A new israeli member of LG, said me that 70W HPS with internal ignitor (Ie. Osram NAV-I, Philips SON-I) can be operated directly from 80W mercury ballasts, despite that they aren't HPS retrofits for mercury ballast.
This means that every 80W mercury lantern that i saw in Israel, that upgraded to HPS (Ie. Globe lanterns, GEC Z8691 and Z8897 lanterns), its 70W HPS with internal glow starter actually operates from the lanterns original 80W mercury ballasts.
How this can be happen???

How can that happen? Easy, someone remove the MV and screw in the selfstarting HPS...

Maybe you wanted to ask "How can that work?"
The 70W HPS/MH ballast is an ~680mH inductor rated for 1Arms, while 80W MV ballast is ~790mH inductor rated for 0.8Arms.
So if you connect the 77V lamp designed for the first onto the second, you get ~0.86A (so ~60W) if the inductance does not change.
As the MV ballast would operate at higher current then rated, it may start to saturate, what would rise the current a bit.
In similar way, the a bil lower current would cause the lamp pressure to be lower a bit, so operate with a bit lower arc voltage.
So bottom line, the 80WMV ballast would operate the 70WHPS lamp at about 60W, what is for most of them still OK.
The ballast would be in a bit worse situation, it's power dissipation would rise by ~15..20%, so it may fail a bit sooner. But 80WMV ballast used to have quite margin there and seconds, the mounting style vary the heat transfer way more, so if decently mounted, it would not be problem as well...

@Medved: So the 70W internal ignitor HPS lamp will be underdriven with the 80W MV ballast?
I will perform a spectral analysis of the 70W internal ignitor HPS lamps in a globe or GEC and AEG former 80W mercury lanterns with my CD-R, and check if the width of the sodium D line is the same as a HPS lamp that operated with its HPS gear.

If brightness is a clue, the lamp does not look too dim or anything



I think what we have here are clones and not real GECs :

The fixture without ballast (that needs a ballast box) is slimmer than the GEC, and the top is simply round, without the angle

The fixture with ballast has slightly different shape of the ballast part - here it is rectangular, in the GEC it is thinner towards the bracket



Some fixtures are converted from 125 MV, in them the ballast is replaced (not allways, i seen a 70 HPS screwed as is in 125 MV fixture, it was really bright and cycled, and died quite fast)

The brightness, the color and the spectra of HPS lamp, plays a large rule.
If the color of the HPS lamp is too orange and dim, the lamp is underdriven, and if the color is bright and yellowish or golden white, the lamp is overdriven.
Check the color of 70W HPS with internal glow starter operated on 80W MV ballast.
I think that the GEC Z8691 and Z8897 lanterns we have are real products and not clones.
There is a local made lantern however that i 100% know that it is a clone of an antique france lantern: The Gaash Oron lantern, which is a clone of the Vulkanlux ASL 150 lantern.

Underdrive by 15 percent would not make huge difference, 10% mains variation would make even more...
The difference would be hidden compare to shifts caused by aging, IR reflection management inside the optics, lamp-to-lamp and mainly maker-to-maker tolerances.
Here many 150W HPS are dimmed (around midnight) to ~70% power and you need quite accurate measurement to tell the difference, so i doubt the 15% reduction would be visible by naked eye, moreover when you do not have any "reference" fixtures (I guess such "adoption" is made on all fixtures of the same type)...