Sources of Error in O-Level Physics Practicals: Experiment-by-Experiment Reference Bank
14 Apr 2026, 00:00 Z
Reviewed by
Chee Wei Jie·Academic Advisor (Physics)
Want small-group support? Browse our O-Level Physics Tuition hub. Not sure which level to start with? Visit Physics Tuition Singapore.
Looking for the full lab practical series? Visit the O-Level Physics Practicals.
Practical course completion-record note
For practical, lab, and experiment courses, Eclat Institute maintains centre-held attendance records and may also issue an internal attendance or completion document based on participation and internal assessment.
- For SEAB private-candidate declarations, the key evidence is the centre's attendance or completion record, not a government-issued certificate.
- This is an internal centre-issued certificate, not an MOE/SEAB qualification or accreditation.
- Recognition (if any) is determined by the receiving school, institution, or employer.
- For SEAB private candidates taking science practical papers, SEAB states you should either have taken the subject before or attend a practical course and complete it before the practical paper date.
View our sample completion document (Current sample layout (design may be refined over time))
Planning a revision session? Use our study places near me map to find libraries, community study rooms, and late-night spots.
> **TL;DR**\
> "Human error" is not an acceptable source of error in O-Level Physics Paper 3. Every error must name the physical mechanism and state whether it is systematic or random.\
> This bank covers DC circuits, pendulum, optics, thermal experiments, and resistance of wire. Each entry includes the error type (systematic or random), the direction of effect, a model ACE sentence, and an improvement suggestion.\
> PDO reminder: "5 cm" is not the same as "5.0 cm" - record to instrument precision every time.
Use this bank alongside the [O-Level Physics Experiments hub](https://eclatinstitute.sg/blog/o-level-physics-experiments) to locate the companion procedure post for each experiment. For ACE writing technique and marking habits, see [O-Level Physics Paper 3 Marking Micro-Habits](https://eclatinstitute.sg/blog/o-level-physics-experiments/O-Level-Physics-Paper-3-Marking-Micro-Habits) and the [O-Level Physics Error and Uncertainty Masterclass](https://eclatinstitute.sg/blog/o-level-physics-experiments/O-Level-Physics-Error-And-Uncertainty-Masterclass).
This page is the per-experiment reference bank: it gives you named errors and model ACE sentences for each Physics experiment. If you are not yet confident about the underlying technique -- what makes an answer valid, the random-versus-systematic taxonomy, five reusable error templates, and how to write a one-sentence mitigation -- read [How to Write a Source of Error in a Singapore Science Practical](https://eclatinstitute.sg/blog/o-level-physics-experiments/How-to-Write-Source-of-Error-Singapore-Science-Practical) first, then return here for the Physics-specific entries.
---
## The rule that earns and loses marks
**"Human error" is not an acceptable source of error in O-Level Physics Paper 3.**
This rule appears in SEAB examiner reports across multiple years. "Human error" tells the examiner nothing about the physical mechanism and therefore cannot be linked to an improvement. Every source of error you write must:
1. **Name the physical cause** (e.g. zero error on the ammeter, parallax on the metre rule, heat loss through the beaker).
2. **Classify it as systematic or random** (see below).
3. **State the direction** (does this make the recorded value too high or too low?).
4. **Propose a specific improvement** (something physically achievable in a school lab).
---
## Systematic versus random errors: a quick distinction
**Systematic errors** shift every reading in the same direction by approximately the same amount. They cannot be reduced by repeating the experiment and averaging. Examples: a zero error on an ammeter that reads 0.2 A when no current flows, a thermometer that consistently reads 1.5 degrees too high.
**Random errors** produce scatter around the true value. Readings are sometimes too high and sometimes too low. They can be reduced by repeating the experiment and averaging. Examples: reaction time when starting and stopping a stopwatch, parallax variation when reading a metre rule without a consistent eye position.
The ACE evaluation should name which type each error is, because the improvement differs. Systematic errors are corrected by calibration, zeroing, or subtracting the error. Random errors are reduced by repeating and averaging.
---
## A note on PDO precision
Before the experiment banks, a precision rule that affects PDO marks directly.
Recording "5 cm" instead of "5.0 cm" when using a metre rule (which reads to 1 mm = 0.1 cm) is a PDO error. It implies the measurement was only made to the nearest centimetre. Similarly, "0.5 A" instead of "0.50 A" on an ammeter that reads to 0.01 A misrepresents the precision of your instrument.
The rule: **record every measurement to the precision of the instrument, not to round numbers**.
This is not about sources of error in the ACE strand. It is a PDO marking criterion. If your table shows "5 cm" where you mean 5.0 cm, you will lose a significant figure mark before the examiner even reads your evaluation. The two issues (PDO precision and ACE evaluation errors) are separate, and both cost marks.
---
## Experiment 1: DC circuits (current-voltage relationships, resistance)
Related posts: [O-Level Physics Electricity and Magnetism Practical Manual](https://eclatinstitute.sg/blog/o-level-physics-experiments/O-Level-Physics-Electricity-Magnetism-Practical-Manual), [How to Connect an Ammeter and Voltmeter in a Circuit](https://eclatinstitute.sg/blog/o-level-physics-experiments/How-to-Connect-Ammeter-Voltmeter-Circuit-O-Level-Physics)
### Error 1.1 - Zero error on the ammeter or voltmeter (systematic)
**Type:** Systematic
**Why it matters:** If an ammeter reads 0.05 A when no current flows (a positive zero error), every current measurement is 0.05 A higher than the true value. This shifts the entire $I$-$V$ graph upward by 0.05 A but does not change the gradient. It produces an incorrect y-intercept and therefore an incorrect calculated resistance from the graph.
**ACE sentence:** "A source of systematic error was a zero error of +0.05 A on the ammeter. Every current reading was 0.05 A higher than the true value, shifting the data points on the I-V graph upward without changing the gradient. The calculated resistance from the gradient was not affected, but any resistance calculated from individual V/I ratios was slightly underestimated."
**Improvement:** Check the ammeter zero before connecting the circuit. If a zero error is present, record it and subtract it from every reading, or adjust the pointer to zero using the zero-adjustment screw if available.
---
### Error 1.2 - Contact resistance at crocodile clip connections (systematic)
**Why it matters:** Crocodile clips may not make a clean metal-to-metal contact if the wire surface is oxidised or the clip jaws are corroded. This adds an extra resistance in the circuit that is not the resistance of the component under test. The measured voltage across the component (or the current through it) will be affected by this additional contact resistance, causing the calculated component resistance to be overestimated.
**ACE sentence:** "A source of systematic error was contact resistance at the crocodile clip connections. An oxide layer on the wire surface added approximately 0.5 ohms to the circuit resistance that could not be distinguished from the component resistance, causing the measured total resistance to overestimate the true resistance of the wire."
**Improvement:** Sand or scrape the wire ends with fine sandpaper before attaching the clips to expose clean metal. Press the clip jaws firmly onto the wire and check that the wire does not slip. Test the connection by looking for stable meter readings before recording any data.
---
### Error 1.3 - Resistance of ammeter and resistance of voltmeter affecting measurements (systematic)
**Why it matters:** An ammeter has a small but non-zero resistance. A voltmeter has a large but finite resistance. If the voltmeter is connected across the component and the ammeter is on the same side, the voltmeter draws a small current. The ammeter reads the sum of the component current and the voltmeter current, overestimating the component current and underestimating the apparent resistance.
**ACE sentence:** "A source of systematic error was the finite resistance of the voltmeter. The voltmeter drew a small current (approximately 0.02 A at 6 V), which was measured by the ammeter as part of the component current. This caused the ammeter reading to overestimate the true current through the resistor by approximately 5 percent, underestimating the calculated resistance."
**Improvement:** For a low-resistance component, use the ammeter-external configuration (voltmeter across the component, ammeter outside the voltmeter loop) to minimise the voltmeter current error. For high-resistance components, use the ammeter-internal configuration. Acknowledge which configuration was used and note its direction of error in the evaluation.
---
### Error 1.4 - Resistance of the component increasing with temperature during measurement (random/systematic)
**Why it matters:** For metallic conductors, resistance increases with temperature. If current is passed through the component for a long period while readings are taken, the component warms up. Later readings in the series reflect a higher temperature and therefore a higher resistance than early readings. The $I$-$V$ graph curves upward at higher currents, making the relationship appear non-linear when it should be linear for an ohmic conductor.
**ACE sentence:** "A source of error was the resistance of the nichrome wire increasing as it was heated by the current during the experiment. At the highest current setting, the wire temperature was noticeably higher than at the start, causing the resistance at high current values to be overestimated by approximately 3 percent and producing a slight upward curvature in the otherwise linear I-V graph."
**Improvement:** Switch off the circuit between readings to allow the wire to cool to room temperature. Take each reading promptly after switching on, before significant heating occurs. Record the duration of current flow and note that heating effects are likely negligible below a specified current level.
---
## Experiment 2: Simple pendulum experiment (measuring g)
Related post: [Simple Pendulum Experiment for O-Level Physics](https://eclatinstitute.sg/blog/o-level-physics-experiments/Simple-Pendulum-Experiment-O-Level-Physics-Measuring-g)
### Error 2.1 - Amplitude of swing not kept small (systematic)
**Why it matters:** The simple pendulum formula $T = 2\pi\sqrt{L/g}$ is derived under the small-angle approximation (angles less than approximately 10 to 15 degrees). For larger amplitudes, the actual period is longer than the small-angle formula predicts. If the pendulum is released from a wide angle, the calculated value of $g$ using $T = 2\pi\sqrt{L/g}$ will be lower than the true value.
**ACE sentence:** "A source of systematic error was using a swing amplitude of approximately 20 degrees, which exceeds the small-angle approximation limit. The actual period at this amplitude was approximately 1.5 percent longer than the small-angle formula predicts, causing the calculated value of g to be underestimated by approximately 3 percent."
**Improvement:** Release the pendulum from an angle of no more than 10 degrees from the vertical. Measure the angle with a protractor clamped to the retort stand before each release, or mark a maximum swing line on the bench.
---
### Error 2.2 - Parallax when measuring the length of the pendulum (systematic)
**Why it matters:** The length of the pendulum should be measured from the pivot point to the centre of the bob. If the rule is not held vertically alongside the string, or if the eye is not level with the measuring point, the recorded length is shorter or longer than the true length. Because period depends on the square root of length, even a 1 cm error in a 50 cm pendulum produces a noticeable error in the calculated value of $g$.
**ACE sentence:** "A source of systematic error was parallax when measuring the pendulum length with the metre rule. The rule was held at a slight angle to the vertical string, causing the recorded length to underestimate the true length by approximately 0.5 cm. This caused the calculated value of g to be overestimated because the formula uses $L$ in the denominator under the square root."
**Improvement:** Use a set square to ensure the metre rule is held vertically. Measure the length in three positions and average. Record the length to the nearest 1 mm (0.1 cm) and check that the measurement is made from the centre of the pivot to the centre of mass of the bob.
---
### Error 2.3 - Timing only one complete oscillation (random)
**Why it matters:** One complete oscillation of a pendulum with a 1-second period takes only 2 seconds. A reaction time error of 0.2 seconds (a reasonable estimate for starting and stopping a stopwatch manually) represents 10 percent of that time, giving a large random error in $T$. This random error does not average out when only one oscillation is timed.
**ACE sentence:** "A source of random error was timing a single oscillation with a manual stopwatch. A reaction time uncertainty of approximately 0.2 seconds represents 8 percent of the 2.4-second period, introducing a large random error in $T$. When $T$ is squared to calculate $g$, this error is amplified, producing a scatter of approximately 16 percent in the calculated values of $g$."
**Improvement:** Time 20 complete oscillations and divide by 20 to find the period. The reaction time error is now 0.2/20 = 0.01 seconds per oscillation, which is less than 0.5 percent of the period. Repeat three times and average for further reduction of random error.
---
### Error 2.4 - Pendulum swinging in an ellipse rather than a plane (random)
**Why it matters:** If the pendulum is not released precisely in the vertical plane of the protractor, it begins to swing in a conical or elliptical path. The effective restoring force is lower than for pure planar oscillation, and the period is slightly different from the theoretical value. This error is difficult to quantify but introduces random scatter between trials.
**ACE sentence:** "A source of random error was the pendulum swinging in an elliptical path rather than a vertical plane. Releasing the bob with a sideways component caused some trials to have a slightly different period than a planar pendulum of the same length, producing random scatter in the $T^2$ vs $L$ graph that could not be reduced by repeating the same trial."
**Improvement:** Use a guide groove cut in a piece of card fixed at the base of the stand to constrain the swing to one plane. Alternatively, release the bob by burning a thread that holds it at rest without giving it any sideways momentum.
---
## Experiment 3: Optics and lens experiments
Related post: [O-Level Physics Optics Practical Handbook](https://eclatinstitute.sg/blog/o-level-physics-experiments/O-Level-Physics-Optics-Practical-Handbook), [Lens Focal Length Experiment](https://eclatinstitute.sg/blog/o-level-physics-experiments/Lens-Focal-Length-Experiment-O-Level-Physics)
### Error 3.1 - Parallax when reading object and image distances on the optical bench (systematic)
**Why it matters:** The object distance $u$ and image distance $v$ are measured from the centre of the lens. If the optical bench scale is read with the eye at an angle, the position of the lens holder or screen holder appears displaced relative to the scale. This introduces a systematic error of several millimetres in each position reading, which propagates into the calculated focal length via the lens formula $\\frac{1}{f} = \\frac{1}{u} + \\frac{1}{v}$.
**ACE sentence:** "A source of systematic error was parallax when reading the position of the lens and screen on the optical bench scale. With the eye positioned 5 cm to the left of the scale markings, the lens position appeared to read 24.8 cm when the true reading was 25.3 cm. This 0.5 cm error in $u$ produced an overestimate in the calculated focal length of approximately 0.3 cm."
**Improvement:** Position the eye directly above and level with each marker on the optical bench scale when taking readings. Use a set square to ensure the component holder is perpendicular to the bench scale before reading its position.
---
### Error 3.2 - Sharpest image is judged subjectively (random)
**Why it matters:** When finding the image position on the screen, the observer moves the screen until the image appears sharpest. "Sharpest" is a subjective judgment: different observers place the screen at slightly different positions, and the same observer may choose different positions on different trials. This introduces a random error in $v$ of several millimetres.
**ACE sentence:** "A source of random error was the subjective judgment of when the image on the screen was sharpest. Different observers moved the screen to positions varying by approximately 0.5 cm before agreeing on the sharpest image, introducing a random uncertainty in $v$ that propagated into the calculated focal length."
**Improvement:** Move the screen to the position of maximum image sharpness from both sides (approaching from shorter and longer distances) and take the midpoint as the best estimate of $v$. This "bracketing" method reduces the random error by approximately half compared with a single judgment.
---
### Error 3.3 - Lens not aligned perpendicular to the optical axis (systematic)
**Why it matters:** If the lens is tilted relative to the optical bench, the light passing through it is refracted asymmetrically. The image formed on the screen will be slightly distorted or shifted from the expected position. The measured $v$ will be shorter or longer than the true image distance, causing a systematic error in the calculated focal length.
**ACE sentence:** "A source of systematic error was that the lens was tilted approximately 3 degrees from perpendicular to the optical axis. This caused the image to form 0.4 cm closer to the lens than expected, underestimating $v$ and producing a calculated focal length that was approximately 0.2 cm shorter than the true value."
**Improvement:** Use a spirit level or set square to check that the lens face is vertical and perpendicular to the bench before each set of readings. Secure the lens holder with a clamp to prevent it rotating between measurements.
---
### Error 3.4 - Ambient light in the room reducing image contrast (random)
**Why it matters:** When using a ray box and lens in a bright room, stray light from windows and ceiling lights reaches the screen and reduces the contrast of the formed image. The edges of the image are less distinct, making it harder to judge the sharpest focus position. This increases the random error in $v$.
**ACE sentence:** "A source of random error was bright ambient light in the laboratory. Stray light from the windows reduced the contrast of the image formed on the screen, making it difficult to identify the sharpest focus position precisely. The random error in $v$ was approximately twice as large as in trials conducted with a darkened room."
**Improvement:** Conduct the experiment in a darkened section of the laboratory or draw the curtains. Position a dark card around the screen to block ambient light from the viewing surface.
---
## Experiment 4: Specific heat capacity (calorimetry)
Related post: [Specific Heat Capacity Experiment for O-Level Physics](https://eclatinstitute.sg/blog/o-level-physics-experiments/Specific-Heat-Capacity-Experiment-O-Level-Physics), [O-Level Physics Thermal Practical Handbook](https://eclatinstitute.sg/blog/o-level-physics-experiments/O-Level-Physics-Thermal-Practical-Handbook)
### Error 4.1 - Heat loss from the block or water to surroundings (systematic)
**Why it matters:** The specific heat capacity calculation uses $Q = mc\Delta T$, where $Q$ is the electrical energy supplied. In practice, some of the electrical energy is transferred to the surrounding air rather than being retained by the substance. The temperature rise $\Delta T$ is therefore lower than it would be if all of $Q$ went into the substance alone. This causes the calculated specific heat capacity to be larger (more energy apparently required per degree) than the true value.
**ACE sentence:** "A source of systematic error was heat loss from the aluminium block to the surrounding air. Approximately 8 percent of the electrical energy supplied was lost to the environment rather than heating the block, causing the measured temperature rise to be lower than expected. This made the calculated specific heat capacity larger than the true value (more energy appeared to be needed per degree Celsius)."
**Improvement:** Wrap the block in thermal insulation (such as cotton wool) and conduct the experiment in a draught-free location. Plot temperature against time and extrapolate the cooling portion of the curve back to the time the heater was switched off to estimate the true maximum temperature that would have been reached without heat loss.
---
### Error 4.2 - Thermal lag between the heater and the thermometer (systematic)
**Why it matters:** The thermometer is placed in a separate hole from the heater in the metal block. Heat must conduct through the block from the heater hole to the thermometer hole. The thermometer reading lags behind the actual temperature of the metal nearest the heater. If readings are taken too soon after switching on, the thermometer underestimates the temperature of the block near the heater, and the apparent $\Delta T$ is smaller than the true average temperature rise.
**ACE sentence:** "A source of systematic error was thermal lag between the heater and the thermometer. The thermometer was in a hole 3 cm from the heater and required approximately 2 minutes to reach the same temperature as the region nearest the heater. Readings taken before thermal equilibrium was established underestimated $\Delta T$, causing the calculated specific heat capacity to be overestimated."
**Improvement:** Wait until the rate of temperature increase has become constant (indicating thermal equilibrium within the block) before recording the temperature. Stir the oil in the thermometer hole gently to improve thermal contact.
---
### Error 4.3 - Inaccurate measurement of electrical energy supplied (systematic)
**Why it matters:** The energy supplied is calculated as $Q = V \times I \times t$, where $V$ and $I$ are the voltmeter and ammeter readings and $t$ is the time. If the current or voltage drifts during the experiment (which it can, as the resistance of the heater element changes with temperature), using a single reading of $V$ and $I$ taken at the start introduces a systematic error. The power is not constant, but it is treated as if it were.
**ACE sentence:** "A source of systematic error was that only the initial voltage and current readings were used to calculate the electrical energy supplied. As the heater resistance increased with temperature, the current decreased from 2.50 A to 2.35 A over the 10-minute experiment. Using the initial current overestimated the total energy supplied by approximately 6 percent, making the calculated specific heat capacity appear smaller than the true value."
**Improvement:** Record the voltmeter and ammeter readings every minute and use the average values of $V$ and $I$ to calculate the energy supplied. Alternatively, use a power meter that integrates voltage and current automatically.
---
### Error 4.4 - Imperfect thermal contact between the heater and the block
**Why it matters:** If the heater element does not fit snugly in its hole, or if no thermal conducting paste is used, there is an air gap between the heater surface and the block. Air is a poor thermal conductor. Some electrical energy heats the air gap rather than the block, and this energy is lost before it reaches the thermometer. The block gains less heat than calculated from $Q = VIt$.
**ACE sentence:** "A source of systematic error was imperfect thermal contact between the heater and the aluminium block. An air gap between the heater cartridge and the hole wall acted as a thermal insulator, diverting some energy away from the block and causing the temperature rise to be lower than expected. This caused the calculated specific heat capacity to be overestimated."
**Improvement:** Fill the gap between the heater and the block with a small amount of thermal conducting oil or paste before inserting the heater. Check that the heater fits tightly with no visible gap.
---
## Experiment 5: Resistance of a wire
Related post: [Resistance of Wire Experiment for O-Level Physics](https://eclatinstitute.sg/blog/o-level-physics-experiments/Resistance-of-Wire-Experiment-O-Level-Physics)
### Error 5.1 - Zero error on the micrometer screw gauge (systematic)
**Why it matters:** The cross-sectional area of the wire is calculated from the diameter measured with a micrometer. If the micrometer has a zero error (it does not read zero when the anvils are closed), every diameter measurement is offset by the zero error value. Since the cross-sectional area is proportional to the square of the radius, even a small zero error in the diameter produces a proportionally larger error in the calculated area and therefore in the calculated resistivity.
**ACE sentence:** "A source of systematic error was a zero error of +0.02 mm on the micrometer. Every diameter reading was 0.02 mm larger than the true diameter. Because cross-sectional area depends on $d^2$, the area was overestimated by approximately 3.5 percent for a wire of true diameter 0.55 mm, causing the calculated resistivity to be underestimated by the same proportion."
**Improvement:** Check the micrometer zero reading before use by closing the anvils gently and recording the reading. Subtract this zero error from every subsequent diameter measurement. Record the zero error explicitly in the table alongside the diameter readings.
---
### Error 5.2 - Crocodile clips not at the measured length markings (systematic)
**Why it matters:** In a resistance of wire experiment, the measured length is the distance between the two crocodile clips. If the clip jaws are wide, the effective contact point may be several millimetres from where the jaw edge aligns with the ruler. The true length of wire in the circuit differs from the ruler reading. This introduces a systematic error in every length measurement that shifts the $R$ vs $L$ graph but does not change its gradient unless the clips are adjusted inconsistently.
**ACE sentence:** "A source of systematic error was that the crocodile clip jaws were 8 mm wide and the contact point was at the centre of the jaw, not the edge that was aligned with the ruler. This caused the true length of wire in the circuit to be approximately 4 mm longer than the ruler reading for both clips combined, shifting the $R$ vs $L$ graph 8 mm along the length axis and causing the y-intercept to be slightly positive rather than passing through the origin."
**Improvement:** Use a pin or thin wire probe to mark the exact contact point of each clip on the wire before measuring the length with the ruler. Alternatively, use a knife-edge contact jig rather than crocodile clips for higher precision.
---
### Error 5.3 - Wire not uniform in diameter along its length (random)
**Why it matters:** A wire drawn through a die may have slight variations in diameter along its length due to manufacturing tolerances. If the micrometer measures only one position, the calculated cross-sectional area represents only that position. The average diameter (and therefore average resistance per unit length) may differ from the measured value, introducing a random error in the calculated resistivity.
**ACE sentence:** "A source of random error was variation in the diameter of the wire along its length. Three measurements of diameter at different positions along a 1 m sample ranged from 0.53 mm to 0.58 mm. Using only the single measurement taken at the midpoint gave a diameter of 0.55 mm, which was not representative of the average cross-section, introducing a random error of approximately 5 percent in the calculated resistivity."
**Improvement:** Measure the wire diameter at a minimum of three positions (each end and the middle) and two orientations at each position. Use the average of all six measurements as the diameter. Record the range to indicate how uniform the wire is.
---
### Error 5.4 - Wire not held taut during length measurement (random)
**Why it matters:** If the wire sags or has kinks between the crocodile clips, the true length of wire in the circuit is greater than the straight-line distance between the clips measured by the ruler. The resistance therefore appears higher than expected for that measured length, producing a graph that lies above the true $R$ vs $L$ line.
**ACE sentence:** "A source of random error was that the wire was not held taut between the clips during length measurement. A sag of approximately 5 mm in a 60 cm length added roughly 0.02 cm of extra wire, causing the resistance at that length to appear slightly higher than the true value for a straight wire. This produced scatter in the upper part of the $R$ vs $L$ graph."
**Improvement:** Stretch the wire along a metre rule taped to a flat bench. Use small pieces of adhesive tape at regular intervals to hold the wire flat and straight before attaching the crocodile clips.
---
### Error 5.5 - Gradient calculated point-to-point rather than from the best-fit line (PDO and ACE link)
**Why it matters:** This is both a PDO error and an ACE link. If the gradient of the $R$ vs $L$ graph is calculated by choosing two specific data points rather than by drawing the best-fit line through all the data, the result depends heavily on which two points are chosen. Two outlier points can give a gradient that differs significantly from the true gradient of the best-fit line, and therefore a significantly different calculated resistivity.
**ACE sentence:** "A limitation of the data analysis was calculating the gradient from two individual data points rather than the gradient of the best-fit line. The selected points included one that lay above the best-fit line and one below it by approximately equal amounts, so in this case the error was small; however, if either point had been an anomalous result, the calculated gradient and resistivity would have been substantially wrong."
**Improvement:** Always draw the best-fit straight line through all the data points by eye (or use a spreadsheet linear regression). Calculate the gradient from two points on the best-fit line itself, not from data points. Choose the two gradient points as far apart on the line as possible to reduce the percentage uncertainty in the rise and run measurements.
---
## Quick reference: error types by experiment
| Experiment | Error | Type | Direction |
|---|---|---|---|
| DC circuit | Zero error on ammeter | Systematic | Shifts all readings by fixed offset |
| DC circuit | Contact resistance at clips | Systematic | Overestimates total resistance |
| DC circuit | Heating of component | Systematic/gradual | Increases resistance at high current |
| Pendulum | Large amplitude | Systematic | Underestimates g |
| Pendulum | Timing single oscillation | Random | Large percentage error in T |
| Pendulum | Elliptical swing path | Random | Scatter in period measurements |
| Optics | Parallax on bench scale | Systematic | Shifts u and v readings |
| Optics | Subjective sharpest focus | Random | Scatter in v measurements |
| Specific heat | Heat loss to surroundings | Systematic | Underestimates temperature rise, overestimates c |
| Specific heat | Thermal lag in block | Systematic | Underestimates DeltaT early in experiment |
| Resistance of wire | Zero error on micrometer | Systematic | Overestimates diameter, underestimates resistivity |
| Resistance of wire | Wire not taut | Random | Overestimates resistance at each length |
| Resistance of wire | Gradient from two points only | PDO/ACE | Unreliable gradient if points are anomalous |
---
For the full Paper 3 procedure guides, see the [O-Level Physics Experiments hub](https://eclatinstitute.sg/blog/o-level-physics-experiments). For graphs, gradient lines, and PDO precision rules in detail, see the [O-Level Physics Graphing and Linearisation Clinic](https://eclatinstitute.sg/blog/o-level-physics-experiments/O-Level-Physics-Graphing-And-Linearisation-Clinic).
---
## Sources
1. [SEAB, GCE O-Level Physics Syllabus (6091) 2026](https://www.seab.gov.sg/files/O%20Lvl%20Syllabus%20Sch%20Cddts/2026/6091_y26_sy.pdf)
