Further to my last post, I found this
Highways England safety bulletin. It is interesting to note, that the failures mentioned in that refer to the canopy detaching and presumeably leaving the frame still attached to the column. However the failures I have seen have been different.
The spigot is attached to the lantern body with stainless steel cap head screws - probably M8. Stainless steel in aluminium is never a good idea due to corrosion of dis-similar metals. I found this image online showing the arrangement.
This image extracted from the Philips datasheet confirms the basic arrangement and 2 types of spigot.
Some of the failures I have seen, have been where those cap head screws have failed - or the casting that they screw into has, leaving just the spigot attached to the column.
A 3rd failure type seems to be failure of the lower frame, with that and the canopy falling to the ground. The failure seems to be in the area along the blue line. Image again from the Philips datasheet.
This results in a section of the back of the frame still attached to the column bracket, but with the canopy missing. For the canopy to be shed, the hinge pin must have also failed, either as the lantern breaks up structurally - or as the HE bulletin has found - the hinge pin is missing to start with.
The HE mitigation of tie-wrapping the canopy to the frame, will only guard against the failure mechanism in their bulletin, not the 2, I have described!
I wonder if the root cause of these failures are horizontal (up/down) oscillations of the lantern caused by the wind - or oscillation of the bracket/column, which then overstresses the rear of the lantern?