On the Number of Squares in an Arithmetic Progression