Subnetting with a 25 bit mask will give you two subnets in your class C address space. The first subnet will have an address range of 0-127, with 0 being the network address and 127 being the broadcast address. That leaves you with addresses 1-126for your clients. The next subnet will have a network address of 128 and a broadcast address of 255. That leaves 129-254 available for use. For example, your two networks will be 172.16.8.0 255.255.255.128and 172.16.8.128 255.255.255.128
I don't know your network setup, but these networks will need to be identified to your router. Typically you would connect one subnet to an ethernet port and assign the port an address in the subnet with the right mask. The router then understands which group of addresses it is listening for. As an example, one ethernet interface would be assigned 172.16.8.1 255.255.255.128 and another interface would have an address of 172.16.8.129 255.255.255.128. If you have a switch with an internal routing module like a cisco 5000 series, you would apply these addresses to the VLANS you have configured.
For more information on this topic, go to www.cisco.com and do a search on IP subnetting, or VLSM (variable length subnet masking). You'll find a lot of information there.
Be sure to leave yourself some room for "fixed" or static ip addressing. You'll need some for the routers, switches and printers on your network. I'd like to be able to help you with your NT config, but I'm more of a router monkey. We use MetaIP here for our DHCP, and from what I understand, the administrator sets up the subnet and mask and identifies the range of addresses to be used for DHCP assignment. I hope this helps.
