hi,
I don't think it matters much what datatype the column is..it's more about the data pattern in that particular field.

I generally don't index fields with non-repeating data..for example, a number field that represented a group number would be ok to index, but that same field containing unique user numbers probly wouldn't...the server would have to search the index, which would reference the table, taking up clock cycles.

in the above example, though, I wouldn't use a number field anyway...i would use a varchar2..

I reckon it boils down to this:

disregard the datatype(other than blob, you're right), pay more attention to the pattern of data in your field.

I assume you're talking about non-unique indexes here, used to speed up reports and such..unique indexes follow different rules.