Page 745 - AP Computer Science A, 7th edition

8. (D) Anytime arithmetic operations are done with floating-point numbers, round-off error occurs. The Math class methods such as pow and sqrt use various approximations to generate their answers to the required accuracy. Since they do different internal arithmetic, however, the round-off will usually not result in exactly the same answers. Note that choice A is not correct because both Math.pow and Math.sqrt return type double. Choice B is wrong because no matter how x was previously calculated, the same x is input to pow and sqrt. Choice C is wrong since round-off error occurs no matter how many bits are used to represent numbers. Choice E is wrong because if x is representable on the machine (i.e., hasn’t overflowed), then its
square root, x1/2, will not overflow.