Some mapping theorems for the numerical range