There is no objection to the use of citrated blood for pernicious anæmia, so that the transfusion can be carried out in the ordinary way described in Chapter VII. It is necessary, however, to utter a warning as to the choice of a blood donor. It is quite clear that in some patients, whose disease has been diagnosed as pernicious anæmia, there is an alteration in the reactions of the serum. The corpuscles may show an agglutination which conforms to one of the group tests described in Chapter VI; nevertheless, it is essential in addition that the patient’s serum should be tested directly against the corpuscles of the proposed donor, even if he belongs to Group IV, whose corpuscles are not agglutinated by the serum of any normal person. I was recently asked to transfuse a patient whose disease had been diagnosed as pernicious anæmia. Her red blood cells had fallen to 600,000 per cmm., so that she was probably in the last stages. Her corpuscles were agglutinated only by serum of Group III, so that she apparently belonged to Group II. Only two donors were available, both of whom belonged to Group IV. Nevertheless, the patient’s serum strongly agglutinated the corpuscles of both of them, so that I considered it inadvisable to carry out the treatment. Similar abnormalities have been noticed by others. It seems to be a universal experience that slight reactions are more commonly met with after transfusion for pernicious anæmia than when it is done for other conditions, although these do not in any way prejudice the results that are obtained. These reactions are possibly to be explained by abnormalities, though of slight degree, in the patient’s serum. In a case such as I have described the reaction would probably be very severe, if not fatal. It is possible also that a well-marked alteration in the serum reaction is not characteristic of the clinical entity constituting true pernicious anæmia, but in reality indicates that there is another underlying cause for the anæmia, such as an undiagnosed carcinoma. Dr. Joekes has recently (August 1921) told me that he believes from his own observations that this is actually the case, but it needs to be established by further investigation. The connexion between malignant disease and abnormal serum reactions is referred to elsewhere (p. 93).