ICND1 Subnetting confusion

I was creating some of my own practice subnetting questions and I'm kind of confused here.  I might just be making a mistake, but when I follow the directions as lined out in video 19 i am coming up with some strange results.

For instance, lets say I'm subnetting a class c network 200.15.32.0 and want to create 8 networks.
So I line out my network and host bits like so
128 - 64 - 32 - 16 - 8 - 4 - 2 - 1
0  -  0  -  0  -  0  -  [1  - 0  - 0  -  0]  -  four bit positions to make 8
so I represent these four bits in the binary and convert

11111111 11111111 11111111 11110000
255.255.255.240
256-240 = 16 for my increment, but this gives me 16 networks instead of the 8 that I need.  
The same thing happens when I want to create a subnet with 4 networks.  If I use network sizes that differ from my bit representations (10, 9, 7 etc.), my math comes out right.  Is this just a quirk in the way the math is done or am I doing something wrong?

Thanks for any help!

-Jake