Treat Paper B preparation as drafting under constraint: write a construction note for each claim, identify the type of objection (novelty, inventive step, sufficiency, clarity, excluded matter, admissibility), then draft primary and auxiliary amendments filtered through the added-matter, support, clarity and admissibility gates, checking each added feature against the filed disclosure rather than against your memory of it.
Build a construction note before you amend anything
Every amendment decision depends on what the claims actually cover, not what you assume they cover. A written construction note for each independent claim pins down term meanings against the description, the embodiments and the cited prior art before any drafting starts.
A construction note records, for each claim term and feature, where it appears in the description, whether it appears in all embodiments or only some, and how a rival could read it. Terms such as 'connected', 'fastening element' or 'substantially' each carry a broad and a narrow reading; a novelty argument that survives the narrow reading may fail the broad one, because an indirect linkage can still fall within 'connected' if the description supports that reading.
Construction also exposes what an amendment silently gives away. If a feature you plan to add appears in only one of three disclosed embodiments, narrowing claim 1 to include it abandons the other embodiments by implication. Check the client instruction: does it tolerate losing that scope, or do you need an amendment sourced from a passage that covers the broader combination? Writing this down before drafting turns a vague unease about scope into a checkable decision.
Separate novelty attacks from inventive step attacks in your skeleton
Novelty asks whether one prior-art document discloses every claim feature directly and unambiguously; inventive step asks whether the skilled person would modify the closest prior art to arrive at the claim. The two require differently structured briefs and differently designed amendments.
For a novelty objection, work feature by feature against a single reference: list each claim feature, mark where it appears in the document, and note whether it is explicitly stated or only implicitly and necessarily present. A feature the reference merely enables or hints at is not, for novelty purposes, the same as a feature it discloses. Whatever remains unmarked is your distinguishing feature, and it drives everything downstream.
An inventive step objection needs the problem-solution structure instead: identify the closest prior art, state the distinguishing features, derive the objective technical problem from the technical effect those features produce, and then assess whether the prior art as a whole would have prompted the skilled person to solve that problem that way. Drafting consequences differ sharply: a novelty response either distinguishes a feature or amends the claim to include one; an inventive step response lives or dies on the effect and the problem formulation, so amendment alone rarely fixes it.
Added matter: the disclosure test that constrains every amendment
An amendment may not extend beyond the disclosure as filed. Every feature you add must be directly and unambiguously derivable, in the context of the whole filed disclosure, for the specific claim you are amending - not merely mentioned somewhere in the description.
Worked scenario: claim 1 to a dispenser comprising a nozzle and a valve controlling flow. The client, facing a novelty objection, instructs you to add 'the valve is a spring-loaded diaphragm'. The description discloses a spring-loaded diaphragm valve only in Example 3; Examples 1 and 2 use a ball valve, and the general passage says the valve 'may be any suitable flow-control valve'. The plausible mistake is to draft the amendment and argue that the feature is plainly disclosed. The real question is whether the filed disclosure makes that feature available for the nozzle-and-valve combination of claim 1, or only as one option in one example. Generalising a single example into a claim limitation is the classic trap known as an intermediate generalization.
The better decision: search the description for a passage that ties the feature to the claimed combination itself, such as a sentence recommending preferred valves for the claimed dispenser. If no such passage exists, treat the amendment as high added-matter risk and look for alternatives - arguing that the prior art lacks a feature the claim already implicitly requires, or amending from a passage that genuinely combines the features. Why it matters: an amendment that adds matter hands your opponent a second, independent attack on the request, so the 'fix' creates a new problem while leaving the original one live.
Choosing and defending a closest prior art
The closest prior art is the document that makes the most promising starting point for reaching the invention - typically the one sharing the most relevant features and purpose with the claim - not simply the document with the most textual overlap or the most shared keywords.
Worked scenario: a claim to a battery pack with a cooling plate between cells and a sensor that adjusts fan speed from cell temperature. D1 discloses a battery pack with a cooling plate and a fixed-speed fan. D2 discloses a general ventilation unit with a temperature-responsive fan but no battery pack and no cooling plate. The plausible mistake is to pick D2 as closest because it discloses the sensor-fan feature the claim rests on. The better decision is D1: it shares the field, the product type and the cooling architecture, so it forms a realistic baseline from which the sensor-based adjustment is the improvement to be assessed.
Why it matters: the closest prior art determines the objective technical problem. Starting from D1, the problem can be framed from the effect - thermal management that responds to actual cell temperature - and D2 then serves as evidence that temperature-responsive fan control was known in another context, which cuts both ways and must be argued, not assumed. Two formulation errors to avoid: restating the objective technical problem as 'providing a sensor that adjusts fan speed', which embeds the solution into the problem, and never justifying the choice of starting point in a sentence. Force yourself to write that one-sentence justification every time; it is the fastest self-audit of whether your inventive-step chain is built on solid ground.
New clarity and sufficiency problems created by your own amendments
An amendment can create defects the original claim never had: relative terms, undefined alternatives or functional language can draw clarity objections, and a narrowed claim can cover variants the description cannot teach the skilled person to carry out.
On clarity, audit every term you introduce. Relative expressions such as 'about', 'substantially' or 'high', alternatives joined as 'A and/or B', and functional definitions can each attract an objection if the description does not give the skilled person a clear scope. Cross-check each amended term against the description: if the filed text never defines the relative term in context, you are importing an ambiguity while trying to cure an objection elsewhere.
On sufficiency, ask whether the disclosure supports the amended claim across its full scope. If you narrow to an embodiment, confirm the description teaches how to make that embodiment work - a claimed parameter range broader than what the examples and general teaching support is a standing weakness. If you draft for the opponent instead, sufficiency attacks should name concrete variants within the claim that the skilled person could not reproduce from the disclosure; a vague assertion that 'the invention cannot be carried out' without identified variants is easily brushed aside.
Excluded and unpatentable subject matter in the claim as a whole
Before relying on any amendment or argument, check whether the claim as a whole falls within excluded categories or lacks the required technical character - for European practice, mathematical methods, programs for computers and business methods are excluded only as such.
The assessment is made on the claim as a whole and in context. Discoveries, scientific theories, mathematical methods, presentations of information, programs for computers and schemes or methods of doing business are excluded when claimed as such. A claim mixing excluded and technical features is not automatically excluded; the question is whether, viewed as a whole, it has technical character or produces a technical effect relevant to the assessment.
This shapes amendment strategy directly. If the client's invention is essentially an algorithm, an amendment adding an abstract 'processing step' gains nothing; an amendment tying the algorithm to its technical application - the cell-temperature sensor driving the fan, for instance - gives the claim a technical anchor. When drafting against such a claim, target whether the technical features do more than transport the mathematical output into an ordinary technical setting, and argue that question explicitly rather than gesturing at the exclusion list.
Request structure, admissibility, and a timed self-check exercise
Every request must be complete, clear and admissible, and each auxiliary request must differ from the previous one for a stated reason. Practice assembling primary and auxiliary request sets with matching arguments, then run a timed construction-and-drafting exercise against a fixed rubric.
Structurally, a request is a full set of claims plus the argument that supports it. When drafting for a proprietor, the primary request answers the objection head-on and each auxiliary request is a genuine fallback, narrower or differently amended, that remains admissible in its own right. When drafting for an opponent, the brief must attack every pending request, not only the primary one - a fallback left unattacked is a request that can still succeed. Sequence your document so that construction, objection analysis, amendment and argument appear in that order, so a reader can verify each step against the last.
Exercise and self-check: pick any published EP application in a field you know, via Espacenet, and write yourself a one-line prompt such as 'an opponent cites D1 against claim 1 for want of novelty'. Timebox it: 30 minutes for the construction note, 60 minutes to draft one primary amendment and one auxiliary request with arguments. Then audit against the rubric: (1) every added feature is traceable to a filed passage that applies to the amended combination, not to a lone example; (2) the objective technical problem states an effect, not the distinguishing feature; (3) the closest prior art choice carries a one-sentence justification; (4) each auxiliary request differs for a stated reason; (5) no relative or undefined term was introduced without support. Expected observations on a first pass: at least one feature traced only to a single example, and a problem formulation that embeds the solution - rerun until both are clean. Your readiness checks are procedural, not predictive of any result: you can produce a construction note unaided, name the filed passage supporting each limitation, state the objective technical problem without the feature in it, and draft two genuinely distinct auxiliary requests inside the timebox.
- Rubric item 1 - added matter: each added feature cited to a filed passage that applies to the amended claim, with intermediate-generalization risk flagged where the source is one example among several.
- Rubric item 2 - inventive step: objective technical problem phrased as a task or effect; the distinguishing feature appears only as the solution.
- Rubric item 3 - closest prior art: a one-sentence justification naming why this document is the starting point.
- Rubric item 4 - request hygiene: every pending request either defended or attacked; each fallback differs for a stated reason.
- Rubric item 5 - clarity and support: no relative, undefined or functional term added without a description anchor.
| Amendment source | Typical benefit | Main risk | Check before using |
|---|---|---|---|
| Feature recited in the original claim | Lowest added-matter risk; combines existing features | May not answer the objection if the feature was already implicit | Confirm the new combination actually distinguishes over the prior art |
| Passage tying the feature to the claimed combination | Strong support for narrowing claim 1 | Scope given up must be acceptable to the client | Verify the passage covers the combination, not one embodiment only |
| Single example or embodiment | Often the only concrete technical detail available | Intermediate generalization - classic added-matter trap | Look for generalising language; otherwise treat as high risk |
| General statement listing options ('may be X or Y') | Shows the feature was contemplated as filed | Availability for the amended combination is still the question | Check whether the option is linked to the claimed combination anywhere |
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.