Dimensions can do more than check whether a proposed equation is consistent. When the relevant physical variables are known, they can sometimes constrain the form of the relation among them.

Suppose a quantity $q$ depends on variables $x_1,x_2,\ldots$ and we expect a power-law relation of the form

$$q=Cx_1^{a_1}x_2^{a_2}\cdots,$$

where $C$ is dimensionless. Dimensional consistency requires

$$[q]=[x_1]^{a_1}[x_2]^{a_2}\cdots.$$

Matching the powers of the fundamental dimensions produces algebraic equations for the unknown exponents $a_i$.

Example: fall time from a height

Suppose the time $t$ for an object to fall from rest through a height $h$ depends only on $h$ and the gravitational acceleration $g$. Assume

$$t=C h^a g^b.$$

The relevant dimensions are

$$[t]=T,$$

$$[h]=L,$$

$$[g]=LT^{-2}.$$

Therefore

$$T=L^a(LT^{-2})^b =L^{a+b}T^{-2b}.$$

Matching powers of $L$ and $T$ gives

$$a+b=0,$$

$$-2b=1.$$

Thus

$$b=-\frac12,$$

and

$$a=\frac12.$$

So dimensional reasoning implies

$$\boxed{t=C\sqrt{\frac{h}{g}}}.$$

The dependence on $h$ and $g$ has been determined without solving the equations of motion.

Dimensional analysis cannot determine $C$. An independent dynamical calculation or experiment is needed for that. For ideal free fall from rest, such a calculation gives

$$C=\sqrt2,$$

and therefore

$$t=\sqrt{\frac{2h}{g}}.$$

The factor $\sqrt2$ was invisible to the dimensional argument because it is dimensionless.

What dimensional inference can reveal

A successful dimensional argument can often show how a result scales. From

$$t\propto h^{1/2}g^{-1/2},$$

we immediately know that quadrupling the height doubles the fall time, while quadrupling $g$ halves it.

This makes dimensional inference useful even when an exact calculation is unavailable: it can identify plausible parameter dependence, expose impossible formulas, and guide experiments or more detailed modeling.

What it cannot determine

Dimensional reasoning has important limits.

It cannot normally determine pure numerical constants such as $2$, $\pi$, or $1/2$. It also cannot distinguish different dimensionless functional dependences. If a problem contains a dimensionless variable $\theta$, dimensions alone cannot tell whether a result contains

$$\sin\theta,\qquad \cos\theta,\qquad \theta^2,$$

or some other dimensionless function.

Most importantly, the answer depends on choosing the correct relevant variables. If an important variable has been omitted, a dimensionally consistent result can still be physically incomplete.

Dimensional inference therefore constrains a physical model; it does not replace the physics needed to decide which quantities matter or to determine dimensionless constants and functions.