This is beginning to worry me!

I have a 26W SOX lamp installed in a street lantern, the lantern is at a slight tilt so that it doesn't look like it's falling off the brcaket. This tilt also means that there's no chance of sodium running down to the electrodes, getting behind them and causing the lamp to fail.

For some reason lots of sodium has managed to get behind the electrodes and I can't work out why! The lamp has never been moved when it's warm and the sodium is in a liquid state so that's ruled that one out. As I've said it's tilted with the cap up higher than the other end so that's ruled out any chance of it running down to the electrode end.

I see there's a gap in the oxide coating at the electrode end, presumably so it doesn't come into contact with the getter and create even more of a problem if the getter is in contact with one of the wires going to the arc tube. I was thinking maybe it's running cooler behind the electrodes which is why the sodium it getting behind there.

So far the glass covering the electrode wires is keeping it from destroying the seals but it's not far away from touching the seals and killing the lamp.

The only solution I've thought of (which I'm testing now) is to wrap aluminium foil around the electrode end of the lamp to try and keep the heat in and force the sodium to migrate away from the electrodes.

Do you guys here have any suggestions to correct this problem or know why this would happen in the first place?

Only one idea:
Didn't it experienced any vibrations hen lit?

No, I really don't know how it's happened.

I've left it on with the foil wrapped around the end where the electrodes are and I just had a look at it and quite a bit of the sodium has gone, looks as though I should be okay!

If you have the option, try letting it run vertical for a while.  Easiest way to get heat behind the electrodes.

If the foil helped, there is some thermal bridge (may be wrong IR reflector composition,...). But that mean, then the lamp is in fact defective, such bridge should not be there...

If a sodium run-down defect like this was occasionally seen during production at GE, we would burn the lamp for half an hour on the ageing racks, then take it out and hold it VBU in one hand and fairly firmly knock the cap end of the lamp against the palm of the other hand.  This was generally enough to encourage the sodium to move down to the bend by gravity.  Afterwards the operator would manually re-distribute the sodium to the correct position, going evenly about 2/3 of the way down each leg.  This method is quite effective but works only if you are quick.  If the sodium has stayed too long behind the electrodes it will begin to react with the glass and then sticks to it so hard that you will never remove it.

The seals on modern lamps are pretty strong though.  Although its better if sodium doesn't get there, in many cases it will not actually cause premature failure.