The small sample of 16 suggests a t-test with 15 degrees of freedom for a 95% confidence interval (2-tailed). From tables, the t value is about 2.13. The pop SD needs to be reduced because of the sample size to 20/√16=20/4=5, and we need to find X where 2.13=(X-485)/5, so X-485=10.65ml. This I interpret to mean that for 95% confidence, if the dispenser outputs more than 10.65ml away from the mean then their decision is unreasonable. 15ml is greater than this, so their decision is unreasonable.