Nine judges, four votes: the Jai Bir Singh judgment on the definition of “industry”
The Court sat a bench of nine to settle what counts as an “industry”, wrote 489 pages across five opinions, and left the law exactly where it found it. Here is what happened, and why the new test everyone is talking about binds nobody.
Case State of U.P. v. Jai Bir Singh & connected mattersUnder review Bangalore Water Supply v. A. Rajappa (1978)Bench Surya Kant, CJI · B. V. Nagarathna, J · P. S. Narasimha, J · Dipankar Datta, J · Ujjal Bhuyan, J · Joymalya Bagchi, J · Satish Chandra Sharma, J · Alok Aradhe, J · Vipul M. Pancholi, JReading time about 20 minutes
Some questions in Indian labour law refuse to die. Whether a hospital, a club, a university, a temple or a forest department is an “industry” under section 2(j) of the Industrial Disputes Act, 1947 is the oldest of them. It went to a bench of five in 1953, a bench of six in 1970, a bench of seven in 1978, and finally to a bench of nine that heard argument over three days in March 2026 and delivered judgment on 20 August.
By then the provision it was asked to interpret had already been repealed. The Industrial Relations Code, 2020 came into force on 21 November 2025 with a new definition of its own. The nine judges were left holding a question about a section that no longer exists, in appeals filed as far back as 2001, on behalf of workmen and employers who had been waiting more than two decades for an answer.
What they produced is unusual. The Chief Justice, writing for himself and three colleagues, took apart the 1978 test, said it had strayed into territory reserved for Parliament, and set out a rewritten version. He then withdrew it. The rewritten test is described in the judgment itself as “only a hypothesis” ¶275 and as “no more than our considered opinion” ¶287(iii). Every pending case, the same paragraph says, goes on being decided under the old test.
Four of the remaining five judges said the old test was right and should not be touched. The fifth said the question no longer needed answering at all. Add it up and the celebrated “calibrated triple test” has the support of four judges out of nine. It is not the ratio of the case. It is not law.
Figure 1 · The bench
Surya Kant, CJI
Lead opinion · 154 pp · ¶¶1–290
Proposes the new test
Satish Chandra Sharma, J
Joins the Chief Justice
Proposes the new test
Alok Aradhe, J
Joins the Chief Justice
Proposes the new test
Vipul M. Pancholi, J
Joins the Chief Justice
Proposes the new test
B. V. Nagarathna, J
Separate opinion · ¶¶1.1–22
Keeps the 1978 test
P. S. Narasimha, J
Separate opinion · 18 pp · ¶¶1–27
Declines the merits
Dipankar Datta, J
Separate opinion · 75 pp · ¶¶1–82
Keeps the 1978 test
Ujjal Bhuyan, J
Joins Datta, J
Keeps the 1978 test
Joymalya Bagchi, J
Separate opinion · 85 pp · ¶¶1–125
Keeps the 1978 test
Five opinions from nine judges. The largest single block is four. On the rewritten test the split is four in favour, four against and one who would not decide, so no version of the count produces a majority for change.
Part one
A forest department and a definition
The lead appeal came out of Uttar Pradesh. The State ran a Social Forestry Department, set up to plant and maintain trees and improve the environment. Workers engaged on it were dismissed and went to the labour court. The State said the department was not an “industry”, so the labour court had no jurisdiction over it and the workers were not “workmen”.
Everything turns on one sentence in the old Act:
“industry” means any business, trade, undertaking, manufacture or calling of employers and includes any calling, service, employment, handicraft, or industrial occupation or avocation of workmen.
Section 2(j), Industrial Disputes Act, 1947
Two halves. The first looks at what the employer does. The second looks at what the workman does. Every word in it is elastic. Narasimha, J puts the problem better than anyone else on the bench: the clause “is written in the grammatical shape of a definition; whether it actually behaves like one has been the question” Narasimha J, ¶8. Read one way it covers a factory and a family kitchen alike. Read another way it leaves out people the Act plainly meant to protect.
He also counted the damage. Benches of four, five, six and seven judges have tried to fix the meaning, alongside at least fourteen three-judge decisions and more than twenty-seven two-judge decisions Narasimha J, ¶7. Nearly eighty years of litigation over the threshold question of whether the tribunal may hear the case at all.
Figure 2 · Seventy-nine years
The line traces how wide or narrow the reading was at each stage. It swings down through the 1960s, hits bottom in Safdarjung Hospital, jumps in 1978 and stays there for forty-seven years. Parliament tried to pull it back in 1982; the executive never notified the amendment, so the line never moved.
Part two
What 1978 actually decided
In Bangalore Water Supply and Sewerage Board v. A. Rajappa, seven judges wrote four opinions. Krishna Iyer, J spoke for himself, Bhagwati and Desai, JJ. Beg, CJ concurred. Chandrachud, J concurred separately, and his reasons came seven weeks after the main opinion. Jaswant Singh and Tulzapurkar, JJ dissented in part. The working majority was five to two.
The core of it is three conditions, which came to be called the triple test. An enterprise is prima facie an industry where there is systematic activity, organised through cooperation between employer and employee, producing or distributing goods and services that satisfy human wants. Profit is beside the point. So is the sector. So is the motive of the person running it.
Krishna Iyer, J then added the parts people forget. “Undertaking” was read down, so that a body outside trade or business could still qualify only if the way it organised employer and employee resembled trade or business. Composite bodies were judged by their dominant nature, taking the enterprise as a whole. Sovereign functions, “strictly understood”, alone escaped, and welfare work by the State did not. Two express exemptions were carved out: ventures that hire employees only in “minimal matters”, and genuinely eleemosynary missions such as free legal clinics, doctors giving spare hours, or an ashram, “even if stray servants, manual or technical, are hired”.
Those two carve-outs matter for what follows.
Figure 3 · The change the Chief Justice proposed
One gate is added and two escape routes are closed. The added gate narrows the test for anything that looks commercial. Closing the two exits widens it for the free clinic and the small charitable body, which under the 1978 guidelines had an express way out. The lead opinion says its purpose is to protect exactly those bodies ¶203.
Part three
How a forest department reached nine judges
The route is worth setting out, because three judges would have thrown the whole reference out on it.
In 1998 two judges in Coir Board, Ernakulam v. Indira Devi P.S. said the 1978 decision had done “more damage than good” and asked for a larger bench. A bench of three, presided over by the Chief Justice, refused: a two-judge bench is bound by a bench of seven and cannot ask for it to be reconsidered.
Four years later a different problem arose. Three judges in Chief Conservator of Forests v. Jagannath Maruti Kondhare had held a State social forestry scheme to be an industry. Two judges in State of Gujarat v. Pratamsingh Narsinh Parmar had held a forest department not to be one. Three judges in Jai Bir Singh saw a conflict and sent the matter up.
The bench of five that received it in 2005 did something else. Instead of reconciling the two decisions or overruling one, it questioned the 1978 decision itself, gave seven reasons, and asked the Chief Justice to constitute a larger bench. In 2017 a bench of seven read those reasons, agreed the doubts were serious, and sent the case to nine. There it sat until 2023, when directions were issued for compiling the papers, and then went quiet again until 2026.
Figure 4 · The route to nine judges
The left branch is the one the Court accepted. The right branch is the one it refused in 1998, and it is the comparison Datta and Bhuyan, JJ press: a bench constituted to resolve a conflict between two smaller benches turned itself into a challenge to a bench of seven.
Six of the nine held the route sound. The Chief Justice went through each step against Central Board of Dawoodi Bohra Community v. State of Maharashtra, found no shortcut, and pointed out that the Chief Justice of India has independent power under Order VI Rule 2 of the Supreme Court Rules to constitute a bench of any strength. Narasimha and Bagchi, JJ agreed.
Three did not. Nagarathna, J read Kondhare and Pratamsingh Parmar side by side and found no conflict at all. The Gujarat case had failed on pleadings, not on principle: the workman had never said what the establishment did or what duties he performed, so the earlier ruling simply had nothing to attach to. On that reading the 2005 bench was chasing a conflict that was never there, and it “adopted a third and strange path of questioning the correctness of the judgment of the seven-Judge Bench” Nagarathna J, ¶18.6.
Datta, J, writing also for Bhuyan, J, went further and tested the 2005 order against Keshav Mills, Shah Faesal and Srimanth Balasaheb Patil, then took its six reasons apart one by one. On the complaint that the 1978 decision was worker-oriented, he answered that a judgment “cannot be characterised as ‘worker-oriented’ merely because its ultimate outcome” favours workers Datta J, ¶38(ii). On the argument that Parliament's unenforced 1982 amendment showed disapproval, he answered that treating an executive decision not to notify as legislative disapproval “is to invert separation of powers”.
Part four
The rewritten test
The Chief Justice's objection to 1978 is narrower than the language around it suggests. He does not say the triple test is wrong. He says it looks at the wrong thing first.
By setting aside noscitur a sociis as an aid to reading section 2(j), the 1978 bench cut the word “industry” loose from the commercial words sitting beside it. What followed, on this account, was that the inquiry stopped asking what the enterprise does and started asking only how it is staffed. “The focus moves away from what the enterprise does and towards how it is organised” ¶219. Bodies with nothing in common except a payroll ended up in the same statutory box.
So commercial character comes back, as what the judgment calls “a principled sieve” ¶230. Profit stays irrelevant; an activity can be commercial in character without chasing a surplus ¶234. The two questions, what the body does and how it engages its staff, become separate pillars rather than one. “To treat them as interchangeable is akin to saying that apples and oranges are the same fruit” ¶253.
Here is the rewrite set against the original, limb by limb.
Limb
1978
2026 rewrite
Change
I(a)(i)
“systematic activity”
“a systematic activity”
Same
I(a)(ii)
organised by cooperation between employer and employee (the direct and substantial element is chimerical)
organised through cooperation between employer and employee
Rider dropped
I(a)(iii)
production or distribution of goods and services calculated to satisfy human wants and wishes. Result: prima facie there is an industry
the output must carry a discernible commercial character that is analogous to trade or business and satisfy material human wants. Result: a prima facie presumption arises
The main change
I(b)
absence of profit motive is irrelevant, in any sector
kept word for word
Same
I(c)
the decisive test is the nature of the activity with special emphasis on the employer–employee relations
the decisive test is the nature of the activity and the existence of employer–employee relations
Now conjunctive
I(d)
a trade or business does not stop being one because of philanthropy animating the undertaking
… does not stop being one merely because philanthropy animates it in its official documentation but on enquiry it is found otherwise
Meaning reversed
II(a)
… provided the nature of the activity, viz. the employer–employee basis, resembles trade or business
… provided the nature of the activity taken together with the employer–employee basis resembles trade or business
Two tests, not one
III(a)
professions, clubs, schools, cooperatives, research bodies and charities are covered if the triple test is met
kept, renumbered (i) to (vii)
Same
III(b)
exemption where the venture hires employees only “in minimal matters”
deleted
Removed
III(c)
exemption for eleemosynary missions, “even if stray servants, manual or technical, are hired”
deleted
Removed
IV(a)
dominant nature test for composite bodies
kept
Same
IV(b)
sovereign functions, strictly understood, alone qualify for exemption, not welfare activities or economic adventures
sovereign functions may qualify for exemption, not the welfare activities of economic adventures
Changed while called unchanged
V
overrules Safdarjung, Solicitors' Case, Gymkhana, Delhi University
not reproduced
Omitted
“It is pertinent to clarify at this juncture that the foregoing is only a hypothesis.” ¶275
Part five
Why it binds nobody
Two reasons, either of which is enough on its own.
The first is the count. Bagchi, J puts his position in his opening paragraph: he agrees the reference was validly made, but is “unable to agree with the reformulation of the triple test”, and concurs instead with Nagarathna and Datta, JJ that the 1978 test “correctly determines the scope and ambit of ‘industry’” Bagchi J, ¶1. Nagarathna, J holds the 1978 majority “is good law and therefore, binding” ¶21(C). Datta and Bhuyan, JJ hold that the statute and the triple test “operate in complete harmony, each reinforcing the other” ¶71. Narasimha, J refuses to reach the merits and warns that modifying the ratio “will have the effect of creating the third category that would apply exclusively to the pending cases” ¶25.
The second is the lead opinion itself. It labels the rewrite a hypothesis at ¶275, calls it “no more than our considered opinion” at ¶287(iii), and in the same breath directs that all pending matters “may be adjudicated in accordance with the Triple Test laid down in Bangalore Water Supply”. Concluded matters are left alone. The Industrial Relations Code is expressly not construed. Nothing is left for the new test to operate on.
Figure 5 · Five questions, five counts
Was the reference to nine judges validly made?
YES · 6
NO · 3
For: the Chief Justice and three colleagues, with Narasimha and Bagchi, JJ. Against: Nagarathna, Datta and Bhuyan, JJ.
Does the 1978 test govern every matter still running under the old Act?
YES · 9
The one point on which the whole bench agrees, including the four who wrote the rewrite.
Is the 1978 triple test correct law?
YES · 4
WRONG IN PART · 4
—
Nagarathna, Datta, Bhuyan and Bagchi, JJ say yes. The Chief Justice and three colleagues disagree in part. Narasimha, J does not decide.
Is the rewritten test adopted?
PROPOSED · 4
REFUSED · 4
—
Four propose it and describe it as a hypothesis. Four refuse it. One does not reach it. No majority either way.
Should section 2(p) of the new Code be read free of the 1978 decision?
YES · 6
NO · 1
OPEN · 2
Nagarathna, J alone says the old construction still counts. Datta and Bhuyan, JJ leave the point open.
Only the second question produces a settled answer of the kind a bench of nine is convened to give. On the question the reference was actually about, the count is four to four with one abstention.
Figure 6 · Which test governs what
Three categories of case, two live sources of law, and one rewritten test with nowhere to land. The lead opinion nevertheless says at ¶287(iv) that the rewrite “shall operate prospectively”.
Part six
The other four opinions
Nagarathna, J
Separate opinion · the reference should never have been made
Her opinion does the reading the 2005 bench did not. She goes through all four 1978 judgments and finds broad agreement rather than fracture: on the triple test, on overruling Safdarjung, on profit being beside the point. Where the judges differed, they differed on application, and Chandrachud, CJ's divergence on charities and sovereign functions was towards a wider reading, not a narrower one.
She then turns to the argument that the new Code should shape how the old section is read, and rejects it flatly. The bench “is unanimous in observing that we have not opined anything about Section 2(p)”, so that provision “cannot be the basis or rationale for narrowing down the definition of ‘industry’ in Section 2(j)” ¶20.2. Telling courts to give the 1978 decision a go-by when they come to construe the Code is, in her words, “not right”. Where the words of the two provisions are the same, the earlier construction “may be highly persuasive, even binding”.
The sharpest line in her opinion is aimed at the lead opinion's own logic. If the bench really has said nothing about the Code, then “the entire exercise of giving a new interpretation to Section 2(j) of the ID Act is wholly unnecessary and only academic. Then the reformulation of the Triple Test would also have no bearing” ¶20.3.
Narasimha, J
Separate opinion · the reference was proper, the answer is no longer needed
The only judge who treats the problem as a drafting failure rather than a doctrinal one. He sets out what a definition has to do to work: it must state a property held in common with nothing else, and survive three tests, no overextension, no underextension, no structural impossibility. Section 2(j) fails all of them, and no amount of interpretation repairs a clause built that way.
He borrows the Court's own reasoning in Yash Developers to argue that reviewing how a statute performs in practice is part of the rule of law, and that constitutional courts are placed to see the fault lines because they watch the statute fail case by case.
On the reference he finds no procedural defect. On the merits he stops, for a practical reason: rewriting the ratio now would produce a third body of law applying to nothing but the leftover cases, and any narrowing would push long-running disputes out of the Act altogether and into civil suits. “It would be a sad story if we were to ask them to file fresh suits before the Civil Court” ¶26.
Datta, J, for himself and Bhuyan, J
Separate opinion · the reference is rejected outright
The longest procedural argument in the judgment, and the one that will be cited in future reference disputes. It measures the 2005 order against the settled standards for reopening a precedent and finds it short on every count, then works through each of its six reasons. The non-unanimity point “missed the woods for the tree”. The reliance on Coir Board was misplaced. The reliance on Aeltemesh Rein was “a plain and simple” error.
On the merits the opinion takes each word of section 2(j) in its ordinary sense, one at a time, and concludes that the statute and the triple test “operate in complete harmony, each reinforcing the other” ¶71. It records that the Attorney General himself accepted the continued validity of the test and asked only for care in applying it.
It also names the flaw in the course the lead opinion took, before that opinion takes it: “If a settled precedent is doubted, the Court must offer a clear and workable alternative. Doubt, without direction, only compounds uncertainty” ¶15. And it is the only opinion that puts the pending appeals somewhere, directing that they be listed before an appropriate bench.
Bagchi, J
Separate opinion · the swing vote
He agrees with the Chief Justice on the reference and against him on everything that follows, which is what makes his opinion decisive. He answers the reference on merits and holds the 1978 test sound: the complaint that it turns every organised human effort into an industry “is misplaced, for the triple test, in fact, carves out just and proper exceptions” for casual work, domestic service, sole practitioners and small unorganised clubs.
His treatment of sovereign functions is the most developed on the bench. Sovereignty has contracted to a core of duties for which the State answers to the electorate rather than to a court. A duty placed on the State by statute, or even by Part IV of the Constitution, does not carry immunity from labour law with it. Articles 309 to 311 “give little solace as they cover a narrow sliver of civil posts” ¶125(III). He holds Narsinh Parmar bad law so far as it raises a presumption against government departments.
On whether to disturb a fifty-year-old precedent he applies three questions: is it plainly wrong, have conditions changed so much that it no longer works, and would reversing it do more harm than good. He answers all three against reconsideration, noting that technological and economic change since 1978 has weakened the case for narrowing the definition rather than strengthened it.
Part seven
Problems in the lead opinion
These are not stylistic quibbles. Each affects how the judgment will be read and cited.
A reference held maintainable, then not answered
The opening paragraphs are firm that the Court has no choice. Deciding “is not discretionary in character” ¶8. Refusing on procedural grounds would be hiding behind “a convenient fig leaf cover” ¶138. Conclusion (i) records that the reference “is accordingly answered on the merits”. What follows answers nothing that anyone can use. Having rejected abstention in terms, the opinion delivers it in substance.
The forest department is never dealt with
Paragraph 1 frames the case as “a singular judicial inquiry, namely, whether the Social Forestry Department of the Appellant State … falls within the ambit of the term ‘industry’”. None of the six conclusions answers it. The order at ¶¶289–290 disposes of interlocutory applications and costs and says nothing about the appeals. Datta, J at least directs that the pending matters be listed before an appropriate bench. The lead opinion leaves twenty-one years of appeals where it found them.
Disagreement with a bench of seven, without overruling it
“We are not in complete agreement with the ratio laid down in Bangalore Water Supply” ¶276. The opinion neither follows that ratio nor displaces it. It publishes a rival version and lets both stand. That is precisely the outcome the rule in Dawoodi Bohra, applied at length at ¶91, exists to prevent. A larger bench affirms or overrules; a standing note of disagreement from above is the one thing the system of precedent cannot digest.
A guideline rewritten while the opinion says it is unchanged
Paragraph 273 concludes that the sovereign functions guideline “warranted no particular change and would continue to read thus”. The text then set out reads “sovereign functions may qualify for exemption”. The 1978 wording, quoted by the same opinion at ¶191, reads “sovereign functions, strictly understood, alone qualify for exemption”. Dropping “strictly understood” and “alone” turns a closed exception into an open discretion. The altered wording appears in all three places the rewrite is set out, so it is not a stray slip. It also sits badly with ¶265, where the opinion declares itself “entirely ad idem” with 1978 on sovereign functions.
The charitable carve-outs are deleted
Paragraph 258 says it would have been enough to keep Guideline III(a) and drop III(b) and III(c). Those two are the only express exemptions the 1978 guidelines gave: for ventures with barely any employees, and for genuinely charitable missions. The same opinion summarises them approvingly at ¶¶34–37 and says at ¶203 that its purpose is to protect small enterprises, charitable institutions and community-oriented activities. On the rewritten test, a free legal aid clinic or an ashram with a cook and a driver has no textual exit left. The stated aim and the drafted result point in opposite directions.
The philanthropy clause now says the opposite of its own reasoning
The reasoning at ¶250 is that a claim of charity cannot change what is in truth a commercial body. The clause drafted to carry that reasoning protects only the case where philanthropy is claimed “in its official documentation but on enquiry it is found otherwise”. Read as written, a trade or business that is genuinely animated by philanthropy now falls outside section 2(j), which is the reverse of the 1978 clause and of the reason given for changing it. The judgment calls the change “a useful clarification”.
The trade-or-business comparison is asked for twice
The new limb I(a)(iii) requires output with “a discernible commercial character that is analogous to trade or business”. Guideline II(a) is kept, and still begins by saying that organised activity satisfying the triple test, “although not trade or business”, may still be an industry provided it resembles trade or business. If limb I already requires the resemblance, II(a)'s opening premise contradicts it and its proviso repeats it. The same question is put at two stages of one test.
Commercial character is made decisive and left undefined
It is the sieve ¶230 and an important indicator ¶243. Profit stays irrelevant. Capital investment is not mentioned. The opinion refuses “exhaustive categories, rigid exemptions, or inflexible criteria” ¶245 and any “ever-expanding catalogue of inclusions and exclusions” ¶259. A tribunal asked to apply it has a decisive element with no content beyond impression, which is the open-endedness the opinion criticises in the 1978 test at ¶220.
The reasoning of an overruled case is preferred without restoring it
Paragraph 231 sets out the passage from Safdarjung Hospital on material services and commercial analogy, and ¶232 holds that it “accords more closely with both the text and structure of the statute”. Safdarjung was expressly overruled in 1978, as the same opinion records at ¶192, and is not restored here. Paragraph 277 declines to revisit any classification decided under the 1978 test. So the better view of the statute, on the lead opinion's own account, is one that no court may apply.
The new Code is disclaimed and directed in the same breath
Paragraph 280 says nothing in the judgment should be read as an opinion on the interpretation of any provision of the Industrial Relations Code. Conclusion (vi) then directs that the 1978 decision “shall not be regarded as the sheet anchor for the interpretation of the IR Code”. Telling courts what may not inform their reading of section 2(p) is an opinion on how section 2(p) is to be read. Nagarathna, J objects to this directly, and her objection carries force because she is describing a direction given by four judges.
A dissent used as authority
At ¶94 the opinion relies on the dissenting opinion of “one of us (Surya Kant, J. (as he then was))” in Aligarh Muslim University v. Naresh Agarwal to support the framework for references. A dissent decides nothing. The same point was available from the majority in that case, which is where Bagchi, J takes it from, noting that the minority there had not doubted Dawoodi Bohra in any event.
“Includes” said to make the definition exhaustive
Paragraph 156 reads: “The inclusion of the term ‘includes’ further characterises the definition as exhaustive; therefore, an activity, regardless of its not being explicitly listed within the definition, but it otherwise satisfies the criteria of an industrial activity, shall also be considered as industry.” The consequence stated is the consequence of an inclusive definition. The sentence contradicts itself in its second clause. Datta, J handles the two limbs correctly at ¶¶63 and 70.
Dates that do not line up
Paragraph 77 dates the three-judge reference 31 January 2001; paragraphs 99 and 105 and footnote 26 date it 31 January 2002. Paragraph 61 calls the amending statute “Act 46 of 1982” and paragraph 62 says it “was passed on 21.08.1984”. On the date the old Act was repealed, paragraph 81 says 2 February 2026, while Nagarathna, J and Narasimha, J both say 21 November 2025 and Datta, J refers to the Code becoming operational in February 2026. Paragraph 142 lists Mr R. Venkataramani as “learned Advocate General for India”; he appears as Attorney General everywhere else. The rewritten guidelines carry typographical corruption into all three settings, including “some of whom Are not ‘workmen’” and “welfare activities of economic adventures”.
The opinion's own history answers the reference's premise
Paragraph 27 records that the triple test “secured broad acceptance among all seven learned Judges, with the divergence being confined not to its formulation, but to the manner and extent of its application”, and ¶28 that even the dissent “substantially accepted the three constituent elements”. That disposes of the non-unanimity ground on which the 2005 bench referred the case. The opinion then relays that ground without comment at ¶¶106–109 and treats it as good enough to support the reference. Nagarathna and Datta, JJ both say plainly that the premise was false.
Part eight
What to do with this on Monday morning
If you have a matter pending under the old Act
Plead the 1978 triple test as it stands. Do not qualify it. If the other side raises the rewritten version, three answers are available and any one of them ends the point: the count of four out of nine; the words “only a hypothesis” at ¶275; and the direction at ¶287(iii) sending every pending matter back to the 1978 test.
If you act for a government department or a welfare body
The sovereign functions exception has not been widened by anything that binds. Bagchi, J holds that a duty imposed by statute or by Part IV brings no immunity with it, and that Narsinh Parmar is bad law so far as it presumes against departments. Nagarathna, J is to the same effect: the State does many things that are not sovereign in the strict sense, and doing them through a department does not put them outside the Act. The softened wording of Guideline IV(b) carries four votes and no more.
If the dispute arises under the new Code
This is where the next fight will be, and the bench does not speak with one voice. Six judges say section 2(p) should be read on its own terms, but they mean different things by it. The lead opinion goes furthest, directing that the 1978 decision not be the “sheet anchor” ¶287(vi). Narasimha, J frees the Code from “the bondage” of that ratio ¶23. Bagchi, J simply confines the old ratio to legacy proceedings. Nagarathna, J takes the opposite view, that where the statutory words are the same the earlier construction may be persuasive and even binding, and that section 2(j) may bear on section 2(p) and not the other way round. Datta and Bhuyan, JJ reserve the point, saying only that nothing in their opinion affects any future challenge to the Code.
Note also that the rewritten test cannot be carried across. It is not law under the repealed Act, and the lead opinion disclaims any view on the Code at ¶280.
If you are arguing about a reference
The most usable material is in the opinions of the three judges who lost on that question. Datta, J's treatment of when a settled precedent may be reopened, at ¶¶32 to 38, is the fullest statement in the judgment. And the finding by both Nagarathna and Datta, JJ that no conflict existed between Kondhare and Pratamsingh Parmar is not contradicted by the lead opinion, which addresses whether the procedure was proper rather than whether the conflict was real.
The short version
The triple test of 1978 survives untouched and governs every case still running under the Industrial Disputes Act. The calibrated test is a four-judge proposal, described by its own authors as a hypothesis, applicable to no proceeding anywhere. Section 2(p) of the Industrial Relations Code starts life with six judges saying it should be read fresh and one saying the old learning still counts.
A closing thought
Datta, J wrote that institutional credibility “lies in respecting finality, not in perpetuating doubt”. The judgment ends by hoping the albatross has finally been lifted from the Court's neck. Whether it has is doubtful. Four judges of the Supreme Court have now said in print that a seven-judge decision is wrong in part, and have declined to do anything about it. The next bench that has to construe section 2(p) will find, sitting in the reports, a rival test with four signatures on it and no legal force, together with a direction not to read the new Code in the light of the old one, given by the same four judges who said they were expressing no view on the new Code at all.
Krishna Iyer, J closed the 1978 judgment by saying he concluded “with diffidence”, because Parliament had not stepped in. Parliament has now stepped in. Whether the Court has stepped back is a harder question.
Frequently asked
Questions about the Jai Bir Singh judgment
What did the Supreme Court decide in State of U.P. v. Jai Bir Singh?
A bench of nine judges delivered five opinions running to 489 pages, reported as 2026 INSC 897 and dated 20 August 2026. All nine agreed on one point: the triple test laid down in Bangalore Water Supply v. A. Rajappa (1978) governs every matter still running under the Industrial Disputes Act, 1947. A rewritten test proposed by four judges was described in the judgment itself as “only a hypothesis” and was not adopted.
Is the Bangalore Water Supply triple test still good law?
Yes. Nagarathna, Datta, Bhuyan and Bagchi, JJ held the 1978 test correct and binding, and the whole bench, including the four who wrote the rewrite, directed that pending matters be decided under it. The seven-judge decision of 1978 was not overruled.
Which judges were on the nine-judge bench in Jai Bir Singh?
Surya Kant, CJI wrote the lead opinion for himself and Satish Chandra Sharma, Alok Aradhe and Vipul M. Pancholi, JJ. B. V. Nagarathna, J, P. S. Narasimha, J, Dipankar Datta, J (for himself and Ujjal Bhuyan, J) and Joymalya Bagchi, J wrote separately — five opinions in all.
Is the new calibrated triple test binding?
No. It has the support of four judges out of nine, the lead opinion calls it “only a hypothesis” at ¶275 and “no more than our considered opinion” at ¶287(iii), and the same paragraph directs that all pending matters be adjudicated under the 1978 triple test. It is not the ratio of the case.
Does the judgment decide the meaning of industry under section 2(p) of the Industrial Relations Code, 2020?
No. Paragraph 280 says nothing in the judgment should be read as an opinion on the interpretation of any provision of the Industrial Relations Code. Conclusion (vi) nevertheless directs that the 1978 decision not be regarded as the sheet anchor for interpreting the Code, and Nagarathna, J objects to that direction.
What changed in the rewritten test proposed by the Chief Justice?
Limb I(a)(iii) would require the output to carry a discernible commercial character analogous to trade or business; guidelines III(b) and III(c), the exemptions for ventures hiring employees only in minimal matters and for eleemosynary missions, are deleted; and the sovereign functions guideline IV(b) is softened from “strictly understood, alone qualify” to “may qualify”. Those changes carry four votes and no more.
What is the citation of the Jai Bir Singh judgment?
State of U.P. v. Jai Bir Singh & connected matters, 2026 INSC 897, delivered on 20 August 2026. The reference was heard by nine judges over three days in March 2026 and the judgment runs to 489 pages across five opinions.