On this page
Most students meet epsilon-delta as the hardest line in the limits chapter: "for every ε > 0 there exists δ > 0 such that...". It looks like a puzzle invented to make calculus harder. It was actually invented to stop calculus from being wrong. The person who made it the standard way of working was Karl Weierstrass, who spent about fifteen years teaching schoolchildren before anyone noticed him. Our post on Augustin-Louis Cauchy covers the earlier, verbal definition. This one covers Weierstrass, the precise version, and how to write two short proofs.
Who was Karl Weierstrass?
The MacTutor biography gives his dates as 31 October 1815, Ostenfelde, to 19 February 1897, Berlin. His father sent him to the University of Bonn to study law and finance. He attended neither those lectures nor the mathematics ones, spent four years fencing and drinking, studied mathematics alone, and left without a degree.
He then trained as a school teacher at Münster, where Christoph Gudermann encouraged him. From 1842 he taught at schools in Deutsch Krone and Braunsberg, covering not only mathematics but physics, botany, geography, history, German, calligraphy and gymnastics. He later described the "unending dreariness and boredom" of those years, with no colleague to talk mathematics with and no library.
A paper on abelian functions in Crelle's Journal in 1854 changed everything. Königsberg gave him an honorary doctorate that year, and in 1856 he moved to Berlin, where his lectures drew students from around the world. He taught Sofia Kovalevskaya privately when the university would not admit her; see our post on Kovalevskaya and the spinning top.
What was wrong with the older definition?
Cauchy had already used inequalities with ε and δ in his proofs. Judith Grabiner's paper Who gave you the epsilon? (1983) shows this, and also points out the gap: Cauchy did not distinguish between "for any epsilon and for all x" and "for any x, given any epsilon". In other words, he did not say whether δ could depend on the point x. That distinction separates ordinary continuity from what is now called uniform continuity, and the confusion led to wrong theorems about series.
What did Weierstrass write?
Weierstrass published little, and his ideas spread through his lectures and students' notes. Galina Sinkevich's history On the history of epsilontics (2015) reports that the earliest known text using the ε–δ technique as a method comes from his summer 1861 lectures, in notes taken by his 18-year-old student H. A. Schwarz. In the translation she quotes, the definition reads: "If it is possible to determine such boundary δ for h that for all values of h, the absolute value whereof is still smaller than δ, f(x+h) – f(x) becomes smaller than any arbitrarily small value of ε", then the function is continuous. Sinkevich concludes that Weierstrass was the first to use the language of ε and δ as a method.
In modern form: f(x) → L as x → a if for every ε > 0 there is a δ > 0 such that 0 < |x − a| < δ guarantees |f(x) − L| < ε.
Worked example 1: limit of 3x + 1 as x → 2 is 7
We want |(3x + 1) − 7| < ε.
Simplify: |3x − 6| = 3|x − 2|.
So 3|x − 2| < ε whenever |x − 2| < ε/3. Choose δ = ε/3.
Check with numbers: for ε = 0.03, δ = 0.01. At x = 2.009, 3x + 1 = 7.027, which is within 0.03 of 7.
Worked example 2: limit of x² as x → 3 is 9
|x² − 9| = |x − 3| × |x + 3|. The second factor changes with x, so first restrict x.
Insist that |x − 3| < 1. Then 2 < x < 4, so |x + 3| < 7.
Now |x² − 9| < 7|x − 3|, which is below ε if |x − 3| < ε/7.
Choose δ = the smaller of 1 and ε/7. For ε = 0.07, δ = 0.01. At x = 3.01, x² = 9.0601, within 0.07 of 9.
What mistakes do students make with epsilon-delta?
Choosing δ that depends on x. δ may depend on ε and on the point a, never on the moving x. In example 2 the bound |x + 3| < 7 is what removes x.
Working backwards and stopping. Finding δ is scratch work. The proof is the forward check: given δ, show the inequality holds.
Thinking it is only for olympiads. The JEE (Main) 2026 syllabus lists limits and continuity without requiring ε–δ proofs, but the idea explains why the standard limits work. Our guide to limits and continuity for JEE has the exam-style practice.
If the definition still reads as symbols rather than meaning, the Fermi Concept Library covers every concept in physics, chemistry, biology, mathematics and computing, including limits and continuity.
What else did his rigour lead to?
MacTutor's history of the function concept records that in 1872 Weierstrass sent the Berlin Academy an example of a function that is continuous everywhere but differentiable nowhere. It contradicted the common belief that continuous functions fail to have derivatives only at special points, and it caused, in Hermann Hankel's account, disbelief. Only a precise definition could prove such a thing, or rule it out.
What can a student take from Weierstrass?
MacTutor says he published little because his critical sense made him rebuild every argument on a firm base. He also kept working through years in which nobody was watching. Write your own proofs as if a sceptic will read every line.
Try it now: prove that the limit of 5x − 2 as x → 1 is 3, and find δ when ε = 0.01.
