When curves became ‘pathological’
Mathematics once called certain curves pathological because they satisfied formal definitions while violating nearly every geometric expectation attached to the word curve.

Continuity without smoothness
Classical curves were imagined as locally straight under sufficient magnification. Weierstrass-type functions and the Koch curve showed that continuity does not guarantee a tangent. Roughness can persist through every scale of an ideal construction.
The Weierstrass function showed that continuity alone does not guarantee a smooth tangent. For nineteenth-century analysis this was not a visual trick, but a test of its definitions: intuition learned from ordinary curves could not be used silently as a theorem. The same idea can now be experienced graphically by adding ever finer oscillations, so that every magnification reveals new corners without introducing a jump in the function.
A line that approaches a plane
Peano and Hilbert described continuous mappings whose limit visits every point of a square. Their finite approximations are still ordinary polygonal paths; the space-filling property belongs to the infinite limit. This forced a separation between topological dimension, geometric measure and visual intuition.
Peano’s 1890 construction challenged a different expectation. A continuous map from an interval can reach every point of a square even though its domain is one-dimensional. This does not mean a finite drawing of a Peano curve already has area. Only the limiting map is surjective; every visible stage remains a long polygonal line. Separating parameter dimension, image set and finite approximation resolves the apparent paradox.
From exception to vocabulary
Fractal geometry did not erase the rigorous distinctions among these objects. It supplied a common language of scaling, dimension and recursive structure. What had appeared to be a cabinet of monsters became a toolkit for studying irregularity.
The label “pathological” weakened as related roughness appeared in probability, dynamics and natural measurement. Brownian paths, for example, are almost surely nowhere differentiable, yet they are central models rather than exotic exceptions. The historical monsters left a durable methodological lesson: when a definition admits a surprising object, the right response is not merely to dismiss the object, but to identify which unstated assumption intuition had supplied.
Sources and further reading
This article summarizes the following specialist sources in original wording. Accessed and editorially reviewed 12 August 2026.
- Koch SnowflakeWolfram MathWorld
- Space-Filling CurveWolfram MathWorld
- Fractal Geometry: Mathematical Foundations and ApplicationsWiley


