To clarify: I claim my randomized rounding method has zero bias. This is based on the mathematical defintion of bias used by statisticians: a quantity whose expectation value is the value it is supposed to be, is "unbiased" otherwise it is "biased."
You can design it to leave X unbiased where X is a variety of things (designer chooses which he wishes to have zero bias for). I claim by this same (standard) definition, all other apportionment methods so far discussed, generically exhibit bias. warren d smith ---- election-methods mailing list - see http://electorama.com/em for list info
