Its funny, I read the list faithfully, but go months without having 
anything to say or ask.  And then come three things in a row!
A friend asked for help, and when I looked at his (business) setup I found 
the following.
One moderate spec machine, now a few years old and in need of renewal, with 
two network cards.  One card is DHCP to the one-port router, which goes to 
the ADSL connexion, which is fairly lightly used.  This is getting an 
address in the range 192.168.0.x, as usual. 
the other card is set up for static addressing, runs off an independent 
hub, and its using the range 172.16.0.1, with a subnet mask of 255.255.0.0. 
I was all set to find that there was some huge network behind this, but in 
fact all there is is two printers.  Yes, this network has one computer and 
two printers and appears to be using an internal Class B address for it, if 
I understand the situation correctly.
This was set up by a local computer store after apparently a great deal of 
head scratching.
Well, I too am scratching my head and trying to figure out could anyone 
have a legitimate reason (ie not simply they failed to understand 
networking) for doing this.  Like, if I really wanted to have the two 
printers on a separate net, why not just use a subnet mask of 
255.255.255.128 and get two subnets that way?  But in any case, why make 
life complicated with subnets, why would I not just assign them addresses 
in the 192.168.0.x range.  What I usually have done in these situations 
(but I'm an amateur at this stuff nowadays) is put the printer on a static 
address of 192.168.0.100, have the router assign addresses to the computers 
in the low numbers, and everyone seems to print just fine. There are only 
two printers and one computer here, so what could the problem be?
It seems that the simplest thing to do is take all this stuff out and use a 
5 port router/modem, and just make the two printers 101 and 102,  why would 
this not work? 
Is this missing something obvious?
Cheers
Al