> Do you have any reference you can point me to ?

Unfortunately no. The graph is just in a parts catalogue, don't
know where they got it from. I would expect any bulb mftr has
something similar, I didn't need to look after I found that one

> If it is the switch on surge that is the predominant failure
> mechanism, then instead of a resistor have you considered
> using a NTC (Negative Temperature Coefficient) Thermistor.

Yes I have considered a thermistor, and will try them. It happened
that I conveniently had enough Rs in the spares box to do a test.
Although it works I'm not over the moon about having a hot resistor,
but 63C isn't an excessive difficult temperature to insulate other
materials from. From what I (think I) can deduce from the pattern
of bulb failures around the house, it's that those that are on for
most of the time are the ones that fail most often at turn-on. This
would be because the filaments are weakened by working hours ?

Is it not necessarily the number of times a bulb is turned on/off (if
you ignore the chance of a peak hitting the cold filament at turn-on)
but rather the degradation of the filament that makes it more likely
to fail ?

> The following deals with the subject of using NTC Thermistors
> for In-Rush Current Limiting

> http://www.epcos.com/inf/50/ap/ntc.pdf

That is very helpful thank you. I don't like fiddling with mains and
every little helps. Cheers

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics