We can only work from
facts, not speculation. I repeat yet again

that we have little concrete evidence, except the 1881 census entry which lists James' daughter as George's niece.
There is an obvious age difference between George and James, which has already led us to propose that they might be half-brothers, with the same father but different mothers.
As we go along, we find additional clues, such as the IGI entries JAP posted today, where 2 of the sons fit very well with George and John, both in age and location. When Paul found the entry for George Robertson Ships Chandler in 1829, a father named George became an obvious possibility, and here we have a father named George. One of the children also has the middle name Moody, which George (1820) later gave to one of his sons.
In the light of this, we have to look again at the contradictory evidence of the 1881 census, and decide what is the more likely explication. I am quite prepared to accept that Margaret might be George's great-niece, not niece, but at the moment that is a
hypothesis, not a fact.
After JAP made the IGI post, we had Pam's info from her Dunbar contact, which states that John and James were brothers, and takes us back to our original idea. (However, we don't know what his source is - it could also be the 1881 census entry.)
Guesswork sometimes cuts corners, but just as often send us off on a false trail....
And Pam's Dunbar contact has confirmed a lot of what we'd already found, so I think we all deserve a pat on the back!
I certainly think we should follow up JAP's George Robertson family. If his wife Mary was quite young when they married, she might have had another child who isn't in the IGI, but we could also look for a second marriage for George 1840-1843.
Sorry if anyone thinks I'm being over-cautious, but keeping this thread on track at times feels like being on a runaway train!

Rambler
CatOne -
your post came up as I was typing!