Every advocate knows the distinction. The ratio decidendi binds. Obiter dicta does not. The difficulty starts the moment you open a hundred and forty page judgment of the Supreme Court, with three separate opinions, forty pages of extracted passages from earlier authority, and one paragraph that says exactly what your client needs it to say. Is that paragraph the holding, or a remark made on the way to the holding? The doctrine gives you two categories. It does not hand you a procedure for sorting a document into them.
This article is about the procedure. The claim it makes is that ratio identification is a falsifiable skill rather than an act of intuition. There are steps. Each step produces something you can write down. Each step has a characteristic failure. Two careful lawyers running the method on the same judgment should mostly converge, and where they do not, they should be able to say exactly which step they parted company at and why. That is what falsifiable means here: you can be shown to be wrong, on the record, by reference to the text.
The consequence of getting it wrong is not academic. Present obiter as the holding and you are inviting the bench to say so, often before you have finished the sentence. Once that happens the rest of your argument is read with suspicion, because the court now knows you either did not read the judgment carefully or read it carefully and hoped nobody else would. The opposite error is quieter and just as expensive: treating a genuine holding as a passing observation, conceding a point you were entitled to win, and never discovering it.
What the Doctrine Actually Asks You to Find
A judgment does not contain a section marked ratio. What it contains is a dispute, a set of facts, one or more questions, reasoning, and an order. The ratio is not a sentence the court wrote. It is a proposition you construct by asking what rule the court must have applied to these facts to reach this order.
That framing does most of the work. It tells you the ratio is tied to two things and only two: the facts the court treated as material, and the order it passed. Everything in between is reasoning, and reasoning divides into the part the order depends on and the part it does not. The classical tests come at this from two directions. One asks which facts the court treated as material, on the view that a rule is only ever a rule about some facts. The other inverts the proposition and asks whether the same order still stands once you remove it. Neither test is mechanical. Together they are close to a procedure.
The ratio is a proposition, not a passage
This is the single most useful correction to make early. You are not hunting for the paragraph that contains the ratio. You are drafting a sentence that states the rule, in your own words, and then testing that sentence against the facts and the order. A judgment can state its own ratio badly, or in three different widths in three different paragraphs. Your job is to work out which width the decision actually supports.
The Method, Step by Step
Run these in order. Skipping ahead is the commonest cause of a confident wrong answer, because the later steps only work on the outputs of the earlier ones. On a short judgment the whole sequence takes ten minutes. On a Constitution Bench decision it takes an afternoon, and the afternoon is the point.
Read the order first, then the facts, then the reasoning
Start at the end. The operative order tells you what the court actually did: allowed, dismissed, remanded, read down, quashed, or directed. Everything you conclude about the ratio has to be consistent with that order. Reading the reasoning first invites you to find the rule you were hoping for and then fit the order around it.
List the material facts as the court treated them
Not the facts as the parties pleaded them and not the facts you find interesting. Write down the facts the judgment keeps returning to, the ones it uses when it explains why the result follows. A fact the court recites once in the narrative and never uses again is background. A fact it repeats when it applies the law is material, and the ratio is a rule about material facts.
State the question the court had to answer to dispose of the matter
Many judgments frame several questions, and the frame is often generous. Ask which of them the order genuinely depended on. If the court answered four questions and the appeal would have been allowed on the first alone, the other three are candidates for obiter. Write the operative question as one sentence, in the narrowest form that still explains the order.
Draft the proposition in your own words, at three widths
Write the rule narrowly, tied closely to the material facts. Then write it broadly, as a general principle. Then write a middle version. You now have three candidate ratios rather than one impression. The rest of the method is a test that eliminates the ones the decision cannot carry.
Apply the reversal test to each candidate
Take the candidate proposition, assume its opposite, and reread the reasoning. If the order could still stand on the remaining reasoning, that proposition was not necessary and is very likely obiter. If the order collapses without it, you are holding ratio. Do this honestly. The temptation is to declare the proposition necessary because you need it to be.
Separate necessary reasoning from illustrative reasoning
Indian judgments are generous with illustration. Comparative material, hypotheticals, historical excursus, observations about what the position would be on facts not before the court: all of it can be excellent and none of it is load bearing. The test is the same. Delete the passage and see whether the order still follows. Illustration explains the rule. It does not constitute it.
Resolve alternative and cumulative grounds explicitly
Where the court gives two independent reasons for the same order, decide, and record, which way you are reading it. This is the step most often skipped, and the one your opponent will exploit. It has its own section below because Indian practice is not uniform here and pretending otherwise will get you into trouble.
Reconcile the opinions on a multi-judge bench
If there are concurrences or dissents, the ratio is what commanded a majority, which may be narrower than any single opinion. Work out what a majority of the judges actually agreed on, as opposed to what the leading opinion says at its widest. A proposition endorsed by two judges on a bench of five is not the holding of the Court.
Fix the paragraph, with its surrounding reasoning intact
Only now go back and identify the paragraph you will cite. Read the two paragraphs before and after it. If the passage reads differently in context than it did in isolation, context wins, and you revise the proposition. Quote enough that the bench can see the reasoning around the sentence rather than the sentence alone.
The Reversal Test, and How It Fails
The reversal test is the workhorse of the method: would the outcome change if this proposition were removed? It is powerful because it converts a question about emphasis, which is unarguable, into a question about logical dependence, which is arguable. Two lawyers disagreeing about whether a passage is emphatic will get nowhere. Two lawyers disagreeing about whether the order survives without it are having a real argument with a real answer.
The question is never how important the sentence sounds. It is whether the order could have been passed without it. Importance is a matter of tone. Necessity is a matter of structure.
It fails in two places, and both are worth knowing before you rely on it. It fails where the court gives several independent reasons, because removing any one of them leaves the order standing, and a mechanical application of the test would make the entire judgment obiter. That result is obviously wrong, which is why the test has to be applied to the set of grounds rather than to each in isolation. It also fails where the reasoning is so entangled that no proposition can be cleanly excised, which is common in constitutional judgments that construct a framework rather than apply a rule.
Broad principle, narrow facts
The most frequent shape of the problem in Indian practice is a judgment that announces a wide principle and then decides on a very narrow set of facts. The principle is stated in language that is quotable, memorable, and considerably wider than anything the case required. Later benches then have to decide how much of it was holding.
The working position is that a decision is authority for the proposition its facts can carry, and that generality in the language does not by itself widen the ratio. That is the safer reading and the one a bench is more likely to accept when the other side is pressing the wide version. But it is not a rule you can apply without judgment, because sometimes a court is quite deliberately laying down a general rule and says so. The signal to look for is whether the court applied the wide proposition to reach its result, or announced it and then decided on something narrower. If it announced and then narrowed, you are looking at a principle with an obiter penumbra around a narrow core.
Alternative Grounds: Where Indian Practice Is Not Uniform
A court dismisses an appeal because the petition was barred by limitation and, in any event, because the claim fails on merits. Which of those is the ratio? There are three defensible answers and Indian practice has used all of them.
- Both are ratio. The court considered both grounds, decided both, and each independently supports the order. On this view a later court is bound by both propositions.
- The first is ratio, the second obiter. Once the matter was disposed of on limitation, everything after that was unnecessary. On this view the merits discussion is a considered observation and no more.
- Neither is safely ratio on its own, and the decision is authority only for the composite proposition that a claim in this posture fails. This is the narrowest reading and the one an opponent will press when the second ground is the one hurting them.
You will find Indian authority pointing in more than one of these directions, and the language courts use, in any event, without prejudice, assuming we are wrong on the first point, matters a great deal. A ground introduced with even if we are wrong on the above is reasoning offered in the alternative, and reads more like a considered dictum than a holding. A ground the court reaches independently, with its own analysis and its own conclusion, reads like a second ratio.
What to do with an unsettled question
You do not resolve this in your own notes and hope. You take a position, you state it openly, and you have the fallback ready. If your authority rests on the second of two grounds, say so: the court decided the matter on limitation and also considered the merits, and the observations on merits are, on the respectful submission, part of the reasoning the decision rests on. A bench respects that framing. It does not respect discovering the point for itself.
Multi-Opinion Benches and the Narrowest Common Ground
On a bench of five, seven, or nine, the ratio of the Court is not the ratio of the leading opinion. It is what a majority of the judges agreed on. That sounds obvious and is repeatedly got wrong, because the leading opinion is usually the longest, the most quoted, and the first thing a search result surfaces.
Count the judges, not the pages
Before reading a word of analysis, work out how many judges signed each opinion. A forty page opinion carrying two votes is a minority view however comprehensive it is, and a two page concurrence carrying three is the Court.
Find the narrowest common ground
Where the majority agrees on the result but reasons differently, the safest statement of the ratio is the narrowest proposition every judge in the majority accepted. Anything wider is a proposition some of the majority did not endorse.
A concurrence is not a dissent
A judge who agrees with the order but reasons differently has voted with the majority. Their reasoning is not the holding unless it is shared, but their vote counts, and their opinion often states the doctrinal issue more cleanly than the leading one.
Partial dissents split the count per issue
On a judgment answering four questions, the majority on question one need not be the majority on question three. Work out the alignment issue by issue. Treating a bench as having one majority across the whole judgment is a real and common error.
Read the operative paragraph at the end
Many multi-opinion judgments close with a consolidated statement of what the Court has held and who agrees. It is not conclusive, and it is the fastest available check on any reading you have built from the body.
Beware the quotable minority
Minority opinions are often better written, because a judge who does not have to carry colleagues can write freely. The eloquence of a passage is uncorrelated with whether it commanded a majority.
The Quotation Trap in Long Indian Judgments
Indian judgments quote earlier authority at length. It is a genuine strength of the tradition and it creates a specific and very common research failure: a passage quoted from an earlier case is not the present court's holding. It is the present court's account of what an earlier court said. Sometimes the present court is adopting it. Sometimes it is setting it out in order to distinguish it, confine it, or explain why it does not apply. Sometimes it is simply recording the submission of counsel.
A keyword search does not know the difference. The words are in the document, so the document comes back. If you take the paragraph number from the search result and cite it as the holding of the later case, you have cited a judgment for a proposition it may have been rejecting. The bench, which knows the case, will notice.
Three sentences of context, every time
Before you cite any passage, read the sentence that introduces it and the sentence that follows it. You are looking for the verb: the court held, we respectfully agree, it was submitted, the earlier bench observed, we are unable to accept. That verb tells you whether you are reading the court's holding, the court's summary of someone else's holding, or an argument the court went on to reject. It takes fifteen seconds and it is the highest yield fifteen seconds in legal research.
Obiter and Its Neighbours: Per Incuriam and Sub Silentio
Three doctrines sit close together and are regularly confused, including in argument. They fail differently and it is worth keeping them apart.
| Doctrine | What is actually wrong with the passage | How you use it |
|---|---|---|
| Obiter dicta | The proposition was not necessary to the decision. The court said it, and meant it, but the order did not depend on it. | Not binding. Cite it for what it is: a considered observation, whose weight depends on the court and the care behind it. Never present it as the holding. |
| Sub silentio | The point was never argued or consciously decided. It was assumed on the way to something else, and the judgment reads as though it were settled. | Not an authority on that point at all. This is your answer when the other side quotes a clean sentence on a question nobody in that case contested. |
| Per incuriam | The decision was reached in ignorance of a binding statutory provision or a binding precedent that would have governed it. | A narrow escape hatch, and it is a later court that declares a decision per incuriam, not a party that labels it. Argue it sparingly and only where the omission is demonstrable. |
| Considered dictum | Necessary to nothing, but reached after full argument and deliberate analysis, often on a point the court chose to settle for the guidance of the field. | Formally obiter, practically weighty, particularly from the Supreme Court. Courts follow it routinely. Present it honestly and its weight will usually carry. |
| Assumption of a party | A proposition both sides accepted, so the court proceeded on it without deciding it. The text may state it flatly. | Not a holding. Read the judgment for the phrase it is common ground or a similar formulation, which is where these hide. |
The practical point about per incuriam is the one most often missed in argument. It is not a property a judgment carries that you can point at. It is a conclusion a later court reaches. Telling a bench that a Supreme Court decision is per incuriam, when you are appearing before a High Court, is a submission with a very high failure rate, and it tends to consume the goodwill you needed for the rest of the case.
What This Looks Like When It Goes Wrong
The exchange is short and it is memorable. Counsel reads a paragraph. The bench asks what the facts of that case were. Counsel does not have them readily, because the paragraph arrived through a search result rather than through a reading of the judgment. The bench observes that the passage was in the context of a different question, or that the court there was quoting an earlier decision, or that the appeal was allowed on an entirely different ground.
Nothing dramatic follows. The authority is simply gone, and so is a measurable amount of credibility for everything else in the brief. The defence against it is not eloquence. It is having run the method, so that when you are asked what the facts of that case were, the answer is already written on the sheet in front of you, because listing the material facts was step two.
Where a Research Platform Helps, and Where It Does Not
Ratio identification is the part of legal research that software is worst at, and it is worth being plain about that rather than gesturing at it. Everything in the method above is an act of judgment about what a court needed to decide. No system performs it for you, and any product suggesting otherwise is describing something it does not have.
What a platform does is get you to the reading faster and stop you reading the wrong thing. CourtMesh runs one search across the Supreme Court, all 25 High Courts, District Courts and Tribunals, sourced from official government portals, with roughly 310M records keyword searchable and roughly 2M semantically searchable. Structured filters for court, year, judge, case type and date range narrow a field before you commit an afternoon to it. Every result links to the judgment text, which matters more here than anywhere else, because the method requires the whole document and not a snippet.
Two capabilities bear directly on this article. Citation relationships, typed as followed, distinguished, overruled and referred across the analysed subset, tell you how later benches have read the judgment, which is often the best available evidence of what the profession treats as its ratio. A decision repeatedly distinguished on the ground that the wide passage was obiter is telling you something you would otherwise take weeks to learn. AI case analysis can summarise issues and outcomes and point you to the paragraphs worth reading. Neither is a holding. Both are leads.
The honest limit
Semantic retrieval and deep analysis cover a subset of the corpus, not the whole of it, and the absence of a signal is not evidence of absence. More importantly, no summary can tell you whether a proposition was necessary to the order, because that requires reading the order, the facts and the reasoning together and forming a view. The issuing court's record is the authoritative one, and the reading is yours. A tool can put the right nine judgments in front of you by lunchtime. It cannot tell you what any of them held.
The Reading Is the Work
Finding the paragraph is the fast half and it has been getting faster every year. Deciding whether that paragraph is the holding is the half that decides cases, and it is done by listing the material facts, stating the operative question, drafting the proposition, and testing whether the order survives without it. CourtMesh exists to shorten the search so the reading gets the time it deserves: one search across the Supreme Court, all 25 High Courts, District Courts and Tribunals from official government portals, structured filters to narrow the field, citation relationships across the analysed subset to show how later benches have treated your authority, and every result linked to the judgment text. Then close the tab and read the judgment.
Explore CourtMesh
