After a week of pretty good success on the windows, today was cold with light snow, and the bird feeder was very active. I realized last week that one of many variables I need to consider when I record bird-strikes-per-hour is bird activity, but I haven't tried to record that yet other than excluding long periods when no birds were present. Obviously, if there are few birds visiting, there are fewer chances for collisions. At the other extreme, like today, when the feeder gets crowded there seem to be birds in the air all the time - hovering and jostling for a perch. In these conditions there are many more chances for collisions.
So today I started observing and recorded 6 window strikes in 1 hour 40 minutes - about one every 16 minutes. Thinking that this was happening because of the faded highlighter I reapplied the lines, and in the next 3 hours 30 minutes there were 8 strikes, or about one every 33 minutes - about half as frequent but still pretty distressing and not really a significant improvement. However (one can always find a hopeful angle) five times during that period birds flew in and landed scratching at the windowsill. Maybe these were birds that would have crashed into the glass, but recognized the danger at the last second and fluttered harmlessly down to the windowsill.
A physics lesson
Some of my assumptions about UV light and what birds perceive were wrong, and I've been educated by some posts on the NikonGear UV photography forum, and some follow-up research. What we see under the black light is fluorescence. Certain compounds are "excited" as they absorb UV energy and then - when electrons return to their normal state - they fluoresce: releasing energy as light in the visible range. So this is why day-glo colors appear intensely bright in sunlight - UV energy from the sun causes fluorescence, and the colors actually glow.
Therefore my experiment with the black light has very little to do with UV other than as the light source. The black light is good for finding fluorescent things, and fluorescence begins with UV absorption, but many things absorb or reflect UV without fluorescing and we simply can't see it. A search for UV-visible substances to mark windows would require much more specialized equipment. You can see some really cool examples of UV photography here.
The fluorescent highlighter should still be visible to birds, because the fluorescent ink will simultaneously absorb UV and release visible light. But the visible fluorescence in daylight may be about as obvious to the birds as it is to us (i.e. not very). So if the highlighter works it may work largely because it absorbs UV, or maybe the combination of absorbing UV and releasing visible light is more obvious to the birds. Either way, that process requires a UV light source, and another issue that I think might explain today's poor results (still searching for that hopeful angle) is that UV wavelengths are blocked by clouds. Apparently "thin" clouds allow about 60-80% of UVA (the longest wavelengths of UV and the range that birds can see) to pass through, but thicker clouds block most UVA. So with little or no UV, such as during today's snowstorm, the highlighter marks would neither absorb UV nor release fluorescence, and the birds' vision would rely on the same visible spectrum that we see. If this is true then all efforts to use UV-related markings on glass will only have limited effectiveness (but don't take my word for it). So I'll have to note weather conditions in the future as I record window strikes to see if the highlighter is more effective on sunny days.
By the way, window glass blocks some shorter UV wavelengths, but at least 90% of UVA apparently does pass through ordinary glass.
So what does this mean?
Well, clearly the highlighter is not the "magic bullet" that everyone is hoping for. I'm still anxious to hear from others who have tried it. So far I have two responses indicating that it works, and I'd like to hear more, but if anyone has tried it without success I'm especially interested to hear about that.
If it does work, it will presumably work best in sunlight, and worst in low light or on overcast days, and unfortunately bird feeders are generally most active at those times.
As I said previously, we can keep trying to find ways to make the reflection in the glass look unattractive to the birds, but there will always be things (like hawks) that look even less attractive and cause birds to try to escape through the window. To truly prevent window strikes will require a barrier such as the BirdScreen.