How Did Joseph Fourier Develop The Fourier Series?

I've always loved the moment when a messy, physical problem suddenly asks for a nice mathematical trick — and that's exactly how Joseph Fourier's story reads to me. He was studying how heat moves through solid bodies and found himself needing to describe an arbitrary initial temperature distribution. Instead of trying to force a single closed-form function onto that mess, he had the bold idea to write the temperature as a sum of simpler, oscillating pieces: sines and cosines. That move turned out to be profound. Using separation of variables on the heat equation, each of those sine/cosine pieces evolves in time in a simple exponential way, so the whole complicated evolution becomes a superposition of easy pieces.

I like picturing Fourier in the early 1800s, jotting down series that looked like sums of sin(nx) and cos(nx) and insisting they could represent very general functions — even ones with corners or jumps. He introduced formulas for the coefficients (what we now recognize as integrals projecting the initial shape onto each sine or cosine mode) essentially by exploiting orthogonality: multiply by a sine, integrate over the interval, and everything but one term cancels. That trick gives the coefficient integrals like a_n = (2/L) ∫ f(x) sin(nπx/L) dx in the usual setting. Fourier published an 1807 memoir and later his famous book 'Théorie analytique de la chaleur' in 1822, where he laid out this whole program.

It wasn't all applause — mathematicians of the day complained that he lacked rigorous proofs about when these series converge and what ‘‘function’’ even meant. But his physical intuition carried the field forward; later giants like Dirichlet and Riemann tightened the foundations. Every time I see a Fourier series on a whiteboard or hear a synth pad decompose into harmonics, I think of that leap: letting physics suggest a new way to represent functions. It still feels a bit like magic to me.

My take is that Fourier was basically daring the math community to be more flexible about what ‘‘function’’ means. Confronted with the heat equation, he used separation of variables to get spatial eigenfunctions (sines and cosines) that satisfy the boundary conditions. The initial condition — any temperature profile f(x) — he expanded as a series ∑(a_n sin(nπx/L) + b_n cos(nπx/L)), and found the coefficients by multiplying by the corresponding sine or cosine and integrating, thanks to orthogonality. That gives the familiar coefficient integrals like a_n = (2/L)∫ f(x) sin(nπx/L) dx for the usual interval conventions.

Fourier presented these ideas in an 1807 memoir and then more fully in 'Théorie analytique de la chaleur' (1822), arguing that these series could represent very general functions and using that to solve the heat equation. Mathematicians at the time worried about his lack of rigorous convergence proofs, but physically his method worked and sparked deeper analysis later. To me it's a beautiful example of physics leading mathematics: you get a tool that models the world well, and then mathematics catches up to explain why it works.

I tend to tell Fourier's origin story like a seasoned teacher telling a classroom anecdote: the problem was heat, the need was representation, and the solution was a radical expansion. Fourier was tackling the heat equation and showed that you can separate variables to reduce the PDE into ordinary differential equations whose spatial parts are trigonometric functions satisfying boundary conditions. Those trigonometric functions form a natural basis for the interval, and so any initial temperature profile can be expressed as a linear combination of them. Practically, you match coefficients by exploiting orthogonality: multiply by a basis function and integrate; all cross terms vanish, leaving the coefficient you need.

Historically, this wasn't invented in a vacuum. Earlier work by Bernoulli, d'Alembert, and Euler on vibrating strings had flirted with trigonometric series, but Fourier pushed the idea much further in a physical context and was bold enough to claim very general representation theorems in his 1807 memoir and then in the 1822 book 'Théorie analytique de la chaleur'. His approach was pragmatic: physical reasoning led him to accept series that mathematicians later wanted to justify more carefully. Questions about pointwise convergence, uniform convergence, and how to handle discontinuities were raised — and those later became central topics in analysis, addressed by Dirichlet, Riemann, and Lebesgue.

If you like, think of Fourier's method as turning a spatial profile into a spectrum: each frequency is a mode that decays over time at its own rate. That spectral viewpoint is now everywhere — signal processing, quantum mechanics, image analysis — and it all ties back to his original insight of expanding arbitrary functions into trigonometric components.