Dunnett's one sided adjustment

Hi everybody;
i want to test H0: µ(B)=µ(A) against H1 : µ(B)>µ(A)
the p-value for this is given by :

adj_p1=1-probmc('DUNNETT1', tValue, ., df, 6);


the second test is H0: µ(B)=µ(A) against H1 : µ(A)>µ(B)


adj_p2=1-probmc('DUNNETT1', -tValue, ., df, 6);


someone can tell me why adj_p1+adj_p2 don't equal 1.


i don't know if the p-values come from a symetrique distribution.
What
about the left bound of the distribution, is it bounded?
Thank's a lot for who giving me a help.

.