England and Wales Court of Appeal (Civil Division) Decisions

IN THE SUPREME COURT OF JUDICATURE
COURT OF APPEAL (CIVIL DIVISION)
ON APPEAL FROM THE HIGH COURT OF JUSTICE
CHANCERY DIVISION (PATENTS COURT)
(NEUBERGER J)

		Royal Courts of Justice Strand, London, WC2A 2LL Monday 29 October 2001

LORD JUSTICE ROBERT WALKER:

Introduction

1. This is an appeal from an order which Neuberger J made on 28 July 2000, determining two preliminary issues in a patent action. Before the judge the claimant was Rohm and Haas Company ("Rohm and Haas") and the defendant was Collag Ltd ("Collag"), which also counterclaimed for revocation of the patent in suit and a declaration of non-infringement. The preliminary issues were concerned only with the counterclaim for a declaration of non-infringement, on which Collag was successful. Rohm and Haas appealed with the permission of the judge.
2. Since the inception of the appeal there have been some changes in the way the battle-lines are drawn up. The claim for revocation of the patent in suit has been abandoned. Moreover the patent has been assigned to Dow Agrosciences LLC ("Dow"). Collag has gone into administrative receivership and Agform Ltd ("Agform") has acquired various rights from its receivers. This court has therefore, by consent, added Dow as a second appellant and Agform as a second respondent. In substance it is now the only respondent, since Collag's receivers have taken no part in the appeal. Agform is interested only in the preliminary issues.
3. All the parties are or were producers of agrochemicals. The patent in suit, European Patent (GB) No.484147 (filed on 30 October 1991 and granted on 26 August 1998, with a priority date of 31 October 1990) is for a process of manufacturing water-dispersible granules whose active ingredient is propanil. Propanil is a broad-spectrum herbicide used for crops such as wheat, rice and potatoes.
4. Farmers normally apply herbicides in liquid form using either a specialised self-propelled sprayer or spraying equipment mounted on or towed behind a tractor. But for obvious reasons of convenience and economy the herbicide is supplied to them in solid form. Water-dispersible granules with a high content of the active ingredient are the most satisfactory form, for the reasons mentioned in the introductory paragraphs of the specification of the patent in suit.
5. The specification explains that granules (typically at least 1 mm in diameter) are preferable to powder (particles whose diameter is measured in microns, or millionths of a metre) because powdered formulations tend to become compacted and so lose uniformity. However granules have their own problems. It is important that they should be readily dispersible in water (the liquid medium most conveniently available to farmers). But granules which are readily dispersible in water tend not to be sufficiently resistant to attrition (that is, they produce dust in the course of being handled and transported).
6. In the case of propanil there is a further manufacturing problem because it starts to melt at a relatively low temperature, such as may be encountered in a grinding or milling process (specification p.2 11.31-5):

"Low-melting solids such as propanil present an especially difficult problem in the preparation of a dispersible granule formulation. The low-melting solid, herein defined as melting below 100^oC, tends to melt or become sticky during or subsequent to the grinding process which is a necessary step in preparing dispersible granules. EP-A-0,252,896 describes a possible solution to this problem which requires micro-encapsulation of low-melting pesticides prior to granulation. However, microencapsulation involves additional pressing steps and adds to the cost of the overall formulation."

7. The specification states what the invention has achieved (specification p.2 11.39-55; lines 50-55 are repeated verbatim as claim 1 at the end of the specification):

"In the case of the herbicide propanil, it has not heretofore been possible to produce a dispersible granular product which combines the features of a high active ingredient content and good suspensibility and dispersant properties, as well as resistance to attrition – thereby avoiding the formation of a dust.

Commercial formulations of propanil dispersible granules are available, but these products have deficiencies in that the active ingredient content is relatively low and/or they fail to provide adequate suspension or dispersibility characteristics or are not resistant to attrition.

We have now discovered a process for producing dispersible granule formulations of propanil herbicide. The process can provide granules which contain more than 60% active ingredient, which also have good suspensibility and dispersibility characteristics, and which resist attrition.

Accordingly one aspect this invention provides a process for producing a dispersible propanil granule comprising:

a) combining one or more surfactants with propanil and milling to a particle size of less than 20 microns to form a pre-mix;

b) adding less than 25 per cent by weight water and optionally a wetting agent to said pre-mix and mixing until a paste is obtained;

c) granulating said paste thereby producing granules; and

d) drying said granules to a moisture content of less than 2 per cent by weight. "

The preliminary issues

8. The two preliminary issues have arisen in the following way. On 20 April 1999 Rohm and Haas commenced proceedings against Collag for infringement of the patent in suit, the alleged infringement being by the manufacture of propanil dispersible granule formulations. By its reamended defence and counterclaim Collag admitted that from December 1998 until May 1999 it had manufactured extruded water-dispersible granules with a propanil content of about 80 per cent. These were marketed under the names of Farm 80 DF and Fitoris. Farm 80 DF was manufactured by two processes, which have been referred to throughout the litigation as Process A and Process B. Collag counterclaimed not only for revocation of the patent in suit (on grounds of lack of novelty and obviousness) but also for a declaration of non-infringement in relation to Process B (as described in Collag's confidential statement of process).
9. On 24 February 2000 Laddie J directed the trial of the following preliminary issues:

"(a) whether the process for preparing propanil granules particularised as Process B in the Defendant's Confidential Statement of Process comprises the step of combining one or more surfactants with propanil and milling to a particle size of less than 20 microns to form a premix; and, if not

(b) whether, by reason of the absence of the said step, the use of Process B (and accordingly any dispersible granules produced by such a process) does not infringe any claim of European Patent (GB) No.0 484 147."

He also gave numerous case-management directions, including a direction for evidence on preliminary issues from one expert witness for each party (subsequently varied to two experts for each party). It is common ground that where the court has to determine the meaning of a technical expression (or term of art) such as 'surfactant' expert evidence may, exceptionally, be admissible on an issue of construction.

10. The judge recognised (in paragraph 28 of his judgment) that however he decided the first issue, there would still be room for argument on the second issue. It is worth spelling out why that is so. The first issue is primarily a question of construction, of the meaning of the term 'surfactant' in claim 1 of the patent specification. As is explained below in more detail, this is essentially a question of whether the term is limited to amphiphilic substances (which produce a chemical reaction only on contact with water) or extends to any non-amphiphilic substance which is active at the interface of two phases so as to modify the surface energy of the substrate. There was no factual dispute about what was involved in Collag's Process B (although the judge was sparing in what he said about it because of its confidential character). But there was room for argument as to whether the substances used by Collag at the dry-milling stage were surfactants (even on the wide construction of that term). The judge decided that argument in favour of Rohm and Haas (see paragraphs 125 to 133 of the judgment).
11. Conversely if (as the judge held) the term 'surfactant' had a narrow meaning limited to amphiphilic substances, there was still room for argument as to whether Process B should be regarded as an immaterial (and obviously immaterial) variant of the patented process, with the consequence that it amounted to an infringement of the patent. This required the judge to pose and answer the three Improver questions (see Improver Corporation v Remington Consumer Products [1990] FSR 181, and paragraph 41 of the judge's judgment). He answered all three of those questions in the affirmative (that is, in favour of Collag) and so reached the conclusion that he should make a declaration of non-infringement.

The patent specification

12. I have already given a brief summary of the patent specification and set out some extracts from it, but it calls for fuller treatment, since (despite Lord Diplock's well-known disparagement in Catnic Components v Hill & Smith [1982] RPC 183, 243 of "the kind of meticulous verbal analysis in which lawyers are too often tempted by their training to indulge") the specification has, inevitably, been closely analysed in counsel's written and oral submissions. It begins with an introductory explanation about water-dispersible granular formulations for pesticides (here used, no doubt, to include herbicides). It contains some references to the prior art and explains the problems which have been encountered, especially with propanil because of its low melting point (see the extracts at paragraphs 6 and 7 above).
13. The specification then gives some indications of preferred versions of the process and defines some of its terms (including 'surfactants', but unfortunately not with any great precision). At p.3 ll.10-14 the specification states,

"In addition to the active ingredient, propanil, the mixture to be formed into dispersible granules contains one or more surfactants and, optionally, flow enhancing agents, dispersants, wetting agents and defoaming agents."

It does not need particularly meticulous analysis to see that this is poorly expressed, since the specification goes on to make clear that 'surfactant' includes dispersants and wetting agents (p.3 ll.25-28):

"The term 'surfactant' is used in the broad sense to include materials which may be referred to as emulsifying agents, dispersing agents and wetting agents, and the surfactant component may comprise one or more surfactants selected from the anionic, cationic and nonionic type."

The reference to "the broad sense" has given rise to much debate.

14. The specification gives examples of surfactants of the three types (anionic, cationic and non-ionic) and continues (p.3 ll.36-57):

"Preferably the surfactant component will comprise at least one wetting agent such as those selected from alkylnaphthalene sulfonates, alkylaryl polyoxethylene ammonium sulfonates phosphate esters, sulfosuccinates and nonionics such as tridecyl alcohol ethoxylate; and /or at least one dispersing agent such as those selected from the group of naphthalene sulfonates, lignosulfonates, polyacrylates and phosphate esters.

Typically the total surfactant component will comprise from 0.1 to 25% and preferably from 1 to 15% by weight of the dry weight of the composition.

In the context of this specification a dispersing agent is a surfactant which facilitates the dispersion of the pesticide particles when the product is added to a liquid, for example water. The dispersing agents used are preferably watersoluble ones. Examples of dispersants preferred for the dispersible granule formulations of this invention include: [various proprietary brands are listed]. Most preferred dispersants are: Reax 85A and Polyfon H.

Examples of surfactants preferred as wetting agents for the dispersible granule formulations prepared according to this invention include [various proprietary brands are listed]. Most preferred wetting agents are Morwet B and Surfactant XN-45S.

All surfactants act as dispersing agents to some degree, and also to some degree as wetting agents; most surface-active agents are, however, more efficient in one capacity than the other. The worker of ordinary skill in the formulation art can select a surfactant most suitable for the purpose in view.

Small particles of a low-melting solid such as propanil often tend to stick together thereby causing flow problems in processing the material. Flow aids such as clay or silica particles may be used to minimize these problems. Flow aids preferred for the propanil dispersible granular include [various proprietary brands are listed]. Most preferred are HiSil 233 and Wessalon 50S. The flow aid content of the dispersible granule may vary from 0 to 10% and preferably from 1 to 8%."

15. The specification then sets out a table of proprietary brands, a more detailed description of the preferred version of the process and three examples. Example 1 describes the preparation of Propanil 80 DG premix with (by weight) 85.9 per cent propanil, 9.5 per cent dispersant (Reax 85A), 4.1 per cent flow aid (Hi Sil 233) and 0.5 per cent defoaming agent (Mazu DF-1300). The premix was then mixed with a wetting agent (Surfactant XN-45S in a 60 per cent aqueous solution) and a small quantity of water. The resulting mixture was extruded into granules and dried. This detailed description corresponds to the four steps in (a), (b), (c) and (d) of claim 1. The example then gives the results of various tests.
16. Example 2 states:

"A number of formulations of propanil DG were prepared with active ingredient content ranging from 60.4% to 90.2%. The procedure for preparing these granular formulations followed the general procedure of Example 1, except that Morwet B, Polyfon H and Barden Clay were substituted for Surfactant XN-45S, Reax 85A and HiSil 233, respectively. No antifoam agent was used."

The significance of this example is that the wetting agent, Morwet B, is not a liquid but a solid. That may explain why the inclusion of a wetting agent at step (b) (a process of wet mixing, whereas step (a) is a process of dry mixing and milling) is optional. The example does not explain in detail how the substitution was effected. Example 3 is a comparison with competitive products.

17. Then come the claims. I have already set out the language of claim 1. This was amended so as to insert the words "by extruding the paste" into step (c), so making the original claim 3 redundant. Some other claims were mentioned in argument but it is better to defer mention of these.

Technical matters

18. 'Surfactant' is an acronym for 'surface active agent'. It is clear from the expert evidence that it does not have a single, clearly-defined meaning. Thus the judge was referred to a relatively narrow definition in the Royal Society of Chemistry's Handbook 'Surfactants Europa':

"Every surfactant possesses the fundamental characteristic of having two essential portions, one being water repellent, usually called hydrophobic (lipophobic), the other being hydrophilic (lipophilic)."

19. He was also referred to a much wider definition in a well-known textbook, Atkins' 'Physical Chemistry':

"A surfactant or surface-active agent is a species that is active at the interface between two phases, such as the interface between hydrophilic and hydrophobic phases. A surfactant accumulates at the interface and modifies the surface tension."

20. The narrow view and the wider view were put forward by the expert witnesses for Collag and Rohm and Haas respectively. The narrow view is stated in the introductory section of the report of Professor Paul Luckham of Imperial College. I set out two passages from that introduction:

"Surfactants are a class of materials, which have in common that they consist of molecules, termed surfactant molecules, that are amphiphilic in nature. An amphiphilic molecule is one that has two parts, one which is hydrophilic, water-liking, and another part which is hydrophobic, water-hating. Surfactants are frequently represented in textbooks as tadpole like molecules, where the head group is water-liking and the tail group is the water-hating component."

The summary at the end of the introduction starts as follows:

"A surfactant is a material comprised of surfactant molecules.

A surfactant molecule is an amphiphilic molecule, where one part of the molecule is polar and water-liking, whilst the other part is apolar and water-hating.

A surfactant molecule actively orients itself at an interface such that the polar, hydrophilic group will be in contact with the more polar, hydrophilic side of the interface and the apolar, hydrophobic part of the surfactant will be in contact with the more apolar, hydrophobic side of the interface.

Surfactant molecules that orient at a surface, as described above will lower the "surface tension" or "surface energy" of that surface."

21. Professor Anthony Ryan, Professor of Physical Chemistry at the University of Sheffield, recognised amphiphilic surfactants in the context of aqueous systems but saw the term as having a wider meaning in other contexts:

"Surfactants also act at solid-air interfaces to reduce the surface free energy. For example an amphiphile whose tail has an affinity for the surface of a solid substrate will act as a surfactant at the substrate-air interface. However, again the surfactant need not be amphiphilic, rather it must have an affinity for the substrate's surface. An adsorbed coating (which can be a monolayer) of the surfactant will confer a lower surface energy than the original substrate surface. Silica and Clays are not amphiphiles but are used as surfactants in a wide range of process industries to assist in milling (a system involving a solid-air interface)."

Professor Ryan put the meaning still wider in cross-examination (Day 2, pages 150-1) but it would be wrong to place much weight on this rather confused passage of evidence.

22. The patent specification nowhere uses the term 'amphiphilic' (or any equivalent expression) but the unchallenged evidence of Professor Luckham was that the particular dispersing agents and wetting agents referred to in the specification were amphiphilic in character. Against that, however, the specification stated that the term 'surfactant' was used in a broad sense. If limiting it to amphiphilics was the broad sense, what was the narrow sense? That was the rhetorical question posed by Mr Henry Carr QC (appearing with Dr Piers Acland for Rohm and Haas).

The right approach

23. Under this heading I propose to deal with four topics which arise either on the grounds of appeal or on Collag's respondent's notice: (a) principles of construction of the specification; (b) burden of proof; (c) expert witnesses; and (d) admissibility, as an aid to construction, of technical information which (as noted on the first page of the specification) was submitted after the application was filed and is not included in the specification.

(a) Construction

24. There was little between the parties as to how the court should approach the task of construing the specification. Neither side criticised the way in which the judge had instructed himself on this. It is unnecessary to embark on lengthy citation of numerous authorities. A clear recent statement, referred to by the judge, is that of Aldous LJ in Hoechst Celanese Corporation v BP Chemicals [1999] FSR 319, 323-4. After referring to s.130(7) of the Patents Act 1977 and to the Protocol on the Interpretation of Article 69 of the European Patent Convention, Aldous LJ said,

"It follows that the court must look at the claims, with the aid of the specification, to ascertain whether there is infringement. To do that attention must be paid to the Protocol which states that a strict literal construction should be avoided as should the use of the claims as a guideline. The correct approach is to arrive at the middle ground which provides fair protection and reasonable certainty.

It is now settled in this court (see Kastner v Rizla Ltd [1995] RPC 585) that purposive construction as propounded by Lord Diplock in Catnic Components v Hill & Smith Ltd [1982] RPC 183 and as explained by Hoffmann J in Improver Corp. v Remington Consumer Products [1990] FSR 181 is the correct means of navigating between Scylla, the rock of literal construction; and Charybdis, the whirlpool of guided freedom as required by the Protocol. It enables the court to arrive at a result which gives fair protection to the patentee with a reasonable degree of certainty for third parties."

25. A purposive construction is however possible only if the purpose of the patentee can be ascertained in an objective way, by studying the specification (and any other admissible material). The specification of the patent in suit is open to criticism not only for a lack of precision in the use of words. It also fails to make clear the main thrust of its inventive purpose. In particular, it identifies the problem of propanil becoming sticky at a relatively low temperature (and so difficult to mill into small particles) but does not appear to provide any inventive solution. It simply contains a brief reference to minimizing the problem by the use of flow aids (which are and were in 1990 familiar products). Nor does the specification contain any clear teaching as to why 'one or more surfactants' were to be added at step (a) (dry mixing and milling) when it is elementary that amphiphilic surfactants have an effect only in a watery environment.

(b) Burden of proof

26. The burden of proof can hardly matter on a pure issue of construction. But on the factual part of the second issue the judge stated (in paragraph 133 of his judgment) that the evidence was unsatisfactory and that he might well have reached a different conclusion on that issue if (contrary to his ruling) the burden of proof had been on Rohm and Haas.
27. The judge based his ruling (paragraphs 30 to 37 of his judgment) on what Scarman LJ said about declarations of non-infringement in Mallory Metallurgical Products v Black Sivalls & Brison Inc [1977] RPC 321, 345, and on the judge's own analysis of the pleadings and the course of the litigation, including open correspondence between the parties' solicitors. The judge concluded (paragraph 34) that it was

"solely the defendant's counterclaim for a declaration of non-infringement from which the instant preliminary issues are derived."

Therefore the burden of proof was on Collag.

28. Mr James St Ville (who was led by Mr Christopher Floyd QC and argued this part of the respondent's notice on behalf of Collag) criticised the judge's interpretation of the correspondence. I can see no force in those criticisms. I think the judge was right in his conclusion and I could not improve on the reasons which he gave for it.

(c) Expert witnesses

29. Originally there were to be only two expert witnesses, Professor Michael Hounslow (Professor of Chemical Engineering at the University of Sheffield) and Mr Alan Knowles. Permission for two more was given when it became apparent that Professor Hounslow, as a chemical engineer, was not really in a position to give expert evidence about surfactants. So Professor Ryan and Professor Luckham came into the case.
30. Mr Knowles is not an academic, although he is a Fellow of the Royal Society of Chemistry. He has for 30 years worked in the field of agrochemicals, first as a senior formulation manager in ICI's Agrochemicals Division and then (since 1992) as a freelance consultant offering advice and training on environmentally friendly pesticide formulations. From 1993 to 1996 he was the chief technical adviser to the United Nations pesticide formulation project in China, and much of the work of that project was concerned with the development of water-dispersible granules.
31. Mr Knowles said in his report that he believed that he was the sort of person to whom the specification of the patent in suit was addressed. That was of course a reference to the well-known principle that a patent specification is (as Lord Diplock said in Catnic [1982] RPC 183, 242),

"a unilateral statement by the patentee, in words of his own choosing, addressed to those likely to have a practical interest in the subject matter of his invention (ie 'skilled in the art') …"

32. The judge accepted that Mr Knowles could aptly be taken as an addressee of the patent. After referring to what Aldous LJ said in Richardson-Vicks Inc's Patent [1997] RPC 888, 895 and after reviewing the qualifications and experience of the four experts the judge said (paragraph 47)

"Of the four witnesses from whom I heard, it appears to me clear that Mr Knowles is closest, possibly by quite a long way, to the notional addressee of the patent in suit."

33. It has not been suggested that the judge was not entitled to form that view. It has been submitted by Mr Carr that the judge's preference for Mr Knowles' approach, based as it was on long practical experience in agrochemical formulation, led him to place too much reliance on Mr Knowles' opinions, and even to abdicate his function of deciding the case himself.
34. I can say at once that I would firmly reject the last criticism. The judge gave an outstandingly clear and careful judgment and the conclusions which he reached (even if he reached some of them with unconcealed difficulty) were his own conclusions. I do think it is possible that some of Mr Knowles' evidence, like that of some of the other experts, went close to the limits of the proper function of expert evidence on an issue of construction.
35. It is however very difficult to define those limits and to keep witnesses within them, especially in the heat of cross-examination. It is the judge's function to decide what the specification means, and the expert's function to assist the judge with evidence as to the meaning of words or expressions which are scientific terms of art. But the specification has to be construed purposively and in its context, and if the term of art has a variety of meanings it is almost inevitable that the expert will get drawn into debate about purpose and context. The transcripts of evidence show instances of this on both sides.
36. For instance Mr Floyd asked Professor Ryan (Day 2, page 142):

"If this patent is really about getting the thing he calls a surfactant to solve the sticky propanil problem, why does he not list Barden Clay and silica as surfactants?"

Similarly Mr Carr had this exchange with Professor Luckham (Day 2, page 218):

"Q …you say the requirement for their presence, this is surfactants, at this stage is most likely to be due to a desire to ensure that they are intimately mixed with the propanil prior to the addition of the water. You say "most likely". This is your best guess, is it?

A That is my opinion.

Q Your opinion. There is nothing in the patent which discusses intimate mixing at all, is there, any passage you wish to point to in the patent regarding intimate mixing?

A No.

Q Nothing?

A No.

Q It is just what you read into it.

A Yes, because the surfactants are not doing anything in the milling stage, so there must be a good reason for them being there. The good reason for them being there in the milling stage is so that they are there ready in step (b) to do the job which they are put in there for."

37. These and many similar passages show how the experts were, almost inevitably, drawn into the heart of the debate on the issue of construction. But in preparing his lengthy reserved judgment the judge had the opportunity to stand back, to review the evidence and the arguments, and to form his own view. I am not persuaded that he did not take advantage of that opportunity.

(d) Admissibility of filed material

38. The filed material referred to in a note on the specification, but not forming part of the specification, includes a letter dated 11 December 1995 from Rohm and Haas' European Operations Patent Department to the European Patent Office. It was in evidence before the judge and is referred to briefly in his judgment (paragraph 114), but only in relation to the second issue. The letter was sent in response to observations made by the European Patent Office, and in paragraph 4 it contains material which (if admissible) is relevant to the construction issue. After describing experiments carried out by Rohm and Haas it stated (paragraph 4(viii))

"Applicants believe that the improved dispersibility and suspensibility, exhibited by the granules produced according to the procedure of the present invention on storage, is attributable to a better, more intimate distribution of the surfactant (ie dispersant) on the surface of the primary particles (ie the propanil granules) during milling of the surfactant/propanil to form the premix. In contrast, the primary particles of propanil milled together with an inert carrier (ie [other] formulations), tend to agglomerate on storage, possibly because the surfaces where the primary particles touch each other do not become covered with dispersant during the post blending procedure used in preparing [the other] formulations and consequently, as the granules age, the primary particles start to fuse together thereby preventing rapid dispersion."

This passage echoes Professor Luckham's reference to 'intimate mixing' which Mr Carr challenged as a 'best guess'.

39. Mr Floyd did not rely on this point before the judge. He candidly told this court that he did not do so because he did not wish to win on this point, but that he told the judge that he wished to reserve it, and it is raised in the amended respondent's' notice. I am sorry not to have the benefit of the judge's views on it, but as Mr Carr did not object to the point being argued I will give my view.
40. There seems to be no clear English authority on the point, even at first instance. In Bristol-Myers Squibb Co v Baker Norton Inc [1999] RPC 253, 274-5 Jacob J has given a useful summary of the problems associated with taking account of what he called prosecution history – that is, the vicissitudes of an application file's progress through the official system – as an aid to construction of the final specification. But Jacob J said that he did not have to decide anything about the point.
41. This court was shown a decision of the Supreme Court of the Netherlands, Ciba-Geigy v Oté Optics (13 January 1995) which contains a helpful statement of principle. In explaining that the Court of Appeal had gone too far in excluding all reference to the file, the Supreme Court said:

"Article 69, paragraph 1 of the EPC as interpreted in accordance with the protocol relating thereto does indeed purport (among other things) to ensure reasonable certainty for third parties, but it does not follow that the information from the granting file that is available to third parties may never be used in support of the interpretation given by the patentee to his patent. The requirement of reasonable certainty for third parties does, however, call for restraint in using arguments derived from the granting file in favour of the patentee. Consequently, a court will only be justified in using clarifying information from the public part of the granting file, when it holds that even after the average person skilled in the art has considered the description and the drawings, it is still open to question how the contents of the claims must be interpreted. In this connection one must also take into consideration that the risk of any ambiguities due to careless wording of the patent specification must in principle lie with the patentee."

42. Apart from the last sentence (which raises a different point, and on which Mr Floyd did not rely) I would treat this as persuasive guidance. The letter to the European Patent Office did not have the same status as published prior art identified in a specification, which is readily admissible. But it did contain objective information about and commentary on experiments which were conducted in response to official observations, and it could be of assistance in resolving some puzzling features of the specification. Although the prosecution process may sometimes superficially resemble a process of negotiation between the applicant and its advisers and the officials who scrutinise the file, it is not the sort of commercial negotiation which is still rigidly excluded in the construction of a written contract (see Investors Compensation Scheme v West Bromwich Building Society [1998] 1 WLR 896, 913). Had it been necessary for the judge to take account of the letter in order to resolve the issue of construction, I consider that he would have been entitled to do so.

The first issue

43. The judge discussed the issue of construction at paragraphs 51 to 105 of his judgment. He set out his conclusions, and a summary of his reasoning (much of which had already emerged in the course of the discussion) at paragraphs 106 to 110.
44. The judge's discussion of the issue is clear, balanced and comprehensive (Mr Carr, in opening the appeal, readily accepted that the judgment was fair and thorough; his only criticism of it was that it reached the wrong conclusions). The judge began, rightly, by reminding himself that the addition of one or more surfactants at the first step (paragraph (a) of claim 1) was (paragraph 53)

"not an optional aspect of the teaching; it is a mandatory aspect."

He went on to ask himself

"the obvious question whether it is possible to discern from the patent in suit, and, to the extent that it is admissible on this point, from the expert evidence, the purpose of adding one or more surfactants at the first step."

45. He proceeded, therefore, to discuss the competing theories put forward by Rohm and Haas and Collag respectively: flow aid to solve the problem of the propanil becoming sticky, or intimate mixing to achieve a better result at later stages. He noted and discussed the three main difficulties in the way of the flow aid theory: there was no teaching about it in the specification, no suggestion of an inventive means of solving the problem, and the linguistic difficulty, on the wording of the specification, of referring to "one or more surfactants" rather than to a flow aid. He continued with a detailed analysis of how some of the evidence and argument had got at cross-purposes about the expressions 'dispersants' and 'dispersing agents'.
46. The rival theory about 'intimate mixing' was open to some of the same objections as to lack of teaching (in this instance, as to the solution of any perceived problem) in the specification. (That lack was partly made good by the Rohm and Haas letter dated 11 December 1995). The judge quoted from Mr Knowles' report (while noting that the only published work seemed to be a book edited by Mr Knowles himself and published in 1998):

"From my experience, the purpose of what is taught in the patent, that is mixing and grinding the surfactants with the active ingredient in a dry grinding process is to ensure that a completely homogenous dry mixture has been obtained. Thus when water is added prior to extrusion the surfactants are in intimate contact with active ingredient particles and can prevent agglomeration of particles during extrusion and drying. Both the wetting agent and the dispersing agent are often added to the powder premix before dry milling for this reason."

(Those remarks apply to a wetting agent which is a solid. If the wetting agent is in liquid form it may be added at step (b), as noted in paragraph 16 above.)

47. The judge stated his conclusion on these purposive arguments as follows (paragraph 109)

"If one confines oneself, as a lawyer, to what is said in the patent in suit, then there is considerable force in Mr Carr's point that it is difficult to discern the purpose of adding one or more surfactants at the first step. For the reasons already given, I am quite unpersuaded by the claimant's contention that it is for the purpose of assisting at the milling stage. There is no teaching, or even any claim to any discovery, to that effect anywhere in the patent. On the other hand, in the absence of expert evidence, the idea that the purpose is to assist intimate mixing of propanil and surfactant to optimise the second step is the purest speculation, albeit not unreasonable speculation. However, Mr Knowles, who seems to me to be close to the epitome of the notional addressee, appears to have had no difficulty in understanding the purpose as being intimate mixing, and while Mr Carr challenged him in cross-examination, and maintained his challenge in his closing speech, I accept Mr Knowles's evidence on the point."

The judge's reference to "purest speculation, albeit not unreasonable speculation" might have been put a little differently (and more favourably to Collag) had he heard submissions about the letter dated 11 December 1995.

48. As well as considering these purposive arguments the judge carefully considered various arguments on the language of the specification. He noted the reference to the term 'surfactant' being used "in the broad sense" but observed that that passage must be read as a whole; and that other linguistic indicators pointed to the term being used so as to refer to amphiphilics, and only to amphiphilics. He also considered the textbooks which had been cited, and the guidance to be obtained from other claims in the specification (claims 4, 5 and 10 before renumbering on amendment) but he concluded that he could get little assistance from those sources. He concluded that in order to be a surfactant within the meaning of the specification a substance must be amphiphilic in nature.
49. As the argument progressed I have become more and more persuaded that the judge was right in this conclusion, and that the reasons which he gave steer the right course between the Scylla and Charybdis of which Aldous LJ spoke in Hoechst. On any view the specification is not an exemplary piece of draftsmanship. The teaching is not clear, and the use of terminology is not as precise and consistent as it should be. But the judge's conclusion is in my view much the most satisfactory way of resolving the various difficulties.

The second issue

50. The second issue raises the Improver questions (or as it may be more correct to call them, the protocol questions). The judge set them out in paragraph 41 of his judgment, but it is convenient to repeat them. Hoffmann J said ([1990] FSR 181, 189) that where the court has to determine

"whether a feature embodied in an alleged infringement which fell outside the primary, literal or acontextual meaning of a descriptive word or phrase in the claim ("a variant") was nevertheless within its language as properly interpreted, the court should ask itself the following three questions:

(1) Does the variant have a material effect upon the way the invention works? If yes, the variant is outside the claim. If no –

(2) Would this (ie that the variant had no material effect) have been obvious at the date of publication of the patent to a reader skilled in the art. If no, the variant is outside the claim. If yes –

(3) Would the reader skilled in the art nevertheless have understood from the language of the claim that the patentee intended that strict compliance with the primary meaning was an essential requirement of the invention. If yes, the variant is outside the claim."

A little later in the same judgment Hoffmann J said that Fox LJ had put the point with great precision (in Anchor Building Products v Redland Roof Tiles 23 November 1988) when he posed the question whether the absence of a feature mentioned in the claim was "an immaterial variant which a person skilled in the trade would have regarded as being within the ambit of the language" (Hoffmann J's emphasis).

51. The judge considered the three questions in turn, first on the basis that (as the judge had just held) Collag was right on the first issue. On that basis, the variant was the use of a non-amphiphile (china clay or Paselli, a form of starch) as a flow aid at the first step, followed by the use of amphiphilic substances at the second step.
52. On the first of the three questions the judge did refer to, and relied on, the letter dated 11 December 1995. When the experiment described in the letter was put to Professor Hounslow in cross-examination he accepted that one of the other formulations mentioned in the experiment (see paragraph 38 above) was "very much akin" to Collag's Process B.
53. Mr Carr pointed out that although this experiment was conducted by Rohm and Haas, Professor Hounslow was not personally concerned with it, and no formal notice of experiments had been given in respect of it. But it does not seem to have been suggested that the material was put to Professor Hounslow in an unfair manner. I consider that the judge was entitled to take the experiment into account on the question of the materiality of the variant. He robustly rejected (paragraph 116) the argument that the variant must be immaterial because Collag's Process A (which did not contain the variant) and Process B (which did) both produced a saleable product:

"In my judgment, however, that will not do. The point would obviously be powerful if it could be shown that the product of the two processes was indistinguishable, but that was not the thrust of the claimant's case. … there is no evidence to show that the products of the two processes are, in fact, identical."

54. The judge concluded that he must answer the first question in the affirmative, that is in favour of Collag. That question is essentially a question of fact (see Improver at p.190) and I am not persuaded that the judge was wrong in his conclusion. On the contrary, I think he was entitled to come to that conclusion on the evidence, and that it was correct.
55. The judge's affirmative answer to the first question made the other questions academic, since Rohm and Haas had to obtain a clear run of three negative answers in order to succeed. But the judge went on to consider the second and third questions also, and answered both of them in the affirmative. I agree with his reasoning on those two questions and with his answers to them.
56. Finally I should record that Mr St Ville, following his leader, addressed the court on two matters raised in the respondent's notice. One, the burden of proof, I have already referred to. The other was the issue of infringement if Rohm and Haas were to succeed on the first issue. That issue does not have to be decided. It is not merely academic but requires something of a mental effort to rearrange one's thoughts in order to address it. In the circumstances I would only say that Mr St Ville's submissions were carefully constructed and detailed, but I was not convinced by them.
57. I would dismiss this appeal.

LORD JUSTICE KEENE

58. I agree.

LORD JUSTICE PETER GIBSON

59. I also agree.
Order: Appeal dismissed. It is appropriate that there should be a deduction form the costs which the appellants will be required to pay to the successful respondents on account of the Respondent's Notice. The appropriate figure is that the appellants should pay 85% of the respondent's costs. There should be no payment by the respondents to then appellants. There should be a detailed assessment of those costs, but it is appropriate that there should be an interim payment to Agform. We will order that the interim payment be £30.000. In respect of Collag's costs, Collag will also get 85% of its costs paid for by the appellant's, but we accept the suggestion of the appellants that such order should be enforced until after the hearing before Jacob J, with liberty to the appellants to apply to the judge for a set-off. (Order does not form part of the approved judgment)