17. There is tremendous variability among the year-classes of northern cod in their survivorship ("recruitment") and other variables, but limitations of data and modeling require that recruitment be handled as a constant in the stock assessment models, based on the calculated average recruitment of past years. However, the average of an extremely stochastic variable is not likely to approximate the true current value. Moreover, the time series of recruitment data used to derive the average is now thought to have been a period of unusually strong recruitment thus adding yet another source of overestimation to the model (Myers and Hutchings 1994). The same problems apply to highly variable growth rates (Templeman 1966) and estimates of natural mortality.