The graphic only shows one send, you could have the signal coming back from that send and going back out to another send before it hits the buss. Can we get confirmation from cakewalk about this?

All the way back to the Project5 days:
 

Yeah That Diagram is clear enough, but it makes no sense to me to have sends in parallel, I always thought that sends and returns were in series. I have to think of a way to test this out.
Thanks Maartian,
 

Seems to me that if they were in series, you'd be stuck. If you want them in series, couldn't you just buss it that way?

stereo vs mono
 
requires different approaches
 
make sure you cover all those bases as well.

sausy1981
I always thought that if you send a vocal to a reverb buss and a delay buss, and the reverb buss is to the left of the delay buss, then the reverb send is on top and the delay send is underneath in the sends section of the track, I thought that some of the vocal is sent via the send to the reverb and returned with the reverb to the track then the vocal with the reverb is sent via the delay send to the delay buss and returned with with the delay on the vocal which had reverb on it to the track, thus you have the reverb feeding the delay. Was I wrong to think this, and is each send a dry send?

All sends on a track are independent of each other. If a send is routed to a bus then whatever is on that bus does not get routed back to the track the send is on. The bus outputs to wherever the bus output is set to, usually the master bus by default. If sends returned to the original track then quite a lot of routing would result in the original track's gain ramping up more and more as buses are added.

It would also cause problems if sends are used to set up multiple monitor mixes if whatever is sent ends up back in the original tracks. Let's say you've three musicians, each wanting a different monitor mix. Not just different levels, but differing reverbs or eq or compression as well. So you set up each track with three pre-fader sends and point each send at a separate bus. On those buses you put the different eqs, dynamics etc then route each bus to a different output to the interface and on to the musician's headphones.

You, in the control room, are listening to the master bus output.

If those busses returned their processed contents to the tracks all kinds of strange things would happen. Not least of which would be a rapid buildup of gain and volume on the master bus as you'd have set up a bunch of feedback loops.

Just think of a Send as a copy of the original audio.
 
And tlw is 100% correct - all sends are completely independent of each other and only get routed to where their output is routed.

You can step in murky, confusing waters here.  Typically, I'll add a delay on the FX/PC bin, say, for main vocals.  Then ship the vocals off through the vocal buss which has the main reverb.  Backing vocals will get their own FX bin delay or more reverb if they need it (they are backing vocals, after all) but still go through the vocals buss.  Just using that matrix keeps things simple.  If I feel I need to break that mental chain I can do it.
 
And remember, you can route sends to sends.  I've been doing  VO work and usually have the music mixed before vocals.  I run all the tracks and sends going to the master buss and route them to a "music" buss, except the VO which continues going to master buss direct.  Makes it easy to get the relative volumes balanced and a means to vol envelope the two sources.
 
@

Ultimately, use your ears to see if it sounds good. It was already mentioned, in a denser mix, it would probably be difficult to discern that level of subtlety. I am working on a mix right now, and I have a common delay bus and that is also being sent to the common reverb. I wanted the delay to have the room sound. It worked and it sounded good, so I stuck with it. It was also already mentioned that you could get away with placing a delay plug-in with the same preset (if that's what you want) on multiple source tracks and then sending that signal to as many reverb busses as you want. If CPU ever becomes problematic, just freeze. I'll say the common reverb is a guideline that will come up more consistently in these sorts of discussion, but this seems like one of those cases whereby if it sounds good, it is good.