Chosun,
That was the sum of what Arek told me. From my own spectroscopy experience, I suspect Arek's previous equipment lacked a suitable or stable reference channel. Readings might be prone to fluctuations in the power of the light source over time for example which might account for some high readings on occasions. As Arek states, the relative transmission would normally be unaffected so those results would be accurate for colour. It would be wrong to interpret those older graphs for peak transmission. Those recent results do appear to be produced on a more advanced machine, but I don't know any more details.
It may be hard for those who cannot readily see binocular colour bias to understand, but to my eyes but in the dozens of Swarovskis and Zeiss binoculars I've tried it is quite clear to me that there is significant sample variation in colour balance. I don't see that any of the individual traces from either Arek of Gijs are necessesarily wrong. In fact statistically they must underrepresent the sample variation of the binoculars I've seen, let alone the thousands produced.
As for that ELSV 10x42. I wouldn't know how accurate the absolute values are, but colour wise, I would say that would be entirely consistant with individual ELSV and SLC samples I've tried which were very obviously blue compared to other Swarovski samples to hand.
David
David,
I meant to reinforce this earlier, since much of our discussion is taking place in the absence of knowledge about error margins. We don't know much of anything about production variability, inter-instrument variability, or test-to-test reliability. We also can't study any of the published distributions analytically.
It's getting boring, I know, but so far there has been no answer from transmission providers (sorry, Gijs) about publishing useful data tables. To be specific, I've attached one of several old transmission reports (of my favorite Swift 804ED) showing what I'm talking about. Unfortunately, even here the PDF format properties prevent moving the data to an Excel or SPSS compatible format — almost as if they are being purposely suppressed. (I'm not usually a conspiracy theorist, but why are the graphs made available and not the data that produced them?)
In this instance, the left and right tubes are are similar pictorially, but the data tables show numerical differences. I have others binocular reports showing greater pictorial differences between tubes. So, how different is different and by who's expert judgment does it make or not make a difference? Without being able to analyze these distributions mathematically/statistically the whole thing is up in the air. I've already mentioned this prevents being able to bridge the gap between the exit pupil and the eye's perceptual system. :-C
Ed