Here is the data for first two crosses:

No changes from the first 4-6 tests. The most striking is the absence of learning for the control. It does not make sense. I also controlled that the genotypes were not inverted by looking at GFP fluorescence which is much weaker in the test line (due to the tubGal80ts presence) than in the control line. Could it be an effect of GFP presence in some neurons? The CS strain should be unguilty, since the CSx c380 had no phenotype in the previous cross…
A third cross is coming, with the elavGal4 control, maybe it is only bad luck… But the results were so reproductible so far, and so solid… it is strange. I got the parental lines out of the stock to get fresh flies for the next cross.
I should also cross new flies now. I am planning these groups (male-female):
UGFP;;d42gal4,chaGal80 x tubGal80ts,U-PKCi
CS x tubGal80ts,U-PKCi
UGFP;;d42gal4,chaGal80 x CS
UGFP,c380;;,chaGal80 x tubGal80ts,U-PKCi