Quote Originally Posted by snaproll
As I noted above, the two drives on my machine still worked together when in cable select mode, I don't think the original ribbon is IDE2 but is is split, [I thought to facilitate routing...] (how can I tell- if it's a smooth ribbon it's not, if its split with a twist it is ?.. )
I would suggest some Google searching, but to be honest I just did that and I come away with more doubts than answers. I found some discussion of a CS IDE cable not with a twist but with one wire cut. Also I found information that directly conflicted with what ockham23 gave below.

Quote Originally Posted by snaproll
I plan to put them back together with the DVD as master & CD as slave. Does it matter which is on the end of the ribbon ? (SCSI can be fussy about that....)
If you jumper for master and slave, it will not matter. Pick which drive you wish to be which and run the cable for best reach to the drive and best air flow. The only concern I can mention here is that you should use as short a cable as reasonable (over 18 inches is considered risky although I have done it on one full size tower case with poorly positioned MB connectors) an if you only have one master device instead of a master and slave the master should be on the end of the cable, not on the connector closest to the computer (it's a transmission line thing to avoid the unterminated end of the wire from picking up signals like an antenna or echoing back signals that hit the end of the wire and bounce back).

Unfortunately, if you use CS at this point I'm not sure how it works. My Google search turned up references that say the master is the middle connector (closer to the MB) but ockham23 says that the master is on the end.
Quote Originally Posted by ockham23
Open the case, set the jumpers of both drives to "cable select (csel)" and put the DVD burner at the end of the IDE cable. As a result, the DVD burner should become the "master" drive for IDE channel #2.
I have no idea who is right, as I just don't use CS, and none of this causes me problems. If the above reference that I found is correct then it's just another reason to not use CS, at least when you only have a master or if the slave might ever be removed, even for testing. That extra unterminated connector hanging off the end of the cable (rather than being in the center, which is ok), can be a source of randon and hard to identify problems.