How exactly does geodatabase replication does this work?

Okay. So.

I just made 2 different checkout replicas from an Enterprise geodatabase, Replica1 and Replica2. They are full/complete data replicas.

I then did the following steps:

1) I made and saved a few edits to Replica1.
2) I deleted the original data from the Enterprise geodatabase
3) I copied the edited data from Replica1 into the now empty Enterprise geodatabase.
4) I versioned all the data.
4) I made and saved a few edits to Replica2.
5) I syncronized the edits from Replica2 to the Enterprise geodatabase

Now this all worked. Where exactly is the replica information being saved? Why am I able to delete the original data from the Enterprise geodatabase and copy the data from Replica1 and everything still works fine?

What's going on in the background? I don't really understand how it all works.

Thanks for any insight.