The often used rule specifying the relationship between a mean anomaly in a block whose side length is ϑ¿ and a spherical harmonic representation of those data to degree ? (i.e. ϑ¿ ?=180¿) is examined by considering the smoothing parameter used by Pellinen <1966>. We have found that mean anomalies computed from potential coefficients without considering the smoothing parameter can be in error by about 30% of the root-mean-square anomaly value. In addition, tests with actual 5¿ mean anomaly data show that there is considerable gravity information above degree 36 in these anomalies. We conclude that the above mentioned rule should be considered only a crude approximation.