The process of rounding, or conversely, of carrying out, is one with which you are certainly already familiar. The general principle is that a digit to the right of a decimal point is rounded up if the one immediately to its right is greater than 5, and it is rounded off—that is, left as is, with the digit immediately to its right simply dropped—if the latter is less than 5. Thus, if you were rounding to two decimal places, 25.68632 would be rounded up to 25.69, while 25.68496 would drop the third decimal digit (and all beyond it) to end up as 25.68. The only potential complication is when there is only one digit to the right of the decimal place to which you are rounding, and that digit is precisely 5; for example, 25.675 and 25.685. For this kind of situation there is a convention that helps to avoid injecting a systematic rounding bias into your calculations. If the next-to-last digit is an odd number, round it up and drop the final 5; whereas, if it is an even number, leave it as it is and drop the final 5. Thus 25.675 and 25.685 would both be rounded to 25.68. The effect of the convention is to insure for this kind of situation that instances of rounding up and rounding off tend to occur equally often and in no particular pattern.