E-Values less than or equal to E0 without units are always kind of seen the same way as percentages, i.e. as a certain proportion of a whole. So, e.g. E-2 is not merely seen as the order-of-magnitude of 10^-2, but as that proportion of 1. This ends up being very convenient for representing values nearly-but-not-quite 1. So if I have on the order of a 99.9% chance of getting rained on today, then that's an E-3^C probability since it is the complement left over after removing an E-3 proportion.