I'm Puzzled

I’m hoping readers of Dissoi Blogoi can help me understand the portion of the following passage from the Republic which I’ve placed in bold type. (Btw, the link to Perseus is here.):

diomologhsa/meno/j g' e)/fhn e)gw/, kai\ a)namnh/saj u(ma=j ta/ t' e)n toi=j e)/mprosqen r(hqe/nta kai\ a)/llote h)/dh polla/kij ei)rhme/na.

[507b] ta\ poi=a; h)= d' o(/j.

polla\ kala/, h)=n d' e)gw/, kai\ polla\ a)gaqa\ kai\ e(/kasta ou(/twj ei)=nai/ fame/n te kai\ diori/zomen tw=| lo/gw|.

fame\n ga/r.

kai\ au)to\ dh\ kalo\n kai\ au)to\ a)gaqo/n, kai\ ou(/tw peri\ pa/ntwn a(\ to/te w(j polla\ e)ti/qemen, pa/lin au)= kat' i)de/an mi/an e(ka/stou w(j mia=j ou)/shj tiqe/ntej, “o(\ e)/stin” e(/kaston prosagoreu/omen.

e)/sti tau=ta.

kai\ ta\ me\n dh\ o(ra=sqai/ famen, noei=sqai d' ou)/, ta\j d' au)= i)de/aj noei=sqai me/n, o(ra=sqai d' ou)/.

panta/pasi me\n ou)=n.

Here are three translations. I can’t make much sense of any of them, or the Greek.

Jowett:

And again, we speak of a self-beautiful and of a good that is only and merely good, and so, in the case of all the things that we then posited as many, we turn about and posit each as a single idea or aspect, assuming it to be a unity and call it that which each really is.

Grube/Reeve:

And beauty itself and good itself and all the things that we thereby set down as many, reversing ourselves, we set down according to a single form of each, believing that there is but one, and call it “the being” of each.

Griffith:

We also say there is a beautiful itself and a good itself. And the same with all the things we then said were “many.” Applying the procedure in reverse, we relate them to a single form or character of each—since we believe it is single—and call it “what each is”.

My difficulties (or some of them):
1. I don’t understand which two procedures are mean to be the reverse of each other. Is it that Plato is presuming that applying the same word to many things is an operation which takes us from one thing to many? So the One Over Many is a kind of inverse operation of predication? Or is it that we couldn't have grouped many things in a single class without having engaged in an operation that is the inverse of the One Over Many?

2. How are we supposed to coordinate the two instances of ‘each’ here (e(ka/stou ... e(/kaston)? Is the first the kind or the class of many and the second the single form corresponding to that class? Or do the two instances indicate the same thing?

(Maybe there’s an easy answer to this. I don’t have any commentaries at hand, as I write this, except Nick White’s, which skips over the passage. But I figure that the passage is likely to be puzzling to others if it is to me and so worth discussing—not as experts, but as interested amateurs--and if someone posts something definitive on it, fine. Naturally I’ve been drawn to the passage by thinking about the One Over Many in Republic 10.596a, in connection with the TMA, and looking at this passage as intimating the same thing.)