> IIRC, it's not possible to represent .1 (or is it .2) accurately in IEEE 754. > Better to use duodecimal. Assuming you mean the decimals 0.1 and 0.2, they c

> IIRC, it's not possible to represent .1 (or is it .2) accurately in IEEE 754. > Better to use duodecimal. Assuming you mean the decimals 0.1 and 0.2, they can't be represented as terminating duodecimals (which is different from not being represented accurately!), either. Whatever base you choose, someone's favourite fraction won't have a terminating expansion.