Looking around at all the cheap surplus beam splitter cubes out there. I've found no description of how things would be expected to behave outside of the spectral bands for which they're designed. Leaving out the transmissivity of the glass itself, should I assume that when relative transmission of the hypotenuse coating goes to zero, relative reflection goes to 100% and vice-versa? Are there any common constructions where (aside from the glass) absorption climbs instead? Eg, looking at typical visible spectrum dielectric beam splitters, the transmissivity curve appears to drop off outside the design band (appears to... no one provides data, so I'm extrapolating from the slope at the edges). Should I assume that as transmission drops, reflection correspondingly increases? Or is the the difference made up substantially in increasing internal losses? Again, this is leaving out the glass itself and any antireflection coating-- that data is easy to find. Thanks!