Dr. Thomas Shinder, MD, MCSE, was kind enough to explain why, with a subnet of 255.255.255.248 we couldn't use a host address of 192.168.0.23. (As mentioned on the subnet page, Dr. Shinder, along with his wife, Debra, has written several excellent books, all with five star ratings on Amazon com. Their "Troubleshooting Windows 2000 TCP/IP" was one of the sources for the subnet article. I not only have a copy of this book at home, but ordered a second copy to keep onsite as a reference. Additionally, I also told my supervisor to buy a copy, which he did, and, after reading, agreed with me that it is excellent.)
He wrote: "Let's solve the mystery of why you can't use 192.168.0.23 with a subnet mask of 255.255.255.248. (11000000.10101000.00000000).(00010)111 = 192.168.1.23 (11111111.11111111.11111111).(11111)000 = 255.255.255.248 So, the network ID is: 192.168.0.16 and Host ID: *7* The valid host IDs would be: 1 thru 6 Note that the Host ID portion is *all ones*. That's not good."Hopefully, this is clear. At this point in your subnet studies, it might not be. The thing to remember is that each 1 in a subnet mask represents the network while each 0 represents a host. What Dr. Shinder is pointing out here is that with a subnet mask of 255.255.255.242, there are only 3 0's left for host IDs. The last three digits of 23 in binary are all 1's--as we know, we can't have host ID's of all ones.
Once you have the basics of how to subnet down, this all becomes much clearer.