Lumenence, or lumenousity, is a somewhat subjective standard I think. Before the advent of charge coupled devices and opto-electrical convertors, it was a lot moreso.

To get a true measurement of the total light output from a device, first you need a standard charge equivalent to compare to. Full moon brightness roughly equates to .25 lux, but it can vary. Establishing what .25 lux means is best down by converting some ambient light constant into an electrical charge that can then be precisely measured using available metering devices. Of course, any variation in color response of the sensors has to be considered, but generally this is easily done through prismatic isolation and measurement of specific wavelengths. So accounting for varying color dispersion, how does one equate flood lighting to spot lighting as a measurement of whole light output? There needs to be some means of ensuring a 100% coupling coefficient between the light source and the sensor, otherwise area dispersion variance would give a greatly unbalanced measurement of light generated.

What this means, then, is that for any given light source to be measured, either all of the light generated needs to be captured in order for the yield to be comparable between different sources, or some means of deriving an average value equivalent must be used. A light bulb's average luminousity per square meter at a given range will be less if the source transmits light in all three axes than if it is focused to some conical transmission plane. There are, of course, mathmatical solutions to compensate for such variances.

Then there is the question of which standard should be used, lux, lumens, or candlepower. I have yet to figure out how to convert one to another. I know this; when any of their values is zero, I can't see anything.

As a practical test, for spotlight sources, we have a stop sign that is about 600 yards out. If the light we are using throws enough downrange that we can see it's effect readily on the sign, then we figure it is bright, if we can't really tell any difference, then it is not so bright. If we hold the light in each others' face at night and we have to squint from the intensity, then we conclude it is bright close up, but not far away.

For flood lights, if we can read a book from the other side of the yard with it, then we say it is bright. If we have to get close enough that we can touch the light before we can read the book then it is not so bright, and if we can't read the book at all, then we say the light is broken.

As for how long the light stays bright, that is our cave test. We take a freshly charged flashlight and go as far in the cave as we can using that light till we can't see where we are going anymore. Then we know how long it will last us before we have to get out the handcrankers and find our way back out again. Another good test is to see if one of our buddies can field dress a hog at night without cutting himself. If he cuts himself, then we know the light ain't worth a dang.