Calculus On Manifolds

Everyone learns the derivative as a slope — the steepness of a tangent line, rise over run, the limit of a fraction. It's a fine picture for a function of one variable, and it falls apart the moment there are two. What is the slope of a surface? In which direction? The partial derivatives give you answers along the axes, but a function can have all its partials and still fail to be differentiable in any meaningful sense, and the standard textbooks papered over this with directional derivatives and warnings.

Spivak's move, which he inherits from the modern tradition and makes elementary, is to throw out the fraction entirely. The derivative of a function at a point is not a number and not a slope. It is a linear map — the best linear approximation to the function near that point. In one dimension that linear map happens to be multiplication by a single number, which is why the slope story ever worked; the slope was the derivative wearing its one-dimensional costume. Strip the costume off and the derivative is revealed as a matrix, a whole linear transformation, and suddenly the several-variable case is not harder than the one-variable case. It's the same case.

Integration is where the classical curriculum sprawls worst. Line integrals, surface integrals, flux, circulation, work — each arrives with its own notation and its own physical story, and the student is left holding a pocketful of formulas that plainly rhyme without being told why. Spivak's answer is the differential form, the second big piece of machinery the book builds, and the one that most repays the effort of learning it.

A differential form is, roughly, the kind of object you are allowed to integrate. That sounds circular until you see what it buys. Forms come with degrees — a one-form is what you integrate over a curve, a two-form over a surface, and so on — and they carry, built into their algebra, the orientation and the signs that in the old treatment you had to track by hand and hope you got right. The wedge product and the exterior derivative, the two operations Spivak introduces, encode the geometry that the classical vector-calculus operators of gradient, curl, and divergence were each capturing in a special dimension. Those three operators, it turns out, are one operator — the exterior derivative — restricted to three dimensions and rewritten three different ways.

The whole book is aimed at one destination, and it's worth naming what makes that destination remarkable. The generalized Stokes' theorem, in Spivak's treatment, says that the integral of a form over the boundary of a region equals the integral of its derivative over the region itself. That's the entire statement. And underneath it sit the fundamental theorem of calculus, Green's theorem, the divergence theorem, and the classical Stokes' theorem — every major integration result of the curriculum — each of them nothing more than this one sentence read in a particular dimension. The catalogue was never a catalogue. It was one idea, seen from several angles, mistaken for several ideas because we lacked the language to see it whole.

This is what the book is really an argument about, beyond its subject. It's a claim about what rigor is for. The common assumption — the one the fat textbooks embodied — is that rigor is a tax on understanding, the price of being careful, and that the more honest you are about the hard parts the longer and harder the book gets. Spivak's hundred and forty pages say the opposite. The rigor and the brevity are the same thing. It was the imprecision of the classical treatment, its reliance on pictures that half-worked, that forced it to keep making exceptions and adding cases. Precision is what let the exceptions disappear.

The book that promised more than its length could deliver turned out to deliver exactly by being that length. Everything in it — the derivative recast as a linear map, the differential form, the manifold — was assembled for the sake of a single sentence about boundaries and derivatives, and once that sentence arrives, the sprawling catalogue of multivariable calculus is revealed to have been one statement all along. The brevity wasn't a stylistic choice imposed on the material. It was what the material looked like once it was told the truth about itself.