If a restaurant owner wants to find the 95% confidence interval around the true mean cost of a pizza, how large of a sample should be studied if the standard deviation from a previous study was 26 cents and he wants his estimate to be accurate to within 15 cents?