SARR Freights: Indian Evidence Law Meets English Rectification

[Sahil Khan is a Trainee at Unum Law in New Delhi]

On 18 September 2026, a Division Bench of the Delhi High Court restrained a London Maritime Arbitrators Association (LMAA) arbitration on the footing that the parties never agreed to arbitrate. In SARR Freights Corporation v Argo Coral Maritime Ltd, the Bench reversed the single judge’s refusal of interim relief.

The outcome is defensible on the facts. But the Bench’s wider claim, that English-style rectification has no place under codified Indian evidence law, is stated more categorically than its reasoning supports. This post argues that the decision is best read as a finding that no mistake existed, not as a rule that English rectification stays out of Indian courts.

The Fixture Recap, the Booking Note and Clause 43

SARR Freights, a Delhi-based logistics group, agreed to move UN peacekeeping cargo from Delhi to Port Sudan via Mumbai. On 4 April 2023, brokers exchanged a Fixture Recap whose clause 23 provided for English law and London arbitration. The Booking Note signed the next day, on the standard form of Argo Coral’s booking agent Ocean7, applied additional rider clauses 20 to 41. The riders ran to clause 43, the only one containing an arbitration clause (a BIMCO clause with LMAA terms). Every page, including the last, carried both signatures.

After the UN declared Sudanese ports unsafe, SARR Freights cancelled, citing force majeure, and Argo Coral commenced LMAA proceedings. By a majority partial award of 18 October 2025, the tribunal treated “41” as a typographical error and rectified it to “43”. The single judge, relying on Govind Rubber v Louis Dreyfus Commodities Asia and Trimex International v Vedanta Aluminium, likewise read an arbitration agreement out of the Recap.

Existence, Not Oppression, Was the Gateway

Following McDonald’s India v Vikram Bakshi and Bina Modi v Lalit Modi, the Division Bench held that a court may restrain even a foreign-seated arbitration where the agreement is shown to be null and void, inoperative or incapable of being performed. It treated a dispute over whether the parties consented at all as the paradigm case. That let it skip any inquiry into vexation.

This sits uneasily with the Supreme Court’s caution against obstructing tribunals in Kvaerner Cementation v Bajranglal Agarwal and NALCO v Subhash Infra Engineers, which the Bench did not address. And because the Bench decided the existence question on an appeal from an interim order, its findings read as conclusive, though a London tribunal has already decided the opposite.

The Facts Did the Heavy Lifting

The Bench’s most persuasive reasoning is factual. Ocean7’s standard form contains no arbitration clause, so any such clause must have been specifically negotiated. The parties edited the Special Terms box, adding four items, yet left the 20–41 range untouched. They also departed from the Recap on vessel substitution, detention triggers and swell delays. Clause 43 differs in substance from the Recap’s clause 23, because an institutional BIMCO/LMAA regime appears nowhere in the Recap. And a printed term of the Booking Note says it prevails over previous arrangements. From this the Bench inferred a deliberate omission.

That is a finding that there was no mistake, and it matters. Even in England, rectification for common mistake requires a shared prior intention from which the document departs, as the Court of Appeal explained in FSHC Group Holdings v GLAS Trust Corporation, the authority Argo Coral itself cited. The tribunal majority and the Bench looked at largely the same material and drew opposite inferences about intention. That disagreement is about fact, not law, which makes the Bench’s extended contrast with English rectification strictly unnecessary to the result.

The Categorical Claim Is Harder To Defend

The Bench held that once a contract is reduced to a signed writing, sections 91 and 92 of the Evidence Act, 1872 bar recourse to prior negotiations unless the language is ambiguous or defective. It added that the English approach “cannot stand under the Indian codified law”. (The Bharatiya Sakshya Adhiniyam, 2023 now carries the same rules in sections 94 to 98, so the analysis should carry over.) Three points deserve scrutiny.

First, proviso (1) to section 92 permits proof of facts that would invalidate a document, expressly including mistake in fact or law, and section 26 of the Specific Relief Act, 1963 separately allows rectification for fraud or mutual mistake. The judgment quotes only proviso (6) and discusses neither. They may not have helped Argo Coral given the finding of no mistake, but an absolute statement is difficult to reconcile with them.

Second, the Bench confined Anglo American v MMTC to cases where plain language fails to fit existing facts. That is a fair reading. But the Supreme Court did approve reading exchanged documents as a connected whole, so the contextual door is open in Indian law. Here the Bench simply found it closed, because it saw no ambiguity in “20–41”.

Third, choice of law. The Bench treated the question as one of Indian law and invoked BSES v Fenner India to set aside foreign authority where Indian law is settled. Yet whether Indian law governs the existence of an arbitration agreement, when the only route to English law is the disputed clause, is arguably the very issue. Roop Kumar v Mohan Thedani, which the Bench quotes for the view that sections 91 and 92 are substantive law, may supply an answer. The Bench did not draw it out.

Supersession as the Load-Bearing Beam

The Bench also relied on Union of India v Kishorilal Gupta and Young Achievers v IMS Learning Resources for the proposition that an arbitration clause in a superseded agreement does not survive the later contract. It distinguished Govind Rubber and Trimex, which allow an agreement to be inferred from correspondence, because those cases involved no signed final document.

One could object that the supersession language here is printed boilerplate rather than negotiated. The Bench answered twice: the parties actively worked over the form, and the Recap was exchanged between agents while SARR Freights itself signed the Booking Note. The lesson is transferable. Boilerplate supersession will hold where the record shows the form was consciously edited, and may hold less firmly where it was filled in mechanically.

Why This Matters

Three questions remain open. First, effect in London. The injunction operates in personam against Argo Coral, but the tribunal is seated in London and its jurisdiction is governed by the law of the seat. The order restrains the reference and does not speak to the partial award, so how the tribunal and the English courts respond is uncertain.

Second, jurisdiction. The Bench upheld Delhi’s jurisdiction because SARR Freights signed in Delhi, has its office there, and any enforcement would occur there. It rejected the argument that only the Mumbai courts, as the port of shipment, could hear the matter. Foreign owners contracting with Indian charterers should assume this exposure.

Third, drafting. A range reference in a special terms box, a clause in a recap and a signature beneath the last rider clause do not together make an arbitration agreement. If the recap says arbitration, the booking note should say so in terms.

Conclusion

SARR Freights is sound on its facts and a useful reminder of the weight Indian courts give to signed writings. Its wider significance depends on whether later benches read it narrowly, as a finding that no mistake existed, or broadly, as a rule that English rectification stays out of Indian courts.

– Sahil Khan

Read More