Part III research fails in a particular way, and it is not the way most research fails. The problem is rarely that you cannot find authority. It is that you find too much, from across seventy-five years, expressing the same test in incompatible terms, all of it quotable and none of it labelled with a date of expiry. A formulation from 1950 reads exactly as authoritative on the page as one from 2017. The document does not tell you that the earlier one was displaced, and if you cite it, the bench will.
Constitutional doctrine layers. A test is laid down, applied for a decade, found too narrow, widened by a larger bench, refined in a later case, and restated in a different vocabulary by a Constitution Bench addressing something else entirely. Each of those judgments remains in the corpus. Several remain good law in part. Some were never overruled, only left behind. The result is a body of authority in which the most famous statement of a test is very often not the currently governing one.
The useful way to think about this is version control. A constitutional test is an artefact with a version history: an original formulation, a series of revisions, a bench strength attached to each revision, and a current head. Researching Part III means reconstructing that history and identifying the head, and the discipline that prevents the commonest error is simply refusing to cite any formulation until you know where it sits in the sequence.
The habit this article is really about
Before citing any constitutional proposition, answer two questions: which bench strength laid it down, and is this the current formulation or a superseded one. Neither answer is visible on the face of the judgment. Both decide whether the citation helps you or ends the argument. Almost every avoidable failure in Part III research is a failure to ask one of them.
Why Part III Research Is Version Control
The clearest illustration is Article 21. In A.K. Gopalan (1950) the Court read procedure established by law substantially at face value, holding that a law depriving a person of life or personal liberty satisfied the Article if the prescribed procedure was followed. Justice Fazl Ali dissented. That reading governed for decades. In Maneka Gandhi (1978) the Court held that the procedure contemplated must be fair, just and reasonable, and cannot be arbitrary or oppressive. The Article's text did not change by a single word. What it requires changed entirely.
Both judgments exist in the corpus. Both contain clean, quotable statements of what Article 21 requires. Only one of them describes the law as it now stands, and nothing on the face of the earlier judgment says so. This pattern recurs across Part III: Article 14, the definition of the State, the standard applied to restrictions on the freedoms in Article 19. In each case the current position is the product of a sequence, and the sequence has to be reconstructed rather than assumed.
A constitutional test is not a sentence in a judgment. It is the current head of a chain of judgments, and the sentence you found may be three versions behind.
Dissents complicate this usefully. Justice Fazl Ali's dissent in A.K. Gopalan and Justice Subba Rao's dissent in Kharak Singh (1963) were both, in substance, vindicated by later benches. A dissent is never authority, and citing one as though it were is a serious error. But a dissent that later became the majority position tells you where the doctrine was heading, and knowing that history is often how you recognise that the majority judgment you are reading has been overtaken.
Articles 12 and 13: The Questions Before the Right
Two threshold questions are decided before any right is reached, and both are commonly skipped by researchers who go straight to the substantive Article. Getting either wrong means the substantive research was wasted.
Article 12 defines the State for the purposes of Part III, including the Government and Parliament of India, the Governments and Legislatures of the States, and local or other authorities within the territory of India or under the control of the Government of India. The litigation is almost entirely about the last limb. A substantial line of authority has developed on when a body that is not obviously governmental is nonetheless an instrumentality or agency of the State. The approach is multi-factor rather than mechanical, looking at matters such as the extent of state financial support, the degree of state control over the body's affairs, whether its functions are of public importance and closely related to governmental functions, and whether it was created by transferring an existing government department. No single factor decides it, and the enquiry is cumulative, which is why the case law is fact-heavy and why a decision about one body is only weakly transferable to another.
Article 13 supplies the consequence. Laws in force before the Constitution, to the extent they are inconsistent with Part III, and laws made after it that take away or abridge the rights conferred by Part III, are void to the extent of the inconsistency or contravention. Two points matter for research. The first is what counts as law for this purpose, which is defined broadly enough to include subordinate instruments and, in terms, custom or usage having the force of law. The second is that voidness operates to the extent of the inconsistency, which is why severability arguments arise and why a challenge that succeeds may leave most of a statute standing. There is also a body of authority on the position of pre-constitutional laws that are inconsistent with Part III, which are not obliterated but rendered ineffective against the rights, a doctrine worth understanding before framing a challenge to an old enactment.
The Tests, and Which Version Governs
Article 14: from classification to arbitrariness
The older approach to Article 14 was the classification test: a law that treats a class differently survives if the classification rests on an intelligible differentia and that differentia bears a rational nexus to the object the law seeks to achieve. That test is not dead. It is still applied, and for many challenges it remains the natural framework.
What changed is that it stopped being the whole of Article 14. A later line of authority, given its most influential expression around the time of Maneka Gandhi (1978), holds that equality is antithetical to arbitrariness, so that state action which is arbitrary is for that reason a violation of Article 14 without needing to be analysed as a classification at all. The practical consequence for research is that there are two available routes and they are not interchangeable. A challenge to a statutory classification and a challenge to an arbitrary executive decision are argued differently, and citing classification authority for an arbitrariness argument produces a submission the bench cannot use.
Article 19: freedoms and their own limiting clauses
Article 19 is structurally different from Article 21 and this shapes the research. Each freedom in the first clause has its own limiting provision, and those provisions are not identical. The grounds on which speech may be restricted are not the grounds on which trade or profession may be restricted. Every restriction must both fall within one of the enumerated grounds for that particular freedom and be reasonable.
So the research has to be pinned to the right pair. Identify the specific freedom said to be infringed and the specific limiting clause the State relies on, and search on that pair. Authority on the reasonableness of restrictions in the context of one freedom is at best analogous when the argument concerns another, because the enumerated grounds differ and the standard of scrutiny applied has varied with them.
Article 21, and the reading together
After Maneka Gandhi (1978), procedure established by law under Article 21 is read as requiring procedure that is fair, just and reasonable. The judgment did something further that is central to how Part III is argued now: it rejected the earlier view that the Articles operate in separate compartments, and held that Articles 14, 19 and 21 are to be read together. A law depriving a person of personal liberty must therefore answer not only Article 21 but also the tests of Article 14 and, where a freedom under Article 19 is engaged, the requirements of the relevant limiting clause.
For the researcher this converts a single-Article question into a multi-Article one. A challenge framed only under Article 21 is usually incomplete, and the research plan should test the impugned measure against each of the three, separately, before deciding which combination to lead with.
Proportionality as the Current Framework
In K.S. Puttaswamy (2017), decided by nine judges, the Supreme Court recognised privacy as a constitutionally protected right and, in doing so, engaged extensively with the framework by which limitations on rights are assessed. Proportionality analysis is now the framework commonly applied when a rights limitation is in issue, and structured limitation reasoning of this kind has been applied across a range of Part III questions since.
Described carefully, and the caution matters because the formulation is not stated identically in every judgment, the analysis proceeds through stages of this general shape:
- A legitimate aim. The measure must pursue a goal of sufficient importance to justify limiting the right.
- A rational connection. The measure adopted must be suitable to achieve that aim, and not merely asserted to be.
- Necessity. There must not be a less restrictive alternative that would achieve the aim comparably well.
- Balancing. The extent of the interference with the right must be proportionate to the importance of the aim served.
- Legality, which several formulations state at the outset: the limitation must have a basis in law, and in the privacy context this requirement did real work.
The research consequence is that a proportionality argument is a four or five part argument, and each part needs its own material. Necessity in particular is an evidentiary stage as much as a doctrinal one, because showing a less restrictive alternative usually means showing that something workable exists. An argument that asserts disproportion as a conclusion, without walking through the stages, is not making the argument the framework asks for.
Article 32 and Article 226: Choosing the Route
The forum question is part of the research, not a formality after it, and the two jurisdictions differ in ways that affect what you must establish.
What each is for
Article 32 is a remedy for the enforcement of the rights conferred by Part III. Article 226 is wider in subject matter: a High Court may issue writs for the enforcement of fundamental rights and for any other purpose, which brings ordinary legal rights within its reach.
The status of the remedy
The right to move the Supreme Court under Article 32 is itself a fundamental right. Article 226 is a constitutional power of the High Courts but is not itself a fundamental right, and both jurisdictions are in practice exercised with discretion.
Territorial reach
Article 226(2) allows a High Court to exercise the power where the cause of action arises wholly or in part within its territory, notwithstanding that the seat of the authority is elsewhere. Where a petition can be filed is frequently a contested threshold issue of its own.
Alternative remedy
The existence of an alternative statutory remedy is a self-imposed restraint rather than a bar. Whirlpool (1998) remains the standard authority for the recognised exceptions, and a petition that does not address the point invites dismissal without reaching the merits.
Tribunals sit underneath
L. Chandra Kumar (1997) held that decisions of tribunals remain subject to scrutiny by the High Court under Articles 226 and 227. So a Part III challenge arising out of a tribunal's decision has a route, and identifying it correctly is part of the research.
Where the doctrine actually is
A great deal of Part III doctrine is worked out in High Court judgments under Article 226, not only in the Supreme Court. Research confined to Supreme Court authority misses the application of the tests to facts, which is usually the part you need.
Right, Limiting Provision, and What Must Be Established
The table below is a planning device rather than a statement of doctrine. Its purpose is to force the research into elements before any searching begins, so that you end up with an authority for each element rather than one case for the whole submission.
| Provision in issue | What limits or qualifies it | What the researcher must establish |
|---|---|---|
| Article 12 (threshold) | Whether the respondent falls within the State | Where the respondent is not plainly governmental, establish the instrumentality factors cumulatively: state financing, control over affairs, public and governmental functions, origin as a transferred department. Find authority on a comparable body, not a general statement of the test. |
| Article 13 (threshold) | Whether the impugned instrument is law, and the extent of voidness | Show the instrument falls within the broad definition, including subordinate instruments. Address severability, because voidness operates only to the extent of the inconsistency. |
| Article 14, classification route | The classification itself must be justified | Identify the class and the differentia, and the object of the statute. Establish that the differentia is intelligible and that its nexus to the object is rational. The object has to be pinned down first, because the nexus is measured against it. |
| Article 14, arbitrariness route | Arbitrariness in state action | Establish that the action is arbitrary rather than merely wrong or unwise. This route suits executive action and is argued differently from a classification challenge. Do not mix the two frameworks in one submission. |
| Article 19(1) freedoms | The limiting clause specific to that freedom | Establish that the freedom is engaged, then that the restriction either falls outside the enumerated grounds for that freedom or is unreasonable. Search on the pair, because the grounds differ from freedom to freedom. |
| Article 21 | Procedure established by law, read as fair, just and reasonable | Establish the deprivation of life or personal liberty, the absence of a valid law or the unfairness of the procedure, and then test the measure against Articles 14 and 19 as well, since the three are read together. |
| The remedy | Article 32 or Article 226 | Establish jurisdiction, including territorial reach under Article 226(2), and deal with the alternative remedy objection in the petition rather than in reply. |
The Method
The order below is designed so that no search is run before the thing being searched for has been defined. Most wasted Part III research is a broad search run too early, which returns a great many famous judgments and no answer.
State the right and the limiting provision precisely
Not Article 19 but the specific freedom, and not restrictions generally but the specific clause and ground relied on. Not Article 14 but which of the two routes you are running. This single act of precision determines every search that follows, and it is the step that separates a targeted research plan from a reading list of constitutional landmarks.
Find the currently governing formulation, not the most famous one
Identify the most recent authoritative statement of the test by a bench competent to state it. Fame and currency are different properties and they diverge often, because the judgment that first laid a test down is usually more celebrated than the one that reformulated it. Start from the most recent Constitution Bench treatment and work backwards, rather than starting from the landmark and working forwards.
Build the version history
For the test you are relying on, write out the sequence: original formulation, each reformulation, the bench strength behind each, and whether the earlier version was overruled, refined, or simply left behind. Distinguish carefully between a formulation that was held wrong and one that remains correct but partial. Both look identical in a search result and they have opposite consequences for citation.
Check bench strength on every proposition
Constitutional propositions are stratified by bench strength. Article 145(3) requires a minimum of five judges for a substantial question of law as to the interpretation of the Constitution, and decisions of five, seven and nine judges sit above smaller benches. A two judge decision that appears to state your proposition perfectly may be inconsistent with a larger bench, and if it is, it is the larger bench that governs. This is the first filter on any Part III authority, not the last.
Check for pending references
Constitutional questions are referred to larger benches and remain pending for long periods. A reference on your exact question does not usually displace the existing position, but it changes how you present the argument and what you disclose. It is also information a Constitution Bench is likely to have, so discovering it in court is worse than the reference itself.
Search High Court applications, not only Supreme Court statements
The Supreme Court states tests. High Courts apply them to facts, and application is usually what your matter needs. Search all 25 High Courts for the test applied to fact patterns like yours. This is also where you find the arguments that have failed, which is information no statement of the test will give you.
Apply the test element by element to your own facts
Take the current formulation apart into its elements and write what your facts establish under each. Where an element is weak, that is the part of the case that needs the work, and knowing it early is the entire point of the exercise. This is also what converts research into a structured submission rather than a collection of citations.
Cite an authority for each element separately
Resist the single-case submission. One judgment cited for the whole argument invites the bench to distinguish it on its facts and take the argument with it. An authority for each element is much harder to dismantle, because each citation has to be answered on its own terms and the argument survives the loss of any one of them.
The superseded formulation problem
A judgment stating a test that has since been reformulated is not marked in any way. It reads as good law, it may still be good law for other purposes, and the sentence you want may be the most elegant statement of the test ever written. Citing it is still a mistake, and it is a visible one, because the bench knows the sequence. This is the most common avoidable failure in Part III research, and the only reliable protection is refusing to cite a constitutional proposition until you have located its current version and the bench strength behind it.
Where a Research Platform Helps, and Where It Does Not
The version control problem has a retrieval half and a judgment half, and it is worth being clear about which is which. The retrieval half is assembling every judgment that has stated or applied the test, across the Supreme Court and all 25 High Courts, in date order. That is exactly what a unified search does well. CourtMesh covers the Supreme Court, all 25 High Courts, District Courts and Tribunals, sourced from official government portals, with roughly 310 million records keyword-searchable, and structured filters for court, year, judge, case type and date range. Filtering by year is more useful in constitutional research than almost anywhere else, because the sequence is the substance.
Citation relationships help with the direction of travel. Across the analysed subset, relationships are typed as followed, distinguished, overruled and referred, so you can see which later benches engaged with a judgment and how. On a doctrine that has moved three times, that traversal is a much faster route to the current head than reading forward through the years. Semantic search matters too, because a judgment reformulating a test frequently does so in vocabulary that does not match the older formulation you searched, and that subset covers roughly 2 million records rather than the whole corpus.
The judgment half does not transfer. Whether a later formulation refined an earlier one or displaced it is an interpretive question argued in open court, and a typed relationship is a pointer to something you should read rather than a conclusion. Bench strength has to be read off the judgment. Whether a passage is ratio or obiter has to be decided by reading it. Deep analysis covers a subset of the corpus, so the absence of a flagged relationship is never proof that none exists. And no tool will tell you which of two available routes under Article 14 fits your facts, which is the kind of question the whole method exists to make answerable by a person.
The official record prevails
Every result links to the judgment text, and the record of the issuing court is the authoritative one. On constitutional propositions this deserves particular care: confirm the bench strength, the date, and the passage against the court's own record before any of it goes into a petition, because a proposition attributed to the wrong bench strength is worse than no citation at all.
Find the current version, not the famous one
Part III doctrine is a sequence, and the sentence you found may be several versions behind. Reconstruct the version history, check the bench strength behind each formulation, and cite an authority for each element of the test rather than one case for the whole argument. CourtMesh gives you one search across the Supreme Court, all 25 High Courts, District Courts and Tribunals from official government portals, year and court filters that make a sequence visible, and citation relationships across the analysed subset. The reading and the judgment stay yours.
Explore CourtMesh
