Some flowers have patterns in ultra-violet light that are invisible to humans. The insects that pollinate these flowers can see the patterns as a different colour than the rest of the flower which is 'unmarked'. The colours have evolved for the insects and not for us, so the UV component is essential to see the 'true' identity of each flower.
Equipment is listed in the EXIF info. The photographs are all taken in natural sunlight, with B+W (Schneider) BG38 and 403 filters to remove infra-red and visible light respectively. Images taken with only the 403 (UV-pass) filter are so contaminated with infra-red that they are practically useless. I have since also discovered that it does not help to replace the low-pass filter with clear glass, as the boost to infrared is too great for the BG38 filter to cope with, and the boost to UV sensitivity is minimal. I solved this problem by the acquisition of a Baader U-filter. What a nice little piece of glass this is! The alternative solution is to revert to the use of two BG38 filters, but having not tested it I cannot comment on its merits.
A tripod is essential, and bellows/extension rings are useful if the subject is in the macro range, like most flowers.
Bjørn Rørslett has in many ways pioneered digital UV flower photography. He has his own vision, which has an artistic foundation, and his methods reflect that. Unlike Bjørn, I take a forensic approach, preferring to view flowers in the way that insects see them. So lately I have started to make YBU composites that is more or less indicative of how flowers appear to insects. My technique is simple. I take two photographs, one in visible light only, and one in UV only. Then I move the channels in Photoshop so that yellow (R+G) goes into the red channel, blue into the green, and finally the UV image converted to greyscale is copied into the blue channel. Because the two exposures differ in magnification (because of the UV refraction change) the one or the other needs to be resized. Then the UV image when copied into the blue channel must be moved so that it fits exactly. If not, the edges will be clearly visible and disruptive.
Some of these and other photographs will appear in the March 2008 issue of Veld & Flora, quarterly of the Botanical Society of South Africa. There I give a much more detailed description of things. If you are interested in more info, leave a request in my guestbook, and I will respond as soon as I get the chance. Please respect the fact that I am a working professional with many other commitments, so my reply may not be immediate. Also bear in mind that this is hard work based on much personal trial and error and financial expenditure. If your results do not come easily, don't be discouraged. Keep trying until you get it right.