Blog Subject guides IGCSE Physics 0625 Paper 6: Precisio...
Subject guides

IGCSE Physics 0625 Paper 6: Precision, Graphs and the Marks Everyone Loses

PapaMarks Team · July 25, 2026 · 10 min read
#Physics #0625 #Paper 6 #Alternative to Practical #Graphs #Errors #Exam technique

Physics Paper 6 is the only paper you sit where you can get the right number and still lose the mark.

Write 1.4 when the column shows 1.40. Head a column Length instead of length / cm. Calculate a gradient from two points that are almost touching. Every one of those is a correct answer that scores zero, and together they account for more lost marks in Alternative to Practical than any genuine physics misunderstanding.

The good news is that Physics Paper 6 recycles a small set of experiments. Pendulums, springs, resistance wires, lenses, cooling curves, density. Once you know the set and you have fixed the precision habits, this becomes the most reliable 40 marks on your timetable.

⚡ The 60-second version
  • 1 hour, 40 marks, 20% of your Physics grade. Same weighting as the practical test.
  • Time 20 oscillations and divide by 20. This one technique appears in a huge share of past papers and students still time a single swing.
  • Repeating and multiple-counting fix different errors — examiners ask which, and most candidates conflate them.
  • Gradient triangles must be big — spanning at least half your line. A small triangle loses the mark even with correct arithmetic.
  • Decimal places must match your instrument and stay consistent down the whole column, including whole numbers.
  • Six experiments cover most of the paper: pendulum, spring, resistance wire, lens, cooling, density.

The six experiments that keep coming back

Cambridge does not have an infinite supply of one-hour experiments that work on paper. Look across five years of 0625 Paper 6 and the same contexts cycle through, lightly re-dressed.

ExperimentWhat you measureThe skill being tested
Simple pendulumPeriod against lengthTiming many oscillations; reaction-time error
Spring / load–extensionExtension against loadReading a scale; proportionality; limit of proportionality
Resistance of a wireV and I, or R against lengthCircuit diagrams; ammeter/voltmeter placement; gradient
Lens / focal lengthImage distance against object distanceRay boxes; sharp-image judgement; repeat readings
Cooling curveTemperature against timeThermometer reading; insulation; anomalies
DensityMass and volumeDisplacement method; meniscus; irregular solids

Refraction through a glass block, moments and balancing a metre rule, and specific heat capacity appear regularly too. If you have worked through all of these once, very little in the exam will be genuinely unfamiliar.

🎯
Do not revise these as topics — revise them as procedures. For each one, write down on a single card: the apparatus, the independent and dependent variables, what must be controlled, the main source of error, and the improvement that fixes it. Six cards. That is most of the paper's non-mechanical marks, and it takes an evening.

The technique that earns the most marks: count many, then divide

If a question asks you to find the period of a pendulum, the answer is never "start the stopwatch, let it swing once, stop the stopwatch."

You time 20 oscillations and divide by 20.

The reason matters, because examiners ask for it. Your reaction time — roughly 0.2 s at each end — is a fixed, absolute error. It does not shrink when the measurement gets shorter. So:

Time 1 swingTime 20 swings
Measured time~0.8 s~16 s
Reaction-time error~0.4 s~0.4 s
Percentage error~50%~2.5%

The absolute error is identical. The percentage error collapses. That is the whole argument, and writing it in that form — "the timing error is the same but is a much smaller fraction of a longer time" — is what scores.

⚠️
Repeating and multiple-counting are not the same thing, and examiners test the difference. Repeating a reading three times and taking a mean reduces random error — scatter around the true value. Timing 20 oscillations instead of one reduces the percentage uncertainty from a fixed timing error. Neither fixes a systematic error such as a zero error on a balance or a ruler with a worn end — only recalibrating or measuring differently does that. Candidates who write "repeat and average" as the answer to every error question lose marks whenever the error is systematic.

Precision: where correct answers score zero

Decimal places follow the instrument, not the number

If you read a ruler to the nearest millimetre, every length in that column has one decimal place in centimetres. A reading that lands exactly on 16 cm is written 16.0, not 16. Dropping the zero implies you measured less precisely than you did, and the mark scheme catches it.

The same applies to a stopwatch reading 12.30 s — the trailing zero is information, not decoration.

Table headings carry the unit, and only there

Cambridge's convention is quantity / unit:

CorrectLoses a mark
length / cmlength
time / stime (seconds) written next to every value
resistance / ΩR
T² / s²T squared

The unit goes in the heading once. It never appears beside individual numbers in the body of the table.

Reading a scale correctly

  • Parallax — read at eye level, perpendicular to the scale. On a ruler, the classic fix is "view the scale from directly above the mark."
  • Meniscus — read the bottom of the curve for water, at eye level.
  • Zero error — check the instrument reads zero before you start. If it does not, subtract the offset from every reading. This is a systematic error and no amount of repeating will remove it.
  • The worn ruler end — start measuring from the 1.0 cm mark rather than the end, and subtract. Examiners like this answer.

Graphs: the four marks almost everyone can get

A typical Paper 6 graph question is worth four or five marks, and they are awarded almost mechanically. Get these in order and they are yours.

  1. Axes — labelled with quantity and unit
    Same convention as the table: extension / cm. Independent variable on the x-axis. This mark requires no physics at all.
  2. Scale — fill at least half the grid
    If your plotted points occupy less than half the printed grid in either direction, you lose the scale mark. Use 1, 2, 5 or 10 units per square. Never 3 or 7 — the arithmetic afterwards becomes a trap.
  3. Plotting — small, sharp crosses
    Accurate to within half a small square. Use a fine pencil and a small neat cross or encircled dot, not a blob covering two squares.
  4. Best-fit line — one thin line, judged by eye
    Roughly equal numbers of points either side. Do not join the dots. Do not force it through the origin unless the physics requires it. If the question expects a curve, draw a smooth curve, not a series of straight segments.
  5. Gradient — a large triangle, drawn on the graph
    Your triangle must span at least half the length of your line. Draw it, label the co-ordinates you used, and show the subtraction. A gradient with no visible triangle usually loses a mark even if the number is right.
📐
Take your co-ordinates from the line, not from your data. The whole point of a best-fit line is that it averages out the scatter. Reading the gradient from two of your original plotted points throws that away — and if you happened to pick the anomaly, your answer is wrong as well as unmarked. Pick two points where the line crosses convenient grid intersections.

Errors and improvements: the two-clause rule

The last part of most Paper 6 questions asks for a source of error, an improvement, or both. This is the biggest mark-loser in the paper, and the fix is a sentence structure.

Name the change, then name the error it removes.

Scores nothingScores the mark
"Use a better stopwatch.""Time 20 oscillations and divide by 20, so the reaction-time error is a smaller fraction of the total time."
"Repeat it.""Repeat each reading three times and take a mean, to reduce random error."
"Stop heat escaping.""Put a lid on the beaker to reduce heat loss by evaporation."
"Measure more carefully.""View the scale from directly above the mark to avoid parallax error."
"Use a thinner wire.""Switch off between readings so the wire does not heat up and change its resistance."

If your improvement is one clause long, it is almost certainly worth nothing. Two clauses, every time.

🔌
Circuit questions have their own recurring marks. The ammeter goes in series with the component; the voltmeter goes in parallel across it. Draw circuit symbols with a ruler and keep wires as straight lines meeting at right angles — a wobbly freehand circuit can genuinely cost you the diagram mark. And if asked why readings drift upward during a resistance experiment, the answer is that the wire warms up.

If you have never done these experiments

Many schools enter students for Paper 6 rather than Paper 5 precisely because they do not have the lab time or equipment. That leaves a real gap: the questions assume you have watched a spring stretch past its limit of proportionality, or seen a ray bend at a glass block.

PapaMarks has free interactive versions of several of these — load a spring and watch the load–extension graph build in real time, measure the density of an irregular solid by displacement, or send a ray through a glass block and watch the refraction angle change. They are quick, and they turn the written descriptions in the exam into something you have actually seen happen.

Common questions

How long is Physics 0625 Paper 6 and what is it worth?
One hour, 40 marks, and 20% of the qualification. It is the alternative to Paper 5, the hands-on practical test, and carries identical weighting.
Why time 20 oscillations instead of one?
Your reaction time introduces a fixed error of roughly 0.4 s regardless of how long you time for. Over one swing of about 0.8 s that is around 50% error; over 20 swings of about 16 s it is around 2.5%. The absolute error is unchanged — the percentage uncertainty is what collapses.
What is the difference between random and systematic error?
Random error causes scatter around the true value and is reduced by repeating and averaging. Systematic error shifts every reading in the same direction — a zero error, a worn ruler end, a mis-calibrated instrument — and repeating does not help at all. You must correct the instrument or change the method.
How big does my gradient triangle need to be?
It should span at least half the length of your best-fit line. A small triangle magnifies any reading error and mark schemes routinely withhold the mark for one. Draw the triangle on the graph and label the co-ordinates you used.
Do I need to memorise formulas for Paper 6?
Paper 6 is about experimental skill rather than recall, but you will need the standard relationships that underpin the recurring experiments — density as mass over volume, resistance as voltage over current, and how a gradient relates to the quantity being investigated. Knowing what the gradient means physically is examined regularly.
What else should I read?
Our complete Paper 6 guide covers the skills shared across all three sciences, and the Chemistry 0620 guide covers qualitative analysis. For the theory papers, see most-tested 0625 topics and how to answer 6-mark Physics questions.

The one-line version

Learn the six recurring experiments as procedures, always count many and divide, keep your decimal places consistent, draw a gradient triangle that spans half the line, and never write a one-clause improvement. None of that is physics knowledge — which is exactly why it is the fastest 20% of your grade to secure.

Put this into practice — free

4,463+ past papers, flashcards and an AI tutor for O Level, AS & A2. No credit card.

Start free →

More from the blog