Osmotic Behavior of Hemoglobin In Vivo and In Vitro

Dear Sir: In 1964 Savitz, Sidel, and Solomon (1964) produced evidence that the relation of red cell volume to the reciprocal of the external osmotic pressure was linear in a range from 480 to 192 milliosmols. From Savitz, Sidel, and Solomon's results, it may be calculated that the mean fraction of cell water which is apparently osmotically active was 0.81 over the hypertonic ranges of their experiments, and 0.80 over the hypotonic ranges. The hypertonic value is in agreement with some previous results, notably those of ~)rskov (1946) and Lefevre (1964), and may probably be accepted (although Sidel and Solomon (1957) and Villegas, Barton, and Solomon (1958) obtained various values, 0.64 in man, 0.97 in the dog, and 0.55 in beef, which are not quoted by Savitz et al. [1964]). However, the hypotonic result is in conflict with many previous studies, which give values in the range 0.90 to 0.99 (Guest and Wing, 1942; Ponder, 1944; 0rskov, 1946; Guest, 1948; Ponder, 1950; Hendry, 1954; Dick and Lowenstein, 1958). Savitz et al. suggested that the discrepancy in the case of Dick and Lowenstein's (1958) results might be due to the immersion refraetometry method employed by the latter authors; it was supposed that this might measure only the osmotically active water instead of the total cellular water content. However, the specific refraction increment of hemoglobin, which formed the basis of the refractometry method, was measured by Stoddard and Adair (1923) and Roche et al. (1932) in terms of the concentration of dry hemoglobin per unit volume of solution. The solid and water concentrations measured by the refractive index method are therefore the same as those obtained by drying the ceils and lead to values of relative cell volume (and of apparent volume of osmotically active water derived from them) identical with those estimated by the hematoerit or other techniques. In a recent paper Gary-Bobo and Solomon (1968) reported that the fraction of apparendy nonosmotic water (column 6 of Table II) varied with tonicity even when referred to a single reference osmolality, ~ri, being less over hypotonie ranges than over hypertonic ranges. They nevertheless claimed (footnote to Table II) that the results were consistent with the linearity of V plotted against 1/~-, which was described by Savitz et al. (1964). This is evidently impossible since from their equation