Light bulbs 1 and 2 with their reflections A and B in a CD, Theuns Jordaan’s Koue Vuur in this case, though it shouldn’t really matter. 
My guess: Globe 1 goes with reflection B.
Quite right … how did you arrive?
Rigil
[spoiler]http://www.thenakedscientists.com/HTML/content/kitchenscience/exp/colours-in-cds/[/spoiler]
I arrived at the right answer via a quite wrong route. I knew that energy saving light bulbs tended to have only discreet wavelengths of light, but I didn’t notice the effect in the photo, for a reason that neatly illustrates how one’s subconscious biases can bedevil thought and indeed even observation. It’s a long story, so bear with me.
I am interested in, among other things, art. On the art message boards I hang out on, there are sometimes discussions about the merits of various light bulbs to use in the studio, or to use to illuminate art in museums. Many artists feel that energy saving light bulbs are no good, because they don’t have the entire spectrum, and thus if you use them in your studio, you may not see the true colours of your paints. I have never experimented with this; I assumed that with some energy saving light bulbs, perhaps entire colours are missing from their spectrum. And this is what was sort of non-explicitly stuck in my head, so I first checked the two spectra to see if any colours are obviously missing in one. I couldn’t see this in your photos, and thus I never paid careful attention to the level of discreetness of the colour bands. It was right in front me, yet I did not notice it because I wasn’t looking for it.
Instead I then looked at the shape of the reflections. I noticed that in one, the reflections were light bulb shaped. In the two light bulbs, the energy saving one is opaque, and therefore its reflection would have a clearly defined bulb shape, whereas in the incandescent bulb, the bulb is transparent, and presumably, with an unclear reflection off a CD, its shape would be too complex to be clearly visible because you would be looking at the glowing tungsten wire itself, plus perhaps bits of the glass bulb. Or at least, that was my argument. It is not quite what your web page says! 
In fact, I can’t quite follow the page, especially its explanation of why the colours are split in the first place.
I have seen a page on the web where they explain how you can use an old CD to make a spectroscope that is good enough to actually show absorption lines in sunlight. Pretty neat!
This may very well be. When colour-matching high end industrial paints to their standards (basically master swatches), the visual comparison under different light sources serves as valuable check for metamerism. Two paints are said to be metameric if they look the same under, say, sunlight but are different under fluorescent light. This usually points to the use of suboptimal pigmentation blends. That is, there may be several combinations of pigments that will match a colour under incandescent light alone, but only a few that will match under both incandescent and fluorescent.
Anyway, the point is that the colour is demonstrably dependent on the type of light falling on it. So I won’t be surprised if an artist with good memory for colour finds that there is something amiss when he looks at his work under a different light source.
For the same reason, if you are buying a car with painted bumpers, make sure that you compare the body with the bumpers not only under the dealership’s in store lighting, but also outside in the sun. If you really want to nit pick, have a look at it under a halogen streetlamp at night as well.
Rigil
I doubt if I’ll ever be able to afford a car. In the days when I could still afford red meat, I learned to take meats out of the supermarket fridge to inspect it - they often have reddish lights in the fridges that make the meat look more red and fresh.
A long time ago I was looking for a (cheap) way to erase an EPROM chip, which is done using UV light.
After a bit of research I discovered that specific fluorescent lights used for disinfecting water can be used, because they act by emitting dangerous (to bacteria and anything else living, including humans) levels of the correct wavelengths of UV light.
Then after a bit more digging I discovered something I find slightly frightening: The “dangerous” fluorescent light used to disinfect water is actually just a normal fluorescent bulb with a clear glass tube instead of a “white” glass tube. Turns out fluorescent lights all create dangerous levels of UV light inside, but the ones we use for lighting are coated on the inside with fluorescent powder. This absorbs the UV light and re-emits it as harmless visible white light. If it weren’t for that, you wouldn’t see much, but you’d get hectic sunburn almost instantly.
As such, I expected the CD with the broken spectrum to be the one from a fluorescent light, since I’d expect a material that is absorbing and re-emitting light to create a distinctive spectral signature.