PE> The modem was never locked at 57600, it merely happened to 
PE> be the speed the com port was set at at the last &W.  

DD> That is how you lock the speed into the Courier . . . .

No it isn't.  You lock the speed by setting your comms software
to a particular speed, whilst making sure that Courier has
&B1 in it (not necessarily in NVRAM either).  Try it on an
outgoing call and you'll find it works a treat.  You can do it
on an ingoing call too, just make sure you have done at ATS0=0
or some other harmless command.

PE> That 
PE> actually happened on my DOS system, and on my OS/2 system 
PE> the rate was set to 38400 (which is what I wanted).  Then I 
PE> had planned on changing it to 57600, and perhaps 115200 one 
PE> day, as an experiment.  So now I've told you the reason why I
PE>  chop and change baud rates as required.  Now you name the 
PE> software that jumps up from 38400 to 57600 after issuing an 
PE> ATZ. 

DD> None that I know of.  

Exactly, it is completely mad.

DD> I still want to know why you would set your software 
DD> to 38,400 after you have locked your modem speed to 57,600?

I set my OS/2 software to 38400, because I was worried about
it impacting my system, but I set my DOS system to 57600 because
I didn't care about it impacting anything, because it was single
tasking.  Even if I had run DOS at 38400, I would STILL have been
increasing my speed to 57600 under OS/2 at a later date.  One
day, if I find that COM.SYS supports 115200, I will be increasing
the speed to 115200.  

Now that I've answered your question AGAIN, at least have the 
courtesy to read my reply rather than asking the same question.

Oh, and then explain why the way the USR is doing it makes any
sense whatsoever, and how it could possibly be better than the
way Rockwell do it.  ie, the USR is fucked by design.

BFN.  Paul.
@EOT:

---