I have forgotten about this article after reading the copy some months back. At that stage I asked whether some follow up could be done, as 2004 is a long time ago. Is anybody in touch with Peter Russell?
Some things that stand out for me
1) Terry Martin mentions in the latter part that a cobalt is darker than a SF violet. I don't quite agree with this statement.
What could possibly cause a violet bird to produce a new dark blue bird, which then again produce a violet bird?
2) The blue hen producing the 4 eggs in 2002 could have been blue misty. Misty could most certainly slip through unnoticed, esp. the type that one only notices in a net or with chicks in the nest. It has been reported that a violet misty looks very cobalt like, and a violet misty would again produce the violet phenotype, the odd dark bird (violet misty) and blue and blue misty (which we'll have to assume goes through as the same thing for this idea to stand ground). Thus, there would exist only a very slight change in the blue birds, and speaking of experience, I often don't pay a lot of attention to plain blue offspring in the same aviary as rare mutations (like violets 10 years back).
3) As Molossus suggested, the NT mutation could also play a role, especially if the first bird was actually a combination between violet and the NT birds. I am not that convinced though, because that combination should probably be a much more purple looking bird than a SF violet. And, the breeding results of the new dark blue hen that again produced a normal SF violet doesn't add up here. Furthermore, if the first bird carried both violet and NT, then the blue offspring indicates non-allelic interaction. Granted, I haven't seen convincing evidence to this regard (yet).
4) As suggested earlier in the topic, perhaps a modifying gene isn't out of the question.