BM, here are my thoughts on your question to Bob.
How long is a length ratio in excess of 3.2:1 for a radiator that is supposed to be close to resonant for the CB center frequency of 27.205 mhz? I estimate it to be about 28.936' feet, based on a 3/4 wavelength being about 27.1275' feet. BTW, the radiator length for my Eznec5 model for my Sigma4 ends up at 28.6' feet long, where the value for X = -J 0.1293 ohms. This will vary a bit due to diameter, but............
If this is so, what does that suggest about radiators at 27.205 mhz that are longer even up to a wavelength as you suggest? 
I don't know how much less gain on the horizon will be realized if we exceed the 3.2:1 ratio, so maybe the resultant problem is insignificant and then I might wonder why raise the issue at all. When I read or hear folks make claims with such limiting conditions attached, I will respectfully ask,
"...how much loss or what is the magnitude of this limiting affect?"
I wonder if the lost gain noted in the article is significant enough so as one can tell from operating our radio, or is it so small that it takes a Spectrum Analyzer on a certified Antenna Test Range to tell the difference?
If not, then how else do we explain the fact that both Bob and Shockwave noted improved gains, over longer distances? I think they both used lengths that exceeded this limiting ratio (3.2:1) by what seems to me to be considerable amount?
I recently got a 72" piece of 1/2" aluminum that is about twice as long as the tip for my Sigma4. I think now I can make my antenna at least 3' longer and see what happens.
On the models below, I added 36" length to the tip of my Eznec5 model for my Sigma4, and it increased the mis-match by about 50%. Even so, I think the gamma should take care of that easily. As far as gain is concerned, it dropped very little and the angle did not change at all. I did however, notice that the currents flowing on the mast increased noticeably, and that may be a problem to consider. I see little to no advantage in adding length to the Sigma4...based on this model, but I hope to test the idea in a real world test real soon.
I also noticed little to no disadvantage in making the antenna shorter either. I did not change the length of the radials however, like Sirio has done with their New Vector 4000. I did not analyze the changes in match here, so maybe that is why Sirio made the radials longer as they made the radiator shorter. Personally, I think the only reason Sirio did this was to try and overcome some of the problems with ice and wind on this very tall radiator while not sacrificing any performance.
How say you?
Notice how the current increased on the image with the longer top element.
View attachment Current differences when making S4 longer..pdf
