Great thread! It's quite fitting that I'm reading this while waiting for my flight out of San Jose, where I've been the last couple of days attending an Optics conference. =)

Hoppy, the 1/R dependence does make sense. For those who are familiar with electrostatics, a good comparison is the expression for the electic field for charges. It has a 1/R^2 dependence for point charges but for an infinite line charge, or for a cylinder (small radius...and if looking at a point far from the cylinder...essentially approximating the infinite line charge situation) you can prove easily with Gauss's law that it will have a 1/R drop off as you mentioned.

There will probably also be some losses from reflection at the top surface of the water...more so for rippling water (more points for the light to hit the surface off normal) although I don't know if that is significant. Another thing to consider is that if the tank is filled with water, the amount of light reaching the bottom will be more than if you had a dry tank because of total internal reflection once the tank is filled with water. The diverging light would hit the tank walls and then be by and large perfectly reflected since they will be incident above the critical angle for the glass-air interface (which is why the sides look like mirrors).

Tom, thanks a lot for the links. Sanjay's article is very well written. I think it can be very informative and useful for all the planted tank folks.