Not incidentally, this is a dried bud from the Sickmeds Green Crack S1 from my last picture post, above.
Obviously, its got some purple in there, and there is a reason I'm bringing up this example (well. . .other than showing off a cool bud!).
The original clone-only line Green Crack does not typically go purple. On the other hand, most accounts I've seen of this S1 (which is just Green Crack [reversed] x Green Crack) say that they're seeing purple.
Green Crack itself is supposed to be Skunk #1 x "unknown" California indica. I've speculated in my grow report of this that because of the purpleness of this S1 offspring, the "unknown" CA indica is probably one of the Norcal "purple lines" (ie Mendo purps).
I'm not going to expand this post too much with complicated genetic descriptions, but here is the point:
If most of the S1 offspring here turn purple (mine did. . .and that seems to be most people's experience with this) then whatever genetics are controlling the purple color CAN'T be controlled by a single simple dominant/recessive Mendelian gene.
The reason is if there were only one purple color "gene" and this gene were truly dominant, then the original Green Crack should be purple. . .but it isn't.
If there were only one purple color "gene" and it were truly recessive then only 1/4 of the S1 offspring should be purple. . .but that doesn't seem to be true either since most growers are seeing purple.
So, assuming the reports are right, in this case, the purple color trait is probably controlled by multiple genes. To really figure out what's going on here is going to require a lot more information, and without it, I don't think you can reliably predict how the offspring of different crosses will or won't go purple.
Bottom line is, if you're crossing unrelated purple and non-purple strains, I don't think you can apply a hard-and-fast rule to predict the color traits of the offspring. You have to do the crosses and see.