As far as the variability of the source, the transmissions should be calculated relative to the source, negating any contribution of non-uniformity in the source. However, you can see in the blue there is not much signal, and the SNR starts to get a bit bad there.

Any body have standard filter factors for 47 and 58 filters? I'll play around tonight with my light meter to calculate them, but they should point to the differences in overall magnitude amongst the filters.