In my experience cracking in the stops result from improper termination; not filling or (with thicker sections) back-stepping. Cracks in the starts could result from inadequate cleaning, lower heat input, and the fact that the base metal is not being warmed by continual welding, as you are placing intermittent welds, and probably staggering them at that. Aluminum dissipates heat pretty rapidly, and when the "almost a metal" (love that) is cold, starts tend to be cold, resulting in all sorts of discontinuities; especially where the joint was not cleaned adequately.
Not only that, but what about the cross sectional thickness of the joint, and was soak time sufficient to get the entire joint and HAZ into solution? Did the PWHT process use a stress relief cycle prior to the artificial age? The shape of the weldment could be playing a role, also, as heat inputs can cause stresses to work on differing thicknesses and geometric attributes.