Lets assume that you have a table of the cumulative standard normal, which
gives the area to the lef of z-values from 0 upwards, so these probabilities
are all greater than 0.5.
In a.) the probability is less than 0.5, so the z-score must be negative, so
we have to do an inverse look up for 1-0.2119=0.7881 then switch the sign
of the corresponding z-score. The z-score giving a p-value of 0.7881 is 0.8,
so the z-score we seek is -0.8.