Dimensional Analysis and Its Applications
On this page
Direct answer
Any equation describing nature must satisfy the principle of homogeneity: both sides carry identical dimensions, and only quantities of the same dimension can be added or equated. Writing [force] = MLT^-2, [energy] = ML^2T^-2, [Planck's constant] = ML^2T^-1 and [G] = M^-1L^3T^-2 lets you check a derived formula, fix the dimensions of an unknown constant, and convert units between systems (1 J = 10^7 erg). The method cannot, however, determine dimensionless numerical factors such as the 2 pi in T = 2 pi sqrt(l/g), and it cannot decide between sin(theta) and theta, because the arguments of trigonometric, logarithmic and exponential functions are dimensionless regardless of which function appears.
What you must remember
- Dimensions that recur in JEE Main: work, energy, torque ML^2T^-2; power ML^2T^-3; pressure, stress, Young's modulus, energy density ML^-1T^-2; frequency and angular frequency both T^-1; Planck's constant and angular momentum ML^2T^-1; surface tension MT^-2; coefficient of viscosity ML^-1T^-1; gas constant ML^2T^-2 K^-1 mol^-1.
- Electrical dimensions: from Coulomb's law F = q^2/(4 pi epsilon_0 r^2), [epsilon_0] = M^-1L^-3T^4A^2, and from B = mu_0 I/2 pi r, [mu_0] = MLT^-2A^-2; the combination 1/sqrt(mu_0 epsilon_0) has the dimensions of speed.
- Checking equations: dimensional homogeneity is necessary but not sufficient — s = ut + at^2 is dimensionally perfect yet wrong (the 1/2 is missing).
- Deriving relations up to a constant: for a liquid drop of density rho, radius r, oscillating under surface tension S, balancing dimensions gives T = k sqrt(rho r^3/S), a result JEE Advanced has tested in all but name.
- Unit conversion: n(2) = n(1) (M(1)/M(2))^a (L(1)/L(2))^b (T(1)/T(2))^c, so 1 newton = 10^5 dyne and 1 joule = 10^7 erg because 1 m = 100 cm enters squared.
- Pattern note: JEE Main almost every year asks the dimensions of one constant or of a combination such as h/e or epsilon_0 mu_0; Advanced prefers asking you to construct a formula from listed variables.
How a dimensional prediction is actually built
Take the oscillating drop question seriously, because it shows the full method. Guess T = k rho^a r^b S^c, where k is dimensionless. Substitute dimensions: [T] = (ML^-3)^a (L)^b (MT^-2)^c. Equating powers of M gives a + c = 0; equating powers of T gives −2c = 1, so c = −1/2 and a = +1/2; equating powers of L gives −3a + b = 0, so b = 3/2. The prediction is T = k sqrt(rho r^3/S), and no amount of dimensional skill can reveal that k is close to 1 for the fundamental mode.
Unit conversion deserves equal respect in numerical papers. To express the Stefan constant 5.67 × 10^-8 W m^-2 K^-4 — that is M T^-3 K^-4 — in CGS, only the M and T powers change: multiplying by 10^3 (gram-to-kilogram) and 10^-6 (second powers inverted) yields 5.67 × 10^-5 erg cm^-2 s^-1 K^-4.
Where students slip
The recurring error is treating homogeneity as proof of correctness; an equation can be dimensionally immaculate and physically wrong, so a "check" can only reject, never certify. The second slip is electrical: forgetting the ampere, or writing epsilon_0 and mu_0 as pure numbers, which destroys the check of Maxwell-type combinations such as E = cB or RC having the dimension of time. A subtler trap is frequency versus angular frequency — both T^-1 — so dimensional analysis can never flag a missing 2 pi in an oscillation formula, a fact Advanced has used to build "which formula can be rejected on dimensional grounds alone" questions. Keep the method in its lane: it filters, it does not confirm.
Frequently asked questions
Can dimensional analysis give the numerical value of a dimensionless constant?
No — it is blind to all pure numbers, which is why the 2 pi in T = 2 pi sqrt(l/g) and the 1/2 in s = ut + (1/2)at^2 must come from calculus or experiment.
What are the dimensions of Planck's constant?
ML^2T^-1, the same as angular momentum, which is why h is the natural quantum unit of angular momentum; h/e carries ML^2T^-2A^-1 and appears in the photoelectric equation.
How do you find the dimensions of epsilon_0?
Rearrange Coulomb's law as F = q^2/(4 pi epsilon_0 r^2) and substitute [q] = AT, [r] = L, [F] = MLT^-2 to get [epsilon_0] = M^-1L^-3T^4A^2.
How is a derived unit converted between MKS and CGS?
Write n(2) = n(1)(M(1)/M(2))^a(L(1)/L(2))^b(T(1)/T(2))^c using the size ratios of the base units; for example 1 J = 10^7 erg because 1 kg m^2 equals 10^3 gram × (10^2 cm)^2.
Why can't dimensional analysis distinguish sin(omega t) from omega t?
Both omega t and sin(omega t) are dimensionless, so any test that only compares M, L and T powers sees identical expressions — the method cannot discriminate between different functions of a dimensionless argument.