So heat may be conducted off the edges of he circuit board, but how much is not obvious.

But the board its self isn't going to transfer any or very little heat.

 

I'm not an expert in LEDs but if the inside units don't fail and the outside ones do then the fault must lie outdoors. Temp? Moisture? Vibration?Something added to the mains line outside but not inside (another light flashing on and off or some odd miswiring)?

 

I'm not an expert in LEDs but if the inside units don't fail and the outside ones do then the fault must lie outdoors. Temp? Moisture? Vibration?Something added to the mains line outside but not inside (another light flashing on and off or some odd miswiring)?

I have not figured out the cause, but my guess goes to different production runs, or possibly a different line voltage.
I did do some measurements today and I read about 3.28 volts across most of the illuminated LEDs, and about 76.5 volts across the whole series string of 24 bulbs. So now I can power one LED at 3.28 volts and read the current and then calculate the series resistor to drop the LED voltage to about 3.1 volts. If the current is 50 Ma and I am dropping 0.2 volts then the power in the resistor will be 10 mW, 10 milliwatts times 24 LEDs, or 240 mW per string. I may be able to use a chip resistor for each string to drop about 24 x 0.2=4.8 volts. That is presuming that the supply is well regulated, which I don't know if it is.
AND, once the current is dropped a bit, hopefully the LEDs will start lasting a long time.

 

Or you could just buy good bulbs.

 

Or you could just buy good bulbs.

I have not seen good bulbs or any at all that match these bulbs. These have a crystal clear cover and a reflector behind the LEDs. Other lights have a diffusing surface of a cylindrical tube and shine light in every direction. So these are sort of specialty bulbs, it appears.