The clinical estimation of hemoglobin has always been a most unsatisfactory procedure. Few scientific data are available concerning the behavior of hemoglobin in health and in disease. Few of the clinical determinations are of any real value, since nearly all are expressed in relative terms and have no exact meaning. There are numerous technical difficulties in the way of accurate hemoglobin work. Perhaps the greatest two obstacles, however, have been the lack of a simple method of calibration for hemoglobinometers, and the absence of a uniform standard. All hemoglobinometers are graduated to read in percentages, yet there is no agreement as to what the 100 per cent. shall be equivalent. One reads in a clinical report that the hemoglobin was a certain per cent., but has no exact idea of what the hemoglobin is in terms of grams. It is usually stated that 100 per cent, on a hemoglobinometer should