Consider this following code in C# 2.0
            int? i = 1;
int? j = i;
= null;

in fact it was complied like this:
1      Nullable<int> nullable1 = new Nullable<int>(1);
2      Nullable<int> nullable3 = nullable1;
3      nullable1 = nullable3.HasValue ? new Nullable<int>(nullable3.GetValueOrDefault() + 1) : new Nullable<int>();
4      Console.WriteLine(nullable1);
5      Nullable<int> nullable2 = nullable1;
6      nullable2 = new Nullable<int>();
7      Console.WriteLine(nullable2.HasValue);

please note line 3, when we increased the nullable i, it actually created a new instance of Nullable<int>
     new Nullable<int>(nullable3.GetValueOrDefault() + 1)
and in line 6, when we assign null to a nullable instance, it also created a new instance with null value.
    nullable2 = new Nullable<int>();

Once we created an instance of Nullable<T>, we can NOT modify the inner value any more. When we modify these value, we  actually create a new instance with the new value. Nullable<T> is an immutable type.

It looks like string, right? (though Nullable<T> is a value type)