For instance a Hebrew or Chinese character might be ᣮ, and since its too big a number to fit into the ASCII/UTF-8 (what's the difference?) charaster set, it breaks down the number into several readable peices (which can number up to 4) and they appear as (4) individual (ASCII readable) characters instead of the single chinese character. So "1 Chinese charcater" (might) = (4 charcaters) A7£æ. A7£æ doesn't make sense on it's own, but with a decoder (I assume) it might combine these 4 charcters and recognize that it had been previously out of range of the original character set and translate it into the chinese character set again.