Section 3(d) vs Section 3(k): Two Very Different Patent Bars
Section 3 of the Patents Act, 1970 lists things that are not treated as inventions at all. Two entries in that list cause more argument than the rest put together: clause (d) and clause (k). Because they sit in the same section and are raised in the same way by an examiner, people often lump them together as the general anti-patent clause. They are not the same thing, and the work needed to answer them is completely different.
Section 3(d) is about substances. It asks a chemistry question: is this already a known substance, and if the applicant has made a new form of it, does that form actually work better? Section 3(k) is about abstraction. It asks an engineering question: is the claimed contribution only a calculation, a commercial idea or a programme by itself, or does it change how a machine behaves?
The practical result is that a 3(d) objection is usually won or lost on data, and a 3(k) objection is usually won or lost on how the claims are written. Confusing the two leads applicants to file comparative test results against a software objection, or to redraft claim language when what the file really needed was a side-by-side efficacy comparison. This page sets out where each bar comes from, what triggers it, and how the two are answered.
Side by side
| Question | Section 3(d) | Section 3(k) |
|---|---|---|
| What it is | A bar on patenting new forms, new properties and new uses of things already known. | A bar on patenting abstract subject matter: mathematical methods, business methods, algorithms and computer programmes standing alone. |
| Fields where it usually bites | Pharmaceuticals, agrochemicals, polymers, food chemistry and other chemical arts. | Software, fintech, e-commerce, artificial intelligence, data processing and telecom. |
| Core question the examiner asks | Is the underlying substance already known, and if a new form is claimed, does it perform better? | Is the contribution only a mental, mathematical or commercial step, or does it change how a machine works? |
| The test that decides it | Enhancement of the known efficacy of the known substance. | A technical effect or technical contribution beyond the mere running of a programme. |
| Evidence that helps most | Comparative test data against the known form, present in the specification from the start. | A clear account of the hardware, the technical problem and the improvement in how the system operates. |
| Built-in deeming rule | Salts, esters, polymorphs, isomers, particle size and similar variants are treated as the same substance unless they differ significantly in efficacy. | The words per se narrow the bar to the programme by itself, but renaming an algorithm as a system does not escape it. |
| Claim that typically fails | A crystalline form of a molecule published years earlier, claimed with no comparison against the old form. | A method of scoring loan applications, run on an ordinary computer, with no change to how the computer works. |
| Claim that has a better chance | A new form supported by data showing it treats the condition better, not merely that it is easier to store. | A method that changes device behaviour, such as cutting memory use or improving signal handling, with the hardware and the effect described. |
| Drafting fix that matters most | Build the comparison into the specification when you file; efficacy data produced late is difficult to rely on. | Frame the invention as the solution to a technical problem in a system, not as a better commercial outcome. |
| Relationship with novelty | Can defeat a claim even when the new form is genuinely new and not obvious. | Can defeat a claim even when the algorithm is original and clever. |
| Where it is raised | Examination report, pre-grant and post-grant opposition, and revocation. | The same stages, but hearings turn mainly on claim wording rather than on laboratory results. |
| Most common misreading | That it bans all pharmaceutical patents in India. | That it bans all software patents in India. |
What Section 3(d) actually stops
Clause (d) targets a specific pattern. Someone owns or knows about a substance. Later, a new version of that same substance appears: a salt, a different crystal form, a purer isomer, a smaller particle size. The chemistry is real work, but the law asks whether the public gets anything more from it. If the new form does not work better than the form already known, it is not treated as an invention.
The clause also covers two other moves. Finding a new property of something already known is not an invention. Neither is finding a new use for a known substance. And using a known process, machine or apparatus in a new way is caught, unless that use produces a new product or uses at least one new reactant.
The word doing the heavy lifting is efficacy. Indian practice reads efficacy in the therapeutic sense for medicines: does the patient do better? Improvements that make life easier for the manufacturer, such as a longer shelf life or a form that is simpler to compress into a tablet, sit uneasily under this test. That is why comparative data, filed with the application, is the centre of a 3(d) answer.
What Section 3(k) actually stops
Clause (k) removes four things from the definition of an invention: a mathematical method, a business method, a computer programme per se, and an algorithm. The intention is to keep pure abstraction out of the patent system while leaving room for real engineering that happens to be carried out in software.
The words per se are the hinge. A programme written on a disc, claimed as a programme, is out. A claim to a machine that behaves differently because of the way it processes signals, manages memory or controls hardware may still be examined on merit. The Patent Office publishes guidelines for examining computer-related inventions, and examiners look for a technical effect: something that improves the working of the system rather than only improving a commercial or administrative result.
Business method objections are the harshest. If the substance of the claim is a way of doing trade, arranging finance, pricing, advertising or transacting, adding a server, a mobile app and a database will not usually rescue it. The test looks at what the invention contributes, not at the hardware listed around it.
Why the two are so often confused
Both clauses live in the same section, both are described in shorthand as the not an invention objections, and both are raised in the first examination report in similar language. Applicants who have only ever seen one of them assume the other works the same way.
There is also a shared feeling behind them. Each clause reflects a policy choice that a patent should be given for a real contribution, not for repackaging something the public already has. But the thing being repackaged is different. Under clause (d) it is a physical substance. Under clause (k) it is an idea.
How each objection is answered
A 3(d) reply is an evidence exercise. You identify the known substance the examiner is relying on, show what the new form does differently, and put numbers against it. The comparison must be like for like, and it must be to the known form rather than to some unrelated benchmark. Because Indian law limits what can be added to a specification after filing, the useful data has to be in the document from the beginning.
A 3(k) reply is a claim architecture exercise. You move the claim away from the calculation and towards the system, describe the technical problem that existed before, and state the effect on the machine. Sometimes the answer is that the claim as drafted really is a business method, and the honest course is to narrow it to the hardware-level improvement or to accept that the subject matter does not fit the patent system.
Neither reply is a formality. Under section 15 the Controller can refuse an application or require it to be amended, and both clauses can be raised again later by an opponent under section 25 or by a petitioner seeking revocation under section 64.
When both clauses appear in one file
Some applications attract both. A digital health tool that claims a known drug delivered through an app-controlled pump can face a 3(d) objection on the drug side and a 3(k) objection on the software side. Diagnostic and bioinformatics filings often sit in the same position.
When that happens, treat them as two separate arguments with two separate proofs. Answering one well does not help with the other, and a reply that blurs them tends to satisfy the examiner on neither.
Which one applies to you
Section 3(d)
- The examiner has cited an earlier disclosure of the same underlying molecule or compound.
- Your claim is to a salt, polymorph, isomer, ester, metabolite, particle size or similar variant.
- The invention is a new use or newly discovered property of something already on the market.
- The value of the invention is that it performs better, and you can measure that against the older form.
Section 3(k)
- The claim recites steps of calculation, ranking, scoring, matching or optimisation.
- The commercial pitch of the invention is about money, trade, logistics or user convenience.
- The only hardware in the claim is a generic computer, server, phone or cloud instance.
- The real contribution is in how a device operates, and that has not yet been written into the claims.
Where people go wrong
- Treating clause (d) as a ban on incremental pharmaceutical research. It is not a ban. It is a condition that the new form must work better than the one already known.
- Assuming that improved stability, easier handling or better bioavailability automatically counts as enhanced efficacy. Indian practice reads efficacy narrowly for medicines, and the connection to how well the product works has to be shown.
- Reading per se in clause (k) as meaning that any software claim wrapped in hardware is allowed. Examiners look at what the invention contributes, not at the list of components.
- Filing comparative data late. Because there are strict limits on adding new matter to a specification, evidence that was not there at filing is much harder to use.
- Believing that surviving examination ends the argument. Both clauses are grounds available in opposition under section 25 and in revocation under section 64.
One company, two objections, two answers
The following is a simplified illustration and not a real matter. Sundar Life Sciences Pvt Ltd in Hyderabad files an application covering a new crystalline form of a fever medicine that has been sold for twenty years, together with a phone app that tells the patient when to take the next dose. The first examination report raises two objections. The first says the crystalline form is only a new form of a known substance. The second says the dosing method is a business method dressed as software. The company answers the first by pointing to a trial table already in its specification, showing the new form reaching a useful blood level faster and lowering fever sooner than the old form. It answers the second by dropping the reminder claims and keeping only a claim to a pump whose control circuit adjusts flow using a sensor reading, describing the circuit and the measured reduction in dosing error. Two objections, two entirely different kinds of proof, in one file.
Simplified illustration only. Actual outcomes depend on the facts.
Questions people ask
Is Section 3(d) the same as an obviousness objection?
No. Obviousness asks whether a skilled person could have arrived at the invention from what was already known. Section 3(d) asks a different question: even if the new form is inventive, does it work better than the substance already known? A polymorph can be genuinely difficult to make and still fall under clause (d) if it gives no improvement in efficacy. That is why the two objections are answered separately. One is met with argument about the prior art. The other is met with comparative data.
Does Section 3(k) mean software cannot be patented in India?
It does not. Clause (k) excludes a computer programme per se, which means the programme standing on its own. Claims directed to a system or device that works better because of the way it processes information can still be examined on merit. What matters is a technical effect: a change in how the machine operates, described clearly in the specification. Claims whose real contribution is a commercial or administrative method are much harder, because adding a generic computer does not change what the invention actually contributes.
Can one application face both objections?
Yes, and it happens often in digital health, diagnostics, agritech and connected devices. A single application may claim a chemical entity on one side and a software-controlled process on the other. Each objection has to be answered on its own terms. Comparative efficacy data does nothing for a clause (k) objection, and rewriting claims into system language does nothing for a clause (d) objection. Treating them as one combined problem usually produces a reply that satisfies neither.
Which clause is raised more often?
It depends entirely on the field. Chemistry and pharmaceutical filings meet clause (d) regularly, because so much work in those areas builds on substances already disclosed. Software, fintech and platform filings meet clause (k) regularly, because their value is often expressed in commercial terms. Neither clause is unusual, and neither ends an application by itself. Both are objections that can be answered, narrowed or, where the subject matter genuinely does not fit, accepted.
Do these objections disappear once the patent is granted?
No. Section 3 grounds are available after grant as well. An opponent can raise them in a post-grant opposition under section 25(2), and a person interested can raise them in a revocation petition under section 64. Courts hearing infringement suits also consider them when validity is put in issue by way of counterclaim. A granted patent is a strong document, but it is not immune, and a weak answer during examination can resurface years later.
Facing a Section 3 objection on your application?
MYCrave Consultancy reviews the objection, the cited art and your claim set, and explains what a reply would need to establish.