05 March 2017 by Paul Harris
A numerical patent claim where the numbers don’t add up
I have something of a thirst for knowledge – for instance, I read all of the English courts’ decisions, as well as those of the Court of Justice of the European Union. But I found the first 2017 case to be decided in the Intellectual Property Enterprise Court of particular interest, as it happens to be one that I was involved with at the start of the case at a previous firm.
The particular case (Jushi Group v OCV Intellectual Capital) concerned glass reinforced fibre used in the manufacture of turbine blades for wind farms. The patentee had broadly claimed a patent relating to a composition for glass reinforced fibre, which used a mixture of numerical values for specific chemical constituents, some of the constituents of which comprised less than 1% of the total weight of the composition.
“By the time the case came to trial, Jushi decided to concede there was infringement, and also dropped the invalidity attack based on common general knowledge and other prior art”
The patentee had also, for good measure, thrown in criteria based on a ratio between two constituents, together with a percentage weight value comprising the summation of one of the key constituents added to two others. Yet there was nothing in the claims which depended on those criteria, such as the mechanical strength of the glass fibre meeting those criteria.
Jushi, wishing to sell its own products in the UK, sought to ‘clear the path’ by seeking a declaration of non-infringement (DNI) as well as invalidating the patent on the basis of prior art and common general knowledge.
By the time the case came to trial, Jushi decided (for reasons best known to themselves) to concede there was infringement (so dropped the DNI part of the case), and also dropped the invalidity attack based on common general knowledge and other prior art. The case hinged, therefore, on a single piece of prior art known as ‘Neely’.
That piece of prior art also comprised a range of exactly the same chemical constituents. However, the range for each constituent varied slightly from that claimed in the patent. Accordingly, some aspects of the prior art fell within the range for certain chemical constituents disclosed in the patent, whereas others overlapped.
Significantly, one of the examples in the prior art fell entirely within the ranges of the chemical constituents but for the ratio and the composition, the criteria fell just outside.
However, shortly before commencement of the case, the Court of Appeal decision in Smith & Nephew v ConvaTec Technologies in 2015 established how numerical values should be construed where some are expressed in whole numbers and others are stated as a decimal to two significant figures (for example, 0.14).
In the Smith & Nephew decision, the Court of Appeal considered that a value of 1 would be read as low as 0.5 and up to 1.5. Taking the same approach meant that the ratio and the composition criteria would encompass the example in Neely in its entirety, and thus the patent would be rendered invalid for lack of novelty.
Having heard each sides’ experts’ evidence, the judge preferred that of the patentee. It was that expert who considered that in the light of common general knowledge – which comprised three standard texts, including a US standard for reinforced glass fibre products – the skilled person (through whose eyes the patent must be judged) would appreciate that in certain instances some numbers would be considered to be specific and so would not be rounded up or rounded down as the case may be.
That being the case, the judge had a ‘get out of jail’ card to enable him to interpret the claim narrowly enough to avoid invalidity based on the example. Had an infringement argument been run where the ranges or ratio fell outside the specific numbers claimed, then the judge would have had a more difficult task in explaining how, on the one hand, the patent was not invalid because it had to be read narrowly in respect of certain numbers, yet was broad enough to catch similar, but not identical, numbers in an infringing product.
It is not known why the DNI was dropped and infringement conceded, or indeed on what basis. That was undoubtedly a costly decision.
Having been able to avoid the prior art for certain criteria, the judge still had to overcome the problem of why overlapping ranges did not render the patent invalid for lack of novelty or obviousness.
“The outcome of this decision is likely to come as a surprise to many – and I am certain it did to Jushi”
This he did by disagreeing with an earlier judge’s decision (which was not binding on him) so that even where the overlap was 99%, the invention in the patent could still claim novelty over the prior art, because it could not be said with certainty that the skilled person would perform the prior art in conformity with the 99% of the range disclosed.
To my mind, although this may avoid the argument of invalidity based on lack of novelty (anticipation by the prior art) it surely could not avoid an obviousness attack. However, the judge concluded it could.
The judge was helped in his decision by the fact that the patentee’s counsel produced a probability figure of 3.9, representing the probability of the skilled person being able to carry out all of the various ranges of the various chemical constituents, which would fall within the scope of the claim of the patent.
The judge did, however, while accepting the probability figure, also consider how the skilled person would approach this, and applied the case law used within the European Patent Office Technical Boards of Appeal. In the TBA’s approach, the skilled person would have to ‘seriously contemplate’ applying the teaching of the prior art document in the range of the overlap.
To the judge’s mind, that would mean the skilled person in the present case would not do so, because the changing of one constituent meant changing one or more of the other constituents in order to balance the equation (so to speak). This in turn would affect the mechanical properties of the glass fibre itself. So the skilled person would hesitate, if not avoid altogether, carrying out all of the variations in the ranges and so not fall within the claim of the patent.
The judge considered that Jushi and its expert did not address why a skilled person would seriously contemplate making a glass fibre having constituents all of which fell within the range in claim one and how the complex compositional juggling would have been mastered. As the judge points out, the burden is on those alleging it would succeed to prove the point.
On the basis that he found the patent valid, then it was also infringed. The outcome of this decision is likely to come as a surprise to many – and I am certain it did to Jushi, its expert and its legal team.
The numerical value Smith & Nephew approach, taking the degree of overlap with one example (save for the ratio and composition criteria) into consideration, both of which fell squarely within the claim, would seem an almost shoe-in for invalidity on anticipation, if not obviousness. Yet the judge has seemingly gone the other way.
I have a feeling this case will be appealed, and in my view, rightly so. I say this because while the judge has acknowledged the numerical value approach, he seems singularly to have ignored it when construing claim one of the patent and the prior art. He has merely said that two of the values (he did not say which two) would not be subject to the approach, but then goes no further. Had he done so, there may not have been any overlap at all, since 50% would equally be 45% or 55%, for example.
It is an odd feeling to read the end-result of a case one was initially involved in. As ever, one’s sympathies have to lie with the client who lost.