I. A brand changes hands

1.   In late 2019, Novartis AG, a Swiss pharmaceutical company, entered into deeds styled as deeds of assignment with three Indian companies, under which the Indian registrations of three product marks, each a brand known to Indian physicians by a single name, were to pass to the purchasers for fixed sums. The name remained on the carton. The trade dress, the manufacturer named on the pack and the entry in the Trade Marks Registry changed. But the deeds also confined the purchasers to India and to specified preparations, barred exports, required the name and the composition of the medicines to remain unchanged, and bound the purchasers to protect the brands' reputation abroad. Was that a sale of the marks, or a right to use them dressed as a sale? The Revenue took the latter view and treated the consideration of about Rs. 433 crore as royalty. The ITAT decision of 12 October 2026 in Novartis AG's case [TS-1671-ITAT-2026 (Mum)] held that the receipts were gains from alienation, taxable only in Switzerland under Article 13(6) of the India-Switzerland treaty.

2.  Every cross-border transfer of an intangible poses three questions, in sequence. What was sold: ownership, or a right to use? How was it paid for: a fixed price, or one that moves with exploitation? And where does the gain go? That is for the residuary paragraph of the capital gains article of the particular treaty to decide. The order answered the first and the third. The India-Switzerland treaty has something unusual to say about the second.

3.  That second question is one the order did not need to answer, and it brings me to the understanding on Article 12 in the Protocol to the treaty. The order cites it, but not as the ground of decision. This article first notes the role the Protocol actually played in the order. It then goes beyond the order: it traces the Protocol to its 1994 text, sets out three ways in which it can be read, evaluates each, and draws some lessons for transactions of this kind. That inquiry is offered as commentary, not as a reading of the order. Statutory references are to the Income Tax Act, 1961, which governs the year in dispute; corresponding provisions of the Income Tax Act, 2025 are noted where relevant.

II. The fault line between use and ownership

4.  The royalty articles of the OECD and UN Models speak of payments "for the use of, or the right to use" the listed property. The words presuppose that the owner keeps the property and parts with something less, and the better reading is that a payment for the property itself falls outside them. But that reading long remained largely implicit. Before 2008, the OECD Commentary stated it only in its paragraphs on software, and there in wider terms than today: where the ownership of rights in a copyright had been "alienated in full or in part", the consideration could not be for their use. The 2008 update deleted the words "in full or in part", added a test of "distinct and specific property", and stated the principle, in a new paragraph 8.2, for all the property the definition covers. Treaty practice was not uniform either: the United States brought contingent alienation gains within royalties, and some treaties, as will be seen, brought every alienation within them. When the India-Switzerland treaty was signed in 1994, the line between use and ownership was therefore a matter of interpretation rather than of settled text.

5. Even on the better reading, the line is hard to draw for an intangible. A trademark or a patent is a bundle of legal permissions, divisible by country, field of use, duration and mode of exploitation. When the whole bundle for one country is transferred outright, the transferee owns something, but so does the transferor. Is that the sale of a distinct asset, or an exhaustive licence of part of a larger one?

6. Paragraph 8.2 of the OECD Commentary on Article 12, introduced in that 2008 update, acknowledges the difficulty where "all rights to the property in a limited geographical area" are transferred "in a transaction structured as a sale". Its guidance is that each case turns on its facts, examined "in the light of the national intellectual property law applicable to the relevant type of property and the national law rules as regards what constitutes an alienation", but that a payment for "the alienation of rights that constitute distinct and specific property (which is more likely in the case of geographically-limited than time-limited rights)" is likely to fall under Article 7 or Article 13 rather than Article 12. It adds, in a sentence carried over from the software paragraphs, that "the essential character of the transaction as an alienation cannot be altered by the form of the consideration, the payment of the consideration in instalments or, in the view of most countries, by the fact that the payments are related to a contingency".

7. The Commentary thus lays down no rule on territorial transfers, only an inclination: it leans towards treating a transfer limited in space as an alienation. On contingent consideration it is firmer, but only "in the view of most countries", a qualification which acknowledges that others, as the American practice discussed below illustrates, take a different view. Nor was the new paragraph accepted by every member. Spain and Italy recorded that they do not adhere to it, holding that a transfer of less than full ownership falls within Article 12; Mexico recorded that rights presented as an alienation, whether geographically limited or time limited, fall within Article 12 because less than the full rights in the property are transferred. These qualifications, the reference to "most countries" and the three dissenting members, are taken as read in what follows. All of this is guidance, and for India, which is not a member of the OECD, persuasive rather than binding. The stakes are considerable. Royalty is taxed at source on the gross amount, at 10 per cent under the Swiss treaty. A gain from alienation is taxed, depending on the treaty, by the residence State alone or under the source State's domestic law.

III. What was sold: where the line falls

8.  Almost no alienation is unrestricted. Land is sold subject to restrictive covenants; a business is sold with a covenant not to compete. The line between alienation and use therefore cannot be drawn between restricted and unrestricted transfers. It is better drawn by asking what the transferor still owns, in the field it has transferred, once the transaction is complete. Restrictions as such do not prevent an alienation; retained ownership does, and some restrictions, by their nature or effect, amount to it.

9.  Several markers help. The clearest is reversion. If the rights return to the transferor at the end of a term, on termination, at its option, or on the happening or non-happening of an event, the transferor never parted with the property; a call option to buy back at a fixed or formula price is an economic reversion to the same effect. A reversion that only secures an unpaid price stands differently: it protects the sale rather than qualifying it, and once the price is paid it falls away. Next is the old distinction between a condition whose breach undoes the transfer and a covenant whose breach yields damages or an injunction. Breach of covenant leaves ownership where it is.

10.  A further marker, often decisive, is whose property the covenant protects. Property law has long accepted that a seller of one plot who keeps the adjoining plot may bind the buyer to covenants that preserve the value of what the seller keeps, without the sale ceasing to be a sale. Brand-protection undertakings in a country-specific assignment work the same way: the assignor keeps the mark elsewhere, and the covenants protect that retained property. Quality control points in the same direction. A licensor must control the quality of goods sold under its mark, because the mark remains its own. An assignor has no such need, and quality covenants in an assignment must find their explanation elsewhere, usually in the spillover of reputation onto the marks the assignor keeps.

11.  The remaining markers are more familiar. Is the transferor itself excluded from the field? Is the transfer for the full life of the right, including renewals, or for a term? Who is recorded as proprietor, renews the registration and sues infringers? And who bears the risk that the asset loses value?

Figure 1: Markers of alienation and use (graphic prepared with Gemini Notebook)

12.  The Tribunal's approach in Novartis AG supra tracked these markers. It read the deeds as a whole, noting the vesting of the marks absolutely, the bar on further use by the assignor in India, the change of proprietor in the Registry, and stamp duty paid as on a conveyance, with the packaging before and after the assignment as corroboration. It treated the restrictions on products, exports, trade dress and reputation abroad as covenants protecting the agreed territorial and product boundaries, enforceable by specific performance and injunction, rather than as a reservation of proprietorship. So far as the order records, the deeds contained no right of reversion, buy-back or termination, so that a breach would give rise to those remedies rather than to forfeiture.

13.  The Revenue's case put the question at its sharpest, and deserves a fair statement. The Panel accepted that word marks had been assigned but held that the brands had not, relying on the restrictions already described and on earlier co-marketing arrangements for the products. The Tribunal's answer was that a word mark is itself property capable of assignment, that sections 37, 38 and 45 of the Trade Marks Act, 1999 contemplate assignment for some only of the registered goods, and that the sale of a mark does not require the sale of the seller's house marks or trade dress. On deeds that vested proprietorship absolutely, that answer is persuasive. The obligations to keep the name and the composition unchanged come closest to a licensor's control, and the order addresses them within its general treatment of the covenants rather than separately. They are explicable, on the reasoning above, as protecting the reputation of the marks the seller keeps abroad; whether that explanation holds in a given case is a question of fact.

14.  The co-marketing arrangements are recorded in the order without their terms; the assessee described them as distinct transactions involving Indian group entities, without the order naming those entities. They are context rather than proof. But the conversion of an existing commercial relationship into a sale is where the substance of a transfer will most often be questioned, and where the position of any group entity whose arrangements come to an end may deserve attention.

15.  The Revenue's distinction will not always be so easy to meet. Where what the seller keeps, whether the global reputation, the formulation, the dossier or a supply relationship, is what gives the Indian mark its value, the question of what was sold shades into the question of how the price is to be allocated across the bundle, including any part that is "information concerning industrial, commercial or scientific experience" within Article 12(3).

16.  What was sold was thus the decisive inquiry in Novartis AG supra. Once ownership was found to have passed, Article 12(3) did not apply on its own terms, and Article 13(6) gave the gain to Switzerland.

IV. The Protocol in the order

17.  The understanding with reference to Article 12 was paragraph 4 of the Protocol signed with the Agreement on 2 November 1994. It became paragraph 6 when the Protocol of 2000 renumbered its provisions, which is how the order refers to it, and paragraph 7 after the Protocol of 2010. Its text has not changed:

It is understood that gains derived from the alienation of a right or a property mentioned in paragraph 3 of Article 12 may be taxed according to Article 7 or Article 13. However, gains derived from the alienation of any such right or property which are contingent on the profits, productivity or use thereof may be taxed according to Article 12.

18.  As the arguments are recorded, the assessee relied on Article 13(6), on "the relevant Protocol" and on paragraph 8.2 of the Commentary, all in support of the distinction between ownership and use; the contingency limb of the Protocol is not recorded as having been argued. The Revenue's case, as recorded, did not engage with the Protocol at all; it rested on the restrictions in the deeds, the earlier co-marketing arrangements and the width of the domestic definition. An order records submissions in summary, as it must, and written submissions and the hearing itself may well have gone further. The observations that follow are confined to what the order records, and nothing in them suggests that anyone overlooked any point.

19.  The Protocol nonetheless finds a place in the Tribunal's own reasoning, at two points. The first is at the outset of the analysis, before the deeds are examined. The Tribunal reads the Protocol as showing that the treaty itself contemplates the alienation of intellectual property as distinct from royalty, and as requiring examination of the terms on which the consideration becomes payable, and adds at once that neither a lump sum nor a geographical limitation is, by itself, determinative. The second comes after the finding that ownership passed. The Tribunal notes that the fixed consideration was not linked to the purchasers' turnover, profits, productivity or extent of use, that no such contingency had been identified by the Assessing Officer or the Dispute Resolution Panel, and that the Protocol's exception therefore did not support treatment of the receipts as royalty. Its conclusion is that the treaty distinction, read with the Protocol, is sufficient to reject the characterisation adopted below.

20. The Protocol thus played a confirmatory role. Its first sentence reinforced a distinction that Article 12(3) already drew; its contingency limb applied only in the negative, confirming that it was not engaged. The same result would have followed from Article 12(3) and Article 13(6) alone. What the Protocol would do in a case where the price was contingent, and how it should be read, did not arise. From here, the article goes beyond what the order decided.

Figure 2: What was sold, how it was paid for, and where the gain goes

V. An American idea, recast in 1994

21. The contingency test is American in origin. Sections 871(a)(1)(D) and 881(a)(4) of the Internal Revenue Code tax a non-resident's gain from the sale of a patent or similar property to the extent the payments are contingent on its productivity, use or disposition: a seller paid by reference to what the buyer earns has, in substance, kept an interest in its exploitation. The formula entered India's Convention with the United States of 1989 inside the royalty definition itself, as "gains derived from the alienation of any such right or property which are contingent on the productivity, use, or disposition thereof". The same language appears in India's treaty with Canada, signed in January 1996, and the treaty with Mexico of 2007 reaches the same result through a Protocol clause which, "notwithstanding the provisions of Article 13", deems contingent alienation payments to be royalties, although Article 13(6) of that treaty otherwise gives residuary gains to the residence State alone.

22.  The 1994 text of the Swiss treaty repays close reading. Its original Article 12, titled "Royalties and fees for included services", follows the India-US Convention closely: the same rates of 15 and 20 per cent for the first five years and 15 per cent thereafter, 10 per cent for equipment royalties, and the same "make available" definition of fees for included services. The royalty definition follows the India-US wording almost word for word, with one striking difference. The clause bringing contingent alienation gains within royalties is not there.

23. It reappears, recast, in the Protocol signed the same day. Four changes stand out. The contingent gains are taken out of the definition of royalties. They are no longer said to be royalties at all, but gains that "may be taxed according to Article 12". A new sentence is placed in front of them, routing gains from the alienation of royalty-type property to Article 7 or Article 13. And the contingencies themselves are rewritten: "disposition" is dropped and "profits" added. A text that closely tracks a known model and departs from it on precisely this point suggests a deliberate choice, although, since the Federal Council's message on the Agreement does not explain the change, that remains an inference.

24. Treaty practice was not uniform on the underlying question, and the closest evidence is India's own. India's Agreement with Singapore, signed on 24 January 1994, some ten months before the Swiss treaty, defines royalties to include "gains derived from the alienation of any such right, property or information", with no contingency qualifier at all: every alienation of royalty-type property, fixed-price or not, falls within the royalty article. Swiss practice could go the same way; Switzerland's Convention with Thailand of 1996 defines royalties as payments "for the alienation or for the use of" the listed property. Including alienation within royalties was, in other words, negotiable currency. Compared with the India-US template, the Singapore text and the Thai treaty, the India-Switzerland Protocol leaves the source State with a narrow taxing right: India secured far less from Switzerland than it had obtained from Singapore earlier the same year, and less than Switzerland itself conceded to Thailand two years later.

Figure 3: The contingency formula, from the India-US Convention to the India-Switzerland Protocol (graphic prepared with Gemini Notebook)

VI. Three ways of reading the Protocol

25.  The Protocol's meaning turns on a prior question: what would Article 12(3) have covered without it? On that question there are three possible starting points, and each yields a different reading of the Protocol. Each deserves to be stated at its strongest before it is tested.

Reading A: an extension of Article 12

26.  On the first reading, "use or the right to use" excludes every alienation, whatever its price. The Protocol's first sentence is then merely declaratory. The second sentence does the real work: it speaks of "gains derived from the alienation", and so presupposes a genuine alienation, yet sends those gains to Article 12. On this view, it is a deeming rule that carries Article 12 beyond its own definition, much as the American clause carries the royalty definition beyond use. Its support lies in the words "gains derived from the alienation"; in the OECD Commentary, which treats a payment for the transfer of full ownership as outside Article 12 and, subject to the qualifications noted above, a contingent price as leaving an alienation intact; and in the argument from drafting practice: treaties that wanted contingent alienation gains in Article 12 said so expressly, which implies that without such words they would fall outside it.

27. Two consequences follow. Under a silent treaty, a genuine alienation, even with a contingent price, would fall outside Article 12 altogether. And under the Swiss treaty, the second sentence would be read literally, so that any contingency on profits, productivity or use, however loosely connected to exploitation, would bring the gain within Article 12.

Reading B: a boundary fixed by agreement

28. The second reading starts from a different place: that before the Protocol, the scope of Article 12 was simply uncertain. "Use or the right to use" could be read by legal form or by substance. Read by form, the question is simply whether title passed under intellectual property law: if it did, the payment is for alienation and falls outside Article 12. Read by substance, the question is whether the seller, whatever the deed says, still shares in the fruits of exploitation: if it does, it is in substance being paid for use. A seller who assigns a mark outright but is paid a percentage of every sale for ten years passes the first test and fails the second. On this reading, the Protocol neither widens nor narrows Article 12. It settles, by agreement, where the line between use and ownership falls, and it chooses the terms of payment as the test. A fixed price for an alienation lies outside Article 12; consideration measured by exploitation lies within it, whatever the instrument is called. The word "alienation" in the second sentence recognises the legal form; the contingency test decides the substance.

29.  Two consequences follow. The first is that the second sentence is read by its purpose rather than literally. Since its office is to mark the point at which a seller is in substance paid for use, the contingency that matters is one that reflects continuing participation in exploitation: a royalty-style earn-out, a share of profits, a payment per unit sold. A one-time payment due on regulatory approval is contingent in a loose sense, but it does not track exploitation, and so stays on the alienation side of the line. The second concerns treaties that say nothing on the point. There, the contingency has no allocating role of its own: a genuine alienation remains outside Article 12 however it is priced. But the terms of payment remain evidence on the prior question, whether there was an alienation at all. A seller paid entirely by a share of sales, indefinitely, looks very much like a licensor, and may be found to be one.

Reading C: an exclusion, with a proviso

30.  On the third reading, the baseline could be argued to be wide, and the Protocol was written to set that apprehension to rest. Absent a specific exclusion, "payments of any kind" for "the use of, or the right to use" property could reach consideration for acquiring the rights outright, since a buyer of intangible rights acquires, among other things, the right to use them. That was no idle argument in 1994. Since 1976, Explanation 2 to section 9(1)(vi) had treated the transfer of all or any rights in such property as royalty, subject to the carve-out for capital gains. India's Singapore treaty of the same year, and Switzerland's own treaty with Thailand two years later, show that alienation could be, and was, brought within royalties in terms.

31.  On this view, the first sentence is constitutive, at least as against the position India could then take; in Switzerland's own eyes, as its Federal Council's message, discussed below, shows, it was declaratory. It takes gains from alienation out of Article 12, where they might otherwise have been claimed, and sends them to Article 7 or 13. The second sentence, opening with "However", is a proviso: contingent alienation gains come back to Article 12. Every word does work, and the structure of rule and proviso is exactly what the text shows. Read against the drafting history, the sequence fits closely: India's template claimed contingent alienation gains as royalties; the text Switzerland agreed to accepts that claim only for the contingent part, states expressly that alienation gains otherwise go to Article 7 or 13, and declines to call even the contingent gains royalties. As a proviso, the second sentence would be construed strictly and confined to its terms.

Figure 4: Three readings of the Protocol (graphic prepared with Gemini Notebook)

Weighing the three readings

32.  Five considerations bear on the choice: the text, including the words "However" and "according to Article 12"; the principle that every term of a treaty should be given effect; the context in 1994, both domestic and in treaty practice; the comparative evidence and its limits; and the consequences each reading would produce.

33. Reading A fares worst. It makes the first sentence close to surplusage. It attributes to a residence State a widening of source taxation beyond a contested baseline. And it sits uneasily with the drafting history, which shows the inclusion of contingent alienation gains being taken out of the definition, not put into it. Its one textual point, the words "gains derived from the alienation", is equally consistent with the other readings. Its external support is also less secure than it appears. The Commentary's general statement came fourteen years after the treaty, and with the qualifications already noted. How far such Commentary governs the interpretation of an Indian treaty concluded in 1994 is not free from doubt.

34.  Reading C best explains why the Protocol was written. It gives every word work and matches the negotiating sequence. As a claim about the baseline, however, it must be confined. The more natural meaning of "use or the right to use" is something less than ownership, and that is how the Commentary has read it, for software before 2008 and generally since. The comparative evidence cuts both ways: an express inclusion of alienation in another treaty may make explicit what was implicit, or may extend a definition that would not otherwise reach it; and a party's treaties with third States illustrate rather than interpret. Nor can the domestic definition carry the argument further. Royalties are defined in the treaty, so Article 3(2) does not import the domestic meaning; Explanation 2 shows what a negotiator in 1994 had reason to fear, not what the treaty definition means. The Singapore text, indeed, shows India writing alienation into the definition when it wanted it there, which tells against an implicit inclusion elsewhere. Most importantly, if alienation were implicitly within Article 12, outright sales of intangibles might be royalty under India's many silent treaties, a conclusion at odds with the settled understanding and with the general proposition this article draws from Novartis AG supra. Reading C is therefore best held in a confined form: that in 1994 the inclusion of alienation within Article 12 was arguable, and the Protocol removed the doubt, not that alienation is royalty wherever such words are absent.

35.  Reading B is the safest account of the Protocol's legal effect. It does not require the baseline to be decided, and it yields sensible results. Its premise draws support from the Commentary itself, which even in 2008 could state its position on contingent prices only for most countries, and whose general statement not every member accepted. Its limitation is that, on its own, it does not explain why the parties thought the first sentence necessary; Reading C supplies that explanation.

36. The better view, I suggest, is a synthesis. Reading C explains why the Protocol was written: to settle a real doubt in the residence State's favour, while keeping exploitation-linked prices within the source State's reach. Reading B describes what it does: it fixes the line by agreement. Reading A is set aside. The Swiss Federal Council's message on the Agreement, dated 3 October 1994, bears this out as far as it goes. It explains paragraph 4 of the Protocol not under Article 12 but under Article 13, as a provision that "confirms" (bestätigt) that gains on the sale of patents, trademarks and know-how are taxable in the seller's State of residence unless there is a permanent establishment in the other State; where those gains are contingent on the profit, yield or use of the rights, they are taxed "according to the principles applicable to royalties" (nach den auf Lizenzgebühren anwendbaren Grundsätzen) (BBl 1994 V 225, at 230). The message records the effect Switzerland attributed to the clause, not the reasons for it, and as one party's explanation to its own legislature it is at most a supplementary means of interpretation. But it is consistent with the synthesis: Switzerland treated residence taxation as the rule and the contingent limb as an exception, and described the contingent gains as taxed like royalties, not as royalties. Its word "confirms" also suggests that Switzerland saw no doubt of its own to resolve; the doubt the Protocol removed was the one India's domestic law and treaty practice had created.

37.  The practical differences are narrower than the theory suggests. On all three readings, a fixed-price alienation falls in Article 7 or 13, and a royalty-style or profit-share earn-out in Article 12. At the edges they converge too. As a proviso under Reading C, the second sentence is construed strictly; under Reading B it is applied by its purpose; and even under Reading A, a deeming provision is ordinarily confined to the purpose for which it was enacted. On any of them, a milestone payable on regulatory approval is likely to stay outside Article 12 unless it tracks exploitation. The sharp divergence lies elsewhere: in what each reading implies for treaties that say nothing.

VII. Beyond the order: a shield, and also a sword

38.  Because the outcome in Novartis AG supra favoured the taxpayer, it is tempting to see the Protocol as a taxpayer's provision. It is more accurately one that cuts both ways, and which way depends partly on the reading adopted.

39.  As a shield, it places a fixed-price alienation outside Article 12 whatever the width of the domestic definition, and where a contingent element is caught, it caps the source State's tax at 10 per cent of the gross. On Reading C, that shield is the Protocol's main purpose.

40.  As a sword, it can operate against the taxpayer where a silent treaty might not. Consider a genuine alienation, with title passed, no reversion and no continuing control, but with part of the price payable as an earn-out on sales. Under the Swiss Protocol, the earn-out falls within Article 12, and India may tax it, even though the property has genuinely changed hands. Under a silent treaty that gives residuary gains to the residence State, such as the treaty with Mauritius, Readings A and B would treat a genuine alienation as an alienation however it is priced, so that the whole gain may go to the residence State. On those readings, a Swiss seller who structures an earn-out is worse placed, on the treaty's allocation, than a Mauritian seller who does the same. The Mauritian seller, however, now faces a principal purpose test, in force since 10 August 2026. The Swiss seller faces no principal purpose test, though the domestic anti-avoidance rule, and, after the decision of the Hon'ble Supreme Court in Tiger Global case [TS-38-SC-2026], the inquiry behind a residency certificate, apply to it as they do to the Mauritian seller. The addition of "profits" sharpens the point, and the word "right" means that the alienation of a carved-out right on contingent terms is caught as well.

41. The sword is also blunter than it first appears. "May be taxed according to Article 12" permits India to tax; it does not itself impose a charge. For a genuine alienation, the domestic definition of royalty excludes consideration chargeable as capital gains, so the earn-out would ordinarily be charged under the head "Capital gains", subject to the threshold question of whether an Indian-registered mark is a capital asset situated in India, with Article 12(2) capping the tax at 10 per cent of the gross. What the Protocol takes from the seller is the residence-only shelter of Article 13(6), not the domestic character of the receipt.

Figure 5: The Protocol as shield and as sword (graphic prepared with Gemini Notebook)

42.  On Reading C's baseline, the comparison with Mauritius turns around. If alienation was arguably within Article 12 absent express words, the Mauritian seller's earn-out, and conceivably the full price, could be claimed as royalty, and the Swiss seller, protected by the first sentence and capped at 10 per cent on the earn-out, is better placed. Against Singapore, no reading is needed. Its royalty definition includes gains from any alienation of the listed property, so a Singaporean seller is within Article 12 even on a fixed price, and the Swiss seller is better placed on every structure. Whether the Protocol is a sword or a shield for the earn-out seller therefore depends on the treaty it is compared with and, for silent treaties, on what one thinks Article 12 means. That is a question the Protocol cannot answer for other treaties, and it is a reason to read silent treaties with care.

43. Two further edges deserve notice. Article 12(2) caps tax at 10 per cent of the gross amount of the royalties, while the Protocol speaks of "gains" taxed according to Article 12, deliberately, it seems, not calling them royalties. For a self-generated mark with no cost, nothing turns on this; for acquired property, whether the cap applies to the gross contingent payment or to the gain net of cost is open. And where a price is blended, the Protocol does not say how to apportion it between its fixed and contingent parts, while the payer must decide at the moment of payment, before the economics have played out, whether a receipt falls within Article 12.

44.  The table draws these strands together. The Singapore pattern, under which every alienation gain is royalty, is omitted because its answer is the same in every row: Article 12, at the treaty rate.

VIII. Lessons for transactions of this kind

45.  Whatever reading one prefers, some lessons hold for any cross-border transfer of intangibles.

46.  Let the deed record a sale, and let conduct confirm it. An assignment should vest title absolutely, exclude the assignor from the field, and avoid any reversion or buy-back. Covenants should protect identifiable property the assignor keeps, rather than give it control over the transferred property. The Registry should be updated promptly, the deed stamped as a conveyance, and the change of ownership visible in the market. Drafting records a transfer; it does not create one.

47. Separate the bundle. Marks, trade dress, formulations, regulatory dossiers, know-how and supply arrangements are different things, and some may fall within Article 12 even if the marks do not. Separate agreements, or at least a reasoned allocation of the price, protect the alienated elements and identify those that are not. An allocation to goodwill now also matters to the buyer, since goodwill has not been depreciable since the Finance Act, 2021, whereas a trademark acquired outright remains a depreciable intangible asset.

48.  Treat the price structure as a treaty decision. Under the Swiss treaty, an earn-out on sales or profits brings that part of the price within Article 12, while a fixed price does not. Under the US and Canadian treaties, a contingent price is royalty by definition. Under silent treaties such as that with Mauritius, a contingent price invites the argument that the transaction is a licence, although the Commentary treats a contingency as leaving the character of an alienation unaltered; under the Singapore treaty, any alienation gain is royalty whatever the price structure. Where a deferred or conditional payment is commercially necessary, tying it to regulatory or other events rather than to exploitation keeps it closer to the fixed-price side of the line, provided the event is commercially real and not a proxy for sales. A milestone keyed to sales thresholds is an earn-out by another name.

49.  Read the residuary paragraph first. Whether any of this matters depends on where the residuary paragraph of the capital gains article sends the gain. Under treaties that leave gains to domestic law, India may tax the gain even if it is not royalty, provided the gain accrues or is deemed to accrue in India, which for a non-resident turns on whether an Indian-registered mark is a capital asset situated in India, a question the Novartis order left open. Where it does, the cost of a self-generated mark is deemed to be nil (section 55(2)(a); now section 90(3) of the 2025 Act), and a sale then mostly brings the tax forward rather than removing it.

50. Count the whole cost, not only India's share. A sale saves Indian tax only to the extent that withholding on royalty would not have been credited at home. For groups, the arithmetic can reverse: a sale to a related Indian enterprise ends a deductible royalty and keeps the brand's future returns in India.

51.  The payer carries the first risk. The Indian buyer must decide, when paying, whether the receipt is chargeable and at what rate. Certificates for lower or nil withholding, the seller's residency documentation and suitable indemnities are the ordinary protections, and they matter most where a price is blended or contingent.

52.  Know where anti-avoidance rules begin. The general anti-avoidance rule does not interfere with a taxpayer's choice of the method of implementing a transaction, as the CBDT has accepted (Circular No. 7/2017, Question 3), and a genuine sale to an unrelated buyer at a real commercial inflection point carries its own explanation. The rule comes into play at three points. The first is migration: an intangible moved, shortly before an Indian disposition, to a resident of a treaty partner that gives residuary gains to the residence State. Since Article 13 of the Swiss treaty carries neither a beneficial ownership requirement nor a conduit clause, and the treaty has no principal purpose test, that question falls principally to the domestic rule; and after Tiger Global supra, a tax residency certificate will not by itself foreclose it. The domestic rule applies even where the treaty is more beneficial (section 90(2A); now section 159(6) of the 2025 Act).

53. The second is disguised contingency, which the Protocol's contingency limb already polices as a specific rule, leaving the general rule less ground to occupy: its proper office is the arrangement the specific rule does not reach. The CBDT's own position is consistent with this; it accepts that general and specific rules coexist, and that where a treaty's limitation of benefits clause sufficiently addresses avoidance there is no occasion to invoke the general rule (Circular No. 7/2017, Questions 1 and 2). The third is a related-party conversion in which the seller continues in substance to control. In treaties that do carry a principal purpose test, Mauritius now among them, a parallel point arises through the test's proviso: where the treaty itself sends fixed-price alienations to Article 7 or 13, granting that treatment to a genuine alienation accords with the object and purpose of the provision. A successful defence under the test does not, in logic, exclude the domestic rule, but it is a material restraint on its use.

54.  For treaty negotiators, the Swiss text is a ready model, with room for refinement. It protects source taxation of exploitation-linked returns while ceding the taxation of genuine dispositions. A refined version might define "alienation", decide whether "disposition" belongs among the contingencies, say whether the cap applies to gross payments or to gains, and provide a rule for apportioning blended prices.

IX. What the price decides, and what it does not

55.  The Novartis order decided what was sold. Its lesson, of general application under any treaty, is that restrictions do not prevent an alienation, but retained ownership does: a geographically limited transfer, hedged with covenants that protect property the seller keeps elsewhere, can be the alienation of distinct property. The principle is general; its application is not. Whether a particular deed leaves ownership with the seller is a question of fact, turning on its terms and on the conduct of the parties, and the order's conclusion on the Novartis deeds does not decide the next one.

56.  The Swiss Protocol speaks to how the property was paid for, a question the order did not need to decide. Its 1994 text shows it to be a recast and narrowed version of an American formula, leaving India a far narrower taxing right than it had obtained from Singapore the same year: written, most probably, to settle a real doubt in the residence State's favour, and operating as an agreed boundary that keeps exploitation-linked consideration within Article 12 and fixed-price alienations outside it. Whether it is a shield or a sword for a particular seller depends on how the price is structured, and on what one thinks a silent treaty means.

57.  The brand is still on the shelf, under the same name. The deeds decided that there had been a sale. The Protocol tells us that, once a sale is established, the terms of payment decide where it is taxed. Characterisation and allocation are distinct steps, and the second is reached only once the first is answered.