The following was supposed to have been sent to Seesat-L on 23 September, but somehow, I managed to mail it to myself! In my post regarding my method to estimate the prediction uncertainty in satellite ephemerides, I made the following erroneous statement: >The constant 876 is simply the number of seconds in day, divided by >100. It enables UNC to be expressed in seconds of time, and converts >%UNC into a fraction. Russell Eberst and Mike McCants have kindly pointed out that 876 is not the number of seconds in a day divided by one hundred. The correct answer of course is 864, so my formula should be: UNC = 864 * %UNC * ndot/2 * dt^2 -------------------------- n0 + 2 * ndot/2 * dt The 2 percent error in the original does not invalidate the basic findings in my article, and it does prove what I said a little later in my message: >I will leave the derivation up to the reader. It involves only high >school math, and I suspect many of us who are long out of school >can use the practice. Yes indeed! Ted Molczan