In review — free for everyone. While a book is in review you are one of its reviewers: read it, use it, and tell us what is wrong. When the reports settle, the Class 11 pass is ₹999 for the year and this book’s PDF is ₹299.
The number is only half the truth — what a measurement is
Physics starts by insisting on something that sounds too obvious to need saying. A number alone measures nothing.
Say a kirana shop weighs out dal and hands over a slip stamped only ‘500’. No unit is written anywhere on it. You cannot act on that slip. Five hundred grams is a small bag. Five hundred kilograms would not fit through the shop’s door. The number promised nothing until a unit joined it.
That is the whole rule this chapter opens with. A physical quantity is measured by comparing it against an internationally agreed standard. It is reported as a NUMBER together with a UNIT: $Q = n \times [u]$. Drop either half and the report says nothing.
Test the rule on something you already know by heart. A cricket pitch is 22 yards long, and it is also 20.12 metres long. Neither number is more correct than the other. The strip of ground between the two sets of stumps never moved. What changed is the unit. The number changed to match it. A larger unit divides the same length into fewer pieces. So it reports a smaller number every time. Measure a school corridor in metres and it reads 30. Switch to centimetres, and the very same corridor reads 3000. Written as a rule instead of an example: $n_1 u_1 = n_2 u_2$. That holds for one quantity read off in two different systems.
A number with no unit has measured nothing at all. Change the unit, and the number is required to change with it. If it does not, the two readings were never describing the same thing.
One language for the whole world — SI units and the scale of things
Physics needs one shared language before anything can be compared. SI (the Système International d’Unités) is that language. It has seven base units, fixed by international agreement rather than derived from anything more basic. The metre, the kilogram and the second are three of them. Four more cover electric current, temperature, amount of substance and luminous intensity. Every other unit in physics is built by multiplying or dividing these seven.
Here is the trap this section exists to correct. Mass and kilogram are not the same thing, and confusing them makes a sentence collapse. ‘The mass is 5’ is incomplete — five what? ‘The kilogram is heavy’ is meaningless. A unit has no weight of its own to be heavy. Mass is the physical quantity; kilogram is only the name given to its unit.
The kilogram used to be a metal cylinder locked in a vault in France. Every scale in every kirana shop traced its accuracy back to that one object. Since 2019, it no longer is. The kilogram is fixed instead by the Planck constant, a fact about nature. A properly equipped lab anywhere can reproduce it without visiting a vault at all. A unit is a reproducible agreement, not an object. An agreement can be revised without the quantity it measures changing by so much as a grain.
Ask what is being measured before you ask what it is measured in. A unit only ever answers the second question.
Only seven units are genuinely basic. Every other unit in physics is a DERIVED unit. It is a product or quotient of those seven. That is because the quantities this subject cares about are rarely independent of one another.
Speed is distance over time, so its unit is metres per second, $m/s$. A bus speedometer reports kilometres per hour instead. That is just a different-looking unit, built from the very same two base quantities. Push a one-kilogram mass so its speed changes by one metre per second every second. The force needed for that gets its own name: one NEWTON (N), equal to $\text{kg} m/s^2$. Do that over a distance. The energy spent is the JOULE (J), $\text{kg} m^2/s^2$. A household electricity bill quietly avoids this unit in favour of the kilowatt-hour. A kilowatt-hour is nothing but the same joule, counted in a more convenient size.
Two derived units are stranger still. The radian (a plane angle) and the steradian (a solid angle) are each a ratio of two lengths of the same kind. For the radian that is arc over radius; for the steradian, area over radius squared. The units above and below cancel completely, and what is left is DIMENSIONLESS.
The standard slip is treating a named unit as though it stood on its own. The unit should instead be checked against what it is built from. A newton that will not rebuild itself as $\text{kg} m/s^2$ is not a newton at all. It is a mistake wearing the right name.
A derived unit’s name is shorthand for an algebra sum. Memorise the name, and you have memorised nothing. Rebuild the algebra, and you have understood the unit.
Physics deals in sizes that no ordinary number line can hold at once. That range runs from what sits inside a nucleus to the width of the observable universe. The only honest way to keep that range in view is by POWERS OF TEN.
Write any quantity in scientific notation, $a \times 10^b$ with $1 \leq a < 10$. The exponent $b$ alone is the quantity’s ORDER OF MAGNITUDE. It is a single number saying how big something is, ignoring everything about its exact value. Two quantities differing by twenty orders of magnitude are not twenty times apart. They are $10^{20}$ times apart. That is a gap no list of digits could make you feel the way one exponent does.
This is also how a working physicist gets a usable answer fast, without measuring anything. Asked how many raindrops fall on a football field in one monsoon season, nobody counts drops. You estimate the field’s area, and multiply by a rough rainfall depth in millimetres. Then round every factor to the nearest power of ten on the way. The answer will not be exact, but its ORDER will be right. That is often the only thing worth knowing. It comes before deciding whether an exact calculation is even worth doing.
The standard misuse runs the other way: treating an order-of-magnitude estimate as though it were the precise value itself. All it ever promised was the right neighbourhood.
An order of magnitude is a promise about the neighbourhood a number lives in, not the number itself. Keep the two apart. A rough estimate then stays honest about how rough it is.
Significant figures — and what they promise
Every measurement carries some uncertainty. Reporting it honestly means writing down exactly the digits you can stand behind. That means every digit you are sure of, plus exactly one more that is a genuine, honest estimate. Together these are the SIGNIFICANT FIGURES.
Read a measuring cylinder in the school chemistry lab. You do not stop at the nearest marked line. You read down to a fraction of the smallest division. You then write that last digit as your best guess — uncertain, but not invented. That final digit belongs in the number as much as any other.
Counting significant figures has nothing to do with where the decimal point happens to sit. Every non-zero digit counts. A zero sitting between two non-zero digits counts too: $1.0008$ has five. A leading zero never counts. It only shows where the decimal point is, nothing about how precisely anything was measured. $0.0006032$ has four significant figures, not seven. A trailing zero counts only when a decimal point is actually written. $4.700$ has four. Plain $4700$ is ambiguous — it could mean anywhere from two to four.
The trap is treating more digits as more precision. Or it is crediting a leading zero with meaning it never carried. Writing more digits than an instrument actually resolved is not extra precision. It is a false claim about how well you know the quantity. That is as much an error as writing the wrong digit outright.
Significant figures count information, not symbols on a page. Scientific notation removes every argument about where the count starts. That is why it is the safer habit.
A result calculated from measurements can never be more precise than the measurements that produced it. That rule wears two different faces, depending on the operation.
Multiply or divide, and the answer keeps as many significant figures as the LEAST precise input carries. A tailor measures cloth to three significant figures, at a rate known only to two. The bill cannot be reported to five figures. The answer is only as good as its weakest input, no better.
Add or subtract, and a different rule applies. The answer keeps as many DECIMAL PLACES as the least precise input, not the same count of significant figures. These are not the same rule wearing two names. Swapping them is the standard mistake.
When a dropped digit is exactly $5$, the convention is ROUND-HALF-TO-EVEN: round so the preceding digit becomes even. So $2.745$ rounds to $2.74$, and $2.735$ also rounds to $2.74$. This is not arbitrary. Always rounding a $5$ upward would quietly push a long calculation’s final answer high. Rounding to even cancels that bias out over many roundings instead of feeding it in one direction.
One more distinction rescues you from over-caution. A pure count or a defined factor — the $2$ in $C = 2 \pi r$ — is EXACT. It carries unlimited significant figures and never drags a result’s precision down. That is because nothing about it was ever measured in the first place.
In a calculation with several steps, round only at the very end, carrying one spare digit through the middle. Rounding after every single step lets small errors compound instead of cancel.
Multiplication counts figures; addition counts decimal places. Confuse the two, and the arithmetic still runs — the physics comes out wrong.
Accuracy, precision, and error
ACCURACY is how close a reading sits to the true value. PRECISION is how tightly repeated readings cluster together, regardless of where that cluster sits. The two sound like they should move together. They do not.
Take a kirana shop’s weighing scale that has drifted forty grams heavy and has never been recalibrated. Weigh the same packet on it five times, and you get five nearly identical readings. The scale is wonderfully precise, tightly repeatable. Yet every single one of them is wrong by the same forty grams. Precise, and inaccurate, at once.
Now take a cheaper scale that has not drifted at all, but wobbles. Five weighings on it scatter loosely above and below the true weight. Average enough readings and you land close to the truth. Yet no single reading on its own was trustworthy. Accurate on average, but imprecise.
The trap here is assuming a more precise instrument must automatically be a more accurate one. It is not. Precision only tells you the instrument agrees with itself; accuracy asks whether it agrees with the world. A fixed bias — SYSTEMATIC error — is invisible to a precision check. It stays exactly where it is, no matter how many times a reading is repeated.
Repeatable and correct are two different questions. A scale can pass one and fail the other. Only a check against a known, trusted standard answers the second.
Measurement error is not one thing. It is two, and they behave completely differently.
A fever thermometer that always reads half a degree high, no matter what, carries SYSTEMATIC error. That is a fixed, one-directional bias. Taking a temperature five times in a row will not fix it. Every reading is high by the same half-degree. Only finding the fault, or checking against a properly calibrated thermometer, corrects it. Averaging more readings does nothing for this kind. There is no scatter to average out. The whole cluster is shifted, together.
RANDOM error looks different. It is the small variation from reading a scale at a slightly different angle each time. Or it is the last digit that has to be estimated. This kind genuinely does shrink as more readings are averaged. The scatter tends to cancel itself out over enough tries.
The gap between a measured value and the true one is reported three ways: absolute error $|Delta a|$, relative error $Delta a / a$, and percentage error. Percentage error is the same relative error, written as a percentage instead of a fraction. The relative error is the one that carries forward into every later calculation. It depends on the size of the number measured, not only on the instrument. A kirana scale accurate to $\pm 10$ grams gives a full ten percent relative error on a small 100-gram packet. It gives only one percent on a full one-kilogram packet. The same instrument gives wildly different honesty about the result, depending on what is being weighed.
The standard confusion is calling any error a “mistake” that could simply have been avoided. Or it is assuming averaging cleans up every kind of error. It cleans up only the random kind. A systematic bias averaged a thousand times is still exactly as wrong as it was the first time.
Ask which kind of error you are looking at before reaching for the fix. Averaging cures scatter; it never cures a bias.
When you combine two measured quantities, their uncertainties combine too. They combine by two different rules, depending on how the quantities themselves are combined.
Add or subtract two measured quantities ($Z = A \pm B$). The ABSOLUTE errors simply add: $Delta Z = Delta A + Delta B$. There is no way around this. Every measurement error nudges the sum a little. The two nudges pile up regardless of whether the quantities were added or subtracted.
Multiply or divide instead ($Z = A B$ or $Z = A/B$). It is the RELATIVE errors that add instead: $Delta Z / Z = Delta A / A + Delta B / B$. This is not the same rule copied over. Adding relative errors for a sum, or absolute errors for a product, is the single most common slip in this topic.
Raise a quantity to a power and the relative error scales with the exponent. For $Z = A^m B^n / C^p$: $Delta Z / Z = |m| Delta A/A + |n| Delta B/B + |p| Delta C/C$. This is where careless measurement gets expensive. Measure the radius of a rubber ball with a vernier calliper. The goal is to compute its volume, $V = (4/3) \pi r^3$. Here $r$ enters as $r^3$ — its relative error is not carried once into the volume. It is carried THREE times. A small, easy-to-shrug-off error in reading the radius comes out of the volume tripled. That happens simply because of the exponent sitting on it.
A quantity’s own relative error is only the starting cost. The exponent it enters a formula with decides the real cost. Dropping that exponent is the slip that makes an otherwise careful calculation quietly wrong.
Dimensions of physical quantities
Strip a quantity of its number and its particular unit. What is left is its DIMENSIONS — the KIND of thing it is. Dimensions are written as powers of the base quantities, inside square brackets. $[M]$ stands for mass, $[L]$ for length, and $[T]$ for time. The same scheme continues through current, temperature, amount of substance and luminous intensity. Most of mechanics lives inside just the first three.
Take the rainfall figure a weather report reads out every monsoon morning, in millimetres. It looks like its own special kind of quantity. But strip away the unit, and what remains is plain LENGTH, $[L]$. That is the same dimension as your height or the width of a field. It is just reported in a size convenient for a thin layer of water.
The DIMENSIONAL FORMULA states these powers explicitly. Force works out to $[M][L T^{-2}] = [M L T^{-2}]$; volume to $[M^0 L^3 T^0]$. Dimensions ignore magnitude and direction completely. Initial velocity, final velocity and plain speed are all $[L T^{-1}]$, however different the three situations feel.
The trap is treating a dimension and a unit as the same thing. $[L]$ is the dimension length. The metre, the foot and the light-year are all units that report it. Choosing between them changes nothing about what is being measured. Angle, strain and refractive index carry no dimension at all. That is not because they lack one by accident, but because each is a ratio of two like quantities. The dimensions cancel completely, rather than merely going unnoticed.
A dimension answers ‘what kind of thing is this’; a unit answers ‘in what size’. Confuse the two questions, and an equation that looks fine can hide a mistake. No arithmetic will catch it.
Dimensional analysis and its limits
One rule governs every equation in physics. Only quantities that share the same dimensions can be added, subtracted, or set equal to each other. This is the PRINCIPLE OF HOMOGENEITY, and it drives three genuinely different jobs.
The first is CHECKING. Every term of a correct equation must carry the same dimensional formula. If they disagree, the equation is wrong — full stop, no further arithmetic needed. But a passed check does not prove the equation right. It is blind to plain numbers. It cannot tell $(1/2) m v^2$ apart from $(3/16) m v^2$, since both carry identical dimensions. And it cannot see inside a sine, a logarithm, or an exponential at all. Whatever sits inside one of those must itself be dimensionless. But the check has nothing further to say about it.
The second is DERIVING a relation up to a constant the method can never supply. Assume a simple pendulum’s time period depends on its length $l$ and on $g$, in some product form. Then match the dimensions on both sides. Out comes $T = k \sqrt{l/g}$. Dimensions alone rule out any dependence on the bob’s mass at all. What dimensional analysis cannot hand you is $k$ itself. Only the actual pendulum experiment in a school lab fixes that at $2 \pi$.
The third is CONVERTING a unit between systems. A quantity’s dimensions never change, even when its unit does. $1 N = 10^5$ dyne, found purely from force’s own dimensions. No experiment is required at all.
The one sentence to carry out of this section is this: a failed check proves an equation wrong; a passed one never proves it right.
Homogeneity tells you an equation is not obviously broken. It has never once told anyone an equation is true.
Dimensional analysis is powerful precisely because it is narrow. Knowing exactly where that narrowness lies is what keeps the method from being misused as proof.
It cannot find a dimensionless constant. Not the $1/2$ sitting in kinetic energy, not the $2 \pi$ in a pendulum’s time period. Both numbers are real and necessary. Dimensional analysis has no way to reach either of them.
It cannot handle a relation that is a SUM of several terms. Nor can it handle one hidden inside a trig, log, or exponential function. The method only ever compares whole terms against each other, never their internal structure.
And it cannot tell apart two quantities that happen to share the same dimensions. Turning a heavy almirah door open by pushing on its handle is one thing. Lifting a sack of rice up onto a shelf is another. Physically the two are nothing alike: one twists, the other lifts straight up. Yet TORQUE and WORK both work out to the identical $[M L^2 T^{-2}]$. Dimensions cannot see the difference a physicist cares about most: what the quantity actually does.
The standard overreach is treating a dimensionally consistent equation as thereby proven. Or it is expecting the method to hand over a number it was never built to supply.
A shared dimensional formula means two quantities are built the same way, never that they are the same thing. That distinction is exactly what dimensional analysis was never asked to make.