Thanks to everyone.. I see what the problem is now... I had thought that these sorts of numbers would be represented as whole numbers and then the decimal place added back in for the final result, so 0.0005 + 0.0005 would be the same as adding 5+5 (or 00000101 and 00000101), and then shifting the decimal point appropriately at the output stage, but now that you've explained that the computer is trying to approximate the actual fraction using powers of 2, it makes a lot more sense as to what is happening.