How to Read a Judgment

Most people read a judgment the way they read an article. They start at the first line, go to the last, and come away with a sentence they liked. Then they put that sentence in a written submission, and a judge asks what the case actually decided, and they discover they do not know.

Reading a judgment is not reading. It is extraction.

You are looking for one thing, most of the document is not it, and the parts that look most quotable are frequently the parts that bind nobody.

This is the long version of one step in how to do legal research in India. That essay treats reading a judgment as a stage in a larger method. This one is about the judgment itself.

What you are actually looking for

The binding part of a judgment is the proposition of law applied to the material facts, on which the outcome rests. That is the ratio decidendi. Everything else is furniture: narration, submissions recorded, authorities discussed and put aside, general observations, and the judge's views about the state of the world.

Those other parts are obiter dicta. Obiter is not worthless. Considered obiter from a large Bench carries real persuasive weight, and advocates cite it constantly. But if you describe obiter as a holding in front of a judge who knows the case, you will lose more than that point.

So the whole exercise reduces to one question, asked repeatedly as you read: does the decision rest on this sentence?

The anatomy of an Indian judgment

Before method, structure. An Indian judgment is a physical document with predictable parts, and knowing them saves you a great deal of time.

The cause title block. The first page carries the court, the case number, the parties, the Bench and the date. On recent Supreme Court judgments it also carries a neutral citation and a QR code in the top left corner. Read this block before anything else. The Bench strength alone tells you how much authority the judgment carries and whether it could have overruled what came before it.

“Leave granted.” Two words at the top of many Supreme Court judgments. They mean the Court has converted a special leave petition into a civil or criminal appeal under Article 136, and is now deciding the appeal rather than deciding whether to hear it. Their absence matters: an order dismissing a special leave petition without granting leave is not a decision on the merits, and it is not a precedent for anything, however tempting the sentence in it.

The paragraph numbers. These are the Court's own, they are identical in every copy, and they are how you will cite. Page numbers belong to a particular printing. Paragraphs belong to the judgment.

The operative portion. Usually the last few paragraphs. What was allowed, dismissed, set aside, remanded, and on what terms. In writ matters this is where the directions live, often numbered.

Separate opinions. Where more than one judge writes, the judgment will say who concurs with whom and on what. This is where most misreadings begin, and it has its own section below.

Read it out of order

Nobody tells students this, and it is the single change that produces the largest improvement.

Do not start at the beginning. Read a long judgment in this order:

1. The cause title block. Court, Bench, date, citation.
2. The last few paragraphs. Learn the outcome first.
3. The paragraph framing the questions for determination. Courts almost always set these out, often numbered.
4. The facts, now that you know which of them mattered.
5. The reasoning, which is finally comprehensible, because you know where it is going.

Reading forwards, you meet forty paragraphs of argument without knowing which side won, so you cannot tell which reasoning survived. Reading in this order, every paragraph arrives with a job.

A related discipline: read the whole thing once without highlighting anything. The urge to highlight arrives early, before you know what matters, and a marked-up judgment on a second reading tells you what a less informed version of you found striking.

The headnote is not the judgment

Headnotes are written by a publisher's editors. They carry no authority. They are a finding aid, and treating one as a substitute for reading is the most common defect in Indian legal research.

Use a headnote for exactly one thing: locating the paragraphs you need in a long judgment. Then close it and read those paragraphs.

This is not a hypothetical caution. Headnotes compress, and compression is interpretation. An editor deciding which of five propositions to lead with has made a judgment about the case, and you inherit it silently if you do not read past it.

Finding the ratio: what the Indian courts have said about it

This is the part most guides skip, and it is the part that settles arguments. Indian courts have addressed how to read their own judgments, repeatedly, and those decisions are citable.

A decision is an authority for what it decides

In State of Orissa v. Sudhansu Sekhar Misra, 1967 INSC 252, the Court put it as directly as it has ever been put:

“A decision is only an authority for what it actually decides. What is of the essence in a decision is its ratio and not every observation found therein nor what logically follows from the various observations made in it.”

Read the second sentence twice. Not every observation, and not what logically follows. A judgment does not authorise the next step in the argument merely because the next step seems to follow from it. That inference is yours, and you must defend it as yours.

What actually constitutes the precedent

Union of India v. Dhanwanti Devi, 1996 INSC 911 is the case most often cited for what does and does not form part of a precedent. It is worth knowing what it was about, because it makes the point better than the citation does.

It was a land acquisition matter, concerning solatium and interest under a Jammu and Kashmir statute. In the course of deciding whether two earlier decisions conflicted, the Court had to explain what the binding element of a judgment is. The passage everyone quotes on precedent therefore sits inside a dispute about compensation for requisitioned land.¹

The lesson is not incidental. The most quoted statements about how to read judgments were mostly made while deciding something else entirely. That is how law works, and it is also a warning: when you lift such a passage, check what the case was doing when it said that.

A judgment is not a statute

In Bharat Petroleum Corporation Ltd. v. N.R. Vairamani, 2004 INSC 569, the Court said something every law student copies out and almost nobody applies:

“Courts should not place reliance on decisions without discussing as to how the factual situation fits in with the fact situation of the decision on which reliance is placed. Observations of Courts are neither to be read as Euclid's theorems nor as provisions of the statute and that too taken out of their context.”

Two working consequences. First, you cannot cite a case without stating how its facts resemble yours. A citation with no fact comparison is decoration. Second, arguments built on the grammar of a single judicial sentence tend to fail, because judicial language is not drafted with a statute's precision and was never meant to bear that weight.

One different fact can change everything

Padma Sundara Rao v. State of Tamil Nadu, 2002 INSC 136, a Constitution Bench, warns against what it calls circumstantial flexibility: a single additional or different fact may make a world of difference between conclusions in two cases.

This is the answer to the student who has found a case with the same section and the same outcome and thinks the work is done. The section is not the case. The facts are the case.

What Article 141 actually binds you to

Article 141 provides that the law declared by the Supreme Court is binding on all courts within India. The working question is what counts as a declaration of law, and the Court has answered it: see Director of Settlements, A.P. v. M.R. Apparao, 2002 INSC 157, on what constitutes law declared for the purposes of that Article.

The short point is that not everything in a Supreme Court judgment is a declaration of law. An order made on the facts of one case, or a direction issued to do justice between particular parties, does not become a rule binding every court in the country merely because the Supreme Court made it.

The removal test, and why it is harder than it sounds

The practical method for separating ratio from obiter is subtraction. Take the proposition out. Ask whether the Court could have reached the same result without it. If the outcome falls, the proposition is ratio. If the Court would have arrived at the same order anyway, it is obiter, and you must call it an observation.

Three complications that the test does not solve on its own.

A judgment can have more than one ratio. Where a court decides an appeal on two independent grounds, each capable of supporting the result, both are generally treated as binding. Some judgments run four issues and decide all four.

Alternative reasoning is not the same as additional reasoning. A court that says “in any event, even if we are wrong about that” is signalling that the result does not depend on what follows.

You will sometimes be arguing about the ratio itself. That is not a failure of technique. Which proposition was necessary to the result is a genuinely contestable question, and appellate arguments are frequently about exactly that.

Material facts, which is the hard part

The ratio is the proposition applied to the material facts, so identifying which facts were material is most of the work.

A fact is material if the outcome would have changed without it. Most facts in a judgment are not. The dates, the amounts, the names of the villages, the procedural history: almost all of it is scaffolding.

The test that works is generalisation. Write the holding at the narrowest level the facts support, then ask how far up you can move before it stops being what the Court decided. A judgment about a driver who was an employee is not automatically a judgment about contractors. It might be. But that is an argument you have to make, not a reading you can assume.

More than one opinion, and the Constitution Bench problem

Here is where confident readers go wrong.

When several judges write separately, the binding element is what a majority actually agreed on, which is not always what the leading opinion says. Judges concur in the result while differing on the reasoning. A proposition in the longest and most eloquent opinion may command no majority at all.

So on a multi-opinion judgment, do this before you read for content: list the judges, note who wrote and who concurred, and work out issue by issue where the majority lies. On a nine-judge Bench, a proposition supported by four judges is not the law, however good it is.

The extreme case is instructive. Kesavananda Bharati v. State of Kerala, 1973 INSC 91 was decided by thirteen judges who produced eleven opinions running to several hundred pages. What is treated as the operative outcome is a short signed statement, “The View by the Majority”, setting out propositions the majority agreed on. Whether that statement is itself part of the judgment, and what precisely it binds, has been argued about ever since.²

There is a practical warning in that. Free full-text versions of very long judgments are sometimes incomplete, and the missing portion is not always obvious. I have found versions of Kesavananda in circulation where the paragraph numbering jumps and the majority statement is simply absent. If a judgment matters to your case, check that the copy you are reading is complete.

Two doctrines that destroy a citation

Both are worth knowing before your opponent uses them on you.

Per incuriam. A decision given in ignorance or forgetfulness of a binding statutory provision or an inconsistent decision of a coordinate or larger Bench is not binding. The doctrine is discussed in A.R. Antulay v. R.S. Nayak, 1988 INSC 123. It is a narrow doctrine and it is over-pleaded. It does not mean “decided without noticing an argument I like”.

Sub silentio. A point passes sub silentio where it was not perceived by the Court or present to its mind, so the decision assumes the answer rather than deciding it. See State of U.P. v. Synthetics and Chemicals Ltd., 1991 INSC 159, which discusses the binding effect of a declaration of law under Article 141 and the weight to be given to a proposition the Court never actually considered.

The practical use of both is defensive. When an opponent cites a judgment for a proposition the Court plainly never turned its mind to, you have a name for that.

Read the dissent

A dissent binds nobody, and students therefore skip it. That is a mistake for two reasons.

A dissent is the best available statement of the case against the majority, written by someone who read the same record. If you need to know the weakness in the authority you are relying on, it is there, argued at its strongest.

And dissents are occasionally the future. The reasoning that lost by one vote in a five-judge Bench is the reasoning a later, larger Bench may adopt. If you are advising on where the law may go rather than only where it is, the dissent is the most informative part of the judgment.

Read it against yourself

You will read a judgment because you hope it helps you. That hope is the main source of misreading.

So make one pass looking only for what is against you. The paragraph where the Court confines the holding to the facts. The sentence beginning “we make it clear that”, which almost always narrows something. The distinction the Court drew that also distinguishes your case.

You will find these anyway, eventually, because your opponent will find them. It is better to find them at your desk than at the podium.

Then check that it still stands

A judgment read perfectly is worthless if it has been overruled. Reading is not finished until you know how later Benches have treated the case: followed, distinguished, doubted, referred to a larger Bench, or overruled.

This is the single feature that separates a serious research tool from a search box, and it is the step most often skipped under time pressure. My longer essay on legal research deals with it in detail.

What the schools actually teach

It is worth looking at what two of the more demanding law schools tell their own first-year students, because they emphasise different things, and the difference is instructive.

Harvard Law School runs a pre-term course called Zero-L for incoming students, built with the university’s learning office and taught across eight modules by faculty including I. Glenn Cohen, its faculty director, Cass Sunstein and Noah Feldman. “How to read a case” is treated as a foundational skill, and the briefing method it teaches has four elements: the facts, the procedural history, the holding and the rationale. Students write those out and then compare their attempt against a model. A separate module asks the question that follows from all of it: precedent, and what the decision means for future cases.³

A correction worth making, since the two get conflated. IRAC, for Issue, Rule, Application and Conclusion, is not the briefing method. It is a framework for writing an answer, and it is taught very widely in the United States for that purpose. Harvard’s own course teaches the four-part brief above for recording a case. Reading a judgment and answering a problem are different tasks, and using the answer framework to take notes is why so many case notes read like half-finished essays.

NLSIU approaches it from another direction. Legal Methods, course code LLT100, is a core first-year course on the three-year LL.B. Its aim is to introduce the legal system and “characteristic ways of reading or reasoning with those sources”, with express attention to how to read statutes, subordinate legislation, case law and commentary. Its fourth stated objective is the one worth sitting with. The course is meant to convey

“the dialogic nature of argumentation and of the central place of interpretation and reasoned disagreement in the legal community.”⁴

Put the two side by side. Harvard teaches you to extract a judgment into a fixed template and then ask what it means for the next case. NLSIU teaches you that you are reading one side of a continuing argument, and that interpretation and disagreement are the normal condition of the material rather than a defect in it.

Both are right, and they correct each other. The template stops you drifting. The dialogue stops you believing that a judgment has one obvious meaning that a careful reader simply collects. If you only have the template you will produce tidy notes and lose arguments about what the case decided. If you only have the dialogue you will have opinions and no record.

What to write down

So take the discipline of the first and the scepticism of the second. For Indian practice, a note on a judgment needs eight lines:

1. Case name, neutral citation, court, Bench strength, date.
2. The question the Court framed, in your own words.
3. The material facts, in two sentences.
4. The holding, with the paragraph number.
5. The reasoning in one sentence: because.
6. What is obiter but useful, marked as observation.
7. What in it is against you.
8. How later Benches have treated it.

If you cannot complete line four with a paragraph number, you have not finished reading. And if a judgment cannot be reduced to these eight lines, that is usually information about the judgment.

I have made this into a one-page sheet you can print, with the prompts and the checks built in: the Judgment Note. It is free, and you are welcome to photocopy it for a class or a moot workshop.

How long this takes

An honest answer, because the guides never give one.

A short two-judge judgment on a settled point takes fifteen minutes. A contested judgment you are going to rely on takes an hour, including the treatment check. A Constitution Bench judgment of any length takes an afternoon, and if it has separate opinions it takes longer, because working out the majority on each issue is itself a task.

Anyone who tells you they read a hundred-page judgment in ten minutes read the headnote.

The mistakes worth naming

• Quoting from the middle of a sentence. Indian judgments are full of long sentences that reverse direction at a comma. Read to the full stop before you quote.
• Citing the Court's summary of counsel's argument as the Court's own view. Judgments record submissions at length. “It was contended that” is not a holding.
• Treating a dismissal of a special leave petition as a precedent. It is not.
• Citing a case for a proposition it assumed rather than decided. Sub silentio.
• Citing without a paragraph number. It tells the judge you did not read it.
• Relying on a headnote. Which is where most of the above begins.

A closing thought

The reason this skill matters more now than it did five years ago is that summaries have become free and abundant, and a summary is an interpretation with the workings removed. Somebody decided which sentence to lead with, and you inherit that decision without seeing it.

A judgment is one of the few documents in public life that shows its own reasoning. It tells you what it decided and why, and it does so in the open, at a paragraph you can point to. Almost nothing else does that.

Reading it properly is not a scholarly nicety. It is the difference between citing an authority and borrowing somebody's impression of one.

Filed under The Legal Profession