Dimensional analysis in physics

In brief

A student proposes that the time a pendulum takes to swing depends on its length L and the strength of gravity g through the formula T equals L plus g. Before touching a stopwatch, this formula can already be rejected, because a length and an acceleration are not the same kind of thing and cannot be added, any more than you can add three apples to four hours and get a sensible answer. This entry is about that kind of check. Every physical quantity carries a dimension, built from a small set of base kinds such as mass, length, and time, and a true physical relationship must have matching dimensions on both sides. This single requirement, applied before any data is collected, can reject wrong formulas, fix missing constants, and sometimes reconstruct the shape of a correct relationship from almost nothing.

The full treatment

First look: apples, hours, and a pendulum

The reason L plus g fails has nothing to do with gravity or pendulums specifically. It is the same reason "3 apples plus 4 hours" has no answer: addition only makes sense between quantities of the same kind. Length is measured in meters, acceleration in meters per second squared, and no rescaling of units turns one into the other, because they answer fundamentally different questions, how far, versus how quickly speed changes. This is the seed of dimensional analysis: track not just the numerical value of a quantity, but the kind of quantity it is, and require that kind to match wherever quantities are added, subtracted, or set equal to one another.

Building the idea: dimensions as a bookkeeping system

Assign every physical quantity a dimension, written using a small set of base symbols: mass M, length L, time T, and a few others such as electric current, temperature, and amount of substance for quantities that need them. A speed, distance divided by time, has dimension L divided by T. An acceleration, speed divided by time, has dimension L divided by T squared. A force, by Newton's relation between force, mass, and acceleration, has dimension M times L divided by T squared. These dimension symbols are not units, a speed could be measured in meters per second or miles per hour, but its dimension, L divided by T, is the same regardless of which units are chosen. Two quantities can only be added, subtracted, or declared equal if their dimensions match exactly. Multiplication and division are always allowed, and combine the dimensions accordingly, since a new kind of quantity, such as energy from force times distance, is exactly what multiplication is meant to produce.

The formal check: dimensional homogeneity

This gives a precise, checkable rule called dimensional homogeneity: in any true physical equation, every additive term must carry the same dimension, and this dimension must match on both sides of an equals sign. Written formally, if an equation claims A equals B plus C, then the dimension of A must equal the dimension of B, which must equal the dimension of C. Applying this to the pendulum, the correct relationship, derivable from the physics of restoring forces, is T equals two pi times the square root of L divided by g. Check the dimensions: L divided by g has dimension L divided by (L divided by T squared), which equals T squared, and its square root has dimension T, matching the dimension of T on the left. The numerical constant two pi is dimensionless, a pure number, and dimensional analysis can never supply such constants, only the surrounding structure of the formula.

What dimensional analysis can and cannot recover

Because dimensions constrain the shape of a relationship so tightly, sometimes they nearly determine it. If you suspect the pendulum period T depends only on length L, gravitational acceleration g, and mass m, dimensional reasoning shows that the only combination of these three quantities with dimension of time, up to a dimensionless multiplier, is the square root of L divided by g, and that mass cannot appear at all, since no combination of M with L and time to the power of negative two can cancel to leave a pure time. This is a real physical prediction, that a heavier pendulum bob does not swing faster or slower, extracted before any experiment, purely from tracking dimensions. What dimensional analysis cannot do is supply the dimensionless constant, here two pi, or guarantee that no other quantity secretly matters; it only constrains relationships among the quantities you have already decided to include.

Lineage

The seed of tracking dimensions as distinct from numerical value appears wherever people compare mixed measurements, but its systematic use in physics grew alongside the development of coherent unit systems in the nineteenth century, when physicists such as James Clerk Maxwell and Lord Rayleigh used dimensional reasoning to check and simplify equations in electromagnetism and fluid mechanics. Percy Bridgman gave the method its first full formal treatment in his 1922 monograph "Dimensional Analysis," establishing the general theorem, now called the Buckingham Pi theorem after Edgar Buckingham's 1914 statement of it, that any dimensionally consistent physical relationship among n quantities built from k independent dimensions can be rewritten in terms of n minus k dimensionless groups. This turned dimensional checking from a spot check into a constructive method for building candidate physical laws.

The strongest case for it

Dimensional analysis costs nothing and catches a great deal. It requires no experiment, no advanced mathematics beyond algebra, and no assumption about the detailed mechanism at work, yet it reliably rejects formulas that are simply impossible and often narrows the space of plausible relationships to a single shape up to a constant. It is used routinely as a sanity check on derivations in every branch of physics, from mechanics to electromagnetism to fluid dynamics, and it generalizes across scale: the same method that catches a wrong pendulum formula also guides estimates of drag on aircraft, the scaling of stars, and the design of scale models in engineering, because the Buckingham Pi theorem shows how a full size system and a scaled model can be related by matching dimensionless groups rather than absolute sizes.

The strongest case against it

Dimensional analysis constrains the shape of a relationship but never proves it correct, and never supplies dimensionless constants or their values, which must come from a genuine physical derivation or an experiment. It also cannot tell you which quantities belong in the relationship in the first place, if you omit a quantity that actually matters, the method will confidently produce a wrong or incomplete answer built only from the quantities you included, and dimensional consistency alone will not reveal the omission. A common misconception is treating a dimensionally consistent formula as thereby physically correct, when dimensional consistency is a necessary condition, not a sufficient one, many wrong formulas are perfectly consistent in their dimensions. Another misconception is applying the method to quantities that are secretly the same dimension by coincidence, such as torque and energy, both mass times length squared divided by time squared, without checking that the underlying physical meaning actually matches.

Where it stands now

Dimensional analysis is a settled, universally used method across all of physics and engineering, formalized by the Buckingham Pi theorem and taught as a routine first check on any proposed physical relationship. Its role is uncontested as a necessary consistency requirement; the only live judgment involved is the ordinary scientific one of deciding, case by case, which quantities are relevant enough to include in the analysis.

Test yourself

A colleague proposes that the speed v of a wave on a stretched string depends on the string's tension F (dimension mass times length divided by time squared), its mass per unit length mu (dimension mass divided by length), and proposes the formula v equals F times mu. Use dimensional analysis to check this formula, show whether it can be correct, and if not, construct the unique combination of F and mu, up to a dimensionless constant, that does have the correct dimension for a speed. Then explain what dimensional analysis alone cannot tell you about your corrected formula, and what further step would be needed to pin that down.