Abstract
A comparison is made between linear programming and weighted least-squares approaches to absolute deviation curve fitting. The computational results show that linear programming substantially reduces solution times while requiring only a relatively small amount of additional storage.