In their online catalog, they states on every CMH that "Product may contain small amounts of Krypton-85 or Thorium – for detailed information please refer to www.MyOSRAM.com. Product will be completely free of Krypton-85 or Thorium by Q1/2012."
I can't provide you a link for a CMH of example because of the way the surfing on thier catalog operates.
But i can provide a line to their catalog: http://catalog.myosram.com .

So we can expect the lamps to not start in total darkness ?

Probably they will change the starting gas of their CMH lamps, to argon+neon (Similar to that of the Philips HPI QMH lamps and the neon HPS retrofits for mercury ballasts).
Ar+Ne penning mixture have a lower ignition voltage than pure argon. Because of this, the Philips HPI-BUS probestart MH lamp, can directly retrofit 400W MV lamps, the rest HPI can be started with a 750V ignitor and the HPS lamps that contains it, to start directly from the mains with a mercury ballast.
Or that they will add an ignition aid.
Ash: Why do you think that there is a relation between the ignition voltage of CMH to the amount of the ambient light?

The photoelectric effect

To get current in gas or vacuum, you need either :
 - Free electrons + field to move them
 - Very strong field to create free electrons by ripping them out of the electrodes

With ambient light, photons hit the electrodes and there is allways some free electron (lots of them) in there, emitted from the photon's impact. But in total darkness, the electrons will just wait in the metal, without free electrons to complete the circuit in the gas or vacuum

This was the case with many early glow starters - The neon or argon lamp in the starter needs to strike at mains voltage. In total darkness, there is 340Vpeak across the glow lamp but there are no free electrons in the gas to carry any current

The result was, that the starters were acting like "afraid of darkness" - you flip the switch and nothing happens, as if there would be no power, though power is supplied and there i voltage up to and including the starter

I guess that one of the reasons to locate the starter behind the lamp end in many old fixtures, was so that the starter would be able to pick up the deepest feeble phosphorescence from the fluorescent lamp's phosphors, maybe several hours or days after it was lit previous time - phosphorescence that we are unable to see, but perhaps there would be some photon once in a while to provide the 1st electron to get the discharge going

And i guess that the hole in many metal can starters was exactly for this - to allow ambient light in

As that happened, manufacturers included trace amount of radioactive material in the filling gas of the glow lamp. This material provides constant mess of particles hitting everything, so would ensure presence of free electrons



To experience this effect yourself, search for a staircase light switch with a neon lamp in it, where the lamp is flickering significantly (discharge is unstable), cover it with your hands or coat etc to make it in total darkness. You will see that the less ambient light is around it, the more its behavior will become dark with only eventual flashes, while in daylight its behavior is much brighter

Another experience which might happen (but its hard to find a place like that), is a tunnel with fluorescent lights and non-radioactive starters. You flip the switch, nothing happens. You light a LED at the 1st fixture, it starts. Its flash provides light for the next one etc.. and you might see a "wave" of light going from you and into the depths

Perfect Start and Programmed Start lamps will not be affected by darkness since the starting system does not rely on discharge, and the lamp filaments are preheated which causes release of electrons by the time the lamp is supposed to start

Osram is using instead a metal strap around the leg of the arc tube nearest the cap, which is connected to the side frame at opposite potential.  Result is that a small corona discharge is established inside the arc tube capillary upon application of the ignitor pulse, rather like with a UV-enhancer.  The UV radiation it generates facililates ignition of the main arc without delay.  All of the EU manufacturers are presently transitioning to radioactive-free lamp designs without krypton-85 or Thorium, this is just the technique being used by Osram for its HCI lamps.  Rather amusingly the patent on this design says that the metal strap is intended for mechanical support to make the lamps physically stronger, but we all know that in fact it is intended for ignition without radioactive fillings!

@ Ash:

Is this why Photomultiplier tubes will be inaccurate in an environment with high radiation? (same reason with the starter)

@ Ash:

Is this why Photomultiplier tubes will be inaccurate in an environment with high radiation? (same reason with the starter)

Depend, how do you look at it. They would register the rays form the radioactivity as the signal to be multiplied. So if this is not the wanted signal to be amplified (as I assume it is your case), then yes, they are inaccurate...

Osram is using instead a metal strap around the leg of the arc tube nearest the cap, which is connected to the side frame at opposite potential.  Result is that a small corona discharge is established inside the arc tube capillary upon application of the ignitor pulse, rather like with a UV-enhancer.  The UV radiation it generates facililates ignition of the main arc without delay.  All of the EU manufacturers are presently transitioning to radioactive-free lamp designs without krypton-85 or Thorium, this is just the technique being used by Osram for its HCI lamps.  Rather amusingly the patent on this design says that the metal strap is intended for mechanical support to make the lamps physically stronger, but we all know that in fact it is intended for ignition without radioactive fillings!

James: Isn't a neon-argon penning mixture like the Philips HPI is a cheaper solution?
In ceramic arctubes, it isn't necessary to fill the outerbulb with a neon-nitrogen when using neon in the arctube like quartz arctubes.
Neon wouldn't diffuse through a hot ceramic arctube like at a hot quartz arctube.

Ash, I have at least two fluorescents right now that don't want to start in total darkness (they have argon glow starters). As far as I know, glowbottle starters in the U.S. don't have any Kr85 in them and I've had a handful that don't start well in the dark. However, I don't know why others of the same brand and year do just fine.

Also, I have two Sylvania probe start metal halides and two GE pulse-start quartz metal halides that will not ignite right away - I'm not sure if that is related to low light, but it definitely isn't completely dark in the room. Other brands start immediately, so I'm not sure why these always do that. One of the GE's even has that glowbottle exciter in the bulb, but I've never seen visible light from it.

Arcblue - how much light it does take to strike them ?

Is it on the order of below what you can even see (down to single photons), or when you can just about see the outline of big objects in the room, or more light is needed ?

And does it eventually strike on its own if left for a while in total darkness ?

In ceramic arctubes, it isn't necessary to fill the outerbulb with a neon-nitrogen when using neon in the arctube like quartz arctubes.
Neon wouldn't diffuse through a hot ceramic arctube like at a hot quartz arctube.

Neon diffuse through nearly everything. I think the PCA is even more "porous" than the quartz, so it would diffuse by faster rate. The PCA is only inert towards the sodium (unlike the quartz, what react with it and form black spots)

So if  the Neon would be used there, it would have to be filled to the outer as well. And if the Neon would be in the outer, it would have to be filled by some ionization stopping gas as well, as with the neon only the discharge would start in the outer (at hot restrike, around lead in wires,...). And filling the outer with other gasses would mean thermal bridge from the arctube, what would mean all the efficiency (converting the electricity into the radiant power) benefits of the ceramic would be gone...

Neon diffuse through nearly everything. I think the PCA is even more "porous" than the quartz, so it would diffuse by faster rate. The PCA is only inert towards the sodium (unlike the quartz, what react with it and form black spots)

So if  the Neon would be used there, it would have to be filled to the outer as well. And if the Neon would be in the outer, it would have to be filled by some ionization stopping gas as well, as with the neon only the discharge would start in the outer (at hot restrike, around lead in wires,...). And filling the outer with other gasses would mean thermal bridge from the arctube, what would mean all the efficiency (converting the electricity into the radiant power) benefits of the ceramic would be gone...

The neon based HPS retrofit lamps for mercury ballasts, haven't an atmosphere of a neon-nitrogen in their outerbulbs. They still have vaccum atmosphere like regular HPS lamps.

metal strip around burner -> like PIA lamps?

what effects can cause neon on discharge?
In fluorescent decreases efficiency.

marcopete87: the argon-neon penning mixture have a lower striking voltage than argon alone or argon-krypton85 in metal halide (MH) lamps, and than xenon in high pressure sodium (HPS) lamps. So MH lamps like the Philips HPI series of MH lamps with an argon-neon starting gas, needs only 750V to strike with a choke ballast (The Philips HPI-BUS have a starting electrode in addition to the argon-neon, allows it to start directly from mercury ballasts). HPS lamps with argon-neon gas, can strike directly from the mains with a mercury ballast and with the assist of a starting coil around the arctube.
regular MH lamps with an argon alone or argon-krypton85 and regular HPS lamps with xenon needs at least 4-5KV to ignite. Ceramic MH lamps would need several tens of KVs, without the krypton85.
A metal strip similar to the PIA antenna of the high xenon pressure HPS lamps, can be a good solution.
Neon-argon as a buffer gas in fluorescent lamps are used in the american T12 VHO fluorescent tubes, where a very high current per unit legnth of the lamp is necessary.

When mixing two gasses, you may lower the ignition voltage. Minimum is for a ratio called "Penning mixture", usually few percents of one gas in the majority of the second one. The gas combination is then steered by another requirements (heavier gasses are easier for electrodes,...)