Here is the proof from p. 48 of the Millennium Edition, corrected 4th printing 2006:
My question: how do we know that there is any ordinal $\theta$ such that $A=\{a_\xi\,\colon\xi<\alpha\}$? Presumably we need to use the fact that $A$ is a set because if $A$ is a proper class I don't think it's true. On the other hand, I don't see how we've used it. Am I missing something? And if I'm not missing something, what is the easiest/most elegant way to patch the hole? If the process keeps going indefinitely, $A$ would have to be a proper class, but I am not clear on how to stick the landing on that approach.
Thank you!