In T 74/98 the board found no reason to round up figures obtained by converting prior art values originally expressed in a different unit of measurement. It was evident that any rounding-up following the conversion of these figures into molar amounts would affect the definition of those compositions. The board noted that it was also evident that the true meaning of a specific disclosure could not be influenced by the units chosen to express it and that the objection of lack of novelty relied on an ambiguity introduced artificially by the appellant.
In T 1186/05 the board drew a distinction between the case before it and the case before the board in T 74/98. In the latter case, the board had considered that rounding up to the next integer (thus matching the lower limit of the claimed "inventive" range) of a component's molar percentage having two decimal places, calculated by conversion from its weight proportion, was not justified because (i) this would lead to a broadening of the claimed range and (ii) reconversion of the rounded molar percentage to the corresponding weight proportion would also imply a modification of the latter, i.e. would alter the true meaning of this specific disclosure.
However, in the case before it, the board in T 1186/05 held that rounding up was required, in order to enable two density values to be compared, each one reflecting a "true" density value having three (or more) decimal places, but expressed to a different degree of accuracy, i.e. one having three and the other one having only two decimal places. Thus, the rounding exercise put the claimed and the prior art density values on the same level by using the mathematical rule existing for that purpose. This exercise had no impact on the density value as disclosed in the prior art document. For the definition of the density range the respondent had chosen to use only two decimal places. This implied that a comparison with the prior art identifying three decimal places could only be made if the prior art values were also reduced to two decimals, that was to say rounded. The board noted that the skilled person reading D1 would thus be obliged to round the disclosed value up to 0.89 for comparison (see also decision T 708/05 of 14 February 2007). The board could not accept the argument of the respondent (patentee) that the value 0.885 could also be rounded down to 0.88 and thus outside the claimed range. The mathematical conventions for rounding values wherein the last digit is 5 (or more) required that the value be increased, or rounded up, resulting in a value of 0.89 in the case at issue.
In T 234/09, the board observed that the claims defined the fibre composition in terms of numerical ranges of its constituent parts' percentage by weight, the appellant patent proprietor having chosen to express most of those ranges' limits as whole numbers. It found that, in the context of the patent in suit, the skilled person would appreciate that the whole numbers defining the limits of the ranges in claim 1 were not to be understood as whole numbers in the mathematical sense, and that they were therefore to be seen as limits deliberately expressed in a less precise form than the decimal numbers set out in the patent specification. Consequently, they also included the decimal numbers from which they resulted on application of the rules on rounding up numbers. It noted that the boards had already taken a similar approach in comparable cases (see e.g. T 1186/05). Two of the decisions cited by the appellant had concerned situations in which, by contrast with the specific circumstances of the case in hand, there had been no particular reason to round up a figure disclosed only implicitly in the prior art (see T 646/05, T 74/98). The board concluded that the percentage in the prior art, as rounded up for the purpose of comparison, could not be distinguished from the upper limit in claim 1 and that, therefore, the subjectmatter of that claim was not novel.
In T 871/08 prior art document D9 related to a method for hydrogen peroxide production by an anthraquinone method involving hydrogenation of a mixture of alkyl and tetrahydroalkyl anthraquinones dissolved in organic solvents. The respondent had argued that the claimed subject-matter was novel, because the value 2.996 was lower than the lower limit of 3:1 defined in claim 1 at issue. This argument was not accepted by the board. When comparing a value from the state of the art (the value "2.996") with those claimed (the range of values "from 3:1 to 9:1"), the state of the art value had to be given the same accuracy as the one claimed. In the case at issue, the values in the claims had been quoted without any digit after the comma, which meant that for comparison purposes, the value 2.996 had to be rounded up to 3, which thus fell into the range of the claimed values. This ruling is consistent with the boards' case law (see, in particular, T 1186/05, T 708/05).
In T 686/96 claim 1 related to a composition with a feature (iv) requiring a Perspex® abrasion value (PAV) in the range from about 12 to about 20 PAV. A prior art document disclosed in example 2 a composition having features (i) to (iii) of claim 1. With respect to novelty it had to be decided whether the known composition also had an abrasion value as required by feature (iv) of claim 1. The board established that the abrasion value of the known composition was somewhat below the lower value indicated in feature (iv) of the claim. Since the lower limit in the claim 1 was defined as "about 12", some interpretation was necessary. The board held that, when deciding on the novelty of the subject-matter of a claim, the broadest technically meaningful interpretation of a claim should be taken into account. In the board's view the scope of claim 1 was to be construed to mean that the indicated lower limit corresponded to the value disclosed in the prior art. Claim 1 was then considered to lack novelty.