जब मध्यस्थता अनुबंध उल्टा काम करे: रोगग्रस्त क्लॉज और भारतीय न्यायालयों की सीमा

An arbitration clause, amongst other things, is intended to remove uncertainty from the dispute resolution process agreed to be adopted by parties in respect of a defined subject-matter. Yet, some clauses achieve precisely the opposite. A provision intended to identify a swift and private forum for future disputes can become the source of a preliminary dispute of its own: Did the parties agree to arbitrate at all, did they agree only to consider arbitration later, did they choose arbitration or some other private decision-making process, and can the mechanism they drafted actually be operated?

The Supreme Court’s decision in South Delhi Municipal Corpn. v. SMS Ltd.1 is a striking illustration. Three concession agreements contained substantially similar dispute-resolution provisions. Different Benches of the Delhi High Court reached different conclusions on whether those provisions amounted to arbitration agreements. By the time the Supreme Court finally resolved the issue, the parties had spent close to a decade litigating over the mode of dispute resolution while the underlying commercial disputes remained unresolved.

Clauses of this kind are commonly described as “pathological arbitration clauses”. The expression is conventionally traced to Frédéric Eisemann2, who identified four functions of a workable arbitration clause, i.e., it should 1) produce mandatory consequences for the parties, 2) exclude merits adjudication by State courts at least until the arbitral process has run its course, 3) confer jurisdiction upon the arbitrators, and 4) permit a procedure capable of producing an effective and enforceable award. The terminology has endured because it captures a recurring commercial problem, but it can also obscure an important legal distinction. Not every pathology is fatal, and not every drafting defect engages the same judicial response.

Recent Indian decisions make that distinction increasingly visible. Some defects go to the very existence of consent to arbitrate. Others arise after consent is already clear and concern only the machinery by which arbitration is to be carried out. A third category concerns the juridical character of the chosen process itself: A mechanism may be binding and adjudicatory in some sense, yet still not be arbitration. The practical question is therefore not simply whether a clause is badly drafted. It is what the court must add, remove or assume before the clause can function as an arbitration agreement.

The present jurisprudence is best understood through that enquiry. The line that Indian Courts are drawing is between interpretation and reconstruction. Courts will ordinarily strive to give effect to an imperfectly expressed arbitral bargain. They will not, however, manufacture the bargain where the parties have not made one.

Section 7 and the anatomy of a pathological clause

Consent is the jurisdictional starting point

Section 7, Arbitration and Conciliation Act, 1996 (Arbitration Act) does not prescribe a model arbitration clause. An agreement may be contained in a contractual clause, a separate agreement, an exchange of communications or, in the circumstances contemplated by the statute, even an exchange of pleadings. What matters at the threshold is that the parties have agreed in writing to submit all or certain disputes arising from a defined legal relationship to arbitration.

The Supreme Court has long explained the substantive content of that requirement. In Bihar State Mineral Development Corpn. v. Encon Builders (I) (P) Ltd.3, the Supreme Court identified, among other essentials, an intention to submit present or future disputes to a private tribunal and an agreement to be bound by its decision. In K.K. Modi v. K.N. Modi4, the Supreme Court elaborated the distinction between arbitration and other decision-making processes by emphasising consent, binding adjudication, impartiality, equal fairness and an enforceable decision on a formulated dispute.

Why the curable-incurable divide is useful but incomplete

Commentary on pathological clauses often divides defects into “curable” and “incurable” categories. The distinction is useful, but it should not be treated as a taxonomy in which a particular drafting error always carries the same result. The same institutional misdescription may be harmless in one contract and fatal in another. If the contract points unmistakably to one arbitral institution, a court may be able to correct the name without changing the bargain. If several institutions or procedural systems are equally plausible, correction would require the court to make a choice which the parties themselves did not make.

Curability therefore depends on degree, not merely type. The closer the proposed judicial correction is to giving effect to an objectively ascertainable intention, the stronger the case for validation.

When arbitration is only a possibility

Using “may” instead of “shall” and creating optionality

The classic Indian authority is Wellington Associates Ltd. v. Kirit Mehta5. The dispute-resolution provision stated that disputes “may be referred to arbitration”, while another clause conferred exclusive jurisdiction on the courts at Bombay. The Supreme Court declined to read “may” as “shall”. Read in its contractual setting, the provision did not impose a mandatory obligation to arbitrate.

The lesson from Wellington Associates is sometimes reduced to a grammatical rule that “may” is permissive and “shall” is mandatory. That is too crude. The real enquiry is whether the contract, read as a whole, records a present commitment to arbitration. Permissive language becomes legally significant when it leaves arbitration as one of the available options rather than the agreed forum for adjudication.

The distinction became sharper in Jagdish Chander v. Ramesh Chander6. There the partnership deed contemplated that disputes would be mutually decided by the partners or referred to arbitration “if the parties so determine”. The Supreme Court held that the provision did not constitute an arbitration agreement because a further consensus was required after the dispute had arisen. The clause did not merely postpone commencement of arbitration, it postponed formation of the arbitration agreement itself.

This distinction is particularly important in multi-tier clauses. Parties are free to require negotiation, consultation, mediation or another preliminary step before arbitration. Such provisions regulate the time or conditions for commencement of an already agreed arbitral process. They become problematic when the last tier says, in substance, that arbitration will occur only if the parties later agree to it. A valid escalation clause may defer invocation. It should not defer consent.

When permissive language falls short of consent

The Supreme Court returned to the problem in Mahanadi Coalfields Ltd. v. IVRCL AMR Joint Venture7. The relevant provision was headed “Settlement of Disputes/Arbitration”, but its substantive terms provided first for settlement at the company level and, if differences persisted, contemplated recourse to court in the case of non-government parties. The Supreme Court held that the heading could not convert the operative text into an arbitration agreement. The decision reinforced a central proposition from Jagdish Chander8: Arbitral terminology is not decisive if the operative clause does not impose an obligation to arbitrate.

That reasoning was reaffirmed and developed in BGM & M-RPL-JMCT (JV) v. Eastern Coalfields Ltd.9. Clause 13 stated, for parties other than government agencies, that “the redressal of the dispute may be sought through the Arbitration and Conciliation Act, 1996 as amended by Amendment Act of 2015”. The Supreme Court rejected the argument that the provision created a unilateral option which, once exercised, bound the other party. The language did not disclose consensus ad idem to arbitrate and required a further agreement before a reference could be made.

The significance of BGM lies in its insistence that a court cannot start with a policy preference for arbitration and work backwards to consent. A provision that merely keeps arbitration available for future consideration is not an arbitration agreement.

What about the word “can”?

The most recent Supreme Court decision on this line of authority is Nagreeka Indcon Products (P) Ltd. v. Cargocare Logistics (India) (P) Ltd.10. Clause 25 of the bill of lading was captioned “arbitration” and provided, in substance, that differences or disputes “can be settled by arbitration” in India or at a mutually agreed place, with each party appointing an arbitrator. The Bombay High Court held that the clause did not impose a mandatory obligation to arbitrate. The Supreme Court affirmed that conclusion.

The Supreme Court treated the word “can”, in the context of the clause, as indicating possibility rather than obligation. It also found the appointment mechanism incomplete. More fundamentally, it held that the words used did not disclose a determination and obligation to arbitrate. Further agreement between the parties would be necessary before arbitration could take place.

Importantly, Nagreeka case11 does not establish that every clause containing the word “can” is invalid. The Supreme Court expressly distinguished the decision in the case of Nagreeka from the Delhi High Court decision in Panasonic India (P) Ltd. v. Shah Aircon12. In the case of Panasonic, “can” was used in relation to the ability of either party to invoke arbitration. The remainder of the clause provided a detailed arbitral framework, and the Delhi High Court concluded that once either party exercised the right of invocation, the reference was binding on the other. In other words, the option concerned who could trigger arbitration, not whether the parties would arbitrate at all.

The Supreme Court also distinguished Zhejiang Bonly Elevator Guide Rail Manufacture Co. Ltd. v. Jade Elevator Components13, where the dispute-resolution provision contemplated arbitration or court proceedings and the Supreme Court had given effect to the claimant’s election of arbitration. The case of Nagreeka, therefore, confirms that modal verbs cannot be interpreted in isolation. The decisive question remains whether the clause as a whole creates a present obligation to submit disputes to arbitration or merely preserves a future possibility of doing so.

When a binding dispute-resolution mechanism is still not arbitration

The existence of a binding private decision-making mechanism does not by itself establish arbitration. Commercial contracts routinely provide for expert determination, certification, valuation, internal review, dispute boards and departmental decisions. Some of those processes may produce outcomes described as “final” or “final and binding”. Their binding quality is relevant, but it is not conclusive of arbitral character.

The older cases such as K.K. Modi case14 drew this distinction by asking whether the agreed decision-maker was intended to determine substantive rights in a judicial and impartial manner and whether the process possessed the attributes of adjudication contemplated by the parties. The recent decision in South Delhi Municipal Corpn. places that enquiry at the centre of modern Indian law.

The three appeals in the case of South Delhi Municipal Corpn.15 concerned Article 20 of concession agreements for parking and commercial projects. The provisions required disputes to be referred to the Commissioner, Municipal Corporation of Delhi (MCD). The process was expressly captioned “Mediation by Commissioner”. In two agreements, the appointed officer could consider documents, interview persons and issue a decision stated to be final and binding. In another agreement, the officer could be appointed from within or outside the MCD.

After reviewing Section 7, Arbitration Act, the decisions in Encon Builders16, K.K. Modi and comparative statutory approaches, the Supreme Court identified three ingredients of a valid arbitration agreement: 1) a clear and mutual intention to arbitrate, 2) a binding adjudicatory process, and 3) compliance with general arbitral norms. Accordingly, the Supreme Court opined that Article 20 presented several features pointing away from arbitration. The process was described as mediation; the words “arbitration”, “arbitrator” and “Arbitration Act” were absent; the reference was to the Commissioner rather than an independent Arbitral Tribunal.

Most importantly, the Supreme Court examined the character of the process. It noted the one-sided control over appointment, absence of a genuinely independent decision-making structure and the limited adversarial safeguards. The provisions did not contemplate oral hearings, examination or cross-examination of witnesses, or a fuller adjudicatory process. Taken together, these features led the Supreme Court to conclude that Article 20 lacked the judicial element and neutrality necessary to characterise it as arbitration.

When the arbitration agreement exists but the machinery fails

Inartistic drafting is not the same as absent consent

The judicial approach changes materially once the agreement to arbitrate is clear. At that stage, the court is not being asked to supply consent. It is being asked to make effective a choice which the parties have already made.

In Visa International Ltd. v. Continental Resources (USA) Ltd.17, the clause provided that disputes which could not be settled amicably would be “finally settled in accordance with the Arbitration and Conciliation Act, 1996”. The drafting did not set out the appointment process or the wider procedural machinery. The Supreme Court nevertheless upheld the agreement because the intention to arbitrate was evident from the clause and surrounding material.

Similarly, in Enercon (India) Ltd. v. Enercon GmbH18, the agreement contemplated a three-member tribunal, but the appointment provision dealt expressly only with the two party-appointed arbitrators and omitted the step by which the third arbitrator would be chosen. The Supreme Court held that the unworkability was confined to the mechanics of appointment. Since the agreement to arbitrate was otherwise clear, the Supreme Court could supply the obvious missing step and make the clause workable. At the same time, it emphasised that interpretation could not become a rewriting of the contract.

Pricol Ltd. v. Johnson Controls Enterprise Ltd.19 provides another familiar example. The clause referred to the “rules of arbitration of Singapore Chamber of Commerce”, although the Singapore Chamber of Commerce was not an arbitral institution having its own rules for appointment of arbitrators. The Supreme Court adopted the most reasonable construction and treated the intended reference as one to the Singapore International Arbitration Centre. The defect institutional identification concerned, not the decision to arbitrate.

Read together, these cases disclose the operative limit. Courts may correct an obvious omission, resolve an inaccurate institutional description or give commercially sensible effect to incomplete machinery where the arbitral destination is unmistakable. They cannot choose that destination for the parties.

Invalid appointment machinery ordinarily does not destroy the agreement to arbitrate

The same distinction explains the treatment of defective appointment mechanisms. In TRF Ltd. v. Energo Engg. Projects Ltd.20, the Supreme Court held that a Managing Director who had become ineligible to act as an arbitrator could not nominate another arbitrator. The decision invalidated the agreed route of appointment without treating the parties underlying commitment to arbitration as having disappeared. The principle was extended in the context of unilateral appointment of a sole arbitrator in Perkins Eastman Architects DPC v. HSCC (India) Ltd.21. A party interested in the outcome could not retain exclusive control over the identity of the sole arbitrator. The statutory appointment jurisdiction could therefore be used to constitute an independent tribunal notwithstanding the defective contractual mechanism.

The five-Judge Constitution Bench in Central Organisation for Railway Electrification v. ECI SPIC SMO MCML (JV)22, disapproving the contrary approach taken in the earlier three-Judge decision in the same matter, has since placed the principle on a broader footing. It reaffirmed that unilateral appointment of a sole arbitrator is impermissible and held that, in the constitution of a three-member tribunal, requiring one party to select its arbitrator from a panel unilaterally curated by the other is inconsistent with the principle of equal treatment. The Supreme Court clarified that PSUs may maintain panels of potential arbitrators. However, what is impermissible is compelling the counter-party to select from such a panel.

Conflicting juridical signals and the seat-venue problem

Pathology can also arise from provisions which clearly choose arbitration but combine inconsistent references to seat, venue, governing law or court jurisdiction. The task in such cases is to identify, through the contract as a whole, the juridical framework within which that arbitration is to operate.

BGS SGS Soma JV v. NHPC Ltd.23 refers to the test of contrary indicia for ascertaining the intent of parties in respect of juridical seat. A geographical reference may, depending on the contractual language and surrounding indicators, amount to designation of the juridical seat rather than a mere hearing venue.

Comparative jurisprudence reflects the same basic tension. In Insigma Technology Co. Ltd. v. Alstom Technology Ltd.24, the Singapore Court of Appeal upheld an unusual hybrid clause combining State Institute for Administrative Careers (SIAC) administration with ICC Rules of Arbitration because the intention to arbitrate was clear and the procedural inconsistency could be made workable. In HKL Group Co. Ltd. v. Rizq International Holdings Pte. Ltd.25, the Singapore High Court similarly adopted a facilitative approach to an inaccurate institutional reference.

Who decides the pathology?

A separate question concerns institutional allocation. Who should decide issues pertaining to pathology?

Vidya Drolia v. Durga Trading Corpn.26 emphasised the principle of kompetenz-kompetenz and described the Arbitral Tribunal as the preferred first authority on questions of jurisdiction and non-arbitrability, subject to a limited prima facie review by the court. The judgment also stated that, in cases of genuine doubt, the court should lean towards reference.

The position has since been further clarified. The seven-Judge decision in Interplay Between arbitration agreements under Arbitration Act, 1996 & Stamp Act, 1899, In re27 distinguished the standards under Sections 8 and 11 and held that Section 11 confines the referral court to examination of the existence of an arbitration agreement on the basis of Section 7. The examination is prima facie and is not a licence for a laborious or contested inquiry. SBI General Insurance Co. Ltd. v. Krish Spg.28 subsequently reiterated that the Section 11 enquiry is limited to the prima facie existence of the arbitration agreement and not to wider questions of arbitrability or merits.

The BGM case29 clarifies that where the alleged arbitration agreement is contained in a single undisputed clause, the referral court may itself examine whether it prima facie satisfies Section 7. Where existence depends on multiple documents or communications, any fuller inquiry should ordinarily be left to the Tribunal. The case of Nagreeka30 also gives an important qualification to the “when in doubt, refer” formulation in the case of Vidya Drolia. The Supreme Court explained that the doubt contemplated there arises in construing an arbitration clause where the parties are otherwise ad idem as to arbitration. Where the parties have not chosen arbitration in the first place, a pro-reference presumption cannot create the missing consent.

Goqii Technologies (P) Ltd. v. Sokrati Technologies (P) Ltd.31 supplies the corresponding caution against enlarging referral-stage scrutiny merely because a claim appears frivolous or mala fide. Such questions are for the Arbitral Tribunal, which may ultimately address abuse through costs.

Drawing the line: A workable Indian framework

Indian law does not contain a statutory test labelled specifically for pathological clauses. The cases have arisen under different contractual settings and procedural provisions. It would therefore be artificial to suggest that they all apply a single formula. Read together, however, the authorities disclose a coherent sequence of enquiry.

1. The first question is whether the contract contains a present and binding agreement to arbitrate.

2. If a binding dispute-resolution process exists, the next question is whether that process is arbitral in juridical character. Finality is relevant but not sufficient. The court must consider the substance of the mechanism, including whether it contemplates an impartial adjudicatory determination of disputes rather than mediation, valuation, certification or internal administrative review.

3. Only after arbitration is identified in principle should the court isolate the defect in machinery.

4. The court should then consider the degree of supplementation required. Interpretation becomes reconstruction when the court must make a substantive contractual choice among materially different possibilities rather than give effect to an objectively ascertainable intention.

5. Finally, the enquiry must remain sensitive to procedural stage. Under Section 11, the referral court undertakes a prima facie examination of existence of an agreement to arbitrate under Section 7, Arbitration Act. Once prima facie existence is established, questions of scope, substantive validity, arbitrability and merits ordinarily belong to the Arbitral Tribunal in the first instance.

The jurisprudence on pathological clauses is best understood not as a contest between strict construction and a pro-arbitration approach, but as an exercise in preserving party autonomy in both directions. Courts should give effect to an imperfectly expressed agreement to arbitrate where the underlying choice is clear, but should not compel arbitration where that choice was never made. The essential distinction is therefore between repairing the means selected by the parties and selecting the means for them.

For parties and practitioners, the drafting lesson is one of precision. The clause should state in clear and mandatory terms that disputes are to be resolved by arbitration. Any negotiation or mediation requirement should regulate only when arbitration may commence, and not whether the parties will later agree to arbitrate. Arbitration should be kept distinct from expert determination, internal review and other contractual processes. The arbitral rules, tribunal constitution, juridical seat and supervisory court should be aligned so that the clause operates coherently. Taking these steps at the drafting stage can prevent avoidable pathology and allow the arbitration agreement to operate as intended, without generating a preliminary dispute over its existence, meaning or implementation.

*Founder and Head, Trinity Chambers, Delhi.

**Counsel, Trinity Chambers, Delhi.

1. (2026) 1 SCC 545 : (2026) 1 SCC (Civ) 69.

2. Frédéric Eisemann, “La clause d’arbitrage pathologique”, in Commercial Arbitration: Essays in Memoriam Eugenio Minoli (Unione Tipografico-Editrice, Torinese, 1974) 129-161.

3. (2003) 7 SCC 418 : (2004) 120 Comp Cas 54.

4. (1998) 3 SCC 573 : (1998) 92 Comp Cas 30.

8. Jagdish Chander v. Ramesh Chander, (2007) 5 SCC 719.

10. (2026) 267 Comp Cas 39 : 2026 SCC OnLine SC 630.

11. Nagreeka Indcon Products (P) Ltd. v. Cargocare Logistics (India) (P) Ltd., (2026) 267 Comp Cas 39 : 2026 SCC OnLine SC 630.

13. (2018) 9 SCC 774 : (2018) 4 SCC (Civ) 574.

14. K.K. Modi v. K.N. Modi, (1998) 3 SCC 573 : (1998) 92 Comp Cas 30.

15. South Delhi Municipal Corpn. v. SMS Ltd., (2026) 1 SCC 545 : (2026) 1 SCC (Civ) 69.

16. Bihar State Mineral Development Corpn. v. Encon Builders (I) (P) Ltd., (2003) 7 SCC 418 : (2004) 120 Comp Cas 54.

17. (2009) 2 SCC 55 : (2009) 1 SCC (Civ) 379.

18. (2014) 5 SCC 1 : (2014) 3 SCC (Civ) 59.

19. (2015) 4 SCC 177 : (2015) 2 SCC (Civ) 530.

20. (2017) 8 SCC 377 : (2017) 4 SCC (Civ) 72.

22. (2025) 4 SCC 641 : (2025) 2 SCC (Civ) 1.

23. (2020) 4 SCC 234 : (2020) 2 SCC (Civ) 606.

24. Insigma Technology Co. Ltd. v. Alstom Technology Ltd., (2009) SGCA 24.

25. HKL Group Co. Ltd. v. Rizq International Holdings Pte. Ltd., (2013) SGHCR 5.

26. (2021) 2 SCC 1 : (2021) 1 SCC (Civ) 549.

28. (2024) 12 SCC 1 : (2025) 3 SCC (Civ) 567.

29. BGM & M-RPL-JMCT (JV) v. Eastern Coalfields Ltd., 2025 SCC OnLine SC 1471.

30. Nagreeka Indcon Products (P) Ltd. v. Cargocare Logistics (India) (P) Ltd., (2026) 267 Comp Cas 39 : 2026 SCC OnLine SC 630.

Powered by Nyaya 247 News