These actions arise as claims by the reclaimers for indemnity against various contractors who were employers of persons killed or injured as the result of the destruction of the Piper Alpha platform by explosion and fire. The indemnity sought is based on contractual provisions. Because of the terms of these provisions, to which I shall turn in detail in due course, in order for the reclaimers to succeed it would have to be established that the disaster arose through some fault or breach of statutory duty on the part of the reclaimers which made them liable to make reparation to the victims. On any view of the relevant indemnities they do not take effect if there was sole fault or wilful misconduct on the part of the reclaimers, and accordingly the reclaimers sought to establish a cause of the disaster which involved both negligence for which they are responsible and also negligence on the part of a third party. The reclaimers’ case on the facts was that the initiating cause was the explosion in module C of a cloud of flammable material which had leaked from the site of a pressure safety value PSV 504, which had been removed for recalibration. The pipe leading to that site should not have contained any condensate or gas and the end of the pipe at the site should have been sealed with a blind flange, which should have prevented any leak. The reason why the pipe contained the condensate was said to be that the lead production operator, Robert Vernon, an employee of the reclaimers, repressurised the condensate injection pump from which the pipe led to the site at a time when he should not have done so because of the absence of the PSV. The reason why there was a leak was said to be that the blind flange had not been properly fitted when the PSV had been removed and that this was the fault of the fitter, Terence Sutton, who was an employee of Score (U.K.) Limited ("Score"). Accordingly, there was fault on the part of Vernon who should have been aware of the absence of the PSV, and also fault on the part of Sutton who should have fitted the blind flange properly. The respondents’ case on the facts, which was maintained before the Lord Ordinary, was that the initial explosion occurred in module B for unknown reasons. They now accept that the initial explosion occurred in module C, but say that the reclaimers’ hypothesis that the source of the leak was the site of the PSV is so bedevilled by improbabilities that it cannot be said to have been proved. It should perhaps at this stage be noted that the respondents’ contention at the Cullen Inquiry was that the most probable source of the leak was indeed the site of PSV 504, but of course they are entitled to change their position on that.

The Lord Ordinary has set out in his Opinion the general background and a detailed description of the machinery and the processes involved in the operation of the platform, and I gratefully adopt that part of his Opinion as set out in the agreed Appendix. I shall deal first with the factual issues relating to the cause of the disaster and then deal with the major legal issues arising therefrom.

The respondents advanced certain legal propositions which they said should be taken into account in evaluating the facts in this case. In the first place reference was made to Benmax v. Austin Motor Company [1955] A.C. 370 for the proposition that once the primary facts have been established, an appellate court is in as good a position as the trial judge to draw inferences from those facts. Accordingly, while due regard should be had to the Lord Ordinary’s conclusions, this court should draw its own inferences from the evidence rather than look to see if the Lord Ordinary’s inferences are such that, on the evidence, no reasonable Lord Ordinary could have come to such a conclusion. I entirely accept that where the primary facts are not in dispute an appellate court may well be in as good a position to draw inferences as the trial judge, but this does not appear to me to be such a clear case. The advantage which a Lord Ordinary has in seeing and hearing the witnesses is not confined to issues of credibility. He also has to consider reliability and a view on reliability can often be formed by an overall impression of the manner in which a witness gives evidence. Further, in the present case I am very conscious of the fact that we have not been referred to very substantial tracts of evidence which the Lord Ordinary would be entitled to take into account in his overall assessment of reliability. This does not of course mean that the Lord Ordinary’s inferences must all be accepted uncritically, but one of the difficulties in this case is that it is only in rare instances that primary facts can be said to be established unequivocally. There is therefore more scope than usual for drawing inferences from the body of evidence as a whole and the Lord Ordinary has the advantage over us that he heard all of the evidence and was able to deal with individual chapters of evidence within the broader context. Accordingly, I consider that we are not entitled to put aside inferences drawn by the Lord Ordinary just because we might have felt able, on the limited information before us, to come to a different conclusion, but can only do so if we are satisfied that the criticisms of the Lord Ordinary’s conclusions are justified and thus the matter becomes open for us to draw our own inferences. On this matter I would only add that I do not derive much assistance from the observations in Mackinnon v. Millar 1909 S.C. 273, which was founded on by the reclaimers. That was an appeal from the decision of an arbiter under the Workmen’s Compensation Act 1906 where an appeal could only be on a matter of law and the arbiter’s decision on the facts was equiparated to the verdict of a jury. Accordingly, the court in that case would have to accept the inferences drawn by the arbiter unless it could be shown that these were wholly untenable.

The next point taken is that the court should not blame the dead without strong evidence, and this is relevant to the cases made against both Vernon and Sutton. This proposition is based on what was said by Viscount Simonds in M’Williams v. Sir William Arroll and Company 1962 S.C. (H.L.) 70:

It is clear, however, from the Lord Ordinary’s Opinion (pages 214 and 855) that he had this very much in mind, and where the evidence clearly satisfies the court that such an inference should be drawn the court cannot shirk its responsibility, however distasteful that may be.

The next proposition is that where material evidence is missing, the court should not simply decide the issue on the basis of which of two or more hypotheses advanced in the proof is the most probable, but must have in mind the possibility that had all the evidence been available a further hypothesis might have proved to be the correct one. Even after a very long and exhaustive proof the court might still find itself in the position of not being able to say that any hypothesis has been proved on a balance of probabilities, in which case, of course, the pursuer must fail. This proposition is based on Rhesa Shipping v. Edmunds [1985] 1 W.L.R. 948. In that case the trial judge considered that he had to make a choice between an extreme improbability and a virtual impossibility and selected the less improbable. It is easy to see how it can be held that a pursuer cannot prove his case on a balance of probabilities by relying solely and exclusively on an extreme improbability. I can also well understand that where a pursuer seeks to prove a particular hypothesis for which there is no direct evidence on the basis of exclusion of all other possibilities, this will be difficult, if not impossible, where many vital witnesses are dead and the real evidence lies at the bottom of the sea, as had the evidence from these sources been available it might have disclosed that one of the alternative possibilities, or even an unanticipated possibility, was the correct one. The Lord Ordinary in the present case, however, has held that despite undoubted difficulties attending certain aspects of the evidence, nevertheless reviewing the whole evidence, he was satisfied that the reclaimers’ hypothesis was the probable explanation. Accordingly, he has not based his conclusion on a balance of improbabilities as was done in Rhesa Shipping. He had clearly had in mind the strictures of the respondents as he makes clear at pages 214-5.

The respondents then founded on what was said in Dickson on Evidence where, in paragraph 108, certain general rules are set out in relation to circumstantial evidence. In particular it is said that

The respondents maintained that on a number of crucial issues the Lord Ordinary had failed to appreciate that the evidence not merely failed to support the reclaimers’ hypothesis but positively destroyed it. In any event the number of inconsistencies even if not individually fatal diminished the proof of the reclaimers’ hypothesis well below the balance of probabilities. I accept, of course, as did the Lord Ordinary, the principles set out in Dickson. The effect of the application of these principles will depend on the precise evaluation of the evidence. I would also accept that there is no direct evidence that Vernon jagged the pump or that Sutton fitted the flange improperly and therefore in respect of these crucial matters this is truly a case of circumstantial evidence.

Finally, the respondents contended that in relation to a number of important issues the witnesses who did give evidence were not asked by the reclaimers questions which would have elicited direct evidence and which would have obviated the necessity to draw inferences. As an example Grieve was not asked if, during the critical time when he was present at the 68 foot level Vernon would have had the opportunity to jag the pump without him noticing. The respondents maintained that by reason of their failure to ask the critical question the reclaimers are not entitled to ask the court to infer that Vernon did have the opportunity when Grieve was in a position positively to confirm or deny the proposition. Reference was made again to Dickson paragraph 108 where it is said:

Again I would accept what is said there but it is to be noted that the context is the failure to call a witness who could give crucial evidence. Where the witness does give evidence it is of course open to both parties to question him. It was not suggested by the respondents that there were any witnesses who could have given crucial evidence who were not called by the reclaimers. Where a witness does give evidence, but is not asked a particular question which at the end of the day turns out to be an important omission, that may be a matter of comment but only in the most extreme case would it be fatal.

In considering the cause of the disaster certain matters were clearly established on the evidence and certain matters have been found as a fact by the Lord Ordinary and are not now challenged by the respondents. It is not now disputed that the initial explosion occurred in the eastern half of module C. Earlier in the day PSV 504 had been removed by Score employees Rankin and Sutton, and the relief line from condensate injection pump A had been fitted by Sutton with a blind flange at the site. The flange fitted by Sutton would have prevented any leak of gas or condensate from the pipe even under full system pressure if it had been properly tightened with the use of combination spanners. Scaffolding had been erected to enable the valve to be removed and this was in position at all material times. Because of intended planned maintenance, pump A had been depressurised and electrically isolated and pump B was running. During the course of the evening pump B tripped. This would not immediately affect oil production from the platform but after about 15 minutes, because of the loss of lift gas, those wells which relied on lift gas would reduce their flow causing an overall loss of oil production of about 20%, and after about an hour, because of the build up of condensate, the whole production process would have to be shut down. Such a trip was not uncommon but usually a pump which tripped could be restarted immediately. On this occasion initial attempts to restart the pump were unsuccessful. Vernon expressed the intention of bringing pump A back into operation and checked with Alexander Clark, the nightshift lead maintenance hand, that no work had yet been done on the pump by way of maintenance. He also arranged with Clark that the pump would be electrically deisolated. The appropriate "red tags" were signed by Vernon and Clark in order that the electrical deisolation could proceed. Vernon was at the site of the condensate injection pumps on the 68 foot level in the minutes before the explosion. If the pump was going to be brought back into operation it would first be necessary to repressurise it and this would be done by a procedure known as jagging. Jagging would produce a flow of condensate under pressure into the pump and thence into the relief line up to the blind flange. The flammable mass necessary to cause the explosion must have resulted from a leak of condensate rather than gas. The pattern of gas alarms spoken to by Bollands, the control room operator, was that there was a low level alarm at zone C3, followed about two minutes later by a flurry of alarms being low level alarms from C2, C3, C4 and C5 and a high level alarm from the area of one of the centrifugal compressors. This is at least consistent with a two-stage release. Because of the direction of wind flow it is also consistent with a leak from the general area of PSV 504. A fireball was seen immediately after the explosion by Grieve in the ceiling of the 68 foot level of a nature and in a position which would be consistent with a leak of condensate from the site of PSV 504.

The combination of the evidence indicating that repressurising pump A could produce condensate at the site of the blind flange in sufficient quantity to produce a flammable mass in the south east quadrant of module C and the evidence indicating that Vernon’s intention was to restart the pump which would initially involve repressurising it would, if it stood alone, give formidable justification for the inference drawn by the Lord Ordinary that the reclaimers’ hypothesis was proved, particularly as there was no evidence to suggest that any other work had been done that day which would affect the integrity of the machinery or pipework in module C. The respondents, however, maintain that when the individual ingredients of this evidence are looked at critically they contain so many improbabilities that the reclaimers’ hypothesis cannot properly be held to be proved.

Certain crucial issues are still very much in dispute. These are: (1) What was Vernon’s state of knowledge as to the absence of the PSV? In terms of the indemnities they will be elided if there has been "wilful misconduct" which is defined as being "intentional and conscious disregard for good and prudent practice". Accordingly, if Vernon knew that the PSV was missing and in that knowledge deliberately repressurised the pump, it is accepted by the reclaimers that this was contrary to prudent and proper practice and would constitute wilful misconduct thus eliding the indemnities. (2) Did Vernon in fact repressurise the pump? (3) Was a leak from the site of PSV 504 consistent with the pattern of gas alarms spoken to by Bollands the control room operator? (4) Did Sutton fail to tighten the flange properly?

The respondents criticised the Lord Ordinary’s approach to this issue, as indeed to other issues, on the basis that he had failed to make findings in fact on crucial matters, and they invited this court to make innumerable findings. The difficulty with this approach, however, is that the Lord Ordinary was faced with a situation where many of the crucial witnesses had not survived the disaster and accordingly there were substantial gaps in the evidence. He was therefore fully entitled to refrain from making positive findings where he was not satisfied on the limited evidence available to him that he would be justified in doing so.

The Lord Ordinary’s approach to Vernon’s state of knowledge was that he was entirely satisfied on the evidence that deliberately to restart pump A, knowing that the PSV was missing, would be something which no lead operator would ever consider doing and therefore such knowledge cannot have been in Vernon’s mind when he repressurised the pump. (It should be said that before the Lord Ordinary one of the respondents’ primary contentions on the matter of causation was that because no experienced lead production operator, which Vernon was, would ever consider doing such a thing it followed that he could not have jagged the pump and therefore the reclaimers’ contention as to the cause of the explosion could not be correct). As this was a factor which the Lord Ordinary considered to be of major importance it is necessary to look at the evidence about it. The evidence on this matter came from four experienced operators, Grieve, Bolland, Murray and Henderson and from the maintenance superintendent, Todd, and the lead maintenance hand, Clark. Grieve (page 9354) was asked:

Bollands (9480) was asked:

At 9564 he was asked:

Murray (15143) was asked:

Henderson (11817) was asked:

In cross-examination (11880) he was asked:

He was then referred by the cross-examiner to the General Safety Procedures manual, section 16, part 3, paragraph 4 which reads:

It was suggested to him that this paragraph contemplated the operation of the system under pressure when a PSV was removed and he was asked if he could explain what is meant by this paragraph. His reply was:

His final position on the interpretation of paragraph 4 was:

I note at this stage that most pipes on the platform which required pressure safety valves were fitted with two such valves, and it may be that the procedure referred to related to the situation when one of two valves had been isolated. Todd, the maintenance superintendent, said that he would not have allowed a condensate injection pump to be run when the PSV had been removed (11145). Clark’s evidence (10016) was that as a maintenance hand if he knew that the valve had been removed he would never have started the pump, although it should be said that the reason he gives is because the pump would have been out of commission plus the fact that there would have been an electrical isolation from another source which he did not have the authority to sign off. While the cause of the tripping of pump B was never ascertained, the evidence in general was to the effect that although these pumps tripped on a fairly regular basis they could usually be restarted without much delay. There was no reason therefore to suppose that pump B could not have been restarted within the hour or so which was available before all production would have to be suspended. Accordingly the situation was not so dire that an operator would even contemplate bringing pump A into operation knowing that its PSV was missing. Further, as Bollands pointed out, it is highly unlikely that Vernon would contemplate repressurising knowing that the PSV had been removed without even checking that a blind flange had been fitted and without pressure testing the pipework to ensure that there would be no leaks. It appears also from Henderson’s evidence that if a pump was being brought back into operation after maintenance involving breaching the integrity of the pipework the practice was to station an operator at flanges to check that there were no leaks when jagging commenced (11771). This would tend to confirm that Vernon did not jag knowing that the PSV was missing as he appears to have made no check at all of the site of the PSV even to be satisfied that a blind flange had been duly fitted let alone pressure testing or stationing an operator to check for leaks. In my opinion the evidence relating to this matter was of sufficient cogency as to justify the Lord Ordinary in concluding at the very least that it was a factor of very great importance in his exercise of weighing the various factors pointing to Vernon’s state of knowledge at the time he repressurised the pump. The question therefore becomes whether the Lord Ordinary was justified in holding that that factor, virtually alone, was sufficient to outweigh all the other factors which the respondents maintained would point to Vernon knowing that the valve was missing when he repressurised the pump.

The respondents advanced several factors which they maintained indicate knowledge. (1) ‘One of the primary functions of a lead production operator is to know the status of all plant as one item out of action could affect decisions on applications for permits to work during his shift.’ Vernon was however aware that pump A was handed over to maintenance and not available for production and accordingly he might not, initially at least, be likely to concern himself with that pump or its associated valves.

(2) ‘At handover Flook, the day shift lead production operator, knew that the PSV had been removed as he had issued the permit and would be expected to record the fact in his notes and mention it to Vernon when he handed over.’ Flook did not survive the disaster and accordingly there was no evidence from him. Again, it may not have been recorded or referred to by Flook at handover because the pump was not available for production.

(3) ‘The control room copies of the permit relating to the PSV if suspended would be on the lead production operator’s desk and Vernon would be expected to look through suspended permits to ascertain the state of the plant.’ However, the evidence was that permits which had been suspended, as opposed to live permits, might only be referred to at a handover if they were relevant or significant (Henderson 11625, Lynch 11182).

(4) ‘If the permit was suspended during or after the handover it would be Vernon himself who dealt with the suspension. If Rankin’s timings are correct the person he would be likely to have seen in the control room would almost certainly have been Vernon, in which case, of course, Vernon would have had direct knowledge of the absence of the PSV.’ The Lord Ordinary’s approach to this aspect of Vernon’s knowledge is not always entirely consistent. At page 424 of his opinion he says

At page 748 he says

In the concluding paragraph of this section of his Opinion the Lord Ordinary says (752):

In the section of his opinion dealing with his conclusions on Vernon’s knowledge he says (774):

The respondents maintain that having regard to these observations there should have been a clear finding by the Lord Ordinary that the permit was in fact suspended and that it was Vernon who suspended it. Accordingly, it could not be said that at all times Vernon was ignorant of the fact that the PSV had been removed. The respondents maintain that instead of making such a finding the Lord Ordinary goes on to contradict what he has previously said. At page 775 he says:

At page 778 he says:

In his final conclusions at 1122 he says:

The position therefore is that it is not clear whether the Lord Ordinary concluded that Vernon was at all times ignorant of the absence of the PSV or that he had been made aware at the beginning of his shift that it had been removed but had forgotten by the time he came to repressurise the pump. His final observation on the matter at page 1448 is:

The Lord Ordinary’s final position therefore seems to be that he is unable to come any definite conclusion one way or the other. The respondents maintained that on the evidence the Lord Ordinary should have made a positive finding that the permit was suspended, that this happened at about 6 p.m., that the person who suspended it was Vernon, and that accordingly Vernon was in fact aware of the absence of the PSV. The reclaimers point out, however, that the time at which the permit was suspended, if it was suspended, depends on Rankin’s evidence and the Lord Ordinary did not find Rankin’s evidence on this matter acceptable. In answer to this the respondents say that the Lord Ordinary’s non-acceptance of Rankin was based on the content of his evidence and not on any criticism of his credibility or reliability based on seeing and hearing the witness. Accordingly, this court is able to consider the content of Rankin’s evidence and come to its own conclusion about its reliability. It appears to me clear, however, that because of the appalling experiences suffered by the survivors of this disaster the Lord Ordinary was understandably reluctant to make any adverse comments about surviving witnesses unless he was absolutely compelled to do so and I am not prepared to assume that the manner in which Rankin gave his evidence was something which had no bearing on the Lord Ordinary’s consideration of it just because the Lord Ordinary does not specifically say that he was founding on the benefit which he had in seeing and hearing the witness. In any event a reading of Rankin’s evidence does show some curious features, in particular the fact that he has no recollection of being in the control room in the morning when he must have obtained a lead production operator’s approval of his permit, and his equivocal reply on the critical issue that "My understanding is that jointly we suspended the permit", although he appears to have no recollection of precisely what was done (page 11485). In the circumstances I do not find it difficult to agree with the Lord Ordinary that no firm conclusion can be reached on the basis of Rankin’s evidence. I shall have to return to this matter later for reasons that will become apparent.

(5) ‘In accordance with normal practice Vernon would walk round the platform either immediately before or shortly after the handover in order to check on what was happening. In the course of that inspection he would see scaffolding which would draw his attention to the missing PSV.’ Clark’s evidence, however, was that there was always a lot of scaffolding, so much so that he eventually took no notice of it. While it is true that a lead production operator would be expected to be more interested in the purpose of any scaffolding than would be a lead maintenance hand, the evidence does not, in my opinion, go so far as to indicate that Vernon would inevitably have had his attention drawn to the absence of the PSV by the presence of scaffolding if he had not been informed of its removal, although I would accept that if he had been so informed the presence of the scaffolding would have acted as a reminder.

(6) ‘When pump B tripped Vernon should have applied his mind to any factors which would affect the restarting of pump A.’ This argument may be valid if Vernon was originally aware of the absence of the valve but had forgotten about it. But if he was never made aware that the PSV was missing there was no reason for him to think that it might be.

(7) ‘Richard, the phase 1 operator (who did not survive the disaster), should have known that the PSV was missing as he would have learned of this at handover and, in any event, as his place of work was module C he would have had ample opportunity to find out. Richard was with Vernon at the condensate injection pumps at the material time immediately before the explosion and could be expected to tell him about the absence of the PSV when Vernon indicated his intention to bring pump A back into operation.’ As neither Richard nor Grant, who handed over to him, survived, again this can only be surmised. There is no doubt that Grant was aware that the valve had been removed as he was present at its removal, but again there is the possibility that because pump A had been handed over to maintenance Grant would not regard the removal of the PSV as a relevant or significant feature worth mentioning at handover. As there is no indication in evidence that Richard did draw the absence of the PSV to Vernon’s attention, this may be a pointer to Richard not being aware himself. It is also of significance that while he was with Vernon, Richard received a message from Bollands that a detector in zone C3 had indicated a gas leak and Richard went off to investigate. Richard would be well aware that zone C3 was downwind of the site of PSV 504 in module C. If condensate was being introduced into pump A and if Richard was aware that the PSV was missing it is unlikely that he would not have associated these two facts. While there is no direct evidence that Richard told Vernon where he was going and why, it is again improbable that he would not have told him. Vernon was, after all, the lead production operator and responsible for safety on the platform and it is unlikely that Richard would keep to himself the not unimportant information from a safety point of view that a gas alarm had annunciated. If Vernon had earlier been made aware that the PSV had been removed, but had forgotten, any information about a gas alarm would have forcibly reminded him and it is inconceivable that he would have continued to repressurise the pump in that situation. This, accordingly, in my view is, if anything, another factor militating against Vernon’s knowledge of the absence of the PSV.

(8) ‘Whoever unloaded and recycled the reciprocating compressors would have been in the immediate vicinity of and would have seen the scaffolding at the site of the missing PSV.’ However, as it is not known who unloaded the compressors this cannot be a pointer to knowledge on the part of Vernon. In any event the control panels from which the compressors would be unloaded and recycled were situated at the north side of the module and separated from the site of the scaffolding on the south side by the substantial bulk of the compressors.

(9) ‘Before restarting the pump the manual isolation valve which was situated within inches of the missing PSV should have been opened.’ The respondents rely on the evidence of Henderson who gives a sequence for restarting a pump, but his sequence is on the assumption that there are two operators available as well as the lead production operator. It was accepted by others that there was nothing unsafe in leaving the opening of the manual isolation valve until after jagging, so long as it was opened before the pump was actually restarted. It is to be borne in mind that until the pump is started the system pressure in the relief line after repressurisation would not exceed 670 psi. When running the pressure in the relief line would normally be 1100 psi. At 1700 psi the pump was designed to trip and it would only be if the pump failed to trip and the pressure reached 1750 psi that the PSV would open and the manual isolation valve would require to be opened for that purpose. Accordingly, the opening of the manual isolation valve would clearly not be a priority at the stage of jagging. The respondents further contend that even if it was not opened at the stage of jagging, the fact that it would have to be opened before the pump was restarted would be in the mind of Vernon and that would have reminded him if he had forgotten that the PSV was missing. I shall deal with this argument later.

(10) ‘As Clark’s evidence was that the maintenance permit for pump A had never been issued, the red tags signed by him and Vernon must have come from the PSV permit (a blue permit). That being so, Vernon must have been aware of the PSV permit.’ This contention however does not square with the evidence of Bollands, who says that Vernon took a pink permit out of the box and that permit had red tags attached (9470/7). Where the evidence of Clarke and Bollands is in conflict, the Lord Ordinary has preferred the evidence of Bollands. Further, Bollands was not cross-examined on this issue.

In this state of the evidence the Lord Ordinary in my opinion was entitled to take the view that while a number of matters could be described as possibilities, the state of the evidence was such that he did not feel able to make positive findings and having regard to the evidence I consider that he cannot be criticised for taking that approach. One firm and clear fact accepted by the Lord Ordinary was that no responsible lead production operator would knowingly start the pump in the absence of the PSV.

There is, however, one aspect of the case on which I find it difficult to support the Lord Ordinary’s reasoning. While he accepts that there is insufficient evidence to warrant a conclusion that Vernon was ever made aware that the PSV was missing, he also appears to have considered that even if Vernon was made aware at about 6 p.m. nevertheless the matter had somehow slipped his mind by the time he came to repressurise the pump. The presence of the scaffolding immediately below the PSV site may not have had much significance to him if he knew nothing of the absence of the PSV but if he did know of its absence he would be unlikely to have missed the significance of the scaffolding. The fact that the manual isolation valve would eventually have to be opened would inevitably have been in his mind at the time of repressurising which again would have reminded him of the absence of the PSV. When Richard was called away for a gas alarm it is likely that he would have told Vernon why he was being called away. The fact that a gas alarm had annunciated in module C would serve to remind Vernon that the PSV had been removed. Apart from anything else the mere fact of asking Clark if he could have the pump back from maintenance would tend to remind him that he could not use it anyway because the PSV was missing. If, therefore, Vernon was made aware of the absence of the PSV at about 6 p.m. I am forced to the conclusion that he could not have been in a state of total ignorance at 9.45 p.m.

The crucial question therefore becomes whether or not it is established that he was made aware at 6 p.m. This in turn depends on what happened to the PSV permit and brings us back to Rankin’s evidence and the Lord Ordinary’s treatment of it. It has to be said that at different parts of his Opinion the Lord Ordinary appears to take differing views of the reliability of Rankin. He deals specifically with Rankin’s evidence at pages 741 to 752.

What then happened to the PSV permit? There can be no doubt that there was such a permit issued in respect of the removal of the PSV. In accordance with normal procedure if the work had not been concluded by 6 p.m. the permit would require to be suspended or renewed. There is no question of the permit being renewed, therefore either the permit was suspended in accordance with the normal practice or it was not. If the permit had not been returned by Rankin to the control room for suspension, the live permit would eventually be alone on the lead production operator’s desk as the sole surviving live permit not extended, cancelled or suspended and, as it was put, it would stick out like a sore thumb. The overwhelming probability therefore is that it must have been suspended. I discount the suggestion that Rankin might have taken all of the copies of the permit away with him. This was not suggested in evidence to Rankin who said he left his copy on the desk. If the permit was suspended, when was it suspended? This depends again on Rankin’s evidence. Although the Lord Ordinary, at page 748, mentions the possibility that the permit was suspended with Flook, he never considers this possibility further. If the permit was suspended with Flook I accept that this would involve Rankin finishing work and going to the control room before about 5.15 p.m. which is when Vernon would come to the control room for the handover. This, however, is not impossible. Even if Rankin did not start work on removing the PSV until 1 p.m. there would have been time for him to finish before 5.15 p.m. This would be consistent with the evidence of McDonald who said that Score employees appeared to know that there was no crane available for the return of the PSV at about 5.30 p.m. It is also more consistent with Rankin saying that there was only one person in the control room when he went there, as it is clear that between 5.25 p.m. and about 5.45 p.m. both lead production operators would be there doing their handover and after 5.45 p.m. is a busy time in the control room with people coming in to suspend or renew permits. I accept that if the permit was suspended with Flook this would be inconsistent with Rankin saying that he did not know the person that he saw in the control room at that time because it must have been Flook who issued the permit to Rankin in the morning. That, however, may be one of Rankin’s lapses of memory. The evidence is certainly not clear enough to make any positive finding to the effect that the permit was suspended with Flook, but at the very least it is one of the possibilities. Accordingly, while I accept that the evidence indicates that Rankin did have a permit and did at some stage have that permit suspended, I would not be prepared on the evidence available to hold that it was positively established that Rankin suspended his permit with Vernon, nor do I consider that the Lord Ordinary can be criticised for not making such a finding. While, therefore, I would not go as far as the Lord Ordinary as to suggest that Vernon may have known of the absence of the PSV but have forgotten about it, I do not consider that this suggestion by the Lord Ordinary vitiates his conclusion that at the time when Vernon started to repressurise the pump, if in fact he did so, he was ignorant of the absence of the PSV.

A major part of the respondents’ case before the Lord Ordinary was that there was no evidence to prove an essential part of the reclaimers’ case, namely, that Vernon repressurised the pump. Indeed such evidence as there was indicated that he would have had insufficient time to do so and in addition would have required to do so in an unusual manner in order to produce a leak of a size which would produce a flammable mass. This involves consideration of the nature of the process. Before a pump is handed over to maintenance it requires to be depressurised by shutting off the inlet gas operated valve (GOV) and thereafter venting the contents of the pump and associated pipework. Before starting the pump again it is necessary to repressurise it. At this stage electrical supply to the pump is not necessary. The first step is to reconnect the 100 psi airline to the GOV. This air supply is used to operate the valve by means of a push/pull button. The valve itself is a ball valve which rotates through 90 degrees. The air pressure operates on a spring loaded piston attached to a Scotch yoke which in turn rotates the valve. When the button is pulled air is introduced and the valve begins to open. It takes about 8 to 10 seconds for the valve to move from fully closed to the fully opened position. When the button is released the air pressure is cut off and a spring closes the valve. (In normal operation when the electrical supply is connected a separate 50 psi airline operates to keep the button in the open position thus keeping the valve open, but as the pump was electrically isolated this function is not material for present purposes). If the pump chamber is at atmospheric pressure there is a possibility of damage to the mechanism if the GOV is fully opened in one movement because of the sudden surge of pressurised condensate. Accordingly, the practice is to jag the pump. This means pulling the button and holding it open until the valve has partially opened and then releasing the button. This has the effect of introducing a small quantity of condensate at a time thus avoiding a pressure surge. There is a dead time of about 2 seconds before the valve begins to open and the button is released very shortly after the valve starts opening. Accordingly, each jag only takes a few seconds. Normally about three or four jags are used to bring the pump up to full pressure. The evidence disclosed a considerable variation in the length of time taken to perform the whole jagging operation, varying from 30 seconds (Murray) to 5 minutes (Henderson). The time mentioned by Murray is however not a reliable guide as he was talking about a situation where the discharge side of the pump was repressurised by use of an equalisation line from the neighbouring pump which was still running. Probably the most reliable guide is the evidence of Grieve, whose estimate was 1 minute to 3 minutes. What is perhaps more important than the exact length of time the operation would take, however, is that it was never suggested to any of the witnesses that the operation would be anything other than continuous. The significance of this is that it is necessary for the reclaimers’ hypothesis as to how the explosion occurred for there to have been two separate injections of condensate into the pump with an interval between them of about 2 minutes. Furthermore for there to have been a sufficiently large leak to provide the necessary flammable mass the button would need to have been held out on the last jag for about 15 to 20 seconds after the pump had been fully pressurised, which would not appear to be part of the normal procedure and would not appear to have any immediately discernible purpose. The respondents therefore say that quite apart from Grieve’s evidence that he did not see Vernon at the GOV button, the whole scenario is highly improbable. The respondents point out that during the evidence of the material witnesses it was never suggested to any of them by the reclaimers’ counsel that the jagging operation would ever be performed in such a way, and in the absence of any evidence to that effect the court should be very slow to draw such an inference.

The reclaimers maintain that the Lord Ordinary was entitled to hold that there was time for Vernon to repressurise the pump and entitled to hold that he did so. They approach the matter, so to speak, from the other end. The Lord Ordinary held that the explosion occurred because of a leak of condensate in module C and although this was hotly contested during the proof, the respondents now accept this finding. Because of the direction of the wind flow and the pattern of gas alarms the leak would be in the south east quadrant which is where PSV 504 is situated. There were a limited number of other condensate lines in the area and there is no indication in the evidence of any reason to suppose that there might have been any leak from any of them. Important evidence, however, comes from Grieve who was at the condensate injection pumps on the 68 foot level at the time of the explosion. He described a fireball in the roof space immediately under the point where the relief line leading from the pump to the PSV went through the roof into module C and this would be entirely consistent with a leak from the flange. Standing this evidence indicating the possible source of the leak and in the absence of any credible competing source the Lord Ordinary was entitled to hold that the source was indeed the flange. The only way in which condensate could have reached that flange would be through repressurisation of the pump. Vernon had undoubtedly expressed an intention to bring pump A back into operation. Accordingly, the only sensible inference is that in pursuance of that intention he had repressurised it. Although there was no direct evidence that he did so, Grieve’s evidence cannot exclude the possibility. While the timing of the jagging operation is unusual there is no requirement that it should be a continuous operation. There is also the possibility that the initial jagging which caused the first leak which initiated the C3 gas alarm was carried out by Richard and that it was only after he was called away that Vernon proceeded with the second stage. The reclaimers accepted that these hypotheses had not been put to the witnesses who might have been able to comment, but maintained that approaching the matter in the way they did all they had to prove was that the leak came from the flange, and if that was established it was inevitable that the pump must have been repressurised whatever the problems of timing may have been. If the respondents wish to show that it was impossible for Vernon to have jagged the pump, it was open to them to suggest to the witnesses that he could not have done so for the reasons which they now advance.

While there are undoubtedly a number of problems involved in the timing, I have come to the conclusion, as did the Lord Ordinary, that the reclaimers’ approach is a legitimate one. Accordingly, the vital question becomes not whether it is clear on the evidence that Vernon did in fact repressurise the pump but whether there was a leak from the flange at the site of PSV 504. If that is established on the evidence it inevitably follows that the pump must have been repressurised and that it would be Vernon who performed that operation.

The respondents maintained that the evidence of Grieve, combined with evidence which he was not asked to give, was fatal to the reclaimers’ contention. Grieve was the phase 2 operator who was working in the gas conservation module which was out of commission while undergoing substantial maintenance. He heard on his radio that there was a problem with a condensate injection pump and as he had no particular work to do at the time, he decided to go to the 68 foot level to see if he could assist. When he arrived there he passed the control panel JCP57 and observed that neither condensate injection pump was running. He then saw Vernon and Richard at the south west corner of pump B. He assumed that they were trying to start that pump. Because of the noise level in the area he could not have any real conversation with them, but he formed the impression that pump B was ready for a restart. He therefore made his way to the local control panel and pressed the start button. The motor turned but when he released the button the motor stopped. He then decided to go round the pump to check that the manual speed control was set to zero. Before he had the opportunity to check that, the explosion occurred. His estimate was that he was only in the area of the pumps for 2 to 3 minutes before the explosion. During that time he was aware that there was a message over the radio for Richard but he was not aware of the content of the message. However, he observed Richard leaving by the stairs to the north west of the pumps. He was asked about Vernon’s position and said that the last recollection he had was of seeing Vernon with Richard between the GOVs . He was not asked if he had seen Vernon jagging pump A nor was he asked if it was possible that Vernon could have been jagging pump A without him noticing. The respondents maintained that the reclaimers could not succeed in proving that Vernon jagged the pump, an essential part of their case, on a purely circumstantial basis when there was an eye witness present at the time when the pump must have been jagged who was not asked if he had seen that happening and more importantly was not asked if it could have happened without him noticing. The failure to ask if it could have happened raises the inference that it did not.

I accept that with the benefit of hindsight it could be said that it would undoubtedly have been preferable for Grieve to have been specifically asked if Vernon could have been jagging the pump without him noticing. However, as the Lord Ordinary points out, Grieve only appears to have noticed Vernon at two stages, namely when he first arrived at the vicinity and saw Vernon at the south west corner of pump B and then his last recollection was of seeing him between the GOVs and he had no idea what he was actually doing. There is nothing in Grieve’s evidence which gives any indication that he kept Vernon in view throughout the whole time he was at the 68 foot level. If such evidence had been given it may be that the failure to ask if Vernon could have jagged would have been fatal, but where there is no such indication I am unable to regard Grieve’s evidence as preventing the inference being drawn if appropriate from other evidence that Vernon did in fact jag the pump.

Important evidence in the case was given by Dr. Davies who carried out tests simulating leaks from various locations in order to try to establish what sort of leak and from where would be capable of producing the pattern of gas alarms spoken to by Bollands. Bollands’ evidence, which was not challenged by either party and was generally accepted by the Lord Ordinary, was that the first alarm to annunciate was a low gas alarm in zone C3. That was the alarm which Bollands intimated to Richard and was the cause of Richard leaving the 68 foot level to investigate. About 2 minutes later there was a flurry of alarms followed immediately by the explosion. Although not absolutely clear, his evidence indicates that in the final flurry the sequence was low gas alarms from C4 and C5, virtually together, followed by C2, followed by a high gas alarm from one of the C detectors all within a few seconds (9493 to 9510). When carrying out his tests in 1989 Davies was obviously concerned to be as accurate as possible about the sequence and timing and he obtained information at the time by way of Bollands’ precognition which is recorded in Davies’ working papers. Davies originally gave evidence as an expert at the Cullen Inquiry. He was thereafter engaged by the reclaimers. He prepared an initial report for the reclaimers in 1989 and a subsequent report some years later, both reports being productions in this case. It is clear, as stated in the first of these reports, that he proceeded on certain assumptions which are set out in section 2. It is also made clear that his brief was to ascertain if a leak from the blind flange could be a cause of the pattern of gas alarms. No mention is made in these assumptions of any scaffolding in the area of PSV 504. 34 simulations were carried out using different flow rates, different ways in which the condensate could escape, e.g. a complete circumferential leak, a partly circumferential leak or a jet, and different locations of a leak within module C, although most were carried out from the location of the PSV. One of these simulations (series 27) was of a downward jet at a leak rate of 37 kg./min. onto a plate situated 1 metre below the source. The conclusion reached in the report was that on the basis of the assumptions in section 2 a small leak of vapour from the condensate stream followed by a much larger release of condensate must have occurred in the vicinity of PSV 504. His evidence in chief was along the same lines. In cross-examination, however, it transpired that one of the assumptions given to him by the reclaimers was that there was a scaffolding platform beneath the site of the PSV. He had been previously aware of that anyway hence the simulation in series 27. He accepted that the result of series 27 was that there was dispersion of the gas escape which led to all C2 alarms annunciating before any alarm from C3. In other words a simulation of a downward leak at the rate of 37 kg./min. onto a scaffolding platform was inconsistent with the observed pattern of gas alarms. Prior to Davies giving evidence a number of witnesses had mentioned the existence of the scaffolding at the site of the PSV. None of them had been asked in any detail about the dimensions of the scaffolding platform. The respondents maintained that such evidence as there was indicated that the platform would have been in such a position that the downward leak would inevitably have struck it and dispersed. The limited evidence bearing on this point can be found in the evidence of Rankin (11457), McDonald (11312), Bagnall (10522 and 10529), McGregor (8967) and Standon (9645). The respondents criticised the Lord Ordinary’s approach to this issue. He dealt with it at three passages in his Opinion. At pages 507 to 509 he says that the exact position and height of the scaffolding is not known but that it was said that it was about 1 metre below the PSV, which is what one would expect. Whether directly below the flange or to one side is not known, but it must have been close enough to allow access.

At pages 559 to 561 the Lord Ordinary deals again with the defenders’ submission that the presence of scaffolding rendered Dr. Davies’ tests valueless. At the end of that passage he says:

The respondents criticised that approach on the basis that it was simply a bald assertion that he did not accept the defenders’ contention. However, the Lord Ordinary at pages 873/ 4 gives more detailed reasons. What he says there is:

Having regard to this passage in the Lord Ordinary’s Opinion it does not appear to me to be a proper criticism of the Lord Ordinary that he simply dismissed the respondents’ contention with the simple assertion that he did not accept it.

The reclaimers’ position was that, assuming a two-stage release, as far as the second stage was concerned, because of the rapid spread of a large gas cloud and the flurry of alarms all within a few seconds the precise pattern of alarms is immaterial, and as far as the first stage is concerned, despite the finding in series 27, the pattern is compatible with a small continuous leak at a rate of about 12 kg./min. The reclaimers referred to series 34 which simulated a fully circumferential leak at the rate of 12 kg./min. That produced a pattern of C3 being the first alarm to annunciate. Davies considered that a downward jet or partially circumferential leak which impinged on a platform could be regarded as being in the same class as a fully circumferential leak in that it would produce a dispersed source. Accordingly, although Davies did not try it, a downward leak onto a platform but at a rate of 12 kg./min. might be expected to produce a result similar to series 34. In that situation the presence of the platform would not result in the triggering of a C2 detector before a C3 detector. A partial repressurisation of the pump to 20 bar would produce a leak at the rate of 15 kg./min. Even if this impinged onto the platform it could still be expected to perform in the same way as a series 34 test and cause a C3 alarm to be triggered first.

The respondents further maintain that Davies’ whole evidence is fatally flawed in that it proceeds on an assumption that the leak was substantially heavier than air, an assumption that not only was not proved but was in fact disproved. His modelling was done with an inert gas which had a molecular weight of 44. The molecular weight of air is 29. The reason he used gas with the molecular weight of 44 was that the molecular weight of the condensate was 42 and the closest hydrocarbon to that is propane with a molecular weight of 44. What the respondents maintained however is that Davies failed to appreciate that not all of the condensate would turn into gas when it was released at the leak source. Because of the process known as differential flashing, when a condensate mixture under pressure is suddenly released into air at atmospheric pressure the lighter ends of the condensate will, within milliseconds, flash into gas but the heavier ends will remain in the liquid phase. All of the methane, nearly all of the ethane, and 50% of the propane will so flash but the remainder will remain liquid. The molecular weight of methane is 16.04, of ethane 30.07, and of propane 44.09. The proportion in the condensate stream of methane is 18%, of ethane is 19% and of propane is 30%. It follows that only about 50% of the condensate stream would flash and because the three hydrocarbons concerned all have relatively low molecular weights the gas cloud so formed would also have a low molecular weight approximating to that of air. The gas cloud would therefore be neutrally buoyant. Davies’ evidence was that if the leak was of neutrally buoyant gas it would rapidly disperse in all directions and if the source was from the PSV site the C2 detectors, particularly G101/1, which was close to the PSV site, would inevitably detect it before it reached any of the C3 detectors. Accordingly, had Davies appreciated the phenomenon of differential flashing he could not have used gas with a molecular weight of 44 for his tests. His tests are therefore of no value. His evidence, however, shows that if he had used a neutrally buoyant gas the inevitable result would have been that the C2 detectors would have picked up the gas cloud first if the leak had in fact come from the site of the PSV.

The reclaimers contended that this argument was not presented to the Lord Ordinary and should not be considered by this court. Before the Lord Ordinary the matter of flashing was argued in relation specifically to the reclaimers’ hypothesis that the first stage leak was a gas leak, and what was argued was that when the pump was repressurised there would initially be flashing of the lighter ends of the condensate inside the pump chamber and that that would be the gas which would rise to the end of the relief line. Accordingly, when gas leaked under pressure from the relief line it would consist of the lighter ends which would have a molecular weight about the same as that of air and would be neutrally buoyant. As Davies’ evidence was that a neutrally buoyant cloud would rapidly disperse such a cloud would first impinge on the G101/1, which was one of the C2 detectors. The submissions made to the Lord Ordinary by the respondents, however, appeared to accept that different considerations would apply at the second stage which, under the reclaimers’ hypothesis, would have to be a leak of liquid condensate.

One thing that is clear is that this is a highly specialised area. Unfortunately the ingenious interpretation of the evidence in relation to the second stage condensate release was not put to Davies for his comments, and I for one would not be prepared to substitute my own views in such a complex matter for that of the many experts who gave evidence in this proof. What is the effect of evaporation, which might not take place in milliseconds but, at least in the case of the balance of the propane and possibly butane, would take place in a second or two? That is a matter which was considered almost in passing with Richardson and unfortunately his evidence on the matter is not entirely clear. At one point he appears to be including evaporation within flashing, while at another point he treats them separately (see pages 16542 and 16639). What may be a significant passage in his evidence is at page 16542 where he is dealing with the behaviour of condensate when it passes through an opening GOV into the pump chamber. He said

It appears from that passage that the reason that more than 50% to 60% of the condensate does not flash into gas when entering the pump chamber is because of the severe temperature drop, together with the fact that inside the pump chamber the pressure is building up. As he points out, however, at atmospheric pressure and room temperature all the condensate may flash. It would appear to follow from this that condensate released from a leaking flange into module C would immediately be at atmospheric pressure and if in the form of aerosol or droplets would rapidly approach room temperature. If this is correct the gas cloud, a short distance from the orifice, would contain far more than just methane, ethane and half the propane.

I agree with the Lord Ordinary that Davies’ evidence cannot in itself prove that the leak came from the site of PSV 504. As the Lord Ordinary recognised, the high water mark of Davies’ evidence was that the reclaimer’s hypothesis is "eminently possible". However, I am equally clear that his evidence cannot, as the respondents maintain, exclude that site as the leak source. However elaborate his modelling techniques were they could not replicate exactly the conditions on the platform on the night of the disaster. Substantial quantities of piping and other equipment between the PSV site and the various detectors, which might have affected the gas flow, were missing from his model which was a substantial simplification of the actual layout. The existence of the scaffolding platform under the site did not form part of his experiments except in the one instance. The molecular weight of the gas he used may not have been precisely the same as that of the leak and if the actual leak contained entrained liquid it may not have behaved in the same way as the gas used by Davies. The position of the vital detector G101/1 was modelled as being at a height of 12 feet, whereas the Lord Ordinary (page 236) has accepted the evidence that it was in fact at a height of 20 feet, a discrepancy which would have some bearing on its ability to detect a neutrally buoyant gas from the site of PSV 504 which was at a height of about 15 feet (Lord Ordinary page 477). I am accordingly satisfied that it cannot be said that on Davies’ evidence the only reasonable hypothesis was that a leak from the PSV site of either gas or condensate would inevitably have triggered a C2 detector before a C3 detector, which would be in direct contradiction of the evidence of Bollands which the Lord Ordinary has accepted as being correct, and thus destructive of the reclaimers’ hypothesis.

The Lord Ordinary accepts that the case against Sutton depends entirely on inference. There was no evidence that he was a careless worker or ignorant of what was required of him. He was alone when fitting the flange and there was no evidence one way or the other as to whether he used combination spanners to tighten the bolts. If the bolts had been properly tightened with the use of combination spanners there could have been no leak. The Lord Ordinary concludes from the other evidence in the case that there was a leak from that site and therefore "I must conclude that Mr. Sutton did not fit the blind flange properly" (page 882). This conclusion was attacked by the respondents on two grounds. In the first place they contended that it was inherently improbable that Sutton would tighten the bolts only finger tight. He had plenty of time to do the job properly and he had nothing else to do during the time that Rankin was carrying out the calibration of the valve. In view of the fact that the flange is now under the sea, as is all the other pipework in module C, the improbability of Sutton having been in wilful dereliction of duty cannot be elevated into a probability on the principle enunciated in Rhesa Shipping. In the second place the respondents argued that the Lord Ordinary’s approach to Sutton’s alleged negligence was flawed. The proper approach was to consider the likelihood, or otherwise, of Sutton having failed to fit the flange properly and then take that into account as one of the factors in a consideration of whether the reclaimers had proved that there was a leak from the PSV site. It was a wrong approach to ignore the improbability of Sutton’s negligence, to decide without taking that factor into account that the rest of the evidence made it probable that a loose flange was the source of the leak, and then conclude that Sutton must have been negligent. In my view these criticisms do not properly reflect the Lord Ordinary’s approach. At pages 779 to 786 the Lord Ordinary considers all the evidence relating to Sutton’s fitting of the flange, including various possible reasons why he may not have fitted it properly on this occasion. He would have expected that the valve would have been refitted during the course of his shift and would be aware that pump A had been taken out of service for planned maintenance. He might, therefore, not have regarded the properly secure fitting of the flange as being of importance. In this context it is perhaps of interest to note that his supervisor, Rankin, appears at least in one point in his evidence (11414/5) to have differentiated between fitting flanges to valves for testing purposes when liquid or gas would be introduced under pressure and fitting flanges to open ended pipework. In relation to the latter he thought that the principal purpose in fitting a blind flange was "to protect the faces of the flange to stop any accidental spillage from the system or to prevent any dirt entering from outside". If this was the supervisor’s view it may be that Sutton was under the same misapprehension. There was also evidence, for what it was worth, from Bagnall to the effect that the constricted space around the PSV would make it difficult to use combination spanners and that the fitting and proper tightening of the two flanges would take a considerable time which would not be far short of the time it would take Rankin to do the necessary work on the valve and be ready to return it to position. In this passage of his Opinion the Lord Ordinary does not come to any specific conclusion. This is just one of the many passages in the Opinion where the Lord Ordinary discusses the different aspects of the evidence which have a bearing on the crucial question of whether or not it has been proved that this was the source of the leak. His conclusion on the cause of the accident is set out at pages 854 to 883 and in considering that there is no reason to suppose that the Lord Ordinary did not have in mind what he had already discussed in relation to Sutton’s actings. The answer to the argument relating to Rhesa Shipping is the same as already discussed, namely that the Lord Ordinary, on a consideration of the whole evidence, concluded that the reclaimers’ hypothesis was probable.

In considering whether the reclaimers have proved their hypothesis as to the cause of the initial explosion a number of factors have to be balanced. The balancing exercise has to be carried out bearing in mind at all times two important considerations. In the first place it has to be remembered that many vital potential witnesses did not survive the disaster and the real evidence is at the bottom of the sea. Accordingly, for the reclaimers to succeed, a balancing exercise must produce a result which is probable and not merely the least unattractive possibility. Secondly, it has to be remembered that as the number and weight of improbabilities increase so the value of the probabilities will diminish.

The factors in favour of the reclaimers’ hypothesis may be summarised as follows. The explosion occurred in the south east quadrant of module C down wind of the site of PSV 504. The necessary flammable mass could only come from a leak of condensate. The only part of the plant in module C where the integrity of the line had been breached was the site of PSV 504. A fireball was seen immediately after the explosion in the roof space at the 68 foot level below the site which could have been fuelled from a condensate leak from the site. The pattern of gas alarms pointed to a two stage leak about 2 minutes apart. This would be consistent with a two stage jagging operation and wholly inconsistent with a leak from some other source which would be likely to be a continuous leak which could not give rise to the pattern of alarms. A few minutes before the explosion Vernon had expressed the intention of bringing pump A back into operation and had taken steps to have the electrical supply deisolated. Before the pump could be started it would have to be repressurised by jagging. Vernon was present at the site of pump A and in the vicinity of the push/pull button which actuated the GOV.

The factors to be weighed against the hypothesis are that there are many ways in which Vernon could have become aware that the PSV was missing and if he was so aware it is highly unlikely that he would have contemplated starting pump A. The only surviving eye witness at the 68 foot level, Grieve, did not see Vernon jagging although he must have been present while it was being done. The timing of a jagging operation necessary to produce a two-stage leak was wholly abnormal. The only type of leak from the site would have to be a downward pointing one in order to avoid setting off G101/1 as the first gas alarm and because of the presence of the scaffolding platform under the site any downward pointing leak would strike the platform and be dispersed, thus setting off that alarm. There was no evidence that Sutton was idle or incompetent or short of time or was unaware that the proper method of fitting a flange was to flog it up with combination spanners.

There appears to me to be one marked difference between these two sets of factors. Those favouring the reclaimers’ hypothesis are to a substantial extent based on facts which are either admitted or have been found by the Lord Ordinary and are now unchallenged. Those which challenge the hypothesis are largely based, not on uncontroverted facts, but on inferences, mainly inferences which the Lord Ordinary was not prepared to draw from the acceptable evidence before him. In that situation it is not surprising that the Lord Ordinary chose to put much greater weight on the factors favouring the hypothesis and in my opinion he was fully justified in doing so. I reject the criticism by the respondents that the Lord Ordinary started with the proposition that Vernon intended to restart pump A and then juggled the remaining facts to suit that proposition. I am satisfied that in coming to his eventual conclusion on the cause of the disaster the Lord Ordinary in his exhaustive Opinion properly weighed all the various factors, and the fact that he chose to put most weight on the factor relating to Vernon’s intention was something which he was entitled to do. In the whole circumstances I am not satisfied that the Lord Ordinary erred in the conclusion to which he came, namely that the initiating cause of the disaster was brought about by negligence on the part of both Vernon and Sutton.

This argument proceeds on the basis that Sutton was negligent in failing to fit the blind flange properly and that thereafter condensate was introduced into the relief line by Vernon. The respondents maintain that Vernon’s actings constituted a novus actus interveniens and as a result Sutton’s negligence ceased to have a causative effect. If this is so the sole negligence would be that of Vernon in which case the indemnities would not apply.

The argument is that a human intervention breaks the chain of causation because on the evidence it would not be reasonably foreseeable that condensate could enter the line by being deliberately introduced. It was accepted that it was reasonably foreseeable that there might be a seepage of condensate through a defective GOV and if that had been the cause of the leak Sutton would be held negligent if he had not fitted the flange properly. What would not, however, be foreseeable would be that an operator would deliberately pressure up the line and such an act would be a novus actus. The only suggestion in the evidence about how condensate could enter the line was that there might be seepage through a valve. Counsel referred to various cases which were said to establish that if the supervening human intervention could be categorised as any one of unlikely, not reasonably to be anticipated, extraneous, extrinsic, a new and unexpected factor, a chapter of accidents and mistakes or not in the ordinary course of things then that intervention is a novus actus. In this case Vernon’s action in repressurising when a PSV was missing could be described by any one of these epithets and accordingly constituted a novus actus interveniens.

The cases referred to could be said to fall into two classes. In the first class were rescue cases such as Haynes v. Harwood [1935] 1 K.B. 146, The Oropesa [1943] P. 32 and Knightley v. Johns [1982] 1 W.L.R. 349. In rescue type cases a wrongdoer who creates a situation where life or limb is at danger is deemed to have foreseen that some public spirited person will attempt to effect a rescue and he will be liable for injuries sustained in the course of that rescue. He will not, however, be liable if the rescuer acts in such a way as to be characterised by one of the epithets used by counsel, because such an act will be a novus actus. In the second class are cases where the true question was how far a duty of care extends. Are post office workers, leaving a manhole open in the road and a lit lamp beside it who admittedly would be liable to a child allured by the prospect who sustained burning injuries, also liable to the same child who is in fact injured by an explosion? Hughes v. Lord Advocate 1963 S.C. (H.L.) 31. Are employers who are admittedly liable to their employee for a leg injury sustained at work liable in respect of a further injury when the employee acted in an unreasonable manner three weeks later, allegedly because of weakness in his injured leg? McKew v. Holland & Hannen & Cubitts 1970 S.C. (H.L.) 20. Are borstal officers liable to persons in the vicinity of a camp through failing to supervise inmates who absconded and caused damage to their property? Dorset Yacht Company v. Home Office [1970] A.C. 1004. Are owners of derelict property liable to neighbours when trespassers set fire to the property causing damage to the neighbour’s property? Maloco v. Littlewoods 1987 S.C. (H.L.) 37. Is a council which should have removed a derelict boat, and which would admittedly be liable if a child playing on the boat fell through rotten planking, also liable to a teenager who attempted to repair the boat by jacking up the bows and was injured when it collapsed on top of him? Jolley v. Sutton LBC [1998] 1 W.L.R. 1546. Again, in such cases, the original wrongdoer will not be held liable for what is done independently by another party unless it is foreseeable that such an independent act will occur. It appears to me, however, that the issue in none of these cases has any material similarity to the issue in the present case. Sutton owed a duty of care to others on the platform. He knew that the purpose of fitting a blind flange to an open pipe was to ensure that there could be no escape of condensate which could, as it did, produce catastrophic results. That he must have been aware that this was the purpose is perfectly clear from the fact that a range of flanges was available to him and he required to fit a flange designed to cope with the particular pressure in the pipe to which it was to be fitted. He also knew that the bolts had to be flogged tight to ensure no leakage. It was not for him to decide that because he could not think of any way in which condensate could enter the pipe he would not flog the bolts tight. This operation was designed, as were many operations on the platform, to be part of a belt and braces system whereby each safety device was, if possible, backed up by another so that even the most highly improbable potential for disaster which managed to circumvent one safety precaution would be faced with another. For some reason he failed in his duty. Had he performed his duty the disaster would not have occurred as a properly fitted flange would have contained the condensate, even under full system pressure. His failure in duty was accordingly a direct cause of the leakage which occurred. It is of course true that Vernon should have acquainted himself with the state of the PSV and should not have repressurised the pump. That, however, is a separate act which would not have caused the disaster had it not been for Sutton’s negligence. The failures in duty of Sutton and Vernon were accordingly concurrent causes jointly contributing to the disaster. This is an entirely different scenario from cases where an original wrongdoer is in effect being held responsible for what someone else subsequently chooses to do. Sutton is not being held responsible for the actings of Vernon. He is being held responsible for what he himself failed to do, his failure being a direct concurrent cause of the disaster. I am accordingly of opinion that this is a straightforward case of two failures in duty combining to cause the disastrous result where neither failure by itself would have caused any problem in the absence of the other. It follows that the concept of novus actus interveniens has no application and is of no assistance to the respondents.

These actions are all based on claims under indemnity clauses in the contracts. The first question that arises is whether the indemnity clauses are operative on the assumption that the injury or death was caused by joint negligence of an employee of the reclaimers and an employee of Score, but with no negligence or breach of statutory duty on the part of the respondents or their employees.

The respondents’ submission was that on a proper construction of the indemnity clauses they were not operative unless there was some negligence or breach of statutory duty on the part of the contractor concerned.

The regulatory regime was such that it was the duty of every person on the rig not to do anything likely to endanger safety. It was the duty of every employer to ensure that his employees complied with the regulations and it was the duty of the operator to ensure that the regulations were complied with. Certain defences are provided in the case of any criminal proceedings but it is specifically provided that these defences are not available in the case of any civil claim. This means in effect that if anything is done to endanger safety, the operator is in breach of an absolute duty. It is to be assumed that the draftsman of the indemnities was aware of this statutory regime.

As far as indemnity clauses are concerned these will always be strictly construed contra proferentem. It will be presumed that such clauses do not cover cases where there has been negligence on the part of the party to be indemnified. The reason for that presumption is the inherent or fundamental improbability of one party being prepared to indemnify against the negligence of the other party. The presumption can be overcome if the clause expressly covers that situation or if the clause is capable of covering that situation and there is no other sensible construction that can be given to the indemnity. The classic test is that set out by Lord Morton of Henryton in Canada Steamship Lines v. The King [1952] A.C. 192. The test applied by the Lord Ordinary was that the strict approach means that it is sufficient for the respondents to provide a sensible contrary construction which the provision was capable of bearing. The respondents accept that the Lord Ordinary correctly identified the test to be applied but contend that he misapplied it.

The contracts in the seven test cases all had different wording to a greater or lesser extent, and it is therefore necessary to consider each clause separately. The indemnity clauses in the seven contracts can be broadly divided into two groups, one group having a long opening paragraph and the other group omitting this paragraph. The respondents in argument concentrated on the first group of which the Eastman Christensen contract was taken as an example. The relevant provisions are in the following terms:

The respondents’ contention was that what is contained in Article 15.1 is a single composite indemnity. There is no doubt that the opening paragraph is concerned only with events occasioned by omission or negligence on the part of the contractor. The respondents say that because what follows is prefixed by "without prejudice to the foregoing generality" sub-articles (a) to (e) are subject to what is said in what the respondents call the preamble and are encompassed within the generality of that preamble. If there was any doubt about that it is resolved by the fact that the preamble concludes with the limitation of £1,000,000 per occurrence. If the reclaimers’ argument is correct that (a) to (e) are separable from the preamble it would follow that there is no limitation in respect of those indemnities except where specifically provided within them. In a case where a contractor’s employee is injured through negligence of the contractor but chooses to sue the reclaimers this would fall under both the preamble and (c) and if it falls under (c) there is no limitation. It would be odd for the indemnity to be limited in a case where the accident was brought about by the contractor’s negligence but unlimited where it was brought about by the joint fault of the operators and some other party with no fault on the contractor’s part.

The respondents argue that other adminicles show that their construction is correct. Article 16 requires employers’ liability, workman’s compensation and general public liability insurance. There is no requirement for insurance to cover a situation where the contractor is not at fault. If the respondents are right there would be insurance cover for an indemnified event whereas, if the reclaimers are right, there may not be. If one contractor’s employee is injured by the negligence of another contractor’s employee he could sue the second contractor and the operator. If the reclaimers are right the operator would get indemnity from the first contractor. So both contractors would have to insure against this risk. If the respondents are right the operator would not get indemnity against the first contractor who would not need to insure but could get indemnity against the second contractor under the third party indemnity. As far as the operator’s indemnities are concerned there is no preamble. This means that there is no precondition of negligence on the part of the operator but it is their condition and if they have not created that precondition that is up to them. The draftsman may have had in mind that whenever there is an injury to one of their employees the operator will inevitably be liable under the statutory regime so there is no need for that specific precondition.

The use of "contributory negligence" poses the question contributory to what? Reading the indemnity as composite there is no problem, contributory means negligence by both contractor and company. Sole negligence does not exclude the possibility of breach of regulations by the contractor, e.g. if the company negligently grants a permit to work in circumstances where the work in fact endangers safety, and the contractor carries out the work causing injury, the contractor would be in breach of regulations but the sole common law negligence would be that of the company.

The cross indemnities cannot properly be described as mutual. It cannot be said that each employer is taking responsibility for his own employees because of the provisions about sole negligence and wilful misconduct. If the preamble is supposed to be free-standing that differs from the company’s indemnities. Even if they were mutual that is no reason for taking a different approach to construction. The observations of Steyn L.J. in E.E. Caledonia Limited v. Orbit Valve Company [1994] 1 W.L.R. 1515 at 1523 show that Lord Morton’s third test in Canada Steamship is applicable to the question of the extent to which negligence is covered. Accordingly, even if negligence was covered in some circumstances the test should still be applied to limit those circumstances. This would apply to the respondents’ contention that it would be limited to cases where they were contributorily negligent and not extended to cases where a third party was contributorily negligent.

Two other contracts in the test cases include similar indemnity provisions. In relation to the Kelvin Catering contract the only difference from the Eastman Christensen contract is that there is no financial limit in the long opening paragraph. There is a £2,000,000 limit of indemnity in sub-article 15.1(d) relating to third party liability and the public liability insurance obligation is for £2,000,000 rather than $5,000,000. In relation to the Wood Group Engineering Off-shore contract again there is a limit of indemnity in Article 15.1(d) of £5,000,000 and a public liability insurance obligation of £5,000,000 rather than $5,000,000. There is, however, one material alteration from the Eastman Christensen contract in so far as the second paragraph starts with the words "In addition but without prejudice to the foregoing generality…". The respondents maintained that this makes no material difference to their argument because the sub-articles do contain an addition, namely that the indemnity in the preamble will still apply even if there is contributory negligence on the part of the operator and that constitutes an extension of the general indemnity.

The contracts in the remaining test cases do not contain the long opening paragraph, nor of course do they contain the words "without prejudice to the foregoing generality". The respondents contend that this makes no difference to the proper construction of Article 15.1(c). While the absence of specific reference in a preamble to act or omission on the part of the contractor leaves the matter more open they maintain that the same problem arises with the use of contributory negligence and sole negligence. They say that contributory negligence means contributory negligence as between the contractor and the company. On any view this is a possible construction which the words are capable of bearing in their context and which is commercially sensible. If the reclaimers wanted a wider indemnity to bring the negligence of a third party into the equation they should have expressed that intention clearly and unambiguously.

The contracts in relation to Northern Industrial Marine Services Company and Northern Coasters Limited (Stena) also have a definition of "sole negligence". That is defined as meaning "the exclusive negligence of the party to be indemnified and shall not apply where any other party bears a proportion of the negligence". The respondents contend that the reference to "any other party" refers to the party to be indemnified which includes not only the operator but a long list of other persons, including participants, suppliers and others. Accordingly, this should not be construed as meaning that sole negligence is excluded where any third party is in some way to blame.

It appears to me that the principal submission by the respondents which requires to be considered is that the words contributory negligence relate only to negligence on the part of the contractor contributing to the loss, injury or damage and that negligence by any third party is irrelevant. The respondents say that even if this cannot be regarded as the obvious construction it is nevertheless a possible and commercially sensible construction which cannot be regarded as fanciful or speculative and therefore applying the contra proferentem rule it is the construction which should be preferred.

The reclaimers’ submission in reply was firstly that the role of the court is essentially one of ascertaining the intention of the parties from the language used, in the light of surrounding circumstances, taken to be within their knowledge (Lord Keith in Smith v. U.M.B. Chrysler 1978 S.C. (H.L.) 1 at page 16). If parties make their intention apparent by "adequate and clear words" the court will give effect to this intention. If there are two plausible constructions of the words the court does not simply adopt that which is less burdensome but still has to decide in the light of all the circumstances which construction is the better one (Steyn L.J. in E.E. Caledonia). When there is a clear reference to negligence this satisfies Lord Morton’s first test and there is no need for further consideration of improbability. Certainly consideration still has to be given to the extent of that negligence but that is done under the ordinary rules relating to contra proferentem. The approach to construction should be from a factual standpoint without being unduly legalistic (Lamport and Holt Lines v. Coubro and Scrutton (The Raphael) [1982] 2 Lloyd’s L.R. 42 per Donaldson and May L.J.J.). In the present case all contractors would know that employees of many other contractors would be working in close proximity and they would be aware of the statutory regime. Regulation 32 of the Off-shore Installations (Safety, Health and Welfare) Regulations 1976 has a much wider impact on the operator than on the contractor as the operator has liability for any act by any employee of any contractor which endangers safety even if acting outwith the scope of his employment and not in relation to work (MacMillan v. Wimpey Offshore Engineers 1991 S.L.T. 515). It is not unreasonable for parties to reach agreement as to how disputes in that environment should be handled. The American case of Fontenot v. Mesa Petroleum 1986 791 F. 2d 1207 which the court described as "a multiparty donnybrook" shows the advantage of trying to avoid such multi-party disputes. That case also described the purpose of the indemnity "as in so many oil field service contracts" as being to divide liability in terms of employment rather than fault.

In the present cases the terms of the cross indemnities in relation to injury to employees are the same. The clear implication is that each employer will meet any claim from one of their employees unless the party who would otherwise be indemnified is guilty of sole negligence or wilful misconduct. The absence of any reference to the acts or omissions of the contractor in the contractors’ indemnity is significant. All the other indemnities refer to acts of or failures on the part of the contractor. This one does not and there is no reason to imply that it should have done. The general practice in the industry of each party bearing responsibility for injuries to their own employees is of significance, - see Daintith and Willoughby U.K. Oil and Gas Law paras. 1-845/6; Sharp Off-shore Oil and Gas Insurance page 104-8; evidence in this case from Mr. Crain (volume 10 1732/3); Nelson v. Atlantic Power and Gas Limited 1995 S.L.T. 102. While the exception here of sole negligence on the part of the indemnitee reduces to some extent the value of this argument, the principle of mutuality would still be of value in a large number of cases.

Contributory negligence is not used as some term of art. Claims arise by reason of injury or death and contributory negligence is negligence contributing to that injury. The indemnity bites whoever contributes to the injury. Reading it in that way it is consistent with the reference to sole negligence. If the respondents’ theory is correct and the use of the words "contributory negligence" carries the implication that the contractor was negligent, there would be no need for any reference to sole negligence. In contracts where there is no opening paragraph there is no conceivable justification for reading into the indemnity relating to employee’s injury a requirement that the contractor should be negligent before the indemnity bites. Where there is an opening paragraph that is a free-standing indemnity which carries a financial limit. It is to be noted that the article is headed "Contractor’s indemnities" (in the plural) whereas the opening paragraph refers to "This indemnity". In the next part of the article most of the opening paragraph is repeated which indicates that the opening paragraph is separate from the other indemnities rather than the ruling indemnity of which the rest are merely examples. "Without prejudice to the foregoing generality" means that nothing that follows will cut down the indemnity given by the contractor in the preceding part. It does not mean that nothing that follows can increase the level of indemnity and this is obvious from the fact that by itself the indemnity in the opening paragraph would not bite if there was any negligence on the part of the reclaimers, in accordance with the Canada Steamship rules, whereas 15.1(c) makes it clear that it does bite in such a situation. In any event if the effect of the opening paragraph was to incorporate automatically into all that follows the concept that there is no indemnity if there is no negligence on the part of the contractor, why the reference to "acts and omissions" of the contractor in 15.1(d)? In the contracts where "without prejudice to the foregoing generality" is preceded by "in addition but…" it is abundantly clear that what follows constitutes an addition rather than merely examples of a general indemnity.

I accept that indemnity clauses fall to be construed contra proferentem. I accept, also, that in general indemnity clauses will not be given effect to in respect of negligence on the part of the indemnitee unless there is specific reference to such negligence or at least there is no other sensible construction. This is because the law regards it as being inherently improbable that any party to a contract would be willing to indemnify against the other party’s negligence; Ailsa Craig Fishing Company v. Malvern Fishing Company 1982 S.C. (H.L.) 14 per Lord Fraser at page 61. If, however, there is a known practice in the industry to apportion liability by reference to employment rather than fault, I see no logical reason for not taking that factor into account in considering the strength of the applicability of the general rule of construction. In the present case there is specific reference to the indemnitee’s negligence. Accordingly, it cannot be argued that some negligence, short of sole negligence, bars the right to indemnity. Because of the contra proferentem rule I accept that that does not necessarily preclude any further consideration of the construction of the clause. In particular I would accept that if a sensible construction of the wording of the clause would permit of the indemnity only being applicable in the event of some breach of duty or negligence on the part of the contractor, that construction would have to be given effect to on the view that it was a sensible limitation of the indemnity. What I am not prepared to do, however, is to read words into the contracts which are not there in order to create an ambiguity. In the articles without the opening paragraph it is in my view significant that sub-articles (a), (b), (d) and (e) all refer to some act or omission on the part of the contractor giving rise to the indemnity whereas (c) has no such mention. If the intention had been to make (c) only effective if there was an act or omission on the part of the contractor it would have been perfectly simple so to provide as has in fact been done in relation to the third party indemnity in (d). I consider that it would be straining the words "contributory negligence" in the context of (c) to imply that a reasonable construction would be that the only form of such negligence would be that it had to be joint negligence with that of the contractor. This is particularly so when on the respondents’ argument negligence in that phrase must mean only common law negligence on the part of the operator, whereas the same word must be taken as meaning either common law negligence or breach of statutory duty on the part of the contractor. Furthermore I regard it as a strained construction to say that the function of the reference to sole negligence is to cover the case where there is a breach of statutory duty on the part of the contractor but some degree of negligence on the part of the operator. This would be particularly so in the two contracts in which sole negligence is defined. It appears to me, in the absence of any reference at all in (c) to either negligence or breach of statutory duty by the contractor, unlike every other indemnity, that the obvious and clear meaning of contributory negligence is some degree of negligence which contributes to the causation of the injury or death of an employee or damage to property and it is used in contrast to sole negligence which means precisely what it says. As far as the contracts with the long opening paragraph are concerned it appears to me that that paragraph stands as a general indemnity to cover cases where the contractor has been in breach of statutory duty or negligent. The words "without prejudice to the foregoing generality" cannot mean, as the respondents contended, that what follows is encompassed within the generality, as it is clear that the reference to negligence on the part of the indemnitee in subsequent sub-articles extends the scope of the indemnity. In my view all that these words mean is that nothing that follows will detract from the indemnity granted in the opening paragraph. This is certainly clear in the Wood Group contract where "in addition" is added. It is also, in my view, significant that the words after "without prejudice to the foregoing generality" are largely repetitive of the opening part of the opening paragraph and it is difficult to see what meaning could be given to these words if in fact (a) to (e) are intended simply to be examples of a single general and comprehensive indemnity granted in the opening paragraph. Further, the reference in (d) to acts or omissions of the contractor would be unnecessary if these words were deemed to be imported into (d) in any event by reason of the opening paragraph. The respondents’ construction also involves contributory negligence in (c) as meaning common law negligence in the case of the operator, but either common law negligence or breach of statutory duty in the case of the contractor. While I can see how a particular word could have different meanings in different contexts, I find it difficult to see how a word used once can have a different meaning depending upon to which party it relates. For these reasons I am satisfied that the existence of the opening paragraph, while it appears to add little to the other indemnities and certainly cannot be classified as an example of lucid draughtsmanship, does not imply on any reasonable construction that (c) is to be regarded as operative only in the event of negligence or breach of statutory duty by the contractor. I therefore consider that the indemnity clauses in all the contracts are apt to cover the situation where the operator has been negligent, provided that there has been negligence by some other party, whether the contractor or otherwise, so that the operator is not solely negligent.

The reclaimers settled the claims of those injured and the relatives of those killed in the disaster. The reclaimers were well aware that there was a serious risk that actions would be raised against them in America and in particular Texas. Jury awards in Texas could be eight to ten times comparable awards in Scotland before even considering the possibility of "runaway" awards. Having taken detailed advice on such matters as jurisdiction and the probable level of awards in Texas, the reclaimers made offers to claimants based on a substantial uplift of what were calculated to be the appropriate Scottish levels. The Lord Ordinary found that the settlements were reasonable in all the circumstances. He also found that if the reclaimers are entitled to be indemnified by the respondents there is nothing in the terms of the basic indemnity clauses to prevent the enhanced awards being recoverable. However, five of the contracts in the seven test cases contained a clause in the following terms ("Article 20"):

The Lord Ordinary held that while this clause would not affect recovery of awards at the Scottish level, the enhancements constituted consequential loss and accordingly could not be recovered. The reclaimers argue that the Lord Ordinary was wrong so to do.

The reclaimers’ argument was that on a proper construction of the contracts as a whole Article 20 does not apply to the indemnities on which the actions are based because it has no relevance to an indemnity granted in respect of loss arising from the reclaimers’ liability in delict to claimants where the extent of