In the 1990 Recommendations of the International Commission on Radiological Protection, the risk-weighted quantity "effective dose equivalent" was replaced by a similar quantity, "effective dose." Among other alterations, the selection of the organs and tissues contributing to the risk-weighted quantity and their respective weighting factors were changed, including a modified definition of the so-called "remainder." Close consideration of this latter definition shows that it causes certain ambiguities and unexpected effects which are dealt with in the following. For several geometries of external photon irradiation, the numerical differences of two possible methods of evaluating the remainder dose from the doses to ten single organs, namely as arithmetic mean or as mass weighted average, are assessed. It is shown that deviation from these averaging procedures, as prescribed for those cases where a remainder organ receives a higher dose than an organ with a specified weighting factor, causes discontinuities in the energy dependence of the remainder dose and, consequently, also non-additivity of this quantity. These problems are discussed, and it is shown that, although the numerical consequences for the calculation of the effective dose are small, this unsatisfactory situation needs clarification. One approach might be to abolish some of the ICRP guidance relating to the appropriate tissue weighting factors for the remainder tissues and organs and to make other guidance more precise.