You need something that has a reasonable adhesion to the quartz/glass in order to form a good seal, it needs to be workable into rather thin foil, it should be weldable, withstand working temperatures when forming the seal and withstand temperatures needed to form the seals. Plus it should not be prohibitively expensive material. Plus it should not be that hard, it should allow (preferably elastic) deformation without needing that high pressures

The cross section needs to have high aspect ratio between width to thickness. Because the trick is to redirect the expansion into the foil thickening, instead of widening, without creating excessive forces to crack the quartz. So then the quartz walls just have to slightly flex out, to create the room for the material expansion. And the thinner the foil, the less the quartz has to flex, so the tougher seal you can get.

"Feathered edges" are there to hold it in place, so the forces related to expansion won't move it and by that lose the adhesion and form possible leak paths. Plus because the thickness is never exactly even, any shift would mean stress concentrations on some spots, sp making it more likely to crack. usually normal burrs from the foil cutting are enough, it means the tooling is explicitely designed to keep them in place and not eliminate them (as is the aim for any normal sheet/foil metal processing).

Gradual seals (so gradually transitioning from one expansion coefficient to another to distribute the expansion) is a technique that was (and I guess sometimes even is) used in lamp making. But it is quite complex process, a lot of steps, so not that much compatible with high volume production. So generally it is avoided whenever possible. And with quartz it is possible...

The problem with metallic thallium is a *significantly* higher boiling point than sodium (1473C vs 883C at 1bar). So while with sodium one ends up with relatively benign conditions in LPS lamp, with thallium fill the arctube temperature needs to be much higher. If you don't trust me, look at somewhat rare thallium spectral lamp construction - it has a quartz burner, shrouded by a metal cylinder to conserve heat, and overall just struggles to work having a very short life. Today, it is probably possible to overcome this by repurposing something like white SON arctube, but why, when existing green metal halide lamps work *so well*!

And I won't always trust atomic-spectra.net for spectral output content. Author uses NIST tables to generate his pictures, but excitation conditions for NIST tables do not necessarily match the ones in the arclamp. So while the wavelengths are of course correct, spectral balance may be way off.

Given how the LPS were fighting with retaining heat in the arctube in order to get some reasonable efficiency, something requiring way higher temperatures would be impossible at such low power densities. So I very strongly doubt.

Plus a iodine lamp needs the arc to have hot enough core to break the salt bonds into atomic metal, as only that is what generates the light (it breaks once the halogenide gas reaches the hot part of the arc, the metal atoms do the light emitting work there and it recombines back to the iodine salt gas once it leaves the hot part of the arc into the colder region).
In high pressure lamps the concentrated arc temperatures are by far sufficient, but with a low pressure the temperatures won't be enough, I'm afraid.

 I'll have to dig out my Jr. I'm gonna say some 2 part epoxy should hold the ballast can together .

So here's a Pic of it when I first got it home

https://www.lighting-gallery.net/gallery/displayimage.php?album=8301&pos=11&pid=242409

As you can see the Knuckle itself luckily isn't snapped, but the circular base of said Knuckle mount has come off the cylindrical ballast can. Looking closely you can see some kind of orange adhesive that gave way. It might be possible to reattach the Knuckle mount to the ballast, but I'm not sure if this is still the best way of going about this.

I should have linked photos earlier but here ya go

  I'll see if I can find them again . They were listed in web finds at one point . IIRC they were a good bit over priced .

 EDIT : I can't seem to find them . I'm assuming the listing expired . If I see them again I'll let you know .

I might get a couple for myself to run off my step up transformer.

Can you send me the link to those 208V self ballasted mercury vapor lamps you found in a PM?

 There's a pair of 208v Self-Ballasted Mercury lamps on eBay right now .

 Look thru this GE catalog on LampTech. https://www.scribd.com/document/113710933/GE-1975-Lamp-Catalog  There's lamps from 6w to 1,500w using all sorts of 120+ voltages from 135v to 300v using bases from Candelabra to Mogul !

 My favorite is the 10w , 250v D.C Bayonet indicator lamp ! 

 There is a 25w & 50w medium based A-19 listed as 300v. The 1st 277v listing is for a 100w A-21

Replaced an old, worn-out Universal Watt-Reducer 446 ballast that I had inside one of my garage lights, which was becoming hard to start. I replaced it with the Advance REL-2S40-SC electronic ballast because I was able to get one cheaply on eBay, and I have had good luck with that model so far in one of my other fixtures. It actually has a true rapid-start startup, which is uncommon for electronic ballasts.

Setting safety aside (assuming it's the 1940s), would it be more efficient?