Let $E$ be the class of ordered pairs such that $a\in B$:
$$E=\{(a,b):a\in b\} . $$
I want to prove that $E$ is not a set. Kelley (General Topology Theorem 104 p.267) does the next:
He supposes $E$ is a set, so that $\{E\}$ is also a set. And because $E\in\{E\}$, the pair $(E,\{E\})$ is in $E$, by definition, so we have the relation
$$ E\in \{E\}\in(E,\{E\})= \{E, \{E,\{E\}\}\, \}\in E. $$
Now he considers the set
$$ \{x:x=E \mbox{ or } x=\{E\} \mbox{ or } x=\{E,\{E,\{E\}\}\}\, \} $$
and says it contradicts the Axiom of Regularity. My question is how it contradicts the axiom.
I mean, if I have sets $$ a\in b \in c \in a, \tag{$\ast$} $$ why the set
$$S=\{a,b,c\}$$
contradicts the axiom?
My attempt:
I want to find an element $x\in S$ such that
$$x\cap S = \emptyset.$$
Suppose $x=a$:
$$a\cap S = \{z:z\in a \mbox{ and } (z=a,b\mbox{ or } c\}.$$
Now, what happens if z=a? I mean, $a\in b\in \cdots$, but for me $\in$ is not transitive, so $a\in a$ not follows from ($\ast$). I suppose my mistake is that, but I cannot understand why.
Thanks