This sounds like homework but I will . Anyway assume the normal approximation to the binomial can be used (is this reasonable?) then the formula for estimating sample sizes based on a given confidence level and a given maximum error is n = z*Sqrt(p*(1-p))/e where z = the z-scores associated with the given confidence level (90% in this case) p = the proportion of successes (bruised apples in this case) e = the maximum error (4% in this case) Your problem is you don't know p since that is what you are trying to estimate. Ask yourself what value of p will make n a large as possible and then you can use this "worst case" solution. Another solution would be to estimate (roughly) the maximum value of p (10%, 20% ...) and use it to find n. Whatever you do you should read up on "Calculating sample sizes for estimating proportions" in any good basic statistics text. GB "@Home" <[EMAIL PROTECTED]> wrote in message b25s7.46471$[EMAIL PROTECTED]">news:b25s7.46471$[EMAIL PROTECTED]... > If you have a confidence level of 90% and an error estimate of 4% and don't > know the std deviation, is there a way to express the error estimate as a > fraction of a std deviation? > > > ================================================================= Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =================================================================