The principles applied by the boards of appeal as part of their established case law on the novelty of selection inventions were developed in particular in T 198/84 (OJ 1985, 209). They are summarised briefly in T 279/89, according to which a selection of a sub-range of numerical values from a broader range is new when each of the following criteria is satisfied:

(a) the selected sub-range should be narrow;

(b) the selected sub-range should be sufficiently far removed from the known range illustrated by means of examples;

(c) the selected area should not provide an arbitrary specimen from the prior art, i.e. not a mere embodiment of the prior description, but another invention (purposive selection).

The Guidelines recall the three criteria (see G‑VI, 8 (ii) – November 2018 version).

The three postulates for the novelty of a selected sub-range are based on the premise that novelty is an absolute concept. It is therefore not sufficient merely for the wording of the definition of an invention to be different. What has to be established in the examination as to novelty is whether the state of the art is such as to make the subject-matter of the invention available to the skilled person in a technical teaching (T 198/84, OJ 1985, 209; see also T 12/81, OJ 1982, 296; T 181/82, OJ 1984, 401; T 17/85, OJ 1986, 406).

With reference to the third criterion, the board in T 198/84 was of the opinion that this view of novelty really entailed more than just a formal delimitation vis-à-vis the state of the art. There would be delimitation only in respect of the wording of the definition of the invention, but not in respect of its content, if the selection were arbitrary, i.e. if the selected range only had the same properties and capabilities as the whole range, so that what had been selected was only an arbitrary specimen from the prior art. This was not the case if the effect of the selection, e.g. a substantial improvement in yield, occurred in all probability only within the selected range, but not over the whole known range (purposive selection). To prevent misunderstanding, the board emphasised, following T 12/81 (OJ 1982, 296), that a sub-range singled out of a larger range was new not by virtue of a newly discovered effect occurring within it, but had to be new per se. An effect of this kind was not therefore a prerequisite for novelty; in view of the technical disparity, however, it permitted the inference that what was involved was not an arbitrarily chosen specimen from the prior art, i.e. not a mere embodiment of the prior description, but another invention (purposive selection).

With reference to the third criterion (criterion "(c)"), several decisions consider that purposive selection is relevant for assessing inventive step but not novelty (see T 1233/05, T 1131/06, T 230/07, T 913/07, T 1130/09, T 2041/09, T 492/10, T 1948/10, T 423/12, T 378/12, T 1404/14, T 261/15, some of which are reported below). However see for example decision T 66/12, in which the board left it open as to whether the first two criteria were satisfied in light of its conclusion that the third criterion was not met (conclusion – lack of novelty). See also the detailed decision T 673/12, which recalled the three criteria for the examination of novelty.

In T 17/85 (OJ 1986, 406) the novelty of the claimed range was denied because the preferred numerical range in a citation in part anticipated the range claimed in the application. A claimed range could not be regarded as novel, at least in cases where the values in the examples given in the citation lay just outside the claimed range and taught the skilled person that it was possible to use the whole of this range.

In deciding the question of the novelty of an invention, the board in T 247/91 emphasised that consideration had to be given not only to the examples but also to whether the disclosure of a prior art document as a whole was such as to make available to the skilled person as a technical teaching the subject-matter for which protection was sought. The board stated that a skilled reader of the cited document had no reason to exclude the range of 85 to 115°C claimed in the patent in suit when carrying out the invention disclosed in the citation. The teaching of the cited document was clearly not limited to the use of the exemplified temperatures, but extended to the whole described temperature range of 80 to 170°C which had been made available to the skilled person as a technical teaching. The subject-matter of the patent in suit lacked novelty.

In T 406/94 the board found that the percentage range cited in the prior art, although numerically close to the claimed range, could not be adduced to anticipate the subject-matter claimed, because the percentage cited in the prior art was based on different starting materials.

In T 610/96 the patentee claimed a magnetoresistive material comprising magnetic and non‑magnetic metallic thin film layers. The board found that the claimed ranges defining the composition of these layers must be considered as a narrow selection of the generic disclosure of prior art document D10, which did not overlap with the sub‑ranges preferred in D10 and which further selected a specific non‑magnetic layer among a group of possible layers. This selection was also sufficiently far removed from the specific examples of D10. Furthermore, the claimed material showed different characteristics of the magnetoresistance change, so that the specific sub‑range was not simply an arbitrary part of the generic disclosure of D10, but was of a different nature and therefore novel. The criteria for selection inventions set out in T 279/89 were thus satisfied. In T 100/12 too, all three criteria were met.

In T 230/07 the board noted that novelty and inventive step are two distinct requirements for the patentability of an invention and therefore different criteria should apply for their assessment. So, the presence or absence of a technical effect within a sub-range of numerical values was not to be taken into account in the assessment of novelty. To establish novelty of a sub-range of numerical values from a broader range, the selected sub-range should be narrow and sufficiently far removed from the known broader range illustrated by means of examples. A sub-range is not rendered novel by virtue of a newly discovered effect occurring within it.

In T 1130/09 the "selection invention" principle had been applied in the contested decision for the purpose of assessing novelty, regard being had to the three criteria developed in T 198/84 (see OJ 1985, 209). The board observed that this principle was applicable where a narrow sub-range was selected from a broader range. The passage on page 9, lines 5 to 7, of document (2) disclosed that the dimensions of the structures were measured in nanometres or micrometres. Therefore, as had already been established in the contested decision, the range specifically claimed was a narrow selection which, in the absence of examples in document (2), had to be considered far removed from the central embodiments in that document. The board thus held that the first two of the criteria defined in T 198/84 were met. The third – that a technical effect of the narrower sub-range claimed had to be demonstrated – could not, however, be considered for the purpose assessing novelty, because novelty and inventive step were two distinct requirements for patentability. A technical effect within the more narrowly claimed range did not confer novelty on a numerical range which was already novel per se, but merely confirmed its already established novelty. Whether or not there was a technical effect nevertheless remained a matter of inventive step (many decisions have applied this approach; see T 1233/05, T 1131/06, T 230/07, T 913/07, T 2041/09, T 492/10, T 1948/10, T 423/12, T 378/12, T 1404/14 and T 261/15).

Indeed in T 378/12, concerning criterion (c), the board was of the view that purposive selection was relevant for assessing inventive step but not novelty. Consequently, since the claimed sub-range met the criteria (a) and (b), the board judged that the subject-matter of claim 1 was new.

In T 261/15 the board stated that in the case of an alloy composition, a disclosure of a combination of the different alloying elements within the claimed ranges is necessary for a finding of lack of novelty. Since the different alloying elements interact with each other to form precipitates and solid solutions, their content ranges were not to be considered in isolation but in combination. Regarding the limit values, the appellant (opponent) argued that these values were explicitly disclosed and that in order to recognise novelty of a sub-range selected from a broader numerical range of the prior art the selected sub-range had inter alia to be sufficiently far removed from the end-points of the known range (Guidelines G‑VI, 8 (ii) (b) – November 2018 version). The approach taken by the appellant was an exercise in cherry-picking within the disclosure of D1, which created a novel combination of features that was not disclosed in D1. In respect of the argument above – recited as a condition in the Guidelines for acknowledging novelty of a numerical selection – neither T 198/84 nor T 279/89 stipulates this condition. In the view of this board, the limit values of a known range, although explicitly disclosed, are not to be treated in the same way as the examples. And concerning criterion (c), which by contrast to the first two criteria (a) and (b), goes beyond the comparison of the invention defined in the claims with the disclosure of the prior art, because it considers the presence of an effect of the claimed invention, the board considered that whether the claimed selection was purposive or not was more a question of inventive step than of novelty. D1 was not novelty destroying.

In T 40/11 the board stated that it may be questioned whether the third criteria mentioned in T 198/84 (see above) has to be considered when assessing novelty.