Science Practical Graph Skills: How to Draw Graphs That Score Full Marks

Study guideUpdated 17 Jul 2026
Marcus Pang
Reviewed by
Marcus Pang·Managing Director (Maths)
TL;DR
Graph drawing is tested in every science practical - Physics, Chemistry, and Biology - at both O-Level and A-Level. Most marks are lost not because of wrong data, but because of presentation mistakes: poor scales, missing units, dot-to-dot lines, and tiny gradient triangles. This guide consolidates every SEAB graph convention in one place so you never lose a graph mark again.

Why a cross-science graph guide?

Graph-drawing rules are nearly identical across Physics Paper 3, Chemistry Paper 3, Biology Paper 3, and all three A-Level Paper 4s. But students rarely see them consolidated - they are scattered across different subject notes, and teachers assume students learned them in another subject. The result is confusion: "Must the line pass through the origin?" is one of the most-asked questions on student forums, and the answer is the same regardless of subject.

This guide covers the universal rules first, then the subject-specific variations.


1 | Line graph vs bar chart - which one?

Use a line graph when both variables are continuous (numbers that can take any value). This covers almost every physics and chemistry practical: time vs temperature, voltage vs current, concentration vs volume.

Use a bar chart when the independent variable is categorical (names or labels, not numbers). This appears mainly in biology: comparing enzyme activity across different substrates, or comparing population counts across different quadrat sites.

If in doubt, look at the independent variable (x-axis). If it is a number with units, use a line graph. If it is a category or name, use a bar chart.


2 | Scale selection

Choosing the right scale is worth a PDO mark in every subject. The rule is consistent across SEAB syllabuses:

Plotted points must span more than half the grid on both axes.

How to achieve this:

  1. Find the range of your data on each axis (highest value minus lowest value).
  2. Count the number of large squares available on the grid (usually 10 - 15 on each axis).
  3. Choose a scale where the data range covers at least 6 - 8 of those large squares.
  4. Use easy intervals. The only acceptable scale divisions per large square are: 1, 2, 5, or multiples of 10 (e.g., 10, 20, 50, 100). Never use 3, 4, 6, 7, 8, or 9 per large square - these make plotting and reading values error-prone.
  5. You do not need to start from zero unless the data includes zero or the physics requires it. Starting from zero when your data goes from 40 to 90 wastes most of the grid.

Example: Your temperature data ranges from 22 °C to 78 °C range=56∘C range = 56 ^{\circ}C . You have 10 large squares on the y-axis. A scale of 5 °C per large square would span 50 °C across 10 squares - that is 11 squares for 56 °C of data (you would start at 20 °C and go to 80 °C). This fills the grid well.

A common mistake is to start at 0 °C: that would put 22 - 78 °C in the top third of the grid only, wasting space and compressing the data. Do not do this unless the question specifically asks for an origin.


3 | Axis labels

Every axis must be labelled with quantity and unit, separated by a forward slash:

  • Temperature / °C
  • Resistance / Ω
  • Time / s
  • Length / cm
  • Volume of gas / cm³

Not acceptable: "Temperature", "Temp", "T (°C)", or unlabelled axes. The forward-slash convention ("quantity / unit") is what SEAB mark schemes specify.

For derived quantities, include the full label:

  • 1/Length / cm⁻¹
  • ln(Voltage / V) - for A-Level linearisation

4 | Plotting points

Use small, neat crosses (×). Each arm of the cross should be about 1 - 2 mm. Do not use:

  • Dots (they are hard to see and verify)
  • Circles (they cover too large an area)
  • Large blobs (the examiner cannot determine the exact position)

Plot to half-a-small-square accuracy. The examiner checks whether each point is within half a small square of its correct position. Use a sharp pencil and count squares carefully.

All points must be plotted. Do not skip data points unless the question tells you to.


5 | Best-fit lines

This section answers the most-asked question on student forums.

Straight best-fit line

Draw a single straight line using a transparent ruler. The line should follow the overall trend of the data, with roughly equal numbers of points above and below the line. It does not need to pass through any particular point, and it does not need to pass through the origin.

When does the line pass through the origin? Only when the physics or chemistry of the experiment requires it - for example, a proportional relationship where zero input gives zero output (Hooke's Law: zero force means zero extension). If the relationship is linear but has an intercept e.g.,y=mx+cwherec≠0 e.g., y = mx + c where c ≠ 0 , the line should not be forced through (0, 0). If you are unsure, let the data decide: draw the best-fit through the points and see where the line naturally intersects the y-axis.

Curved best-fit line

If the data clearly curves, draw a single smooth freehand curve through the trend. Do not use a ruler. The curve should be smooth and continuous - not wobbly or kinked. It should follow the overall shape of the data, not pass through every point.

What NOT to do: dot-to-dot

Never connect plotted points with short straight segments. This is the single most common graph mistake and earns zero marks for the best-fit line. A dot-to-dot line does not represent a trend - it just joins data points in order, including any measurement errors.


6 | Anomalous points

If one point clearly deviates from the trend of the other points:

  1. Circle it with a pencil.
  2. Ignore it when drawing the best-fit line.
  3. Comment on it in your ACE evaluation: "Point 4 was anomalous, possibly due to [specific cause]. It was excluded from the best-fit line."

Do not force the best-fit line through an anomaly. Do not delete the point from your table.


7 | Gradient calculation

When a question asks you to find the gradient of a straight-line graph:

  1. Use the best-fit line, not your plotted data points. Pick two points that are on the line itself.
  2. Draw a large right-angled triangle. The horizontal and vertical sides of the triangle should span at least half the length of the line. A tiny triangle introduces rounding errors and loses the mark.
  3. Read the coordinates of the two points from the graph and write them down.
  4. Calculate: gradient = (y₂ − y₁) / (x₂ − x₁)
  5. Include units. The gradient has units: [y-axis unit] / [x-axis unit]. For example, if the y-axis is Resistance / Ω and the x-axis is Length / cm, the gradient has units Ω cm⁻¹.
  6. Quote to 2 - 3 significant figures, consistent with the precision of your data.

8 | Rate from a graph

Students frequently confuse two different calculations. The question wording tells you which one to use.

"Find the rate at time t" → draw a tangent

  1. At the specified time, draw a straight line that just touches the curve (a tangent).
  2. Extend the tangent line until it spans a large area of the grid.
  3. Calculate the gradient of the tangent line using the method above.
  4. This gives the instantaneous rate at that point.

"Find the average rate between t₁ and t₂" → total change ÷ time

  1. Read the y-values at t₁ and t₂ from the graph.
  2. Calculate: average rate = (y₂ − y₁) / (t₂ − t₁)

These are different calculations with different answers. Check the question carefully.


9 | Subject-specific graph notes

Physics

  • Linearisation is common at A-Level: plotting ln(y) vs ln(x) or y² vs x to produce a straight line from a power-law or quadratic relationship. See the Physics Practical Maths Toolkit for linearisation techniques.
  • Extrapolation (extending the line beyond the data) should be shown as a dashed line, not a solid line.

Chemistry

  • Titration curves (pH vs volume) use a smooth S-curve, not a straight line. The equivalence point is at the steepest part of the curve.
  • Reaction rate graphs (volume of gas vs time) typically start steep and level off. The initial rate is the gradient of the tangent at t = 0.

Biology

  • Bar charts should have equal-width bars with gaps between them (unlike histograms, which have no gaps).
  • Error bars may be required at A-Level: draw them symmetrically above and below each data point or bar.
  • For scatter plots of ecological data, a line of best fit is appropriate only if the question asks for one.

10 | Common graph mistakes - before and after

MistakeWhy it costs marksFix
Starting both axes at zero when data starts at 40Data compressed into one corner; fails the "more than half the grid" ruleStart the axis at a sensible value near the lowest data point
Scale of 3 per large squareImpossible to plot or read values accuratelyUse 1, 2, 5, or 10 per large square only
Axis labelled "T" with no unitExaminer cannot verify what the axis representsWrite "Temperature / °C"
Dot-to-dot lineNot a best-fit; does not represent a trendDraw one smooth line through the trend
Gradient triangle using two data pointsData points have measurement error; only the best-fit line represents the trendUse two points on the best-fit line
Gradient triangle spanning 2 squaresToo small; large rounding errorSpan at least half the line length
Tangent confused with average rateDifferent calculations give different answersCheck the question wording: "at time t" = tangent; "between t₁ and t₂" = average

Related guides


Sources