood stains were found on the front of the rear carriage.”

Mr Norris submitted that the judge has wrongly assumed that the respondent fell from the platform at or near the point where the blood was found on the shoe beam on the rear carriage. It was from that finding that the judge concluded that the respondent had walked 14 metres from where he alighted from the train.

Mr Norris submitted that the assumption is not justified and that the evidence at the trial was wholly inconsistent with the respondent having walked 14 metres. On the first of those points, Mr Norris relies on the agreement, following the expert evidence, that the gap between platform and train was sufficiently wide to enable the respondent to fall at any point between the doors from which he alighted and the point of impact with the shoe beam of the rear carriage where blood was found. Mr Metcalfe, the appellants’ expert, said that it was easily wide enough to fall through and, in the joint statement, Dr Cox, the respondent’s expert, agreed that it was sufficiently wide to allow a fall.

It was not until the closing speech on behalf of the respondent that the 14 metre walk was asserted. It had not been debated in evidence. A 14 metre walk of course enhances very considerably the probability of the behaviour observed by witnesses having commenced before 42.53, that is while the guard was still on the platform.

Mr Norris submitted that the evidence was strongly against a 14 metre walk and that there was no need for such a journey because the respondent could have fallen at any point between his point of exit and where the blood was found. There was no evidence from the witnesses that he fell at the point where the judge found he fell. Moreover, had he fallen at that point, submitted Mr Norris, Mr Gregory would have had little difficulty in pulling him out because of easier access to the gap between carriages. In his statement, Mr Gregory said:

“I reached down to help him and at this point the train started to move.”

Before considering the evidence in more detail, I make clear that I do not accept the submission of Mr Melville QC, for the respondent, that the absence of blood on the rear wheels of the middle coach demonstrates that the fall must have occurred to the rear of that point. There was more room to avoid the moving middle carriage because its wheels had no shoe beam and the gap between the train and the side of the platform was therefore greater than that at the rear coach. Moreover, there was a shoe beam on the rear wheels of the rear coach and it appears that the respondent also avoided those.

The judge’s task at the trial would have been easier had Mr Gregory been available to give oral evidence, because he was in the best position to assist, but the court had to rely on his written statement. It was accepted that the eye witnesses were each attempting to tell the truth. Mr Norris relied on Mr Gregory’s account of a very short walk:

“The male walked approximately 1 metre along the platform. He then walked back towards the train that he had just got off.”

Mrs Wileman accepted that, by reference to the CCTV photographs, some of the contents of her police statement were not correct. Mr Norris relied on the absence of any mention in the statement of the respondent walking a substantial distance. Following cross-examination of her in relation to her later statements, it was submitted, Mrs Wileman was not contending that there had been a walk for a substantial distance. Moreover, when cross-examined, Mrs Wileman said that the first time she saw the respondent was at 42.57, when a photograph shows her waving to her friends, and that she saw him just moments before he fell, which was (as the judge found) at 42.58 or 43.00. Mr Norris submitted that the evidence was contrary to a finding that the behaviour had commenced before 42.53. It indicates, to the contrary, that only a very short time elapsed between the beginning of the unusual behaviour and the fall.

As to Ms Steenkamp, she turned around as soon as she heard the banging. In cross-examination, Ms Steenkamp said that the respondent fell “a moment later” and had stumbled for “a couple of paces. I mean, just a little”. She persisted in that, agreeing that the respondent took “4 or 5 steps, but anyway it is not more than a few metres” and “then he just folded”. Ms Steenkamp said she had the respondent in sight for “a couple of seconds” before he fell.

Mr Norris submitted that the evidence is consistent only with a short distance walked and a short period of bad behaviour. It is consistent with that behaviour having occurred between 42.53 when the guard re-boarded the train and between 42.58 and 43.03 when the respondent fell. Ms Steenkamp did not see the guard when she first heard and noticed the respondent which indicates that the guard had left the platform. The judge has not sufficiently analysed the evidence of the eye witnesses, it was submitted.

Mr Sanjayan’s evidence was that he saw nothing untoward on the platform that night. He said that the platform was busy at the time. There was considerable evidence, including that from the photographs, that it probably was not busy.

The judge disbelieved the guard’s evidence. Mr Sanjayan’s belief that the platform was busy “does cast doubt on whether he was paying proper attention on the night” (paragraph 64). The judge found, at paragraph 63:

“I find that either for some reason the claimant’s presence and behaviour just did not register with him, or it did and he chose to ignore it and assume that the claimant would be safe.”

In seeking to uphold the judge’s finding, Mr Melville relied on other parts of the evidence. Mr Gregory’s account strongly suggests, it was submitted, that the respondent’s behaviour began soon after the respondent had alighted from the train. Moreover, in the midst of her other answers, Ms Steenkamp twice referred to the respondent as “walking down the platform”. When she first saw him, “he was already walking down the platform banging on the window”.

Mrs Wileman waving to her friends, as shown in the photograph at 52.57, was not inconsistent with bad behaviour having already occurred, it was submitted. In her police statement, Mrs Wileman twice referred to the respondent “walking” on the platform. In evidence, Mrs Wileman said that the respondent was “walking down the platform”.

Mr Melville relied on the total length of time the respondent could have been on the platform, at least 22 to 24 seconds if he alighted as soon as the door opened and at least 12 to 14 seconds if he alighted just before the doors closed. It is unlikely, submitted Mr Melville that, given the respondent’s apparent state of mind, the behaviour was limited to the last few seconds after the guard had returned to the train.

As to the place of fall, Mr Melville relies on the agreed statement of the expert witnesses that the respondent fell “somewhere between the rear door of the centre carriage and the leading edge of the rear carriage (inclusive of the gap where the centre and rear carriages are coupled)”. The appellants did not resile from that agreement at the hearing, though the significance of the distance walked or not walked was not appreciated at the hearing until a very late stage. At the time the engineers made the agreement, they were working on a very limited factual matrix. They do not appear to have accepted the relevance, in relation to the duration of the bad behaviour, of the distance traversed.

I agree with Mr Norris that, on a consideration of the evidence, the respondent could have fallen at any point within the 14 metre parameter. That it was possible was agreed by the engineers. I also agree with Mr Norris that, on the evidence as a whole, it is not possible to infer a 14 metre walk. However, the evidence does not, in my judgment, exclude the possibility and it permits a finding that some distance was covered.

The central question is when the bad behaviour began. Cogent though Mr Norris’s submissions were, in my judgment the judge was entitled, on balance of probabilities, to conclude, as he did at paragraph 57, that bad behaviour along the side of the train began at the latest at 42.49. That finding was not dependent on the earlier reference to 14 metres, though the judge clearly regarded that figure as material. The judge fairly relied on the inference which could be drawn from Mr Gregory’s evidence as to when the behaviour began.

The evidence of Mrs Wileman and Ms Steenkamp was, understandably, less than clear, and contained inconsistencies, but, overall, the judge was entitled to infer that the behaviour went on for a significant period of time, and for some distance, and began while the guard was still on the platform. The judge was entitled, in declining to accept the guard’s evidence, to draw the adverse inference he did from the guard’s erroneous insistence that the platform was busy. This is not a case in which I am prepared to reverse the finding of fact by the judge who heard the evidence.

That being so, the appeal on the first ground must fail. It is accepted that, if the bad behaviour began while the guard was still on the platform, he was negligent in signalling the driver to start the train and in failing to take other action.

Notwithstanding the finding on ground 1, it is necessary to consider the adverse finding made by the judge on the appellants’ methods of operation. As recorded by the judge at paragraph 70:

“The claimant asserts that a safe system for the dispatch of trains at Gunnersbury would have involved staff on the platform, or the use of CCTV monitors and/or mirrors for the guard or driver, similar to those provided for the drivers of the London Underground trains, or a local door which would remain open until the train starts moving, or a window in the local door which opens and through which the guard can look until the train starts moving.”

The “window” is the time between the guard re-entering the train and the train starting. In the present case it was 14 seconds.

The judge rejected the other allegations but found the last of them proved. He stated, at paragraph 114:

“I therefore find that it is established that to deal with the clear risk of such an accident occurring in the 10 to 20 seconds window of time in question, the defendant should have had a system whereby the guard looked out of an open window in the local door up to and just beyond when the train commenced its departure from the station. It must follow that I find that in that regard the defendant failed to measure up to the appropriate standard of care and was negligent.”

The judge’s reasoning was:

“However, it does seem to me, and I so find, that to deal with the risk of a person falling between a train and platform in the relevant window of time, it would have been very straightforward to have a system whereby the guard looks out of an open window in his local door up until or beyond the time when the train moves. In my judgment, this would require a relatively modest modification to the local door and I am unpersuaded by Mr Metcalfe's view that the Railway Inspectorate would not approve it. It is of significance that other trains such as the class 321 do have such an opening window which the guard does look out of prior to and at the time of the train's departure. Indeed, the Master Rule Book requires that to be done where such a window is available.

In my judgment, the risk of the guard's head being caught as the door was operated or of a passenger assaulting him or trying to jump in the window are low and of far less gravity than the possible consequences of someone falling in the gap between the platform and the train in the way the claimant did.”

In seeking to uphold the judge’s finding, Mr Melville relied on a basic principle in the industry:

“The principle is that trains should not be moved until it is established it is safe for the train to do so.”

The appellants had produced for general use a Railway Safety Case as required by the Railways (Safety Case) Regulations 2000. Mr Melville relied on the figure supplied, under the heading Risk Assessment for “passengers falling off platforms (including struck by train)”. The risk of fatal accidents per 100 years is put at 56.2, which means, submitted Mr Melville, an anticipated fatal accident every 1.8 years. An explanation is given, at 3.7.6.1 of the Case:

“Passengers falling off platforms (including struck by train) (TE 12A) – 56.2EF/100yrs

The figure has been derived in part from Silverlink’s own statistics. In assigning such a high risk, three other issues were taken into consideration:

•

the problems of overcrowding experienced on many of Silverlink’s stations, and;

•

the unpredictable and uncontrollable behaviour of a minority of passengers late at night, particularly those under the influence of alcohol and drugs;

•

a significant potential for fatalities arising from such an incident.”

In later paragraphs, reference is made to the comprehensive system for training of staff and the existence of competence and fitness requirements for staff, including guards. The performance of guards is to be monitored. In his evidence, Mr Sanjayan said that he had not been given training about the behaviour of customers on the platform but that “if I had seen an unusual person, or person who’s under the influence I would have spotted him”.

In his report of the incident on 26 June 2003, Mr K Peterson, driver/manager at Euston, recommended that consideration be given to “require conductors to move from the door panel to the driver window after the ready to start has been given, when despatching from off-side platforms. The conductor should remain at the driver window until the train is clear of the platform”.

Mr Peterson subsequently resiled from the recommendation and the judge did not follow it. The driver window is in a compartment separate from the guard’s and is where the driver would sit if the train was being driven in the other direction.

Reference was made to the Station Safety Performance report produced by the Rail Safety & Standards Board. It includes figures over the whole rail network. The frequency per year of passengers falling from platforms and being struck by a train is put at 5.67, and the fatal risk at 2.36. (This may be compared with 1169.67 and 6.85 for workforce assault). The appellants’ projected contribution would be a significant proportion, it was submitted. The frequency of falls between train and platform was put at 204.35. In that category, as a whole, the risk of fatal accidents is assessed as extremely small but it becomes substantial, 2.36 of 5.67, that is about 40%, when passengers who fall are struck by a train. The risk of fatal accidents is an even higher proportion (about 85%) when persons crossing the track are struck by a train in a station.

The respondent’s expert, Dr Cox, is an expert in risk assessment though he has little operational experience of railways. He stated that people falling between platforms and trains is “a recognised hazard in the industry”. Dr Cox referred to the risk of people falling between the platform and the train while the train was unobserved prior to departure. He was asked about the magnitude of the risk:

“Q. Most people on a platform, by the time of this window, will have either boarded or alighted from the train, will they not? Most people anywhere near the train will have boarded it or alighted from it.

A.

I wouldn’t agree with that. I think there are always people, latecomers, who are trying to get onto a train when the doors are closed, that kind of thing.”

Mr Melville also relied on the acceptance by Mr Metcalfe, who had very considerable experience of railway operations, that with Class 321 trains, also operated by the appellants, “the guard can stand at the window in the driving cab and easily reach and operate the controls on the door control panel in the driving cab whilst looking out of the driving cab window”. It is unsatisfactory, submitted Mr Melville, that Class 313 does not have a similar system.

It was submitted that the risk of a fall is considerable and the appellants accept the gravity of accidents of this kind when they occur. The judge rightly commented that “if such an accident does occur there can be fatal or very serious consequences” (paragraph 107). Fitting opening windows would present no serious problems, it was submitted. Mr Metcalfe accepted that it would be cheaper than CCTV.

Acknowledging Mr Metcalfe’s great railway experience and expertise, Mr Melville submitted that, in declining to follow his opinion as to the safety of the system, the court should rely on the decision of the House of Lords in Morris v West Hartlepool Steam Navigation Co Ltd [1956] AC 552. Notwithstanding evidence that there was a general practice in ships at sea not to erect guardrails in the prevailing circumstances, the decision was, by a majority, in the claimant’s favour, reversing the Court of Appeal. Lord Reid stated, at page 575:

“Here the likelihood of an accident may have been small, but at least it was sufficient to prevent the respondents from maintaining that the accident could not have happened without the appellant being negligent. And the consequences of any accident were almost certain to be serious. On the other hand, there was very little difficulty, no expenses and no other disadvantage in taking an effective precaution. Once it is established that danger was foreseeable and, therefore, that the matter should have been considered before the accident, it appears to me that a reasonable man weighing these matters would have said that the precaution clearly ought to be taken. I am therefore of opinion that the appeal should be allowed.”

Mr Norris submitted that it was not necessary, to achieve the appropriate standard of care, for doors on Class 313 trains to be modified to enable the window in the guard’s vestibule to open. The risk of someone falling from the platform during the time between the guard returning to the train and the train departing was very small. The proposed modification also carried risks of its own, it was submitted.

Mr Norris relied on the evidence of Mr Metcalfe. Asked whether he knew of any such accident, Mr Metcalfe stated:

“In all of my experience as a local manager it never happened to any of my staff. When I was Traction Train Crew Manager of the British Railways Board I had access to the daily incident logs and I never came across it then. When I was Head of Safety in the British Railways Board I never came across it then; never discussed it in any of the meetings that I attended where we were talking about the adequacy of the rules and procedures and the safety measures that the industry needs to adopt.

I have to say that since I have been a consultant I do not have access to those daily incident logs. I am not saying it does not exist, but to the best of my knowledge it is not on the industry radar, if you could call it that, as a risk that needs to be addressed.”

Mr Metcalfe was asked about another type of train, Class 150, which had a window. He stated:

“It has a window in it to enable the guard to look out. The window has benefits, there is no question about that, the guard can look out, but it has disadvantages. It is an opportunity for people to try and dive in through the window.

I know it sounds ridiculous, but I remember when I was at British Rail in about 1994 (or it might have been 1995) there was a very sad case of somebody going for a job interview. He was late arriving at Oxford Station and he tried to get in through the window. He fell under the train as the train was moving and he died. You could argue the guard can see out and stop the train, but that takes time.”

In answer to another question, he added:

“There is always this residual risk, and there are various options open to the industry, but you have to balance the pros and cons of all the different methods of course. Yes, there is a benefit in the guard being able to look out, but there are many cons associated with that.

If we think about the physical abuse that guards have in certain types of area. I have to say it is not widespread across the whole of the country, but on this type of operation at Gunnersbury it is a common occurrence I am sad to report; the kind of areas that Silverlink cover. That is with a window.”

Mr Metcalfe said that there was a trapping hazard for the guard if the door was opened while he was looking out the window. There was a door control inside the train but also one outside. Mr Metcalfe stated:

“That is a very good point. I would say the likelihood of it happening is quite low with a guard, with that kind of arrangement, but if somebody was outside, a passenger for example, and the guard is looking out of that window and a passenger attempted to join the train after the final safety check had been carried out and all the passengers doors are secured, if they knew they could go to the external device that opens that door and power it open, the risk is from somebody external to the train coming along at the last minute and attempting to open that door using an external device.”

Mr Metcalfe considered the question further:

“But what you would do is you would do a risk assessment to assess the level of risk; the residual risk of the train being unsupervised for that period of time, albeit 15 second, and in assessing that risk you would use data that exists of Silverlink services to establish the likelihood that somebody falling between the train and the platform during that short period of time, for all conditions; I’m not just saying for somebody who is in a drunken state, but for all passengers.

And so you would assess the likelihood of that event and you would assess the consequence of that event. And I would put it to you that if you were to do that it is likely, given that at the lower frequency of these events, albeit the significant consequences, that you would still fall into the tolerable area which would mean you would have to do a risk assessment to determine whether any additional controls were reasonably practicable. If you were then to do a cost benefit analysis for this additional equipment that you are limited to, you would find that it would not be reasonably practicable because the cost of that additional equipment to manage that small, very small, residual risk just will not add up.

Now if that was the case, you would find that across the railway network here in the UK, train operators would be providing that kind of equipment for services operated by a guard which plainly they had not done and the rail industry has not required them to do it, either be it through Railway Group Standards or through the Railway Safety Principles and guidance which the Railway Inspectorate require.”

Later, he added:

“Because they made their position quite clear when we were improving the design of trains through the 1980s that they wanted a train to be secure with power-operated doors, with no hazards. The objective is to try and eliminate hazards in design rather than introduce risk for the staff who work those trains. Of course, if you have a window then there is that residual risk of the guard being attacked whilst they are looking out of the window at certain stations.”

Mr Cox did not suggest an override on the internal control and it does not appear to have been explored in evidence. That would not eliminate the possibility of staff error. It would make impossible, the need, which could arise in a different type of emergency, to obtain access to the controls and the carriages from outside.

Mr Metcalfe did not cost the suggested modification but said:

“. . . it definitely would not be cheap”

He referred to the risk as a “very, very small level of risk”.

Asked about accidents of the kind that happened in this case, Mr Metcalfe stated:

“It is not an event that I personally have ever come across in all my railway career. And it is not one that I have ever heard discussed with the people who I deal with. It is definitely not on the radar for the industry. I am not saying that there has never been an incident of this nature before but the kind of incidents that you get are not of this type.”

The curvature of the platform was acknowledged but Mr Metcalfe said that there are many stations where the gap is greater than at Gunnersbury. All the measures required by HMRI’s Safety Principles and Guidance and Railway Group Standards to manage this risk were in place. Mr Metcalfe also drew attention to the requirement for approval by the company which leased the train to the appellants to any engineering design change or modification.

Mr Norris submitted that the figure of 56.2 for the risk of fatalities over 100 years for passengers falling off platforms was of limited significance because it was not sub-divided between the periods when such accidents could occur. It did not include falls as trains approached and, in particular, it did not include falls while passengers are boarding or alighting, clearly the most vulnerable period, the period when most people are involved and accidents are most likely to happen. Warnings were in place to address the risk.

In an assessment of the requirements upon the appellants, an important feature is the absence of evidence of previous accidents, over the entire network, during the period of 10 to 20 seconds between guards leaving the platform and trains moving. A few passengers may be liable to vent their frustration at not being able to board after the doors have closed and rash efforts to board trains may sometimes be made but the possibility of a prospective passenger, even if drunk or eccentric, taking action which leads to a fall between platform and train at the time when the doors have been closed and the lights put out is, on the evidence, extremely small.

The standard of care is to be set by the court, on the evidence before it. The court is not, of course, bound by an expert’s opinion or by practice in the industry, though that is a factor to be considered. I accept the careful analysis of Mr Metcalfe and that he has accurately stated practice in the industry. The risk of injury during the relevant period is extremely small. Mr Metcalfe has also referred to the risks an opening window would itself present; the possibility, albeit a small one, of the guard being trapped, the possibility of abusive approaches to him through the open window and even the possibility of a passenger seeking entry.

In my judgment, in not providing an opening window in the guard’s vestibule on Class 313 trains, the appellants have not fallen below the standard of care to be expected of a railway company and negligence in this respect is not established.

I would find in the appellants’ favour on this ground.

This court is reluctant to interfere with a trial judge’s assessment of contributory negligence. I bear in mind that the issue is now to be approached on the basis that the appellants are only vicariously liable for the negligence of their guard and not for negligent system. I am not prepared to interfere with the judge’s assessment, though it was, in my view, about as favourable to the respondent as it could be, having regard to his behaviour.

For the reasons given earlier, I would dismiss this appeal.

I agree.

I also agree.