Smith–Volterra–Cantor set

The smith–volterra–cantor set is a subset of the real line obtained by repeatedly deleting open intervals from the unit interval while arranging that the total deleted length is strictly less than one. It is a standard example of a set that is closed, nowhere dense, and of positive Lebesgue measure. The example demonstrates that smallness in the sense of Baire category does not imply smallness in the sense of measure.

Unlike the usual Cantor set, which has measure zero, the smith–volterra–cantor set has measure (1/2). It is consequently known as a fat Cantor set, although that term also applies to many related constructions with other positive measures.

Construction

Let (S_0=[0,1]). The first stage removes the open middle interval

[ \left(\frac{3}{8},\frac{5}{8}\right), ]

whose length is (1/4). The remaining set (S_1) consists of two closed intervals.

At stage (n), an open interval of length (4^{-n}) is removed from the center of each of the (2^{n-1}) closed components surviving from the preceding stage. The resulting sequence

[ S_0\supset S_1\supset S_2\supset\cdots ]

is decreasing, and the smith–volterra–cantor set is

[ S=\bigcap_{n=0}^{\infty}S_n. ]

The total length removed at stage (n) is

[ 2^{n-1}4^{-n}=\frac{1}{2^{n+1}}. ]

Thus the combined length of all deleted intervals is

[ \sum_{n=1}^{\infty}\frac{1}{2^{n+1}}=\frac12. ]

Because the deleted intervals are pairwise disjoint and measurable, countable additivity gives

[ m(S)=m([0,1])-\frac12=\frac12, ]

where (m) denotes Lebesgue measure.

Topological and measure-theoretic structure

Each (S_n) is a finite union of closed intervals. Their intersection (S) is therefore closed and, as a closed subset of the compact interval ([0,1]), is also compact. Every surviving interval is subdivided at a later stage, so no point of (S) is isolated. The set is consequently perfect.

The lengths of the component intervals of (S_n) tend to zero. Hence no nonempty open interval can remain entirely within (S), and the set has empty interior. Since (S) is closed, its closure is (S) itself, making (S) nowhere dense. Its boundary satisfies

[ \partial S=S. ]

The set contains no nondegenerate interval and is totally disconnected. As a topological space, it is homeomorphic to the ordinary Cantor set. The distinction between the two examples is therefore not topological but measure-theoretic: the ordinary Cantor set has measure zero, whereas (S) has positive measure.

Every perfect subset of the real line is uncountable, so (S) has the cardinality of the continuum. This remains compatible with nowhere denseness because cardinality, category, and measure describe different aspects of a subset’s size.

Historical development

Henry John Stephen Smith published an early construction of a nowhere-dense set of positive measure in 1875 while examining the integration of discontinuous functions. His construction established that the interval-deletion method could leave a substantial measurable remainder even when every open subinterval was eventually disrupted.

Georg Cantor later developed the ternary Cantor set in connection with the structure of point sets and uniqueness questions for trigonometric series. Cantor’s version removes intervals whose total length is one, producing a null set rather than a positive-measure remainder. Its formal resemblance to Smith’s construction contributed the third component of the compound name.

In 1881, You Watanabe analyzed the quarter-length deletion scheme in the setting of exceptional sets for derivatives. Watanabe expressed the retained measure through the convergent geometric series of deleted lengths and identified the resulting set as simultaneously perfect, nowhere dense, and of positive content. This formulation supplied the interval arithmetic used in contemporary applications of the construction.

Relation to Volterra’s example

Vito Volterra employed a positive-measure nowhere-dense set in his study of differentiation and integration. The resulting Volterra function is differentiable, but its derivative is discontinuous on a set of positive measure. A bounded function is Riemann integrable precisely when its set of discontinuities has measure zero, so such a derivative need not be Riemann integrable.

This application separates two statements that coincide under stronger regularity assumptions. A function can occur as an everywhere-defined derivative while failing the discontinuity criterion for Riemann integrability. The smith–volterra–cantor set provides the positive-measure exceptional set required for that separation without containing any interval on which the irregularity becomes locally unrestricted.

The name “smith–volterra–cantor set” consequently records three connected strands of nineteenth-century analysis. Smith supplied the early positive-measure construction, Volterra incorporated that type of set into an analytic counterexample, and Cantor’s related zero-measure construction provided the standard topological model. The quarter-length version described above became the conventional representative of this family.

Generalized fat Cantor sets

The particular value (1/2) is not essential to the underlying phenomenon. Interval deletions can be chosen so that their total length equals any prescribed value strictly between zero and one, while the lengths of surviving components still tend to zero. The remainder is then closed, perfect, nowhere dense, and of the complementary positive measure.