The LI-LII screening doublet separation in the relativistic approximation up to Z20 has been compared with that obtained experimentally by the transitions MIVLII (Lβ1) and MIVLI (Lβ10). The experimental and theoretical values deviate considerably. If one retains terms up to Z4, an apparently close fit with the experimental values is obtained. It is shown that such a fit disappears on applying the field theoretical corrections and this stresses the need of retaining the terms up to Z20. The justification of neglecting the still higher powers of Z lies in the smallness of their magnitude when compared to the experimental errors. The field theoretical corrections modify the relativistic curve that contains terms up to Z20 in such a way that one again gets a good fit with the experimental values. This shows how the various field theoretical corrections and the contributions from relativistic terms with powers of Z greater than four compensate each other.