Smith & Nephew Plc v Convatec Technologies Inc & Anor

The question in this action is whether Smith & Nephew’s Durafiber Ag product infringes Convatec’s European Patent (UK) No. 1,343,510 entitled “Light Stabilized Antimicrobial Materials”. The patent is concerned with the silverisation of gel forming fibres used in wound dressings.

This is the second time I have had to consider this patent at a trial between these parties and the third time I have had to consider a patent dispute between Convatec and Smith & Nephew about Durafiber.

Convatec have a wound dressing product called Aquacel. It is based on gel forming fibres made from carboxymethyl cellulose (CMC). It has been a major success both commercially and medically. Smith & Nephew wish to sell wound dressings which compete with Aquacel. Instead of using CMC, the gelling fibres in Durafiber are based on cellulose ethyl sulphonate (CES). Durafiber Ag is a version of the Durafiber product which contains silver as an antimicrobial agent. The silverised version of Aquacel is called Aquacel Ag.

At the first trial Convatec contended that Smith & Nephew’s Durafiber product infringed three Convatec patents (EP (UK) 0 927 013, EP (UK) 1 085 912 and EP (UK) 0 616 650) and was the result of a misuse of confidential information on the part of Smith & Nephew’s supplier of the CES fibres. By the trial the confidential information claim and the 912 and 650 patents had fallen away one way or the other. The only issue remaining was about the 013 patent. I held that Durafiber did not infringe a valid claim of that patent ([2011] EWHC 2039 (Pat)). That conclusion was upheld in the Court of Appeal ([2012] EWCA Civ 520). The result of that first action is that Smith & Nephew are free to make and sell Durafiber in the UK.

The first trial did not address silverisation. In parallel with the first action, Smith & Nephew started an action to revoke Convatec’s silverisation patent No. EP (UK) 1,343,510. That led to the second trial. At the second trial I held that Convatec’s 510 patent was valid ([2012] EWHC 1602 (Pat)). That conclusion was upheld by the Court of Appeal ([2012] EWCA Civ 1638). The outcome of the second trial was a blow to Smith & Nephew. They had developed a process for making silverised Durafiber and had obtained CE mark regulatory approval to sell that product in the EU but the manufacturing process infringed (or at least risked infringement of) the 510 patent. So after my judgment was handed down in June 2012 Smith & Nephew developed a new process which they thought would not infringe and in January 2013 they began this action for a declaration of non-infringement of the 510 patent. Convatec counterclaimed alleging infringement and the action has now come on to trial.

The claim I need to consider is the same as the one considered at the previous trial. It is:

The first issue I have to decide is the meaning of the phrase “the agent being present in a concentration between 1% and 25% of the total volume of treatment”. There are two points arising: the scope of lower boundary of the claim and the meaning of “total volume of treatment”.

On the lower boundary there are three possibilities. Convatec’s primary case is that the claim would be understood as requiring rounding to the nearest whole number and so anything more than 0.5% is caught. Smith & Nephew’s primary case is that the claim “means what it says” and anything less, for example 0.9999%, is outside the claim. Smith & Nephew’s fall back is that 1% is expressed to one significant figure and so anything greater than or equal to 0.95% is covered. This argument about the lower boundary of the claim did not arise at the previous trial and the judgment does not rule on the matter. There was no (or no clear) suggestion at the first trial that the lower limit was anything other than 0.95%.

Total volume of treatment is dealt with in paragraph 48 of my previous judgment. Smith & Nephew submitted that paragraph 48 meant that the construction they contend for on this occasion had been accepted. There is some force in that submission but I prefer to rule on the meaning of the patent rather than having a meta-debate about the meaning of my previous judgment, where the point was not clearly decided.

Once I have decided the issues of construction there are two infringement questions. The first is about the process Smith & Nephew now wish to use commercially. In this method the relevant agent (sodium chloride) has a concentration of 0.77%. Clearly this only infringes if the boundary of the claim is 0.5%.

The second issue only matters if the boundary of the claim is 0.95%. The issue is about Smith & Nephew’s development work. In that development work Smith & Nephew carried out certain development experiments which were used to make product to be used to generate data to be used to obtain regulatory authorisation to sell the product. Smith & Nephew did not contend these experiments were the subject of the experimental use exception to patent infringement in s60(5)(b) of the 1977 Act. The experiments have been given numbers. The relevant ones are 40, 43, 44 and 48. Convatec contend that the experiments used a salt concentration at a level more than 0.95% and therefore within the claim. Smith & Nephew contend that in fact, when considered properly and taking into account a phenomenon called “carryover”, all the runs of these experiments were at a lower salt concentration, outside the claim. I need to resolve that factual dispute.

Finally I need to resolve the question of springboard relief. This only arises if the claim boundary is 0.95% and if the development experiments infringed the patent. If so then Convatec contend that since Smith & Nephew obtained CE mark approval for Durafiber Ag using data generated as a result of those experiments, it follows that they obtained that regulatory approval by infringing the patent. Smith & Nephew are only able to launch Durafiber Ag today in the EU as a result of having that approval. What ought to have happened, say Convatec, is that Smith & Nephew ought only to have been in a position to apply for CE mark approval in January 2013 and that the approval process would have taken until September 2014 to complete. So even though selling Durafiber Ag today would not infringe the patent (on the relevant premise) nonetheless Smith & Nephew should be restrained from selling Durafiber Ag until September 2014 in order to prevent them from deriving an unfair profit from their infringement.

Smith & Nephew deny all this. They contend that no such injunction should be granted at all and they challenge the dates proposed by Convatec. Convatec submit that even if the injunction should not last until September 2014, depending on my findings a springboard injunction of even as little as two months duration should be granted. Given the very high value of the market for these products, arguing about an injunction for only two months is worthwhile from each side’s point of view. The market for Aquacel (which I recognise is not just Aquacel Ag) in the UK alone was said in the first action to be worth £17 million per year.

As before I heard technical expert evidence from Prof Kennedy (called by Convatec) and Prof Burrell (called by Smith & Nephew). Their backgrounds are explained in my previous judgment.

Smith & Nephew said that Prof Kennedy’s evidence lacked objectivity, referring to his “resolute adherence” to the nearest whole number construction of the patent, his “dogged resistance” on carryover, and an episode in which Smith & Nephew submit he “flushed red in the face (an indication that he suddenly realised he had been caught out)” when confronted with a change in his evidence about the total volume of treatment.

Convatec’s cross-examination of Prof Burrell put to him that he had changed his evidence from that given in the previous action.

I reject both sides’ criticisms of the other party’s technical expert. Smith & Nephew submitted in defence of Prof Burrell that it is not a valid criticism of an expert that he (or she) develops their thinking on an issue if the circumstances bring that issue into sharper focus and greater scrutiny (as happened in this action). Smith & Nephew are right. The observation applies to both experts. Read carefully it is possible to see that in the previous case each made statements about points which have great significance in this action but which were not central last time and which, when looking back, can be interpreted as inconsistent. In my judgment both gentlemen were seeking to give objective and dispassionate evidence on this occasion.

As for Prof Kennedy’s complexion, I did not detect anything about his demeanour at any stage which I could take into account as being relevant. As for his refusal to accept certain propositions put by counsel, in my judgment they were simply a reflection of his genuinely held views.

Convatec also submitted that the nature of Prof Burrell’s expertise as compared to that of Prof Kennedy meant that on questions relating to polymers I should unhesitatingly prefer the evidence of Prof Kennedy. I would not go that far, but I do acknowledge that Prof Kennedy has a much more extensive experience of polymers, particularly cellulose and the development of cellulose derivatives, than Prof Burrell. It is a matter I will take into account.

Smith & Nephew also called David Stephenson. He is a project manager at Smith & Nephew. Until two years ago he worked in new product and process development. He attested to Smith & Nephew’s Process Description and gave detailed evidence about the manufacturing process for Durafiber Ag. He also explained the history of the Durafiber Ag project and dealt with the experiments undertaken during the development. The preparation of his evidence clearly involved a huge amount of work. Convatec rightly did not criticise Mr Stephenson. He was a good and careful witness.

Convatec called Neil Armstrong as an expert witness. He is the CEO of MeddiQuest Limited, a regulatory affairs consultancy which specialises in bringing novel healthcare products to the market. He has 12 years experience in obtaining regulatory approval for medical devices. He was asked to give an opinion about how long Smith & Nephew would take to obtain regulatory approval for Durafiber Ag. He was a good witness.

Smith & Nephew also called Stewart McKinlay. He has been the Vice President of New Product Development for Smith & Nephew Healthcare Ltd for over six years. His evidence was related to the springboard issue. He explained what work Smith & Nephew would have to do to obtain regulatory approval and estimated how long the exercise would take. He was a good witness.

I will not rehearse the background to this technology. I refer to my previous judgment. As I said in paragraph 8 of that judgment, the patent relates to a method of preparing a light stabilized antimicrobial material for use in wound dressings and medical devices. Essentially the method is one whereby a wound dressing material can be made which comprises silver as an antimicrobial agent but which is stable in the presence of light.

Paragraph 29 of my previous judgment identified the relevant person skilled in the art, which would be a team comprising a biomedical engineer, a chemist and a materials scientist. Nothing has changed in that respect.

The common general knowledge will include the matters I addressed in my previous judgment. Two further topics are relevant to this case, number conventions and fibres.

The common general knowledge includes the conventions associated with rounding numbers, decimal places and significant figures.

A point of detail arose about the ways of rounding a number ending in 5. The elementary rule is that the number 5 is always rounded up. A more complicated rule is that 5 is rounded to the nearest even digit. This is sometimes called the unbiased rounding rule and is particularly applicable in large data sets since on average, if they are random, half the numbers will round up and half will round down. A yet further refinement takes account of all the information available. So 1.42501 to three significant figures would be 1.43 because the “01” after the 5 shows that the number is more than halfway to the next digit. Whereas if all one had was 1.425, then the convention would apply and the number be rounded to the even number.

The unbiased rounding rule is discussed in textbooks addressed to quantitative chemical analysis however it is not universally known in this area of science since Prof Kennedy had not heard of it. On that basis I find that as a matter of common general knowledge, while the skilled person would obviously round 1.42501 to 1.43 if required to round to three significant figures, if they had to consider what to do about a number with no further information after the 5 (say 1.425 to be rounded to three significant figures) they would round up.

The precision with which a number is associated can be expressed by indicating the number of significant figures. The number of significant figures used is not always stated expressly. When a number is understood to be expressed using significant figures but that fact is not stated in terms, the rules are:

Non-zero digits are always significant;

Zeros between non-zero digits are always significant;

Leading zeros are never significant;

If a decimal point appears in the number then trailing zeros are significant (before or after the decimal point); and

In the absence of a decimal point, trailing zeros are not generally significant unless stated otherwise either expressly or with a bar over the zero.

Save for the qualification “generally” in point (v), in setting this out I have accepted Convatec’s submissions. The reason for the qualification is that truly trailing zeros in those circumstances are often ambiguous. They are generally not regarded as significant but they may be, for example 10 is often taken to be stated to two significant figures.

The common general knowledge also included the knowledge that fibres such as the ones in issue in this case contain moisture. Air dried CMC fibres contain about 12% free water. This free water is available to act as a solvent and can be removed. Such fibres can also hold water in such a way that it is not available to act as a solvent. This is called bound water. It is held by strong chemical interactions and as a matter of common general knowledge it cannot be easily removed.

There was a dispute at trial about whether, when fibres absorb water from a liquid, they can take up additional bound water. Prof Kennedy’s view was that they could. It was put to him that the manufacturing process for the fibres concerned, which involved mercerisation with a strong alkali, would mean that all the sites for bound water would be occupied before the fibres were used in the process of this invention. Prof Kennedy was not prepared to accept that this was necessarily so. Prof Burrell’s view was that all the sites would be occupied. Since this is a point more closely related to Prof Kennedy’s experience than Prof Burrell’s experience, I find that the skilled person would not be sure that all the sites for bound water would be occupied but the tenor of the evidence leads me to conclude that the skilled person would not expect any take up of bound water to be large if it happened at all.

The fibres in this case are made in batches and the skilled person would know that they would be likely to exhibit variations in their characteristics from one batch to another.

The way the claimed invention works is that the material comprising the gel forming fibres is first treated with a solution comprising an organic solvent and a source of silver. An example of a suitable solution would be an ethanol/water solution of silver nitrate. This is the silver loading step. Next a facilitating agent is used to render the material photostable. An example of a suitable agent is sodium chloride.

The two issues I have to decide relate to the meaning of the phrase “the agent being present in a concentration between 1% and 25% of the total volume of treatment”.

By using the words “during or after” the claim covers two ways of carrying out the process which it is convenient to refer to as the “during” process and the “after” process.

The “during” process can be considered this way. The material could be placed in a bath containing 1 litre of an ethanol/water solvent containing silver nitrate. After a period the agent is added to the bath by adding 1 litre of a sodium chloride solution. The total volume of treatment will be the sum of the volume of solvent plus the added volume of agent. So it would be 2 litres in this example. Thus if the concentration of sodium chloride in the added solution was 10%, the concentration based on the total volume of treatment would be 5%.

In the “after” process things are different. As before the material would be placed in a bath containing 1 litre of an ethanol/water solvent containing silver nitrate. Now, after a period the ethanol/water solution is removed. This may be by taking the material out of the bath or leaving the material in the bath but draining out the solution. Now the agent is added by adding 1 litre of a sodium chloride solution. Convatec say in this example the total volume of treatment will be 1 litre – the added volume of agent. Smith & Nephew disagree. They point out that the material will be wet. Indeed since gel forming fibres are able to absorb large amounts of water, there is likely to be a substantial amount of fluid carried over from the first step to the second step. They contend that the skilled person would realise this and would understand that it needed to be taken it into account in order to decide what the total volume of treatment is.

Convatec’s two best arguments were the following. First, the skilled person would also know as a matter of common general knowledge that air dried CMC fibres would also contain about 12% free water. Thus the logic which requires carryover to be taken into account requires this to be taken into account as well. And yet Smith & Nephew’s construction of carryover in relation to the “during” process does not accommodate this material. So Smith & Nephew’s construction is inconsistent as between the during process in which this water is not taken into account and the after process in which it is taken into account as water carried over from the first step to the second step. Second the skilled person would also see that actually working out what the carryover was in a given case was very difficult. The specification says nothing about the calculation or how to attempt it. If the patentee had intended the skilled person to calculate carryover he or she would have said so. They did not.

Despite Convatec’s arguments, I prefer Smith & Nephew’s construction. Any skilled person who saw a method in which substantial amounts of fluid were being carried over from the first step to the second step would think that this fluid should be included in the “total volume of treatment”. The total volume of treatment at the point at which the agent is added is obviously going to include the volume added which has the agent in it and any fluid present. It is true that there may be an inconsistency when one then considers the likely free water inside dry CMC but that point is not strong enough to lead a skilled person to ignore carryover altogether. It is also true that in this case determining what the carryover actually is has proved to be difficult but I do not accept this as a reason why the skilled person would ignore it.

The key decisions on construction in general are Kirin Amgen[2005] RPC 5 and the Court of Appeal’s summary of the approach in Virgin v Premium[2009] EWCA Civ 106. As regards numerical ranges in particular, I was referred to a number of cases in the English courts and the Technical Board of Appeal in the EPO. Counsel pointed out that in Kirin Amgen Lord Hoffmann referred to numerical ranges at paragraph 65, observing that the notion of strict compliance with the conventional meanings of words or phrases sits most comfortably with the use of figures, measurements, angles and the like, when the question is whether they allow for some degree of tolerance or approximation.

Convatec’s counsel submitted that no English court had ever adopted an interpretation of a claim containing a numerical limit in accordance with Smith & Nephew’s primary construction. In most cases an approach based on significant figures has been taken. The cases were:

Lubrizol v Esso (13 November 1996 in a passage not reported in the report at [1997] RPC 195) in which Jacob J as he then was decided that a lower limit of 1.3 should be read as “1.3 to two significant figures” and so would include 1.27.

Goldschmidt v EOC Belgium[2000] EWHC Pat 175 in which David Young QC sitting as a Deputy Judge of the High Court held that a pH of 4.6 was within the claimed pH range of 5 to 8 since “5” was stated to one significant figure and was not, for example, stated as “5.0”.

Halliburton v Smith[2006] RPC 8 in which Pumfrey J construed “between 31% and 35%” to include from 30.5% to 35.4% since the figures in the claimed range were stated to two significant figures.

FNM Corp. v Drammock [2009] EWHC 1294 in which Arnold J held that figures up to 45.4% were embraced by a limit expressed as “45%” since that was expressed to two significant figures and was not “45.0%”.

I was also referred to Auchinloss v Agricultural & Veterinary Supplies[1997] RPC 649 (Peter Prescott QC sitting as a Deputy judge of the High Court). There the learned judge held that a product with 10 parts sulfamic acid did not infringe a claim limited to “3 to 8 parts sulfamic acid” on the basis that a numerical range admits of no variant in the Catnic sense or that the skilled reader would understand that strict compliance with the parameters of the range was an essential requirement of the invention. However these findings are not concerned with the issue of significant figures, rounding and numerical precision, as Mr Prescott himself recognised in the passage at p690 ln 5-17.

Finally I should mention PLG v Ardon[1993] FSR 197 as the other English case emphasised in argument. Aldous J (as he then was) held (obiter) that a thickness of 60-72% would have infringed a claim limited to a minimum thickness of 75% following an application of what were then called the Improver questions. I do not have to decide whether, following Kirin-Amgen, the court would reach a similar conclusion today. The point is that PLG v Ardon does not support Smith & Nephew’s narrow construction of the claim.

Smith & Nephew placed emphasis on a decision of the EPO Technical Board of Appeal in T74/98 (19 Oct 2000) in which the Board held that a claim limited to 5-50 mol% monomer was novel over a disclosure of 4.98 mol % monomer. The Board refused to apply the rounding up rules since they thought that would expand the scope of the claim beyond the indicated limits (paragraph 3.2, first point).

Convatec submitted that the settled jurisprudence of the Technical Boards of Appeal in the EPO, as reflected by the current (Sept 2013) edition of their Case Law textbook, did not support Smith & Nephew’s primary construction. This was based on cases T1186/05 (6^th Dec 2007) in which a lower limit of 0.89 was held to be satisfied by 0.885 because 0.885, when rounded appropriately, gave 0.89; T234/09 (1^st June 2012) in which an upper limit of 5% was held to include 5.2%; and T871/08 (8 Dec 2011) in which 2.996 had to be rounded up to 3, the number in the claim.

Convatec also submitted that the Boards of Appeal clearly regard T74/98, the case relied on by Smith & Nephew, as a case which can be distinguished from these other cases (the board in T1186/05 doing exactly that). The distinction relating to a particular aspect of T74/98 in that rounding the figure for the prior art in that case would alter the true meaning of the prior disclosure. I accept Convatec’s submissions about the jurisprudence of the EPO. The outcome in T74/98 may be said to favour Smith & Nephew but the reasoning in that decision is not persuasive.

Smith & Nephew submitted that the true principle was that the number had to be perceived to be a measured or calculated value before the skilled reader would understand that a significant figures approach was to be taken. Applied to this case, Smith & Nephew submitted that the skilled reader would see from the specification of the 510 patent that the range in the claim was one of a group of sub-ranges disclosed, none of which would be thought to have a technical basis. The skilled reader would understand that the only reason for the sub-ranges was as something drafted by a patent attorney to provide a reservoir of possible fall-back positions in case prior art was cited against the patent which required a limiting amendment to one of the more limited sub-ranges.

I do not accept Smith & Nephew’s submission about the principle to be applied. I do not see why the significant figures approach must only be applied to values in a patent claim which can be perceived to be the result of a measurement. This case is concerned with a value which acts as the limit of a range of concentrations. The actual concentrations used in the processes which have to be compared to the claim are likely to be measured or calculated. The scope of the limit of that range should not to be treated differently depending on whether it is possible to see in the specification that it derives from an actual measurement as opposed to being the product of a choice by the patentee. When, as here, the limiting value functions as a target against which a measured or calculated value is to be compared, then in principle I can see no reason why the skilled person should not use a significant figures approach.

However despite all these cases it is important to remember that the true interpretation of numerical limits is a matter of the construction of the particular patent specification in its own particular context. The fact that in many earlier cases the courts in England and the EPO Boards of Appeal have held that the question of whether a value should be held to fall within a numerical limit should be decided by rounding that value to the same number of significant figures as the range is expressed in the claim, does not mean that this must always be the conclusion to be reached.

The common general knowledge is part of the relevant context. In T708/05 the claimed range was from 1 to 5 μm. The Board of Appeal held that in the relevant technical field (multilayer films) the values concerned would be rounded to one decimal place and so 0.9625μm would be rounded to 1.0 μm. Separately the Board accepted evidence that it was not possible to determine the layer thickness with an experimental error of less than ± 5 to 10% and so 0.9625μm would be understood as “about 1μm” and within the claim on that basis.

Smith & Nephew invited me to hold that the claim’s lower limit was precisely 1% so that any concentration which was less than unity fell outside the claim. I should adopt that construction because the claim “means what it says”. I reject this argument for the following reasons.

First, to say that the claim “means what it says” does not help. The question is then – what does it say? If 1% is understood to say that it is a target to be judged to one significant figure then what it means is that a measured concentration of 0.96% falls within the claim. In that respect it is important to understand significant figures correctly. Taking an example away from the facts of this case, if one knows that in the relevant context a number is to be stated to two significant figures, then all three of 23.8, 24.0 and 24.3 are the same. They are not merely close to each other, they are each 24 in the relevant context. So if the claim is understood to be referring to 1% to one significant figure then a measured or calculated concentration of 0.96% is 1%. The fact that 0.96% is not 1.0% (because 1.0% is a number taken to two significant figures) is irrelevant. To put things another way: to say 1% and to say 1.0% is to say two different things in the context of significant figures.

Second, the Protocol on the Interpretation of Art 69 does not require such a reading. The Smith & Nephew’s interpretation gives no more certainty for third parties than the other two rival constructions. Each is as certain as the other, albeit the boundary is in a different place. Moreover to read the claim as if 1% meant 1% to an infinite level of precision, does not seem to me to give fair protection for the patentee.

Third, Smith & Nephew referred to some passages in text books about exact numbers. The skilled person would understand that some numbers are indeed exact numbers, such as the value of 2 in the equation L = 2πr for the circumference of a circle. There are other examples of exact numbers but none of them apply to this range.

Fourth, Smith & Nephew are right that the skilled reader would understand that the patentee had chosen these values and not measured them. It is obvious that they have not been directly measured but that does not mean the skilled reader would think they were meaningless. The reader would think that the patentee had selected these values in order to act as the limits of the claimed range of concentrations.

Fifth, nothing in the common general knowledge mandates Smith & Nephew’s reading of the claim. Smith & Nephew submitted the evidence was that in this art the skilled person would measure concentrations to two significant figures. Although that does not justify the absolute approach to 1%, it would, if correct, run counter to the two alternative constructions. However that was not the evidence. The evidence was that the skilled person could readily measure to a higher level of precision than two decimal places if they wished to do so. The common general knowledge of how concentrations are measured or calculated does not help Smith & Nephew.

Convatec’s primary case was that the claim’s lower limit was 0.5%. I reject this argument for the following reasons.

First, its principal foundation was Prof Kennedy’s evidence that when he looked at the ranges it appeared to him that the numbers stated were whole numbers or zero decimal places. So he thought one should round to the nearest whole number. On that basis anything from 0.5% should be rounded to 1%. However I do not believe a skilled reader would take this approach because they would see immediately that the range in claim 1 is one of the ranges described in paragraph [0028] of the specification. In that paragraph the widest range disclosed is from 0.01% to 50%. Other sub-ranges are referred to, many of which start from 0.01% or from 0.1%. Thus self evidently the patentee was not using whole number percentages.

Prof Kennedy accepted this in cross-examination but he maintained that the disclosure was using what he described as “clear” numbers but which I understood really to mean round numbers. So 0.1% is not a whole number but it is a round number. After all none of the values in paragraph [0028] are numbers like “2.234%”. I agree that the skilled person would see that the numbers are round numbers. However that does not justify rounding 0.5% up to 1%. I doubt the skilled person would think the observation that round numbers had been chosen told them anything about how rounding should be done. However if the skilled person did think that the use of round numbers such as 0.01%, 0.1%, 1%, 5% and so on told them anything about rounding then for values less than 1%, rounding to the nearest round number would take a value like 0.14% to 0.1% and a value like 0.77% to 0.8%. It would not justify rounding 0.77% to 1%.

In other words to see 1% to 25% as a range expressed to the nearest whole number or to zero decimal places is to consider that range out of its context in the specification.

At one stage in the parallel EPO proceedings it appears to have been contemplated that paragraph [0028] should be amended to delete the sub-ranges which are outside the claim in its now amended form. It is a familiar idea that consequential amendments to a patent specification are made when a claim is amended but this case illustrates the risks which can occur when that is done. Removal of the references to 0.01% and 0.1% from this part of the specification could be said to risk altering the true interpretation of the claim, expanding its scope and contravening Art 123(2) and Art 123(3) EPC.

Second, another key aspect of Convatec’s case was Prof Kennedy’s opinion that the skilled person would think the difference between a 0.5% and 1% concentration was unlikely to make any difference to the way the invention worked. This was because the skilled reader would see that the range was very broad. A concentration of 25% is more than four times a concentration of 1% and as a matter of chemistry, Prof Kennedy’s view was that a skilled person would think the mechanism of action could well be different in different sub-ranges between these limits. He thought the ends of the ranges were unlikely to be important. He pointed out that the method is described in the specification in very general terms. For example it mentions a wide range of different sources of silver, wide ranges for the quantity of silver, the times and the temperatures. Paragraph [0028] states in general terms that the quantity of agent to be used will depend on the amount of polymer containing material and the total volume of solution but does not provide specifics.

I accept Prof Kennedy’s opinion that the skilled reader would see that the range was claimed broadly and that the specification is written in general terms, but I was not convinced that the skilled reader would think the ends of the ranges were unimportant. For example 1% salt is twice the concentration of 0.5% salt. The generality of the specification would mean that the skilled reader would not be surprised if the lower limit of the claim range had been expressly stated to be 0.5% but it does not help in deciding how to interpret the limits actually defined.

Smith & Nephew’s fall back position was the construction which put 0.95% as the lower limit of the claim. Nevertheless they both advanced arguments why it was wrong in support of their preferred construction. Each advanced arguments why a significant figures approach to the lower limit of the claimed range would be rejected.

Smith & Nephew’s argument against this construction of the claim was that any use of rounding for the lower limit of the claim meant that the skilled person would be required to round numbers twice, which was not scientifically appropriate. It is true that when performing calculations using measured values, the skilled person knows that one should only carry out rounding once, at the end of the calculation. Intermediate values should not be the subject of rounding. That is part of the common general knowledge.

Smith & Nephew suggest that this means that the skilled person would first measure the concentration and state the result to whatever significant figures they normally used (so e.g. determining an unrounded measured concentration to be 0.978% and stating the result as 0.98%, rounded to two significant figures) and then, when it came to compare the result to the claim, they would have to round again, this time to one significant figure (1%). Although in this example no problem arises, when the numbers are close to the boundaries difficulties can arise. For example 0.847% rounded to one significant figure is unambiguously 0.8%. Rounding the same starting value first to two significant figures, and then rounding that result (0.85%) to one significant figure can be 0.8% or 0.9% depending on the rounding convention.

However the skilled person knows that the correct way to assess a measured value to a given level of significant figures is to make the measurements, perform the relevant calculations without rounding the raw data and then perform rounding to the appropriate degree of precision required by the claim. That does not involve impermissible double rounding.

There is nothing about the facts of this case which makes double rounding any more of a problem than it might be in every case with numerical limits in a patent claim. I do not accept that Smith & Nephew’s double rounding point is sufficient to reject construing the claim using significant figures.

Convatec’s argument against this construction of the claim was as follows. They argued that the correct way to reason was that once one had rejected Smith & Nephew’s exact numbers approach, the real issue was the degree of precision intended by the patentee in the term “between 1% and 25%”. On that score Convatec pointed out that both sides agreed that at the top of the range, 1-25 extends to 25.5 because 25 extends between 24.5 and 25.5 on either party’s construction. The only question is: what range of numbers rounds up to 1%? Convatec’s key argument was that the 0.5% approach does not produce anomalies whereas the one significant figure (0.95%) approach does.

The main anomaly relied on by Convatec is that rounding powers of ten produces an asymmetric result. This is undeniable. It arises from the numbers themselves. The range for 1% to one significant figure is from 0.95% to 1.5% (or strictly ≥0.95% to <1.5%). Given that the significant figures approach is a convention to deal with normal experimental error and that experimental error is generally expressed as a symmetrical range (say ±5%) because it is regarded as random error, there is an inconsistency.

The various factors to consider in assessing this argument are the following:

First, I accept that the underlying purpose of expressing a value to a certain number of significant figures is to give an indication of precision and that precision is generally expressed more fully as x ± y (i.e. symmetrically). However the asymmetry is simply a mathematically inherent quirk in this very widely used significant figures convention. It is not a reason to reject its use.

Second, Convatec contended that the problem of being faced with a rounding range of from 0.95 to 1.5 was a problem of principle which the skilled person, who is meticulous and careful, would consider in deciding which of the whole numbers/ decimal places approach or the significant figures approach to adopt. Convatec contended that significant figures produced inexplicable inconsistencies within the specification of the patent. I was not persuaded by this. It is true that the asymmetry in rounding powers of ten produces odd results when one looks into the sub-ranges in paragraph [0028] but Convatec magnify the significance of this too much.

Third, in the worked example in the patent (paragraph [0036], which itself uses ranges for some things rather than actual values) it is clear that there has been no consistent handling of significant figures. This fact does not support either party’s construction.

Fourth, the fact the claim is defined by a range means that the asymmetry, although it exists, has no significance. The upper asymmetry does not matter because the claim covers everything “between 1% and 25%”. The asymmetry makes no practical difference as a result. Thus even if a skilled person spotted the point at all (and I am doubtful about that but will assume it in Convatec’s favour) they will see it does not matter.

In my judgment the correct construction of “between 1% and 25%” is one based on significant figures. It is true that the patentee could have said “about 1%” if it wanted a broad construction (as Smith & Nephew might say the patentee should have) but the converse is also true. The patentee could just as readily have said “1.0%” or “1.00%” if it had intended the limit to be construed to a higher degree of precision.

Adopting a significant figures approach to these numbers would be the natural approach of the skilled person. None of the reasons to prefer the other constructions advanced were persuasive.

The significant figures construction sees 1% as a number stated to one significant figure and puts the lower limit at 0.95% while 25% is a number stated to two significant figures and puts the upper limit at less than 25.5%. In addition I think Mr Mellor was correct when he observed in closing that an advantage of this approach is that the size of the relative error margins it produces are comparable at both ends. On this basis the relative error margin at the bottom end of the range is 5% (0.05 as a fraction of 1) while the error margin at the top end is 2% (0.5 as a fraction of 25). These are not the same but they are comparable and they are both small. Whereas Convatec’s construction, although it has the same absolute value for the difference at each end (0.5), the relative error margins are very different. The relative error margin at the lower limit is 50% (because 0.5 is half of 1). In my judgment applying the Protocol directly, either rounding method produces the same degree of certainty for third parties as I have said already, but a 5% relative error margin at the lower limit produces much fairer protection for the patentee than a 50% relative error margin.

The consequences of my findings on construction are that the commercial process which Smith & Nephew wish to employ to make Durafiber Ag, in which the salt concentration is 0.77%, does not infringe the 510 patent. No issue of carryover can make a difference to that conclusion.

I now need to decide whether the four development experiments carried out by Smith & Nephew infringed the patent. Stated without carryover, the concentrations of salt used in the experiments were as set out below. When a range is given that is because the experiment included a number of runs. The numbers are:

Thus without carryover, these experiments would have infringed on any view.

In the original form of the Process Description (including supplements and updates) Smith & Nephew put forward two ways of calculating the concentration taking into account carryover which applied to the commercial process. These two methods can be called the theoretical method and the empirical method. The two methods work in the same way. The mass of salt in the final step of the process is known. The volume of liquid added to the reactor is known. The concentration in weight/volume (g/ml) stated without carryover is just the mass divided by that volume. In order to determine the effect of carryover, the extra volume carried over must be worked out. Then the concentration is just the mass divided by the sum of the added volume plus the carried over volume. The calculation has various details, e.g. a density correction factor for mixtures of ethanol/water.

The theoretical method determines the carryover volume by taking a reported value for the absorbency of lyocell (one of the fibres in Durafiber) and applying it to the mass of material. A figure of 4.5 g/g is used (i.e. 1 gram of fibres is taken to absorb 4.5 grams of water). Thus 1 kg of fibres would carryover 4.5 kgs of water. This theoretical approach shows that the amounts of fluid probably carried over are not trivial. For example in the commercial process the final step uses a volume of about 120 litres and a 2kg mass of fibres, producing a theoretical carryover of about 9kgs or about 10 litres, in other words about 8% of the volume and so an approximately 8% reduction in the stated salt concentration. In the commercial process, if the salt concentration in the final step stated without carryover is 0.77%, the concentration determined on the theoretical method would be 0.71%.

The empirical method was based on performing some tests on the commercial process to try and weigh the amount of fluid being carried over by weighing the wet material. The test was done on the material at the very end of the process after the salt addition step because it was easier to do it that way. The figure used for the weight carried over was an average from nine results. It produced a volume carried over of 19.86 litres, i.e. approximately double the volume arrived at by the theoretical method. In other words the amount carried over on this basis is roughly 16% of the 120 litres volume. In the commercial process the empirical method would reduce the 0.77% concentration figure to 0.66%.

In the Process Description Smith & Nephew described both the commercial process and also the various experiments which had been conducted. In relation to the experiments however, the Process Description only applies the theoretical method to determine carryover to the historical experiments. After all the empirical method was based on testing the commercial process, not the experimental set ups. In the Process Description the effect of carryover applied to the figures for the relevant experiments was stated to be:

Thus stated this way, the experiments would all infringe on the construction I have accepted.

However very shortly before trial Smith & Nephew decided to contend that carryover for these experiments could be determined using the data derived from the empirical method. The argument was that the figures determined in the empirical method can be used to derive a ratio for the mass carried over per unit mass of fibres. So knowing the mass of fibres in the experiment, one can derive a new figure for the volume carried over and adjust the figures for final salt concentrations accordingly. The calculations were set out in a short witness statement of Mr Stephenson and in his exhibit DS-15. The salt concentration values for the four experiments ranged from 0.92% to 0.93%. That is outside the claim.

Convatec objected to this course. They submitted it was far too late in the day to permit such a significant change. It raised numerous issues and would require significant evidence in response.

Smith & Nephew recognised that the amendment was very (very) late and that Convatec must be able to respond but they submitted that in truth this had happened as a direct result of a change in position by Convatec. Until just before trial Convatec were alleging that a large number of the experiments in Smith & Nephew’s experimental programme infringed (many more than the four above). Smith & Nephew argued that the others were covered by the experimental use exception (s60(5) of the 1977 Act) but Convatec did not admit that this was so. Applying the empirical method to all the experiments then in issue was not worthwhile because it left many still within the claimed range. Just before trial Convatec stated that they would not contest the experimental use point. As a result the only experiments which now conceivably mattered were just No.s 40, 43, 44 and 48. When Smith & Nephew focussed on those in isolation they realised that using the empirical carryover method would mean they did not infringe.

I decided that the least bad option was to permit Smith & Nephew to advance this new case and permit Convatec to answer it but to keep the trial on track. One option was to adjourn this point but since it relates to the springboard injunction, it could not be adjourned for long. Such a short adjournment would not solve anything and would just increase the costs. To refuse to let it in altogether risked a grave injustice to Smith & Nephew since although they were putting forward a new point in one sense, the circumstances in which it arose meant it was not fair to blame either side for what had happened. Convatec could hardly be blamed for dropping a point and limiting the issues but equally there was force in Smith & Nephew’s point that it was only when Convatec decided not to contest the experimental use that the focus turned so closely on the four experiments. Although it would put Convatec to a lot of work, I thought Convatec’s legal team and Prof Kennedy would be able address the new issues raised fairly. In the result they did so. Both sides served further experts reports and Smith & Nephew also served further evidence from Mr Stephenson.

The evidence from Smith & Nephew consisted of reports from Prof Burrell and statements of Mr Stephenson.

Mr Stephenson provided the carryover calculations and an explanation of the relevant circumstances. He explained and compared the equipment and methods used in the commercial process and in the experiments. He described how the fabric behaved in the process and how the fluid drained away. Mr Stephenson explained why, in his opinion, it was appropriate to apply the carryover calculation derived empirically on the commercial process to the experiments. His firm opinion was that the carryover calculation involved a conservative estimate of the amount of fluid carried over.

Prof Burrell had dealt with carryover in general terms in his main evidence in the debate about whether carryover would be difficult to quantify. One of the issues was about water being bound physically or chemically into the fibres, the extent to which this occurred and its likely significance. A question which arose relating to the debate whether the empirical carryover values determined on the commercial process were applicable to the experiments was a point about differences in fluid absorption. It was common ground that the degree to which fibres made of CMC and CES would absorb fluid will depend on the salt concentration. Prof Burrell was asked to consider whether the difference in salt concentration between the commercial process (0.77%) and the experiments (say 1.02%) had any effect. His opinion was that the higher salt concentration in the experiments would reduce the absorbency by about 5%. A 5% reduction was taken into account in a further set of figures exhibited to Prof Burrell’s report at REB 7. In cross-examination Prof Burrell accepted that the true value could be higher than 5% but said he would be surprised if it was.

On the carryover issue, Convatec relied on Prof Kennedy. Like Prof Burrell, Prof Kennedy had considered carryover in his main reports. His view, that it was difficult to calculate, was advanced by Convatec to support their argument on construction that carryover would not be included in the claim. On the question of Smith & Nephew’s application of the empirical carryover method to the experiments, Prof Kennedy’s opinion was that the information available about the empirical method was not sufficient to make a proper assessment in order to say whether it was fair or reasonable to apply it. He explained the information he would require in his fifth expert report. Additional information he wanted included information about how the material was handled, the moisture content of the starting material, how the draining step was conducted, the rate of draining, the degree to which the fibres are dispersed during emptying, how long the draining took, whether there was a measurement of the amount of fibres washed away down the draining and the dissolution of the fibres. Mr Stephenson’s eighth witness statement sought to provide answers to the questions raised by Prof Kennedy.

One point was that the variation in weights measured in the nine runs in which the wet fibres were weighed at the end of the commercial process needed to be taken into account. This was a point Mr Stephenson accepted. Another point was that the mass of fibres in the nine runs was slightly different (save in one case) from the mass used in the experiments. Again Mr Stephenson accepted this.

In the light of the evidence Smith & Nephew made adjustments to the figures taking into account points they and their witnesses were prepared to accept. The final figures for the experiments emerged during the trial and were stated clearly in an annex to Smith & Nephew’s closing. The numbers are:

As Convatec pointed out, these numbers have moved very close to the boundary of the claim. As Smith & Nephew pointed out, they have not reached 0.95% taking on board all the points which Convatec raised which Smith & Nephew were prepared to accept.

In cross-examination Prof Kennedy said that Smith & Nephew’s estimates of carryover were reasonable approximations. Smith & Nephew relied heavily on this opinion but in my judgment that reliance was misplaced. In saying what he said Prof Kennedy made it clear that his view was that the estimates of carryover by Smith & Nephew were indeed reasonable approximations but they were not precise enough to decide whether the experiments were outside the claim or not. The problem for him was that the numbers are too close.

For example if the experimental uncertainty associated with a value of 0.94% is as little as ± 0.02%, the range will be 0.92% to 0.96% and in that case the true value could not be said to be outside the claim. Prof Kennedy did not accept that the information was sufficient to say that the carryover in the experiments was outside the claim.

If one could be confident that the estimate of carryover was conservative then the uncertainties would not really matter but Prof Kennedy did not agree with Mr Stephenson about that. Prof Kennedy maintained that despite Mr Stephenson’s evidence and explanations, there was not enough information available to him to be able to express an opinion to help the court decide that question.

In cross-examination Prof Burrell accepted that if he was seeking to ascertain the degree of conservatism of an estimate about this process, he would want to see the process run.

I remind myself that the question I have to decide is whether the four experiments infringe the claim. On that the burden of proof is on Convatec. The question of fact is whether the figures for the four experiments, calculated using the empirical carryover approach derived from the commercial process, can fairly be relied on. If they can be relied on, the experiments do not infringe. The question is whether I accept those figures as a fair representation of true salt concentration in the experiments.

On the basis of the Process Description provided by Smith & Nephew, the four experiments infringe. The figures in the Process Description, using the only carryover estimate Smith & Nephew advanced as applicable to them until just before the trial, are within the claim. The figures are more than 0.95%. In my judgment, although the legal burden to prove infringement rests with Convatec, Smith & Nephew bear an evidential onus to establish that the figures they derive using the empirical carryover method are fair or at least that the true figures must be further from the claim. They are the ones seeking to show that the empirical carryover method can be applied in this way.

I am not satisfied that it is more likely than not that the figures advanced by Smith & Nephew can be relied on in this way. I am not satisfied they are right nor am I satisfied that they are a sufficiently conservative estimate so that uncertainties do not matter. My reasons are as follows.

First, the context in which this debate is taking place is that the numbers are very close to the claimed boundary. The way in which the figures moved even during the trial illustrates how tight things are. Very small differences in the percentage concentrations are enough to take the figures into or out of the claim.

Second, Smith & Nephew submitted Prof Kennedy’s opinion was based on speculation and exaggeration of very small effects. I do not agree. One of the problems for Smith & Nephew was that given the way things turned out, Prof Kennedy was never given a chance to view the processes in question. All he had to go on was the description given by Mr Stephenson. While I have no doubt Mr Stephenson was expressing his genuinely held views and I recognise that he is the person best able to describe and comment on Smith & Nephew’s own work, his evidence was not sufficient to satisfy me that Professor Kennedy’s concerns were misplaced or exaggerated. The Professor explained that in order to answer the questions he would need to see what was going on, ask questions on the shop floor and do some measurements. In my judgment this reflected the Professor’s genuine concerns and I accept them.

Third, the empirical measurements on which the method is based were undertaken on the commercial process but are now being applied to the experiments. The experimental conditions were different and the differences did matter. This was shown by the point on the difference in absorbency caused by the different salt concentrations which led Prof Burrell to recalculate the figures. I am not satisfied that the magnitude of the effect of the differences in the conditions is sufficiently small to mean that that I can be satisfied the results are outside the claim.

Fourth, the empirical measurements on which the method is based were experiments conducted for litigation. The fact Convatec did not ask for a repeat at the time the results were presented in the Process Description is not very significant since at that stage Smith & Nephew were not deploying the empirical results in the way they did at trial. The experiments did not then matter that much. If Smith & Nephew had made the change to their case at an earlier stage, Convatec would have been entitled to require a witnessed repeat. I permitted Smith & Nephew to rely on these unwitnessed experiments and I have no doubt they were honestly conducted. Nevertheless when the outcome depends on such tight margins, the fact that no witnessed repeat was possible is a factor going to weight which I am entitled to take into account.

I conclude that the only reliable figures I have to go on in relation to the experiments are the figures in Smith & Nephew’s Process Description. On that basis the four experiments infringe the 510 patent.

The facts can be stated shortly. On the basis of my finding above, Smith & Nephew infringed the patent when running the four development experiments. These experiments produced substantial amounts of silverised Durafiber material which was used to conduct further tests to generate data which was provided to the relevant regulator. The data was a necessary part of the application for regulatory approval. For example Experiment 40 was used to make samples for use in biocompatibility studies.

The experiments took place between November 2009 and July 2010. In October 2010 Smith & Nephew applied for CE mark approval for Durafiber Ag. In June 2012 CE mark approval was granted. The specification of the product to which this CE mark approval applies is wide enough to mean that product made using the non-infringing process is also covered by the existing CE mark approval. So from a regulatory point of view, Smith & Nephew does not need to obtain another authorisation to sell Durafiber Ag in the UK and elsewhere in the European Union.

Convatec submit that on these facts Smith & Nephew should be restrained from selling Durafiber Ag in the UK and elsewhere in the European Union until the date Convatec submit they would have obtained CE mark approval if they had not infringed the patent. To decide this I need to first consider the law on injunctions of this kind. Then against that background I will consider the disputes about what would or could have happened. This will involve the evidence of Mr McKinlay, Mr Stephenson and Mr Armstrong. Finally I will decide what to do.

Normally final injunctions are granted at trial to prevent a defendant from committing in future an act which the defendant was threatening and intending to commit but which the court has now decided would be unlawful. At the trial the defendant will have argued that the act was lawful but the court has decided to the contrary. The injunction is final because it is granted after the parties’ legal rights have been finally determined (subject to appeals). It is different from an interim injunction which is granted pending trial and subject to a cross-undertaking in damages. Interim injunctions are granted at a point in the legal process when one does not know whether the restrained acts are lawful or unlawful.

Sometimes final injunctions are called “permanent” injunctions because they are not pending trial and in that sense are permanent. In a patent case a normal final injunction is time limited in that it will only last until the patent expires.

The word “springboard” is used to refer to an injunction granted in particular circumstances. Usually what has happened is that a defendant has built up a bridgehead or springboard into a market and the springboard injunction is designed to deprive the defendant of that benefit. The term “springboard” comes from Terrapin v Builders Supply Co (Hayes)[1967] RPC 375 which was an interim injunction application. However the sort of springboard injunction sought in this case is a final injunction.

Often the acts restrained by a final springboard injunction would otherwise be lawful. If the acts restrained would be unlawful anyway then no conceptual difficulty arises. The interesting question is whether and in what circumstances the court should grant a final injunction which restrains what would otherwise be a lawful act.

In opening Smith & Nephew submitted that it was clear on the authorities that springboard injunctive relief should never be granted to restrain lawful activity. That submission goes too far and in closing Mr Mellor accepted it went too far. He accepted that the jurisdiction existed but submitted it should be treated with real care.

I will start with the terms of the Directive on the Enforcement of Intellectual Property Rights (2004/48/EC). The relevant sections are Section 1 and Section 5. In Section 1, Article 3 (1) sets out the general obligation to provide measures, procedures and remedies which are fair and equitable, not unnecessarily complicated or costly or entail unreasonable time-limits or unwarranted delays. Article 3(2) provides that the measures, procedures and remedies shall be effective, proportionate and dissuasive, shall avoid creating barriers to trade and provide safeguards against abuse.

Section 5 on measures resulting from a decision on the merits of the case contains three provisions, Articles 10 to 12. Article 10 relates to corrective measures such as orders for delivery up. The article contemplates such corrective measures being applied to goods found to be infringing and tools for making them. Article 11 relates to injunctions. In terms it provides for an injunction against the infringer “aimed at prohibiting the continuation of the infringement.” Article 12 is concerned with alternative measures. It provides that in a proper case the court may award pecuniary compensation to the injured party instead of making an order for an injunction or delivery up.

The springboard jurisdiction has been recognised in patent cases. In a proper case the court has jurisdiction to grant an injunction which restrains a defendant from selling a product for a period after the patent has expired. For example in Generics v Smith Kline & French (C-0316/95, 9^th July 1997) the European Court of Justice heard a reference from the Netherlands in which the Dutch court had granted an injunction to restrain Generics from selling a drug (cimetidine) for 14 months after the expiry of the relevant patent as a consequence of pre-expiry acts of infringement. Generics had submitted a sample of the product to the regulator in order to obtain a marketing authorisation. That act had been an act of patent infringement. As the ECJ put it (paragraph 27) “if Generics had respected SKF’s patent right, it could not have submitted the cimetidine samples until that patent had expired. SKF would thus have been able to continue to market its product without competition […] throughout the period required to obtain the marketing authorisation.” The remedy granted by the Dutch court was not disproportionate (paragraph 28) and the ECJ held that “an order of a national court prohibiting the infringer from marketing such a product for a specified period following expiry of the patent in order to prevent him from deriving any unfair profit from his infringement” was not an impermissible barrier to intra-Community trade contrary to then Arts 30 and 36 E.C. (paragraph 29).

In Union Carbide v BP Chemicals[1998] FSR 1 Jacob J (as he then was) accepted that post-expiry injunctions were available in patent cases on the strength of Generics.

In Dyson v Hoover (No. 2) His Honour Judge Fysh QC (sitting as a Deputy Judge of the High Court) thoroughly analysed the law on post-expiry injunctions starting from their recognition by Lord Brougham in Crossley v The Derby Gas Light Co. (1834) 4 L.T.Ch 25 in paragraphs 16 to 46 of his judgment. He noted that in Gerber v Lectra[1995] RPC 383 Jacob J held that damages for acts committed post expiry which were foreseeable damage caused by infringements committed before expiry (a springboard) were recoverable. This conclusion was upheld by the Court of Appeal ([1997] RPC 452). Judge Fysh granted the narrower of two post-expiry injunctions sought by the patentee in Dyson.

Smith & Nephew referred to the judgment of Arnold J in Vestergaard Frandsen v BestNet Europe[2009] EWHC 1456 (Ch). On appeal in that case ([2011] EWCA Civ 424) there was no challenge to the judge’s analysis of the law which Smith & Nephew relied on although there was a challenge to his decision to refuse certain injunctions as a result. The appeal on that point was dismissed (paragraphs 51-57).

Smith & Nephew submitted that in Vestergaard Arnold J had comprehensively reviewed the authorities and refused to grant a form of springboard injunction. They submitted this judgment addressed the relevant legal principles but I do not accept it is that simple. The authorities the learned judge reviewed were the cases which bore on the legal principles relevant to the matter before him, which was a case of misuse of confidential information and a claim for an injunction to restrain making and selling a product whose making and selling did not amount to a continued misuse of confidential information. When he came to summarise his conclusions at paragraph 93, the judge stated them expressly by reference to misuse of confidential information. I am sure that was no accident.

For example since the case before Arnold J was about misuse of confidential information, which raises its own special factors as can be seen from that judgment itself, the parties do not appear to have cited Generics v Smith Kline & French. The judge does not refer to it in terms although he does refer to Dyson which itself refers to it. No doubt the parties did not cite Generics to the court in Vestergaard because that judgment of the ECJ is concerned with a post-expiry patent injunction and not with relief in a confidential information case.

Convatec also referred to the judgment of Neuberger J (as he then was) in Kirin Amgen v TKT[2005] FSR 41. There the court found that the defendant’s activities would infringe the patent. It was agreed that the final injunction should be stayed pending appeal but there was a dispute about the terms of the stay. The point was that in the meantime the defendant was going to carry out work to generate data for use in applications for European regulatory approval. That work would have been prevented by the final injunction. Neuberger J held that the injunction should be in unqualified form. His reasoning included accepting the point that if the defendant was able to obtain regulatory approval through infringing acts which could only have taken place as a result of the stay then, if it lost its appeal, it should not be entitled to keep the benefit of the European regulatory approval. If the appeal failed, the net effect would have been to deprive the defendant of the benefit of a regulatory approval obtained using data generated as a result of infringements. The defendant would not be able to sell product lawfully in the UK and throughout Europe.

Convatec also referred to my judgment in Merck Canada v Sigma (No 2)[2013] RPC 2. This was a patent case in which I ordered the defendant to deliver up stock imported in infringement of a patent. The defendant wished to retain the stock in order to sell it once the SPC expired. Sale after the SPC expired would be lawful. I ordered delivery up. One reason I gave was because there was a springboard (paragraph 24), I held that the defendant’s “retention of unlawfully imported stock would give it an unwarranted advantage when the SPC expires.” Convatec did not refer to the judgment of the Court of Appeal in Merck v Sigma. The defendant appealed and on the main issue the matter has been referred to the CJEU. However the defendant also specifically appealed the delivery up order. The Court of Appeal upheld that order ([2013] EWCA Civ 326; Patten, Black, Kitchin LJJ). In paragraph 94 of his judgment Kitchin LJ said that:

The facts of Merck v Sigma are different from the present case and the remedy is different too, although an order for delivery up is really just a mandatory injunction. Nevertheless this seems to me to illustrate the essential point that the court can intervene to deprive an infringer of an unwarranted advantage gained from their act of infringement.

I am satisfied that the court can, in a proper case, make an order of the kind sought by Convatec. The cases make that clear. As for the Enforcement Directive, a springboard order does not fall neatly within the words used in either Art 10 or 11 of the Enforcement Directive but in a proper case it would fall within Art 3. Such an order can be fair and equitable, effective and dissuasive. The ECJ (as it then was) recognised in Generics that such a remedy can be proportionate and is not a barrier to intra-Community trade. Moreover it seems to me that the scope of the damages available in Art 13 (appropriate to the actual prejudice suffered) would include the sort of damages recognised as being available in Gerber. If damages are in principle recoverable then I do not see why an injunction should not be able to be granted to prevent the harm occurring in the first place. In any event the Enforcement Directive does not limit the remedies which are available, it provides a minimum standard.

I turn to look at the considerations which condition the exercise of the jurisdiction. Taking the three patent cases above in which springboard relief has been awarded:

In Generics the Dutch court granted an injunction to restrain sales of generic product for 14 months. The 14 month period appears to have been chosen because it usually took at least 14 months from regulatory application to registration (see Jacobs AG paragraph 15). In fact the approvals in that case had taken more than 14 months to obtain (see Jacobs AG paragraph 8).

In Dyson a 12 month injunction was granted by reference to a particular product found to infringe (paragraph 51 and 63). An injunction in wider form was refused (paragraph 48). The 12 month period came from the fact that it took the defendant 12 months to develop and launch the product in question (see paragraph 9 (including footnote 14) and paragraph 16). There was evidence that the normal period for development was 16-19 months.

In Merck v Sigma the order for delivery up only prevented the defendant from selling a particular batch of product and thereby entering the market more quickly then it might otherwise do. The defendant was free to sell other batches as fast as it could post-expiry.

The start date of injunctions to restrain sales in these cases is easy to identify, it was the date of expiry of the patent. The period chosen in Generics was the least period the application might have taken and was shorter than the actual periods. The period in Dyson was the period actually taken, which was shorter than the normal period. In Dyson and in effect in Merck v Sigma, the relief did not prevent the defendant from competing with the patentee altogether, it only had the effect of preventing certain activity defined narrowly.

Overall in my judgment that the following factors are relevant in considering final springboard relief in a patent case:

Caution is required before a final injunction is granted restraining an otherwise lawful activity. Nevertheless in a proper case it will be.

The nature of any unwarranted advantage relied on should be identified. The precise relationship between the unlawful activity in the past and the later acts which are said to exploit that unwarranted advantage needs to be considered.

If an injunction is to be granted it must be in an appropriate form and for a duration which is commensurate with the unwarranted advantage relied on.

The court must be particularly careful not to put the claimant in a better position than it would be if there had been no infringement at all, especially if otherwise lawful competitive activity will be restrained.

In considering what relief to grant, the availability of other remedies apart from an injunction needs to be taken into account, not only damages but, as in Vestergaard, the availability of an account of profits should be considered too.

With these principles identified I will consider what to do.

I will address the following topics:

Start date;

Date of filing for CE mark approval;

Date of obtaining CE mark approval;

Looking at all the factors together.

Smith & Nephew are not selling Durafiber Ag in the UK or elsewhere in Europe pending this action but they intend to start if they can as soon as this judgment is delivered. I was told by Smith & Nephew’s Counsel that they are selling in the USA pursuant to approval from the FDA but the injunction sought by Convatec does not apply to that.

In a sense the start date is clear since it would be the date of judgment, but the period of the injunction is not referable to that date. The injunction sought is to run until September 2014 irrespective of when it starts. That is because Convatec contends the springboard starts from October 2012. That is the date on which Smith & Nephew had finished developing the non-infringing process. Starting from that point, Smith & Nephew would be ready to apply for CE mark approval in January 2013 and the September 2014 date follows.

Smith & Nephew contend that the start date of the springboard ought to be much earlier. They submit it should be either May 2009, when Smith & Nephew started developing a CES based product in the first place or September 2011, when Convatec unconditionally amended the claims of the 510 patent to introduce a limitation to “between 1% and 25%”, thus making it possible for a process running below that limit to avoid infringing. It is common ground that if the start of the springboard is either of those earlier dates, then on any view no injunction should be granted. The end date of any advantage to Smith & Nephew will be in the past.

The rationale advanced by Smith & Nephew for the earlier dates derives from the fact that the patent as granted did not contain the 1% lower boundary and moreover has been accepted to be invalid in its unamended form. Smith & Nephew contend that if the patent had been granted in the amended form from the outset then they would have undertaken a very different approach. They would have started their development using the non-infringing process from the outset and would not have invested money, as they did do, in a process which fell within the amended claim. Although Smith & Nephew accept that as a matter of law, amendments are deemed to relate back to the date of grant, Smith & Nephew contend that in assessing what would have happened, the date the amendment was made has to be taken into account. A small point is that May 2009 cannot be right on any view since it is before grant but the same logic would work for Smith & Nephew from May 2010 (the actual date of grant).

In his witness statement Mr McKinlay said that if the patent had been limited to reflect the amended claim from grant then Smith & Nephew would not have developed the earlier process. Instead they would have developed the non-infringing process. He also addressed what happened after September 2011 when Convatec unconditionally applied to amend. He said that at that stage it was two months before the date when the validity trial of the 510 patent was listed to start (end of Nov 2011). The trial was so close that Smith & Nephew decided to wait and see rather than amend their existing process. What then happened was that the November fixture was adjourned on the eve of the trial because Convatec’s expert, Prof Qin, was ill. It was not heard until March 2012 and judgment not given until June 2012. Only in June 2012 did Smith & Nephew start developing the non-infringing process.

Mr McKinlay’s cross-examination showed that in fact Smith & Nephew were convinced the proposed amended patent was invalid. That was the reason they made no attempt to design around it even though it was known that salt concentration was of little consequence. This testimony undermines Smith & Nephew’s case.

Smith & Nephew submitted that in re-examination Mr McKinlay qualified this testimony. He explained that in his opinion the resources available were also a factor in the decision not to change the process after September 2011. Although I accept that resources were reduced in that period (since Smith & Nephew had to relaunch Durafiber having been kept off the market as a result of undertakings in the first action) I do not accept that this was a factor of any real significance.

A further point was whether the actions in September 2011 were a guide to what would have happened at the outset had the patent been granted in the amended form. I accept that the outset of a project is different from the position after one has invested heavily in a development exercise but nevertheless I do not accept that Smith & Nephew would have done anything different in this case if the patent had been granted in the amended form. I find that Smith & Nephew were convinced the patent was invalid at all material times until the judgment of the Patents Court went against them and acted accordingly.

So the start of the springboard on any view should be seen as the point at which the Patents Court gave judgment.

The next issue to consider is the period of the springboard. Starting from June 2012, it took Smith & Nephew until October 2012 to finish developing the non-infringing process. That is not disputed. There was then a dispute about when Smith & Nephew would have been ready to file for CE mark approval. Convatec submitted Smith & Nephew would have been ready in February 2013 and Smith & Nephew contended they would have been ready in December 2012.

The debate is about whether the various tests to be carried out (biocompatibility, stability and validation) would be done in parallel or in series. Mr Stephenson’s view was that the biocompatibility studies took about three months but he maintained they would be done in parallel with the other tests. Convatec put to him that originally the tests were done in series but Mr Stephenson did not agree that this meant Smith & Nephew would do them in series on this occasion.

I find that Smith & Nephew could have been ready to file for CE mark approval in mid December 2012. If they wanted to get a move on, they could have run the tests in parallel.

To be lawfully sold in the EU under the Medical Devices Directive (93/42/EC and amended by 2007/47/EC) a medical device must bear a CE mark. For this the manufacturer must sign a declaration that the device meets the essential requirements of the directive and that it is fit for its intended purpose.

Durafiber Ag is classified as a class III medical device. It is a device with an ancillary medicinal component (silver). The relevant process for obtaining CE mark approval requires the applicant to prepare a technical design dossier in considerable detail. The dossier has to describe the design, manufacture, quality assurance mechanisms, safety and performance of the product and must contain the documents needed to assess whether the product conforms to the Medical Devices Directive.

To obtain CE mark approval for such a device the application must be submitted to an organisation (a “Notified Body”) which is able to grant CE mark approvals for medical products in general (such as the British Standards Institute (BSI)). Since the device has an ancillary medicinal component, the Notified Body will then file an application to a “Competent Authority” such as the Medicines and Healthcare products Regulatory Agency (MHRA). Once the Competent Authority has issued an opinion, the Notified Body must take the opinion into account but is not bound by it. The Competent Authority is meant to issue its opinion within a period of 210 days.

The original CE mark application for Durafiber Ag was submitted to the BSI as the Notified Body and the MHRA were the Competent Authority. The period of time the process actually took was 20 months. However Smith & Nephew contended that this was a poor guideline. Mr McKinlay’s experience at Smith & Nephew was that such applications normally took no more than 10 months. This 10 month estimate is based on the 210 day period applicable to the MHRA with suitable extra periods either side for the work done by the BSI. Mr McKinlay also produced a table showing the actual periods for all the new Smith & Nephew class III devices since 2005. The actual periods (save for Durafiber Ag) were 5, 7, 9 and 10 months.

Mr McKinlay’s view was that special factors had caused the period taken by Durafiber Ag to be much longer than the norm. The special factors were: (i) de-prioritisation of the Durafiber Ag CE mark application in favour of other products because at the time Durafiber was subject to undertakings to Convatec and litigation uncertainty, (ii) lack of resources at the MHRA and (iii) unnecessary interventions by the BSI. I will address these in turn.

The point on de-prioritisation was that Mr McKinlay said that one reason the process took longer than normal was because Smith & Nephew decided to ask BSI to ask MHRA to prioritise two other products, Iodosorb and Allevyn Ag, over Durafiber Ag. Convatec submitted that Mr McKinlay had accepted that this only accounted for about 6 weeks of the period of 20 months but that is not a fair reflection of his evidence. It does not take into account the periods which the MRHA need internally to feed in and out of projects and therefore are not able to pick up another one. Convatec also submitted that although there was documentary proof of deprioritisation over Allevyn Ag, there was none for Iodosorb and he was wrong about this. Mr McKinlay maintained his evidence in cross-examination and explained that he knew what happened because the work was carried out in a small team. I accept his evidence.

There was a major debate about lack of resources at the MHRA leading to delays. It is clear that there was a problem of lack of resources, it is clear that in June 2011 and up until May 2012 the BSI were referring to this problem as a reason why delays might occur and it is also clear that it was being said by the MHRA that steps were being taken to try and address these delays. However it is not clear what effect this would have had in the hypothetical world in which a CE application is made in December 2012. Mr McKinlay maintained that the MHRA delays were to be rectified although, as Convatec pointed out, it was based on hearsay from unnamed sources and as Convatec submitted, Smith & Nephew were in a position to call better evidence on this point if they had wished to do so. Mr Armstrong doubted whether new resources would have the desired effect within the relevant timescale but he was not in a position to say what the MHRA had actually done either or what effect it might have had. The point Mr Armstrong makes is that new resources take time to have an effect. This is a fair point but it seems to me to be unlikely that the MHRA would not have taken steps, starting in 2012, to address the delays. By December 2012 I think it is more likely than not that the delays would have been reduced.

The point on the intervention from the BSI was that one cause of some delay in the process of approval of Durafiber Ag was said to be that a reviewer in the BSI intervened in Smith & Nephew’s proposed response to the MHRA’s questions and recommended additional work be done. Mr McKinlay’s view was that this was unnecessary and untypical and so could not be expected to delay the hypothetical CE mark application under consideration. Convatec submitted that it could not be ignored. I disagree. I accept Mr McKinlay’s point.

Considering the period overall, Mr Armstrong’s original opinion was that application would take 20 months, using what actually happened as a guideline. Then when he saw Mr McKinlay’s evidence about the special factors which led to delays in the original Durafiber Ag application, he gave his best estimate as 11-14 months based on Mr McKinlay’s evidence. Convatec submitted that they had undermined Mr McKinlay’s special factors in cross-examination and so, subject to 6 weeks de-prioritisation, Mr Armstrong’s initial estimate stands. I do not accept this for two reasons. First I do not accept that Mr McKinlay’s special factors were really undermined at all. Second Mr Armstrong was clearly prepared to accept an 11-14 month period as a realistic estimate, albeit based on assumptions based on what happened in practice.

Looking at this issue in the round, I find the position is as follows. One does not know how long a CE mark application made in December 2012 would take. I am sure that 20 months is too long. I am sure on the evidence that the longest realistic period one could consider is 14 months.

I am also sure that although the process is capable of taking less than 10 months, I doubt it would have done that if it had started in the relevant period. There is not sufficient evidence about the extent to which the MHRA has rectified its delays to establish that.

However I am less sure about the argument between 10 months and 14 months. This spans Mr McKinlay’s estimate and Mr Armstrong’s estimate of 11 to 14 months. Smith & Nephew submitted I should prefer Mr McKinlay’s opinion to that of Mr Armstrong because Mr McKinlay had much greater experience and because Mr Armstrong’s opinions added nothing to what was apparent from Smith & Nephew’s documents. I am not prepared to do that. Both men are experienced in the regulatory affairs. The estimates they each have made are based on what happened in the past and their own experience. A major imponderable is what the effect might be in 2013 of lack of resources and delays experienced in the MHRA at an earlier period. In my judgment both witnesses were essentially speculating about that. I am also wary of manufacturing a false kind of certainty by picking a concrete period out of these estimates. I will return to this point below.

A point urged by Smith & Nephew is that they gave undertakings to Convatec in the first action which lasted about a year and which Convatec were not entitled to. There is to be an inquiry on the relevant cross-undertaking. These undertakings slowed down the whole project. Smith & Nephew submit this ought to be taken into account but I am not convinced it is a strong point on the facts. The undertakings do explain part of the chronology in this case but they had been discharged well before June 2012 and, apart from providing part of the explanation why the CE mark process was de-prioritised, they do not otherwise have a direct impact. Smith & Nephew may (or may not) recover on the cross-undertaking. If I deduct a period from a springboard injunction to correspond to the period of the undertakings, where does that leave the inquiry? Would there need to be a discount? The parties did not address this and I doubt it is legitimate. In my judgment, beyond the detailed point on the facts about de-prioritisation, there is no sufficient nexus between the undertakings and the length of any springboard to take it into account.

I would not refuse an injunction as a result of the availability of other remedies. Financial remedies such as damages or an account will be difficult to quantify in any case. An account of profits might well not compensate Convatec properly since when selling into a (former) monopoly market the new competitor’s profits will often be lower than the former monopolist’s losses because the market price will be likely to drop. Moreover awarding a financial remedy instead of an injunction for a limited period does not avoid grappling with the problem of the period since a period would need to be determined for the financial remedy too. An injunction for a limited period would be a fairer remedy.

The scope of the injunction sought by Convatec is broad. It would prevent Smith & Nephew from selling a product which is not the subject matter of the invention. By that I mean to draw a distinction from Generics and Dyson. In those cases the product which could not be sold was a product within the claim. The reason the activity restrained by the injunction was otherwise unlawful was purely temporal. This injunction would prevent Smith & Nephew from doing something which does not and never would have infringed a valid patent. This is not a reason to refuse springboard relief altogether but it is a factor to take into account in deciding if the unwarranted advantage obtained would support it.

The relationship between the infringing acts and the later non-infringing acts is not direct. Whereas in Generics samples were submitted (an act of infringement) and in Dyson a number of infringing acts were committed including actually launching the product on the market (paragraph 49), in this case the link between the infringement depends on the submission of data to a regulatory authority. I think this is an important point. Although, as Neuberger J recognised in Kirin-Amgen, an injunction to restrain infringements has the effect of preventing the generation of the regulatory data, the fact remains that data does not infringe the patent and that submitting data is not itself an act of infringement. I do not have to decide whether this alone would be a sufficient ground to refuse any springboard relief but again it seems to me to be a factor which should be taken into account.

The evidence has left the length of the putative CE mark application period in doubt. I am satisfied that the length of the period it would have taken to obtain CE mark approval would be between 10 and 14 months. The least period it would have taken is 10 months and the most is 14 months. Starting from December 2012, grant of CE mark approval in February 2014 is no more likely than a grant in October 2013.

By obtaining CE mark approval in June 2012 Smith & Nephew secured an advantage which was unwarranted because they obtained it as a result of infringing the patent. If Smith & Nephew had launched Durafiber Ag in reliance on that CE mark approval when the non-infringing development was finished in October 2012 then they might have been exploiting that unwarranted advantage. They might have been liable for damages or an account of profits now if that is what they had done. But they did not. They have stayed off the UK/European market until now. By staying off the market until the judgment despite the fact that their commercial process does not infringe, Smith & Nephew have in effect submitted to a form of voluntary springboard remedy in any case.

If Smith & Nephew started selling Durafiber Ag in the UK and Europe today, the question I have to ask myself is whether they could be said to be exploiting an unwarranted advantage gained from the infringements in 2009/2010?

In my judgment they would not. If Smith & Nephew had applied for CE mark approval in December 2012 it is just as likely that they would have obtained it by now as that they would not. To regard sales of non-infringing Durafiber Ag in the UK or EU from now on as being the product of an unwarranted advantage would be unfair. To prevent those sales or award damages or an account in respect of them would overcompensate the patentee for the harm caused by the infringements in the past.

If the infringements in the past had not taken place, Smith & Nephew would have had to obtain CE mark approval after they finished developing the non-infringing process. They could have got it by now. The least period it could have taken to get approval has now passed. Unless I am satisfied that they would not have got the CE mark by now, I should not award springboard relief. I am not so satisfied.

There is a further reason why it would be unfair to regard the ability to sell Durafiber Ag after today as something unwarranted or illegitimate. In the post-expiry injunction cases the start date for the springboard is clear. In this case it is open to argument. To stigmatise sales taking place from now until February 2014 attributes too much significance to the dates associated with the validity trial. I have found that Smith & Nephew would not have started the non-infringing development until they had the first instance validity judgment. In fact that was in June 2012 but if Prof Qin had not fallen ill, the trial would have been in November 2011 and the relevant realisation would have come at least four months earlier. The knock on effect of this is to bring all the dates forward. The longest period I am prepared to contemplate, 14 months, would expire in Oct/Nov 2013. On that basis Smith & Nephew’s decision to stay off the European market until judgment in this trial (in December 2013) has already more than compensated Convatec. There is no good reason why the impact of the adjourned trial should fall on Smith & Nephew alone.

I find that any springboard gained by Smith & Nephew from the infringements in 2009 and 2010 has now expired. As a result I do not have to resolve the argument between 10 and 14 months.

I find that Smith & Nephew’s commercial process for making Durafiber Ag does not infringe Convatec’s 510 patent. Some of the work carried out by Smith & Nephew in 2009/2010 which led to data being generated for use in the application for regulatory approval did infringe the 510 patent. However I refuse to grant an injunction to restrain sales of Durafiber Ag.

After receiving this judgment in draft form both parties identified a number of typographical and other minor errors in the judgment which I have taken into account. I am grateful to them. In particular I am grateful to them for identifying an elementary slip in the equation in paragraph 55. I have corrected it.

Smith & Nephew submitted that in paragraph 111 I should not characterise the amounts produced by the development experiments as “substantial” because (i) there was no evidence as to whether the amounts were substantial or not, (ii) it was not an issue which had to be decided, and (iii) the question of substantiality is one of context and can change from one situation to another. I disagree with point (i). There was evidence about how much was made. Moreover many of the runs were on a pilot scale. I disagree with point (ii), the amount made is part of the relevant context in deciding the springboard issue. I agree with point (iii) and made the finding in that context. Thus I have not altered paragraph 111.