To the excellent list of reasons to go MH that George Booth and David
Whittaker posted, I'd just add something implicit in both: you can't
really get the intensity out of fluorescent that you can with MH. Yes, you
can get lots of lumens/watt from fluorescent, but it comes in a big linear
package, and you don't get many lumens/lamp. Thus for higher intensities
MH ends up cheaper on lamp costs (just don't buy the $100 lamps they sell
to reef keepers), and physically much more practical. Just don't build the
luminaire out of wood.
Some clarifications: the lumens I quote are all mean lumens over rated
lifespan (except for incandescent lamps). Thus, some of those lamps do
give you high efficiencies at 8000 hours (the T8s have a rated life of
20000 hours, the biaxials 15000 hours, both on the basis of 3-hour starts
on EM ballasts). And the high-pressure sodium lamps I quote are all
high-CRI, low efficiency lamps, on the assumption that we're interested in
looking at our tanks; the CRI 22 lamps used for street lighting (and grow
lamps) are much more efficient.
Now if we could just get the biologists to agree on the spectral
proclivities of our plants, we could (using our *enormous* market clout)
convince the lamp manufacturers to rate their lamps with a Plant Happiness
Index, or invent the plant equivalent of the lumen--the phumen, perhaps?