Standard price indexes fix on a set of goods, each very precisely specified as to what it is and where sold. Measurements of the change in price of each good are made, and these combined by formula to get an overall measure of change. In a "dynamic universe", goods disappear and new ones appear, making a fixed product index difficult to construct. In cross area indexes, in which we compare the overall prices of goods in one area to those in another, there may be no common specific goods available. In both cases, indexes are constructed using hedonic regression. In hedonic regression, prices are modeled on the properties of the goods, and the properties serve as the basis of comparison from one time period to the next or one area to another. We here investigate the behavior of dynamic price indexes compared to standard indexes, and of estimators of dynamic indexes based on samples.