Shannon Fano Encoding
There are generally two descriptions of this technique available online.
One description will tell you that the most frequent symbol encountered is assigned the code 0, then the next most frequent code is assigned the code 10, then 110, and so on. This is not a very efficient compression scheme, but it's a good way to introduce the PrefixProperty to beginners.
The other description will tell you that all the symbols encountered are split into two subsets with roughly equal total frequency, assigning one the prefix 0 and the other the prefix 1. These subsets are recursively split with the same technique, each time adding a 0 or a 1 to each subsets' code, until each subset only contains a single symbol.
The data compression faq describes the latter http://www.faqs.org/faqs/compression-faq/, so it's probably safe to assume the former is just a misunderstanding that has propagated all about the internet..