It is the same with plagioclase, which is a solid solution between albite and anorthite and is often expressed as "Abxx" or "Anxx", where xx is the molar percentage of considered end-member.
Edit: As for the range, e.g. 95-90, I'd say they analyzed several olivine grains in the same sample, and the range represents the variability of compositions they found. It could also be the analytical error on a single composition, but I doubt it would be that large, although it could be in some old papers when the precision of chemical analyses was not as good as today.