England and Wales Court of Appeal (Civil Division) Decisions

IN THE SUPREME COURT OF JUDICATURE
COURT OF APPEAL (CIVIL DIVISION)
ON APPEAL FROM QUEEN'S BENCH DIVISION
COMMERCIAL COURT
The Hon Mr Justice Tomlinson

		Royal Courts of Justice Strand, London, WC2A 2LL
		10th March 2004

Lord Justice Rix:

This is the judgment of the court to which each member of the court has contributed.

1. On 9 March 2000 at about 1420 a fire began in the engine room of the vessel Kastor Too. Fifteen hours later at between 0500 and 0600 on 10 March she sank in deep water between the island of Socotra and the coast of Yemen. She had been on a voyage from Aqaba to Vizagapatnam with a cargo of phosphate. She was a steel-built four hold motor bulker of 17,665 tonnes deadweight, 148m in length, built in Japan in 1977.
2. Kastor Too had been insured under a contract of marine insurance contained in a slip policy on the MAR 1991 form for a period of 24 months at 30 December 1999. Her agreed value was US$3 million. The defendants, in this court the appellants, subscribed to 80 per cent of that risk. We shall call them the "insurers". The claimants, here the respondents, are respectively the owners and the mortgagees of the vessel. We shall call them the "owners". The insured perils included fire and perils of the sea. However, it is well known and has at all times been common ground that perils of the sea do not include the mere entry of seawater because of ordinary wear and tear and in the absence of some accident or fortuity.
3. The owners' case was that the fire had led in turn to explosions and to such effect on the vessel's structure as to cause her to sink. There was no separate case that the entry of the seawater had been caused by any other fortuity. It was the insurers' defence, however, that the fire had not caused the vessel to sink, and that the vessel had been lost because of entry of seawater from merely unexplained causes. They had no positive case of scuttling, but they were clearly suspicious and wished to put the owners to proof that the vessel's sinking had been caused by fire. They did not at first admit any fire, but, if a fire had occurred, they again had no positive case that it had been started deliberately. On the eve of trial, however, they did admit a fire and ultimately, by the end of the trial, they also accepted that it was accidental. But they always contended that the fire and the entry of seawater were independent, if coincident, events, and that only the seawater was causative of the vessel's loss. The logic of the dispute was thus as follows. If the vessel was lost because of the fire, there was cover under the policy, but if it was not, then the insurers were not liable. The insurers relied among other things upon the common ground between experts that a steel ship such as Kastor Too should not sink as a result of an engine room fire and that for her to have sunk within 15 hours or so of the outbreak of fire would be very remarkable.
4. When the owners began proceedings, on 1 November 2000, their only claim was that the vessel had become an actual total loss. Their claim was that the vessel's loss was caused by "(a) the fire and/or (b) the explosions and/or (c) the fortuitous ingress of seawater". In August 2001, however, they added an alternative case that the vessel had suffered a constructive total loss, because "prior to her sinking" she was so damaged by fire that the cost of repairing her would have exceeded her value when repaired. They estimated the cost of repairs to damage suffered by the time of sinking to be $7/8 million.
5. At trial Tomlinson J held that the alternative constructive total loss claim succeeded, but the actual total loss claim failed: [2002] EWHC 2601 (Comm), [2003] 1 All ER (Comm) 277, [2003] 1 Lloyd's Rep 296. He agreed with the insurers that the fire had not caused the sinking, which rather was due to an independent entry of water from some unexplained cause not shown to be fortuitous. However, he agreed with the owners that the fire had led to the constructive total loss of the vessel at some material time prior to her sinking. He therefore awarded the owners (the appropriate percentage of) the agreed value of the vessel plus interest, a total of $3.49 million. He gave the insurers permission to appeal. He reserved the question of costs.
6. At a subsequent hearing the judge heard detailed argument on costs. The owners had obtained judgment on their claim, but they had lost on the issue which had taken most time at trial and caused the greater part of the expense of litigation. In a second, reserved judgment he explained his reasons for ordering the insurers to pay only 15% of the costs of the action to the owners and for ordering the owners to pay 85% of the costs of the action to the insurers. In effect, on the assumption that the parties' costs were in the same amount, the owners, despite their overall success in the litigation, would have to pay the insurers 70% of their costs: [2003] EWHC 472 (Comm), 17 March 2003. Since the costs expended on both sides were very substantial, the owners obtained something of a Pyrrhic victory. The judge gave them permission to appeal on the issue of costs.
7. The owners by their respondents' notice on the (insurers') main appeal also sought to support the judge's order, in the alternative, on the ground that he had erred in rejecting their case of actual total loss. If the judge had stated in his order that it was made on the basis of a finding of constructive total loss, the owners would probably have had to seek permission for their cross-appeal on actual total loss (see Cie Noga d'Importation et d'Exportation SA v. Australia and New Zealand Banking Group Ltd [2002] EWCA 1142, [2003] 1 WLR 307). As it is, the owners having advanced their cross-appeal on the ground of actual total loss withdrew it shortly before the oral hearing. On the merits therefore only the insurers' appeal from the finding of constructive total loss remained for argument, and the owners' cross-appeal on the question of costs would be relevant only if the insurers' appeal failed. It is nevertheless relevant both to certain aspects of the insurers' substantive appeal and to the owners' costs appeal to set out the way in which the constructive total loss issue emerged in the course of the litigation.

The emergence of the constructive total loss issue in the course of the litigation

8. Members of the public are, we suspect, familiar with the concept of the writing-off of motor-cars which have suffered an accident causing damage that is not worth repairing. However, it is only marine insurance that formally recognises the concept of a constructive total loss. Outside marine insurance, the only total loss which is acknowledged is an actual total loss. Marine insurance, however, has two kinds of total loss, actual and constructive. Both are equally total losses. Constructive total loss can itself come in several forms, but the one claimed by the owners in this case is that mentioned in section 60(2)(ii) of the Marine Insurance Act 1906 (the "Act"), which applies when a vessel "is so damaged by a peril insured against, that the cost of repairing the damage would exceed the value of the ship when repaired".
9. Normally, a claim for a constructive total loss has to be notified to the insurers in the form of a "notice of abandonment". Section 62(1) of the Act says in terms that the assured "must give notice of abandonment" and that if he fails to do so "the loss can only be treated as a partial loss". This is so that the insurer who is on risk and is willing to pay can be put in a position to avail himself promptly of the rights and remedies which accompany his taking over the vessel. Exceptionally, however, a notice of abandonment is unnecessary, in particular where at the time the assured learns of the loss "there would be no possibility of benefit" to the underwriter from such a notice being given to him (section 62(7)). It is common ground that that can arise where, as in this case, an assured only learns of the constructive total loss after the vessel has also become an actual total loss.
10. The insurers' response to this alternative plea was to deny that there had been any constructive total loss. Initially, the defence went no further than to put the owners to proof of the essential ingredients of the constructive total loss relied upon. Thus there was no admission of the fire, or of its extent, or of its causing any damage or in particular of its causing damage extensive enough to make the ship not worth repairing. Similarly, it was not admitted that a notice of abandonment could be dispensed with.
11. By the time the trial had begun, however, a fire had been admitted (although its extent and course of development had not). Moreover, the parties' expert marine surveyors had agreed that "had the vessel not sunk, the cost of damage repairs would probably have substantially exceeded US$3,000,000". (The figure of $3 million was taken, rather than "the value…when repaired", because the policy provided by clause 19.2 of the Incorporated Institute clauses that no claim for constructive total lost based upon cost of repair would be recoverable unless such cost exceeded the agreed value.)
12. The trial began on Monday, 1 June 2003. On the previous Thursday, Mr Steven Berry QC, who appeared below as he did again on this appeal for the insurers, served his opening skeleton in which the insurers for the first time disclosed what their real defence case in relation to the plea of constructive total loss would be. First, they admitted that "prior to her sinking the theoretical cost of repairs exceeded her insured value" and thus that if the vessel had survived "there may have been a CTL claim". As the vessel sank, however, the claim was bad "in law" and there then followed three points in explanation of that submission: the first was that there was one loss and one casualty, by sinking; the second was that because the vessel was an actual total loss by sinking, the owners had suffered no loss by reason of the fact that, before and when lost, she was fire damaged; and the third was that "even if there were two separate casualties and losses", nevertheless since the earlier fire damage had not been claimed as a constructive total loss before the sinking, the earlier fire damage had "merged" into the later loss by sinking.
13. The owners responded overnight by serving a supplemental skeleton argument dealing with the constructive total loss claim and requesting a preliminary issue on what was asserted as the "only issue" between the parties, namely "whether the fact that the vessel sank after she had become a constructive total loss defeated the claim". It was asserted that this was a "pure point of law". It may be doubted whether this was an entirely accurate analysis of the insurers' position at that time, but as that position had been put so newly, briefly and obscurely, we think that the owners' response, which was served on the Friday before the trial was due to begin, was understandable. The owners' document detailed the legal merits of their alternative claim on the basis that the vessel was constructively lost by fire before her remains sank.
14. On the first day of trial, Monday 1 June 2002, the insurers also served a supplemental skeleton argument to deal with the owners' application for a preliminary issue. The insurers submitted that whatever merit that application might have had at an earlier stage, it was now too late. As for the merits of the constructive total loss claim, the document responded to the owners' detailed submissions. It emerged that the factual basis of the constructive total loss claim remained disputed – despite the admission of a fire and the cost of repairs. Thus the insurers put the matter in this way:

"7. The factual situation giving rise to the CTL claim is that, in the scintilla temporis before the sinking, the theoretical repair cost of the fire damage exceeded the insured value. At this time she was about to, and bound to, sink."

15. This was the first time that the insurers made it clear that they were testing the question of a constructive total loss at the instant of or before sinking, expressed by reference to a "spark of time" or what might today be called a nanosecond. However, by a footnote to that paragraph the insurers revealed the inherent premises of this submission, thus:

"The Claimants' case, and the Defendants' admission as to repair cost, relates to the fire damage "prior to" sinking, and the case goes no further than the moment immediately prior sinking. If the point had been investigated, the likelihood is that the ship was doomed by ingress of water when the crew abandoned and in any event well before the time the repair cost reached US$3 million."

16. In other words, the insurers were asserting that the constructive total loss claim was to be considered only at the moment of sinking, when the vessel was in any event doomed to loss by sinking; and that even if it were to be considered at some earlier time, such as about 1700 on 9 March when the crew entered the life boats, the vessel was doomed to sink already then; and that at such earlier time the repair costs had not yet reached $3 million.
17. In a ruling delivered on 2 July 2002, at the conclusion of oral submissions on the question of a preliminary point, the judge decided that the trial should go ahead in the ordinary way and that there should be no preliminary issue. He was principally concerned that the full trial would be lost by reason of the necessity to reserve judgment and/or to allow for appeal on that issue, and so the witnesses would have to stand down and there would be a wastage of costs. He was not even prepared to hear submissions on constructive total loss at the outset of the trial. There was no appeal from that ruling.
18. The judge also ruled that the insurers should formally plead their newly developed defence to the constructive total loss claim. Their re-re-reamended defence is dated 8 July 2002. This disclosed (inter alia) the following addition to paragraph 5B:

"Further it is denied that the vessel was a CTL before she was an ATL by the Claimants being irretrievably deprived of the ship; that the Claimants elected to treat the ship a CTL before she became an ATL; and that the Claimants were entitled to treat the vessel as a CTL in the absence of abandonment before she became an ATL. Further the claimants by issuing the Claim Form herein claiming only an ATL elected irrevocably not to treat the fire damage before sinking as a CTL."

19. Thus there was a general plea that the actual total loss preceded any constructive total loss, and further that there was no abandonment of the vessel before she became an actual total loss. There was a new case of election barring the constructive total loss claim by reason of the issue of the claim form. There was a formal, but limited admission of a fire. The owners' pleaded details of the fire remained a matter for them to prove, and all that the insurers were prepared to admit was that "there was a fire in the engine room starting at about 14.20 hours".
20. This amendment resulted in a request for further information. On 19 July 2002, well into the trial, the further information was served. The document made it clear that in two respects the insurers' case was limited to the time "immediately prior" to the sinking. First, the experts' agreement that the vessel had suffered damage which would have cost more than $3 million to repair was glossed into an admission that that only became the case immediately before sinking, thus –

"It is admitted that (as agreed by the expert marine engineers) had the vessel not sunk the cost of damage repairs would probably have exceeded US$3 million, and therefore that, immediately prior to her sinking, the theoretical cost of repairing that damage would have exceeded the insured value of the vessel."

21. Secondly, however, the insurers, having been asked to clarify their pleading as to when they said the owners were irretrievably deprived of their vessel, now limited their case of when the vessel was bound to sink also to the point of time immediately prior to sinking, thus –

"Immediately prior to her sinking the Claimants were irretrievably deprived of their vessel because the event shows that, at that time, she was bound to sink."

22. It follows that the insurers had no case that the vessel was bound to sink or was irretrievably lost before the moment of her sinking, but at that moment (and that moment only) they admitted that repairs to the fire damage would have cost more than her value when repaired. They did not admit that the cost of such repairs exceeded her value before that point, but they had no positive case that the cost did not. They had also pleaded that the vessel was an actual loss by sinking before she had become a constructive total loss by fire; but it seems to us that that plea was narrowed by the further information.

The judgment below

23. Tomlinson J dealt with the constructive total loss claim at the outset of his judgment. The critical finding of fact, whether the vessel had become a constructive total loss by fire before the vessel had become or was doomed to become an actual total loss by sinking by reason of the entry of seawater, which had been left open on the parties' pleaded cases, was addressed in paragraph 19 as follows:

"Mr Berry pointed out that the admission by underwriters of the occurrence of the CTL related only to the scintilla temporis before sinking, at which time he said the vessel was doomed to become an actual total loss so that it was unreal to speak of her becoming a CTL before she was in fact an actual total loss. It is true that in their Amended Particulars of Claim the insured do not plead precisely when before the sinking the vessel became a CTL. The claimants' expert marine engineer, Mr Bevis, estimated the cost of repairs to damage caused by an uncontrolled fire burning for 15 hours or so as US$7.9m if repaired in Colombo, US$5.8m if repaired in the Far East. He made no allowance for salvage or contingency. Mr Todd had this report of Mr Bevis before preparing his own. Without condescending to detail he considered that the likely cost of repairing damage likely to have been sustained prior to sinking would have "appreciably exceeded US$3m". On this state of the evidence and bearing in mind the hypothesis that the fire was not causative of the sinking I do not believe that I am justified in concluding that the vessel was already an actual total loss or bound to become one at the moment at which she became a CTL. Some unexplained event occurring after that point may well have tipped the balance between sinking and not sinking. I do not believe that my conclusion in this regard on the particular facts of this case throws up the spectre of owners being able, in most cases of actual total loss, to claim years after the event as for a CTL occurring a moment earlier than the actual loss. The problem only arises in the very unusual case where underwriters do not admit that the fortuity which caused the CTL caused the actual loss. In a case where actual total loss immediately follows CTL, such a situation must be very rare. A more realistic example would be a vessel which has become a CTL by fire thereafter being struck by an Exocet missile which causes her to sink. I cannot see why in such circumstances the insured should not recover from their marine risks underwriters…The fact is that whereas the Exocet example is demonstrably two casualties and two losses, so the present case less obviously but for the purposes of the argument must likewise be regarded as two casualties and two losses. The premise upon which I am considering a claim for CTL is that the actual total loss was brought about by a coincidental but unconnected cause."

24. Since the judge was dealing with the constructive total loss claim first, he had to assume, against the owners but in favour of the insurers' case, that the actual total loss by sinking was not due to the fire but to the independent but unexplained cause of entry of seawater. Hence he spoke of his premise that the sinking had been brought about "by a coincidental but unconnected cause". By the end of his judgment, he had agreed with the insurers' case on actual total loss.
25. On the factual basis thus explained, the judge rejected the insurers' submissions on constructive total loss. It followed necessarily that the constructive total loss by fire preceded the actual total loss by sinking by a material time in the sense that it occurred at a time when the vessel was not bound to become an actual total loss. It is common ground, and is said to have been so below, that no notice of abandonment needed to be served (section 62(7)). In any event the judge held that where, as in this case, there was simply no opportunity to give notice of abandonment before the vessel had become an actual total loss by reason of another peril, the CTL claim could be maintained without giving such notice. In the circumstances the judge also held that there was nothing else that the owners needed to do as a condition precedent to claiming for the constructive total loss. The constructive total loss had become an accrued cause of action prior to the actual total loss, and the actual total loss did not prevent that cause of action being maintained. There were two losses, and there was no merger of the first with the second. The doctrine of merger of unrepaired partial loss with a subsequent total loss (see section 77(2) of the Act) did not apply to a prior unrepaired constructive total loss overtaken by an actual total loss. Therefore, in the absence of merger, it could not be said that the owners had suffered no loss by fire, for the loss was complete before the actual loss by sinking. The owners had never treated the constructive total loss by fire as a partial loss and had not elected to do so by reason of originally claiming only for an actual total loss. In either case the claim was for a total loss and involved the owners in ultimately abandoning the vessel to insurers against payment. Finally, the judge dealt with an additional defence which insurers raised pursuant to a specific clause of the policy, clause 18.2, which said that in no case should the insurers be liable for "unrepaired damage in the event of a subsequent total loss": but the judge said that the claim was not for unrepaired damage but for a total loss.

The Marine Insurance Act, 1906

26. Before turning to the insurers' detailed submissions on this appeal, it is first necessary to set out the relevant provisions of the Act.

Partial and total loss

56. (1) A loss may be either total or partial. Any loss other than a total loss, as hereinafter defined, is a partial loss.

(2) A total loss may be either an actual total loss, or a constructive total loss…

(3) Where the assured brings an action for a total loss and the evidence proves only a partial loss, he may, unless the policy otherwise provides, recover for a partial loss.

Actual total loss

57. (1) Where the subject-matter insured is destroyed, or so damaged as to cease to be a thing of the kind insured, or where the assured is irretrievably deprived thereof, there is an actual total loss.

(2) In the case of an actual total loss no notice of abandonment need be given…

Constructive total loss defined

60. (1) Subject to any express provision in the policy, there is a constructive total loss where the subject-matter insured is reasonably abandoned on account of its actual total loss appearing to be unavoidable, or because it could not be preserved from actual total loss without an expenditure which would exceed its value when the expenditure has been incurred.

(2) In particular, there is a constructive total loss –

(i) Where the assured is deprived of the possession of his ship or goods by a peril insured against, and (a) it is unlikely that he can recover the ship or goods, as the case may be, or (b) the cost of recovering the ship or goods, as the case may be, would exceed their value when recovered; or

(ii) In the case of damage to a ship, she is so damaged by a peril insured against that the cost of repairing the damage would exceed the value of the ship when repaired…

Effect of constructive total loss

61. Where there is a constructive total loss the assured may either treat the loss as a partial loss, or abandon the subject-matter insured to the insurer and treat the loss as if it were an actual total loss.

Notice of abandonment

62. (1) Subject to the provisions of this section, where the assured elects to abandon the subject-matter insured to the insurer, he must give notice of abandonment. If he fails to do so the loss can only be treated as a partial loss.

(2) Notice of abandonment may be given in writing, or by word of mouth, or partly in writing and partly by word of mouth, and may be given in terms which indicate the intention of the assured to abandon his insured interest in the subject-matter insured unconditionally to the insurer.

(3) Notice of abandonment must be given with reasonable diligence after the receipt of reliable information of the loss, but where the information is of a doubtful character the assured is entitled to a reasonable time to make inquiry.

(4) Where notice of abandonment is properly given, the rights of the assured are not prejudiced by the fact that the insurer refuses to accept the abandonment.

(5) The acceptance of an abandonment may be either express or implied from the conduct of the insurer. The mere silence of the insurer after notice is not an acceptance.

(6) Where a notice of abandonment is accepted the abandonment is irrevocable. The acceptance of the notice conclusively admits liability for the loss and the sufficiency of the notice.

(7) Notice of abandonment is unnecessary where, at the time when the assured receives information of the loss, there would be no possibility of benefit to the insurer if notice were given to him.

(8) Notice of abandonment may be waived by the insurer.

(9) Where an insurer has re-insured his risk, no notice of abandonment need be given by him.

Effect of abandonment

63. (1) Where there is a valid abandonment the insurer is entitled to take over the interest of the assured in whatever may remain of the subject-matter insured, and all proprietary rights incidental thereto…

Total loss

68. Subject to the provisions of this Act and to any express provision in the policy, where there is a total loss of the subject-matter insured –

(1) If the policy be a valued policy, the measure of indemnity is the sum fixed by the policy…

Partial loss of ship

69. Where a ship is damaged, but is not totally lost, the measure of indemnity, subject to any express provision in the policy, is as follows:-

(1) Where the ship has been repaired, the assured is entitled to the reasonable cost of the repairs, less the customary deductions, but not exceeding the sum insured in respect of any one casualty…

(2) Where the ship has not been repaired, and has not been sold in her damaged state during the risk, the assured is entitled to be indemnified for the reasonable depreciation arising from the unrepaired damage, but not exceeding the reasonable cost of repairing such damage, computed as above…

Successive losses

77. (1) Unless the policy otherwise provides, and subject to the provisions of this Act, the insurer is liable for successive losses, even though the total amount of such losses may exceed the sum insured.

(2) Where, under the same policy, a partial loss, which has not been repaired or otherwise made good, is followed by a total loss, the assured can only recover in respect of the total loss…

Right of subrogation

79. (1) Where the insurer pays for a total loss, either of the whole, or in the case of goods of any apportionable part, of the subject-matter insured, he thereupon becomes entitled to take over the interest of the assured in whatever may remain of the subject-matter so paid for, and he is thereby subrogated to all the rights and remedies of the assured in and in respect of that subject-matter as from the time of the casualty causing the loss.

(2) Subject to the foregoing provisions, where the insurer pays for a partial loss, he acquires no title to the subject-matter insured, or such part of it as may remain, but he is subrogated to all rights and remedies of the assured…

27. It will be observed that the Act provides that a loss may be either total or partial; that any loss other than a total loss is a partial loss; that, as stated above, a total loss may be either actual or constructive, but both are forms of total loss; that a constructive total loss is defined (in section 60) separately from the requirement of notice of abandonment (section 62); that the effect of a constructive total loss may depend on how it is treated by the assured, so that if the assured treats it as a partial loss (section 61), or if the assured fails to give a notice of abandonment save where one is excused (section 62(7), (8) and (9)), then it can only be treated as a partial loss (section 62(1)); that where an assured wishes to claim a constructive total loss he must be prepared to abandon the vessel to the insurer so that the insurer can take the benefit of whatever remains (sections 61 and 63), and that the insurer who pays for an actual total loss is entitled to the same benefit (section 79(1)); and that an insurer is liable for successive losses save in the case where an unrepaired partial loss is followed by a total loss, in which case the assured can recover only in respect of the total loss (section 77(2)).

The insurers' submissions

28. In his skeleton argument for this appeal Mr Berry on behalf of the insurers makes submissions under seven separate headings, although it would appear that some of the headings are multi-stranded and others are perhaps different ways of expressing the same thought.
29. First, Mr Berry submits that there was no abandonment under section 61 and thus can be no claim for a constructive total loss. It is vital, he submits, to distinguish the election to abandon or abandonment under section 61 from the additional and separate requirement of notice of abandonment under section 62. Although he accepts, or now accepts, that in this case there was no need for a notice of abandonment and that section 62(7) applies, he submits that the owners never got to first base in the absence of an initial abandonment, which he describes as a "mere mental decision" or "making up his mind" (see Court Line, Ltd v. The King (1945) 78 Ll L Rep 390 at 397). This is the most protean of Mr Berry's submissions. It sometimes appeared as something almost factual, premised on the consideration that the vessel was an actual total loss before the owners had an opportunity to consider their reaction to events. Sometimes it reflected the different factual consideration that the owners' initial claim was for an actual total loss and the contention that such a claim was inconsistent with any election to abandon since it presupposed that the owners were continuing to treat the vessel as their own right down to her actual loss by sinking, rather than as abandoned to insurers. Sometimes it appeared to be premised on the assertion that it is always too late to claim a constructive total loss after an actual total loss, since an assured must always be in a position to abandon the vessel to underwriters while it is still in being or in specie, which it ceases to be after its actual loss: for this purpose Mr Berry emphasised the language of section 61 in speaking of abandoning "the subject-matter insured".
30. Secondly, Mr Berry submitted that there was only one casualty, the actual loss by sinking. On the premise that the loss by fire was only separated from the loss by sinking by a scintilla temporis, he disputed the judge's finding of separate losses and separate casualties as wholly artificial.
31. Thirdly, he submitted that there was in any event no proof of a prior constructive total loss. He again relied on the premise that only a scintilla temporis separated the constructive from the actual total loss. It therefore followed that at that time the vessel was already bound to sink and an actual total loss. The owners had borne the burden of proving a materially prior constructive total loss, and had not even sought to assert one, let alone managed to prove it.
32. Fourthly, the owners had suffered no loss, since their true loss resulted from the vessel's sinking, an event which was not covered by the policy.
33. Fifthly, the constructive total loss was in any event excluded by clause 18.2 of the Institute Clauses incorporated into the policy. The judge had simply erred in finding that unrepaired damage due to a constructive total loss was not within the clause.
34. Sixthly, Mr Berry submitted that the common law doctrine of merger of unrepaired damage, reflected in but extending wider than section 77(2) of the Act, meant that even unrepaired damage which amounted to a constructive total loss and went beyond mere partial loss, was merged into and swallowed up by a subsequent total loss, so that an insured's recovery depended entirely on whether that total loss was covered by the policy or not. He relied on Livie v. Janson (1810) 12 East 647, Woodside v. Globe Marine [1896] 1 QB 105, and British & Foreign Insurance Co Ltd v. Wilson Shipping Company Limited [1921] 1 AC 188.
35. Seventhly and finally, he submitted that the owners' initial limitation of their claim to one of actual total loss was an irrevocable election not to claim on the basis of a constructive total loss which barred their subsequent attempt at amendment. It was put as matter of estoppel, but also reflected one of the strands of his first submission.
36. We will revert to these submissions after considering the principal authorities relied on before us.

The authorities

37. The earliest case relied on was Hamilton v. Mendes (1761) 2 Burr 1199 which concerned the ship Selby. She was first captured by a French privateer, then retaken by an English man-of-war and brought safely to England. Thereafter the owner abandoned the ship to his underwriter, who declined the offer but said he was ready to pay salvage or other losses. The ship herself was undamaged. The issue, as stated by Lord Mansfield, was –

"whether the plaintiff, who at the time of his action brought, at the time of his offer to abandon, and at the time he was first apprized of any accident having happened, had only, in truth, sustained an average-loss, ought to recover for a total one."

38. The answer given, which remains the law today, is that, unless an offer of abandonment had previously been accepted by the insurer, no greater loss can be claimed than exists at the time action is brought. If, therefore, a ship which had been lost is recovered before action is brought, the total loss claim is defeated, although a claim may remain for a partial loss: see Polurrian Steamship v. Young [1915] 1 KB 922. The reason is that given by Lord Mansfield (at 1210) namely that –

"It is repugnant, upon a contract of indemnity, to recover as for a total loss, when the final event has decided that the damnification, in truth, is an average, or perhaps no loss at all."

39. In the present case, however, there was no recovery, and what Mr Berry has rather sought to use the case for is to respond to the submission that there was a vested right to sue for a total loss immediately upon the capture. Lord Mansfield said (at 1211) that such an idea –

"is fictitious only, and not founded in truth. For the insured is not obliged to abandon, in any case: he has an election. No right can vest as for a total loss, till he has made that election. He can not elect before advice is received, of the loss: and if that advice shews the peril to be over, and the thing in safety, he cannot elect at all; because he has no right to abandon, when the thing is safe."

40. Mr Berry submits that this shows the importance of the initial election to abandon under what is now section 61 of the Act. Just as it is too late to abandon a vessel after she is safe, so, he submits, it is too late to abandon her after she has become an actual total loss. It was therefore wrong of the judge to speak as he did (at para 9) of "an accrued cause of action" for constructive total loss.
41. We would comment that in a case where notice of abandonment was served, it could well be mistaken to discriminate between an election to abandon and the notice itself; that the language of Lord Mansfield's judgment shows that abandonment is also spoken of as an "offer to abandon"; and that it does not necessarily follow from the fact that a recovery of the subject matter insured will defeat a subsequent claim for a constructive total loss that its destruction will do the same. In the former case the assured's need for an indemnity has been answered whereas in the latter case he would seem to remain in need of one.
42. In Livie v. Janson (1810) 12 East 647 the American ship Liberty was first stranded, seized the next day by the American authorities for breaking an embargo, then refloated, and ultimately condemned by the authorities for that breach. The policy contained a warranty free from American condemnation. It was held that the assured could not recover for the partial loss caused by the stranding since that was overtaken by the total loss, nor for the total loss by condemnation since that was an excepted peril. Lord Ellenborough CJ said –

"The object of a policy is indemnity to the assured; and he can have no claim to indemnity where there is ultimately no damage to him from any peril insured against" (at 654)

and also –

"In the present case, as the immediately operating cause of total loss was one from which and its consequences the defendant is by express provision in the policy exempted; and as the other antecedent causes of injury never produced any pecuniary loss to the plaintiff; and as there never existed a period of time, prior to the total loss, in which the assured could have practically called on the underwriters for an indemnity against the temporary and partial injury sustained by the property insured; we are of opinion that such prior partial injury forms in this case no claim upon the underwriters of this policy…" (at 656).

43. Although counsel in that case sought to argue that the vessel might have been viewed as a total loss by stranding, it is plain that Lord Ellenborough was prepared to consider her as only partially damaged. Since that damage was unrepaired, the assured was not out of pocket, as he might have been if he had made prior "Actual disbursements for repairs" (at 655). Thus in Le Cheminant v. Pearson (1812) 4 Taunt 367, where the partial loss was repaired prior to total loss, the assured recovered for both. The same issue of unrepaired partial loss arose again in British and Foreign v. Wilson Shipping (see below).
44. However, the alternative case of a total loss by perils of the sea followed by a seizure was considered in Hahn v. Corbett (1824) 2 Bing 206, where Livie v. Janson was distinguished. It concerned a policy on goods rather than on ship. The goods were carried in a ship which stranded and was totally lost. The master and crew having escaped to Fort San Carlos (in the West Indies), he applied the very next day after the stranding for assistance to enable him to save the cargo, but found that the Fort had been overrun by the Spanish, who imprisoned him and the crew and immediately went to unload the cargo as prize of war. The cargo was partly sound and partly damaged. Intelligence of the loss did not reach England for several months, and the assured then gave notice of abandonment to underwriters. The policy on goods contained an exception against capture and seizure. It was held that the loss was by perils of the sea and not by capture or seizure and the assured recovered. Best CJ said (at 210):

"But what are the facts here? the ship was totally lost; and though the goods are afterwards recovered in a damaged state, they fall into the hands of an enemy, and, but for that circumstance, would also have been totally lost. It has been asked when the goods can be said to have been lost? – when the ship was totally lost…It was just as if they had been cast on a rock, and had been completely out of reach. The goods, therefore, were lost when the ship was lost, and what happened afterwards makes no difference in the case…The case of Livie v. Janson is materially distinguishable from the present. In that case, though, perhaps, the facts would have warranted the statement of a total loss, the loss is stated not to have been total…"

45. The question might arise whether this was a case of constructive or actual total loss by reason of the stranding. The question was not explicitly addressed in the judgments, but the facts stated in the report would suggest that it was a case of constructive total loss, and it was as such that Tomlinson J treated it. Arnould, Law of Marine Insurance and Average, 16^th ed, 1981 at para 1180 footnote 74 would likewise seem to treat it as a case of constructive total loss. The case report ends with a note by the reporter, Peregrine Bingham, whose reports of this period are well regarded, to the effect that "The same judgment was given in another case, which only differed from the present, in the circumstances that the goods were undamaged" (at 288). That again suggests that the total loss was a constructive total loss. We would accept, however, that it is by no means clear that that is how the court viewed the matter. On the basis that the loss of cargo was a constructive total loss, Hahn v. Corbett is an authority against the insurers.
46. Perhaps another way of looking at Hahn v. Corbett is to ask whether the proximate cause of the loss was the stranding or the capture. That was the ground relied on expressly by Burrough J, and it is also the explanation of the case given in Andersen v. Marten [1908] 1 KB 601 at 608 (CA). Subject to the formalities connected with constructive as opposed to actual total losses, since Kastor Too was, commercially speaking, lost by fire at a material time prior to her sinking, there seems to be no difficulty in concluding that she was proximately lost by fire, even if, as the judge also found, the fire was not the proximate cause of the sinking. In Andersen v. Marten [1908] AC 334, the House of Lords emphasised that the ship there was a total loss by capture before she became a wreck by perils of the seas: and in Pesquerias y Secaderos de Bacalao de Espana SA v. Beer (1945) 79 Ll L Rep 417 at 434, Atkinson J said "That was, of course, a constructive total loss by capture."
47. Roux v. Salvador (1836) 3 Bing (NC) 267 is indisputably a case which discusses the distinction between an actual and a constructive total loss, and is a leading case on the distinction. It again concerned an insurance on goods, namely hides shipped from Valparaiso to Bordeaux and warranted free of particular average. On the voyage the hides began to putrefy as a result of a leak in the ship. They were sold for a quarter of their value at Rio de Janeiro on the basis that they would have been totally destroyed before they reached their destination in Bordeaux. The assured only heard of their condition at the same time as he heard of their sale. There had therefore been no abandonment before the sale. However, it was argued on behalf of the insurer that since the hides had not been an actual total loss at the time of sale, the assured could not throw on the insurer by a premature sale a liability as for a total loss. Moreover, even if the hides had been a constructive total loss at the time of sale, they were represented by the salvage price which they obtained on their sale and there ought to have been a notice of abandonment of that salvage, which there had not been. The court of common pleas accepted these or similar arguments and gave judgment for the insurer, but the exchequer chamber gave judgment instead for the assured. The basis of that judgment was that the hides were already doomed at the time of sale to a process of putrefaction which could not be reversed and which would have resulted in their complete destruction, their ceasing to be in specie, before the end of the voyage: they were therefore already at the time of sale what would now be called an actual (there an absolute) total loss and no question of abandonment arose.
48. Mr Berry seeks, we think, to rely on this case as suggesting that, if the exchequer chamber had found that the loss was only a constructive one, then the claim would have failed for lack, not of notice of abandonment, but of any election to abandon, of abandonment itself, prior to an actual total loss. Thus he referred to the concept discussed in the judgment of Lord Abinger CB (at 277/8) of converting a partial loss into a total loss by means of abandonment. However, Lord Abinger is there discussing a submission which he in any event rejects; moreover it seems to us that "abandonment" is often used as a shorthand for "notice of abandonment". Mr Berry also appears to rely on Lord Abinger's discussion for the submission that the abandonment necessary to a constructive total loss has to take place while the thing insured is still in specie. We think there is no such principle. On the contrary, the different point is made that if, in the absence of actual total loss, an insured wishes to rely on a constructive total loss of something which therefore still exists, in whole or part, then –

"the very principle of the indemnity requires that he should make a cession of all his right to the recovery of it, and that too, within a reasonable time after he receives the intelligence of the accident" (at 286).

49. In other words it is simply assumed that, since ex hypothesi the thing is not an actual total loss, it still exists. Nothing in Roux v. Salvador in any event deals with the different point which is the origin of section 62(7) of the Act as to what if anything needs to be done when a notice of abandonment would be of no benefit to the insurer, as for instance when the thing insured has already been actually as well as constructively lost before intelligence is heard of any casualty.
50. That question was discussed in Rankin v. Potter (1873) LR 6 HL 83. It was a case in which the judges were called upon to advise the House of Lords, and it is therefore particularly rich in dicta. It concerned a policy on freight payable for a voyage from Calcutta to England. The ship concerned was damaged on her outward voyage and temporarily repaired in New Zealand: at Calcutta, however, the charterer refused to load on the ground of the vessel's condition and the owner then examined the vessel further and concluded that the ship was a constructive total loss, whereupon he gave notice of abandonment to her underwriters. The question was whether the owner could recover for the loss of his freight. No separate notice of abandonment was given to the underwriters of the freight. For present purposes it is the second of the four questions posed by Lord Chelmsford at 153 that is most relevant, namely whether notice of abandonment, either of freight or ship or both, was necessary in order to recover for a total loss of freight. It is not entirely clear whether that question assumed that the answer to the first question, which was whether there was an actual total loss of freight, was no; or whether the submission of the underwriter was that a notice of abandonment was necessary in every case of total loss, whether constructive or actual. In any event, if the answer to the second question was that a notice of abandonment (of the freight) was unnecessary, then the absence of such a notice, and the submission that the notice of abandonment of ship (if necessary at all for recovery of an indemnity for the freight) was invalid either because it was too late or because the owner had already treated the damage as a partial loss (the subject matter of a third and fourth questions), were equally irrelevant.
51. The House of Lords held, first, that the loss of freight was an actual total loss. As to the second question, they held that whether the loss of freight had been an actual or a constructive total loss, no notice of abandonment was necessary because it could not have benefited the underwriter. It could not have benefited the underwriter because the facts remained that the vessel was unable to perform the voyage, the cargo had not been loaded, there was no right to substitute the ship, and thus there was simply no way in which the freight could have been earned. In the circumstances definitive answers to the remaining questions were unnecessary.
52. It may also be observed that there was in fact a constructive total loss of the ship (and thus an actual loss of the freight), even if the position had been that, on the arguments raised on the third and fourth questions, the owner had been unable to claim in respect of such a loss of the ship: see Lord Chelmsford at 155. That emphasises the distinction drawn between the definition of a constructive total loss (section 60) and the right to claim for one (sections 61 and 62).
53. The case also contains dicta which emphasise the ambiguous way in which "abandonment" is used to cover both cession to underwriters and the notice of abandonment: see for instance per Blackburn J at 118, per Martin B at 144 and per Lord Chelmsford at 156. The dictum of Martin B at 144 possibly contains the clearest definition of the concept of abandonment at common law. It was referred to in Royal Boskalis Westminster NV v. Mountain [1997] 1 LRLR 523 at 556 and by Tomlinson J below (at para 14), namely –

"…a cession or transfer of the ship to the underwriter, and of all his property and interest in it, with all the claims that may arise from its ownership, and all the profits that may arise from it, including the freight then being earned. Its operation is as effectually to transfer the property of the ship to the underwriter as a sale for valuable consideration, so that of necessity it vests in the underwriter a chattel of more or less value, as the case may be."

54. There is nothing here however to support Mr Berry's submission that even where a notice of abandonment is unnecessary there nevertheless has to be, and therefore has to be time for, the making of some mental decision amounting to an election to abandon or abandonment.
55. Rankin v. Potter also contains, in the advice tendered by Blackburn J at 120, a good example of how an assured might prevaricate in the treatment of a loss as partial or total and of the role of a notice of abandonment for policing such matters:

"If before giving this notice the assured have already indicated by their acts, or if the circumstances are such that they indicate by their silence, that they have elected to adhere to the adventure as their own, the notice of abandonment obviously comes too late. A very good example of such a case is afforded by Mitchell v. Edie 1 T.R. 608, as explained in Roux v. Salvador 3 Bing, N.C. 266. There a ship laden with sugar, and bound for London, was captured and finally taken into to Charleston, where the sugar was sold and the proceeds lodged in the hands of a person resident in Charleston. From the state of political affairs at that time, sugar was dear at Charleston, and, as Lord Abinger conjectured, the sugar had come to a very good market, and the assured was satisfied, and took the proceeds. A year afterwards, the person in whose hands the money was, became insolvent, and after that it was, with obvious justice, held that it was too late to come upon the underwriters for a total loss."

See also in this connection Royal Boskalis at 557/8.

56. In Kaltenbach v. Mackenzie (1878) 3 CPD 467 the ship, which had been insured for £4000, stranded, was immediately refloated and returned to her loading port of Saigon for inspection. The conclusion was that she was a constructive total loss and at trial the assumption was made that that was so. The owner knew of this by 7 February, but did not act promptly in giving notice of abandonment. Instead, the vessel was sold on 23 February, repaired for £50 and sailed to Singapore, there resold and repaired for a further £500. All this happened before notice of abandonment was given to underwriters on 11 March. It was argued that even if notice had been given, the underwriter would not have had time to stop the sale and that the case was therefore within the doctrine of Rankin v. Potter that a notice of abandonment was in such circumstances unnecessary. It was held, however, that the owners should have acted promptly and that there had been "a fatal gap" (at 478) and the claim therefore failed. It was not for the courts to speculate on how much use the underwriters would have made of the notice. It was not as though the assured had set out to prove that by the time the underwriter would have been able to act on a prompt notice the subject-matter of insurance would have absolutely perished, a point the relevance of which in any event should be left open (at 475).
57. In the course of his judgment Brett LJ included an extended passage at 470/475 which has become a classic explanation of the concepts of abandonment and notice of abandonment. Tomlinson J set out the full passage at paragraph 16 of his judgment. In the circumstances we would merely emphasise the following matters.
58. First, Brett LJ made it plain that whereas a notice of abandonment was, together with the doctrine of constructive total loss, a peculiarity of marine insurance, the doctrine of abandonment on the other hand was endemic to all total losses, constructive or total, and to all contracts of indemnity. Thus he said (at 471):

"Whenever, therefore, there is a contract of indemnity and a claim under it for an absolute indemnity, there must be an abandonment on the part of the person claiming indemnity of all his right in respect of that for which he receives indemnity…Abandonment, however, is applicable to the claim, whether it be for an actual total loss or for a constructive total loss…Or where goods have been totally lost, as in the case of Roux v. Salvador, but something has been produced by the loss, which would not be the goods themselves, if it were of any value at all, it must be abandoned. But that abandonment takes place at the time of settlement of the claim; it need not take place before."

59. Secondly, Brett LJ explained the rule in Rankin v. Potter, changing the facts to fit the facts of his case, in these terms (at 474):

"In Rankin v. Potter the law was established that where at the time when the assured receives information which would otherwise oblige him to give notice of abandonment, at the same time he hears that the subject-matter of the insurance has been sold so as to pass the property away, inasmuch as there was nothing of the subject-matter of the insurance which he could abandon, notice of abandonment was not necessary...In those particular circumstances it was held that notice of abandonment need not be given because there was nothing to abandon. That is in one sense true; but if goods had been sold it is obvious there must be something to abandon, that is the proceeds of the sale; the money which is the proceeds of the sale, when the insurance is settled, is abandoned; but where there is nothing of the subject-matter of insurance to abandon, there is no ship to abandon, there are no materials of the ship to abandon, there are no goods to abandon, notice of abandonment under those circumstances was said to be futile."

60. It will be observed that those final remarks apply to this appeal; and also that the language "nothing of the subject-matter of insurance to abandon" may well have directly informed the statutory language of section 61's "abandon the subject-matter insured". In other words, the language does not make the survival of the subject-matter insured a condition for the existence of a claim for a constructive total loss: rather it is there to distinguish the case of treating the subject-matter insured in a particular way (as a partial or alternatively constructive total loss) and the case where nothing of the subject-matter remains for treatment in any way.
61. Thirdly, a remark of Cotton LJ in Kaltenbach at 481/2 was referred to by Tomlinson J as being relevant to the instant appeal. Cotton LJ said:

"I give no opinion on the question which arises when the state of the thing insured is such that before the communication could have reached the underwriters it must, so far as human probability goes, have ceased to be in specie."

The judge said (at para 16) that Cotton LJ had there left open "the very point which arises in this case". We would respectfully doubt that that is so. Our case is dealt with by Brett LJ as being a case within the rule in Rankin v. Potter, viz the case of a constructive total loss (prima facie requiring a notice of abandonment) overtaken by the complete destruction of the subject-matter of insurance before the time when the insured first learns of the loss: in such a case the insured is relieved of any obligation to give notice of abandonment in order to claim. What Cotton LJ is talking about, on the other hand, is we think a case where, even though the thing insured still existed as at the time when a notice of abandonment would otherwise need to be given, the assured could prove that it would have perished before anything could be done about it by the underwriters, the same case as that left open by Brett LJ at 475.

62. Fourthly, Brett LJ explained the unique marine insurance requirement for a notice of abandonment in cases of constructive total loss as deriving from a consensually recognised need in the international and uncertain context of a marine adventure to protect underwriters from an insured's attempt to play the market: see at 472 where he said:

"The underwriter in general can receive no notice of what has occurred, unless from the assured, who is the owner of the ship or the owner of the goods and there would therefore be great danger if the owner of a ship or of goods – that is the assured – might take any time that he pleased to consider what the state of the market might be, or many other circumstances, and would throw upon the underwriter a loss if the market were unfavourable, or take to himself the advantage if the market were favourable. These are the reasons why I think the assured and underwriters came to the conclusion that it should be a part of the contract and a condition precedent that, where the claim is for a constructive total loss, there must be notice of abandonment, unless there were circumstances which excused it."

63. That is consistent with the opinion expressed by Blackburn J in Rankin v. Potter (see at para 55 above). It would seem to follow from the views of these great commercial judges that, where a vessel's actual loss follows closely upon her constructive loss, so that there would be no benefit to underwriters from a notice of abandonment and thus no need for one and equally no opportunity for the assured to play the market by prevaricating between treating his loss as partial or total, there could then be no objection in practice to the assured claiming for a constructive total loss despite the subsequent actual total loss. Indeed, unless the constructive loss has to be viewed as simply swallowed up in the subsequent actual loss, or, as Mr Berry would put it, merged in that subsequent loss, a constructive total loss is a complete loss of the vessel so far as the assured is concerned, so that any subsequent total loss by a further peril, whether that loss is constructive or actual, simply becomes irrelevant.
64. Woodside v. Globe Marine Insurance Company Limited [1896] 1 QB 105 is an important authority for Mr Berry's purposes, relied on by him as answering the issue in this appeal in his favour. The ship Bawnmore was insured against loss by fire. She stranded, and an assumption was made at the request of the underwriters that she was a constructive total loss. However, the assured at no time treated her as one, and indeed did not admit that she was. Thirty-six hours after she stranded she was destroyed by fire. The assured claimed a total loss by fire, the underwriter defended on the ground that she had already been lost by stranding, for which (the case proceeded on the basis that) she was not covered. Matthew J said (at 106/8):

"It was ordered that the question should be argued as a preliminary point of law on the assumption that the vessel when stranded was still capable of being floated and repaired, although at an expense exceeding her repaired value. Upon the argument it was agreed by both sides that the case must be dealt with as if the owners were uninsured against the perils of the seas which caused the stranding…For the defendants it was argued that if the fire had occurred first, and the damage had not been repaired, the underwriters would not have been liable. The damage by fire, it was said, was, as it were, merged in the previous total loss. The assured must be treated as if they had been indemnified and had received the value of their ship, and they ought not to be permitted to recover twice over for what was one loss.

"I am of opinion that the plaintiffs are entitled to judgment. The loss by stranding would only become total if the assured gave timely notice of abandonment. If none were given, the loss would be a particular average, and it would seem clear law that a particular average loss, however serious, could not impair the right of the assured to recover for a subsequent total loss upon the basis of the valuation…

"In the case of a ship, in the same way, the vessel may from many causes be worth much less at the time of loss than the agreed value; but the valuation determines the amount of the underwriters' liability…

"I am unable to distinguish such a loss as the present from any other in which, while the subject-matter of insurance still exists and there is no suggestion of fraud, a diminution in its value is relied upon to exonerate the underwriter from the whole or from part of his liability. It is impossible to draw a line between a greater or less diminution of value. I therefore give judgment for the plaintiffs."

65. Thus the judgment proceeded on the basis that, although the loss by stranding was assumed to be a constructive total loss, it had never been treated as such by the assured by service of a timely notice of abandonment. It also seems to have been assumed that there would have been time for such a notice. The decision does not appear to have involved at all any argument that a notice of abandonment was unnecessary on Rankin v. Potter (ultimately section 62(7)) grounds. Therefore, despite the assumption of a constructive total loss, the judge treated the unrepaired damage as merely partial loss. He did not expressly put the case on the basis of the doctrine of merger of the partial loss in the total loss (ultimately section 77(2)), but rather on the irrelevance of depreciation in value by reason of unrepaired damage in the case of an agreed value. It would seem that if the value had not been agreed, then the answer given might have been different.
66. Mr Berry submits that Woodside is authority in support of the insurers' position in the present case. Here, however, the owners have treated the loss by fire as a constructive total loss, if they are permitted to do so, and it is common ground that there was no requirement to serve a notice of abandonment. It therefore seems to us that, whatever be the proposition for which it may stand, Woodside is not authority for the present case. It may be noted that in Arnould at para 1180, where the learned authors consider the problem of successive total losses, Woodside is cited under the second of the three rules proposed, whereas Mr Berry sought to bring the present case under the third rule (to which we will return below). Thus Woodside is there cited only for the following proposition:

"Secondly, if the first casualty was not covered by insurance, or if notice of abandonment was not given in due time, the fact that the operation of the first peril produced a situation of constructive total loss, although not recoverable as such, does not preclude a claim from being made in respect of the second casualty, assuming it to be caused by an insured peril."

Such a rule would not cover the present case.

67. The question of unrepaired partial loss was considered further in British and Foreign Insurance Company Limited v. Wilson Shipping Company Limited [1921] 1 AC 188. We are now, for the first time, in a period after the 1906 Act. The case was concerned with what were in effect two separate covers on the ship Eastlands under charter to the Admiralty. An insurance policy covered marine risks only, and war risks were covered by an Admiralty indemnity, but there was no agreed value. Partial loss by reason of marine perils was suffered and part of that damage was left unrepaired. During the currency of the marine policy the vessel suffered a total loss by war risks. The Admiralty deducted £1770 which was the estimated cost of repair of the unrepaired damage from the sound value of the ship. The owner now tried to recover the sum of £1770 from the marine underwriters. The argument centred around the precise language of section 77(2), for that was limited to the case of unrepaired partial loss followed by a total loss "under the same policy". However, section 91(2) also provided that the rules of the common law should continue to apply save in so far as they were inconsistent with the express provisions of the Act. The question was therefore as to the principles of the common law, and they were to be found in the principle of indemnity of which Livie v. Janson was one example. As applied to the particular case of unrepaired partial loss followed by a total loss before the expiry of the policy those common law principles had produced the following rule, per Lord Birkenhead LC at 199, namely –

"When a vessel, insured against perils of the sea, is damaged by one of the risks covered by the policy and before that damage is repaired she is lost, during the currency of the policy, by a risk which is not covered by the policy, then the insurer is not liable for such unrepaired damage."

It was not necessary therefore for the partial loss and the total loss to be suffered "under the same policy" and the common law rule was in this respect wider than its statutory embodiment.

68. Mr Berry submits that Lord Birkenhead's rule, applied in that case in circumstances wider than the specific provisions of section 77(2), is wide enough to encompass the case of an unrepaired constructive total loss followed by an actual total loss, as in the present case. The fact, therefore, that section 77(2) may refer to and be limited to "a partial loss" is not determinative. The rule can extend to a constructive total loss, for that too comes within the language of Lord Birkenhead's rule "When a vessel…is damaged…and before that damage is repaired she is lost".
69. We would recognise that if a constructive total loss has been treated as a partial loss, either because that is what the owner has elected to do or because, in a case where any claim for a constructive total loss can only be made good by service of a timely notice of abandonment and no such notice has been served, the loss can only be treated as a partial loss, then the case is within both the language of section 77(2) and, if it is necessary to say this which we doubt, within the rationale of any wider common law rule laid down in British & Foreign Insurance v. Wilson. If therefore for any reason, such as Mr Berry's first or seventh submission, the owners in the present case are not entitled to treat the vessel's loss by fire as a constructive total loss, then it must fall to be treated as a partial loss (see sections 61 and 62(1)). However, it seems to us that all this is to beg the essential question which, subject to those arguments, arises in the present case, which is whether a constructive total loss which the assured is entitled to treat as such is nevertheless to be treated as within the merger rule dealing with unrepaired damage. On that question Mr Berry prays in aid British & Foreign Insurance v. Wilson: but we consider that the House was there simply not considering even a potential case of constructive total loss, nor were any of the common law authorities there reviewed concerned with such a case. We have already said above that Livie v. Janson must be viewed as a case of partial loss, and it was for a rule concerned with partial loss followed by total loss that it was so regarded by their Lordships: see for instance Lord Birkenhead at 194 and Lord Sumner at 211/212.
70. It is true that in one sense the owner in British & Foreign Insurance v. Wilson was unfortunate, which is why a strong court of appeal (Bankes, Warrington and Scrutton LJJ) found in his favour [1920] 2 KB 285. The owner suffered a total loss, but received £1770 less than the full value of his ship! That does not seem like a full indemnity. His difficulty, however, was that he fell between two stools: he had "insured" with the Admiralty for war risks not on terms of agreed value, but on terms which permitted the Admiralty to pay only for the actual value of the vessel as she was at the date of loss, and that value was depreciated from her sound value by the cost of repairing the partial loss, for which the Admiralty was not responsible. On the other hand the rule which had been established in marine insurance was that there was no claim for unrepaired partial loss until the end of the policy period. We suppose a different rule might have been established, but the rule makes good sense in a relationship where a ship is usually insured at an agreed value which is paid on total loss whatever might be her true value and there is therefore the danger that an owner will be overcompensated if he can claim both for unrepaired partial loss and for a total loss all within the policy period. As Lord Ellenborough said arguendo in Livie v. Janson at 652/3:

"It therefore seems to me useless to be seeking about for odds and ends of previous partial losses which might have happened to a ship in the course of her voyage, when at last there was one overwhelming cause of loss which swallowed up the whole subject-matter."

71. Therefore the owner of Eastlands had no claim for the unrepaired partial loss under his policy with marine underwriters, because until the end of his policy such unrepaired partial loss simply did not count, and before the end of the policy the vessel had become a total loss. As Lord Sumner said (at 213/4):

"I do not see anything intrinsically in conflict with a contract of insurance in the proposition, that perils insured against have not caused any direct loss where although the ship was damaged the owner had nothing to pay, and although the ship was depreciated, she ceased to exist as a ship before the time came for measuring the amount to be recovered for it. At that date there was no pecuniary loss against which the owner could be indemnified."

72. Lord Sumner went on to suggest that it was the owner's own fault for not contracting as between his two insurers for a full indemnity (at 213/215), and it is true that that was the combined effect of the two contracts, the facts and the established law. But it was also bad luck for the owner. Mr Berry sought to rely on this passage to support his case that there would be nothing wrong or strange in applying the doctrine of unrepaired partial loss to a situation of constructive total loss. But that is the issue, and unless the case of constructive total loss treated as such followed by actual total loss is in truth the same, in principle, as unrepaired partial loss followed by total loss, then we do not see what Mr Berry can get out of this passage.
73. Robertson v. Petros M Nomikos Limited [1939] AC 371 concerned a loss of freight policy. The ship in question suffered a constructive total loss under the terms of that freight policy, which stipulated that the value when repaired was to be taken as the insured value. The cost of repairs was greater than the insured value but less than the actual market value when repaired and so the owner elected to treat the loss as a partial loss and keep the vessel. He therefore served no notice of abandonment. The repairs prevented the performance of the charter under which the insured freight was to be earned. When the owner therefore sought to recover on the freight policy, the underwriter said there had been no constructive total loss of the ship because there had been no notice of abandonment. However, the House of Lords pointed out that the existence of a constructive total loss (section 60) was different and anterior to the right to abandon and claim for a total loss (sections 61/62). Lord Wright said (at 381/2):

"The appellant's argument confuses two different concepts, because it confuses constructive total loss with the right to claim for a constructive total loss. The right to claim except in certain cases depends on due notice of abandonment under s. 62 of the Act. The distinction is explicitly stated in s. 61…[which] makes it clear that the right to abandon only arises where there is a constructive total loss in fact. That is the necessary precondition to a right to abandon…It is a superimposed right of election where there is a constructive total loss…"

See also Rankin v. Potter at 155, and para 52 hereof above.

74. On this basis Mr Berry wishes to say that there are two conditions precedent to a right to claim for a constructive total loss: one is the existence of the loss itself, but the second is the exercise of the right of election to abandon, a right which he submits is lost once an actual total loss has occurred.
75. It is in this connection that Mr Berry also relies on Court Line v. The King (1945) 78 Ll L Rep 390. The case concerned a different kind of abandonment, namely that of a ship by her master and crew for the purposes of section 60(1) of the Act. The distinction between that kind of abandonment, and the abandonment which is spoken of in sections 61 and 63, was underlined by Scott LJ at 397. The one is the abandonment of a ship to her fate, the other is the abandonment of a ship to her underwriters, even though, where the second kind of abandonment discussed in section 60(1) is concerned, the abandonment of the ship to her fate may, nowadays at any rate, be "a mere mental decision by the owner that he will exercise the option which Sect. 61 allows him". That, however, is still establishing the fact of a constructive total loss, rather than what in Robertson v. Nomikos was described as the second and subsequent concept. We do not accept, therefore, that Scott LJ's "mere mental decision" is a reference to how the section 61 election is itself carried out. Scott LJ continued:

"In Sect. 61 the word "abandonment" seems to import an act on the part of the assured, but in truth it amounts to nothing more than his making up his mind to give notice of abandonment to the insurer under Sect 62(1), at the peril of losing his right of election under Sect 61."

76. We would seek to put the matter under section 61 somewhat differently. The word "abandonment" does not itself appear in section 61, which speaks rather of the choice available to the assured to "abandon the subject-matter insured to the insurer". What is being chosen is therefore a legal concept rather than an act, physical or mental. The legal concept is the cession of the ship to the insurer (see the dictum of Martin B in Rankin v. Potter cited above and now sections 63 and 79 of the Act). That cession occurs in both forms of total loss, actual as well as constructive (Kaltenbach v. Mackenzie, see para 58 above). The cession, however, only occurs upon payment, and only if the underwriter is willing to accept the cession: for even in the case of an actual total loss, where the cession operates as a matter of law, the language of section 79(1) speaks of the insurer being "entitled" to take over the interest of the assured: see Allgemeine Versicherungs-Gesellschaft Helvetia v. Administrator of German Property [1931] 1 KB 672 at 687/688 per Scrutton LJ. In the case of a constructive total loss, because of the need for a notice of abandonment save in exceptional cases, the underwriter's decision will effectively occur when he accepts, if he does, a notice of abandonment (see section 62(6)), but even so the abandonment only takes effect on payment. It is then retrospective to the time of the casualty: see section 79(1). That is abandonment.
77. The notice of abandonment, however, is the normal means by which an election to abandon, is notified, and we would suggest, made. We are not attracted by Mr Berry's submission that the election is a matter of the mind. English commercial law does not favour the merely mental, subjective or internal, not even something as comparatively physical as an unsent letter in the desk drawer. Where relations between contracting or would-be contracting parties are concerned, there needs to be something objective, something said or done (and exceptionally, where some action is required and mere silence or a failure to act becomes a form of speaking silence or conduct, some omission) which, as it were, "crosses the line". Hence the general requirement of notice, which nevertheless as section 62(2) makes clear need not be formal or in writing. That notice, however, does not effect abandonment, even if it is valid. It is essentially an offer to cede, a formal or even informal recognition that, if the insurer so elects, he will, upon payment, be entitled to a complete interest in the thing insured or what remains of it as from the time of the casualty. If that offer is accepted it becomes irrevocable. If it is rejected, like any offer, it can be withdrawn: Royal Boskalis at 556. If it is neither accepted nor withdrawn, but nevertheless valid, then in due course a judgment of the court will confirm its validity, with the consequences provided for under section 63.
78. And what then of the election to abandon spoken of in section 62(1) and the choice allowed under section 61? And what of the exceptional case where a notice of abandonment is not required to make a claim for a constructive total loss? Is this where the mere working of the mind comes in? Again, we do not think so. The election is rather a legal concept which is produced by the assured's conduct or omission to act (the latter is specifically provided for in the last sentence of section 62(1)). The emphasis is on how the assured treats the loss. If he treats the loss as a partial loss, for instance by continuing to treat the thing insured as something in which he alone is interested, for instance by selling it without reference to the underwriter, then he cannot go down the road of claiming a constructive total loss. If, however, he shows that he is treating the thing "as if it were an actual total loss" (section 61), the paradigm and in general the only way of doing which is to give prompt and due notice of abandonment to the underwriter, then he has exercised his choice to abandon. It is spoken of as an election: but it may not be an election of classic form, which is generally regarded as being irrevocable. On the contrary, an unaccepted notice of abandonment may be revoked; and even if a claim for a constructive total loss, premised on a notice of abandonment, fails, the assured may still be able to recover for a partial loss (section 56(4)). It is rather that, in the absence of treating the constructive loss as a total loss, the assured may be taken to have elected to have treated it as a partial loss (section 62(1)).
79. It seems to us that this as it were "default" position is analogous to situations found in the general law where a party has an election, for example to accept the repudiation of a contract. If that party chooses to maintain the contract rather than to bring it to an end, then he may recover damages only on the basis of the properly attributable damages flowing from an actual breach, and not on the basis of a merely anticipatory breach or on the basis of the repudiatory loss of the contract. If, however, he purports to accept the repudiatory breach as bringing the contract to an end, then, if he succeeds in his allegation of repudiation, he can recover damages on an anticipatory basis and in any event on a repudiatory basis, and even if he fails in his allegation of repudiation, he may still recover damages for any actual breach (but not any merely anticipatory nor repudiatory breach). Similarly, if he takes too long to indicate his acceptance of a repudiation, he may lose the chance to bring the contract to an end.
80. It is in this context that the next question arises: what of the case of a constructive total loss overtaken by an actual total loss, where the assured is excused the need for a notice of abandonment to demonstrate that he has not treated the loss as a partial loss? Is he in such a case to be dealt with on the "default" basis that he has treated the loss as a partial loss? If so, then the excuse from giving a notice of abandonment is nugatory, or even worse, it is deceptive. This is the question which to our mind is at the root of the various submissions advanced on this appeal and we will revert to it below.
81. Pesquerias v. Beer (1946) 79 Ll L Rep 417 was a Spanish civil war case. The constructive total loss of a trawler (the Vendaval) by reason of a risk within the policy (riots or civil commotions) was followed by another alleged seizure by the Basque government (that would have counted as an excluded war risk) and, after the policy expired, by an actual total loss by reason of an excluded war risk (sinking by Nationalist enemy fire). The original constructive total loss by seizure was in August or September 1936, the second alleged seizure was in October 1936, the policy expired in December 1936, notice of abandonment was given in February 1937 and the sinking was in March 1937. It was held that the events subsequent to the original constructive total loss were irrelevant (at 434). That case did not involve the unusual feature of the instant case whereby the actual total loss followed very closely upon the constructive total loss and excused the giving of a notice of abandonment. Nevertheless, Atkinson J held that once there had been a constructive total loss within the policy, there was a right to claim. He said that that followed from Andersen v. Marten, adding (at 434) –

"That was, of course, a constructive total loss by capture. That is the reverse case [sc because that capture was an excluded war risk, which preceded an actual total loss by perils of the seas], but it is just as good an authority as if it were not, that, once you have got a constructive total loss, the mere fact that it may ultimately become an actual total loss because of some event that is not within the policy does not affect your right at all to claim as for a constructive total loss."

82. The distinction between the existence of a constructive total loss and a right to claim a total loss indemnity was emphasised in a more recent case where matters had to be considered from the point of view of a defence based on limitation of time. In Bank of America National Trust and Savings Association v. Chrismas, The "Kyriaki" [1993] 1 Lloyd's Rep 137 Hirst J held that for limitation purposes the assured's cause of action arose at the date of the CTL casualty, that a notice of abandonment was not an essential ingredient of that cause of action but rather a notification of an election between two alternative quantums of damage.
83. We are now in a position to deal with the insurers' submissions, and, in the light of our treatment of the authorities, to do so relatively briefly.

(1) No abandonment under section 61

84. We refer to the submission outlined at paragraph 29 above. In essence Mr Berry submits that without an election to abandon, the claim can only be for a partial loss (which becomes merged with the actual total loss); and that such an election can only validly be made before an actual total loss and while the subject matter insured is still in existence as such (in specie). Although a notice of abandonment can be excused, the prior election to abandon cannot be excused. In any event the original pleaded claim for an actual total loss was inconsistent with a claim for a constructive total loss and thus excludes the election to abandon which is fundamental.
85. In our judgment this submission is flawed. The first question is whether there has been a constructive total loss by fire (section 60). Subject to submissions (2) to (6) below, which all represent various ways of saying that there has been no loss by fire at all, or rather no loss that can stand separately from the actual loss by sinking, there is no dispute that there has been such a loss by fire, as found by the judge. We will deal below with the submissions that there has been no loss by fire that can be recognised in law. For the present it must be at least assumed that the existence of a prior constructive total loss by fire has been established. That fulfils section 60.
86. The current submission is founded on section 61. That section, however, does not in terms require an election to abandon. It is certainly not concerned with any election to treat a total loss as constructive as distinct from actual. It is rather concerned with eliminating from the universe of total losses a case where the assured elects to treat a constructive total loss as a partial loss only. As section 62(1) makes clear, unless a notice of abandonment is given then (subject to cases where such notice is excused) the loss can be treated only as a partial loss. The second question therefore is whether a notice of abandonment was needed, for if it was then in its absence section 61 and section 62(1) entail that the constructive total loss by fire can only be treated as a partial loss. In such a case it is common ground that such an unrepaired partial loss founds no recovery in the light of the subsequent actual loss by sinking: section 77(2) and clause 18.2. However, it is also common ground that a notice of abandonment was excused under section 62(7). That answers the second question.
87. The third question then is whether the constructive total loss by fire has been treated as a partial loss in some way other than by failure to serve a notice of abandonment. It has not. It is indeed hard to imagine how an owner assured could treat the loss by fire in the circumstances of this case as a partial loss. The vessel is at the bottom of the sea, without hope of salvage. In what way is it suggested that the owners have treated the loss by fire as a partial loss? There is no answer to that, save the point which is the focus of Mr Berry's seventh submission, viz that the initial limitation of the owners' claim to one of actual total loss was an irrevocable election not to claim on the basis of the constructive total loss. That submission, however, is misconceived. The owners' claim for an actual total loss cannot amount to treatment of the constructive total loss as a partial loss. A claim for a total loss is essentially inconsistent with treatment of the vessel as a partial loss.
88. Nevertheless, Mr Berry submits that the actual total loss claim is inconsistent with treating the vessel as a constructive total loss because the owners were thereby representing themselves as treating the vessel in the short period between constructive and actual loss as being their vessel, and that was inconsistent with abandonment of the vessel to the insurers for that period of time. However, we regard the submission as mistaken. By claiming for a total loss, albeit initially an actual loss, the owners demonstrated that they were willing to abandon the vessel to the underwriters. Moreover, by claiming for an actual total loss the owners were representing themselves as compliant with an ultimate abandonment of the vessel. As the authorities make clear, abandonment, as distinct from notice of abandonment, operates in cases of actual as well as constructive total loss. In this case, because a notice of abandonment was not required, the owners did not have to declare promptly by such a notice that they were willing to abandon the vessel to the insurers. But they were willing, and at no relevant time represented themselves as unwilling. The fact that a claim for an actual total loss may be said to involve a proposition that the vessel at the time of such a loss belonged to the owners is neither here nor there. It is not inconsistent with a willingness at all times to cede her to the insurers. Nor is it treating the constructive total loss as a partial loss. In truth, the owners never treated the loss of their vessel as a partial loss; nor did they ever have the opportunity to do so.
89. Moreover the owners' pleading was that the actual total loss, consummated by the sinking of the vessel, was a loss by fire (and/or explosions and/or perils of the seas). We do not understand how this is in any way inconsistent with a subsequent alternative claim, that if the actual loss was not by fire then a previous constructive loss was a total loss by fire. The original plea was not treating the constructive total loss as a partial loss; it was merely concentrating on one mechanism of total loss in the absence of making claim, for the present, by reference to another mechanism of total loss. To regard the owners' claims in any other light is, in our judgment, to use the intensely practical rules which the common law has developed and the Act has enshrined regarding constructive total loss as a destructive theoretical tool.
90. In sum, the owners have never treated or represented themselves as treating the damage caused by the fire as amounting only to a partial loss and have never represented themselves as being unwilling to abandon the vessel to the insurers. Nor have the insurers ever relied upon such a representation.
91. The fourth question is whether nevertheless it is always impossible to make claim for a constructive total loss after the actual total loss of the vessel. At one point in his oral submissions Mr Berry said that his case in a nutshell was that it is impossible to have a constructive total loss under section 61 without an election to abandon or an abandonment, and it is impossible to have an election to abandon or an abandonment after a ship is actually lost. We can see no reason why this should be so. The cases demonstrate that vessels may well be subject to successive losses, first constructive and then actual. The assured's essential cause of action arises at the time of the casualty: The Kyriaki. In most cases a notice of abandonment may be a condition precedent to the right to measure the indemnity on a total loss basis, but that does not apply in the present case. In such a case the matter must be looked at as at the time of the casualty. Was there then a right to claim for a constructive total loss? If there was, we do not see why the cause of action should have been lost, just because the vessel subsequently becomes an actual total loss.
92. Mr Berry's fall-back position is that the right to claim for a constructive total loss disappears at any rate if the actual loss precedes the assured's opportunity to elect, by some mental process, to treat the casualty as a total loss. In this connection he relies on dicta in the cases, such as Hamilton v. Mendes, Roux v. Salvador and Rankin v. Potter, which speak of the difference between abandonment and notice of abandonment. We have sought to show how those dicta do not affect this case. It may be true that, as here, an assured never has the opportunity, prior to the actual total loss, to address the question whether he wants to treat the constructive loss as partial or total: but that is beside the point. The focus of section 61 is the choice between treating the loss as partial or total: and where the vessel is already actually lost before the assured has the opportunity to choose, the choice is meaningless. The underwriters have lost nothing by the absence of an offer of abandonment (the notice of abandonment), and as for actual abandonment, that will follow as a matter of law even in the case of actual total loss. We have not been impressed by the submission that what is essential is a mere mental decision. It does not follow from the need for a notice of abandonment in the normal case that anything further is required in a case where notice of abandonment is excused than the absence of an inconsistent treatment of the damage as a partial loss. Nor does it follow from the fact that timely recovery of a captured vessel may remove a claim for a constructive total loss that an untimely actual loss following hard upon a constructive total loss destroys the opportunity to claim for the latter. The reference to "the subject-matter insured" which Mr Berry emphasises at this point of his argument may merely reflect the general assumption that the thing has survived in some form. It reflects in other words the standard, as distinct from the exceptional, case where the vessel survives and is the subject of a notice of abandonment. Or it reflects the case which is the concern of section 61 where it is possible to choose to treat a constructive total loss as a partial loss (see para 60 above). But, as section 62(7) makes clear, the general assumption may not always apply; and in any event there appears to be no essential difference between "the subject-matter insured" (section 62(1)) and "whatever may remain of the subject-matter insured" (section 63(1)), nor for these purposes between constructive total loss (sections 62(1) and 63(1)) and actual total loss, which falls within section 79(1) where the statute speaks equally both of "whatever may remain of the subject-matter [insured]" and "that subject-matter".
93. We do not think that the existence or otherwise of the subject-matter insured as something in specie is relevant to the current problem. That question may be relevant to whether or not an actual total loss has occurred, for instance to issues which could arise under section 57(1) as to whether the subject-matter insured has been destroyed or "so damaged as to cease to be a thing of the kind insured". The court has not been concerned with such issues. Even in the case of an actual total loss, however, anything left over, even if not in specie, must be abandoned (per Brett LJ in Rankin v. Potter at 471, see para 58 above). Nor do we think that the special rule which exists where the thing insured is recovered (see eg Hamilton v. Mendes) is relevant to the case where a constructive total loss is overtaken by an actual total loss.

(2) One casualty

94. Under this heading Mr Berry submits in effect that there cannot be two casualties separated by a scintilla temporis and that the judge's finding of two separate casualties was unrealistic.
95. That, however, is a question of fact and one that is answered by paragraph 19 of the judgment below, where Tomlinson J found that there were two independent causes, a true case of overlapping coincidence, but not identity of coincidence. It was the insurers' own case that the fire was an independent cause of loss from the ingress of water. Critically, the judge found that since the fire caused damage well in excess of $3 million and was not causative of the sinking he was unable to conclude that the vessel was already an actual total loss or bound to become one by sinking before she became a constructive total loss by fire. As explained in that part of the judgment above headed "The emergence of the constructive total loss issue in the course of the litigation", by the time of trial there were plain issues both as to the extent and development of the fire and as to whether the vessel had only become a constructive total loss by fire at a time when the vessel was already doomed to actual loss by sinking. The judge would have had these issues fully in mind while considering the evidence in the case. Even though the insurers' only concession related to the point of sinking, the judge was entitled to make inferences and findings upon the evidence as a whole. We are unable to say that the judge was wrong in his critical finding, nor has Mr Berry sought to argue this submission as a matter of the evidence. Rather, his standpoint was ultimately that the only issue that was before the judge related to the point of time when or immediately before the vessel sank. That became clearer under the insurers' third submission (see below). Subject to that, we conclude that the present submission must be rejected.

(3) No proof of prior constructive total loss

96. This submission was very much bound up with the second submission, and was in effect the legal route by which Mr Berry sought to turn the flank of the judge's finding of fact as to a separate and prior constructive total loss by fire.
97. It emerged under this heading that Mr Berry went so far as to submit that there was no issue at trial as to any other point of time apart from the moment of sinking, ie the scintilla temporis before sinking. As such, this was, in logic, the most basic of all the insurers' points and ought, in theory, to have been the first. In effect, Mr Berry was complaining that the judge had decided a matter which was not in issue. He submitted that the owners' constructive total loss case was tied exclusively to the moment of sinking, when the vessel was in any event bound to sink and thus an actual total loss. If the point was a good one, it was another explanation of why the insurers submitted (see under submission (2) above) that there had been only one casualty, not two.
98. When the point emerged in the course of Mr Berry's oral reply, it caused some surprise because Mr Berry had not raised it in the course of his opening submissions nor then prepared the ground for it as one might have imagined that he would, in particular by taking the court to the pleadings and skeleton arguments which were before the judge. As it happened, this was not done until Mr Bernard Eder QC, counsel for the owners, commenced his respondent oral submissions, when he took us to the relevant material in large part in order to support his subsequent submissions on the owners' cross-appeal as to costs. Nevertheless, it is fair to say that Mr Berry's point can be discovered in the insurers' grounds of appeal, for instance under the "one casualty" heading, where the single unexplained sentence can be found: "Indeed, it was not the Claimants' case that there were two casualties"; and again under the current heading of "No proof of prior constructive total loss", where it was said that "There was no case advanced by the Claimants…nor any evidence, that the cost of repairs to the Vessel would have exceeded her insured value at any time before the scintilla temporis before her sinking". In the insurers' skeleton argument the point was addressed under the current heading, where it was submitted that it was common ground that the only factual basis of the claim for the vessel's constructive total loss related to the moment of her sinking and that there was simply "no allegation that the Vessel had become a constructive total loss at some earlier time".
99. We are satisfied, however, that these are not accurate statements of the position. We refer to the exposition set out above of the pleadings and pre-trial skeleton arguments exchanged between the parties. It is true that the only admission ultimately made by the insurers in connection with the plea of constructive total loss related to the scintilla temporis before sinking, ie to the moment of sinking; and also true that the insurers' forensic position during the trial was that it was common ground that it was only at that point that the claim for a constructive total loss fell to be determined. However, neither the insurers' admission nor their forensic stance determined the issues before the court. The owners' plea was the general one that the constructive total loss occurred "prior to" the sinking, and we think that that plea, which went back beyond the point of sinking, was properly understood, for instance when the insurers amended during the course of trial to deny generally that the vessel "was a CTL before she was an ATL" (see para 18 above) even though that plea was subsequently narrowed by their further information (see paras 19/22 above). Ultimately, it was for the judge to determine any "pleading point" seriously raised at the trial. Inasmuch as such a point emerged at trial, and we are not satisfied that it did, Tomlinson J determined it by his finding that the vessel was a constructive total loss by fire at some material point prior to her actual loss by sinking and at a time when she was not as yet bound to sink (para 19 of his judgment). That finding is simply inconsistent with Mr Berry's present submission.
100. We therefore proceed to consider the essential point raised by the submission of "No proof", which is that the judge simply did not have the evidence to justify his critical paragraph 19 finding. There is an allied submission that the judge placed the burden of proof on the insurers, whereas it ought to have been borne by the owners.
101. This is a short point. The experts agreed that "had the vessel not sunk" she was a constructive total loss by a substantial margin. The cost of repair figures before the court demonstrated just how wide that margin was. The judge was entitled to find, as we have said above, having considered any remaining disputed assessments of the extent and development of the fire, that the vessel was a constructive total loss by fire a material time before her loss by sinking. We think that he was entitled to draw that inference and make that finding even if the burden of proof was on the owners. The burden of proving a constructive total loss was on the owners, subject to this, that if the insurers wanted to say that the vessel was already doomed to be an actual loss by sinking before she became a constructive total loss, a point which they had pleaded, then the burden was on them.
102. We conclude therefore that this submission fails.

(4) No loss

103. It is not entirely easy to determine how this submission differs from the previous submissions or from submission (6), which depends on the doctrine of merger. We will consider that doctrine separately below. Mr Berry simply asserts that the owners' true loss resulted from the vessel's sinking and not from the fire. He says that the fact that when she sank she was damaged by fire was of no concern to the owners. He relies on Livie v. Janson, but that was an example of the doctrine of merger in the case of partial loss. It remains to be seen whether that doctrine applies to the case of a constructive total loss. Mr Berry also relies again on the absence of abandonment, but that takes us back to submission (1). In the meantime, the facts as found by the judge were that the vessel was not merely damaged by fire, but totally lost by fire a material time before her loss by sinking. There was a cause of action for that loss as from the time of that prior and separate casualty. If the vessel had not sunk, she would have been lost to the owners in any event. Subject to the doctrine of merger we do not see why the owners have not suffered a true loss.

(5) Clause 18.2

104. For the sake of convenience we set out clause 18.2 of the Institute Clauses again, but in its full context:

"18. Unrepaired damage

18.1 The measure of indemnity in respect of claims for unrepaired damage shall be the reasonable depreciation in the market value of the Vessel at the time this insurance terminates arising from such unrepaired damage, but not exceeding the reasonable cost of such repairs.

18.2 In no case shall the Underwriters be liable for unrepaired damage in the event of a subsequent total loss (whether or not covered under this insurance).

18.3 The Underwriters shall not be liable in respect of unrepaired damage for more than the insured value at the time this insurance terminates."

105. This clause forms part of the new Institute Clauses of 1 November 1995. There is a helpful discussion of it at Arnould, 16^th ed, Vol 3, 1997 at paras 134/138 of Part One, at pp 89/91. The following matters should be noted. The clause is headed "Unrepaired damage" and speaks throughout of unrepaired damage. It does not refer to constructive total loss. On the contrary, constructive total loss is dealt with separately in clause 19. Moreover, in clause 18.2 itself "unrepaired damage" is contrasted with "a subsequent total loss". Such a total loss might itself be a constructive total loss.
106. Subject to submission (6) and the doctrine of merger below, we consider that it is plain that the clause is dealing with partial loss and not constructive total loss, although we would concede that if the doctrine of merger embraces a constructive total loss followed by an actual total loss, then our opinion of the clause might have to be revised: for it seems reasonably obvious to us that the clause is intended in essence to reflect the Act and the common law, even if clause 18.1's reference to "market value" is also a clarification of previous controversy (and adopts a solution rejected in Irvin v. Hine [1950] 1 KB 655). Thus Arnould comments (at para 138 at p 91):

"Clause 18.2 does not alter the position which would otherwise obtain, in the absence of such provision. Its wording expands upon what was previously included in Clause 16, of the 1970 Clauses. In both instances, the purpose would appear to have been to inform prospective assureds, and clarify the meaning and scope of their insurance. Clause 18.2 draws attention to section 77(2) of the 1906 Act, and to principles which are well established by reported cases. For further discussion of the point, the reader is referred to paragraphs 1126-1129 in the 16th edition."

107. It is instructive that the reference in that passage is to paras 1126/9 of the main text: for those paragraphs are headed "Successive losses: partial followed by total" and were not relied on by Mr Berry; as distinct from the different passage at para 1180, which is headed "Successive total losses", to which we will refer below.
108. It is also instructive that clause 18.1 and 18.3 refers to "the time this insurance terminates". This reflects the common law rule implicit or referred to in cases such as Livie v. Janson and British & Foreign Insurance v. Wilson that the time for measuring the loss represented by unrepaired damage is at the end of the period of cover and not before. Before that time, unrepaired partial loss counts for nothing.
109. We bear in mind that clause 18.2 begins with the words "In no case", on which Mr Berry laid stress. That does not, however, assist him if, as we consider, unrepaired damage only relates to partial loss and does not embrace a constructive total loss which has not been treated as a merely partial loss.
110. We therefore conclude that, subject to submission (6) to which we now come, clause 18.2 does not assist the insurers.

(6) Merger

111. Mr Berry's submission is that a constructive total loss which is overtaken by an actual total loss during the period of the insurance falls to be treated like any other unrepaired damage and is merged and swallowed up in the subsequent total loss. As stated above, he relies principally on Livie v. Janson, Woodside, and British & Foreign Insurance v. Wilson. He acknowledges that section 77(2) of the Act is expressed in terms of partial loss, so that it does not assist him. However, he submits that the common law rule is wider than section 77(2) and extends to all unrepaired damage whether that damage constitutes a partial loss or a constructive total loss. In this respect he relies on the formulation of Lord Birkenhead in British & Foreign Insurance v. Wilson at 199 (cited at para 67 above). He submits that Woodside, on the assumption of a constructive total loss there made, is a decision in his favour. He also submits that clause 18.2 of the Institute Clauses, with its reference to "unrepaired damage" rather than to partial loss, reflects the wider common law doctrine.
112. We have discussed those authorities in detail and in context above. In our judgment they do not assist the insurers. To summarise, Livie v. Janson (see paras 42/43 above) is a case of partial loss. If Hahn v. Corbett, where Livie v. Janson was distinguished, is correctly viewed as a case of constructive total loss followed by actual loss, then it goes wholly against the insurers' submission and in favour of the owners (see paras 44/46 above). Woodside proceeded on the basis that, despite the assumption of a constructive total loss, that loss had been treated, in the absence of a notice of abandonment, as only a partial loss (see paras 64/66 above). Moreover Woodside has not been treated in Arnould as authority for Mr Berry's proposition (see at para 66 above). As for British & Foreign Insurance v. Wilson, we would not regard their Lordships there as contemplating the case of total as distinct from partial loss (see para 69 above). The use of the expression "unrepaired damage" by Lord Birkenhead was therefore, in context, a reference to damage which was or had been treated as a partial loss. And although their Lordships considered that the common law rule of merger went beyond the section 77(2) restatement of that rule, nevertheless that was only in the context of the limitation in the section to successive losses "under the same policy". British & Foreign Insurance v. Wilson of course raised a problem under separate contracts of indemnity. We do not think therefore that their Lordships had it in mind to widen the rule substantively to successive total losses. On the contrary, Lord Birkenhead himself contrasted damage with total loss when he spoke of a vessel which "is damaged…and before that damage is repaired she is lost". Where, however, a vessel is properly treated as a constructive total loss, there is no question of her being repaired (at any rate by the assured).
113. We consider that Arnould, as far as it goes, supports this view of the authorities. We have already made the point that the merger rule is dealt with at paras 1126/9 under the heading "Successive losses: partial followed by total" and that successive total losses are dealt with separately at para 1180 (see para 107 hereof above). We have also made the point that at para 1180, Woodside is not cited for the proposition relied on by the insurers (see para 66 hereof above). At para 1180, Arnould emphasises the important difference between a partial loss followed by a total loss and true cases of successive total losses. The text reads:

"As we have already seen, an unrepaired partial loss merges with a subsequent total loss happening as the result of a later casualty…

"Although the point has never been decided, there are strong grounds for asserting that the rule is different in cases where there are successive total losses, each caused by separate casualties…

"In those cases, where there are, in the true sense, successive losses, the following rules may be proposed. First, where the original casualty is covered by insurance and the assured has given notice of abandonment in due time, any casualty which may happen thereafter is irrelevant, so long as the original notice of abandonment is maintained…

"Thirdly, it is submitted that where the first casualty is treated as a partial loss, the right to recover in respect of that casualty is governed by the rules prevailing in cases where a partial loss is succeeded by a total loss. In other words, if the loss was in fact constructively total, but not claimed as such, is left unrepaired, the loss merges with a subsequent total loss by perils operating during the currency of the policy, but if the second casualty is caused by perils operating after the policy has run off, no merger takes place."

114. Mr Berry submits that the instant case falls under the third proposition. That, however, is a mistaken view, because the third proposition is expressly dealing with the situation where the first casualty is "treated as a partial loss". In such a case, the situation falls back into the merger rule governing unrepaired partial loss followed by total loss. The fact that the constructive loss could have been treated as a total loss is irrelevant where the loss has in fact been treated as a partial loss. Nevertheless, Mr Berry went on to submit that the second sentence of the third proposition was in any event dealing with the instant case. However, in our judgment that is not so: the second sentence begins "In other words": it merely draws out further the situation inherent in "treated as a partial loss".
115. In our judgment, this case falls rather under a logical development of Arnould's first proposition. That is dealing with a constructive total loss which has been duly treated by the assured as such, by reference to a due notice of abandonment. Naturally enough, Arnould is dealing with the paradigm case, not the exceptional case where a notice of abandonment is excused. In this case, however, there is of course no notice of abandonment, but its need is excused. Therefore, unless there is some special reason why the constructive total loss by fire cannot be treated as such and must be treated as a partial loss, this case fits under the first and not under the third proposition. If, however, a constructive total loss where notice of abandonment has been excused had to be treated as a partial loss, then the excuse of a notice of abandonment would be nugatory. We have covered this ground. There is no special reason why the constructive total loss by fire cannot be treated as a total loss. We have rejected the various attempts made by the insurers, discussed principally under submission (1) above, to say that the fact that the actual loss by sinking forestalled any possibility for the owners to exercise some mental decision in relation to the prior constructive total loss by fire meant that the latter could not be made the subject of a valid claim. We have also rejected the submission made by the insurers that the initial claim for an actual total loss by fire was inconsistent with a subsequent claim in respect of a constructive total loss by fire.
116. Those are the authorities. They may not support Mr Berry, but is his submission nevertheless well founded? We do not think so. Our reasons are essentially these. First, the statute, reflecting the common law rule, even if narrower than it to the extent described above, is expressly in terms of partial loss. Secondly, we consider that there is no case at common law which decides that the merger rule goes as wide as Mr Berry needs to submit it does. Thirdly, as a matter of principle, we do not see why a total loss, even one premised on the excessive cost of repairs, should follow the same merger rule as unrepaired partial loss. If a constructive total loss has not been treated as a partial loss so that an assured is entitled to put it forward as a total loss, then the thing insured has been lost and assuming the claim to be a valid one the insurance is exhausted, subject to any question of sue and labour. Once a vessel has been totally lost, and again assuming that the loss has not been treated as a partial loss, we cannot see why or how any subsequent loss is relevant. That is not to say that an assured may not maintain a number of separate total loss claims in the alternative: but once a total loss has been established, it exhausts the insurance. If the first total loss falls outside the cover, the assured cannot recover; but if it falls within the cover, he can. All the cases are consistent with that. Moreover, we do not see why, as long as there is dispute about the facts and the parties' responsibilities, an assured cannot maintain alternative claims for a total loss, even if some of them were to be formally inconsistent. That is a common occurrence. What an assured cannot do is to treat a loss as a partial loss and then recover for that loss as a total loss. Fourthly, the merger rule is premised on the consideration that to allow an insured to recover for an unrepaired partial loss as well as for a subsequent total loss would be to give the assured more than an indemnity. So it would in any case of an agreed value; and, independently of that point, so it would where the rule for unrepaired loss is that it has no value until the end of the period of cover. In our case, however, there is no possibility of the owners getting more than an indemnity.
117. We therefore reject the insurers' submission that the merger rule applies to the constructive total loss by fire in this case.

(7) Irrevocable election and estoppel

118. Mr Berry submits that the initial claim for an actual total loss by sinking excluded a subsequent claim for constructive total loss by fire. We have already given our reasons under submission (1) above for saying that there was no irrevocable election by reason of the initial claim. A fortiori, in the absence of any reliance by the underwriters, there could be no estoppel.

Conclusion on the insurers' appeal

119. It follows that the insurers' appeal fails. We have upheld the judge's judgment on the merits of the constructive total loss claim, essentially for the reasons given by the judge.

The cross appeal on costs

120. We have already referred to the order for costs which the judge made and how the constructive total loss issue emerged in the litigation. In his costs judgment the judge also referred to the history of the actual total loss claim, which was as follows.
121. After the vessel sank the crew were rescued and taken to Sri Lanka where they were interviewed in the presence of solicitors for the interested parties. Once the interviewing was complete it was apparent that if a claim for the actual total loss of the vessel was to be pursued the owners would have to show how a fire in the engine room could have caused the vessel to sink within fifteen hours of the outbreak of fire. In the proceedings started on 1 November 2000 the owners relied on the crew interviews and a report from Mr Charlton of Dr J.H. Burgoyne and Partners to the effect that there had been an explosion in a topside tank which ripped open the sides of the vessel at or below the water line and so caused the engine room and the no. 4 hold to flood. The judge described this as a theory which might have explained why the vessel sunk but one which "lacked any other merit". This evidence formed the basis of an application for summary judgment which failed on 23 May 2001.
122. As we have said the constructive total loss claim was added by amendment on 17 August 2001. By further amendment on 6 March 2002 the claimants abandoned their reliance on Mr Charlton's theory of causation. They initially resisted attempts to get them to say what their case on causation now was, but were ordered to do so and on 28 March served reports from Dr Foster (a fire expert), Mr Bevis (a marine engineer) and Mr Parry (a naval architect). These experts advanced two new theories of causation to support the actual total loss claim namely a double bronze valve failure on the day oil tanks and "melting" of the engine room bulk head. These new theories raised questions of metallurgy but the owners resisted the introduction of evidence from metallurgists. When metallurgists were consulted they agreed that the "melting" theory was untenable and it was not pursued at trial.
123. The trial took 17 days. The judge records that the factual evidence from the crew took 3 days and the expert evidence 9 days. In the section of his judgment dealing with the actual total loss claim the judge dealt comprehensively with and rejected the case on causation based on valve failure. It is not necessary to go into detail, but in his costs judgment the judge said of the owners expert evidence that it "was seriously flawed in that it failed to address the question how, realistically, sufficient water can have entered the ship within the required time scale in consequence of damage caused simply by fire and explosion".
124. There is one further event which the judge had to consider when making his order for costs. On 11 June 2002 the owners made a Part 36 offer to settle for $2.8m inclusive of interest. As the owners obtained judgment for $3.49m they bettered this offer. The insurers accepted that this was a valid offer, open for their acceptance until the first day of the trial.
125. In deciding what order for costs to make the judge considered what would have happened if no actual total loss claim had been made. This was, he said, to some extent an exercise in rewriting history but he concluded:

"…I entertain no real doubt that if the claimants had never made a claim for an actual total loss, restricting themselves to a claim for a constructive total loss, the course of the litigation would have been different. Naturally the initial evidence gathering exercise involving the interviewing of the crew members, would have been the same. But had the proceedings issued in November 2000 related only to a claim as for a CTL, the whole focus of the debate would have been different, and in particular it would never have been necessary to investigate at such length and at such expense implausible theories as to how, in consequence of the fire, sufficient water might conceivably have been admitted in sufficient time to cause the vessel to sink within 15 hours of the outbreak of fire."

126. He added that such a claim could probably have been determined summarily very much earlier, but if it had gone to trial it was inconceivable that the owners would have called oral factual evidence. Expert evidence would have been confined to the extent and cost of repairs issue on which there would have been little if any scope for debate.
127. This led the judge to say:

"I have no doubt that the greater part of the costs expended on both sides will have been directed to proof and disproof of the suggestion that the ingress of sea water that caused the vessel to sink was caused by the fire and such explosions as that may in consequence have caused. That was the purpose of the great bulk of the experts' work. The work of the naval architects (and of the metallurgists although their costs must pale into insignificance) was devoted to nothing else and the issue as to the extent and cost of the fire damage was only a very small part of the work of the fire experts and of the marine engineers. At trial the argument as to the CTL was entirely of a legal nature without the need even to refer to the factual or expert evidence."

He appears to have accepted that the argument on the constructive total loss issue occupied only about 1½ out of the 17 days of the trial. The rest of the time was spent in investigation of the claim that the fire caused the vessel to sink.

128. Based on these findings the judge said that he was satisfied that the making and pursuit of the actual total loss claim substantially lengthened and increased the costs both of the action and of the trial. It was simply inconceivable that a claim for a constructive total loss alone would have generated costs on anything like the scale which had in fact occurred. This conclusion did not involve a finding that the owners had acted unreasonably or improperly but the judge referred to his view of the owners expert evidence "should such considerations be relevant".
129. Mr Eder takes issue with the way in which the judge has re-written history. He submits that the insurers assertion that at the moment the vessel became a constructive total loss she was about to and bound to sink raised factual issues which were intertwined with the issues arising on the claim for an actual total loss. The insurers assertion involved consideration of the size and spread of the fire and how much water was entering the vessel and from where. The judge was therefore quite wrong, Mr Eder submits, to conclude that the course of the litigation would have been different if there had been no actual total loss claim.
130. In paragraphs 8 to 22 we have referred to the way in which the insurers' challenge to the factual basis of the constructive total loss claim was developed and refined. It was accepted that during the trial no evidence was directed to establishing when the vessel was doomed. There was evidence about the size and spread of the fire, but that was directed to whether the fire caused the vessel to sink. Nevertheless Mr Eder submits that the "doomed to sink" point hung over the trial and in the event the judge made a finding of fact in paragraph 19 of his judgment in the owners favour.
131. In dealing with this point in his costs judgment the judge said:

"It is of course true that it was the defendants case that one reason why the CTL claim must fail is because the vessel was doomed to sink by reason of substantial water ingress from unknown causes. However this point was only the corollary of their very limited concession to the effect that, on the figures, the cost of repairing damage exceeded the value of the vessel only in the scintilla temporis before sinking. The point was therefore a limited one. On the figures it is was not a point of any substance. I dealt with it fairly summarily in paragraph 19 of my judgment …. On the figures it is obvious that an uncontrolled fire burning in the engine room and in the accommodation block in the manner described by the witnesses would readily and quite quickly cause damage, the cost of repair of which would exceed the insured value."

132. We think this shows that the judge's finding was based to an extent on the evidence he had heard about the size and spread of the fire. But this does not lead us to conclude that the judge's assessment of what would have happened if there had been no actual total loss claim was wrong. He was in the best possible position to make that assessment and we think the appeal must be considered on the basis that he was right.
133. Broadly the question on the appeal is whether the judge's order can be supported having regard to the fact that the owners lost on one major issue but recovered the full amount they had claimed and in doing so beat the Part 36 offer. There is however a discrete issue of construction of CPR 36 .21 which it is convenient to deal with first.
134. The relevant part of CPR 36 .21 says:

"36.21. (1) This rule applies where at trial –

(a) a defendant is held liable for more; …

than the proposals contained in a claimant's Part 36 offer.

……….

(3) The court may … order that the claimant is entitled to –

(a) his costs on the indemnity basis from the latest date when the defendant could have accepted the offer…

(4) Where this rule applies, the court will make the order referred to in paragraph … (3) unless it considers it unjust to do so…."

135. The owners submitted that "his costs" in CPR 36.21 (3)(a) meant "all his costs". If they were right about this it would not be unjust to give effect to the rule because if the insurers had accepted the offer there would have been no trial.
136. In dealing with the point of construction the judge said:

"… the rule is concerned with the basis of assessment of such costs as are ordered to be paid not with the basic incidence of costs. It would be surprising if a rule drafted in terms which appeared to focus on the basis of assessment should have been intended to bring about a rebuttable presumption as to the incidence of all costs incurred after a certain date, irrespective of the issue upon which they had been expended and of the relative success of the parties on that issue."

He went on to say that if he was wrong, he would have decided that it was unjust to award the owners all their costs on an indemnity basis from 1 July 2002.

137. The owners submit that the judge's construction of the rule was wrong. The rule does not say "such costs as he is awarded" and to restrict the meaning of the rule in this way is to emasculate the beneficial Part 36 regime.
138. We think the judge's construction of the rule was right for the reason he gave.
139. The judge's discretion had therefore to be exercised in accordance with the provisions of CPR 44.3, the relevant parts of which say:

"44.3 (1) The court has discretion as to –

(a) whether costs are payable by one party to another; ..

(2) If the court decides to make an order about costs –

(a) the general rule is that the unsuccessful party will be ordered to pay the costs of the successful party; but

(b) the court may make a different order. …

(4) In deciding what order (if any) to make about costs, the court must have regard to all the circumstances, including –

(a) the conduct of all of the parties;

(b) whether a party has succeeded on part of his case, even if he has not been wholly successful;

(c) any … admissible offer to settle made by a party which is drawn to the court's attention (whether or not made in accordance with Part 36)

(5) The conduct of the parties includes – …

(b) whether it was reasonable for a party to raise, pursue or contest a particular allegation or issue;

(6) The orders which the court may make under this rule include an order that a party must pay –

(a) A proportion of another party's costs ..

(c) costs from or until a certain date only; …

(f) costs relating only to a distinct part of the proceedings; and …

(7) Where the court would otherwise consider making an order under paragraph (6) (f), it must instead, if practicable, make an order under paragraph (6) (a) or (c)."

140. The judge explained his decision by saying:

"I turn finally to the order which I should make. My order must reflect that the claimants should in principle recover their costs of the CTL issue, and moreover should recover such of these costs as were incurred on and after the first day of the trial on an indemnity basis. However since so small a part of the trial was concerned with that issue the adjustment necessary on account of the latter consideration must be very small. The defendants should in principle recover their costs of the actual total loss issue. I do not consider it relevant that the claimants attempted on the first day of the trial to bring about a saving in trial costs. That attempt was in the event unsuccessful and the claimants must live with the consequences of pursuing and losing a claim which significantly increased the length and cost of the trial. I must if I can make an order that requires payment of a proportion of the costs of the other party rather than leaving to detailed (and expensive) assessment the question which costs related to which issue – see CPR 44.3 (7). Mr Berry submitted that an assessment that 10% of all of the costs of the action were attributable to the CTL issue would be generous to the claimants and would comfortably take into account the small uplift to which they are in principle entitled pursuant to Part 36. I consider that an assessment that 15% of the claimants' overall costs relate to the CTL claim eliminates all possibility that it might be an underestimate. It does not of course logically follow from this that 85% of the defendants' costs relate to the claim for the actual total loss, indeed I am inclined to think that a higher figure might well be justified. However Mr Berry did not argue for a figure higher than that presented by the difference between the proportion of the claimants' cost attributable to the CTL claim and 100%. Therefore an award to the defendants of 85% of all of their costs involves yet further generosity to the claimants, which is perhaps not inappropriate given that without resort to litigation they would not have recovered the US$3.million to which I have held them entitled."

141. So the judge gave the owners their costs of the constructive total loss issue and on an indemnity basis after the first day of the trial and underwriters their costs of the actual total loss issue. Each side's costs were about the same (£850,000) so the net effect of the judge's order was that the owners would bear all their own costs and pay 70% of the insurers' costs. In the last paragraph of his judgment the judge said he should stand back and ask himself whether this outcome was:

"in the light of the claimants recovery of $3m., together with interest, so plainly wrong that it must be regarded as perverse."

142. He answered this question by saying:

"In the light of the considerations identified by the Court of Appeal in the cases to which I have referred which must now inform the approach of the court to questions of this sort I do not consider that this outcome can be regarded as plainly wrong. It is the logical outcome of considering costs on an issue basis. Comparisons are dangerous, but if costs are in some cases to be considered on an issue basis, it is difficult to see how the present cannot be an appropriate case for such treatment."

143. It is trite to state but important to bear in mind that the rules prescribe the way in which the court's discretion as to costs should be exercised rather than any decision of this court on the facts of any particular case. The general rule is that the "unsuccessful party will be ordered to pay the costs of the successful party" (CPR 44.3 (2) (a)). Does this mean successful party on any particular issue or successful party in the litigation? As a matter of construction it must mean the latter. Where the rule refers to part of a case or a particular allegation or issue it says so.
144. Mr Berry suggested that in Summit Property Ltd. v Pitmans [2001] EWCA Civ 2020 Chadwick LJ had, as he put it, "redefined who has won". In that case the judge had ordered the successful defendant to pay 65% of the unsuccessful claimant's costs. This court confirmed that under the CPR it was no longer necessary for a party to have acted unreasonably or improperly before he could be required to pay the costs of the other party of a particular issue on which he had failed. In the course of his judgment Chadwick LJ said:

"An issue based approach requires a judge to consider, issue by issue in relation to those issues to which that approach is to be applied, where the costs on each distinct or discrete issue should fall. If, in relation to any issue in the case before it the court considers that it should adopt an issue based approach to costs, the court must ask itself which party has been successful on that issue. Then, if the costs are to follow the event on that issue, the party who has been unsuccessful on that issue must expect to pay the cost of that issue to the party who has succeeded on that issue. That is the effect of applying the general principle on an issue by issue based approach to costs."

Longmore LJ who gave the leading judgment in this case (with which Tuckey LJ agreed) did not put it in this way, but Chadwick LJ's general approach has subsequently been approved by this court in Stena v Irish Ferries Ltd. [2003] EWCA Civ 214.

145. We do not read Chadwick LJ's judgment in the way Mr Berry suggests. He does not refer to the terms of CPR 44.3 (2) (a). All he is saying is that if in carrying out the exercise required by CPR 44.3 (4) the court is considering the question of costs on an issue by issue basis and decides that costs should follow the event on a particular issue then "the party who has been unsuccessful on that issue must expect to pay the costs of that issue to the party who has succeeded…". This was said in a case where it had been submitted (based on Elgindata principles) that before the judge could make an order which not only deprived the successful party of his costs on an issue, but also required him to pay the costs of the other side, he had to be satisfied that there had been some element of unreasonableness or impropriety.
146. So we do not therefore think that Chadwick LJ said anything to cast doubt on what we think is the plain meaning of CPR 44.3 (2) (a). The owners were undoubtedly the successful party in this case. They had recovered the full amount of their claim plus interest. In A. L. Barnes Ltd. v Time Talk (UK) Ltd [2003] EWCA Civ 402 Barnes had obtained judgment for a substantial part of their claim but were ordered to pay 50% of the other side's costs because the judge had made a finding that one of its directors was dishonest, an issue which had taken up "the great bulk of court time". This had led the judge to conclude that the defendant was the successful party. In allowing the appeal Longmore LJ (with whom Clarke and Ward LJJ agreed) said:

"It does seem to me that the judge has, with the greatest respect, fallen into an error of principle. In what may generally be called commercial litigation… the disputes are ultimately about money. In deciding who was the successful party the most important thing is to identify the party who is to pay money to the other. That is the surest indication of success and failure."

147. Mr Eder's principal submission is that the judge fell into much the same error in this case. He did not start from the general rule that the owners were entitled to their costs. He gave effect to the outcome of his issue by issue analysis too mechanistically without having regard to all the circumstances of the case and then tested his conclusion by applying the wrong test.
148. We remind ourselves that the rules confer a wide discretion on trial judges on questions of costs. This court must exercise restraint when it is asked to upset orders for costs and should only do so if they are flawed. As Chadwick LJ said in Summit:

"The first question for this court is not whether it would have made the order which the judge made. The first question is whether this court is satisfied that the basis upon which the judge reached the conclusion that he did has been shown to be flawed. It is only if that question is answered in the affirmative that this court can properly interfere with the exercise of the judge of the discretion entrusted to him. It is only then that this court will go on to consider what order it will make in the exercise of its own discretion."

149. For the reasons the judge gave we do not think he can be faulted for adopting an issue by issue approach to this case. Nor was there anything wrong in principle with an order which resulted in the successful party having to pay the unsuccessful party a substantial proportion of its costs. The "winner takes all" principle no longer applies. But the question in this case is whether the judge was right to apply the issue by issue approach so rigorously ("logically" as he put it) to the exclusion of any other factors. This may be entirely justified in some cases. But was it justified in this case and, if it was not, does this mean that the judge's order was flawed?
150. The judge gave effect to the fact that owners had succeeded in recovering the amount they claimed in full and that in doing so they had beaten their Part 36 offer simply by giving them the costs of this issue. But the insurers had faced the constructive total loss claim from August 2001. As is apparent from the history to which we have referred, and as the judge was prepared to accept, they met the claim by making the minimum possible concession at the latest possible moment in the hope that something might eventually turn up which would justify them in making an allegation of scuttling. If they had admitted the factual basis of the constructive total loss claim earlier and analysed it correctly, it could have been dealt with as a preliminary issue at a much earlier stage when none of the reasons which later persuaded the judge not to take this course would have existed. This would have been possible even if the insurers maintained their positive factual case for the purposes of the "doomed to sink" point. The preliminary issue could have been tried on assumed facts or the factual dispute could have been resolved. As the judge said, that dispute did not raise a point of any substance and the answer was obvious. The important point here is that if the constructive total loss claim had been decided as a preliminary issue as it should have been there would have been no trial of the actual total loss claim. So on this view of the matter it was the insurers' reluctance to face up to the constructive total loss claim which led to both claims being tried together.
151. This is not a case where the issue on which the successful party lost was a separate head of claim: it was a separate basis for putting the successful party's only claim. Accordingly, unlike in many cases involving issue based orders for costs, this was a case where the issue on which the successful party lost would not have been litigated if the unsuccessful party had conceded the issue on which the successful party won. The force of that point in relation to costs in the present case is underlined by the existence of the owners' Part 36 offer.
152. These factors might have justified an order that the insurers should pay all or a very substantial part of the owners' costs. But this is not to say that the judge was wrong to take into account the fact that the owners had been wholly unsuccessful on the actual total loss claim and that this had occupied most of the time at trial and generated much of the cost before it. They had started with this claim and pursued it without ever having a sustainable case on causation. They made the choice to pursue it at trial when they need not have done. This is obviously a factor which pulls the other way. On its own it obviously justified the order which the judge made.
153. But we think it was an error of approach by the judge simply to visit the mathematical outcome of the issue by issue approach on the owners. This took no account of the other factors to which we have referred. The rules required him to have regard to all the circumstances of the case and it does not seem to us that he did so, at least not in the way the rules required. He should have stood back from the mathematical result and asked himself whether in all the circumstances, including the other factors to which we have referred, it was the right result. In fact he asked himself whether it was a perverse result. This was not the right question. If he had asked the right question we doubt that he would have decided that the owners should be deprived of all their costs and pay 70% of their unsuccessful opponents' costs.
154. For these reasons we think that the judge's decision was flawed. It is therefore our difficult task to decide what order should have been made. There is much room for manoeuvre if one takes the parties' submissions. Mr Eder submitted that the owners should have 75% of their costs; Mr Berry that the judge was too generous to the owners and that the insurers should have been awarded 90% of their costs (i.e. a net order of 80% in their favour). Obviously we should make a net order as we are told that both parties costs are the same and CPR 44.3 (7) exhorts us to do so if we can.
155. How then should we set about this task given that we have abandoned the winner takes all approach and rejected the mathematical or logical certainty of a strictly issue by issue approach in this case? Like many other judicial tasks the exercise is one which must involve weighing the various factors which have to be taken into account. There is nothing unprincipled about such an approach, although it can certainly be said that the result is less predictable than a strictly mathematical one. But if the latter approach does not produce the right result then obviously as a matter of principle it should not be followed.
156. We have already identified the main factors to be taken into account. It is helpful, we think, to start by looking at the range of orders which could be made. On the basis of what we have said above these are: an order that owners should have say 85% of their costs (to reflect the late appearance of the claim for a constructive total loss) and the judge's order that the insurers should have 70% of their costs. On these figures if the factors to be taken into account pulled in equal but opposite directions the result would be a small percentage in favour of owners. The task is then to see whether one set of factors pull more strongly than the other so as to adjust the percentage appropriately within the range. In this case we think the factors are very much evenly balanced. This view is best reflected we think in no order for costs either way. That is a fair result having regard to all the circumstances.
157. To this extent we allow the cross appeal on costs.

Order:

158. Main appeal on the CLT issue dismissed.
159. Cross appeal on costs allowed
160. Appellants to pay 70per cent of the respondent's costs of the appeal
161. Leave to appeal to the House of Lords refused.
(Order does not form part of the approved judgment)