Units, SI System, Significant Figures and Errors: JEE Main
Reviewed Last updated 2026-11-21. Syllabus topic: Units, SI system, significant figures and errors.
In short. A measurement is a number with a unit and an uncertainty. SI has seven base units and all others are derived. The least count limits the reading, and for a formula the maximum fractional error is the sum of each fractional error times its power.
- Seven SI base units: metre, kilogram, second, ampere, kelvin, mole and candela; radian and steradian are supplementary and dimensionless.
- Least count of vernier = 1 main scale division - 1 vernier division; of screw gauge = pitch/number of circular divisions; always correct for zero error.
- Sum or difference keeps the fewest decimal places; product or quotient keeps the fewest significant figures.
- Sum or difference: absolute errors add. Product or quotient: fractional errors add. A power n multiplies the fractional error by n.
Syllabus line: Units and Measurements Units of measurements, System of units, SI Units, fundamental and derived units, least count, significant figures, Errors in measurements. Dimensions of Physics quantities, dimensional analysis and its applications. (NTA JEE (Main) 2026 syllabus, Physics, Unit 1 (https://jeemain.nta.nic.in/document/syllabus-2026/))
Before you start
Prerequisites
- Limits and differentiability: derivative as a small change (lesson planned): Error propagation is differentiation of a formula. Read this first if dy = f-prime(x) dx is not clear.
- Binomial theorem: approximation for small x (lesson planned): The approximation (1 + x)ⁿ is about 1 + n x for small x underlies the rule that a power multiplies the fractional error.
- Experimental skills: using instruments (lesson planned): Vernier callipers, screw gauge and pendulum experiments use the least count and error rules met here.
Related topics
- Dimensional analysis and its applications
- Experimental Skills (lesson planned)
- Limits, Continuity and Differentiability (lesson planned)
- Kinematics (lesson planned)
The concept step by step
What a measurement is
To measure a physical quantity is to compare it with an agreed standard called a unit. The result is a number multiplied by a unit, such as 4.2 m. The number has no meaning without the unit, and it has no honesty without an uncertainty, because no instrument reads exactly. The International System of Units (SI) is the system JEE Main uses, so every numerical answer must be written in SI unless the question asks otherwise.
Base units, derived units and prefixes
A fundamental (base) quantity is defined independently; the seven SI base quantities are length (metre), mass (kilogram), time (second), electric current (ampere), temperature (kelvin), amount of substance (mole) and luminous intensity (candela). Every other unit is derived: newton is kg m s⁻2, joule is kg m² s⁻2, pascal is kg m⁻1 s⁻2. Prefixes scale units by powers of ten: milli 10⁻3, micro 10⁻6, nano 10⁻9, kilo 10³, mega 10⁶. Convert areas and volumes with squared and cubed factors.
Least count and instruments
The least count is the smallest value an instrument can read. For a metre scale divided in millimetres it is 1 mm. For a vernier, if N vernier divisions equal N-1 main scale divisions, the least count is one main scale division divided by N. For a screw gauge it is the pitch divided by the number of circular scale divisions. Reading = main scale reading + coinciding division x least count, and the zero error is subtracted with its sign: a positive zero error reduces the reading, a negative one increases it.
Significant figures and rounding
Significant figures are the digits known reliably plus the first uncertain digit. All non-zero digits count; zeros between non-zero digits count; leading zeros never count; trailing zeros count only after a decimal point. So 0.02030 has four and 5.00 x 10³ has three. In addition and subtraction the result has as many decimal places as the term with the fewest. In multiplication and division it has as many significant figures as the factor with the fewest. Round only at the end.
Types of error and how to quote them
A systematic error pushes every reading the same way (zero error, a wrongly calibrated scale). A random error scatters readings and is reduced by repeating and averaging. For n readings, the mean is the best value, the absolute error of each reading is its distance from the mean, and the mean absolute error is the average of those distances. The relative (fractional) error is the mean absolute error divided by the mean, and the percentage error is that times 100.
Combining errors
For Z = A + B or A - B the absolute errors add: dZ = dA + dB, even for a difference. For Z = A B or A/B the fractional errors add: dZ/Z = dA/A + dB/B. For Z = A^p B^q / C^r the maximum fractional error is p dA/A + q dB/B + r dC/C, with every term positive. A measured quantity that appears with a high power, such as a diameter squared, dominates the error budget and deserves the most careful measurement.
Worked examples
Each solution shows the marks every step would earn. The sanity check tests units and size of the answer.
Example 1 (easy)
A screw gauge has a pitch of 1.0 mm and 100 divisions on the circular scale. Find its least count in mm. With the studs touching it reads +5 divisions (positive zero error). A sphere gives linear scale 3 mm and circular scale 47 divisions. Find the corrected diameter in mm.
| Step | Marks |
|---|---|
| Least count = pitch/number of divisions = 1.0/100 = 0.01 mm. | 1 |
| Observed reading = 3 + 47 x 0.01 = 3.47 mm. | 1 |
| Zero error = +5 x 0.01 = +0.05 mm; corrected diameter = 3.47 - 0.05 = 3.42 mm. | 1 |
| Total: 3 marks | 3 |
Answer: 3.42 mm
Fast method: Reading = MSR + n x LC - zero error: 3 + 0.47 - 0.05 = 3.42 mm.
Sanity check: Units: mm. The correction is small (0.05 mm), comparable with a few least counts, as expected for a good instrument. A positive zero error always lowers the answer.
Example 2 (easy)
Find the area of a plate of length 4.2 m and breadth 1.35 m, written with the correct number of significant figures, in m².
| Step | Marks |
|---|---|
| Raw product = 4.2 x 1.35 = 5.67 m². | 1 |
| 4.2 has 2 significant figures and 1.35 has 3, so the product keeps 2. | 1 |
| Area = 5.7 m². | 1 |
| Total: 3 marks | 3 |
Answer: 5.7 m²
Fast method: Count figures: 2 and 3, keep 2. 5.67 rounds to 5.7.
Sanity check: Units: m x m = m². The rounded value 5.7 is within about 0.5 percent of 5.67, as expected from a 2-figure rule.
Example 3 (medium)
A car moves at 72 km/h. Express this in m/s, and express the acceleration due to gravity, 9.8 m/s², in km/h².
| Step | Marks |
|---|---|
| 72 km/h = 72 x 1000 m/3600 s = 20 m/s. | 1 |
| For 9.8 m/s², multiply by 3600² s²/h² and divide by 1000 m/km. | 1 |
| 9.8 x 12 960 000/1000 = 127 008 km/h². | 2 |
| Total: 4 marks | 4 |
Answer: 20 m/s and about 1.27 x 10⁵ km/h²
Fast method: Per second squared to per hour squared multiplies by 3600² = 1.296 x 10⁷; kilometre divides by 1000.
Sanity check: Units: the time factor is squared because time appears squared. A metre per second is 3.6 km/h, so 72 km/h is 20 m/s: 20 x 3.6 = 72, consistent.
Example 4 (medium)
Two lengths are measured as (25.0 +/- 0.2) cm and (10.0 +/- 0.2) cm. Find the difference with its absolute error and the percentage error in the difference.
| Step | Marks |
|---|---|
| Difference = 25.0 - 10.0 = 15.0 cm. | 1 |
| For a difference the absolute errors add: 0.2 + 0.2 = 0.4 cm. | 1 |
| Percentage error = 0.4/15.0 x 100 = 2.67 percent, so the result is (15.0 +/- 0.4) cm. | 2 |
| Total: 4 marks | 4 |
Answer: (15.0 +/- 0.4) cm, 2.67 percent
Fast method: Errors never cancel: add them, then divide by the new value 15.0, not by 25.0.
Sanity check: Units: cm. The percentage error of the difference (2.67) is larger than that of either reading (0.8 and 2.0), which is the signature of subtracting nearly equal quantities.
Example 5 (medium)
In a pendulum experiment the length is L = 64.0 cm with a least count of 0.1 cm, and 20 oscillations take 32.0 s with a least count of 0.2 s. Use g = 4 pi² L/T² and find the maximum percentage error in g.
| Step | Marks |
|---|---|
| Fractional error in L = 0.1/64.0 = 0.156 percent. | 1 |
| Fractional error in the time for 20 oscillations = 0.2/32.0 = 0.625 percent; T is the total time divided by 20, an exact number, so the percentage error in T is the same. | 1 |
| g depends on T², so the percentage error in g = 0.156 + 2 x 0.625 = 1.41 percent. | 2 |
| Total: 4 marks | 4 |
Answer: about 1.4 percent
Fast method: Percentage error of g = %L + 2 x %T = 0.16 + 1.25 = 1.4.
Sanity check: Units: percentages are unitless. The time term dominates because it is squared and has the larger fractional error. Measuring more oscillations would reduce it.
Example 6 (hard, multi-step)
The resistivity of a wire is rho = R pi d²/(4 L). Measured: R = (5.0 +/- 0.1) ohm, d = (0.50 +/- 0.01) mm and L = (1.00 +/- 0.01) m. Find rho and the maximum percentage error in it.
| Step | Marks |
|---|---|
| rho = 5.0 x 3.1416 x (0.50 x 10⁻3)²/(4 x 1.00) = 9.82 x 10⁻7 ohm m. | 2 |
| Fractional errors: R 0.1/5.0 = 2 percent; d squared gives 2 x 0.01/0.50 = 4 percent; L 0.01/1.00 = 1 percent. | 2 |
| Maximum percentage error = 2 + 4 + 1 = 7 percent, which is 0.7 x 10⁻7 ohm m, so rho = (9.8 +/- 0.7) x 10⁻7 ohm m. | 2 |
| Total: 6 marks | 6 |
Answer: rho = 9.8 x 10⁻7 ohm m with a 7 percent error
Fast method: Write the exponents: R¹, d², L¹. Add 1 x 2 + 2 x 2 + 1 x 1 = 7. The diameter contributes more than half the error.
Sanity check: Units: ohm x m²/m = ohm m. A resistivity of this order is that of an alloy wire, and a 7 percent error from a 2 percent d-reading shows why the diameter needs the best instrument.
Example 7 (hard)
Resistors R1 = (100 +/- 3) ohm and R2 = (200 +/- 4) ohm are connected in series, and separately in parallel. Compare the percentage errors of the two combinations.
| Step | Marks |
|---|---|
| Series: R = 300 ohm, absolute errors add: 3 + 4 = 7 ohm, so the percentage error = 7/300 = 2.33 percent. | 2 |
| Parallel: R = 100 x 200/300 = 66.67 ohm; dR = R² (dR1/R1² + dR2/R2²) = 4444.4 x (3/10⁴ + 4/(4 x 10⁴)) = 1.78 ohm. | 2 |
| Percentage error in parallel = 1.78/66.67 = 2.67 percent, larger than the series value 2.33 percent. | 1 |
| Total: 5 marks | 5 |
Answer: series 2.33 percent, parallel 2.67 percent
Fast method: Differentiate 1/R = 1/R1 + 1/R2: dR/R² = dR1/R1² + dR2/R2².
Sanity check: Units: ohm. Both percentage errors lie between the two individual percentage errors, 2 and 3 percent, and below their sum, as a weighted combination should.
Formulas and units
| Result | Formula | Valid when | Units or note | Worked example |
|---|---|---|---|---|
| SI base units | m, kg, s, A, K, mol, cd | not applicable | length, mass, time, current, temperature, amount, luminous intensity | none |
| Vernier least count | LC = 1 MSD - 1 VSD = MSD/N | vernier scale where N vernier divisions equal N-1 main-scale divisions | mm or cm | none |
| Screw gauge least count | LC = pitch/number of circular divisions | screw gauge; pitch in mm, divisions on the circular scale | mm | Example 1 |
| Corrected reading | MSR + n x LC - zero error (signed) | zero error taken with sign: positive if gauge reads above zero when closed | mm | Example 1 |
| Sum or difference | dZ = dA + dB | Z = A + B or A - B; maximum-error method, absolute errors add | same unit as Z | Example 4 |
| Product or quotient | dZ/Z = dA/A + dB/B | Z = A x B or A/B; maximum-error method, fractional errors add | dimensionless | Example 5 |
| Power law | Z = A^p B^q / C^r: dZ/Z = p dA/A + q dB/B + r dC/C | Z is a product of powers; maximum-error method, small errors, exponents taken as positive | dimensionless | Example 5 |
| Mean absolute error and percentage error | mean of |x_i - mean|; 100 x that/mean | repeated readings of the same quantity | unit of x; percent | none |
| Significant figures: sum and product | sum: fewest decimal places; product: fewest significant figures | not applicable | n/a (rule) | Example 2 |
Common mistakes and speed
- Subtracting errors when two measurements are subtracted. Why it tempts: The result is a difference, so it feels natural to subtract the uncertainties; but uncertainties describe the worst case and always add.
- Forgetting the power in the error formula for a squared or cubed quantity. Why it tempts: The measured quantity is a diameter or a time, and the error looks like one reading; but it enters as d² or T² and counts twice or more.
- Adding a positive zero error to the observed reading. Why it tempts: The word error suggests a correction in the plus direction; the correct rule is corrected = observed - zero error, with sign.
- Applying the product rule (figures) to a sum (decimal places), or the reverse. Why it tempts: Both rules are about rounding, and the wrong one often gives a close number, so the wrong option looks right.
- Dividing the mean absolute error by the wrong quantity or dividing by n twice. Why it tempts: Several averages appear in one problem (mean value, mean error); mixing them gives a value close to a distractor.
Time per question and fast method
Target: about 80 seconds per question.
For error problems write the exponent of each measured quantity, convert each error to a percentage, then add exponent x percentage with no minus signs. For a vernier or screw gauge write LC, observed reading and zero error on three lines and subtract the zero error with its sign. For significant figures decide first whether the operation is a sum (decimal places) or a product (figures).
Recall quiz
Pick an option for each question. Your answers stay on this page and are not sent anywhere.
Score: 0 of 6 correct. Pick an option for each question.
Practice
- A vernier callipers has a main scale graduated in millimetres, and 10 vernier divisions coincide with 9 main… Verified (T1)
- A screw gauge has a pitch of 0.5 mm and 50 divisions on its circular scale. With the studs touching, the 46th… Verified (T1)
- Three measurements are recorded as 0.02030 m, 5.00 x 10³ kg and 0.0040 s. The total number of significant… Verified (T1)
- A cyclist covers three stretches of length 12.11 m, 18.0 m and 1.012 m in a total time of 4.0 s. Reporting… Verified (T1)
- The density of a solid cylinder is found from rho = 4 m/(pi d² h). The mass is measured as 20.0 g with a… Verified (T3)
- Five readings of the time of a fall are 2.63 s, 2.56 s, 2.42 s, 2.71 s and 2.80 s. Using unrounded… Verified (T1)
- Two resistors R1 = (100 +/- 3) ohm and R2 = (200 +/- 4) ohm are connected in parallel. The maximum absolute… Verified (T1)
Previous-year questions
Previous-year questions are not imported yet. Previous-year questions are not imported yet; see the official papers on the NTA site. Official NTA papers page.
Going deeper and revision
SI system: base units, supplementary units and standards
Seven base units; the radian (plane angle) and steradian (solid angle) are dimensionless. Definitions now rest on fixed constants: the second on a caesium transition frequency, the metre on the speed of light, the kilogram on the Planck constant. A derived unit is a product of powers of base units, and the same base units appear when you check a formula. Remember the common derived units with their base forms: N = kg m s⁻2, J = N m, W = J/s, Pa = N/m², C = A s, V = J/C, ohm = V/A, F = C/V, T = N/(A m). Angle and solid angle are the exceptions that carry no dimension, so their unit can be added or dropped freely.
Least count, zero error and instrument formulas
Vernier: LC = value of one main scale division x (1 - (N-1)/N) when N vernier divisions equal N-1 main scale divisions; equivalently 1 MSD - 1 VSD. Screw gauge: LC = pitch/number of circular divisions. Zero error is read with the jaws or studs closed. For a vernier, a positive zero error equals (coinciding division) x LC. For a screw gauge, if the zero of the circular scale is above the reference line the error is negative, the reading n gives an error of (n - N) divisions. Correct reading = observed reading - zero error, always with sign. A common slip is to add a positive error. The least count of a stopwatch or ammeter is its smallest division, and it is the minimum absolute error for a single reading.
Significant figures: rules and edge cases
Powers of ten do not change the count: 4.50 x 10³ has three. A number such as 1200 with no decimal point is ambiguous, so exam questions write 1.2 x 10³ or 1.200 x 10³. Exact numbers (a count of 5 oscillations, the 2 in 2 pi r) have unlimited figures and do not limit the answer. For a sum take decimal places, for a product take figures. Rounding: look at the first dropped digit; keep the last digit if it is below 5 and raise it if it is above 5; the neutral case 5 is rounded to the even digit in most books, so avoid questions that depend on it. In JEE Main numerical answers are usually asked to a stated number of decimal places, so follow the instruction.
Absolute, relative and percentage error; mean of readings
Absolute error = |measured - true| or, in an experiment, |reading - mean|. Mean absolute error = average of the absolute errors, so it is never the sum and never negative. Relative error = mean absolute error/mean. Percentage error = relative error x 100. When the least count is the only uncertainty, the absolute error is the least count. The errors are quoted with the answer, such as (2.62 +/- 0.11) s, with the error to the same decimal place as the mean. The maximum error of a combined quantity assumes all errors add in the worst way, which is the convention in JEE Main unless a statistical rule is named.
Propagation formulas and where the traps are
Z = A + B or A - B: dZ = dA + dB. Z = A B or A/B: dZ/Z = dA/A + dB/B. Z = A^p: dZ/Z = p dA/A. Z = A^p B^q / C^r: dZ/Z = p dA/A + q dB/B + r dC/C. For a more general function Z = f(A, B) use dZ = |df/dA| dA + |df/dB| dB, as for resistors in parallel where dR/R² = dR1/R1² + dR2/R2². Traps: subtracting errors for a difference; leaving out the power on a squared or cubed term; mixing absolute and percentage errors; treating the percentage error of a difference as the sum of percentage errors; and for a difference of nearly equal quantities, the relative error becomes large even when each reading is precise.
What examiners test
Frequent items are: reading a vernier or screw gauge with zero error; counting significant figures and rounding a sum or a product; maximum percentage error in a derived quantity (density, resistivity, g from a pendulum, Young modulus, refractive index); mean absolute error from several readings; converting units with prefixes and squared factors; and statements on systematic and random errors. Speed matters: the usual answer is a sum of three or four small percentages, so write the exponents first and add. Keep every quantity in a consistent unit before computing the percentage, because a percentage is unitless but the absolute error must be in the same unit as the reading.
Memory hook
Suppose a kirana shopkeeper weighs rice on a spring scale whose smallest mark is 10 g, and the pointer rests a little past the 1.45 kg mark.
- The shopkeeper can only say the weight is 1.45 kg, give or take about 10 g: the smallest mark sets the smallest uncertainty, which is the least count.
- If the scale pointer already sat above zero before anything was placed on it, every weighing is too large by the same amount: that is a zero error and it must be subtracted.
- If three bags are weighed and then the total is announced, the total cannot be quoted more precisely than the coarsest bag reading, and the uncertainties add up bag by bag.
The smallest scale mark is the least count, a pointer off zero is a zero error to subtract, and errors of the parts add when a total is formed.
The scale cannot promise more than its smallest mark, so correct the zero, then add the errors.
Revision card
Try to answer each without looking above.
- Name the seven SI base units and the two dimensionless supplementary units.
- Write the least count of a vernier and of a screw gauge, and the sign rule for zero error.
- State the rounding rule for a sum and for a product.
- What is the error in A - B if A and B have errors dA and dB, and why?
- Write the maximum fractional error in Z = A^p B^q / C^r.
- How do you get the mean absolute error and the percentage error from five readings?
Sources and review
Reviewed Reviewed on 2026-11-20: every number recomputed by code and independently reviewed; textbook section references pending.
- Method: Every number in worked examples, recall checks and notes recomputed by code; claims recomputed independently by a separate verifier run using other numerical methods; textbook section references not yet attached; wording defects found were fixed afterwards and re-linted (code checks re-run) but not re-reviewed by a second run; syllabus scope compared with the NTA JEE Main 2026 syllabus text (by the verifier; see the report)
- Reviewer: independent verifier agent run (not the author)
- Date: 2026-11-20
- Defects found: 2; fixed: 2
- Report: content/reviews/learn8-uom-2026-11-21.md
Textbook references
- NCERT Physics Part 1, Class 11 (reprint 2026-27), Chapter 1: Units and Measurement, Summary SUMMARY (read 2026-11-22)
- NCERT Physics Part 1, Class 11 (reprint 2026-27), Chapter 1: Units and Measurement, 1.2 The International System of Units (read 2026-11-22)
- NCERT Physics Part 1, Class 11 (reprint 2026-27), Chapter 1: Units and Measurement, 1.3 Significant Figures (read 2026-11-22)
- NCERT Physics Part 1, Class 11 (reprint 2026-27), Chapter 1: Units and Measurement, 1.3.1 Rules for Arithmetic Operations with Significant Figures (read 2026-11-22)
- NCERT Physics Part 1, Class 11 (reprint 2026-27), Chapter 1: Units and Measurement, 1.3.2 Rounding off the Uncertain Digits (read 2026-11-22)
Sources used
- Syllabus: NTA JEE (Main) 2026 syllabus, Physics, Unit 1 (https://jeemain.nta.nic.in/document/syllabus-2026/), imported into syllabus/lines/jee-main/2026.json
- Standard results of school-level units, measurement and error analysis; no textbook section reference attached yet (to be sourced at verification). Content written fresh for this site.
How labels work: see how answers and lessons are verified. Found a mistake? Use Report an error on any question page.