Let
$\unicode[STIX]{x1D6FA}$ be a domain in
$\mathbb{R}^{m}$ with nonempty boundary. In Ward [‘On essential self-adjointness, confining potentials and the
$L_{p}$ -Hardy inequality’, PhD Thesis, NZIAS Massey University, New Zealand, 2014] and [‘The essential self-adjointness of Schrödinger operators on domains with non-empty boundary’, Manuscripta Math.150(3) (2016), 357–370] it was shown that the Schrödinger operator
$H=-\unicode[STIX]{x1D6E5}+V$ , with domain of definition
$D(H)=C_{0}^{\infty }(\unicode[STIX]{x1D6FA})$ and
$V\in L_{\infty }^{\text{loc}}(\unicode[STIX]{x1D6FA})$ , is essentially self-adjoint provided that
$V(x)\geq (1-\unicode[STIX]{x1D707}_{2}(\unicode[STIX]{x1D6FA}))/d(x)^{2}$ . Here
$d(x)$ is the Euclidean distance to the boundary and
$\unicode[STIX]{x1D707}_{2}(\unicode[STIX]{x1D6FA})$ is the nonnegative constant associated to the
$L_{2}$ -Hardy inequality. The conditions required for a domain to admit an
$L_{2}$ -Hardy inequality are well known and depend intimately on the Hausdorff or Aikawa/Assouad dimension of the boundary. However, there are only a handful of domains where the value of
$\unicode[STIX]{x1D707}_{2}(\unicode[STIX]{x1D6FA})$ is known explicitly. By obtaining upper and lower bounds on the number of cubes appearing in the
$k\text{th}$ generation of the Whitney decomposition of
$\unicode[STIX]{x1D6FA}$ , we derive an upper bound on
$\unicode[STIX]{x1D707}_{p}(\unicode[STIX]{x1D6FA})$ , for
$p>1$ , in terms of the inner Minkowski dimension of the boundary.