reply: We do not concede this. That would be so only if it were not the case that each member depended on another in an infinite regress. It cannot be objected that the totality of what the aggregate depends upon must be either the aggregate itself, something within it, or something external to it; that the first is impossible, since otherwise something would be its own cause; that the second is impossible, since otherwise some of the parts would suffice for the aggregate; and that the third is therefore correct. We reply: If by ‘the totality of what the aggregate depends upon’ you mean all the things of which it can individually be said that the aggregate depends upon them, why do you say that this cannot be the aggregate itself? What demonstrates this is that the totality of the things upon which the necessary and the contingent depend is neither internal to the aggregate, since it depends upon every one of its parts, nor external to it. It is therefore the aggregate itself. If, however, you mean the efficient cause, why do you say that it follows that some of the parts must suffice for the aggregate?”

And if the aggregate is contingent in itself, it depends upon something other than itself. What the aggregate depends upon must therefore be either the aggregate itself, something within it, or something external to it. The first is impossible, since otherwise something would be its own cause. The second is impossible, since otherwise some of the parts would suffice for the aggregate. For if the aggregate is contingent and depends only upon some of its parts, those parts must be what brings the aggregate into existence. It would follow that they bring both themselves and their cause into existence. And if that which