I was playing with a super worn Sylvania that I could see the arc still (the bulb and oxide coating dulled some of the lite down) and I saw in the green arcstream a few little orange balls suspended within the arc. It was almost like an atomic projection, two were floating around while the other swirled across the arc. Is this just sputtering from startup or have I discovered a quark MV lamp?

The mess sputtered from the electrodes becomes noticeable part of the arc plasma, so radiates according tho it's composition...

I used to wonder why MV lamps would 'flare up' when warming up or when bumped. It wasn't until I used a welder's glass to see the arc stream that I realized that minute particles were falling from the uppermost electrode and glowing brightly as they fell through the arc stream. Sometimes they would even be suspended for a few seconds before dropping or disappearing. I figured it was particles of loose emitter materials.

Ah so that must explain the pink when you smack it or even in the warmup process.

Interesting.  I wonder if this is why MH lamps tend to extinguish if bumped/moved while lit?

One I've been wondering about for years now is why the krypton-filled energy saver fluorescents (34w F40T12, 60w energy saver F96T12, etc) striate when cold.  I've always wanted to somehow remove the phosphor from a lamp (I did start once with a GE Watt-Miser whicb had/has a piece of loose glass inside but worked fine) and got a fair amount but not enough to see the arc-stream. 

The MH extinction comes from another reason: The arc heats up the gas, the heated gas expands, so becomes lighter than the surrounding. Because of the gravity acceleration, the less dense gas rises against the rest, forming a gas flow, bowing the arc upwards, so against the acceleration vector. That is well known, so obvious to anyone at least somehow familiar with high pressure discharges.

What is not that obvious: Any bump is in fact a surge of very high acceleration. And the same as gravity, this acceleration moves the gasses of different density in similar manner. Because the bump may easily be way greater in it's peak acceleration level than the gravity, it yields way faster gas flows within the arctube. And this faster gas flow then bows the arc in a way greater extend and even quite likely just blows it off, so the arc extinguish as a result.

By the way this is one of the factors, why a simple mirror optics never catch up with HID automotive lighting: The vibration cause the arc to move wildly around it's position and that would make the beam shaking in the same way. The projection technique with dedicated rigid shield was the only viable way around the problem, even though the vibration still does modulate the light intensity. But at least it does not cause glare...