Why Water Dispute Tribunals Take Decades to Decide

Two statements about water dispute tribunals in India are repeated constantly, and both are true. The first belongs to the Central Government: adjudication under the Inter-State River Water Disputes Act, 1956 is time-bound, because a tribunal must forward its report within three years of its constitution, a period the Centre may extend by two more. The second belongs to the record: disputes referred under that Act have run for decades, and one reference has remained unfinished, in one stage or another, since 1986. The contradiction is apparent rather than real, and it dissolves once the intervals are separated. The statutory clock governs one stretch of a water dispute, the stretch between a tribunal’s constitution and its report. The decades accumulate in the stretches on either side of it, before a tribunal exists and after it has reported, where the Act sets no clock at all.

Water dispute tribunals and the decades of delay in inter-State river adjudication in India - Insight Crunch

When the Minister for Jal Shakti, Gajendra Singh Shekhawat, presented the official position to the Lok Sabha on 31 July 2019, the figures he gave were these. Nine tribunals had been constituted under the Act. Four of them had given awards. The time taken by those four ranged from seven years to twenty-eight. The Ravi and Beas matter had been pending for thirty-three years, and the Cauvery matter for twenty-nine. Every one of those figures is defensible, and every one of them measures something different from the five-year limit against which public argument habitually sets them. A duration is meaningless without its starting point, and the recurring failure in commentary on water disputes is to quote a span measured from a complaint, or from a tribunal’s constitution, or from an award, as though it were a span measured against the one period the statute actually prescribes. This article keeps the intervals apart. It sets out the periods the Act contains and the periods it omits, walks through the record of the nine tribunals with each duration tied to its endpoints, and then gives the four causes the record supports, before turning to the amendment attempt of 2019, which Parliament passed in one House and then allowed to die for want of a corrected clause.

The clock the statute sets, and the two stretches it leaves timeless

The machinery begins with a complaint. Under Section 3 of the Act, a State Government that considers that a water dispute has arisen, or is likely to arise, by reason of the acts of another State may request the Central Government to refer the dispute to a tribunal for adjudication. The request does not itself start any adjudication. Under Section 4(1), the duty to constitute a tribunal arises only when the Central Government is of opinion that the water dispute cannot be settled by negotiations. The opinion is the gate, and the statute prescribes no period within which the Centre must form it. A request may be met with years of conferences, committees and correspondence, all of it consistent with the Act, because the Act is silent on how long the Centre may take to conclude that negotiation has failed. The detailed working of this scheme, provision by provision, belongs to the statute’s own treatment in the pillar article on the Inter-State River Water Disputes Act, 1956, and what follows takes from it only the periods, because the periods are this article’s subject.

One period does attach to constitution, and it arrived late. As originally enacted, the Act fixed no time limit for constituting a tribunal even after the opinion was formed. The Sarkaria Commission, whose report examined inter-State river water disputes in its seventeenth chapter, recommended that constitution be made mandatory within one year of a State’s application, that the award of a tribunal become effective within five years of its constitution, that a tribunal’s award carry the same force as an order or decree of the Supreme Court, that the Union be empowered to constitute a tribunal of its own motion where a dispute existed in fact, and that a data bank and information system be established at the national level, with States required to supply data to it. A sub-committee of the Inter-State Council accepted four of the five recommendations. The Inter-State River Water Disputes (Amendment) Act, 2002, in force from 28 March 2002, carried the core of that package into the statute. Section 4(1) acquired its one-year requirement: where the Centre is of the requisite opinion, the tribunal must be constituted within one year of receipt of the State’s request. The one-year period, however, runs from a receipt that the Centre has accepted as raising a dispute negotiation cannot settle, and the formation of that opinion remained, as before, unbounded by any period. The amendment put a clock on the second half of the pre-constitution stage and left the first half exactly as timeless as it found it.

The reporting period is the one the public argument knows. Section 5(2) requires a tribunal to forward its report, setting out the facts as found and giving its decision, within three years from the date of its constitution. The Central Government may extend that period by such further period as it considers necessary, not exceeding two years. Three and two make the five years of the standard account. Before 2002 there was no reporting period of any kind: the first generation of tribunals worked under a statute that asked for a report and attached no date to the request, which is part of the explanation for their spans, though only part, as the record below shows. The five-year figure also conceals a feature of the extension power. The Centre grants extensions by notification, tribunal by tribunal, and extensions in the history of the Act have been granted serially, in tranches, so that a tribunal whose statutory period has run out continues to sit on the strength of successive administrative renewals rather than on any single considered decision that more time is justified.

Section 5(3) then builds a further stage into the path of every award. After a report is forwarded, the Central Government or a State Government may, within three months, refer the matter back to the tribunal for explanation or guidance on specified points. The tribunal must forward a further report within one year of such a reference, and the Centre may extend that year by such further period as it considers necessary, with no outer limit written into the provision. The three-month window for seeking the reference is a real period, and references are in practice made within it. The one-year period for the further report is not matched by any consequence for its expiry, and the extension power that qualifies it has no ceiling. In the record, the reference back has been used in nearly every major dispute, by the Centre as well as by States, and it functions less as an occasion for clarification than as a second, shorter adjudication bolted onto the first, on points the losing side selects.

Publication is the stage on which the award’s legal life depends, and it has no period at all. Under Section 6, the Central Government must publish the tribunal’s decision in the Official Gazette, and on publication the decision becomes final and binding on the parties. Since the 2002 amendment, the published decision carries the same force as an order or decree of the Supreme Court, which was the Sarkaria Commission’s fourth recommendation given effect. The provision, however, prescribes no time within which publication must occur. A report may therefore sit with the Central Government, complete and unanswerable, while the legal position of every farmer in the basin remains formally undecided, and nothing in the Act is breached by the sitting. The decree force that commentators cite as the system’s guarantee of finality attaches only at the end of this unbounded interval, not at the moment the tribunal decides.

Section 6A completes the picture of what the statute does and declines to do. It empowers the Central Government, where it considers it necessary, to frame a scheme for giving effect to a tribunal’s decision, and a scheme may constitute an authority charged with implementation. The power is permissive in terms. No period governs its exercise, and for most awards no scheme has ever been framed. Where a scheme and an authority have come, in the Cauvery matter, they came because the Supreme Court directed the Centre to frame one, a point the enforcement discussion below takes up. The statute, in short, prescribes three periods with any firmness: one year for constitution after the opinion is formed, three years extendable by two for the report, and three months for a reference back. It prescribes no period for forming the opinion, none for the further report that an extension cannot indefinitely prolong in practice, none for publication, and none for anything that happens after publication. The governed interval is the middle of the dispute’s life. The ends are ungoverned.

How many water dispute tribunals have been constituted under the Act?

Nine tribunals have been constituted under the Act: for the Krishna in 1969 and again in 2004, and one each for the Godavari, the Narmada, the Ravi and Beas, the Cauvery, the Vansadhara, the Mahadayi and the Mahanadi. The Minister gave that count to the Lok Sabha on 31 July 2019; four had delivered awards by then.

The count repays a moment’s attention, because it is a count of tribunals and not of disputes. The Krishna appears twice because the first tribunal’s award was made subject to review after a date, and a fresh reference produced a second tribunal a generation later. The Godavari and Krishna tribunals of 1969 were constituted on the same day, 10 April 1969, with a common chairman in Justice R. S. Bachawat, to take up disputes that the Sarkaria Commission later recorded as having lingered since 1956. A ninth tribunal, for the Mahanadi, followed in 2018. Whether a dispute that has passed through two tribunals and several rounds of litigation is one dispute or several is a question the official count does not answer, and the durations in the next section are therefore given tribunal by tribunal, stage by stage, so that no two unlike intervals are ever added together or compared.

The record of nine tribunals, stage by stage

The first generation shows both the system’s original slowness and its occasional capacity to finish. The Krishna tribunal was constituted on 10 April 1969 and forwarded its report on 24 December 1973, a span of four years and eight months for the adjudication itself. The Central Government and the three riparian States then filed references under Section 5(3), the tribunal heard them, and the decision was published in the Official Gazette on 31 May 1976. From constitution to publication, the Krishna dispute’s first adjudication took seven years and a month, of which the reference back and the publication interval account for two years and five months after the report was already written. The award allotted 2,060 thousand million cubic feet of Krishna water at 75 per cent dependability among Maharashtra, Karnataka and Andhra Pradesh, and it was in substance accepted and acted upon by all three States for the remainder of the century. Its one deliberate incompleteness was a review clause: the arrangement could be revisited by a competent authority or tribunal at any time after 31 May 2000, which converted a settled award into the seed of a second reference.

The Godavari tribunal, constituted the same day in April 1969 under the same chairman, took longer and finished differently. It made its award on 27 November 1979, more than ten years after its constitution, gave a further award under Section 5(3) on 7 July 1980, and saw the decision published in the Gazette in July 1980. The span invites a qualification that matters for the argument of this article. During the nineteen seventies the riparian States of the Godavari basin negotiated a series of bilateral and multilateral agreements among themselves, and the tribunal’s award largely gave legal form to allocations the States had already accepted in those agreements. The ten years the Godavari file records were not ten years of continuous adjudication. They were years in which adjudication and negotiation ran in parallel, and the negotiation, for once, delivered much of the answer. The Godavari case is the standing counterexample to any simple claim that the pre-tribunal and extra-tribunal stages are wasted time.

The Narmada tribunal sits between the two. Gujarat’s complaint had set the process in motion, and the Sarkaria Commission recorded that the Narmada dispute had been continuing for six years before the tribunal was constituted on 6 October 1969. The tribunal’s progress was interrupted by negotiation at the highest level: an agreement reached in July 1974 among the Chief Ministers of Madhya Pradesh, Maharashtra and Rajasthan and the adviser to the Governor of Gujarat, with the assistance of the Prime Minister, settled the assessment of available water and left the tribunal a narrower set of questions, including the height of the dam at Navagam. The tribunal declared its award under Section 5(2) on 16 August 1978, just under nine years after its constitution. All five interests, the Union and four States, filed references under Section 5(3), and the tribunal’s final order came on 7 December 1979, with publication in the extraordinary Gazette on 12 December 1979. Constitution to publication: ten years and two months. Implementation then passed to the Narmada Control Authority, a standing body created to give effect to the decision, and the award’s allocations and the dam height it fixed governed the valley’s development thereafter. The award, like the first Krishna award after it in the comparison this article has drawn, wrote its own succession into its text: by its sixteenth clause the shares it fixed, the full reservoir level and the canal’s full supply level were made subject to review at any time after forty-five years from the award’s publication in the Gazette, a period that ran from December 1979. The Narmada award was litigated later, fiercely, on rehabilitation and environmental questions, but the inter-State allocation itself held. The two oldest completed awards in the system thus share a feature the system’s critics rarely credit: each contemplated its own revision and provided the date from which revision could lawfully be sought, and each then operated, undisturbed in its essentials, for the whole of the period it had set itself. Duration cuts both ways in this record. The same system that takes decades to decide has produced decisions that endure for decades, and a reform that shortened the first kind of duration by destroying the second would be a poor bargain.

The Ravi and Beas reference had a prelude of its own, and it explains the tribunal’s strange career. The dispute was created by reorganisation: when Punjab was divided in 1966, Haryana acquired a claim to the waters of the Ravi and Beas that the undivided State had used alone. The Central Government allocated the surplus waters by an order of 1976, the three States, Punjab, Haryana and Rajasthan, revised the arrangement by an agreement of 1981, and the Sutlej-Yamuna Link canal was begun in Haryana’s territory to carry Haryana’s share. Punjab’s politics then turned against the arrangement: the State’s assembly resolved against the agreement in the mid nineteen eighties, construction in Punjab’s territory stalled, and the Punjab settlement of 1985, the accord between the Prime Minister and Sant Harchand Singh Longowal, sought to close the question by referring the quantum to a tribunal whose findings on the parties’ shares would settle what negotiation had not. The Ravi and Beas tribunal thus began life as the instrument of a settlement rather than as a substitute for one, charged with verifying the riparian shares of the States concerned and with the availability of water, and its report of January 1987, delivered in under ten months, reflected the narrowness of that charge. What it could not survive was the reopening of the politics around it. The references back of August 1987 asked it to explain and clarify, and while it deliberated the settlement that created it decayed, until by the time of the Termination Act of 2004 the question was no longer what the tribunal meant but whether Punjab’s obligations existed at all. A clarificatory reference addressed to a tribunal whose parent settlement has collapsed is a letter to an address from which the occupants have moved, and the further report’s forty-year gestation is in part the record of a tribunal waiting for a political world that would make its answer operative.

The Ravi and Beas tribunal inverts the usual pattern: its adjudication was the fastest on record and its completion the slowest. Constituted on 2 April 1986 to give effect to the water provisions of the Punjab settlement of 1985, the tribunal submitted its report and decision under Section 5(2) on 30 January 1987, inside ten months. The report was forwarded to the States on 20 May 1987. The Central Government forwarded its own references, together with those received from the party States, to the tribunal on 19 August 1987, seeking explanation and clarification under Section 5(3). The further report has never been given. The tribunal’s tenure has instead been extended on a yearly basis, a state of affairs the Jal Shakti Ministry was still recording in its parliamentary answers of 2022 and 2023, and the reference remains formally alive more than three and a half decades after it was made. Around that unfinished reference the dispute continued by other routes. Punjab enacted the Punjab Termination of Agreements Act on 12 July 2004, purporting to annul its water agreements and discharge its obligations to deliver water to Haryana and Rajasthan. The President referred the Act’s validity to the Supreme Court on 22 July 2004, and the Court’s advisory opinion of 10 November 2016 held the enactment constitutionally invalid. The opinion settled the law of the Termination Act. It did not produce the tribunal’s further report, and the Sutlej-Yamuna Link canal, the physical work at the heart of the quarrel, remained unbuilt in Punjab’s territory. When the Minister told the Lok Sabha in 2019 that the Ravi and Beas matter had been pending for thirty-three years, the figure measured from the tribunal’s constitution in 1986, and it was accurate. What it could not convey is that the pending item is not an adjudication that has run long but a clarification that has never been answered.

The Cauvery tribunal supplies the fullest demonstration of the stage structure, because in that dispute every stage was used and several were used twice. Tamil Nadu’s demand for a tribunal ran through years of failed negotiation after the 1924 agreement between Madras and Mysore lapsed in 1974, and the tribunal was constituted on 2 June 1990, following proceedings in the Supreme Court brought from Tamil Nadu seeking that step. Its interim order of 1991 provoked a Karnataka ordinance designed to nullify it, which the Supreme Court declared unconstitutional. The final award came on 5 February 2007, sixteen years and eight months after constitution, apportioning 740 thousand million cubic feet in a normal year among Tamil Nadu, Karnataka, Kerala and Puducherry. The Centre then held the report. Publication, which Section 6 requires and no period governs, did not follow until the Supreme Court directed it on 4 February 2013; the decision was notified on 19 February 2013, six years after it was written. Appeals against the award then occupied the Court for five further years. In its judgment of 16 February 2018 the Court raised Karnataka’s share to 284.75 thousand million cubic feet, reduced Tamil Nadu’s to 404.25, left Kerala at 30 and Puducherry at 7, and reduced the monthly releases to be measured at Billigundlu to 177.25. The Court also directed the Centre to frame a scheme under Section 6A, and the Cauvery Water Management Scheme was notified on 1 June 2018, constituting the Cauvery Water Management Authority and the Cauvery Water Regulation Committee. From the tribunal’s constitution to the judgment that closed the appeals, the Cauvery adjudication ran for twenty-seven years and eight months. From constitution to the Authority’s creation, it ran for twenty-eight years less a day.

The second Krishna tribunal shows the post-2002 Act performing no better than its predecessor. Constituted on 2 April 2004, with the amended time limits in force, it gave its report and decision on 30 December 2010, six years and nine months after constitution and therefore outside even the extended five-year period the amended Section 5(2) describes, its tenure having been renewed by successive extensions. Andhra Pradesh challenged the report in the Supreme Court, and by an order of 16 September 2011 the Court restrained the Central Government from publishing the decision in the Gazette, so that the report entered the one stage with no period, publication, and was held there by litigation rather than by administrative inaction. References under Section 5(3) produced a further report on 29 November 2013. The bifurcation of Andhra Pradesh then reopened the allocation itself: Section 89 of the Andhra Pradesh Reorganisation Act, 2014 extended the tribunal’s term and required it to address the sharing between the successor States, and the tribunal’s work continued on that enlarged reference. A report delivered in 2010 had thus not, a decade later, produced a published and operative allocation for the basin.

The second Krishna reference also carries a lesson of its own about how awards age. The first tribunal, unusually, had written its award’s mortality into the award itself, providing that the arrangement could be reviewed by a competent authority or tribunal at any time after 31 May 2000. The clause was a recognition that allocations made on the flow record of the nineteen sixties could not bind a basin forever, and it gave the basin a lawful road to reopening rather than leaving discontent to find an unlawful one. The road was taken at once: demands for a fresh tribunal followed the date, and the second tribunal of April 2004 was its product. What the clause could not schedule was everything after the reopening. The second tribunal worked on a longer flow record and assessed the basin at 65 per cent dependability where its predecessor had used 75, an illustration of how a generation’s additional data changes the size of the thing being divided. Its report of December 2010 then entered the publication stage and was stopped there by the Supreme Court’s restraint of September 2011, the references back ran to November 2013, and the State whose territory the basin chiefly concerned was divided in 2014, so that the allocation the tribunal had made among three States required a further allocation between two of them, under a term extended by the reorganisation statute. A review clause designed to keep an award current instead delivered the basin into every ungoverned stage the Act possesses, in sequence. The contrast with the first tribunal’s career is the sharpest in this article’s record: the award of 1976, imperfect and incomplete as it was, was published, accepted and worked for twenty-five years, while the award of 2010, produced under a statute Parliament had reformed in the interval, had still not acquired operative force a decade after it was written.

The three most recent tribunals repeat the pattern in miniature. The Vansadhara tribunal, constituted on 24 February 2010 for the dispute between Andhra Pradesh and Odisha, submitted its report on 13 September 2017, seven and a half years after constitution. The dispute turned on Andhra Pradesh’s project at Neradi, a barrage across the Vansadhara to which Odisha objected, fearing submergence and loss of flow on its side of the boundary the river forms between the two States. The report permitted the barrage to be constructed, subject to conditions, and apportioned the river’s waters between the States. Odisha and the Central Government filed references under Section 5(3) in December 2017, and the further report remained pending on the record the Ministry placed before Parliament, so that a project approved by a tribunal in 2017 awaited, years later, the clarification that would settle the terms on which it could proceed: the reference back performing, in a small basin, the same suspension it performed in the Ravi and Beas. The Mahadayi tribunal’s history divides into the two ungoverned stretches. Goa sent its request under Section 3 to the Central Government in July 2002, as the Jal Shakti Ministry confirmed in a Lok Sabha answer on 29 July 2021; negotiation efforts followed, and the tribunal was constituted only on 16 November 2010, eight years and four months after the request. It submitted its report on 14 August 2018. The three States filed references under Section 5(3) and also carried the report to the Supreme Court by special leave petitions. Publication came on 27 February 2020, under an order of the Supreme Court dated 20 February 2020, and expressly subject to the outcome of those petitions. The Mahadayi reference repays a closer statement of its subject, because its scale inverts the usual proportions of a water dispute and shows that duration does not track magnitude. The river is short, rising in the Western Ghats in Karnataka and reaching the sea through Goa as the Mandovi, and the quantities are small by the standards of the Krishna or the Cauvery: the tribunal assessed the waters generated in the Karnataka and Maharashtra catchments and apportioned 40.125 thousand million cubic feet of consumptive use among the three States. Karnataka’s object was the Kalasa-Banduri project, a diversion of Mahadayi waters into the Malaprabha basin to supply drinking water to Hubballi-Dharwad and the districts of Belagavi and Gadag, for which it sought 7.56 thousand million cubic feet. The award of 14 August 2018 allotted Karnataka 13.42 thousand million cubic feet in all, including 3.9 for diversion out of the basin and a share for power generation, Goa 24 and Maharashtra 1.33. A dispute over a diversion smaller than a single medium reservoir on the Krishna had by then consumed eight years before its tribunal existed, eight years in the tribunal, eighteen months awaiting publication, and the years since in references and petitions, a quarter of a century after Goa first asked for adjudication in 2002. If the system’s delays were proportioned to the difficulty of the questions, the Mahadayi file would be thin. It is among the thickest, because the delays are properties of the stages, not of the rivers.

The Mahanadi tribunal is the one instance in which the pre-constitution stage was cut short from outside the Act: Odisha’s complaint of 19 November 2016 was followed by a Negotiation Committee whose report came on 27 June 2017, by a direction of the Supreme Court on 23 January 2018, and by the tribunal’s constitution on 12 March 2018. Even with a court’s hand on it, sixteen months separated complaint from constitution, in a dispute over a river whose projects were being built while the parties argued. The Mahanadi complaint concerned the effect on Odisha, and on the Hirakud reservoir’s inflows, of the dams and barrages Chhattisgarh was constructing in the river’s upper reaches, and the adjudication the tribunal inherited is therefore of the familiar kind: a contest over the flow record of a river being altered by one of the contestants while the record is made.

The long prelude: what happens while the Centre negotiates

The record above begins, in each case, at a tribunal’s constitution, because that is where the Act’s own counting begins. The disputes themselves are older, and the pre-tribunal years deserve their own examination, because they are the years in which the pattern of the whole system was set. The Sarkaria Commission, reviewing the Krishna and Godavari references a generation after they began, recorded the anatomy of the prelude in terms that later disputes have repeated almost exactly. The two disputes had been lingering since 1956, the year of the Act itself, as the reorganisation of States redrew the riparian map and made the allocations of the past unworkable. A scheme for reallocation was drawn up by the Central Water and Power Commission and was not acceptable to the States. An inter-State conference in September 1960 failed to settle the matter. In January 1962 Mysore applied under Section 3 for a reference to a tribunal. In March 1963 the Union Minister for Irrigation and Power told the Lok Sabha that the agreement of 1951, on which part of the existing distribution rested, was legally ineffective and unenforceable, a statement that removed the legal floor from the negotiation without putting any adjudication in its place. The Union Government made further attempts at settlement. Fresh applications for reference were made by Maharashtra, Mysore, Orissa and Madhya Pradesh in 1968. The tribunals came in April 1969, thirteen years after the disputes had begun and seven years after the first formal application for one. The Commission’s comment on the sequence was restrained and exact: the dates of application and the dates of constitution do not indicate the actual span of the disputes, and in the three cases it examined the entire process took an unduly long time, during which development of irrigation and power could not wait and did not wait.

The Cauvery prelude was longer and its structure identical. The dispute in its modern form descends from two agreements between the Madras Presidency and the princely State of Mysore, of 1892 and 1924, the second of which ran for fifty years and rested on the principle that the upper riparian would not construct works on the river without the lower riparian’s consent. When the 1924 agreement lapsed in 1974, Karnataka took the position that it was free of the consent regime and proceeded with its projects, and Tamil Nadu took the position that the established pattern of use was protected. Sixteen years separated the lapse of the agreement from the constitution of the tribunal in June 1990, and they were filled with the same instruments the Krishna prelude had used: conferences, correspondence, draft agreements that failed, and finally litigation by interests from Tamil Nadu to compel the step the Centre would not take unpressed. Nothing in those sixteen years was unlawful. Every month of them was consistent with a statute that makes adjudication conditional on the Centre’s opinion that negotiation has failed, and every month of them strengthened the position of the State in possession of the water and of the works built with it. When the tribunal finally sat, it inherited a basin in which the upper riparian’s utilisation had grown for a decade and a half under the protection of the delay itself, and the protection of existing uses, a principle no tribunal abandons, converted part of that growth into entitlement.

The prelude has a second, less noticed property: it is the stage at which the dispute is least visible. A tribunal files reports, and its existence can be counted and its duration measured. A negotiation has no registry. The years in which a complaint rests with the Centre are recorded, if at all, in the answers to parliamentary questions, which is why the Mahadayi prelude can be dated at all: Goa’s request of July 2002 and the tribunal of November 2010 survive in the record because a member asked and the Jal Shakti Ministry answered, in July 2021, with the dates. For most disputes the prelude’s length can only be reconstructed from commissions of inquiry and the prefaces of awards, and official presentations of the system, which begin their arithmetic at constitution, are able to pass over it in silence without falsifying anything. A measurement convention that starts the clock when the tribunal is born will always acquit the stage in which most of a dispute’s life is spent before its birth.

Where the years go: the six stages of a water dispute

Where the years go. The table below sets each stage of a water dispute under the Act against the period, if any, that the statute prescribes for it, and against an observed duration drawn from the record set out above, with the source of the observation named. Its purpose is the purpose of this article: to show that the governed stage is not the stage in which the decades accumulate. The record on which it rests, of Gazette notifications, tribunal reports and answers given to Parliament, circulates as scanned documents and office files, and a reader assembling it for study can work through the material in the ReportMedic office file viewer, which opens the Word, Excel and presentation formats in which such documents are exchanged.

Stage	Statutory period	What the record shows
First formal complaint to the Centre’s opinion that negotiation has failed	None. Section 4(1) conditions the duty to constitute a tribunal on that opinion and fixes no period for forming it	Goa’s request of July 2002 produced a tribunal in November 2010, a gap of eight years and four months (Jal Shakti Ministry, Lok Sabha answer, 29 July 2021). Odisha’s Mahanadi complaint of November 2016 produced a tribunal in March 2018, and only after a Supreme Court direction of January 2018
Centre’s opinion to constitution of the tribunal	One year from receipt of the request, since 28 March 2002. Before the 2002 amendment, no period existed	The Sarkaria Commission recorded the Krishna and Godavari disputes as lingering since 1956, thirteen years before the tribunals of April 1969; the Narmada dispute had continued for six years before the tribunal of October 1969
Constitution to the first report under Section 5(2)	Three years, extendable by up to two, since 2002. Before the amendment, no period existed	Krishna I: April 1969 to December 1973, about four years and eight months (the tribunal’s report as recorded in Supreme Court proceedings). Cauvery: June 1990 to February 2007, sixteen years and eight months. Krishna II: April 2004 to December 2010, six years and nine months, on successive extensions
The reference back under Section 5(3)	A reference must be sought within three months of the report. The further report is due within one year, extendable by the Centre without a stated outer limit	Narmada: references by the Union and four States carried the award of August 1978 to a final order in December 1979, about sixteen months. Ravi and Beas: references of August 1987, further report still awaited in the Ministry’s parliamentary answers of 2022 and 2023
Report to publication in the Official Gazette under Section 6	None. Publication is required, and the decision takes effect and acquires the force of a Supreme Court decree on publication, but no period is prescribed	Krishna I: report December 1973, published May 1976, two years and five months. Cauvery: report February 2007, published February 2013, six years, and only under a Supreme Court direction. Mahadayi: report August 2018, published February 2020, under a Supreme Court order
Publication to the conclusion of proceedings in the Supreme Court	None. Section 11 bars court jurisdiction, but special leave petitions against awards have been entertained and decided	Cauvery: notified February 2013, appeals decided February 2018, five years. Krishna II: publication itself restrained by a Supreme Court order of September 2011 before the Gazette stage was reached

One feature of the table deserves emphasis before the analysis resumes. The only stage in which the observed duration regularly matches the statutory design is the one the standard criticism ignores: the three-month window for a reference back, which parties observe because it is framed as a condition of their own right to reopen. Periods in this statute bind when they are conditions on the party who must act, and drift when they are duties owed by an institution to no one in particular. That distinction, more than any shortage of resources, is what the six rows display.

Why the five year limit is not breached on its own terms

The most common sentence written about Indian water disputes is that tribunals routinely violate the five-year limit Parliament imposed. The sentence is, in the form in which it is usually written, not accurate, and the inaccuracy is instructive. The five years of Section 5(2) govern the interval between constitution and report. Measured on that interval, and leaving aside the extensions the same section authorises, most of the famous spans of the water dispute record do not even purport to be measurements of it. The Cauvery dispute’s twenty-nine years, as given to the Lok Sabha in 2019, run from the tribunal’s constitution in 1990 to a point at which the appeals had been decided and the implementation authority created: the reporting interval inside that span is sixteen years and eight months, and the remainder is publication delay and appellate time, neither of which Section 5(2) touches. The Ravi and Beas figure of thirty-three years is, on inspection, a measure of an unfinished reference back, an interval for which the Act prescribes one year and then qualifies it with an extension power that has no ceiling. To say that the five-year limit is breached is to treat the limit as a limit on the dispute. It is a limit on one stage of the dispute, and conflating the two misdiagnoses the system in a way that guarantees the wrong remedy.

This is the seventh occasion on which this series has had to separate a measurement from the boundary that defines it, and the third on which the divergence between the official figure and the observed figure has turned out to be structural rather than arithmetical. The pattern appeared in the sanitation cluster, where the implementation record of the manual scavenging legislation diverged from the official account because the two were counting different populations under the same name, and again in the rural employment cluster, where the MGNREGA implementation record showed official and independent series parting company because each measured the programme from a different point in its delivery chain. The water dispute case belongs to the same family and is in one respect its purest member. In the sanitation and employment cases the parties at least purported to measure the same quantity. In the water case, the official limit and the observed duration measure different quantities altogether: the one a statutory interval inside a longer process, the other the process itself. Both are accurate. Neither answers the other’s question, and any reform debate that does not begin by distinguishing them will spend itself on the wrong target, tightening a clock that is not the one running slow.

The measurement point can be fixed by working one completed and one unfinished case through the statutory arithmetic. Take the first Krishna reference. The dispute, on the Sarkaria Commission’s finding, had lingered since 1956: thirteen ungoverned years before the Act’s process began in earnest. The tribunal’s adjudication, from April 1969 to the report of December 1973, took four years and eight months, inside the five years the amended Act would later prescribe, though no such period then existed. The reference back and the interval to publication took two years and five months more, for a constitution-to-publication span of seven years and one month, and a dispute-to-award span, measured from the lingering of 1956, of twenty years. Every figure in that sentence is accurate, and a writer may choose among seven, thirteen and twenty years for the same adjudication according to the point being made, which is why the choice must always be declared. Take, against it, the Cauvery reference: sixteen years and eight months inside the reporting stage alone, a figure that no reallocation of stages can rescue, followed by six years to publication and five in the appeals. The two cases together mark the honest range of the system’s defence. Where the adjudication stage itself is concerned, the early tribunals were slow but not grotesque, and their spans were recorded under a statute that set them no period. It is the stages around adjudication, in 1969 as in 2019, that convert years into decades, and the Cauvery case shows the further truth that a sufficiently contested dispute will overflow even the one stage that has a clock.

The public account of all this is episodic, which is a final feature of the measurement problem. No annual return lays the state of the references before Parliament as a series; the durations in this article have had to be assembled from ministerial answers to individual questions, from the Gazette, from judgments and from the records of commissions, each giving its figures for its own purpose and its own date. A citizen asking how long the Vansadhara references have been pending, or what stage the Mahadayi further report has reached, will find the answer, if at all, in an answer given to a member on a particular day, accurate for that day and silent thereafter. Systems that must report themselves regularly acquire a discipline that systems reported on by accident do not, and the water disputes system’s immunity from routine account is of a piece with its directory periods and its reasonless extensions: at no point in the ordinary course is any institution required to stand up and own the whole of a dispute’s age.

What does a reference back under section 5(3) do to a water dispute?

A reference back lets the Central Government or any party State, within three months of a tribunal’s report, send specified points back to the tribunal for explanation or guidance. The tribunal must give a further report within one year, a period the Centre may extend without a stated outer limit. In the record it reopens the adjudication itself.

The mechanism was designed as a clarifying device, a means of resolving genuine obscurities in a long technical award before the award hardened into a decree. Its operation in the record has been wider. Because the points referred are chosen by the parties, and because a party that has lost on allocation can almost always formulate points on which explanation is genuinely needed, the reference back operates in practice as a structured opportunity to re-argue. The tribunal that hears it is the tribunal that decided, asked to explain itself to the dissatisfied, and the further report it produces may confirm, gloss or adjust the original decision, as the Godavari tribunal’s further award of July 1980 adjusted its award of the previous November. The device also interacts with publication in a way the Act does not spell out: a government holding a report under reference has a reason, or at least an excuse, to defer the Gazette notification, so the three months in which references may be sought can stretch the unbounded publication stage behind them. In the Ravi and Beas matter the device failed altogether, producing neither clarification nor closure across more than three decades, because nothing in the provision compels the tribunal to answer within the year or attaches any consequence to its silence. A stage designed to take a year, and in the Narmada case taking sixteen months, has in the hardest case taken the entire life of a generation of the dispute.

The opinion with no deadline, and the deadline that bends

The first cause of delay is the gate at the front of the Act. Constitution waits on the Centre’s opinion that a dispute cannot be settled by negotiations, and no period governs the forming of that opinion. The design is intelligible: negotiation is cheaper and more durable than adjudication, and a Centre that referred every complaint on receipt would take settlement off the table in disputes that settlement could have closed. The cost of the design is written across the record. The Krishna and Godavari disputes were, on the Sarkaria Commission’s finding, lingering from 1956, yet the tribunals came in 1969. The Narmada dispute ran for six years before its tribunal. Goa’s request of 2002 waited eight years for the Mahadayi tribunal of 2010. Tamil Nadu’s demand for a Cauvery tribunal ran through the nineteen eighties and was met only in 1990, after litigation pressed the point. In each case the interval was occupied by conferences, committees and ministerial correspondence that the Act neither requires nor limits, and in each case the underlying works did not wait: projects proceeded on the ground while the question of their water was formally under negotiation, so that by the time a tribunal sat, it faced a basin already partly built, and facts on the ground that no award could fully undo. A delay before adjudication is never neutral between upper and lower riparians. The upper State holds the water in the interim, and time itself is a party to every water dispute. The unbounded opinion stage therefore does more than postpone decisions. It rewards the party satisfied with the existing distribution and taxes the party seeking change, and it does so without any institution ever having decided that this allocation of advantage is the right one.

The second cause is that the periods which do exist operate, in practice, as directory rather than mandatory. A mandatory period is one whose breach carries a consequence: the act done late is invalid, or the authority to do it lapses. Section 5(2) attaches no consequence to a late report. The tribunal that has outrun its three years and its two years of extension does not dissolve; the Centre notifies a further extension, and the adjudication continues, lawfully in every practical sense, because no court will treat a basin’s allocation as void for lateness at the suit of a State that has itself sought extensions and references along the way. Krishna II’s six years and nine months to its first report, across the very period in which the amended limits were in force, is the demonstration. The Vansadhara tribunal’s seven and a half years is another. The Cauvery tribunal’s sixteen years and eight months belong mostly to the unamended Act and so indict the old design more than the new, but the new design’s own tribunals show that adding a period without adding a consequence changes the vocabulary of delay rather than its duration. Extensions are recorded, year by year, as administrative routine; the Ravi and Beas tribunal has been extended annually for a further report that remains ungiven, an arrangement which converts an extension power meant for exceptional prolongation into the ordinary mode of the tribunal’s existence.

The extension habit deserves a closer look, because it is the mechanism by which directory periods are administered, and its paperwork is revealing. An extension is not a single grant of the time a tribunal needs. It is a series of short renewals, granted as each previous renewal runs out, so that a tribunal’s legal existence is perpetually a few months deep. The Mahadayi tribunal’s further report, awaited since the references of 2018, has been carried in exactly this way: the Jal Shakti Ministry’s answers to Parliament record the tribunal’s term extended for a further period of one year at a time, an extension of July 2022 running from August of that year, and the series of renewals still running in the Ministry’s later answers, with the tribunal adjourning its own proceedings on the references while the special leave petitions against its report are pending in the Supreme Court. The Ravi and Beas tribunal’s yearly extensions have already been described. The administrative meaning of the pattern is that no decision is ever taken, at any point, that a reference shall have five more years, or ten; instead a decision is taken, repeatedly, that it shall not end yet, which is a decision no one need defend on its merits because it decides nothing. A system that wished its periods to bind would put the extension decision at a level and in a form that required reasons proportioned to the time sought. A system content with directory periods grants them by routine notification, and the notification, renewed often enough, becomes indistinguishable from the tribunal’s constitution itself: a standing arrangement wearing the dress of a temporary one.

There is a further feature of the pre-constitution stage that the statutory text cannot show and the record cannot hide: the Central Government is not a disinterested timekeeper. The Centre that must pronounce negotiation hopeless is a political executive with interests in the riparian States themselves, frequently governing in one of them, allied with the government of another, or courting both. To constitute a tribunal is to set in train a process that will, in the ordinary course, disappoint at least one State government and be blamed by it, at a moment not of the Centre’s choosing. To continue negotiating is to disappoint no one on the day it is continued. The opinion stage gives that preference a legal form, and the absence of a period for the opinion means the preference can be indulged indefinitely without any decision ever being recorded against anyone. The pattern of the record fits the incentive uncomfortably well: the references that moved fastest were those a settlement had already framed, as with Ravi and Beas in 1986, or those a court pushed, as with Cauvery in 1990 and Mahanadi in 2018, while references resting on a State’s complaint alone, Goa’s of 2002 being the cleanest example, waited for years in the negotiation stage. It would be wrong to reduce the Centre’s hesitation to calculation, since the same period is genuinely used to attempt settlements and some attempts succeed. It would be equally wrong to design a system on the assumption that the gatekeeper has no stake in the gate, and then to express surprise at how seldom it opens promptly.

An award is not the end: the court after the tribunal

The third cause is the incompleteness of the award’s finality. The constitutional and statutory design intended water adjudication to end with the tribunal. Article 262 empowers Parliament to exclude the jurisdiction of every court, including the Supreme Court, over inter-State water disputes, and Section 11 of the Act exercises that power, barring the Supreme Court and every other court from jurisdiction over a dispute referred, or to be referred, to a tribunal. The doctrine of the ouster, and the line of authority by which the Court has in practice entertained proceedings touching awards, is treated in the statute’s own article and is taken here only for its effect on time. The effect is plain. Special leave petitions against tribunal decisions have been filed, entertained and decided. In the Cauvery matter the route carried the dispute from a report of 2007 through a publication that itself required a direction of the Court in 2013 to a judgment of 2018 which did not merely review the award but altered its central figures, raising one State’s share and reducing another’s, and directed the creation of the implementation machinery the statute made optional. In the Krishna matter the route reached in before publication and stopped it, the Court’s order of September 2011 restraining the Gazette notification of the second tribunal’s report while the petitions were heard, so that the decree force which publication confers was withheld for years by the very process meant to be excluded. In the Mahadayi matter the award was published in February 2020 expressly subject to the outcome of the petitions pending against it, a formula which records, on the face of the notification, that publication no longer means settlement.

A reader is entitled to ask why States litigate awards that the Constitution sought to place beyond litigation, and the answer is not cynicism. The sums at stake in a basin allocation are measured in the drinking water of cities and the irrigation of districts, and an award binds not for a season but for decades, in the Krishna case by design until a review the award itself scheduled. A government that accepted an adverse allocation without exhausting a remedy the Court has shown itself willing to entertain would be answerable to its own legislature for the surrender. The ouster failed, as a practical matter, because the Court’s appellate jurisdiction under Article 136 is constitutional in source and the bar in Section 11 is statutory, and when the two have met, the statutory bar has not held the line the framers of the Act drew. The consequence for duration is that the tribunal stage, however long, is followed by a judicial stage with no period of any kind, conducted at the pace of the Supreme Court’s docket, and capable of revising the substance of what the tribunal decided. Finality in water adjudication is therefore not a property of the award. It is a property of the exhaustion of the parties, reached when a judgment of the Court closes the last petition or when a State calculates that further proceedings will gain it less than compliance costs. The Cauvery dispute reached that point in 2018, twenty-eight years after its tribunal was constituted. The Ravi and Beas dispute, travelling by an original suit over the canal and a Presidential reference over the Termination Act rather than by appeal against an award, has not reached it at all.

Why can a tribunal award wait years for Gazette publication?

Because Section 6 requires the Central Government to publish a tribunal’s decision in the Official Gazette but prescribes no period within which it must do so. Publication is the act that makes an award final, binding and possessed of the force of a Supreme Court decree, so a report awaiting publication is a decision suspended between making and effect.

The Cauvery report waited six years in that suspension, from February 2007 to February 2013, and the wait ended not because any period expired but because the Supreme Court directed publication. The Mahadayi report waited from August 2018 to February 2020 and was published under a court order as well. The Krishna report of 1973 waited two years and five months, across the references back, for the Gazette of May 1976. Publication delay has a politics of its own. A report is almost always unwelcome to at least one State government, frequently to two, and the Centre that publishes it spends political capital with the disappointed while gaining nothing it can point to from the satisfied, who receive only what they say was always theirs. Where Section 5(3) references are pending, deferral wears the dress of procedural propriety. Where they are not, it wears no dress at all, and the Act offers no remedy against it, because the duty to publish is unaccompanied by a date and therefore, in the ordinary run of administration, unenforceable until a court supplies the date the legislature omitted. The 2002 reform sharpened the stage’s difficulty in a way its authors can scarcely have intended. By giving a published decision the force of a decree of the Supreme Court, the amendment raised the legal consequence of the Gazette notification from the commencement of an obligation to the creation of an executable instrument of the highest rank. A government contemplating publication after 2002 is therefore not merely announcing a result; it is arming the winning State with a decree. The rational hesitation that the unbounded stage always invited acquired a stronger reason, and the two longest publication delays in the record, the Cauvery’s six years and the restraint on the second Krishna report, both belong to the period after the decree force was conferred. The Sarkaria Commission had recommended the decree force and the five-year effectiveness of awards as a pair, force being the sanction that would make the timetable real. The amendment severed the pair, enacting the force and leaving the timetable’s outer stages open, and so produced a system in which awards are harder to publish than they were when they bound less.

It is worth noting what the omission costs beyond time. A basin cannot plan on an unpublished award, a State cannot lawfully build on shares that have been decided but not made binding, and the tribunal’s own authority decays as the interval lengthens, so that when publication finally comes it arrives as the revival of an old quarrel rather than as its conclusion.

The facts are also in dispute

The fourth cause is the least discussed and the most fundamental. An allocation of river water is a division of a quantity, and the quantity is not given by nature in a form the parties accept. A tribunal must first establish the flow of the river across a run of years, decide the dependability at which the available water is to be assessed, measure existing utilisation in each State, and project the needs against which shares are to be set. Every one of those findings is contestable, and every one is contested. The flow series depends on gauge and discharge sites whose records the States compile through their own irrigation establishments and do not compile identically. Dependability is a statistical convention with distributive consequences large enough to fight over: the Krishna tribunal assessed availability at 75 per cent dependability and arrived at 2,060 thousand million cubic feet, while the second Krishna tribunal, working two generations later on a longer record, assessed the basin at 65 per cent dependability and divided a different river. The Narmada tribunal proceeded on an assessed availability of 28 million acre feet at 75 per cent dependability, a figure the 1974 agreement had fixed for it. A difference of a few points in the dependability figure moves more water than many of the projects the States are building, so the arithmetic is the litigation.

The choice of dependability is not uniform across basins, and the variation is itself a finding about the system’s method. The Cauvery tribunal assessed that basin’s yield at 50 per cent dependability, arriving at the 740 thousand million cubic feet it apportioned, a markedly less conservative convention than the 75 per cent at which Krishna and Narmada were assessed, and one that builds a larger measure of expected shortfall into the award and correspondingly more work for the distress-sharing arrangement that accompanies it. Nothing in the Act prescribes the convention. Each tribunal chooses the statistical basis of its award afresh, on the parties’ rival submissions, and a State that loses on the choice has lost on a question that determined the size of the divisible pool before a single share was argued. A standing body, or a statutory data institution of the kind Sarkaria proposed, would at least have made the choice once, basin by basin, on a common record, rather than leaving the foundations of each award to be relitigated as a preliminary issue in every reference.

No authoritative common dataset stands behind the tribunal to settle these matters in advance. The Sarkaria Commission saw the point in 1988 and recommended a national data bank and information system, with a statutory obligation on States to supply data and power in the tribunal to compel it. The Act that emerged from the reform process of 2002 contains the time limits and the decree force, and no data bank. A tribunal therefore begins by adjudicating the facts it needs in order to adjudicate the dispute, receiving rival series from the parties, appointing assessors, and testing one State’s gauges against another’s, all before the first question of entitlement is reached. Its difficulties are compounded by the movement of the subject matter: a basin does not hold still during a decade of proceedings, since States continue to build, utilisation grows, and the existing uses the tribunal must protect are larger at the end of the case than at its start, so delay does not merely postpone the answer but changes the question. The protection of existing uses is itself a principle with a distributive edge that rewards the swift builder. A tribunal dividing a river does not start from an empty basin; it starts from the utilisations each State has established, which equitable apportionment treats as weighty facts, so that water a State succeeds in using while the case proceeds tends to harden into water the award will recognise. Every party’s engineers understand this, and the race to establish uses during a reference is not an abuse of the system but a rational response to its principles, which is why the data contest and the construction contest are the same contest conducted in different registers. A tribunal must therefore police the interim position, through orders of the kind the Cauvery and Mahadayi tribunals made, while knowing that its policing is partial and that the basin it finally divides will be the basin the delay has made. It is this cause, above the others, that is nobody’s fault in particular. Hydrology is genuinely hard, records are genuinely imperfect, and a tribunal that decided shares quickly on data the parties reject would produce an award neither side could implement with confidence. The cause explains why water adjudication is slow everywhere it is attempted. What it does not explain, and what the other three causes supply, is why Indian water adjudication has no mechanism that brings even a well-evidenced dispute to a close: a clock on the opinion, a consequence for the late report, a date for the Gazette, and an end to proceedings after it.

Inside a tribunal: how the record that takes years is made

Some part of the reporting interval is consumed by the sheer mechanics of a proceeding unlike any other in Indian law. A water tribunal sits with a chairman and members drawn from the higher judiciary, advised by assessors drawn from the engineering services, and its evidence is not testimony in the ordinary sense but a mountain of hydrological and project material: gauge and discharge observations reaching back decades, rainfall series, project reports for works built and works proposed, utilisation statements for every significant diversion in the basin, and the replies and counter-replies of States each maintaining that the other’s figures are unsound. The tribunal must hold this material in a form in which it can be decided on, which means directing the parties to file data in common formats, testing one State’s series against another’s at the points where they overlap, visiting the basin, and returning, often repeatedly, to the Central Water Commission and the States for records that exist in incompatible series or do not exist at all. The Cauvery reference generated a record of a size that dwarfed ordinary appellate litigation, and the Krishna references were of the same order. None of this is bureaucratic sloth. It is the cost of deciding a question whose answer is a set of numbers, where the numbers are the parties’ own and the parties disagree about them.

The mechanics explain the shape of the reporting interval even where they cannot excuse its length. A tribunal cannot sit continuously as a court does, because its members’ judicial duties, the assessors’ other work and the parties’ preparation of technical material all proceed at their own pace, and because the evidence itself is seasonal: flows are observed across water years, and a question about dependability in a dry cycle may have to wait for the cycle to declare itself. Successive benches compound the difficulty. Members retire, are elevated, or complete their terms mid-reference, and a reconstituted tribunal must master a record it did not hear made, on technical material that cannot be skimmed. The extensions that carry a reference through these interruptions are granted, as described earlier, by routine notification, and no one reading the Gazette trail of a long reference can tell where genuine evidentiary need ends and institutional drift begins, which is itself an indictment of the reporting system: a period that is extended without published reasons, year after year, gives the public no instrument by which to distinguish a tribunal working through a hard record from a tribunal that has ceased to work at all. The 2019 Bill’s proposal for a standing tribunal spoke to this failure as much as to the constitution stage, since a permanent body with a permanent registry could at least have been asked, in public and in one place, what it was doing with the years.

The dispute during the dispute: interim orders and facts on the ground

Between constitution and award there runs a second, shadow adjudication, over what may be done while the case is being heard, and it consumes years that the stage table attributes to the tribunal’s reporting period. A water tribunal can grant interim relief, and in a dispute where the subject matter is a river in flood and recession, interim questions are not preliminary skirmishes but the substance arriving early. The Cauvery tribunal’s interim order of 1991, directing releases during the pendency of the reference, produced the gravest constitutional episode of that dispute’s middle years: Karnataka’s ordinance purporting to nullify the order, the Presidential reference, and the Supreme Court’s declaration that the ordinance was unconstitutional, a sequence that settled the tribunal’s authority to act interim while settling nothing about the shares. The Mahadayi tribunal’s interim order of 17 April 2014 restrained Karnataka from diverting or using the river’s water until a revised project report was prepared and approved and the permissions required in law obtained, an order Goa’s government was still invoking in February 2020 as the operative protection of its position while the award awaited publication and the petitions awaited hearing. Interim arrangements of this kind are necessary, since a tribunal that could not hold the position would be deciding a river that one party had already taken. But they have a cost in the accounting of delay, because every interim contest is litigated with the energy of the main case, appealed with the same energy, and remembered by both sides when the final award asks them to accept less than the interim position gave them.

The deeper difficulty is that the basin does not freeze while the lawyers argue. Irrigation departments build in the shadow of a pending reference, and each completed work presents the tribunal with an existing use claiming protection, so that the adjudication’s raw material appreciates against the claimant with every season. The first Krishna award illustrates the mechanism across a generation. The Bachawat tribunal’s report set out two schemes, of which the published decision carried Scheme A into force, while Scheme B, which addressed the sharing of surplus waters and recommended a Krishna Valley Authority to manage it, was not published with the decision. The Central Government’s attempt to evolve an implementation mechanism for the award came to nothing, Karnataka maintained that Scheme B formed part of the decision and required notification, and in 1997 Karnataka sued on the point, in proceedings in which the Supreme Court recorded the whole history of report, references and the Gazette of 31 May 1976. A tribunal that reported in 1973 was therefore still generating litigation about the content of its own decision in the last years of the century, and the second Krishna reference of 2004 grew from soil that the first award’s unfinished business had prepared. The Mahanadi dispute presents the same mechanism in live form: Odisha’s complaint of November 2016 was directed at the barrages and projects Chhattisgarh was constructing on the river, and every month of the negotiation and adjudication that followed was also a month of construction, so that the tribunal constituted in March 2018 faced a different river from the one complained of sixteen months earlier. Delay in water adjudication is not time subtracted from a fixed problem. It is time added to the problem, and the interest compounds in concrete.

The interim regime has one further effect that outlasts every order made in it. A generation of farmers plants to the pattern of releases the interim arrangements permit, and the cropping pattern of a delta, once settled, becomes a political fact of the first order: the second crop that an interim order made possible in a run of good years acquires cultivators, credit, mills and constituencies, and its removal by a final award is experienced not as the correction of a provisional arrangement but as an expropriation. Tribunals know this, and the knowledge tempers awards toward the protection of what the interim years established, which is the mechanism already described by which delay enlarges the entitlements it postpones. The interim period is therefore misnamed in the system’s own vocabulary. It is not a pause before the real distribution; it is a distribution, provisional in law and cumulative in fact, and the decades this article has counted are decades in which rivers were governed, crops were grown and cities were supplied under it. The final award arrives, when it arrives, not to divide an untouched river but to regularise, at a discount, the division the delay itself has made.

What the delay is not: the case for the system’s defenders

An honest account must give the counter-reading its strength, because part of the delay is the price of things worth having. Negotiation, whose open-endedness this article has counted as the first cause, has real settlements to its name. The Godavari award rested on agreements the riparian States negotiated among themselves in the late nineteen seventies. The Narmada tribunal’s task was cut down by the agreement of July 1974, reached with the Prime Minister’s assistance, which fixed the assessment of available water and withdrew appeals then pending. The Ravi and Beas tribunal itself was a creature of settlement, constituted to implement the water paragraphs of the Punjab accord of 1985. A Centre obliged to constitute a tribunal on every complaint, or within a year of one regardless of the state of talks, would harden positions earlier and foreclose some of these outcomes, and the States that demand speed in a dispute they expect to win are not always the States that demanded it in the last one. Nor is adjudication itself the failure it is sometimes drawn as. The Krishna award of 1976 was accepted and worked for a quarter of a century. The Godavari award was published and governed. The Narmada allocation survived, and the machinery built under it, the Narmada Control Authority, shows that an implementation body created at the time of the award can hold an allocation in place across decades. Measured against the alternative of no machinery at all, in which every inter-State river quarrel would be a standing invitation to unilateral construction, the Act’s record contains durable successes.

The counter-reading also has an institutional point that reformers glide past. A large share of the post-award delay is not delay in reaching a decision but delay in accepting one, and that is a feature of the federal politics of water rather than of the Act’s drafting. A State government that loses water loses it visibly, to named districts and crops, and gains nothing it can campaign on by complying promptly. The incentive to litigate, to seek references, to await a friendlier season in Delhi, is built into the subject matter, and it would survive any conceivable amendment that left the parties as States and the stakes as rivers. These considerations, taken together, explain a great deal of the observed duration: the genuine difficulty of the facts, the genuine value of settlement, and the genuine politics of acceptance. What they do not explain is the design choice to leave the explanation’s beneficiaries without any outer bound. Negotiation can be valuable and still be given a year, or three, after which a tribunal is constituted whether or not the Centre has pronounced negotiation dead. A report can require references and still be published on a date. The counter-argument establishes that some delay is the cost of a federal system dividing rivers. It does not establish that the cost must be unbounded, and on the evidence of the nine tribunals the unbounded stages, not the bounded ones, are where the decades are.

One structural feature holds the whole pattern in place and deserves separate statement: the parties are repeat players, and their positions reverse from basin to basin. Karnataka, the upper riparian resisting releases on the Cauvery, is the middle riparian claiming against Maharashtra on the Krishna and the State whose projects Maharashtra and Goa resist on the Mahadayi. Madhya Pradesh, the generous contributor of the Narmada settlement, is a claimant against Maharashtra on the Godavari. Odisha, the lower riparian complaining of Chhattisgarh’s barrages on the Mahanadi, is the upper riparian whose objections Andhra Pradesh overcame on the Vansadhara. No State experiences the system from one side only, and the consequence is a curious stability in the rules of the game. A precedent that hurries adjudication along in the basin where a State is the sufferer will hurry it in the basin where the same State is the beneficiary of delay, and the States’ collective advocacy, in the Inter-State Council and in the debates on reform, reflects that ambivalence: every government is for speed in the dispute it is losing and for deliberation in the dispute it is winning. The Centre’s hesitation, described earlier as a first cause, is in part the sum of these State ambivalences arriving in Delhi. It follows that reform cannot be carried by the riparian interest; it has to be carried, if at all, by the general interest in decided questions, which is diffuse, unorganised and represented in the process chiefly by the courts when a State or a litigant contrives to bring them in. The history bears this out: the two accelerations in the entire record, Cauvery in 1990 and Mahanadi in 2018, were both court-made, and the one legislative attempt at acceleration, the Bill of 2019, died unattended in a House whose members’ States had no common interest in reviving it.

The 2019 attempt to rebuild the machinery, and its unusual death

The standing tribunal proposal answers three of the four causes directly, and its career in Parliament is the second finding of this article. The Inter-State River Water Disputes (Amendment) Bill, 2019 was introduced in the Lok Sabha on 25 July 2019 and passed by that House on 31 July 2019, six days later. Its design replaced the ad hoc tribunal, constituted dispute by dispute, with a single standing Inter-State River Water Disputes Tribunal, composed of a chairperson, a vice-chairperson and other members, sitting in benches as required. The existing tribunals were to stand dissolved and their pending disputes transferred to the new body, so that the accumulated experience of the institution would no longer be wound up with each award. Before adjudication, a dispute was first to go to a Disputes Resolution Committee, constituted by the Central Government from persons of experience in the field, which was to attempt a negotiated settlement within one year, extendable by six months, so that the negotiation stage acquired for the first time a period of its own and an institutional owner. A dispute the Committee could not settle would pass to the Tribunal, which was to give its decision within two years, extendable by one. The decision was to be final and binding and to carry the force of an order or decree of the Supreme Court. The Bill also returned to the Sarkaria Commission’s unfinished business, proposing a transparent system of data collection and a data bank for each river basin, to be maintained through an agency the Centre would appoint. On paper, the open opinion stage was closed by the Committee’s deadline, the constitution stage was abolished altogether, the reporting period was shortened, and the data cause was met by the one reform the 2002 round had left behind.

The Bill then died in a manner almost no published account records. After its passage in the Lok Sabha it was found that a further amendment to one of its clauses was required: the provision transferring pending proceedings had not dealt with the Ravi and Beas reference, whose tribunal rests on its own footing in the existing scheme, and without a correction the transfer machinery would not reach it. An amendment of that character, moved in the Council of States, required the approval of the Cabinet, and Cabinet approval for it was not obtained, as Ministry sources recounted when the Bill’s position was reported in July 2022. The Bill was therefore never taken up in the Rajya Sabha at all. It was not defeated, not amended, not referred to a committee and not withdrawn. It lapsed on the dissolution of the Seventeenth Lok Sabha, a lapse the Rajya Sabha’s bulletin recorded in June 2024. The single standing tribunal, which had passed the House of the People within a week of introduction, ceased to exist as a legislative proposal because a drafting correction could not be carried through the executive’s own clearance process. The episode belongs in any account of why water adjudication reform does not happen: the obstacles are not only federal resistance or partisan disagreement, of which the Bill encountered remarkably little on the record, but the ordinary friction of a legislative machine in which a Bill between the Houses is nobody’s responsibility, and in which a defect discovered after passage has no owner charged with repairing it.

The flaw that killed the Bill deserves one further paragraph, because its content is diagnostic. The Ravi and Beas reference was not an ordinary pending matter that a transfer clause could sweep up with the rest. Its tribunal had reported in 1987 and existed, on yearly extensions, for the single purpose of a further report, and its subject matter was entangled with the Punjab settlement, with agreements a State had purported to terminate, and with a Presidential reference on which the Supreme Court had advised in 2016. A clause drafted to transfer the proceedings of the existing tribunals to the new standing body, in terms that did not reach this singular case, would have dissolved the machinery around the longest-running reference in the system while leaving the reference itself legally stranded, its tribunal gone and its successor unseised. The correction was therefore not a technicality; it went to the Bill’s central promise, which was continuity. That the correction could not be made is the point on which the episode turns. Between the two Houses a Bill is amended by the government that owns it, on the advice of the ministry that drafted it, with the approval of the Cabinet whose programme it carries, and each of those steps was available at any time across the five years the Bill waited. The water disputes system is often described as one in which nothing is anyone’s responsibility for long. Its reform died of exactly that condition.

The merits of the standing tribunal deserve the same two-sided treatment as the delay itself. For the proposal it is said, correctly, that a permanent body would retain the expertise that the ad hoc system discards: the assessors, registry staff and accumulated hydrological record that a Krishna or Cauvery tribunal builds over a decade are dissolved with it, and the next tribunal begins by relearning the basin science its predecessor mastered. A standing body also removes the constitution stage entirely, the stage at which the Mahadayi dispute spent eight years, and it permits a uniform procedure and a consistent jurisprudence of allocation across basins, where the ad hoc system has produced tribunals that differ in method from one river to the next. Against the proposal it is said, also correctly, that a single tribunal for every basin in India could become a bottleneck of a new kind, its benches queued with disputes each of which demands years of attention, and that permanence did not feature in the Bill alongside any answer to the post-award problem: the special leave route, which the record identifies as the principal source of delay after report, would have run against the standing tribunal’s awards exactly as it ran against the ad hoc ones, and the publication stage would still have awaited the Centre’s convenience. The Bill was a serious attempt on three of the four causes and silent on the stage where the Cauvery award spent six years and the Krishna report has spent longer. Its death by drafting defect, unrepaired across the remainder of a Parliament, is in its own way the emblematic water dispute outcome: a decision reached, a correction identified, and no period within which anyone was required to act.

From tribunal to Supreme Court judgment, how long did the Cauvery dispute run?

The Cauvery tribunal was constituted on 2 June 1990 and gave its final award on 5 February 2007. The award was published in February 2013, under a Supreme Court direction, and the Court’s judgment on the appeals came on 16 February 2018: twenty-seven years and eight months from constitution to judgment.

The arithmetic is worth setting against the Minister’s figure of twenty-nine years for the same dispute, given to the Lok Sabha in 2019, because the difference between the two numbers is precisely this article’s subject. The Ministerial figure runs onward from the judgment to the constitution of the Cauvery Water Management Authority in June 2018 and the continuing work of implementation, and on that basis it too is accurate. A dispute’s duration is not a fact but a measurement, and the Cauvery dispute can be stated at seventeen years to its award, at twenty-three to its publication, at twenty-eight to the judgment that closed the appeals, or at twenty-nine and counting to the machinery that administers the outcome, each figure truthful and each answering a different question. What no honest version of the arithmetic can produce is a figure anywhere near the five years of the statutory design, because the design’s interval is the only one the dispute did not use up several times over.

After the award: enforcement, boards and the limits of a decree

An award that has been published is, since 2002, final and binding and possessed of the same force as an order or decree of the Supreme Court. The sentence sounds like the end of the matter, and it is worth being precise about what it does and does not supply. It supplies the award with the juridical status of a decree, which means that disobedience is disobedience of an instrument of the highest legal rank, and that the Supreme Court’s contempt jurisdiction stands behind it in principle. It does not supply a bailiff. The Act contains no penalty for a State that declines to release water in accordance with an award, no machinery of execution against a State government, and no officer charged with measuring the flows at the border and reporting default. Section 6A offers the Centre a power, not a duty, to frame an implementation scheme and constitute an authority, and across the life of the Act that power slept until the Supreme Court woke it in the Cauvery case in 2018. The result is an enforcement regime whose strength is entirely borrowed: borrowed from the Court’s authority, which the Act elsewhere purports to exclude, and from the Centre’s willingness to spend political capital in a basin where it usually has reasons of its own to hesitate.

Where implementation machinery has been written into law from the start, the contrast is visible. The Andhra Pradesh Reorganisation Act, 2014, dividing a State whose rivers were already under adjudication, did what the water disputes Act does not: it constituted the Krishna River Management Board and the Godavari River Management Board as standing bodies charged with the administration of the successor States’ water arrangements, and by Section 89 it extended the term of the second Krishna tribunal and directed it to make project-wise allocations between Andhra Pradesh and Telangana, as the Andhra Pradesh Reorganisation Act, 2014 records in its own scheme. Implementation there was not left to a discretionary scheme framed after the award; it was legislated alongside the allocation process itself, by a Parliament that had plainly learned, from the Cauvery experience running concurrently, what an award without machinery is worth. The Narmada Control Authority offers the older version of the same lesson, created to implement an award at the time the award was made and credited with holding the Narmada allocation in working order thereafter. The pattern is consistent enough to state as a rule of the record: awards accompanied by a standing implementation body acquire a working life, and awards left to the parties’ goodwill acquire further proceedings.

The Cauvery Authority’s working record shows what an implementation body does with an allocation once it exists, and why its absence elsewhere matters. In the years following its constitution on 1 June 2018, the Authority and its Regulation Committee settled into the unglamorous routine the basin had lacked for three decades: the Committee met repeatedly through each water year to monitor the hydro-meteorological position, sixty-one of its meetings having been held by April 2022 on the Ministry’s account, and prepared the annual and seasonal water accounts for the basin, while the Authority apportioned releases month by month, including the monthly apportionment of the seven thousand million cubic feet due to Puducherry, agreed at its meeting of 25 February 2020. None of this is adjudication, and none of it makes headlines of the kind the tribunal made. It is the conversion of an award from a judgment into an administration, and it is precisely the stage the Act left to a discretionary scheme and a direction that took twenty-eight years to arrive. Where the Authority’s existence has been tested, in the distress seasons when the monsoon fails over the catchment, the argument between Karnataka and Tamil Nadu has continued, but it has continued inside a body charged with the figures, before a committee holding the basin’s accounts, rather than as a fresh dispute about whether the award binds at all. That difference, between arguing about implementation and arguing about obligation, is the whole of what enforcement machinery buys.

The distress seasons of the Cauvery dispute before the Authority existed show the same work being done badly by improvisation. In the dry season of 2012, with the award published only the following year and no implementation body in being, the quarrel over releases ran through the Cauvery River Authority, a monitoring body chaired by the Prime Minister, and then through the Supreme Court, which directed daily releases to Tamil Nadu while Karnataka’s government protested that its own reservoirs could not bear them. In September 2016 the pattern repeated at a higher temperature: interim directions for the release of Cauvery water were followed by violence in Bengaluru and a curtailment of the releases the Court had ordered, with the Court adjusting its directions as the season and the unrest developed. These episodes are not adjudication and not enforcement; they are the management of a live river by a court and a committee in the absence of the machinery the statute made optional, conducted under the pressure of a failing monsoon and a standing crop. They belong in the ledger of delay because they are what delay purchases: not the postponement of conflict but its recurrence, season after season, in the least deliberative forum available.

The Authority’s sternest test is the distress year, and the distress year exposes a final dimension of delay that no stage table captures: a water dispute does not age evenly. In a normal year an unresolved allocation is an abstraction argued by advocates. In a dry year it is a standing crop in the delta and a city’s reservoirs, and the political temperature of the dispute rises with the thermometer and falls with the reservoirs. The Cauvery tribunal anticipated the problem in its award, providing that in years of distress the available water be shared in proportion to the shares it had fixed, and the Supreme Court retained that arrangement in 2018, leaving its operation to the Authority and its Regulation Committee, which monitor the hydro-meteorological situation through the season and apportion the shortfall. The machinery works because it exists before the crisis. Where no machinery exists, the dry year produces the crisis politics that have repeatedly surrounded the Cauvery and the Mahadayi: emergency sittings, demands for the Prime Minister’s intervention, litigation moved at the pace of a failing monsoon before courts whose calendars are set in ordinary time. A system’s delay is therefore not fully measured by the years it takes. It is measured also by where in the cycle its failures land, and a process that spends its decades in the ungoverned stages will reliably deliver its hardest questions to the driest seasons, when they are least capable of patient resolution.

The Sutlej-Yamuna Link dispute is the limiting case of enforcement, because there the instrument defied was not an award but a series of agreements and the decrees of the Supreme Court itself. Haryana sued Punjab in the Court’s original jurisdiction, and the Court decreed in 2002 that Punjab complete its portion of the canal, returning to the matter in 2004 as construction stalled. Punjab’s answer was the Termination of Agreements Act of July 2004, annulling the agreements on which its obligation rested, and the Presidential reference that followed ended in 2016 with the Court advising that the enactment was constitutionally invalid. Twelve years separated the Act from the opinion that undid it, and at the end of them the canal remained unbuilt, because a decree against a State that controls the territory, the police and the construction season is executed, in the end, by politics or not at all. This is the honest outer boundary of the enforcement discussion. The water disputes system can produce decisions and can clothe them with the highest legal force the Constitution knows. It cannot compel a State government to open a headgate, and any account of tribunal delay that stops at the award has examined only the delays that institutions can, in principle, repair.

A tribunal that dissolves: institutional capacity and the standing alternative

Running beneath the four causes is a quieter one, of institutional design. A water disputes tribunal is an ad hoc body. It is assembled for one dispute, with a chairman and members drawn from the sitting or retired judiciary, assessors borrowed from the engineering services, a registry improvised for the purpose, and it is wound up when its work is done. The expertise it accumulates is basin-specific and unusually deep: by the end of a long reference its members and assessors command the flow records, the project histories and the utilisation data of a river system more completely than any other body in the country. All of it disperses at dissolution. The contrast with the standing adjudicatory institutions Parliament has built in other fields is instructive: in company law, the National Company Law Tribunal and its appellate counterpart sit permanently, accumulate a jurisprudence, and carry their registries and records from case to case, so that the hundredth matter before them benefits from the ninety-nine that preceded it. Water adjudication, dealing with subject matter at least as technical and stakes at least as large, restarts from nothing with every dispute. The second Krishna tribunal could draw on the first’s published award but not on its working knowledge, its tested data or its staff, a generation of learning having been formally closed in 1976.

The system did once, briefly, experiment with continuity. The tribunals of April 1969 for the Krishna and the Godavari were constituted on the same day under a common chairman, Justice R. S. Bachawat, so that the two great Deccan references proceeded in parallel under one judicial mind, with the basins’ interlocking claims on the same States argued within a single institutional memory. The experiment was not repeated and was not converted into a design: when the Krishna returned for its second reference in 2004 it came before a new tribunal with no inheritance from the first beyond its printed award, and the Godavari’s settlement was left to stand on the agreements it recorded. The episode matters because it disposes of the suggestion that continuity is foreign to the Act’s tradition. The tradition contains the seed of the standing model, planted in 1969 and never cultivated, and the half century since has been a demonstration, conducted at the scale of nine tribunals, of what the ad hoc alternative costs.

The comparison that stings most is not with company law but with the water sector’s own statute book. In the same year that Parliament enacted the water disputes Act, it enacted the River Boards Act, 1956, empowering the Centre to establish boards for inter-State rivers that would advise the riparian governments, coordinate development and help prevent disputes from arising at all. No river board was ever brought into being under it. The preventive institution was legislated and left unbuilt, while the adjudicatory institution was built, dispute by dispute, and disbanded after each. The authorities that do exist in the sector, the Narmada Control Authority, the Cauvery Water Management Authority, the Krishna and Godavari River Management Boards, were each created outside the 1956 design, by an award, by a court direction or by a reorganisation statute, which is to say by improvisation after the event. A legislative scheme that contains an unused preventive Act, an adjudicatory Act without clocks at its ends, and a shelf of improvised implementation bodies is not a scheme that failed in execution. It is a scheme whose working parts were never assembled, and the decades are the sound of the unassembled machine.

The staffing model compounds the continuity failure. A tribunal is chaired by a judge drawn from the higher judiciary and staffed by members of the same cadre, which gives its award an authority no body of administrators could command, and the price of that authority is that the tribunal’s scarcest resource, judicial time at the summit of a State’s legal system, is borrowed rather than owned. A sitting judge carries a court’s docket beside the reference; a retired judge carries a tenure that the reference will usually outlast, whereupon the membership turns over mid-case and the record must be re-mastered by a successor who did not hear it made. The assessors who supply the tribunal’s engineering judgement are drawn from the services on deputation, with postings that rotate on the services’ own schedule rather than the reference’s. Every element of the staffing is therefore temporary in a different way and on a different clock, and the reference, which is permanent until it ends, must be carried across the joins. A standing tribunal under the 2019 design would not have abolished these frictions, since its members too would have been drawn from the judiciary, but a permanent registry and a bench that sat for the class of disputes rather than for one dispute would at least have confined the turnover to the bench, leaving the institution’s memory, its data and its files in continuous custody.

Capacity also fails at the level of continuity within a single reference. Tribunals sit part-time in their members’ working lives, lose members to retirement and elevation, and require reconstitution mid-reference, with the new member obliged to master a record that may run to hundreds of volumes of evidence. The extensions recorded against the Ravi and Beas tribunal, renewed annually across decades for a further report that never arrives, are the administrative shadow of this design: a body kept notionally alive because closing it would admit that the reference had failed, and because no institution exists to which its unfinished business could be handed. A standing tribunal would not by itself shorten the stages the statute leaves timeless, and the 2019 Bill’s silence on publication and on the appellate route stands as the measure of its limits. It would, however, convert adjudication from a sequence of expeditions into an institution with a memory, and it would give the system’s unfinished business an address other than a file in the Ministry. The capacity argument and the timeline argument are distinct, and both point the same way: the Indian water dispute system is not slow because its people are slow, but because it is rebuilt for every dispute and disbanded after every award, while the disputes themselves run on.

The verdict

The record supports a verdict, and it is narrower than the standard complaint and harder for the system’s custodians to answer. The statutory limit on water dispute adjudication is real, and it is mostly not the limit that is being exceeded. The periods Parliament enacted govern the middle of a dispute’s life, and even those periods operate without consequences for their breach. The periods Parliament omitted, for the Centre to decide that negotiation has failed, for an award to reach the Gazette, and for the litigation that follows publication to run its course, are the periods in which the decades accumulate, and they were omitted by choice rather than by oversight, because the Sarkaria Commission named three of the four gaps in 1988 and the amending Act of 2002 closed only the one that cost the Centre least. The four causes compound one another: an unbounded start rewards the State the status quo favours, a directory reporting period removes any institution’s reason to hurry, an unclocked publication stage lets a finished decision wait on politics, and the appellate route converts every award into the opening of a further case. Against these, the system’s genuine achievements, the accepted Krishna award, the negotiated Godavari settlement, the implemented Narmada allocation, show that the machinery can close a dispute when the stages happen to align. A system that works when everything aligns and has no answer when anything does not is not a system under strain. It is a system whose design assumes the cooperation it exists to replace. Reform that deserves the name would attach a period to the opinion and to the Gazette, give the data a statutory home, and make the implementation authority a consequence of the award rather than a favour after it. The periods, to be worth enacting, would need the quality this article found in the one period the parties observe, the three-month window for a reference back: they would need to be conditions, with a stated consequence on expiry. An opinion not formed within a fixed time of a State’s request could be deemed formed, so that the duty to constitute arose by default rather than by grace. A report not published within a fixed time of its receipt could be deemed published, so that its force did not wait on a notification no one is charged to issue. An extension of a tribunal’s period could be required to state its reasons and to be laid before Parliament, so that the accumulation of years acquired a public author and a public account. None of these devices is exotic, and between them they would convert the statute’s duties owed by institutions to no one in particular into duties whose breach changes the legal position without anyone having to sue. The 2019 Bill attempted the first of these objects in part, through the Committee’s deadline, and the third not at all, and then lapsed, unrepaired, in the manner its subject matter would have predicted. The rivers, meanwhile, continue to be divided by the arithmetic this article has described: three years and two for the stage in the middle, and no number at all for the stages in which the decades are.

Frequently Asked Questions

Q: How long have water dispute tribunals actually taken?

Measured from constitution to published award, the completed adjudications have taken from about seven years to about twenty-eight. The Krishna tribunal of 1969 reported in December 1973 and was published in May 1976, a span of seven years. The Narmada tribunal took ten years, the Godavari tribunal eleven, and the Cauvery tribunal, counting the appeals that followed publication, occupied the system for twenty-eight years from its constitution in 1990 to the Cauvery Water Management Authority’s creation in 2018. The Minister for Jal Shakti told the Lok Sabha on 31 July 2019 that the four tribunals which had given awards took between seven and twenty-eight years, figures consistent with the record when each is tied to its endpoints.

Q: Why is the statutory time limit for a water disputes tribunal not met?

The familiar answer, that tribunals ignore the five-year limit, misstates where the time goes. The five years of Section 5(2), three years extendable by two, govern only the interval between a tribunal’s constitution and its report. That interval itself has overrun in practice, as with the second Krishna tribunal’s six years and nine months, because the Act attaches no consequence to a late report and extensions are granted serially. The larger part of every long dispute, however, sits outside the limit entirely: the years before constitution, while the Centre forms its opinion that negotiation has failed, the unbounded interval between report and Gazette publication, and the litigation that follows an award. The limit is not so much defied as surrounded by stages to which no limit applies.

Q: What delay happens before a water disputes tribunal is even constituted?

The pre-constitution stage has two halves, and only the second has a clock. Before the Central Government can constitute a tribunal it must form the opinion that the dispute cannot be settled by negotiations, and the Act fixes no period for forming that opinion. Years of conferences and correspondence can pass inside it lawfully: the Krishna and Godavari disputes lingered from 1956 to the tribunals of 1969, Goa’s request of July 2002 waited until November 2010 for the Mahadayi tribunal, and Odisha’s Mahanadi complaint of 2016 produced a tribunal in 2018 only after a Supreme Court direction. Since 2002 the Centre must constitute the tribunal within one year of the request once the opinion exists, but the opinion itself remains timeless.

Q: How does the Supreme Court route affect the finality of a tribunal award?

The Act, acting under Article 262, bars the Supreme Court and every other court from jurisdiction over a referred water dispute, and a published award is final, binding and possessed of the force of a Supreme Court decree. In practice, special leave petitions against awards have been entertained and decided, so an award’s finality is provisional until the petitions end. The Court restrained publication of the second Krishna tribunal’s report in 2011, directed publication of the Cauvery award in 2013, and in 2018 altered the Cauvery allocation itself, raising Karnataka’s share and reducing Tamil Nadu’s. An award therefore opens a judicial stage with no time limit, and finality arrives only when that stage closes.

Q: Why do states dispute the hydrological data in a water dispute?

Because the data decides the case before the law is reached. A tribunal divides a quantity: the river’s flow across a run of years, assessed at a chosen dependability. The Krishna tribunal assessed 2,060 thousand million cubic feet at 75 per cent dependability; the second Krishna tribunal, on a longer record, assessed the basin at 65 per cent. A few points of dependability move more water than many contested projects, so States fight over gauge records, flow series and utilisation figures compiled by their own irrigation departments, which do not compile them identically. With no authoritative common dataset behind the tribunal, the facts must be adjudicated first, and that adjudication is slow, technical and genuinely difficult to shorten.

Q: What happened to the 2019 water disputes amendment bill?

The Inter-State River Water Disputes (Amendment) Bill, 2019 was introduced in the Lok Sabha on 25 July 2019 and passed there on 31 July 2019. It proposed a single standing tribunal, a Disputes Resolution Committee to attempt settlement within a stated period, shorter decision timelines and a national data bank. After passage it was found that a further amendment to one clause was required, concerning the transfer of the pending Ravi and Beas proceedings, and the Cabinet approval needed for that amendment was not obtained. The Bill was accordingly never taken up in the Rajya Sabha, and it lapsed on the dissolution of the Seventeenth Lok Sabha in 2024.

Q: Would a single standing tribunal for water disputes fix the delay?

It would fix some of it. A permanent tribunal abolishes the constitution stage, at which the Mahadayi dispute spent eight years, and it retains the hydrological expertise, staff and records that ad hoc tribunals lose on dissolution, while allowing a consistent method across basins. The case against is that one body for every basin could queue disputes behind one another, and that the 2019 Bill which proposed it left untouched the stages where much of the measured delay sits: it set no effective answer to the special leave route against awards, and publication of a decision would still have awaited the Central Government. A standing tribunal is a reform of the adjudication stage. The decades accumulate in every stage, not only in that one.

Q: What happens when a state does not comply with a tribunal award?

The Act supplies status rather than machinery. A published award is final and binding with the force of a Supreme Court decree, so defiance is defiance of an instrument of the highest rank and the Court’s contempt jurisdiction stands behind it in principle. The Act, however, prescribes no penalty for non-compliance, creates no execution machinery against a State, and makes the implementation scheme under Section 6A a discretionary power of the Centre rather than a duty. In practice enforcement has depended on further litigation, as in the Sutlej-Yamuna Link decrees of 2002 and 2004, on Presidential references, and on political pressure. The Cauvery Water Management Authority exists because the Supreme Court directed the Centre to frame a scheme, not because the statute required one.

Q: How long did the Narmada tribunal take from constitution to a notified award?

Ten years and two months. The Narmada Water Disputes Tribunal was constituted on 6 October 1969, following a dispute the Sarkaria Commission recorded as already six years old. It declared its award under Section 5(2) on 16 August 1978. The Union and the four riparian States then filed references under Section 5(3), the tribunal gave its final order on 7 December 1979, and the decision was published in the extraordinary Gazette on 12 December 1979. An agreement of July 1974 among the Chief Ministers, reached with the Prime Minister’s assistance, had narrowed the tribunal’s task midway by settling the assessed availability of the river’s water.

Q: Why did Punjab enact a law terminating its river water agreements in 2004?

Punjab enacted the Punjab Termination of Agreements Act on 12 July 2004 to annul its river water agreements with neighbouring States and to discharge itself of the obligation to deliver water, including the obligation underlying the Sutlej-Yamuna Link canal, which the Supreme Court had decreed Haryana entitled to have completed in judgments of 2002 and 2004. Punjab’s stated case rested on riparian ownership of its rivers and on changed circumstances since the agreements. The President referred the Act’s validity to the Supreme Court on 22 July 2004, and the Court’s advisory opinion of 10 November 2016 held the enactment constitutionally invalid, since a State legislature cannot by unilateral law override obligations and decrees of that character.

Q: Did the 2002 amendment to the water disputes Act make tribunals faster?

Not on the evidence of the tribunals constituted under it. The 2002 amendment, in force from 28 March 2002, introduced the one-year period for constitution after the Centre’s opinion, the three-year reporting period extendable by two, and the decree force of a published award, carrying the Sarkaria Commission’s core recommendations into the Act. The second Krishna tribunal, constituted in 2004, took six years and nine months to report, on successive extensions. The Vansadhara tribunal took seven and a half years. The amendment changed the vocabulary of delay, giving every overrun a statutory instrument to be measured against, but it attached no consequence to breach and left the stages before constitution and after report without periods, so the durations it governed continued to expand.

Q: What was the Disputes Resolution Committee that the 2019 bill proposed for water disputes?

It was the Bill’s answer to the unbounded negotiation stage. Under the Inter-State River Water Disputes (Amendment) Bill, 2019, before any adjudication the Central Government would constitute a Disputes Resolution Committee for the dispute, drawn from persons with experience in the field, and the Committee would attempt to resolve the dispute by negotiation within one year, a period extendable by six months. Only a dispute the Committee failed to settle would pass to the standing Tribunal the Bill proposed. The design gave the settlement stage an institutional owner and a deadline for the first time, so that negotiation could no longer absorb years without any body being charged with concluding it. The Bill lapsed before the Committee ever existed.

The Commission’s report, in its seventeenth chapter, made five recommendations. Constitution of a tribunal should be mandatory within one year of a State’s application under Section 3. The Union should be empowered to constitute a tribunal of its own motion where a dispute exists in fact. A data bank and information system should be established at the national level, with States required to supply data and the tribunal vested with powers to obtain it. A tribunal’s award should become effective within five years of the tribunal’s constitution, extendable by the Union on the tribunal’s reference. And an award should carry the same force and sanction as an order or decree of the Supreme Court. Four of the five were accepted in the Inter-State Council process, and the 2002 amendment enacted the time limits and the decree force.

Q: How many years passed between Goa asking for a tribunal and the Mahadayi tribunal being set up?

Eight years and four months. Goa sent its request to the Central Government in July 2002 under Section 3 of the Act, a date the Jal Shakti Ministry confirmed in a Lok Sabha answer on 29 July 2021. Efforts at negotiated resolution followed, as the Act’s design contemplates, and the Mahadayi Water Disputes Tribunal was constituted on 16 November 2010. The interval fell entirely within the stage for which the Act prescribes no period, the formation of the Centre’s opinion that the dispute could not be settled by negotiations, and it exceeds, by itself, the whole of the five-year period the statute allows for the adjudication that followed it.

Q: Why was the second Krishna tribunal’s award kept from publication?

Because the Supreme Court restrained it. The second Krishna tribunal, constituted on 2 April 2004, gave its report on 30 December 2010. Andhra Pradesh challenged the report by special leave petitions, and on 16 September 2011 the Court ordered the Central Government not to publish the decision in the Gazette while the petitions were heard. Publication is the act that makes an award binding, so the restraint suspended the award’s legal life altogether. References under Section 5(3) produced a further report on 29 November 2013, and the bifurcation of Andhra Pradesh in 2014 then reopened the allocation between the successor States, with Section 89 of the Andhra Pradesh Reorganisation Act extending the tribunal’s term for that purpose.

Q: What did the Supreme Court change in the Cauvery tribunal’s allocation in 2018?

By its judgment of 16 February 2018, the Court raised Karnataka’s share of the Cauvery waters to 284.75 thousand million cubic feet, an increase of 14.75 over the tribunal’s award, and reduced Tamil Nadu’s share correspondingly to 404.25. Kerala’s 30 and Puducherry’s 7 were left undisturbed, as were the tribunal’s reservations for environmental purposes and for flows to the sea, and the monthly releases to be measured at Billigundlu on the Tamil Nadu border were reduced to 177.25 thousand million cubic feet. The Court grounded the increase in Bengaluru’s drinking water requirement and in the tribunal’s underestimation of available groundwater in Tamil Nadu, and it directed the Centre to frame an implementation scheme under Section 6A.

Q: What is the Cauvery Water Management Authority and when was it created?

The Authority is the implementation body constituted under the Cauvery Water Management Scheme, which the Central Government notified on 1 June 2018 in exercise of the power in Section 6A of the Act, together with the Cauvery Water Regulation Committee that assists it. It was created to give effect to the Cauvery tribunal’s decision as modified by the Supreme Court’s judgment of 16 February 2018, the judgment having directed the Centre to frame such a scheme. The Authority supervises the regulation of releases among Karnataka, Tamil Nadu, Kerala and Puducherry under the modified allocation. It is the only implementation authority of its kind produced under the Act, and it arrived twenty-eight years after the tribunal it serves was constituted.

Q: Have any tribunal awards under the Act been accepted and worked without a court fight?

Yes, and the exceptions are instructive. The first Krishna tribunal’s award, published in May 1976, was accepted by Maharashtra, Karnataka and Andhra Pradesh and governed the basin’s use for the remainder of the century, its own review clause scheduling the reopening that came after 2000. The Godavari award of 1980 largely embodied agreements the riparian States had themselves negotiated, and it was published and worked on that basis. The Narmada award was implemented through the Narmada Control Authority, a standing body created for the purpose, and its allocation held even while other aspects of the Sardar Sarovar project were litigated for decades. The common feature of the awards that worked is machinery or consent built alongside them, rather than legal force alone.

Q: Can an inter-State water dispute be settled without a tribunal at all?

Yes, and some of the Act’s most durable outcomes were settlements that a tribunal later recorded. The Act itself assumes settlement is possible: constitution of a tribunal waits on the Centre’s opinion that the dispute cannot be settled by negotiations. The Godavari States negotiated the agreements of the late nineteen seventies that the tribunal’s award then embodied. The Narmada dispute was narrowed by the Chief Ministers’ agreement of July 1974. The Ravi and Beas tribunal existed to implement the Punjab settlement of 1985. The Mahanadi dispute passed through a Negotiation Committee in 2017 before its tribunal. Settlement fails as a system not when it is attempted but when it is attempted without a deadline, since the party favoured by the existing distribution has little reason to conclude.

Q: Which water dispute has been pending the longest under the Act?

The Ravi and Beas reference has the longest unfinished life inside the Act’s machinery. Its tribunal was constituted on 2 April 1986 and reported with unusual speed, on 30 January 1987. The references for explanation and clarification forwarded to it in August 1987 under Section 5(3) have never been answered, and the tribunal’s tenure has been extended year by year to receive a further report which, on the Jal Shakti Ministry’s parliamentary answers, remained outstanding more than three and a half decades later. The Minister for Jal Shakti described the matter to the Lok Sabha in 2019 as pending for thirty-three years. The underlying quarrel over the Sutlej-Yamuna Link canal is older still, running through agreements of 1976 and 1981, decrees of the Supreme Court, the Punjab Termination of Agreements Act of 2004 and the advisory opinion that struck it down in 2016.