The "algorithm" isn't strictly linear...it figures on reported numbers of 2000 the first 6 hours, 1000 more over the next 24 hours, and the final 800 over the next week... So it's linear for each of those subsets of time... I don't know that a strict exponential algorithm would really be any more accurate...not that I'm calling what I have accurate. I think I'm going to do another estimate column based on the ratios of our sample data and see what that looks like... h34r: