I sort of misspoke slightly, I should have said that you can't represent 0.1 in binary floating point. Much like you can't represent 1/3 in our decimal number system.