Measure Theory

Measure theory is the mathematics of assigning sizes to sets—lengths, areas, volumes, probabilities, or more abstract "weights"—and then integrating functions against those sizes.

10 frameworks, from 1854 to today · Analysis · Mathematics

still a working view ended · each leads to its entry below

Overview

Measure theory is the mathematics of assigning sizes to sets—lengths, areas, volumes, probabilities, or more abstract "weights"—and then integrating functions against those sizes1. Its central problem is that not every set can sensibly be measured, and not every function can be integrated by the methods of elementary calculus. Measure theory answers both difficulties at once, and in doing so it rebuilt the foundations of integration, probability, and large parts of analysis. It is the language in which modern probability is written, the tool that makes Fourier analysis rigorous, and the framework behind the geometry of irregular sets.

Integration Before Measure: Riemann's Sums

In the 1850s Bernhard Riemann gave the first rigorous general definition of the integral. The idea was to slice the horizontal axis into small intervals, approximate the function by a step function on each slice, and take a limit as the slices shrink2. This "Riemann integral" works for continuous functions and for many others, and it is still what students learn first2.

But it has limits. Riemann's construction fails for functions with too many discontinuities: the Dirichlet function, equal to 1 on the rationals and 0 on the irrationals, is not Riemann integrable, because every interval contains both kinds of point and the upper and lower approximations never meet. Worse, the Riemann integral does not behave well under limits3. A sequence of integrable functions can converge pointwise to a function that is not integrable, and one cannot in general swap a limit and an integral. Since analysis constantly needs to pass to limits, this was a serious obstacle.

Borel, Lebesgue, and Measuring Sets

The fix came from measuring sets rather than slicing axes. Émile Borel, in a series of papers around 1898, developed a theory of "measurable" sets: start with intervals, whose length is obvious, and close up under countable unions, countable intersections, and complements. The resulting collection—today called a σ-algebra (a family of sets closed under complement and countable unions)—is rich enough to include almost every set one meets in practice.

Henri Lebesgue, in his 1902 dissertation Intégrale, longueur, aire, turned this into an integral45. Instead of slicing the x-axis, he sliced the y-axis: for a nonnegative function, measure the set of points where the function exceeds each height, and integrate by summing the areas of these horizontal slices. This "Lebesgue integral" integrates the Dirichlet function without trouble (its integral is 0, since the rationals have measure zero) and, crucially, obeys powerful convergence theorems3. The monotone convergence theorem and the dominated convergence theorem say, under mild hypotheses, that limits and integrals can be exchanged—exactly what Riemann's theory could not guarantee67.

Lebesgue's construction also produced a notion of measure on the line: the "Lebesgue measure" of a set generalizes length4. Some sets, it turned out, cannot be assigned a consistent measure at all—the Vitali construction of 1905 exhibits a set with no Lebesgue measure, showing that the restriction to measurable sets is not a technicality but a necessity.

Carathéodory's Abstract Machinery

Lebesgue's theory was concrete, tied to the real line and Euclidean space. Constantin Carathéodory, around 1914, gave it an abstract form that could be transplanted anywhere. His key idea was an "outer measure": a function assigning a size to every set, monotone and countably subadditive, from which one singles out the "measurable" sets by a splitting condition—a set is measurable when it divides every other set additively89. This construction, the Carathéodory extension theorem, lets one build a measure from a pre-measure defined on a small collection of sets, and it became the standard way to construct measures in general10. It is the backbone of the abstract measure spaces—a set equipped with a σ-algebra and a measure—that modern texts present1.

Daniell and Stone: Integration First

An alternative route ran through integration rather than measure. Percy Daniell, in 1918, started from an "integral" defined on a space of functions—a linear, order-preserving functional—and extended it to a larger class by a limiting process, recovering measure as a by-product. Marshall Stone generalized this in the 1940s into the Daniell–Stone theorem, which characterizes integration as a positive linear functional on a lattice of functions. The approach is attractive because it begins with the objects analysts actually manipulate (functions and linear functionals) and derives the measure afterward. It remains a standard alternative to Carathéodory's construction and connects measure theory to functional analysis.

$L^p$ Spaces and the Riesz Representation Theorem

Once integration was secure, attention turned to the spaces of functions it defines. For \(1 \le p < \infty\), the space \(L^p\) consists of functions whose \(p\)-th power is integrable, with a norm measuring their size; \(L^\infty\) consists of essentially bounded functions. These spaces, developed in the early twentieth century and systematized in the 1910s and 1920s, are complete—every Cauchy sequence converges—which makes them Banach spaces and, for \(p = 2\), Hilbert spaces. This completeness is the payoff of Lebesgue's theory: the space of Riemann-integrable functions is not complete, but \(L^p\) is.

The Riesz representation theorem, proved by Frigyes Riesz in 1907 for the Hilbert-space case and extended by others, identifies the continuous linear functionals on these spaces with integration against a fixed function (or, in the general measure version, with integration against a measure). It ties measure to duality and became a cornerstone of functional analysis.

Haar Measure and Analysis on Groups

In 1933 Alfréd Haar proved that every locally compact topological group carries a measure invariant under the group's translations—a "Haar measure"—unique up to a positive constant. On the circle group this is ordinary arc length; on the real line under addition it is Lebesgue measure; on matrix groups it is the natural volume. Haar measure made it possible to do Fourier analysis on general groups, integrating functions over a group and decomposing them into characters11. This became the foundation of abstract harmonic analysis, developed through the mid-twentieth century, and it shows how measure theory travels far beyond the number line.

Kolmogorov and Measure-Theoretic Probability

Probability had long been a somewhat informal calculus of chances. In 1933 Andrey Kolmogorov's Grundbegriffe der Wahrscheinlichkeitsrechnung recast it entirely: a probability space is a measure space of total measure 1, a random variable is a measurable function, and expectation is an integral. Events are measurable sets; independence and conditional expectation get precise definitions112. This axiomatization unified probability with analysis and made deep results—limit theorems, martingale theory, stochastic processes—provable with measure-theoretic tools. It is the single most consequential application of measure theory outside analysis itself.

Hausdorff Measure and the Geometry of Irregular Sets

Felix Hausdorff, in 1918, introduced a family of measures designed to size sets that ordinary length and area cannot handle1314. Hausdorff measure generalizes the idea of covering a set by small balls and letting their radii shrink; varying a parameter yields a notion of "dimension" for sets like the Cantor set, which has Lebesgue measure zero yet a fractional Hausdorff dimension. This gave mathematicians a precise way to talk about the size of fractals and irregular sets, and it seeded geometric measure theory—the study of rectifiable sets, currents, and minimal surfaces—developed substantially by Herbert Federer and others from the mid-twentieth century onward.

Descriptive Set Theory and Measurability

A parallel tradition asked which sets are definable and how definability interacts with measurability. Descriptive set theory, growing from the work of Borel, Lebesgue, and later Polish-school mathematicians such as Wacław Sierpiński and Kazimierz Kuratowski, classifies sets by the complexity of their definitions—Borel sets, analytic sets, projective sets—and studies which of them are measurable. Results here are delicate: some definable sets are provably measurable, while for others measurability depends on axioms beyond the standard ones. This line of work keeps the boundary between the measurable and the unmeasurable under close scrutiny.

The Present

Measure theory today is a working toolkit rather than a settled monument. The Carathéodory and Daniell–Stone constructions coexist as standard ways to build measures; \(L^p\) spaces and the Riesz representation theorem are everyday language in analysis; Haar measure underpins harmonic analysis on groups; and Kolmogorov's axioms are simply what "probability" means. Geometric measure theory continues to develop the sizing of irregular sets and the calculus of variations in weak settings. Open questions remain at the edges—about which definable sets are measurable, about the fine structure of measures on fractals and metric spaces, and about extending integration to settings where classical measure theory strains. The field's history is one of successive widenings: from the line to abstract spaces, from sets to functions, from analysis to probability and geometry.

References

  1. Measure (mathematics) (introduction, Definition), Wikipedia
  2. Riemann integral, Wikipedia
  3. Lebesgue integral, Wikipedia
  4. Lebesgue measure, Wikipedia
  5. Henri Lebesgue, Wikipedia
  6. Dominated convergence theorem, Wikipedia
  7. Monotone convergence theorem, Wikipedia
  8. Outer measure, Wikipedia
  9. Carathéodory's criterion, Wikipedia
  10. Carathéodory's extension theorem, Wikipedia
  11. Locally compact group, Wikipedia
  12. Measurable function, Wikipedia
  13. Hausdorff measure, Wikipedia
  14. Hausdorff dimension, Wikipedia

The frameworks of Measure Theory

Each framework is a way of working the field has used: when it held, and the question it answers.

  1. 1854-1930

    Riemann integration

    How can areas under curves be computed rigorously?

    Fly to its tower in Nousfarer

  2. 1902-Present

    Lebesgue measure and integration

    Which sets have a well-defined size, and how can we integrate over them?

    Fly to its tower in Nousfarer

  3. 1907-Present

    L^p spaces and Riesz representation

    What are the complete spaces of integrable functions and their duals?

    Fly to its tower in Nousfarer

  4. 1914-Present

    Abstract measure spaces and Carathéodory extension

    How can measures be constructed abstractly on arbitrary sets?

    Fly to its tower in Nousfarer

  5. 1917-Present

    Descriptive set theory

    Which definable sets are measurable, and how complex can they be?

    About Descriptive set theory
    Descriptive set theory enters measure theory at the point where the naive question "which sets can be measured?" meets a deeper, structural question: "which sets can be described?" The framework’s organizing idea is that the pathological sets which resist measurement—like the Vitali set or the non-measurable sets constructed with the axiom of choice—are not merely rare or exotic. They are, in a precise sense, indescribable by the countable operations that build up the Borel hierarchy. Descriptive set theory provides the classification of sets according to how complex their definitions are, and this classification turns out to be the natural boundary for measurability.
    The central problem this framework addresses is the gap between the Borel sets—those generated from open sets by countable unions, intersections, and complements—and the full power set of a Polish space. The Lebesgue measure, as developed in the early twentieth century, assigns a measure to every Borel set, and then completes the measure by adding all subsets of null sets. But the completion process is opaque: it does not tell you which additional sets become measurable. Descriptive set theory supplies the missing vocabulary. It asks: if a set can be obtained as a continuous image of a Borel set, is it measurable? If it can be obtained by projecting a Borel set in a product space down to one coordinate, is it measurable? The answers to these questions, developed between 1917 and the 1930s, form the core of the framework.
    The essential concepts build in a natural order. First, one needs the notion of a Polish space: a complete, separable metric space, such as the real line, the Cantor space, or the Baire space. These are the natural arenas for descriptive set theory because their topology is rich enough to support the hierarchy of definable sets. Next come the Borel sets, which form the smallest σ-algebra containing the open sets. The Borel hierarchy stratifies these sets into levels: Σ⁰₁ (open), Π⁰₁ (closed), Σ⁰₂ (countable unions of closed sets), Π⁰₂ (countable intersections of open sets), and so on, transfinite through countable ordinals. This hierarchy is the backbone of the framework, and it predates the formal development of descriptive set theory, having been implicit in the work of Baire and Lebesgue.
    The crucial step beyond Borel sets is the class of analytic sets (also called Σ¹₁ sets). A set is analytic if it is the continuous image of a Borel set, or equivalently, the projection of a Borel subset of a product of two Polish spaces. The fundamental theorem, due to Souslin in 1917, is that every Borel set is analytic, and a set is Borel if and only if both it and its complement are analytic. This theorem is the hinge of the framework: it shows that the analytic sets form a natural extension of the Borel sets, and that the boundary between the two is exactly the boundary of complementation. The measurability theorem for analytic sets—that every analytic set is Lebesgue measurable—is the direct payoff for measure theory. It guarantees that projections of Borel sets, which arise constantly in analysis and probability, are always measurable, even though they need not be Borel.
    The framework then extends to the coanalytic sets (Π¹₁), the complements of analytic sets, and beyond into the projective hierarchy. The key result here is the regularity property: analytic and coanalytic sets are universally measurable, meaning they are measurable with respect to every Borel measure on the Polish space. This is a strictly stronger statement than Lebesgue measurability, and it is the reason descriptive set theory is indispensable to abstract measure theory. The framework also introduces the Luzin separation theorem, which states that disjoint analytic sets can be separated by a Borel set, and the perfect set property, which holds for analytic sets and implies that every uncountable analytic set has the cardinality of the continuum.
    The intellectual history of this framework is tightly bound to the French and Russian schools of analysis in the early twentieth century. The foundational work was done by Nikolai Luzin and his student Mikhail Souslin in Moscow, beginning around 1917, when Souslin discovered the analytic sets and proved that they are not all Borel. Luzin’s 1930 monograph Leçons sur les ensembles analytiques codified the theory. In parallel, Wacław Sierpiński in Poland contributed fundamental results on the hierarchy and on the cardinality of analytic sets. The framework’s local period, from 1917 onward, is precisely the era in which these results were established, and they remain the core curriculum today. Later developments, such as the use of determinacy and large cardinals in the 1960s and beyond, are refinements of this framework rather than its origin, and they belong to a different chapter of the subject.
    The framework differs from nearby approaches in a decisive way. Abstract measure theory, as developed by Carathéodory and Fréchet in the 1930s, treats measurability as a purely algebraic property: a set is measurable if it belongs to a given σ-algebra. Descriptive set theory instead treats measurability as a definability property: a set is measurable because it can be described by a sufficiently simple formula in the language of topology and countable operations. This shift in perspective has practical consequences. In probability theory, for example, stochastic processes are often defined as projections of Borel sets in path spaces, and the measurability of these projections is guaranteed by the analytic set theorem, not by any explicit construction of the σ-algebra. Similarly, in geometric measure theory, the measurable sets that arise from continuous images of compact sets are always analytic, and hence always measurable for any Borel measure.
    The prerequisites for understanding this framework are modest but real. The reader should be comfortable with the basics of point-set topology: open and closed sets, compactness, separability, and completeness. A working knowledge of Lebesgue measure and the definition of a σ-algebra is also necessary, since the framework is built on top of those notions. No prior exposure to descriptive set theory itself is required, and the framework can be learned directly from the classical sources, such as Luzin’s monograph or the later textbook treatments by Yiannis Moschovakis and Alexander Kechris. The essential mathematical maturity is the ability to handle transfinite induction and countable ordinals, since the Borel hierarchy is indexed by them, but this can be acquired alongside the material.
    The continuing significance of descriptive set theory in measure theory lies in its role as the boundary between the measurable and the non-measurable. Every set that arises in ordinary analysis—every set defined by a formula involving limits, integrals, or continuous functions—is either Borel or analytic, and hence measurable. The framework provides the guarantee that this remains true under projection and complementation, the two operations that most often produce pathological sets. It also explains why the axiom of choice is needed to construct non-measurable sets: such sets are not merely complicated, they are indescribable in the countable language of the Borel hierarchy. The framework thus gives measure theory its epistemological foundation: we can measure what we can describe, and the descriptive hierarchy tells us exactly what that means.
    The 7 concepts its tower teaches
    • Polish spaces
    • Borel sets
    • Analytic sets
    • Coanalytic sets
    • Measurability of analytic sets
    • Luzin separation theorem
    • Perfect set property

    Fly to its tower in Nousfarer

  6. 1918-Present

    Daniell integral

    Can integration be founded on linear functionals rather than measures?

    Fly to its tower in Nousfarer

  7. 1918-Present

    Geometric measure theory

    How can we measure and integrate over irregular geometric sets?

    Fly to its tower in Nousfarer

  8. 1918-Present

    Hausdorff Measure

    How can we assign a dimension to sets of zero volume?

    Fly to its tower in Nousfarer

  9. 1933-Present

    Haar measure

    Does every locally compact group have a translation-invariant measure?

    Fly to its tower in Nousfarer

  10. 1933-Present

    Kolmogorov measure-theoretic probability

    What are the axioms that make probability a rigorous mathematical theory?

    Fly to its tower in Nousfarer

How the ideas connect

Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.

Riemann integrationLebesgue measure and integrationL^p spaces and Riesz representationAbstract measure spaces and Carathéod…Descriptive set theoryDaniell integralGeometric measure theoryHausdorff MeasureHaar measureKolmogorov measure-theoretic probabil…

Beyond the field

Where Measure Theory's ideas came from in other fields, and what they led to there: the strongest 12 of 62 links.

A city in Nousfarer: a red-carpeted stair climbing terraces to a crowned tower on an acropolis, more towers round it, forest and snowy mountains beyond.
A city in Nousfarer, the game played in the browser.

Measure Theory is a planet in Nousfarer

Its city has a tower for each of these 10 frameworks. Land, walk in, and climb one concept at a time.

Land on Measure Theory

Related fields

Fields its ideas are linked to

More in Analysis

See where Measure Theory sits on the Map of Human Knowledge