Rod, at 12:52 on Apr 15 1996, you wrote to Bill Grimsley...

RS> 2400     30
RS> 2550     31
RS> 2700     31
RS> 2850     32
RS> 3000     33
RS> 3150     34
RS> 3300     35
RS> 3450     38
RS> 3600     45
RS> 3750     54

BG> Christ, any wonder that I was having trouble with Paul as well! 

RS> You are having a brain fart there, the problem between you
RS> and Paul is primarily because YOUR line is pretty poor.

Not at all.  I'm fully aware that my line is fucking awful, but your own
stats show around a 10dB decrease in signal strength to Paul, when compared
with Dave's stats, and as previous testing from here indicated that the
Rockwells show around a 5dB shift, the other 5dB drop must be due to his
own line.

BG> Given FO between exchanges, that extra 10dBm can only
BG> be due to the line between Paul and his own exchange
BG> being pretty piss-poor, as I already suspected.

RS> Thats actually the 10db quirk seen when there isnt a Courier
RS> on the other end of the call. The evidence for that is the
RS> call Russ made from FNQ with a Courier on Pauls line.

It ain't a full 10dB though.  Closer to 5dB in fact, and don't forget that
a 5dB drop is as near as dammit to HALVING the signal level.  With my
dreadful lines, that's more than enough to create major problems to Paul.

RS> Also did some very interesting calls to the USR BBS in the USA to see the
RS> effect of the 4 different ways you can call the US. Very fucking 
RS> interesting indeed actually, but I will post it in Aust_Modems coz its 
RS> likely to be interesting to others on the OS calling question. Looks very 
RS> interesting indeed graphing all 4 calls on the one graph in a spreadsheet. 
RS> Very bloody obvious why 2 got 26K type speeds and two got 16K.

I look forward to reading it.  Let's just hope that it gets up here...

Regards, Bill

--- Msgedsq/2 3.20