How to deal with a bifactor solution with negative loadings on group factors

#1
I´m a little confused: I tried to calculate a bifactor solution and afterwards omega like in e.g. "The Rediscovery of Bifactor Measurement Models; Multivariate Behavioral Research, Volume 47, Issue 5, 2012" http://www.tandfonline.com/doi/abs/10.1080/00273171.2012.715555

Now some of the variables show negative loadings on the group factor.
To be honest I´m not sure about two things: 1) How to interpret these results? I mean on the general factor everything is fine, and these are only a few items... 2) How to calculate omega with this: Do you take the absolute values or do you leave the items out or...???

Any help would very much appreciated, best,

Pete