The Thomas Duplex is an LPS bulb. And your (correct) DIY filtration filtered out virtually all of the blues, and almost but not quite all of the greens, while passing a usably high percentage (but not 100%) of the sodium yellow, and all of the red. This is because the Roscoe #19 is not truly a red filter. It's transmission level spikes upward just before the sodium D-line, and so passes a fair amount of it.
Click
here to see the transmission graph of the Roscoe #19 Fire filter. Note how much yellow is actually passed.
So if you were printing on graded papers that are sensitive only to blue, you would have very long safe times. My own pre-fogged tests with Kentmere Bromide #3 showed similar results to yours, at least 30 minutes safe, at which point I stopped the test.
If you are printing on VC paper that is sensitive to both blue (hard contrast) and green (soft contrast), the tiny amount of green that still makes it through the Roscoe #19 Fire filter will still fog your paper over time. But the paper will tolerate a far longer working time than without any filtration at all.
The mitigation in this case is to slide a sheet of black board into the upper vanes, then partially close those vanes down to dim the light. The problem for most who use the Thomas is that these very small blue/green spikes are visually lost in the sea of sodium yellow, so they cannot normally be seen by the naked eye, and no danger is perceived.
However, there is a way to see them. Find an old recorded CD or DVD disc, fire up the LPS bulb in a totally darkened room so the LPS output is all there is, then closely observe the light from the bulb as it is reflected from the disc. If you look carefully you will see the blues and greens as faint bands that stand out from the overwhelming yellow background.
It's also a good idea to test LEDs intended to be used as safelights in the same way. They too are not always simply pure red or pure yellow. Sometimes (often) they also emit blue and/or green spikes, and thus can also fog papers.
The mitigation for impure red LEDs is to filter their output through a single sheet of Rubylith. After doing so while using 635 nm red LEDs, my own pre-fogged tests of Ilford MGIV FB and RC papers showed safe times out to at least 60 minutes.
The expected mitigation for impure yellow LEDs would be to do the same with a sheet of Amberlith. However, I have not personally tested that combination, and I believe Amberlith may no longer be manufactured.
Ken
