For multiple sclerosis patients we wish to estimate the mean age at which the disease was first diagnosed. We want to be 95% confident that the point estimate is within 5 years of the true mean. If the population variance is 120, how large should our sample be?