The EPO’s treatment of numerical subranges is undergoing a significant shift. Following T 1688/20 and the 2026 Guidelines update, novelty of numerical subranges is now assessed using the gold standard of direct and unambiguous disclosure, replacing the traditional “narrow and sufficiently far removed” criteria. This alignment with the standard used for other types of features has implications for applicants, patentees and opponents defending or attacking numerical subranges.

What are numerical subranges?

Some inventions are defined by numerical ranges. This is particularly common in chemistry, where ingredient concentrations may be expressed by a lower and upper limit, for example “from 2 % to 47 %”. Sometimes, however, the prior art discloses a broader range that fully encompasses the claimed range without disclosing the claimed endpoints, for example “from 1 % to 50 %”.

A recurring question is whether a claimed numerical subrange is novel over a broader prior-art range. For decades, numerical subranges were treated as a special case when assessing novelty.

Background: “narrow and sufficiently far removed”

In the 1980s, the Boards of Appeal of the EPO developed a three-step approach for assessing the novelty of a claimed subrange over a broader prior art range that fully encompassed it without explicitly disclosing its endpoints.[1] Under this approach, the claimed subrange was considered novel if it met three criteria: first, it had to be narrow compared to the known range; second, its endpoints had to be sufficiently far removed from the endpoints of the known range and from the examples of the prior art; and third, the selected area had to represent a purposive selection, i.e. another invention rather than an arbitrary embodiment of the prior art.

In more recent years, however, the Boards began to consider the question of “purposive selection” to be a matter of inventive step rather than novelty, because it concerned the presence of a technical effect of the claimed invention.[2] Following T 261/15, the third criterion was removed from the list of criteria set out in the Guidelines for Examination at the EPO.

Until 2025, in order to be considered novel, a subrange thus only had to be “narrow” compared to the prior art range and “sufficiently far removed” from any specific example disclosed in the prior art. The meaning of “narrow” and “sufficiently far removed” had to be determined on a case-by-case basis. However, T 261/15 did not completely remove the skilled person’s judgment from this assessment: until 2025, the Guidelines still stated that “it must be assessed whether the skilled person, in the light of the teaching of the prior art, would seriously contemplate working in the selected subrange”.[3]

At the same time, novelty of other claimed features is, in principle, assessed based on the “gold standard”. Subject-matter is disclosed if, from the standpoint of the skilled person, it is directly and unambiguously derivable from the prior art. According to the Boards of Appeal, the gold standard must be applied uniformly when assessing novelty, the validity of the priority claim (i.e. whether the priority application relates to the same invention as the subsequent application) and added matter (i.e. whether the claimed invention is directly and unambiguously derivable from the content of the application as filed).[4]

A question then emerged: does the “narrow and sufficiently far removed” approach, under which a subrange may lack novelty without an explicit disclosure of its endpoints, align with the “directly and unambiguously derivable” standard?

T 1688/20 and the recent update of the Guidelines

In T 1688/20, the Board of Appeal answered this question in the negative. According to the Board, “the relative terms ‘narrow’ and ‘sufficiently far removed’” do not “provide objective, solid and consistent criteria for establishing novelty of a selected sub-range”. The terms are “generally open to such a broad interpretation that the decision whether criteria (a) and (b) are met not only depends on the factual circumstances of each case, but could also depend on the subjective perception of the deciding body on which values are to be considered ‘narrow’ or ‘sufficiently far removed’”.[5] The Board held that the gold standard should also be applied to subranges. In other words, for a claimed subrange to lack novelty, it must be directly and unambiguously derivable from the range disclosed in the prior art.

This decision had sufficient impact on European practice that the 2026 update to the Guidelines for Examination replaced the previous “narrow and sufficiently far removed” criterion with a reference to the gold standard.[6] Several subsequent decisions have already followed the reasoning outlined in T 1688/20,[7] suggesting that the age of special treatment for numerical subranges is over.

What does this mean for applicants, patentees and opponents?

To the extent that a numerical subrange was previously considered to lack novelty in view of a broader prior art range that did not disclose its endpoints, the alignment of the assessment of novelty with the gold standard has important consequences.

A subrange is novel if neither its endpoints nor values within the subrange are disclosed. This may render such subranges more difficult for opponents to attack.

The other side of the coin, however, is that applicants and patentees may face challenges when it comes to amendments: an amendment extracting a subrange from a broader range may be difficult to defend under the already strict EPO standards for added matter if the subrange is not explicitly disclosed in the application as filed. A similar challenge may apply to the validity of priority.

Inventive step may also become more central.[8] A subrange that previously was not considered novel may now become the sole distinguishing feature. Patentees and applicants may therefore face the challenge of showing a technical effect linked to the selection of specific endpoints, as opposed to the remaining part of the broader range disclosed in the prior art.

Applicants should keep this in mind when drafting new applications. If the claimed range is central to the invention, it may be useful to collect experimental data showing a technical effect compared with values outside the claimed range but close to its endpoints. Such data may prove helpful if prior art later emerges disclosing a broader range. To preserve amendment flexibility, it is also advisable to include explicit basis for further subranges in the description.

 

[1] T 198/84, Headnote, and T 279/89, Reasons 4.2.

[2] T 261/15, Reasons 2.2.2.

[3] Guidelines for Examination at the EPO, April 2025 Ed., G-VI, 7.

[4] G 2/10, Reasons 4.6.

[5] T 1688/20, Reasons 3.2.1.

[6] Guidelines for Examination at the EPO, April 2026 Ed., G-VI, 7.

[7] T 667/23, T 1132/22, T 989/22, T 377/22.

[8] This applies where the prior art reference is full prior art under Article 54(2) EPC, rather than prior art under Article 54(3) EPC, which is not available for inventive step.

This article first appeared in the Hoffmann Eitle Quarterly, June 2026, and is used by permission of the author.

---