1815 – 1897 • The Father of Modern Analysis
The schoolteacher who spent 15 years in obscurity before revolutionizing calculus with absolute rigor — and shocked mathematics with a continuous, nowhere-differentiable function.
Born on October 31, 1815 in Ostenfelde, Westphalia, Karl Weierstrass had an inauspicious start in mathematics. His father, a civil servant, sent him to study law and finance at the University of Bonn in 1834.
Weierstrass spent four years at Bonn without earning a degree, reportedly spending his time on fencing and beer-drinking. However, he was secretly studying the works of Jacobi and Abel on elliptic functions.
He then trained as a secondary school teacher at the Munster Academy (1839-1842), where he studied under Christoph Gudermann, one of the few experts on elliptic functions. This private study laid the foundation for his later revolution.
From 1842 to 1854, Weierstrass taught mathematics, physics, German, geography, handwriting, and gymnastics at secondary schools in provincial Prussia. He had no access to a research library and no mathematical colleagues.
During these 12 years in obscurity, Weierstrass developed an entirely original approach to analysis, working in complete isolation. He published in an obscure school journal that no mathematician would normally read.
Weierstrass published "Zur Theorie der Abelschen Funktionen" in Crelle's Journal in 1854. The mathematical world was stunned: here was work of the highest caliber from a completely unknown schoolteacher. The University of Konigsberg immediately granted him an honorary doctorate.
Appointed to the University of Berlin in 1856, Weierstrass became the most influential mathematics teacher in the world. His lecture courses on analysis, elliptic functions, and calculus of variations set the standard for rigor that persists today.
Weierstrass rarely published formally. His ideas were disseminated through carefully prepared lecture courses, which students copied and circulated. Many fundamental results of modern analysis first appeared in these lecture notes.
His students included Cantor, Schwarz, Mittag-Leffler, Kovalevskaya, Killing, Holder, and Hilbert. Through them, the Weierstrassian approach to analysis spread across all of European mathematics.
By the mid-19th century, calculus was enormously powerful but logically shaky. Newton and Leibniz had used "infinitesimals" — quantities infinitely small but not zero — without rigorous definition. Cauchy had begun the reform with his definition of limits, but gaps remained.
Mathematicians routinely assumed that continuous functions were differentiable (except perhaps at isolated points), that convergent series of continuous functions had continuous sums, and that various limit operations could be interchanged freely.
All of these assumptions were wrong. Weierstrass' mission was to rebuild analysis on absolutely firm foundations, replacing intuitive arguments with precise epsilon-delta proofs.
Riemann used geometric intuition brilliantly but sometimes lacked rigor. Weierstrass represented the opposite extreme: algebraic and arithmetical precision. Their approaches were complementary, and the tension between them drove analysis forward.
The rivalry between Weierstrass' Berlin school (algebraic rigor) and the Riemann/Klein Gottingen tradition (geometric intuition) shaped mathematics for decades. Both approaches proved indispensable.
Weierstrass perfected the epsilon-delta definition of limits, continuity, and uniform convergence. While Cauchy had the basic idea, Weierstrass formulated it with complete precision:
f is continuous at a if:
For every ε > 0, there exists δ > 0
such that |x - a| < δ implies |f(x) - f(a)| < ε
He distinguished pointwise from uniform convergence and showed that only uniform convergence preserves continuity under limits. This resolved paradoxes that had plagued analysis for decades.
Abel (1826) had noted that Cauchy's "theorem" — that a convergent series of continuous functions has a continuous sum — was wrong. Fourier series provided counterexamples. But nobody could precisely state what additional condition was needed.
The missing condition is uniform convergence: the rate of convergence must be independent of the point. With this distinction, Weierstrass proved correct versions of all the limit-interchange theorems that had been used carelessly.
If |f_n(x)| ≤ M_n for all x, and ∑M_n converges, then ∑f_n converges uniformly. This simple but powerful test remains the standard tool for establishing uniform convergence in analysis courses worldwide.
Weierstrass based his entire theory of analytic functions on power series, avoiding Cauchy's integral-based approach. He showed that every analytic function equals its Taylor series in a disk of convergence, providing a purely algebraic foundation for complex analysis.
In 1872, Weierstrass presented his most shocking result: a function that is continuous everywhere but differentiable nowhere.
f(x) = ∑ a^n cos(b^n π x)
where 0 < a < 1, b is odd,
and ab > 1 + 3π/2
This "pathological" function shattered the widespread belief that continuous functions must be smooth except at isolated points. Hermite famously wrote: "I turn away with fear and horror from this lamentable plague of continuous functions which do not have derivatives."
Far from being a curiosity, such functions later proved central to fractal geometry and the study of Brownian motion.
Each term a^n cos(b^nπx) is smooth, but b^n oscillates faster than a^n decays. The series converges uniformly (giving continuity), but the derivative series diverges everywhere. The function has infinite total variation on every interval.
The Weierstrass function is approximately self-similar: zooming in on any portion reveals similar jaggedness at every scale. This makes it an early example of a fractal, decades before Mandelbrot coined the term.
Every bounded sequence has a convergent subsequence. This theorem, which Weierstrass proved rigorously, is fundamental to real analysis and is the key to proving that continuous functions on closed intervals attain their maximum.
Every continuous function on a closed interval can be uniformly approximated by polynomials (the Weierstrass approximation theorem). Its vast generalization by Stone applies to any compact space and any separating algebra of functions.
Weierstrass developed his own approach to elliptic functions based on the Weierstrass ℘-function (the Weierstrass p-function), defined as a sum over a lattice in the complex plane:
℘(z) = 1/z^2 + ∑' [1/(z-ω)^2 - 1/ω^2]
This function satisfies the differential equation:
(℘')^2 = 4℘^3 - g2℘ - g3
The ℘-function parameterizes elliptic curves, connecting analysis to algebraic geometry. Every elliptic function can be expressed rationally in terms of ℘ and ℘'.
Beyond elliptic functions, Weierstrass attacked the far harder problem of abelian functions — functions of several complex variables with multiple independent periods. His product theorem shows how to construct entire functions with prescribed zeros:
Every entire function can be written as a product over its zeros, multiplied by an exponential factor. This infinite product representation generalizes the factorization of polynomials to transcendental functions.
The Weierstrass preparation theorem provides a canonical form for analytic functions of several variables near a zero, reducing local questions to polynomial algebra. It is foundational in algebraic geometry and several complex variables.
"A mathematician who is not also something of a poet will never be a complete mathematician."
— Karl WeierstrassWeierstrass' defining characteristic was insistence on complete logical precision. Every statement required proof; every proof required explicit quantifiers. He eliminated geometric intuition from the foundations of analysis, replacing it with arithmetic.
Where Riemann used geometry and Cauchy used integrals, Weierstrass built everything from power series. This algebraic approach had the advantage of extending naturally to several variables and abstract settings.
Weierstrass used pathological examples not as curiosities but as essential tools for understanding the limits of theorems. His nowhere-differentiable function showed precisely where intuition fails and what additional hypotheses are needed.
Unusually, Weierstrass transmitted most of his work through lectures rather than publications. This created a dynamic, evolving body of knowledge that was constantly being refined and improved through the teaching process.
Weierstrass' relationship with Sofia Kovalevskaya, his most brilliant student, combined deep mathematical collaboration with personal affection. He fought tirelessly to secure her academic positions in a world hostile to women in mathematics.
Weierstrass endured one of the most painful rivalries in mathematical history. Leopold Kronecker, his colleague at Berlin, became an increasingly hostile critic of Weierstrass' analytical methods.
Kronecker rejected the use of irrational numbers, infinite sets, and non-constructive existence proofs — the very tools Weierstrass had used to build modern analysis. He reportedly said he wanted to "drive Weierstrass out of Berlin" and used his editorial power to delay publications.
The conflict was both personal and philosophical. Kronecker's constructivism anticipated 20th-century debates about foundations, but his attacks on Weierstrass and Cantor were often cruel.
Weierstrass admired yet competed with Riemann. He found a gap in Riemann's proof of the mapping theorem (the use of the Dirichlet principle) and insisted on providing rigorous alternatives. This productive tension improved both their legacies.
When Kovalevskaya arrived in Berlin (1870), women were not admitted to the university. Weierstrass taught her privately and championed her career. Their correspondence reveals a deep intellectual partnership that transcended the conventions of the era.
Weierstrass suffered from poor health throughout his career, including long periods confined to a wheelchair. Despite this, he continued lecturing into his late seventies, carried to the podium by assistants.
Every modern analysis course teaches Weierstrass' epsilon-delta definitions. His approach to rigor became the standard, and every student who studies limits encounters his framework.
The Weierstrass factorization theorem, product representations, and the theory of entire functions remain central. His approach via power series complements Cauchy-Riemann methods.
The Weierstrass function was the first fractal, a century before the term was coined. Mandelbrot explicitly credited Weierstrass as a precursor of fractal geometry.
The Weierstrass ℘-function links elliptic curves to complex analysis. Every elliptic curve over C is C/L for some lattice L, parameterized by ℘ and ℘'.
The Stone-Weierstrass theorem is the foundation of approximation theory, with applications in numerical analysis, signal processing, and machine learning.
Weierstrass' work on convergence, approximation, and function spaces paved the way for 20th-century functional analysis and the theory of Banach and Hilbert spaces.
The Weierstrass approximation theorem justifies using polynomials to approximate functions in computation. Chebyshev approximation, splines, and finite element methods all rest on this foundation.
Brownian motion paths are almost surely continuous but nowhere differentiable — exactly like the Weierstrass function. Modern stochastic calculus (Black-Scholes, etc.) deals directly with such paths.
Fracture surfaces and crack propagation in materials exhibit self-similar roughness analogous to the Weierstrass function. Understanding these "pathological" geometries is essential for materials engineering.
Neural network approximation theorems are descendants of the Weierstrass approximation theorem. The universal approximation property of neural networks echoes Weierstrass' result for polynomials.
Elliptic curve cryptography uses the Weierstrass normal form y^2 = x^3 + ax + b to define curves. Every elliptic curve can be put in this canonical Weierstrass form.
Uniform convergence guarantees (from Weierstrass' theorems) ensure that truncated Fourier series faithfully approximate signals, providing the mathematical backbone of digital audio and image processing.
Weierstrass / ed. R. Siegmund-Schultze (1988) — Selected chapters from Weierstrass' function theory lectures, including his treatment of the nowhere-differentiable function. Primary source.
Hairer & Wanner (1996) — Beautifully illustrated account of the development of analysis, with extensive coverage of Weierstrass' role in the rigor revolution.
David Bressoud (2nd ed., 2007) — Teaches real analysis through its historical development, making Weierstrass' contributions accessible to undergraduates.
Various biographies — The Weierstrass-Kovalevskaya relationship is covered in biographies of both mathematicians, illuminating the human side of 19th-century mathematics.
"A mathematician who is not also something of a poet will never be a complete mathematician."
— Karl WeierstrassFor every ε > 0, there exists δ > 0...