The purchase of 5000 electric bulbs was negotiated by a company. If 100 bulbs rested at random have a mean power and standard deviation of 100 watts and 20 watts respectively, what are the 95% confidence limits for estimating the mean power of the remaining 4900 bulbs?

How large a sample of bulbs should we take in the preceding problem so that we can be 90% confident that the error of the estimate will not exceed 0.5 watts?