Scottish Court of Session Decisions

In August 1993, the first pursuer Mr. Ronald Dewar and his wife were living at 6 Edenbank Road, Cupar, which had been their home for many years. On the morning of Friday, 27 August, at about 8.20 a.m., there was an explosion and fire in the room where Mr. and Mrs. Dewar were sitting. Mrs. Dewar sustained severe burn injuries, from which she died on 29 August. Mr. Dewar was also injured.

There was a gas-fired boiler in the house, which was used to heat radiators and provide hot water. On 25 August, the pilot light on the boiler had gone out, and Mr. Dewar could not get it to work. On 26 August, he went to the premises of the defender, Stewart Winton, who is a plumbing and heating engineer, and asked him to come and see to the matter. That afternoon, the defender came, and replaced the thermocouple. When he left, the boiler was running, the pilot light had not gone out, and the central heating system thereafter continued to function. It was set on a timed programme, running until 10.15 p.m., and coming on again at 6.45 a.m. On the morning of 27 August, the radiators had come on and there was hot water, indicating that the boiler was functioning on its morning cycle. The explosion occurred because a gas rich mixture had built up, and was ignited when a cigarette was lit. Gas had escaped from a gap which had been created at the coupling between two pipes which brought gas to the boiler. At the coupling, there was a "union nut" on one pipe, which would hold the two pipes together by being screwed onto a thread on the other pipe. On examination shortly after the explosion, it was found that this nut was completely unscrewed, resulting in the gap from which gas escaped.

In this action, Mr. Dewar (both as an individual and as his wife's executor) and their son Mr. Murray Dewar, seek damages from the defender Mr. Winton, upon the basis that it was he who unscrewed the nut, thus causing the escape of gas and the explosion. The sheriff having found for the pursuers, and having granted decree for agreed sums of damages, the defender appeals.

Averments that the defender was in breach of certain statutory duties are not insisted in: the case for the pursuers is founded upon negligence at common law. It is not disputed that the explosion and its tragic consequences were a result of the union nut being unscrewed. And it is not disputed that the nut was unscrewed by human intervention, rather than any external factor such as vibration. Moreover, it is not disputed that the act of unscrewing the nut, and leaving it unscrewed, would be negligent. On behalf of the pursuers, it is of course accepted that it is for them to prove, on a balance of probabilities, that the defender unscrewed the nut.

In submitting that the sheriff's finding in favour of the pursuers could not stand, counsel for the defender and appellant advanced three main arguments, which may be summarised as follows. First, the sheriff stated expressly that his finding was on the basis of res ipsa loquitur: that was unsound in law, that maxim having no application in the present case; and it was moreover clear from the sheriff's findings that without recourse to that maxim, and the consequential shift of onus to the defender, the sheriff would have held the pursuers' case not to have been proved. Secondly, in dealing with the evidence, the sheriff had accepted that no person other than the defender had anything to do with the nut, either before or after the defender's departure, without taking into consideration, before reaching that conclusion, the strong evidence that the defender himself had not had anything to do with the nut: that was a fundamental error, effectively imposing an onus on the defender, or a presumption against him. And thirdly, the sheriff had made three findings in fact (40 to 42) which had no sufficient foundation in the evidence, and appeared to derive from reports which were productions but had not been proved: these findings were crucial in relation to the vital question of whether, if the defender had unscrewed the nut early on the Thursday afternoon, the joint could have remained sufficiently gas tight for the boiler to continue in operation as it had, that day and the next morning: in the absence of these findings, the sheriff's conclusions could not stand, and a finding in favour of the pursuers would not be justified.

On behalf of the pursuers, it was acknowledged that the sheriff had erred in invoking the principle of res ipsa loquitur. It was, however, submitted that in this respect, he had merely gone too far: he had apparently had recourse to that principle because the pursuers had "failed to prove the exact reason for the union nut coupling having become completely loosened" whereas his prior reasoning effectively established that it was the defender who had unscrewed the nut - which was sufficient for a finding of negligence, even if the reason for his doing so could not be discovered. In relation to the sheriff's handling of the evidence, it was submitted that the pursuers' case had always been founded upon inference, in the absence of direct evidence as to what the defender had done; and that the inference that it was the defender who unscrewed the nut could entirely properly be drawn from the exclusion of all others, such as Mr. Dewar or other tradesmen, or intruders, who in theory might have interfered with the nut. In the sheriff's handling of the matter, any shift of onus to the defender was merely a normal evidential shift, the sheriff having concluded that unless it could be shown on behalf of the defender that there was some countervailing evidence, the probability was that it was the defender who had unscrewed the nut. In relation to findings 40 to 42, it was accepted that the sheriff had apparently relied on material which he was not entitled to take into account, in passages in reports which had not been referred to. It was however submitted that in the other findings and in the evidence, there was a sufficient basis for the finding which the sheriff had made.

We do not find it necessary to examine the sheriff's reasoning in detail, as we are satisfied that it is flawed, and that his decision cannot simply be left standing. In invoking res ipsa loquitur, as a means by which the pursuers might establish negligence, the sheriff appears to us to have turned his attention from the factum probandum - that the defender unscrewed the nut - to the question of why the defender might have unscrewed the nut. Difficulty in answering that question, or indeed its possible unanswerability, inevitably bears upon the primary question of whether the defender unscrewed the nut: in the absence of any satisfactory explanation as to why he should have done so, a court might feel unable to conclude that he had. But difficulties of that kind, which might lead to the pursuers failing to discharge the onus of proof incumbent upon them, cannot be elided or overcome by invoking the principle of res ipsa loquitur. And while there is some force in the argument advanced on behalf of the pursuers and respondents, to the effect that the sheriff's reasoning had apparently brought him to the stage of considering that the defender had unscrewed the nut, his actual findings stop somewhat short of an express finding to that effect, and in any event it appears to us that the flaw in his reasoning, inherent in his reliance on res ipsa loquitur, is so central to his disposal of the issues that his decision cannot stand upon this alternative interpretation. We are moreover satisfied that the other two attacks on the sheriff's reasoning are also well-founded: if the sheriff had stopped at saying (as he does) that he considered it highly unlikely that Mr. Dewar or any other party interfered with the system after the defender left, and that the explanation that some unknown third party sneaked into the bungalow and undid the coupling appeared to him to be farfetched, we do not think that he could be criticised for reaching views of that kind, which would then require to be set against the unlikelihood of the defender himself unscrewing the nut. The evidence could then be weighed as a whole, and a decision could be reached as to whether the pursuers had established their case, on a balance of probabilities, or not. But the sheriff has not done that. Having made these observations, he says: "In short, I accept that right up to the time of the explosion no other person had anything to do with the nut coupling either before or after the defender's departure". Similarly, in his findings in fact and law, he says that the occurrence of the accident per se, "in circumstances in which the possibility of it having been caused by the first pursuer or any other third party has been discounted" gives rise to the reasonable inference that it was caused by the negligence of the defender. These passages appear to us to show that the sheriff has erred in his approach to weighing the evidence, and has reached a firm conclusion based on the unlikelihood of Mr. Dewar or others interfering, and with inferences against the defender's interests, omitting to weigh all possibilities one against the other upon the basis that it would be the pursuer who would fail if (as might happen in such a situation) none of the unlikely possibilities could be held to be probable. So far as findings 40 to 42 are concerned, we are again satisfied that the sheriff's recourse to material which was not evidence means that his decision cannot stand. While counsel for the defender and appellant submitted that it was evident that apart from these errors, the sheriff would have granted absolvitor, we do not feel able to proceed upon the hypothetical basis of what he would have done. An appeal court will of course be slow to differ from the conclusions and findings of a sheriff, as the judge who heard the witnesses, on issues of fact and indeed the inferences to be drawn from them. But we are satisfi

Before we turn to the parties' submissions upon the substantive issues, it will be convenient to consider in rather more detail certain matters which are not in dispute. Gas was supplied to the boiler by means of a horizontal pipe, which entered the casing of the boiler at the lower right front corner. At that point, this pipe was joined to a vertical pipe, by which the gas was taken to the burner of the boiler. By finding in fact 37, the sheriff has held that "The pipe joint is a brass coupling which was a machined finish fitting, the two faces of the joint having been manufactured to fit together and not require jointing paste. Once tightened, the two brass faces seal together and remain gas tight until disturbed". This finding was acknowledged by counsel for the defender and appellant to have a proper foundation in the evidence, and we were not asked to find otherwise. Further details of the elements which constituted the coupling were undisputed. At the bottom end of the vertical pipe there was a projecting "male" element, designed to fit into a corresponding female element, which was set within a short vertical element at the end of the main horizontal inlet pipe. It is the brass faces of these male and female elements which are referred to in finding 37. On the outside of the short vertical part of the main inlet pipe, (around the female part of the joint) there was a screw-thread; and around the other, vertical pipe there was the "union nut" with a corresponding thread on its inner face. The bottom end of the vertical pipe was flanged, so that when the nut was screwed onto the thread on the main inlet pipe and tightened, the male element would be brought fully home into the female element, and so long as the nut was not unscrewed, the faces of these elements could not be disturbed, and would remain sealed together and gas tight as described in finding 37. Thereafter, unscrewing the nut would make it possible for the two elements to come apart, but would not ipso facto make them do so.

The problem with the pilot light, which led to Mr. Winton being called to the house, turned out to be a faulty thermocouple; and when Mr. Winton arrived, in the early afternoon of 26 August, he was expecting to deal with a faulty thermocouple, as that was the most likely cause of the pilot light not working. He accordingly cut off the source of the gas supply into the thermocouple, by closing a valve. This valve is at the end of the horizontal pipe, immediately below the thread onto which the nut is screwed. To open or close the valve requires only a quarter turn of a quite separate nut on the side of the joint, using a small spanner. For this operation, there would be no need to interfere with the union coupling nut, but it has been found by the sheriff, and is not disputed, that if the connection at that point had already been loosened, the defender would have noticed this. In fact, Mr. Dewar had already turned off the gas at the meter, and after Mr. Winton had replaced the thermocouple, he restored the gas supply from the meter, as well (although it is not clear when) as re-opening the valve. He did not notice anything untoward about the union coupling nut at that juncture. In addition to the findings as to what he would have noticed or did notice, there is an undisputed finding that if by the time of his repair, the coupling had been loosened or completely undone by some other person, then the force used by him to open and shut the valve would have disconnected the two faces of the coupling, and the system would not have functioned. (In the terminology of finding 37, which we have already mentioned, to the effect that the two brass faces remained gas tight "until disturbed", this is effectively a finding that the opening and shutting of the valve would "disturb" the brass faces, so that they would not remain gas tight). The system in fact functioned after Mr. Winton's departure, both that evening and on the morning cycle next day.

We have already noted that it is agreed that the nut must have been loosened by human hand. Despite some variation in the evidence it is further apparent that it would take more than three, and perhaps up to six, turns of the nut to disengage it completely from the screw-thread from a fully secure position; and that for the final quarter-turn, moving from the nut being finger-tight to the union being gas-tight, a tool would be needed. The sheriff has held that the nut would have required a spanner or stillson wrench (pipe wrench) to loosen it from a fully secured position. Leaving aside for the moment an issue as to certain scratch marks subsequently found on the nut, it appears that Mr. Winton had adjustable spanners which could have turned the coupling nut, whereas Mr. Dewar in evidence spoke only to having a little work wrench, opening to about one inch, the expert evidence being that the nut was 35 millimetres across - i.e., too big for Mr. Dewar's wrench.

While unscrewing the nut from a fully secure position would thus require initial use of a suitable tool, followed by further unscrewing either by using the tool or manually, the evidence does not suggest that unscrewing of the nut could occur without such human actions except, possibly as a result of severe vibration caused by expansion and contraction of the fitting, to the extent of perhaps one eighth of a turn. As the coupling was found to be completely unscrewed, with a gap between the two pipes, shortly after the explosion, it thus appears that it must have been very nearly completely unscrewed, if not completely so, by human intervention, even if the final one eighth of a turn resulted from such vibration. The possible significance of vibration is thus not in relation to the unscrewing of the nut, but as something which might "disturb" the two brass faces, making the joint no longer gas-tight, at a stage when they were no longer held tight together by the nut, the nut having been unscrewed either completely or to such an extent that they could thus be disturbed and part in this way. Given that the nut was completely unscrewed when found, and that the brass faces could have ceased to be gas-tight, if somehow disturbed, once the nut was wholly or even almost wholly unscrewed, it is not necessary to consider just how little unscrewing from a fully secure position would be required to make such disturbance possible. But since it would apparently only be the last quarter turn of tightening which would render the joint gas-tight, it would appear that the first quarter turn of unscrewing might suffice to make it possible to disturb the brass faces so that they were no longer gas-tight. In the circumstances of the present case, however, with the nut completely unscrewed shortly after the explosion, it is at least at first sight probable that the complete or almost complete unscrewing of the nut occurred before the two faces were disturbed and ceased to be gas-tight. One important question in the present case is "How long before?" and that question has to be asked along with a related question: "How long was it before the explosion that the faces were disturbed and ceased to be gas-tight?".

Before considering the general or theoretical evidence on these matters, it is convenient to return to the question of the scratch marks which were found on the nut: this specific matter has a possible bearing upon the broader issue as to when the nut was unscrewed. The marks were photographed on 14 July 1994, quite some time after the explosion. If made after the explosion, they are of no significance. On behalf of Mr. Winton, the submission was that there was no indication that they were post-explosion; there was no reason to suppose that they were there before Mr. Winton arrived; there was undisputed evidence that they were not made by the tools he actually used at the time of the repair; and the probability was thus that they were made after Mr. Winton's visit, and before the explosion. They were thus indicative of human intervention during that period, and moreover of inept or amateurish intervention. While they did not directly point to the person who made the marks being the person who loosened the nut, they constituted evidence of an unexplained intervention, and raised the possibility that these were marks made by the "culprit" who had loosened the nut - he had in effect left his fingerprints. So far as events after the explosion were concerned, it was true that others had apparently tightened the nut, and that the possibility of the marks having occurred then could not be ruled out. But those involved had been sent by British Gas, and as one of the expert witnesses Dr. Keith Borer had said in evidence, "You would expect British Gas to send a competent person who has a competent tool bag". The marks were not likely to have been made in such circumstances. For the pursuers on the other hand it was pointed out that Detective Chief Inspector Douglas Potter, who had attended the scene shortly after the explosion, had described the nut as seen in the photographs as looking a lot cleaner than when he had seen it at the time, and had moreover said that he had co-operated with a great number of experts who came to look at the boiler, and that "these people carried out examinations on this, looked at it, may have cleaned it up, perhaps applied tools on it and it can't really in fairness be anything like it was that day". Despite Dr. Borer's optimistic view of British Gas employees, it was clearly possible that the marks had been made after the explosion. It was not a matter of attributing them to Mr. Winton; but he himself had said that they were not there when he did the work, and they would only be relevant if they had occurred, and thus pointed to human intervention, during the next 18 hours before the explosion. There was no positive indication that they had occurred then, and every possibility that they had occurred later.

We turn to the broader issues of timing. In his grounds of appeal, the defender and appellant founds upon the evidence as to the normal functioning of the boiler during the period between the completion of the repair by him and the occurrence of the explosion. It is contended that this evidence is not only unchallenged, but inconsistent with the defender and appellant having loosened the union nut at the time of the repair. On behalf of the pursuers, it is accepted that if the coupling itself, in the sense of the two brass faces, had been separated at the time of the defender's repair work, the boiler would not have functioned normally thereafter, as it in fact did. The contention for the pursuers is not that this occurred, but that the nut was loosened at that time, with the two brass faces separating later, on being "disturbed". It is plainly important to distinguish clearly between the loosening of the nut and a separation at the gas-tight coupling itself. Standing finding 37, the possibility or probability of the two brass faces continuing to provide a gas-tight seal after the loosening of the nut becomes in practical terms a question as to the likelihood of some event which will disturb them. On behalf of Mr. Winton, counsel took the position that the nut was required not merely to bring the two faces together, but to keep them together. Once the nut was loosened, the joint must be expected to part. It was accordingly for the pursuers to establish that after a supposed loosening of the nut by Mr. Winton, a period of some 18 hours could be accepted not merely as possible but as probable, with no disturbance severing the joint. It was submitted that there was virtually no evidence on this matter, and that it could not be taken as established. Some of the material contained in unproved reports touched on these matters, and had resulted in findings 40, 41 and 42. But leaving these out of account as one must, the pursuers had simply failed to prove what they had to prove. On behalf of the pursuers, it was acknowledged that the evidence was not strong. But since the nut could be loosened without severing the joint, and since some extraneous disturbance would be required to sever the joint, the length of any lapse of time was not really the issue. If the evidence pointed to the nut having been loosened by the defender, that could be taken into account, with the subsequent occurrence of the explosion effectively shifting the onus to him, to show that a disturbance could not have occurred as late as, for example, the start up of the morning cycle with some attendant vibration. While findings 41 and 42 could not be supported on the basis of the reports from which they evidently derived, they had a sufficient foundation in evidence. Finding 41 was effectively implicit in finding 37, stating as it did that if the defender had inadvertently left the coupling nut completely undone "but with the seat faces of the coupling remaining together" then the joint would have been gas-tight for test pressure purposes - unless and until there was a disturbance. And again, so long as there was no disturbance, finding 42 was plainly sound - "subsequent operation of the central heating system on the morning of the explosion would have been sufficient to have widened the joint left in that condition and thus allowed gas to escape". The start up of the system had been seen in evidence as possibly causing a small unscrewing movement by the nut itself, and could plainly result in the two faces of the coupling parting, if they had not already done so. There was no basis for envisaging an appropriate "disturbance" during the hours between the end of the evening cycle and this start up of the morning cycle; and since there was no basis for seeing the loosening of the nut as ipso facto disturbing and severing the joint itself, the only question was whether disturbance and severan

Whatever the time at which the nut was unscrewed, it seems clear, in accordance with findings 43 to 46 which we were not invited to change, that the gas rich mixture which resulted in the explosion had started to occur by about 7.15 a.m., which would be consistent with a sufficient period of functioning to supply hot water and heat the bungalow, with severance to a degree sufficient to interrupt operation of the boiler occurring within about half an hour of the boiler starting to warm up on its morning cycle. The position of the defender and appellant was again that it was for the pursuers to show that this could be seen as a possible and indeed probable set of events. Counsel for the pursuers submitted that it made good sense, if there had been no prior disturbance, and the expansion or contraction which might break the joint was to be associated, as it had been by Mr. Walker, with heat from the combustion chamber and the boiler.

One other matter which requires brief comment is the possibility that at some stage the unscrewing or partial unscrewing of the nut might be followed by some minor degree of separation between the two faces of the coupling, so that it was no longer fully gas-tight, and permitted some leakage of gas and lowering of pressure, but not enough to stop the boiler working. (Finding 40 related to this matter). When explaining how he had checked the pressure, the defender was unfortunately asked a question which perhaps produced confusion between the unscrewing of the nut, on the one hand, and the coupling itself being disconnected, on the other. However, the defender gave clear evidence that when he checked the pressure the reading was satisfactory, and in any event there does not appear to be any evidence indicating that thereafter there was any degree of leakage until, after a significant degree of severance the following morning, escaped gas could be smelled, and the gas rich mixture began to build up. On the basis that once tightened, the two brass faces would seal together sufficiently to remain gas-tight until disturbed, we do not see the possibility of slight leaks or drops in pressure as throwing any light on the present case, or in particular upon the question of when the nut may have been unscrewed.

The question of whether the pursuers have succeeded in identifying the "culprit", on a balance of probabilities, does not of course depend only on the time-related matters which we have discussed up to this point. In submitting that it had been proved that it was the defender and appellant who had unscrewed the nut, counsel for the pursuers relied inter alia upon evidence which was said to indicate that it was no one else, and thus to go towards identification of the defender by a process of elimination. Mr. and Mrs. Dewar had been in the house during the crucial period between Mr. Winton's departure and the explosion. In that sense, they had the opportunity to interfere with the nut. But the evidence showed that Mr. Dewar was scared of gas, and safety conscious; and this was confirmed by the way he had acted on the previous two days after the failure of the pilot light. He was thus unlikely to have interfered with the nut at his own hand. More particularly, the matter had been attended to professionally by Mr. Winton, and the system was thereafter apparently working satisfactorily. The Dewars thus had no cause to interfere, and in evidence the defender himself had been quite unable to suggest why they might have done so. It did not appear that Mr. Dewar had the kind of tool which would be needed to initiate the unscrewing of the nut. And finally, he had categorically denied having touched the nut, and with no reason for having been near it, could not be seen as someone who had forgotten unscrewing it, or had unscrewed it without appreciating what he was doing. He and Mrs. Dewar could be entirely excluded as possible "culprits". So far as outsiders were concerned, the first pursuer's denial that there had been any outsider at the nut during the crucial period was not to be doubted - but in any event, there would be no reason at all for calling in some other tradesman to do anything to the boiler, since it was operating satisfactorily; and the theoretical possibility that some lay outsider might have sneaked into the house, unknown to the Dewars, and while leaving no other trace of his visit might have unscrewed the nut was entirely fanciful - and all the more fanciful in the absence of any indication that there was anyone who might have done something of the kind out of malice towards the Dewars. The evidence effectively excluded the whole world, apart from Mr. Winton.

It was of course accepted on behalf of the pursuers that excluding the rest of the world was not enough. As was evident from the arguments advanced in relation to the sheriff's handling of the matter, one could not treat the evidence which excluded others as establishing that Mr. Winton was the culprit without first considering also the evidence in relation to Mr. Winton himself: evidence indicating that it was not Mr. Winton, or an absence of evidence indicating that he had, or was in any way likely to have, unscrewed the nut would have to be considered, and it might well be that taking all the evidence into account, a court might either conclude that on a balance of probabilities the culprit had not been Mr. Winton (even if no culprit could be identified or even envisaged) or simply that the pursuers had failed, on a balance of probabilities, to establish that the culprit had been Mr. Winton. None of that was in dispute.

However, it was submitted that there was evidence which, taken along with the evidence excluding others, could and should be held to have established that it was indeed Mr. Winton who had unscrewed the nut at the time of his visit. It was submitted that he plainly had the opportunity, in a very immediate sense, to unscrew the nut, since he was not merely present, and working on the boiler, but in relation to the valve had plainly been working at this very joint. Moreover, he had the ability to unscrew the nut, both in terms of the physical means (since he had appropriate tools if he chose to use them) and in terms of "know-how". It was true - and was the main problem in the case - that he not only denied emphatically that he had unscrewed the nut, but had no reason at all for unscrewing it having regard to the nature of the actual problem and his own expectation as to what the problem might be. Nonetheless, he (unlike the first pursuer) could at least be envisaged as having unscrewed the nut unthinkingly, even if unnecessarily; and he (again unlike Mr. Dewar) might have no recollection of having done something which was thus not part of the series of steps which would be appropriate and deliberately carried out. It was thus possible in relation to Mr. Winton (as it was not in relation to Mr. Dewar) to conclude that he might have unscrewed the nut, without holding him to be lying in his evidence. Moreover, and specifically, while it was accepted that he had no reason to unscrew the nut in relation to what he was doing, the possibility that he might have done so mistakenly or unthinkingly was not simply a general or theoretical possibility. During his visit, Mr. Winton had been asked by the first pursuer to change some water-tap washers, and had done so. It was at least possible that this had caused some degree of distraction or interruption of concentration on the job at the boiler. And while unscrewing the nut would have no meaning in relation to this type of boiler, there was evidence that in relation to another type of boiler, familiar to Mr. Winton, unscrewing the equivalent nut would be an appropriate step, on a job of this kind. Some quite momentary forgetfulness could therefore be pictured, with Mr. Winton doing on this machine something which would have been appropriate on such another machine. While these matters of distraction and confusion were of course in a sense entirely speculative, they could give one what counsel described as "a more comfortable feel to his making a mistake". Having regard to Mr. Winton's opportunity, know-how, means, and possible honest error, the court should hold that he was not excluded, as the rest of the world was, and that putting these whole matters into the context of the history of events and the time-related issues already discussed, the case advanced by the pursuers had been established. That was the position both upon the view that there had been an evidential "shift of onus" to the defender, and upon the somewhat different approach of simply looking at the evidence in its entirety, and deciding whether the pursuers' case had, on a balance of probabilities, been established. The court was not being asked to "choose" between Mr. Dewar and Mr. Winton or between Mr. Winton and others, or to decide which was the more probable of two alternative possibilities. It was acknowledged that the onus upon a pursuer meant that a court might conclude that the case had not been proved on a balance of probabilities, being unable to reach any conclusion. But it was submitted that in the present case the pursuers had established their case. Looking at the evidence as a whole, it was probable that it was Mr. Winton who had unscrewed the nut.

On behalf of the defender and appellant, the submission that the pursuers had not proved that it was he who had unscrewed the nut entails in a sense the proposition that it was someone else who did so; and in effect this proposition was given the more precise form of an assertion that it must have been the first pursuer, or some other untrained and unqualified person, who unscrewed the nut at some point after the defender had left, with the actual coupling becoming disconnected thereafter. This was effectively the submission made to the sheriff. But in relation to this court's consideration of the matter de novo, counsel for the defender and appellant understandably emphasised the importance of adopting the correct approach to the evidence, and to questions of onus. Not only had there been no shift of onus away from the pursuers. Even if the court was inclined to think it highly unlikely that the nut had been unscrewed by the first pursuer or some third party, it was equally or more unlikely that the defender and appellant would have done so, and the extreme improbability of his being the culprit was compounded by the substantial period of satisfactory operation after he departed, and the long interval between his departure and the time when actual disconnection of the joint apparently occurred. The alternative theories or possibilities must be looked at together, with no implication that the culprit must be the defender and appellant merely because other possibilities had been looked at first and might seem highly unlikely. On a balance of probabilities and looking at the whole evidence, it was submitted that the culprit had probably been the first pursuer or some third party; but there was no need for the defender and appellant to go so far. If both theories seemed highly improbable, it might well be that there was no sufficient basis for preferring one to the other - far less for holding that the pursuers had established, on a balance of probabilities, their positive case that the culprit was the defender and appellant. It was no doubt true that a court could usually reach a conclusion, on evidence heard, as to what had probably happened in a particular case. But there could be cases in which it was not possible for the court to reach such a conclusion, so that a defender would be assoilzied upon the simple basis that the pursuer had failed to discharge the onus upon him. A case such as this, with no real explanation as to why anyone should have interfered with the nut, and with the high degree of unlikelihood in all possible theories, was such a case: the pursuers had failed to discharge the onus upon them, and the defender should be assoilzied.

We are satisfied that the submissions made on behalf of the defender and appellant, as to the proper approach to evidence and questions of onus, are well-founded. We were referred to certain authorities, but the main issues as to the proper approach are not disputed. As regards the question of a "shift of onus", we are satisfied that this is not a case where that is a useful concept or expression. At no stage and in no sense was there a positive obligation upon the defender to establish anything; and while, after looking at some evidence, a court might consider that it proved the pursuers' case "for aught yet seen", it does not seem useful to us, once all the evidence is available, to approach matters in that way or to adopt the terminology of shifts of onus. We are also satisfied that it is right for us, in considering the matter de novo, not to consider everything "from scratch": we proceed upon the basis that both Mr. Dewar and Mr. Winton were honest, and that the facts found by the sheriff are to be taken as the basis of our own decision except to the limited extent that either party sought change.

Furthermore, we accept the submissions to the effect that on the basis of Mr. Winton's qualifications and experience, one would not expect him to be inattentive or careless, or to do something which was simply not in point for the work he was engaged upon. One would not expect him to be unaware of what he had done, either at the time or on subsequent thought, and he is categoric that he did not unscrew the nut. Leaving aside for the moment issues as to the subsequent functioning of the boiler and the lapse of time, we are satisfied that it can be said that it was "highly unlikely" that Mr. Winton would unscrew the nut on this occasion. The possibility of distraction over tap washers does not seem to us to alter that position; and while the possibility of his unscrewing the nut unnecessarily and mistakenly seems somewhat more realistic, given experience with the other type of boiler, we do not see that as materially altering the general improbability of Mr. Winton unscrewing the nut - and leaving it unscrewed.

Against such a background, great caution is appropriate before proceeding to hold the pursuers' case proved. But "highly unlikely" events sometimes occur; and when they occur, and perhaps cannot be "explained" in terms of purpose, reason, motive or the like, it may nonetheless be possible to say (as a matter of probability or indeed perhaps with great certainty) what actually happened and who did what was done. In the present case, someone unscrewed the nut. Out of all the world, Mr. Winton and apparently only Mr. Winton had been at the boiler and working at the joint, with the physical means and "know-how" to unscrew the nut, so that one can say that he could and at least hypothetically might have unscrewed it. If he had been there shortly before the explosion, then even with no "explanation" for his unscrewing the nut, it would be almost unavoidable that he would be held to have unscrewed it. The gap in time between his visit and the explosion of course increases the possibility of someone else having intervened in the interval; but we do not in the whole circumstances regard that possibility as indicating a realistic alternative to Mr. Winton as the "culprit" if the unscrewing of the nut which resulted in the explosion can reasonably be regarded as having occurred as early as Mr. Winton's visit. The gap in time between his visit and the explosion of course raises the question of whether the visit was "too early" in terms of probable causal link. But we are satisfied in the whole circumstances that if the nut was unscrewed when Mr. Winton was present, the joint itself might not separate until the following morning cycle had begun. And on a balance of probabilities we are satisfied not only that Mr. Winton could have unscrewed the nut, but also that he did. The appeal is refused.