1 track. put a loop on it.
hot but not clipping level.
create and Assign output of loop track to it's own "loop" buss.
Assign loop buss to "master"
 
create new buss called realmaster (outputs to soundcard)
send master buss output to realmaster.
 
check levels.
select mono on realmaster.
 
master clips.
???????????
 
Edit:
Ok, there seems to be some confusion in what I typed.
I think the word Master may be throwing some off.
In this case master is not last in the chain. realmaster is.
(master feeds realmaster).
when I mono realmaster, it affects master. That shouldn't happen.
 
In case you're wondering why do this, ARC is on realmaster.
 
 
 
 

Selecting mono sums the stereo track. Reduce levels by 3 dB before summing to mono.

Please re-read Craig.
(master is before realmaster)

I reported a long time ago that RMS meter levels in buses drop when the interleave of a downstream bus is changed to mono. This sounds like another manifestation of that, but in my case I don't think it affected peak levels, and it didn't affect vertical meters that I usually use.
 

Same here. I routed audio, peaking at -0.8dB to Preview buss, which is routed to Master buss. When Master buss is set to mono it is clipping, meter shows 2.2dB. Preview buss is also clipping, peaking at 2.2dB.

Soundwise
Same here. I routed audio, peaking at -0.8dB to Preview buss, which is routed to Master buss. When Master buss is set to mono it is clipping, meter shows 2.2dB. Preview buss is also clipping, peaking at 2.2dB.

That would be normal Susan. See my edit above in the OP.

Okay, now I understand...the master isn't the "master." I can in fact duplicate what you are experiencing, and I agree it is anomalous. However, I have never experienced it so I was taken aback. But the reason why I haven't experienced it provides the solution.
 
When I want to send from bus to bus, I set the "sending" bus output to "None" and use a send. This is probably force of habit ("I'm sending to a bus, therefore I will use a send to a bus"). If you do this, there are no level changes, upstream or downstream. (Aside from the anomaly you described, I also noticed that the first time I send a bus output to another bus, the second bus has a hotter signal level. If I use a send, there is no issue.)
 
Here's the screen shot that (I believe) sets up your scenario but I'm using sends to get from the Loop bus to the Master bus, and another send to get from the Master bus to the Realmaster bus. Because the Realmaster is set to mono, you'll note that it distorts for the reason I mentioned previously caused by summing stereo to create mono, but all the other bus levels are as expected.
 

 
So the bottom line seems to be...if you're sending from one bus to another, use a send and you won't run into problems. As to why chaining bus outputs to inputs causes issues, I don't know. The only possibility I can think of would be panning law issues that affect bus output pan but not send pan.

I did a quick check and sending the upstream bus to the downstream bus instead of outputting did not eliminate the apparent summing of signals on the upstream bus when the downstream bus is switched to mono.
 
Moreover, it's not just a metering issue. The output - and any sends - from the upstream bus actually becomes mono both audibly and in the meters if the you start with a true stereo signal.

brundlefly
I did a quick check and sending the upstream bus to the downstream bus instead of outputting did not eliminate the apparent summing of signals on the upstream bus when the downstream bus is switched to mono.

Any chance the bus output was inadvertently assigned? What you're saying seems to go against what the screen shot shows. As long as I used sends and set outs to "none"...no problems. So now I wonder what's different in your setup compared to what I was doing.

 

Anderton
So the bottom line seems to be...if you're sending from one bus to another, use a send and you won't run into problems. 

What? Are you serious?
 
Maybe we need a Buss Hud.