This is common with new colour lights but the semaphore is probably heavier so it would have needed a new and stronger design of support, and the cost of getting that done and approved would far outweigh the saving when it would only ever be applied to a handful of signals. The operating wire and balance weight also complicate things.
According to the Kitchenside and Williams book I used to have, it was because the driver might see a green colour light and assume it meant the same as a normal green colour light, that the train was safe to accelerate to line speed and if the next signal was red it would have a yellow (or a distant at caution) at braking distance before it.
Were they really that recent? I understood that they died out in the first half of the century when colour light signals became the obvious choice for lines where traffic was too dense and signal spacing too short for each stop signal to have its own separate distant. The three-position ones were upper quadrant to distinguish the caution aspect from the proceed aspect of a normal stop signal, and getting rid of them allowed most companies to standardize on two-position upper quadrants.
Thanks for those dates. I was given that book at the age of about 12 but haven't seen it for several decades, so please forgive me for not remembering chapter and verse 40-odd years on! Railsigns mentions that three-positions were motor worked. I imagine a wire pull would have been a bit like a derailleur gear, difficult to adjust to get the middle position reliably, and with the risk of thermal effects etc causing caution to be displayed as clear. Which makes them all the more pointless in my view, eliminating one of the few advantages of the semaphore over a colour light, that it doesn't need an electric supply.
And there was a much smaller (A5-ish) volume back in the 60s with a red cover, which might be what edwin_m refers to (I lent mine to someone....).
I think I had two, the blue and the green. The dates looks about right.
