Study each subject as a three-step routine: elementize the claim, identify which statutory test the facts trigger, then apply the named conditions in order. Novelty fails when a single reference discloses every element; inventive step weighs combinations only after novelty survives. Amendments must be traceable, phrase by phrase, to the original specification. Infringement analysis runs literal scope first, then the five equivalents conditions and their limiting doctrines. Treaty questions turn on which effective date controls each disclosure. Rehearse all of this on paper hypotheticals, and keep a written error log of every decision you got wrong.
Novelty trap: when one disclosing reference makes inventive-step argument pointless
Anticipation in Korean practice asks a narrow question: does one prior reference disclose every claim element, arranged as claimed? Confirm this before any combination argument, because a combination argument cannot restore novelty that a single reference has already destroyed.
Anticipation analysis asks whether a single prior reference — a publication, a granted patent, or a public act — contains every element of the claim, arranged as the claim arranges them. Figures and embodiment descriptions count as disclosure, so read them before the body text. This is why elementizing the claim into a table comes first: you compare row by row, not impression against impression. Only when the table shows at least one element missing from the reference does the inventive-step discussion begin.
Worked scenario: a claim recites a solar module with a bypass diode across each cell string and a temperature sensor driving a relay. The examiner cites one reference describing exactly that arrangement. A plausible mistake is to file an argument combining that reference with a second one and stressing housing differences. The better decision: build the element table first; where disclosure is complete, move to a narrowing amendment lifted from the original description. Why it matters: the misframed argument burns the response cycle, while the amendment is the route that actually changes the outcome.
Three tests, three questions: separating 신규성, 진보성, and expanded prior art
Novelty asks whether one reference already contains the claim. Inventive step asks whether a skilled person could readily arrive at the claim. Expanded prior art under Article 29(2) screens publicly known, worked, or described items before combination reasoning starts.
The three tests differ in reference count, standard, and available defense. Novelty is binary and reference-specific; inventive step is evaluative and asks about the skilled person's path, including motives to combine and teaching-away signals. Article 29(2) is a screening step, not a license to cite general knowledge at large: the item must qualify as publicly known, publicly worked, or publicly described, and even then the combination still needs its own inventive-step justification. Use the table below as the order of questions to ask in every rejection analysis.
Keep Article 30, the exception to loss of novelty, conceptually separate: a qualifying disclosure by the applicant can restore novelty, but it does not remove the disclosure from expanded prior art, so an obviousness rejection may still stand. A recurring conflation is treating 'not literally disclosed' as 'inventive' — that leap skips the actual test. Name the test you are answering before you write a word, and your analysis will not drift between standards mid-paragraph.
| Test | Reference count | Question it asks | Typical defense when it fails |
|---|---|---|---|
| 신규성 (novelty, Art. 29(1)) | One reference or prior public act | Does a single prior disclosure contain every claim element as claimed? | Narrow amendment; Article 30 exception if the facts qualify |
| 진보성 (inventive step) | Usually two or more, or one plus common knowledge | Could a person of ordinary skill readily arrive at the claim from the references? | No motive to combine, teaching-away, or an unexpected effect |
| 확장된 선행기술 (Art. 29(2)) | Items publicly known, worked, or described | Is the invention merely a slight modification of what was already circulating? | Distinguish circulating art from what a skilled person would actually combine |
The new-matter bar: writing amendments the original specification can support
An amendment may not add subject matter beyond what the original specification, claims, and drawings disclosed. The safe technique is quoting limiting language from the detailed description and checking the amended claim stays inside the disclosed embodiments.
The new-matter (신규사항) test is not 'the idea appears somewhere in the application.' Ask whether the amendment reaches subject matter the original disclosure did not convey to a skilled reader: a feature that appears once, inside a single example, may support a narrow claim covering that example, while a feature appearing only in your head at amendment time supports nothing. Write amendments by copying the original wording, then reread the amended claim as a subset of what the description taught.
Worked scenario: a claim to a 'metal fastener' is amended to 'a stainless-steel fastener with a knurled head and a protective coating.' The description mentions a knurled stainless-steel example once. The mistake is bundling in the coating — invented at the desk that afternoon — because that phrase is new matter and can sink the amendment. The better decision: amend only with the description's own words and preserve the pre-amendment scope as a dependent claim, so the broader version can be restored if the examiner resists. Why it matters: overreaching amendments convert one rejection into a structural problem.
Office action, refusal, or trial: choosing the procedural door before drafting
Examination-stage responses, trial against refusal, correction trial, and invalidation trial differ in who may file, what amendments remain possible, and what the outcome is. Decide the route before writing, because the route limits the wording you may use.
Examination (심사) and trial (심판) are different institutions: the examiner examines the application, while the patent trial body hears trial-level disputes such as refusals, corrections, and invalidations. A trial against refusal responds to a refusal decision, not to an examiner's opinion — the opinion stage is where amendment and argument are cheapest and least restricted. Correction-trial and invalidation-stage amendments are confined to narrower categories, such as scope narrowing or curing plainly stated defects. Divisional applications follow their own timing windows, so place them on the timeline of your specific case rather than assuming they stay open.
Scenario: after two office actions flag the same ambiguity, a client instructs argument only. The mistake is following that instruction verbatim; when the examiner has twice identified a specific unclear limitation, an argument without amendment invites a third opinion and then a refusal. The better decision is a paired response: a narrowed independent claim drawing language from the description, plus an argument addressing the remaining grounds. Why it matters: once a refusal decision issues, free-form examiner responses are gone, and the remaining routes — a trial against refusal or a divisional filing within its statutory window — each carry their own timing limits and tighter amendment categories.
Equivalents checklist: five conditions and the timing mistake that flips conclusions
Claim interpretation precedes infringement: determine scope under Article 126 first. Where literal infringement fails, run the settled five-condition equivalents framework, then apply the limiting doctrines — prior-art restriction and prosecution-history bar — before concluding.
The five conditions drawn from settled Supreme Court precedent: the differing element does not change the task-solving principle; a person skilled in the art could substitute it; the substitution produces the same effect; the substitution is easy at the time of infringement; and the exchanged element is not an essential part. Alongside literal analysis, remember indirect infringement (Article 127) for supplying key parts suited to making the patented invention, and the burden-shifting presumption in product-by-process situations. Order the memo: interpretation, literal, equivalents, indirect.
Worked scenario: the accused module swaps the claimed relay for a transistor switch, and the memo concludes no literal infringement and stops there. The mistake is treating the equivalents checklist as optional decoration. The better decision runs the five conditions and then the prior-art restriction, which asks whether the substitution was obvious to a skilled person at the filing date — not at the infringement date. If transistor switching was conventional at filing, equivalents fail however similar the products are. Misplacing the timing of that one condition reverses the conclusion.
Paris priority versus PCT national phase: tracing dates on one timeline
Priority questions compare a later filing with an earlier one within the statutory priority period; PCT questions track one international application into national-phase entry. Practice both as timeline exercises where every disclosure receives an effective date.
Under the Paris Convention right of priority, the earlier filing date can serve as the effective date against disclosures that land in between. PCT procedure layers a different timeline onto the same facts: one international application, international search and publication, then entry into the designated national phases. The concepts interact — an international publication is itself a disclosure — so a timeline showing the applicant's own publications, the priority claim, and each national-phase entry is the fastest diagnostic tool for both question types.
Worked scenario: the applicant posts a technical paper after the priority date but before national-phase entry, and a memo treats the PCT publication date as controlling. The mistake is substituting one date for another: with a valid priority claim, the paper's effective date is the priority date; without one, the paper becomes intervening prior art against the later filing. The better decision is to trace each disclosure to the earliest date that actually counts for it. Drawing that timeline once per practice case makes the error visible immediately.
Ethics conflicts and the weekly case-file drill with a self-check rubric
Confidentiality and conflict obligations constrain which matters you may accept across prosecution, trials, and litigation. Rehearse them inside the same paper cases as patentability and infringement, so ethics becomes one paragraph of every memo you write.
Representative problems: taking an invalidation trial against a patent whose prosecution you know from earlier work for another client; disclosing a former client's technical information while drafting for a competitor; and recommending an amendment you cannot reconcile with statements already on the record. Frame each as a conflict-and-confidentiality question — identify the representations, the information held from each, and whether the new matter is materially adverse — rather than as a memorized list of prohibitions.
The weekly case-file drill ties all subjects together. Take one hypothetical invention and produce four pages: an element table against a given reference; one amendment with a source line for every phrase; an infringement memo ending with the equivalents checklist; and a conflict memo naming the representations. Expected observations: amendment sentences you cannot trace to the original text, equivalents conclusions written before the limiting doctrines, and conflict memos describing feelings instead of representations. Administrative note: the reference fetched for this guide carries no exam-administration details, so confirm schedule, eligibility, and fees in the current official notice.
- Weeks 1–3: read the core Patent Act provisions while elementizing one granted claim per day into a table.
- Weeks 4–6: patentability hypotheticals only; write the novelty analysis before any inventive-step argument.
- Weeks 7–8: drafting and amendment drills where every phrase cites a line of the original specification.
- Weeks 9–10: infringement memos with the equivalents checklist, plus priority and PCT timeline problems.
- Weeks 11–12: mixed case-file drills under time, then reweight the sequence toward your weakest subjects.
- Readiness check: you can state the three patentability tests' differing questions from memory.
- Readiness check: you can draft a compliant amendment in one sitting with every phrase traceable to the disclosure.
- Readiness check: in every memo, the equivalents conclusion is the final sentence after the limiting doctrines.
- Readiness check: you can map a priority and PCT timeline, assigning each disclosure an effective date, without pausing.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.