scale factors

When you use curvilinear coordinates — polar, cylindrical, spherical, or any custom system — moving one coordinate by a small amount does not generally move you by that same amount of actual distance. Turning the angle theta by a step d-theta carries you a physical distance r d-theta, not d-theta. The scale factors are the conversion numbers that tell you, for each coordinate, how much real length corresponds to a unit change of that coordinate. They are the bridge between abstract coordinate steps and honest lengths, areas, and volumes.

For an orthogonal coordinate system with coordinates (u1, u2, u3) and position vector r, the scale factor h_i is the length of the partial derivative of r with respect to u_i — that is, h_i equals the magnitude of partial r over partial u_i. An infinitesimal displacement then has length-squared given by the line element ds^2 = (h1 du1)^2 + (h2 du2)^2 + (h3 du3)^2. The volume element is the product dV = h1 h2 h3 du1 du2 du3, which is exactly the Jacobian written in geometric form. For plane polar coordinates the scale factors are h_r = 1, h_theta = r. For spherical coordinates they are h_rho = 1, h_phi = rho, h_theta = rho sin(phi) — and their product rho times rho sin(phi) gives back the familiar rho^2 sin(phi).

Scale factors are the clean, unifying way to handle any orthogonal coordinate system at once. They give you the gradient (divide each component by its scale factor), the divergence, the curl, and the Laplacian in curvilinear coordinates through a single set of formulas, instead of re-deriving each operator from scratch. This is the machinery behind the spherical and cylindrical Laplacians that mathematical physics relies on for the heat, wave, and Laplace equations. The one limitation to keep in mind: the simple ds^2 = sum of (h_i du_i)^2 form and the product volume element only hold for orthogonal coordinates, where the coordinate directions meet at right angles; a general (non-orthogonal) system needs the full metric tensor with cross terms.