This paper deals with the optimization problem of calibrating the periodic errors of Lacoste-Romberg model "G" Land gravity meter by weight method. Minimization of the trace of the variance-covariance matrix of adjusted parameters,∑x
, is adopted as the criterion for the optimization. It is presented that the optimum calibrating weight and observational distribution for detecting the periodic errors with one wave-length. For several wave-lengths, the author derives the general principles of the optimization design for the calibration and corresponding practical procedure from simulative computations. The problem of the required number of observations and the baseline methods used in the field and buildings are also briefly discussed.