This is an appeal against a costs order made by HHJ Rich QC, sitting as a Deputy Judge of the High Court. The order was made shortly after the beginning of the trial of an action for infringement of copyright in computer software. The parties settled the action on detailed terms set out in the schedule to a Tomlin Order.

As the parties were unable to reach a compromise on costs, they agreed that the costs of the proceedings should be decided by the court. HHJ Rich QC knew what the case was about. The first day (15 July 2002) had been set aside for pre-reading the trial bundles. On 16 July the judge had heard an oral opening from each side. Some of the claimant’s witnesses, including the claimant’s chairman and managing director, Mr Jowitt, were cross examined on their witness statements. The judge was then informed that, following an open offer by the defendant, the case had been settled, save for costs. New skeleton arguments confined to costs were submitted by each side. Oral submissions were made. The claimant, BCT Software Solutions Limited (BCT), which had agreed to ask the judge to embark on this difficult exercise, was aggrieved by the judge’s adverse costs decision given on 18 July. On a renewed application on 12 November 2002 BCT obtained permission from the Court of Appeal to pursue this appeal.

This court has been informed that the total costs of the appeal alone are in the region of £120,000. The costs of the action are likely to exceed £300,000 on each side. The total sum recovered by appellant BCT under the terms of the settlement was just over £10,000. The amounts involved speak volumes. Judicial comment is superfluous.

The arguments advanced on this appeal have demonstrated the real difficulties inherent in asking a judge to exercise his discretion in respect of the costs of an action, which he has not tried. There are, no doubt, straightforward cases in which it is reasonably clear from the terms of the settlement that there is a winner and a loser in the litigation. In most cases of that description the parties themselves will realistically recognise the result and the costs will be agreed. There will be no need to involve the judge in any decision on costs. If he becomes involved, because the parties cannot agree and ask him to resolve the costs dispute, the decision is not usually a difficult one for him to make.

There are, however, more complex cases (and this is such a case) in which it will be difficult for the judge to decide who is the winner and who is the loser without embarking on a course, which comes close to conducting a trial of the action that the parties intended to avoid by their compromise. The truth often is that neither side has won or lost. It is also true that a considerable number of cases are settled by the parties in the belief that the terms of settlement represent a victory, or at least a vindication of their position, in the litigation, or in the belief that they have not lost; or, at the very least, in the belief that the other side has not won.

In my judgment, in all but straightforward compromises, which are, in general, unlikely to involve him, a judge is entitled to say to the parties “If you have not reached an agreement on costs, you have not settled your dispute. The action must go on, unless your compromise covers costs as well.”

The disposition of a judge to help parties in negotiations for a settlement is understood and applauded. Good intentions are not, however, risk free. If acted upon too readily, commendable judicial intentions can make things far worse than they would have been if the judge had adopted the unpopular stance of requiring the parties to confront the realities of their litigation situation. The judge has a discretion to decline to do what the parties ask him to do. If, on the one hand, the action is for damages, it will be relatively easy for the judge to tell from the size of the settlement sum and from the litigation history (offers, payments in and so on) how the costs should be borne. As I have already said, it would be relatively unusual for the parties themselves not to agree on the costs of such cases. In more complex cases, however, involving a number of issues and claims for discretionary equitable relief, the costs position is much more difficult for the judge to resolve without actually trying the case.

This court is entitled to approach an appeal against a costs order, which has been made as part of a compromise, with an even greater degree of reluctance than is usually the case when it is asked to interfere with the discretion of the trial judge. (It has even been said that there is no appeal against an order for costs made by a judge in a case in which, as part of a compromise, it has been agreed by the parties that he should decide the issue of costs: Denne v. Denne (1977) CAT 4743, which is mentioned in footnote 2 to paragraph 9-03 on p 158 of The Law and Practice of Compromise 5^th Ed by David Foskett QC. In my view, there is no such hard and fast limit to the jurisdiction of this court.) If there is a point of principle in this case, which I very much doubt, it does not arise from the way in which the judge exercised his discretion, but from whether he should ever have embarked on this particular exercise at all. As both parties agreed that he should undertake the task, it is reasonable to expect them to accept his decision, unless it can be shown that the result is, in all the circumstances, manifestly unjust. I would certainly not be inclined to interfere with the judge’s decision simply because it is possible to detect imperfections in his approach or in his reasoning.

In my judgment, this court should only interfere with the costs order in this case, if BCT makes out a case of manifest injustice. It has not succeeded in that. I would dismiss the appeal. It is unnecessary, for the purposes of deciding the appeal, to deal, point by point, with every detailed argument advanced by each side. In the course of their submissions both Mr Peter Susman QC for BCT and Mr Henry Carr QC for C Brewer and Sons Ltd (Brewer) dipped into the detail of the compromised action to such an extent that, at times, there was an uncomfortable feeling that this court was being asked to try the action, which, because it was settled, has never been tried by anyone. The court was referred to pleadings, to correspondence, to parts of witness statements and to transcripts of evidence, as well as to the 14 paragraphs of the schedule to the Tomlin order.

BCT is a computer software house. It licences the use of software. On 19 September 2000 BCT took over the business of a predecessor BCT company, on its going into receivership. Brewer, which has a growing number of branches in its business as a paint retailer and distributor of building materials, was a licensee under an agreement of 25 May 1988 granted by the old BCT company in respect of software (the PACS system). It paid a one-off fee for a computerised stock control and accounting system, on which it came to depend at its various branches, and quarterly payments for software support. A dispute arose as to whether Brewer was making use of the PACS computer software outside the terms of the licence and whether it should pay a larger annual fee and additional fees for new branches or users. Proceedings were started on 7 June 2001. BCT claimed infringement of copyright in its computer software.

Almost a year later on 19 June 2002 and only a month before the trial was due to begin, BCT radically amended its claim. The issues were narrowed to four specific claims. BCT was ordered to pay the costs thrown away by reason of the amendments.

In the circumstances already described the action was settled shortly after the trial began and the judge made a costs order. The agreed overall effect of the order (after a correction under the slip rule) was that BCT was ordered to pay, on the standard basis, (a) 85% of all Brewer’s costs incurred before the amendments and (b) 62.5% of Brewer’s costs after the amendments. The judge described the process by which he decided that BCT should pay the bulk of the costs as “extremely broad brush.” There were four issues in the case. He assumed that each of the four issues generated 25% of the costs. He then divided the costs incurred by four. BCT had lost on three of the four issues. So prima facie the costs should be borne 75%/25% as between BCT and Brewer. Under the settlement BCT had only recovered a small sum compared with the total size of its original claim and the total costs of the proceedings. So the judge decided to halve the recovery of 25% on the issue on which BCT had won and awarded 12.5% to Brewer.

BCT has not appealed against the order that it should pay all its own pre-amendment costs and 85% of Brewer’s pre-amendment costs. (Brewer only challenges that order by its respondent’s notice, if BCT’s appeal on the post- amendment costs succeeds). As to the post-amendment costs, Mr Peter Susman QC for BCT criticised the approach of the judge as erroneous in principle and as producing an end result (i.e. that it should pay all its own post-amendment costs and 62.5 % of Brewer’s post-amendment costs) that was plainly wrong. The submissions may be summarised as follows:-

The judge erred in principle in notionally attributing the costs equally to the four issues identified in the very late amendments in June 2002. It was accepted that the judge was entitled to conclude that BCT had failed on three of the four specific issues and that it had succeeded on only one of them; but it was submitted that he was wrong to treat the four issues as equally significant in the context of costs and to attribute the costs equally among them. He had adopted an approach that was too “broad brush.” The majority of the costs incurred since the amendments was attributable to the one issue on which BCT had succeeded. That issue, raised in paragraph 12.2 of the Particulars of Claim, involved serious disputes of fact concerning 35 disputed additional users of the software material supplied to Brewer. Preparations for trial on that issue involved much time, effort and expense. At least half of the post-amendment costs were attributable to that issue. The other three issues, on which BCT lost, only involved short points of construction. The costs incurred were less than on the factual issue. In those circumstances the sensible order for the judge to have made was no order as to costs. (Mr Susman accepted, however, that the judge was not asked by either party to make no order as to costs.)

The judge had made a double deduction from BCT’s 25% portion of the costs on the issue on which it had succeeded. He had halved it again. That was an inadmissible double penalty.

BCT had sought in the action to vindicate its copyright in the material licensed to Brewer. After the opening of the case at trial, Brewer had finally admitted infringement and agreed to pay for additional users. BCT had to bring the action to protect its copyright.

I am not persuaded by any of BCT’s points, whether taken separately or cumulatively, that this court should interfere with the judge’s order. In general the appellate function in relation to judicial discretion on costs is that described in AEI Ltd–v- PPL [1999] 1 WLR 1507 at 1523 C to D:

There are the additional special circumstances mentioned earlier. In the absence of manifest injustice, an appellate court should not interfere with a discretion, which has not been exercised at the end of the trial, as is usually the case, but with the agreement of the parties when they have settled the case.

The obvious question is: what costs order ought the judge to have made different from the order that he has actually made? It was not contended by BCT that the judge had erred in failing to make an issues-based order i.e. separate orders as to the costs of each of the four specific issues according to who had won or lost on the issue, rather than making an overall apportionment of the total costs on a percentage basis. It was accepted that the proper approach, which the parties had urged him to follow, was to make a percentage apportionment of the totality of the costs between the parties. In those circumstances it is difficult to identify an error of principle in the judge’s approach. BCT’s complaint is directed to the proportions of the costs assessed by the judge in the exercise of his discretion rather than to the legal principles on which the costs order was based.

BCT’s submission that the judge should have made no order as to costs is particularly difficult to accept. That form of order certainly has its attractions in cases where it is difficult to fathom from the terms of a settlement who is the winner and who is the loser. It is an obvious candidate for serious consideration. There is, however, no convention that such an order should be made whenever the court is asked to decide costs on the settlement of the case: see Brawley v. Marizynski (No 1) [2002] EWCA Civ 756 at [18]; [2003] 1 WLR 813 at 819B. It is difficult to hold that a judge was wrong in principle or has committed a manifest injustice by not making an order which he was never asked to make. BCT had sought to secure an order more favourable than that to reflect its claimed victory in the proceedings.

The judge did his best to achieve justice between the parties without requiring them to incur the additional costs of fighting the whole case in order to establish the merits. If I had been the trial judge I do not think that I would have embarked on the exercise at all. If I had, out of a well-intentioned, though ill-advised, wish to assist the parties, I would probably have ended up making no order as to costs. I very much doubt whether I would have started out by dividing the costs into four equal portions. But the appeal is not about what I would have done in the judge’s place. It is about whether what the judge has done was legally erroneous and has produced a manifest injustice. I am persuaded by the submissions of Mr Henry Carr QC that the judge’s order was neither erroneous nor unjust and that there were features of the litigation which entitled the judge to make the order for the post-amendment costs against BCT.

BCT had sought an injunction to restrain the use of computer software, on which Brewer was dependent at its existing branches, and which, if granted, would have had a serious effect on its business. The litigation would put pressure on Brewer to accept new standard terms and conditions of the licence, which involved payment of a large annual fee. Flagrant damages were also claimed. It was also alleged that Brewer intended to infringe copyright by making a rival product in co-operation with ex-employees of BCT. The main part of the original case, including the allegation of making a rival product, was abandoned just one month before the trial after a large amount of costs had been incurred. The case was then almost totally recast. It raised four specific issues, all of which required some factual investigation, some more than others. The action was then settled on terms under which BCT recovered only £10,400 plus VAT for the additional users claim (13 users at £800 per user, compared to the 35 users claimed), a small amount when compared to the sums it had originally sought to recover and to the large amount of post-amendment costs incurred. Further, there was no question of a double penalty or of BCT having to proceed in order to vindicate its copyright. The dispute concerned the extent of the BCT licence to use PACS software. There was no dispute about the existence of copyright in PACS as a whole and there was no dispute that the number of users by Brewer had increased. BCT’s title was admitted by Brewer once an assignment from the old BCT and its receivers curing the defect of title had been produced at trial.

In those circumstances there was sufficient material before the judge to entitle him to conclude that BCT ought to pay the bulk of the post-amendment costs. The order was not manifestly unjust in all the circumstances. I would dismiss the appeal.

I agree that this appeal should be dismissed. I add some observations of my own only in order to emphasise that - as has already been said by Lord Justice Mummery in his judgment – a trial judge should be cautious before making an order as to costs in litigation in which all other issues have been compromised without a full trial.

The power to make an order as to the costs of civil proceedings is conferred by section 51(1) of the Supreme Court Act 1981. It is in the discretion of the court whether, in any particular case, that power should be exercised. That is made clear by CPR 44.3(1)(a). It finds expression in the opening words of CPR 44.3(2) – “If the court decides to make an order about costs -”. The first question for the court – in every case – is whether it is satisfied that it is in a position to make an order about costs at all.

In addressing that question the court must have regard to the need (if an order about costs is to be made) to have a proper basis of agreed or determined facts upon which to decide, in the light of the principles set out under the other provisions in CPR 44, what order should be made. The general rule, if the court decides to make an order about costs, is that the unsuccessful party will be ordered to pay the costs of the successful party – CPR 44.3(2)(a). But the court may make a different order – CPR 44.3(2)(b). Unless the court is satisfied that it has a proper basis of agreed or determined facts upon which to decide whether the case is one in which it should give effect to “the general rule” - or should make “a different order” (and, if so, what order) – it must accept that it is not in a position to make an order about costs at all. That is not an abdication of the court’s function in relation to costs. It is a proper recognition that the course which the parties have adopted in the litigation has led to the position in which the right way in which to discharge that function is to decide not to make an order about costs.

In a case where there has been a judgment after trial, the judge may be expected to be in a position to decide whether one party or the other has been successful overall; whether one party or the other has been successful on discrete issues; whether the fact that the party who has been successful overall but unsuccessful on some issues calls for an order which reflects his lack of success on those issues; and whether - having regard to all the circumstances (including conduct) as CPR 44.3(4) requires – the order for costs should be limited in one or more of the respects set out in CPR 44.3(6). But where there has been no trial – or no judgment – the judge may well not be in a position to reach a decision on those matters. He will not be in a position to decide those matters if they turn on facts which have not been agreed or determined. In such a case he should accept that the right course is to decide that he should not make an order about costs. As the arguments on the present appeal demonstrate, it does the parties no service if the judge – in a laudable attempt to assist them to resolve their dispute – makes an order about costs which he is not really in a position to make.

It does not, of course, follow that there will be no cases in which (absent a judgment after trial) the judge will be in a position to make an order about costs. There will be cases (perhaps many cases) in which it will be clear that there was only one issue, that one party has been successful on that issue, and that conduct is not a factor which could displace the general rule. But, in such cases, the answer to the question which party should bear the costs of the litigation is likely to be so obvious that, as Lord Justice Mummery has pointed out, the judge will not be asked to decide that question. It will be agreed as one of the terms of compromise.

The cases in which the judge will be asked to decide questions of costs - following a compromise of the substantive issues – are likely to be those in which the answer is not obvious. And it may well be that, in many such cases, the answer is not obvious because it turns on facts which are not agreed between the parties and which have not been determined. The judge should be slow to embark on the determination of disputed facts solely in order to put himself in a position to make a decision about costs. As Lord Justice Mummery has put it, the better course may be to require the parties to confront the realities of their litigation situation; to point out to them that, if they have not reached an agreement on costs, they have not settled their dispute and the action must proceed to judgment.

I share Lord Justice Mummery’s view that this is a case in which the judge could not have been criticised if he had taken that course. For my part, I think he would have been wise to do so. But it is not open to the appellant to complain that the judge set out to do what both parties had asked him to do – that is to say, to make an order about costs and to decide what order to make on the material before him and without determining disputed facts. Nor is it open to the appellant to complain that, in seeking to perform that task, the judge adopted an approach which he, himself, described as “broad brush”. It is difficult to see what other approach the judge could have adopted in the circumstances.

The judge approached his task on the broad basis that there were four issues in the case; that the appellant had lost on three of those issues and won on the fourth; and that each party should have the costs of the issues or issue on which it had been successful. To my mind that approach is, aptly, described as “issue-based”. There is no doubt that an issue-based approach of that kind was open to the judge – see CPR 44.3(6)(f) – and the judge cannot be said to have erred in principle in adopting that approach.

Nor can it be said that the judge erred in failing to carry that approach through to what might be seen as the logically consistent result – an order for the separate assessment of the costs of each issue with consequent orders for payment of the costs so assessed. It has been recognised by the rule-making body that the separate assessment of costs, issue by issue, is likely to be time-consuming, difficult and expensive; and so add to the ultimate burden on the paying party. So, where a court is considering an issue-based order under CPR 44.3(6)(f), it must, instead, make an order under paragraphs (6)(a) or (c) if practicable – CPR 44.3(7). Paragraph (6)(a) enables the court to make an order that one party pay a proportion of the other party’s costs. In a context where the court would – but for CPR 44.3(7) – consider it appropriate to make an issue-based order under paragraph (6)(f), the proportional order which it makes under paragraph (6)(a) ought to reflect the underlying premise that each party should have its own costs of the issues on which it has been successful and should not bear the other party’s costs of those issues. In that context the court should determine the proportion of the overall costs to be paid by one party to the other by asking itself what proportion of the overall costs should be attributed to each the respective issues and making the appropriate adjustments to reflect success or failure on those issues. That is what the judge did in this case.

I agree with both judgments.