R5RS specifies only the characters necessary to handle parsing Scheme source code itself, leaving almost everything else unspecified. The character ordering only guarantees that the characters a..z, A..Z and 0..9 are in order, and the actual integer values are left unspecified (although in practice almost all implementations use ASCII).
Unicode is fast becoming the most widely used character set, and full Unicode compliance was specified in the R6RS. However, Unicode is expensive to support on small and embedded systems, is slower and more space intensive than most native character sets, and there are competing unified character sets which should not be dismissed easily. Given this, while one would expect a general-purpose featureful Scheme implementation to support Unicode, it is not immediately clear that we should _require_ Unicode support unconditionally in the WG1 standard.
Given this, we need to determine what and how much we specify about characters and strings, including their integer values, orderings and case mappings.