Its mostly because CGFloat can be of different sizes (float or double) depending on architecture, so can be assumed to be neither a double nor a float exclusively.