For this one thing to know ,
no of subnets that can be formed using n bits = 2n - 2 and
same result for no of hosts in a given subnet as well given n bits meant for a given subnet.
Here we are following classful addressing scheme since it is mentioned in the question that the address is a class B address.
So we know according to default mask of class B,
no of bits for network id = 16 bits
no of bits for host id = 16 bits
Now for subnetting , the subnet id bits are taken from the host bits of the network.
So keeping in mind the above mentioned points ,
given no of subnets in a given network = 200
with 7 bits we can have : 27 - 2 = 126 subnets hence not sufficient.
So we need to take 8 bits since with 8 bits max no of subnets = 256
So remaining bits for hosts i.e. no of host bits for a subnet = 8 bits
Now we can have continuous as well as discontinuous subnet masks.
In continuous masks , we mask the bits and set the 0's of host bits to 1 but that should be continuous and just after the network bits whereas this is not necessary for discontinuous subnet masks.Any bits out of the given host bits can be set to 1 , not necessarily continuous.
To mention , we can take the examples in the option itself.
Option A is a continuous mask as all 8 1's just after the 16 1's of default mask of class B (first 16 bits of class B address belongs to network id and default mask of class B is 255.255.0.0)
Whereas option B and C are discontinuous subnet masks.In option B bits 1-4 and 9-12 after ending of first 16 bits are set to 1 and for option C , 1-2 and 9-14 bits are set to 1.
Hence D option is not a subnet mask and hence D) option is correct option.
Moreover as mentioned earlier , D) option i.e. 255.255.0.0 is a default mask for class B and hence not a subnet mask.
