By the main judgment in this case ([2004] EWCA Civ 1568) we upheld Pumfrey J’s finding that the SKB patent was not infringed but reversed his findings that it was invalid by reason of anticipation or obviousness over ‘407 and obviousness over the erythromycin patent. Because we were not asked to do so, we did not rule on further questions of construction and attacks of invalidity though we expressed the view that some of these had “considerable force.”

When the case came back before us for oral argument as to the form of order there was a considerable debate as to the appropriate order for costs. Other matters were disposed of readily and I need not recite them. An order was made dealing with these. We reserved judgment on the question of costs and matters consequential upon it. This judgment deals with the appropriate order as to costs.

The overall result of valid but not infringed was reached by decision of a host of issues and sub-issues. Both Pumfrey J and we had to consider a number of further issues which in the event it was not necessary to decide (see e.g. para. 115 of my judgment). On top of that there were issues which Apotex had to prepare to deal with (claims 3 and 8) which SKB had to abandon at the commencement of trial because of the decision of the Court of Appeal in the BASF case. This required a concession that these claims were invalid and in due course the necessary consequence of amendment down to the original claims 10 and 11. Many of the issues overlapped, for instance the denial of claim coverage of the Apotex process being coupled with an assertion of invalidity if the position were otherwise.

The result of all this is that it is just not possible to disentangle the various issues and make a precise allocation of costs as to who was the winner of each. For that reason the parties both agreed that the thing to do in principle was to decide to reach an overall conclusion as to whether one side should be paying the other costs, and if so to set a proportion of those costs allowing for all appropriate factors by way of reduction. The parties were also agreed that the costs of first instance and appeal should be dealt with separately on such a basis.

We agree. To set a costs judge with the case the task of unravelling the costs of the issues would be to ask the impossible. We must do the best we can to decide now which side is a payer and the appropriate percentage if it is.

The disparity between the parties’ positions is substantial. Apotex say they should have 65% of their costs below and 60% of the costs of the appeal. SKB say they should have 18% of their costs below and 46% of the costs of the appeal, each to be on an indemnity basis.

Before getting down to the apportionment exercise it is necessary to deal with a number of points of principle.

Following judgment in the BASF action, upheld on appeal, Pumfrey J certified that “the validity of claim 11 of the Patent was unsuccessfully contested.” Claim 11 became claim 2 of the D version of the patent, the version before us. SKB say accordingly s.65 of the Act applies. It provides:

“65(1) If in any proceedings before the court or the comptroller the validity of a patent to any extent is contested and that patent is found by the court or the comptroller to be wholly or partially valid, the court or the comptroller may certify the finding and the fact that the validity of the patent was so contested.

(2)

Where a certificate is granted under this section, then, if in any subsequent proceedings before the court or the comptroller for infringement of the patent concerned or for revocation of the patent a final order or judgment or interlocutor is made or given in favour of the party relying on the validity of the patent as found in the earlier proceedings, that party shall, unless the court or the comptroller otherwise directs, be entitled to his costs or expenses as between solicitor and own client (other than the costs or expenses of any appeal in the subsequent proceedings).”

It is not in dispute that although the section speaks of “costs as between solicitor and own client” the reference should be taken to mean “costs on the indemnity basis” within the meaning of CPR 44.4(1). We were told that this would in practice raise the level of recovery from about 70% to 90% of the actual fees paid by a client to his solicitor.

The fact there is this very old-fashioned reference in the section shows its age. According to Mr Blanco White’s book Patents for Inventions, 4^th Edn. (1974) provisions of this sort go back to 1835 with various amendments. I doubt that they really serve any useful purpose these days – particularly in primary legislation which is inflexible and hard to change. The principles of the award of costs are now much more sophisticated and flexible than they once were. They are well able to cope with the vice at which the provision is aimed – vexing a patentee with essentially the same points on which he has previously won. Nonetheless, so far as the provision applies, it is relevant here.

Mr Watson disputes that it applies. First, so far as the appeal is concerned, he draws attention to the closing words of s.65(2). He says they have the effect that where there is a certificate of validity from a previous action, in a subsequent action where the patentee is again victorious, he can have his costs of the victory on an indemnity basis, but only at first instance. The provision, he submits, does not apply to appeals.

The conclusion is completely irrational – why should a patentee vexed twice get a special costs order at first instance but not for an unsuccessful appeal? Mr Watson did not suggest any rational basis for his construction – he submitted that is simply the effect of the words used.

No one doubts the principle that one leans heavily against a construction which is irrational, the more irrational it is the harder one leans. But in the end if one is forced to the conclusion that Parliament intended that which seems irrational, then that is the law. Here I think Mr Watson is right: I can find no way of reading “subsequent proceedings” as meaning anything other than the second, vexing, proceedings.

I wondered whether the predecessor provisions to s.65 might help. Section 64 of the Patents Act 1949 read:

“64(1) If in any proceedings before the court of validity of any claim of a specification is contested, and that claim is found by the court to be valid, the court may certify that the validity of that claim was contested in those proceedings.

(2)

Where any such certificate has been granted, then if in any subsequent proceedings before the court for infringement of the patent or for revocation of the patent, a final order or judgment is made or given in favour of the party relying on the validity of the patent, that party shall, unless the court otherwise directs, be entitled to his costs as between solicitor and client so far as concerns the claim in respect of which the certificate was granted:

Provided that this subsection shall not apply to the costs of any appeal in any such proceedings as aforesaid.”

The proviso, as Mr Blanco White observed at para. 12-145, was ambiguous: did “any such proceedings” refer to the original proceedings only or did it refer to all appeals? It was possible to read the excluding proviso as making it clear that only an appeal in the original action was not subject to the special costs rule. But that possible way of reading that provision cannot apply here – the word subsequent, which appears for the first time in the 1977 Act, is explicit. It is the appeal in the subsequent action which is exempted from the special costs rule.

Nor is much help to be gained from even earlier legislation. I looked at s.35 of the Patents Act 1907 as amended in 1919. This read (the 1919 amendments being in bold):

“In an action for infringement of a patent, the court may certify that the validity of any claim in the specification of the patent came into question; and if the court so certifies, then in any subsequent action for infringement the plaintiff in that action on obtaining a final order or judgment in his favour shall, unless the court trying the action otherwise directs, have his full costs, charges and expenses as between solicitor and client so far as that claim is concerned.”

Neither in 1907 nor in 1919 did Parliament explicitly deal with appeals at all. The focus on “the court trying the action” seems to be a reference to the first instance court only. If that were right, then in those days the special costs rule simply did not apply to appeals at all. I note in passing the remarkable fact that prior to the 1919 amendment, a special costs order could be obtained, subject to the court’s discretion, against a defendant who did not even seek to re-challenge the claim.

The upshot is therefore that Mr Watson is right about s.65(2). SKB cannot pray it in aid directly so far as the costs of the appeal before us is concerned. It does not follow, however, that the general principles under the CPR may not not apply to have the same effect. To that I will turn in due course. Before I do so I should mention Mr Watson’s other points on s.65 since they would apply to the costs below.

Mr Watson pointed out that Pumfrey J’s certificate related only to original claim 11, which became claim 2. In issue here was also original claim 10(i) - which became claim 1. No certificate extends to that. Mr Watson submitted that the special costs rule could not apply to claim 1 – that the inflexible system of s.65 applies only so far as it goes.

Consider the problem generally. One can suppose a case where an attack on a certified claim was wholly unrelated to an attack on an uncertified claim. It would be odd indeed to make a challenge to the uncertified claim subject to the special costs rule just because the certified claim was challenged too. That indicates that Mr Watson is right in principle. On the other hand suppose (as was the case here) the case on the certified claim and that on the uncertified claim was essentially the same. Why should the special costs rule not apply? The answer to my mind is that it does – only costs unrelated to the certified claim are excluded from the rule. So although Mr Watson is right in principle, in the present case it makes no difference. It follows that the rule does apply to the costs before Pumfrey J of the fight over uncertified claim 1 because it was the same fight as the fight over certified claim 2 – there were no extra claim 1 costs.

Mr Watson’s other point was based on s.63 of the Patents Act 1977. Subsections 1 and 2 of the section read:

“63(1) If the validity of a patent is put in issue in proceedings for infringement of the patent and it is found that the patent is only partially valid, the court or the comptroller may, subject to subsection (2) below, grant relief in respect of that part of the patent which is found to be valid and infringed.

(2)

Where in any such proceedings it is found that the patent is only partially valid, the court or the comptroller shall not grant relief by way of damages, costs or expenses, except where the plaintiff or pursuer proves that the specification for the patent was framed in good faith and with reasonable skill and knowledge, and in that event the court or the comptroller may grant relief in respect of that part of the patent which is valid and infringed, subject to the discretion of the court or the comptroller as to costs or expenses and as to the date from which damages should be reckoned.”

Mr Watson submitted that applied here because, until the start of the trial when SKB disclaimed any reliance on them, claims 3 and 8 were in issue. The patent was then duly amended to delete them. So he submitted in essence it was found that the patent was only partially valid and accordingly no costs could be awarded unless it was proved that the specification was framed in good faith and with reasonable skill and knowledge. That would have to be considered at subsequent inquiry and, he submitted, his clients had material to show that SKB were in possession of information which ought to have told them that claims 3 and 8 were invalid. Hence there was a lack of reasonable skill and knowledge.

This is a false point. In the second action Pumfrey J did not make any finding that the patent was partially valid. SKB conceded it in view of the decision of the Court of Appeal. Mr Watson submitted that if “found that a patent is only partially valid” did not include the court acting on a concession to that effect, then a patentee could always avoid the application of s.63(2) by making a concession. That, he submitted, could not be right. I do not see why. It is not the case that every time a patent is amended (for instance in proceedings before the Comptroller) the patentee can only recover damages for infringement of the amended claims if he proves good faith and reasonable skill and knowledge. Patentees often apply to amend their patents, whether in proceedings or without. In so doing they may well accept invalidity of the original claims. Where they do there is no question of s.63(2) applying. I can see no difference where the concession happens to be made in proceedings for infringement. In my opinion it is only where the court has made a finding as to partial invalidity – which involves determination of a dispute about the point – that the subsection applies. Acting on a concession is not making a finding.

That being so, Mr Waugh’s other answer to the point does not come into play. Since it was argued I will deal with it nonetheless. It was of an extremely technical nature. He submitted that if Pumfrey J did make a finding in the present proceedings it was solely in the claim for revocation and not in the subsequent counterclaim (brought by a separate action) for infringement. Hence it could not be said that validity was put in issue in proceedings for infringement and the section does not apply. I regard that as wholly without merit both as a formal matter and as a matter of substance. As a formal matter Apotex actually pleaded invalidity of the patent in its defence to SKB’s infringement action and thus indeed did in that very action put validity in issue. As a matter of substance validity was raised as a defence to the infringement proceedings.

Mr Watson submitted that the Court should take account of the fact that SKB, even though they ultimately lost on infringement, almost certainly did very well out of this case. An interim injunction ran for about 14 months. Although SKB will have to compensate Apotex for holding them off the market during the currency of that injunction (assuming Apotex can prove that was its effect), the amount involved is likely to be very much less than the benefit SKB had of holding on to the whole market at an undisturbed monopoly price. Mr Watson may be right about this, but I think it is wholly irrelevant to any question of allocation of costs incurred. A party who obtains an interim injunction when it turns out that he had no claim is not thereby to be treated as a wrongdoer (see e.g. Air Express v Ansett Transport (1979) 146 CLR 249). That being so, I cannot see that the fact he obtained (or benefited from) an interim injunction is relevant as to how the discretion as to costs of the ultimate trial is to be exercised.

In respect of all intellectual property matters the general rule is that the CPR and associated practice directions apply, unless a rule in Part 63 or its practice direction provides otherwise (CPR 63.2). There is no such rule. So, subject to the two statutory provisions discussed above, the general rules apply as much to patent actions as to any other action.

It follows that all the factors and matters set forth in CPR 44.3 apply to how the court should exercise its discretion as to costs. Prior to the CPR a party who was successful overall was not normally deprived of its costs of an issue it took unsuccessfully unless it has done so unreasonably, see Re Elgindata (No. 2) [1992] 1 WLR 1207. But since the CPR a more issue-by-issue approach is appropriate, see Summit Property v Pitmans [2001] EWCA Civ 2020, applied in a patent action, Stena v Irish Ferries [2003] EWCA Civ 214. Even before the CPR an issue-by-issue approach was, as an exception to the Elgindata approach, applied in patent actions because of the “large number of issues and the very extensive costs that can be incurred” per Aldous LJ in Rediffusion v Singer Link [1993] FSR 369 at 410.

An issue-by-issue approach is therefore one that should be applied so far as it reasonably can. On the other hand such an approach is not the be-all and end-all. Whether or not “it was reasonable for a party to raise, pursue or contest a particular allegation” remains a relevant factor to be taken into account as part of the conduct of the parties (see CPR 44 rule (3)(a) and (5)(b).

Before turning to this particular case I should say something about this. Although an issue-by-issue approach is likely to produce a “fairer” answer and is likely to make parties consider carefully before advancing or disputing a particular issue, it should not be thought that it is capable of achieving a “precise” answer. The estimation of costs, like that of valuation of property, is more of an art than a science. True it is that one can measure certain things (such as pages of witness statements or transcript devoted to a particular issue) but they can only be indicia to be taken into account. It would be dangerous to rely upon them as absolutes. Indeed brevity of a document, or a cross-examination, may be the result of great care: was it Hazlitt who apologised for the length of a letter, excusing himself on the grounds that he had not enough time to compose it?

It follows that there is no “precise” figure of costs which, in theory with perfect measurement tools, one could reach. The best that can be achieved is an estimate which is necessarily going to be somewhat crude. The costs in this case are very great, reflecting the much larger sums at stake. We were told the total figure is about £8m for both sides all in (trial, appeal, interlocutory matters). This reflects the much bigger sums at stake for a top-selling pharmaceutical. Mr Watson pointed out that a 1% difference given to one side or the other could amount alone to £80,000. He is right, but that does not mean that anything like an accuracy of 1% can be achieved.

Pumfrey J awarded Apotex 76% of their costs. Although they won on infringement and validity, he deducted 10% for an issue called the “Ferrosan Memorandum” and 14% because Apotex failed on their non-infringement argument based on the proposition that their process started with an anhydrate.

Both parties accept that the figure should be varied to reflect this court’s judgment. Apotex say that the 14% deduction should go because this court held that their point had “considerable force” (see para. 115). They accept that the 10% deduction should remain undisturbed because the Ferrosan memorandum never featured in the appeal. Apotex accept there should now be a deduction because they lost on invalidity over ‘407 and erythromycin patent. That, they suggest, should be 30%. Hence they reach the 65% figure.

SKB say the 14% deduction should remain. And there should be such a big “deduction” for ‘407 and erythromycin issues that the position moves to 18% in their favour on an indemnity basis. They invoke s.65 because of the certificate of contested validity.

First then the 14% deduction. The Judge’s reason for the deduction was more than just the question of construction. He found that Professor McClelland’s evidence was not satisfactory. We did not go into this aspect of the evidence and were not asked to. In the circumstances I would not disturb the Judge’s deduction.

Next, the indemnity basis. I think SKB are entitled to invoke s.65 for the first instance costs for the reasons given in paragraph 18 above. It makes no difference that the uncertified claim 2 was also in issue – the 407 and erythromycin attacks would have been the same if it were not.

Moreover quite apart from s.65 I think an award on an indemnity in respect of these issues is justified on quite different grounds. We were told that Apotex had an observer in court throughout the BASF proceedings. Given the commercial importance of the case it is near certain that they knew of those proceedings from a very early stage. They did not apply to join in, preferring to let BASF try to knock the patent out. Only when that failed did they bring their own proceedings. These include founding their case for invalidity on the same major piece of prior art (‘407) as in the earlier case. So the re-litigation of the ‘407 attack was as a result of Apotex’s deliberate decision to hold back and wait and see. That they were entitled to do, but having lost that issue I think the general discretion to award indemnity costs in respect of it can and should be invoked.

What is the appropriate amount to attribute to these two attacks? Apotex say that about ¼ of their costs were taken up with them. SKB estimate that 42% of its costs relate to all invalidity issues of which by far the dominant element relates to ‘407. They point out that these costs are subject to the indemnity principle. It is, of course, SKB’s costs of these issues which matter, Apotex’s being something of a cross-check. Allowing for the fact that some of the SKB costs on validity do not relate to these issues I think the appropriate figure to take is 40% of SKB’s costs on an indemnity basis which translates into about 50% of SKB’s costs on a standard basis (we were told that the difference between standard and indemnity costs is up to 20%). Translating this to a deduction from Apotex’s (and allowing for the fact that SKB’s overall costs are somewhat greater than Apotex’s) gives one a deduction which can be fairly estimated as 60%. This would give Apotex overall 16% of their costs below on a standard basis.

That, I think, is about the right figure. It also reflects who was the overall commercial winner who is entitled to some general costs of the action as a whole. I am conscious that SKB have put in calculations suggesting even bigger figures for the deduction – figures big enough to put them in “credit.” The calculation proceeds on the basis of working on numbers said to represent actual figures of costs on both sides. Whilst they of course must be borne in mind I do not accept they are conclusive. The figure of 16% upon which I have settled reflects the overall justice of the case, namely that Apotex were the commercial winners. As such I think it is they who are entitled to the general costs of the proceedings which a pure issue-by-issue approach ignores.

The appeal took six days. Most of the time (about 4 days) was spent on the ‘407 issue (with a little added for “Bodmin” and erythromycin). SKB were successful on that. The remaining time concerned the finding of non-infringement (upheld) and attacks on validity which, in the event, we were not asked to and did not rule upon.

Although the greater proportion of time involved ‘407 I think it fair to take into account that when it was being argued the Court was, necessarily, proceeding rather slower than it would have done if the point had not been dealt with first. Patent actions always speed up as the court becomes more immersed in the technology and the patent.

I have decided (see above) that s.65 does not apply directly to the appeal. But for the reasons I have indicated above I think the general powers of the court to award indemnity costs nonetheless apply to the arguments about ‘407.

Taking all these matters into account, I think the fair figure to be reached so far as the appeal is concerned is that SKB should have 25% of their costs of the appeal on a standard basis. This allows for the fact that about ½ of the time (allowing for speeding up) was spent on ‘407 and that this time carries indemnity costs.

Accordingly I would award Apotex 16% of their costs below and SKB 25% of their costs of the appeal. Given this conclusion I would hope that the parties can agree consequential matters which include:

I respectfully agree with the order that Jacob LJ proposes and gratefully adopt his reasons, save in the following minor respects.

In my judgment, it is clear from section 65(2) of the Patents Act 1977 taken on its own that the expression “subsequent proceedings” in the final words in parenthesis refer to proceedings (a) which are separate from the proceedings in which a certificate is granted under section 65(1) and (b) in which the relevant order for the purposes of section 65(2) is made subsequent to the grant of the certificate. For my own part, differing with diffidence from the view expressed by Jacob LJ, that conclusion is not irrational. It is open to Parliament to take the view that only the costs of the trial before the court or the Comptroller-General of Patents, Designs and Trademarks (“the comptroller”) should be subject to the special rule in section 65(2) and that no provision needs to be made for the costs of the appeal. Indeed, as Jacob LJ points out, there is power in the appellate court under Civil Procedure Rules to award costs on an indemnity basis, and the existence of section 65(2) would no doubt be a relevant consideration in the exercise of the appellate court’s discretion in this regard. Parliament cannot wholly dispense with section 65 since it deals not only with proceedings before the court but also proceedings before the comptroller. His powers to award costs are conferred not by the Civil Procedure Rules or by the practice of the court but by the provisions of the 1977 Act. The rule-making power in section 123 of the 1977 Act does not appear to confer powers to make rules under the 1977 Act in relation to costs. Even as regards proceedings before the court, Parliament clearly intended to create a presumption in favour of indemnity costs in every case in the circumstances there specified. On that basis, section 65(2) needed to be included in the 1977 Act and (for the reason already given) to refer to both the court and the comptroller. Similarly, section 106 of the 1977 Act creates a special rule as to costs in infringement proceedings which has no counterpart in the Civil Procedure Rules.

I am in agreement with Jacob LJ.

The court is always very reluctant to award costs on an issues basis for the reasons that this court gave in English v Emery Reimbold & Strick Ltd [2002] 1 WLR 2409 at [115] and [116]:-

“... we would emphasise that the Civil Procedure Rules requires that an order which allows or disallows costs by reference to certain issues should be made only if other forms of an order cannot be made which sufficiently reflect the justice of the case: see rule 44.3(7), above. In our view there are good reasons for this rule. An order which allows or disallows costs of certain issues creates difficulties at the stage of the assessment of costs because the costs judge will have to master the issue in detail to understand what costs were properly incurred in dealing with it and then analyse the work done by the receiving party’s legal advisers to determine whether or not it was attributable to the issue the costs of which had been disallowed. All this adds to the costs of assessment and to the amount of time absorbed in dealing with costs on this basis. The costs incurred on assessment may thus be disproportionate to the benefit gained. In all the circumstances, contrary to what might be thought to be the case, a “percentage” order, under rule 44.3(6)(a), made by the judge who heard the application will often produce a fairer result than an “issues based” order under rule 44.3(6)(f). Moreover such an order is consistent with the overriding objective of the Civil Procedure Rules. ...

Wherever practicable, therefore, the judge should endeavour to form a view as to the percentage of costs to which the winning party should be entitled or alternatively whether justice would be sufficiently done by awarding costs from or until a particular date only, as suggested by rule 44.3(6)(c).”

In response to the question at the end of paragraph 27 of the judgment of Jacob LJ, I am unaware of any citation from Hazlitt on this point, but a similar comment was made Blaise Pascal:-

“Je n’ai fait celle-ci plus longue parce que je n’ai eu de loisir de la faire plus courte.” (Lettres Provencales (1657) No.16).

As regards paragraph 36, I am in agreement with the figure of 16%. I have reached this figure as a result of a consideration of the issues that were won and lost by either party and the evaluation made by Pumfrey J. I am not in a position to say Apotex were the “commercial winners”.

Subject to the aforesaid, I agree with the judgment of Jacob LJ.

The provisions of the 1977 Act mentioned above have been amended by the Patents Act 2004, but the relevant provisions of that Act have not yet been commenced and do not in any event affect the matters in this judgment.

I can well understand why solicitor and client/indemnity costs should be awarded if the party relying on the certified validity of his patent successfully wards off a new attack upon it in subsequent proceedings. What is more difficult to fathom is why that should be the rule for the trial but not for any appeal. Yet that, it seems to me, is the literal effect of s. 65 of the Patents Act 1977 and I can find no justification for displacing the ordinary meaning of the words in parenthesis in s. 65(2) despite the anomaly created by that conclusion. CPR 44 does, however, permit indemnity costs to be awarded in the appeal and I agree that they should be awarded in the circumstances of this case for the reasons given by Jacob L.J.

This is a case where the court has to do its best to allocate costs on an issue by issue basis, imprecise though that approach may be. I agree with Jacob L.J.’s assessment and with the order he proposes.