It’s quite cool to be honest, even the simple Pygmy lamp is still showing me new things in the lighting world! 

Even traditional light sources still have a few surprises 

To me it looks like the filament is touching the support wire (not only within the hoop, but a bit down as well), which is then shorting out the few mm section of the filament (between the hoop and the touching point). The thermal expansion makes that section to move, so making/breaking the connection there.

Very strange behaviour.  I think Medved’s explanation is the most likely, and wonder if there is a very fine splinter of wire there which is short-circuiting some turns.  The only other possibility I can think of is a rare phenomenon in which there is a locally increased concentration of potassium in the filament (GLS filaments are in fact a tungsten-potassium alloy).  It is known that when that happens, a small cloud of potassium can form around the filament, and that vapour is more conductive than the filament which can lead to the formation of an arc.  Usually that is destructive and the lamp fails, but I suppose its possible that it could happen on a limited scale and only short-circuit part of the filament and be ballasted by the rest of the filament.

Please bring it along to the June lamp collectors meet at Kev’s place - if the UK is open again by then I will come with a small digital microscope and then we could try to see what’s happening in more detail.

That’s exactly what I was going to suggest James!