Chebyshev expansions for the error and related functions

New 30 decimal place Chebyshev expansions, which can be used to obtain approximations up to that accuracy, for the error function and its complement are presented. Bilinear and biquadratic mappings of the independent variable have been employed, in one case to improve convergence and in the other to increase the basic approximation range. The practical effects of these mappings are discussed, and the method used to generate the tabulated expansions is outlined. The expansions here presented form the basis of the routines in the S15 chapter of the NAG library. Introduction. Clenshaw [1] presents Chebyshev expansions, accurate to twenty decimal places, 20D, for the error function, erf(x). These expansions are of the form* erf(x) =x 1> arTr(t), IxlS4, I =2( )2 -1