1875 – 1941 • The Measure of All Things
He redefined the integral from the ground up, creating measure theory and transforming real analysis into a rigorous, powerful discipline that underpins modern mathematics.
Henri Leon Lebesgue was born on 28 June 1875 in Beauvais, Oise, in northern France. His father was a typesetter and his mother a schoolteacher. The family was modest but valued education deeply.
Tragedy struck early: his father died of tuberculosis when Henri was still young, and he himself suffered from poor health throughout his life. Despite this, he excelled at school, particularly in mathematics.
He entered the Ecole Normale Superieure in Paris in 1894, where he studied under Emile Borel and was influenced by the work of Camille Jordan on the theory of "content" (a precursor to measure).
Even as a student, Lebesgue was drawn to the pathological functions that defied Riemann integration — functions like Dirichlet's that were "everywhere discontinuous."
Lebesgue's cohort at the Ecole Normale included Borel, Baire, and Denjoy — the "French school of analysis" that would remake the foundations of real analysis.
By 1900, Riemann's integral had served mathematics well for decades, but its limitations were becoming painfully clear. Functions arising naturally in Fourier analysis, point-set topology, and the theory of real functions resisted Riemann integration.
Emile Borel had introduced measurable sets in 1898, and Rene Baire had classified discontinuous functions. The stage was set for someone to build a comprehensive theory of integration based on measuring sets rather than partitioning domains.
The French school of analysis (Borel, Baire, Lebesgue, Denjoy) was also engaged in a philosophical debate about constructivism, with Borel and Baire holding more restrictive views than Lebesgue about what constituted a valid mathematical object.
The characteristic function of the rationals in [0,1] has no Riemann integral. Fourier series of well-behaved functions could converge to non-Riemann-integrable functions. The fundamental theorem of calculus had gaps.
Cantor's set theory provided the language. Jordan and Borel provided proto-measure concepts. Lebesgue fused them into a complete theory.
Instead of partitioning the domain into intervals (Riemann), Lebesgue partitioned the range into intervals and measured the size of the preimage sets.
For a non-negative function f, the Lebesgue integral is:
∫ f dμ = sup { Σ y_i · μ(E_i) }
where E_i = {x : f(x) ≥ y_i} and μ is Lebesgue measure. The "size" of these preimage sets is measured using Lebesgue measure, which extends the notion of length to vastly more sets than intervals.
"I have to pay a certain sum. I can take the bills from my pocket and hand them over one by one (Riemann). Or I can sort them by denomination first, then pay (Lebesgue)."
Lebesgue measure extends "length" from intervals to a vast σ-algebra of measurable sets.
Lebesgue's construction proceeds in stages:
Not every subset of R is Lebesgue measurable. The Vitali set (1905) was the first counterexample, constructed using the Axiom of Choice.
This was deeply unsettling: it showed that a "natural" measure on R cannot consistently assign a size to every set. The Banach-Tarski paradox later dramatised this in higher dimensions.
Lebesgue's ideas were later abstracted by Caratheodory, Radon, and Kolmogorov. Kolmogorov's 1933 axiomatisation of probability is built directly on Lebesgue measure theory.
If f_n → f pointwise and |f_n| ≤ g for some integrable g, then:
lim ∫ f_n = ∫ lim f_n
This is arguably the single most useful theorem in analysis. It provides a simple, checkable condition for passing limits through integrals.
If 0 ≤ f_1 ≤ f_2 ≤ ... and f_n → f pointwise, then:
lim ∫ f_n = ∫ f
No domination condition needed for increasing sequences. This builds the integral "from below" in a natural way.
For non-negative measurable functions:
∫ lim inf f_n ≤ lim inf ∫ f_n
A one-sided inequality that serves as the key lemma underlying the other convergence theorems.
Under Riemann integration, exchanging limits and integrals required uniform convergence — an extremely restrictive condition. Lebesgue's theorems provide far more flexible tools, enabling modern PDE theory, probability, and functional analysis.
Find where Riemann breaks
Partition range, not domain
Construct measure on sets
Derive powerful theorems
Lebesgue always emphasized geometric intuition. His thesis title, "Integrale, longueur, aire" (Integral, length, area), reveals his approach: start from concrete geometric problems (measuring lengths and areas) and build abstract theory to solve them.
His famous "paying bills" metaphor (sorting by denomination rather than order) made his abstract ideas accessible.
Lebesgue began with specific pathological examples — functions that defeated Riemann integration — and built a theory general enough to handle them all. His approach exemplifies the French analytical tradition: rigorous, systematic, building from clear foundations to powerful abstractions.
In 1905, Borel, Baire, Lebesgue, and Hadamard exchanged famous letters on the Axiom of Choice and the nature of mathematical existence. Lebesgue and Borel were sceptical of Zermelo's well-ordering theorem.
Ironically, Lebesgue himself had unwittingly used forms of the Axiom of Choice in his own work — a fact later pointed out by others, causing considerable embarrassment.
Lebesgue had tense relations with several contemporaries over priority. He clashed with Borel over who deserved credit for measure theory, and with Fubini over the double integral theorem.
These disputes, while painful, reflected the rapid, revolutionary nature of the field. Multiple mathematicians were converging on similar ideas simultaneously.
Lebesgue suffered from chronic ill health throughout his life and experienced periods of depression. He worked during both World Wars, witnessing the devastation of French intellectual life.
Kolmogorov's 1933 axiomatisation of probability is built on Lebesgue measure. Every probability space is a measure space.
L^p spaces (Lebesgue spaces) are the fundamental Banach spaces. L^2 is a Hilbert space central to quantum mechanics and signal processing.
Weak solutions, Sobolev spaces, and distributional derivatives all rest on Lebesgue integration. Modern PDE theory is inconceivable without it.
The Lebesgue integral resolved foundational questions about Fourier series convergence that had motivated the theory in the first place.
Birkhoff's ergodic theorem, the Poincare recurrence theorem, and all of dynamical systems theory use Lebesgue measure.
The Lebesgue integral has entirely replaced Riemann in advanced mathematics. It is the starting point for graduate real analysis worldwide.
Digital audio, image processing, and telecommunications all rely on L^2 theory (energy-finite signals). The Fourier transform is defined via the Lebesgue integral.
Wave functions live in L^2(R^3). The Born rule (probability = |ψ|^2) requires Lebesgue integration. Quantum measurement theory uses measure-theoretic probability.
Stochastic calculus (Ito integrals), option pricing (Black-Scholes), and risk theory all rest on measure-theoretic probability founded on Lebesgue's work.
Statistical learning theory, convergence of empirical measures, and PAC learning bounds use measure theory as their foundational language.
Inder K. Rana (2002). Clear graduate text covering Lebesgue measure and integration with historical context.
H.L. Royden & P.M. Fitzpatrick (4th ed., 2010). The classic graduate text, building measure theory from the ground up in Lebesgue's tradition.
Thomas Hawkins (1975). Definitive historical account of how Lebesgue's theory developed from its 19th-century precursors.
G. de Barra (2003). Accessible introduction that follows Lebesgue's original approach closely.
Henri Lebesgue (1903/1928). The original lectures — remarkably readable and available in French. A masterclass in mathematical exposition.
"Reduced to general theories, mathematics would be a beautiful form without content. It would quickly die."
— Henri LebesgueHenri Lebesgue • 1875–1941 • He measured the unmeasurable