On the significance of perihelion shift calculations