I don't think this quite true. Correct me if I am wrong. A laser directed directly into the water would not have any reflected light. The angle if incidence is to small. If we assume an angle of incidence of 30 degrees (anyone know the real number) and ideal cone of light (equal distribution equal spread, not true but). Then a 60 degree lens would have not loss light since all light striking the surface is less than 30 degrees. Where as the 125 degree bare LED would. If my number above were correct the 125 has about a 7 inch radius the 60 about 3. Lets assume 1 par per square inch The light entering would be related to pir^2 or 3.14*3*3 = 28.3 the total would be 3.14*7*7=153.9. More than 80% is reflected. Now there are a lot of false physic statements here to try and make explaining it easier, but I believe a lens will have less reflected light from the surface of the water.
Ha! I really had to laugh. One of the best comebacks I've read! I wasn't planing on using lasers, were you? Any surface would have some reflective qualities and, yes, even a laser would be reflected, after all, that is what a laser show is all about, reflection and refraction. With such a pure light source, calculation of reflective surfaces would probably be easier, come to think of it. Not really my area. In principle, if there is no angle of incidence, <90°, laser light should not be reflected by a transparent substance. In the real world, I can't see it happening. We are not perfect enough. Then, water is a flowing surface, which means reflection can't be avoided, but may be possible to equalize (+ and - angles of incidence), still, one is splitting hairs.
As you noted, if the light arrives at the surface of the lens (optic), with a much smaller angle than the lens, then the light cannot be appreciably further bundled (ok, some of this would have to do with the lens type and quality). This wouldn't be the case for our tanks, would it? But seriously, light being bundled by an optic will always loose some of its intensity, if you want to split hairs, simply due to the reality of the materials involved. Then the loss through the angle would be additional. Both of these factors are what manufactures list as transmission efficiency. After this loss, you will still have the loss associated with entering the water and refracting/reflecting from the glass.
Again, we come back to what are we attempting with our leds. Illuminate an aquarium as efficiently as possible...plus a bit of aesthetic. Unless you are aiming for columns of light dancing upon the rockery, I think we can chuck the lenses and imitate a single light source by clustering the leds. Of course, spotlighting special corals would be cool, although the coral may not like it. I find it very easy to over-light (photoinhibition) corals with leds. Hmmm. Where does this leave us?
I'm off to read another good book
