# Recursion in pi-calculus

In the book by Sangiorgi and Walker ("The $\pi$-calculus: A theory of Mobile Processes"), Subsection 3.2 is devoted to recursion. They state the following constraint (pages 132-133):

"The environment is finitary, in the sense that any process depends on only finitely many constant definitions. (Formally, the transitive closure of the relation that relates two different constants $K$ and $K'$ if an instance of $K$ occurs in the definition of $K'$, is well-founded.)"

I do not understand how the two sentences are related. Right after there is an example with the definitions of $B_0$, $B_1$ and $B_2$: instances of $B_1$ and $B_0$ occur in the definition of $B_0$; instances of $B_2$ and $B_0$ occur in the definition of $B_1$; and an instance of $B_1$ occurs in the definition of $B_2$. Thus the environment is "finitary" in the sense that each one of the constants $B_0$, $B_1$ and $B_2$ only depends on three constant definitions. But how is "the transitive closure of the relation that relates two different constants $K$ and $K'$ if an instance of $K$ occurs in the definition of $K'$" well-founded? This transitive closure should relate any of the three constants $B_0$, $B_1$ and $B_2$ to any of these same constants and thus, if I consider the set $\{ B_0, B_1, B_2 \}$, I cannot find a minimal element.

What should be ruled out is the presence of infinite descending chains involving infinitely many constants $K$. This is however a different requirement from well-foundation.