Consider the following proof from the book "The Banach-Tarski Paradox" by Stan Wagon:
In the second paragraph he seems to implicitely use that every bounded set $E\subseteq\mathbb{R}^2$ is contained in a unique disc of minimal radius. Now the uniqueness of such a disc is actually easy to see, but I don't see why it must necessarily exist.

Let's suppose that $D(p_0,r_0)$ is a closed disc with center $p_0$ and radius $r_0$ that contains $E$.
Let $CR \subset \mathbb R^2 \times [0,\infty) \subset \mathbb R^3$ be the set of ordered pairs $(q,s)$ such that the closed disc $D(q,s)$ contains $E$ and such that $s \le r_0$.
Once you believe that the set $CR$ is closed and bounded, it follows that $CR$ is compact, and therefore the projection function $CR \mapsto [0,\infty)$ has compact image and hence has a minimum.
The set $CR$ is bounded, because if $(p,r) \in CR$ then $r \le r_0$, and then by picking any $q \in E$ one sees that $d(p,p_0) \le d(p,q) + d(q,p_0) \le r + r_0 \le 2 r_0$.
The set $CR$ is closed, because if $(p,r) \in \mathbb R^2 \times [0,\infty)$ is a limit of a sequence $(p_i,r_i) \in CR$ then for each point $q \in E$ we have $d(q,p_i) \le r_i$, and since distance is continuous we may pass to the limit to deduce that $d(q,p) \le r$.