Error Analysis in Physics Experiments

Why Error Analysis Matters

Every physical measurement has uncertainty. A vernier callipers reads to $0.01\text{ cm}$ , a stopwatch to $0.01\text{ s}$ , a digital multimeter to its last digit. When we combine measurements through formulas, those uncertainties propagate. Knowing how much our final answer might be off is just as important as the answer itself.

This is the first chapter of NCERT Class 11 Physics for a reason: nearly every numerical we solve later carries an implicit assumption that we know how to handle errors. JEE Main asks at least one error-propagation question every year. NEET asks about significant figures or instrument least counts.

We’ll cover types of errors, how to quote a measurement, how errors propagate through arithmetic, and the rules for significant figures.

Key Terms & Definitions

Accuracy: how close a measurement is to the true value. Affected by systematic errors.

Precision: how close repeated measurements are to each other. Affected by random errors.

Least count (LC): the smallest division a measuring instrument can read. Vernier callipers: $0.01\text{ cm}$ . Screw gauge: $0.001\text{ cm}$ . Stopwatch: $0.01\text{ s}$ (digital).

Absolute error: $|x_{\text{measured}} - x_{\text{true}}|$ . Has the same units as the measurement.

Relative error: $\Delta x / x$ — dimensionless. Often expressed as a percentage.

Mean absolute error: average of absolute deviations from the mean: $\Delta\bar{x} = \frac{1}{n}\sum |x_i - \bar{x}|$ .

Types of Errors

Systematic errors push measurements consistently in one direction. Causes: zero error in instrument, calibration drift, parallax in reading scales, environmental effects (temperature, humidity). These can be reduced by careful technique and calibration.

Random errors scatter measurements unpredictably around the true value. Causes: hand tremor, fluctuations in ambient conditions, noise in electronic instruments. These can be reduced by averaging many readings.

Gross errors are flat-out mistakes — misreading a scale, transposing digits. The cure is to repeat the measurement.

Random errors decrease as $1/\sqrt{n}$ when averaging $n$ readings. Systematic errors don’t average out — averaging more readings doesn’t help unless we change technique.

Error Propagation Rules

Rule 1: Sum or Difference

If $z = x + y$ or $z = x - y$ :

$\Delta z = \Delta x + \Delta y$

Always add absolute errors, even when subtracting. The error doesn’t get smaller just because we subtracted.

Rule 2: Product or Quotient

If $z = xy$ or $z = x/y$ :

$\frac{\Delta z}{z} = \frac{\Delta x}{x} + \frac{\Delta y}{y}$

Rule 3: Powers

If $z = x^a y^b / w^c$ :

$\frac{\Delta z}{z} = a\frac{\Delta x}{x} + b\frac{\Delta y}{y} + c\frac{\Delta w}{w}$

All exponents become positive coefficients of the relative errors. Higher powers contribute more to error.

Significant Figures

Rules for counting:

All non-zero digits are significant. (123 → 3 sig figs)
Zeros between non-zero digits are significant. (1003 → 4 sig figs)
Leading zeros are not significant. (0.0123 → 3 sig figs)
Trailing zeros after a decimal are significant. (1.230 → 4 sig figs)
Trailing zeros without a decimal are ambiguous (write in scientific notation).

Rules for arithmetic:

Addition/subtraction: result has the same number of decimal places as the input with the fewest decimal places.
Multiplication/division: result has the same number of significant figures as the input with the fewest sig figs.

Solved Examples

Example 1 (CBSE)

Find the relative error in $g$ if $g = 4\pi^2 L/T^2$ where $L = (20.0 \pm 0.1)\text{ cm}$ and $T = (0.90 \pm 0.01)\text{ s}$ .

$\frac{\Delta g}{g} = \frac{\Delta L}{L} + 2\frac{\Delta T}{T} = \frac{0.1}{20} + 2 \cdot \frac{0.01}{0.90} = 0.005 + 0.0222 = 0.0272$

So percentage error in $g$ is about $2.72\%$ . Notice $T$ contributes more because of the power 2.

Example 2 (JEE Main)

The resistance $R$ of a wire is found from $R = V/I$ where $V = (5.0 \pm 0.1)\text{ V}$ and $I = (2.0 \pm 0.1)\text{ A}$ . Find the percentage error in $R$ .

$\frac{\Delta R}{R} = \frac{0.1}{5.0} + \frac{0.1}{2.0} = 0.02 + 0.05 = 0.07 = 7\%$

Example 3 (CBSE)

Two lengths $L_1 = (10.0 \pm 0.1)\text{ cm}$ and $L_2 = (12.0 \pm 0.2)\text{ cm}$ . Find $L_1 + L_2$ and its uncertainty.

$L_1 + L_2 = 22.0\text{ cm}$ . Error: $0.1 + 0.2 = 0.3\text{ cm}$ . Result: $(22.0 \pm 0.3)\text{ cm}$ .

Example 4 (JEE Advanced)

The period of a pendulum is measured 5 times: $2.05, 2.03, 2.06, 2.04, 2.05\text{ s}$ . Find mean, mean absolute error, and percentage error.

Mean = $2.046\text{ s}$ . Deviations: $0.004, 0.016, 0.014, 0.006, 0.004$ . Mean deviation = $0.0088\text{ s}$ . Percentage error = $0.0088/2.046 \approx 0.43\%$ .

Exam-Specific Tips

JEE Main: Always asks for percentage errors. Powers in formulas are danger zones. If a quantity appears squared, its error contributes double.

NEET: Often asks “which instrument has the least error” or “find least count from given data.”

CBSE: Mean absolute error from a table of readings is a standard 3-marker.

JEE Advanced: Sometimes uses $\Delta z = \sqrt{(\Delta x)^2 + (\Delta y)^2}$ for independent random errors. NCERT uses simple addition; check the question.

Common Mistakes to Avoid

Subtracting errors when subtracting quantities. Errors add, even for subtraction. The relative error of a difference can be huge if the two quantities are nearly equal.
Forgetting the power rule’s coefficients. If $T$ enters as $T^2$ , the error is $2\Delta T/T$ , not $\Delta T/T$ .
Confusing significant figures with decimal places. $0.005$ has 1 sig fig but 3 decimal places.
Mixing absolute and relative errors. Sums use absolute; products use relative. Don’t mix.
Quoting too many digits. A measurement of $g$ as $9.81234567\text{ m/s}^2$ when the error is $\pm 0.5$ is silly. Quote to the digit where the error sits.

Practice Questions

Q1. Find percentage error in $V = (4/3)\pi r^3$ if $\Delta r/r = 1\%$ .

$\Delta V/V = 3 \cdot 1\% = 3\%$ .

Q2. $L = (5.0 \pm 0.1)\text{ cm}$ , $B = (3.0 \pm 0.05)\text{ cm}$ . Find error in area $A = LB$ .

$\Delta A/A = 0.1/5 + 0.05/3 = 0.02 + 0.0167 = 0.0367 \approx 3.67\%$ .

Q3. Vernier with main scale $1\text{ mm}$ divisions and 10 vernier divisions. What is the least count?

LC = $1\text{ mm}/10 = 0.1\text{ mm} = 0.01\text{ cm}$ .

Q4. A screw gauge has pitch $0.5\text{ mm}$ and 50 divisions on circular scale. Least count?

LC = $0.5/50 = 0.01\text{ mm} = 0.001\text{ cm}$ .

Q5. Mass $m = (5.0 \pm 0.1)\text{ kg}$ and velocity $v = (10.0 \pm 0.2)\text{ m/s}$ . Find percentage error in KE = $\frac{1}{2}mv^2$ .

$\Delta\text{KE}/\text{KE} = 0.1/5 + 2 \cdot 0.2/10 = 0.02 + 0.04 = 0.06 = 6\%$ .

Q6. Find sig figs in $0.00450$ .

3 sig figs (the leading zeros don’t count, but the trailing zero after the decimal does).

Q7. Why don’t we average to reduce systematic error?

Systematic errors push every reading in the same direction by roughly the same amount. Averaging just gives the same biased value. We need to fix the cause (calibrate the instrument, correct for zero error).

Q8. A length is $(2.30 \pm 0.05)\text{ m}$ . State the relative error.

$0.05/2.30 \approx 0.0217 \approx 2.17\%$ .

FAQs

Q: Should I use $\Delta z = \Delta x + \Delta y$ or $\Delta z = \sqrt{(\Delta x)^2 + (\Delta y)^2}$ for sums? NCERT and CBSE use the simple sum. JEE Advanced sometimes uses the quadrature sum (Pythagorean) for independent random errors. Default to NCERT unless the question specifies.

Q: How does least count relate to error? The least count gives the smallest reading, but the actual error is at least half the least count, sometimes the full least count.

Q: Why does a power propagate as a multiplied error? By calculus: $d(\ln z) = a\, d(\ln x)$ if $z = x^a$ , so $dz/z = a\, dx/x$ . The factor $a$ appears as a coefficient.

Q: What’s the difference between accuracy and precision? Accuracy is closeness to the true value (a darts player whose darts hit the bullseye on average). Precision is closeness of repeated readings to each other (a darts player whose darts cluster tightly, even if off-centre).

Q: How do I report a measurement? Quote the value, the uncertainty, and the unit: $L = (5.20 \pm 0.05)\text{ cm}$ . The uncertainty digit fixes the last meaningful digit of the value.

Q: Can errors be negative? The signed deviation $x - \bar{x}$ can be negative. The absolute error $|x - \bar{x}|$ is always non-negative. We usually quote $\pm$ both ways unless we know the bias direction.