How to Write a Lab Report: What Each Section Is Actually For

The experiment took three hours. The report is due at nine tomorrow and you have a table of numbers, a graph your software produced, and a page and a half that mostly restates the lab manual.

The part you're stuck on is the discussion, which is also the part carrying the most marks. That combination is not a coincidence — it's the section where the writing has to do something the data can't do by itself.

Lab reports are a formulaic genre, which sounds like bad news and is actually the opposite. A formula means the assessment criteria are stable and knowable. Here's what each section is being judged on.

1. The mark distribution nobody tells you about

Rubrics vary, but the pattern across most undergraduate science courses looks roughly like this:

The discussion is the largest single component and the one students spend the least time on. Reports are usually written front to back, and the discussion is what gets written at 1am with the energy that's left.

Reverse the order. Write the discussion when you're fresh, and the methods section — which is nearly mechanical — when you're tired.

Discussion section: a 5-step self-checklist

Run this on your discussion before you submit. Each step takes a minute, and together they cover most of what a marker looks for in the section that carries a third of the grade. The worked example runs through one experiment — a pendulum measurement of g — so you can see each check applied to the same data.

If step 3 or 4 fails, that is where the marks are. Those two are the difference between a discussion that reports and one that reasons.

2. Introduction: the funnel, and the hypothesis

An introduction moves from general to specific in three or four paragraphs:

1. The broader phenomenon and why it matters
2. The specific principle this experiment tests, with the relevant theory stated
3. What's known and what the experiment will establish
4. The hypothesis, in a form that could turn out false

The last one is where marks go missing. "We investigated the relationship between temperature and reaction rate" is an aim, not a hypothesis. A hypothesis makes a directional prediction that the data can contradict: "Reaction rate will increase approximately exponentially with temperature, consistent with the Arrhenius relationship."

Include the governing equation here rather than in the discussion. The introduction is where you establish what the theory predicts; the discussion is where you compare that prediction to what happened. Splitting them this way makes the discussion much easier to write.

3. Methods: the reproducibility test

One question governs this section: could a competent person in your field repeat this and get comparable data?

That standard resolves nearly every judgment call about what to include:

• Instrument model and settings — yes, this is what makes results comparable
• Concentrations, volumes, temperatures, durations — yes, with the precision you actually used
• Number of trials and how you handled repeats — yes, and this is commonly missing
• "We obtained a beaker from the cupboard" — no

Two conventions worth getting right because they're easy marks. Past tense, and usually passive or impersonal — "the solution was heated to 60 °C" — though a growing number of journals and courses now accept "we heated." Check your department's guidance; both are defensible, inconsistency is not.

Describe what you did, not what the manual said to do. If the manual said 60 °C and your bath held 58, the report says 58. Deviations belong in the methods, and they frequently explain something in your results — which turns a mistake into an analysis point.

4. Results: presentation without interpretation

The discipline of this section is negative. You are showing what you measured and not yet saying what it means.

• Every figure and table gets a number and a caption that can be understood on its own.
• Figure captions go below, table captions above, in most conventions.
• The text points to the data and states the pattern: "Rate increased with temperature across the range tested (Figure 2)."
• Uncertainties and units appear everywhere. A number without units is not a result.
• Raw data goes in an appendix if it's bulky.

Two things reliably cost marks here. Interpreting early — "this shows the reaction is endothermic" belongs in the discussion. And duplicating a table as a graph: pick whichever representation carries the point, because presenting both signals you don't know which one does.

Significant figures matter more than students expect. Reporting 12 decimal places from a calculator when your instrument resolves to two claims a precision you don't have, and markers in physical sciences treat that as a conceptual error rather than a formatting one.

5. Discussion: the section that carries the grade

Six moves, in roughly this order. Most weak discussions contain the first and skip the rest.

1. Did the results match the prediction? State it directly, with numbers. "The measured value of 9.6 ± 0.3 m/s² agrees with the accepted 9.81 m/s² within uncertainty."

2. If not, by how much and in which direction? Percentage error, and — importantly — the sign. A result consistently below the accepted value points at a systematic effect, and that's a far more interesting sentence than "there was some error."

3. What are the sources of error, specifically and quantitatively? This is the highest-yield paragraph in the report and the one most often wasted on generic filler.

"Human error" earns nothing. Name the mechanism and estimate the size:

"Heat loss to the surroundings during transfer would bias the measured final temperature downward. Given the roughly 15-second transfer and the observed cooling rate, this could plausibly account for 1–2 °C, which is the direction and approximate magnitude of the discrepancy observed."

That single paragraph structure — mechanism, direction, magnitude, comparison to the observed gap — is what separates a strong report from an average one, and it takes fifteen minutes.

4. Distinguish random from systematic. Random error scatters your repeats and shrinks with more trials. Systematic error shifts every measurement the same way and doesn't. Confusing the two is a conceptual mistake markers watch for.

5. Limitations of the design itself. Not just execution. What could this apparatus never have shown?

6. What would you change, concretely. "Be more careful" is not an improvement. "Insulate the calorimeter and record the transfer time so the correction can be applied" is.

6. Where tools help with a lab report

The bottleneck in most lab reports is not the writing. It's understanding the physical principle well enough to reason about the error, which is exactly what the discussion demands.

The theory you're comparing against. If the governing relationship is one you half-follow, the discussion will stay generic no matter how long you spend on it. Video Explanation under Ask Sovi walks a specific derivation or principle through in about a minute, which is usually the difference between "there was some error" and a paragraph naming a mechanism.

The lab manual and the lecture material. These carry the constraints your report is graded against — the required method, the expected uncertainty treatment, the format. Running the actual manual through AI Notes keeps your preparation anchored to your course's version rather than the internet's general treatment of the experiment, which often uses different apparatus entirely.

Checking your own arithmetic. Propagation of uncertainty is where quiet errors live. A computer algebra system or a spreadsheet you built yourself is a genuinely independent check; asking the same assistant that produced a number to confirm it is not.

Once the error analysis is written, Smart Writing's revise-and-polish side is a reasonable last pass on prose you produced — lab reports are a conventional genre, and tightening a discussion you reasoned out yourself is editing rather than substitution.

What to keep away from a lab report: generated discussion text. Error analysis is specific to your apparatus, your measurements and your deviations, and a general assistant that wasn't in your lab will produce plausible, generic error sources — precisely the thing the rubric penalizes. It's also the section where a viva-style question ("why did you think heat loss explained the sign?") is most likely to come.

7. Six errors that cost easy marks

• A hypothesis that can't be wrong. If no possible result would contradict it, it isn't one.
• Interpretation in the results section. Keep the wall between describing and explaining.
• "Human error" as an error source. Name a mechanism, give a direction, estimate a size.
• Units and significant figures. Free marks, lost through haste, in every cohort.
• Reporting the manual's procedure instead of yours. Deviations are data.
• Writing the discussion last and tired. It's a third of the mark. Write it first.

<br>

Frequently Asked Questions

Q1: What goes in a lab report discussion?

Six things, in order: whether the results matched the prediction, the size and direction of any discrepancy, specific error sources with estimated magnitudes, a distinction between random and systematic effects, limitations of the design itself, and concrete improvements. The error paragraph carries the most weight and is the most commonly wasted — naming a physical mechanism, its direction, and roughly how much of the observed gap it explains is what separates a strong discussion from a generic one.

Q2: How do I write about sources of error without saying "human error"?

Identify a specific physical process, say which way it would bias the measurement, and estimate its size. Heat lost during a 15-second transfer biases the final temperature downward and might account for one or two degrees. Parallax on a meniscus read from above biases volume readings one consistent way. The pattern is mechanism, direction, magnitude — and then compare that magnitude to the discrepancy you actually observed.

Q3: Should a lab report be written in the passive voice?

Traditionally yes for the methods section — "the solution was heated" — and many courses still require it, but a growing number of journals and departments accept first person. Check your course guidance and then be consistent, because inconsistency is the version that reliably costs marks. Whatever you choose, methods are written in past tense, because you're reporting what was done rather than issuing instructions.

Q4: How much detail belongs in the methods section?

Enough that a competent person in your field could repeat the experiment and obtain comparable data. That test resolves almost every borderline case: instrument settings, concentrations, durations and number of trials are in; routine glassware handling is out. Record what you actually did rather than what the manual specified — deviations belong here, and they often turn out to explain something in your results.