ABSTRACT
******************************In this chapter we take up a number of issues connected with the estimation and control of the error. For this purpose we must go beyond the simple convergence results of Chapter 5, which merely tell us the order of the error, and actually approximate the error. This hardly seems possible without being much more specific about the methods used, how the integration is started, and how the step sizes are selected. Numerical experiments encourage us to investigate the behavior of the error because they show that it can be quite regular:
