VOLUME OF BLOOD 



31 



by using this relationship. If, for example, a plasma 

 label were used, and PV measured, then 



CV,' = PV X 



H' 



H,' 



(5) 



where CVh' is Cl\ corrected for differences between 

 the mean and the central hematocrit, and H/ is H„ 

 estimated from the observed value for He, using the 

 predicted relationship between //„ and H^. The ratio 

 H,n/Hc is termed Feeiia by Reeve, Gregersen, and 

 their collaborators (191)- 



Since the ratio of He to its complement appears in 

 equation 3, the error in estimating the unlabeled 

 compartment from a single label injection, using this 

 equation, is considerably greater than would be sug- 

 gested by the ratio H„/Hc. When He is 0.50, the per- 

 centage error in equation 3 is 2 X 100(1 — Hm/Hc), 

 and the error increases with increasing values of He- 

 For the same reason, when equation 5 is used, the 

 percentage error due to individual variations in the 

 relationship between //„, and He is greater than the 

 coeHicient of variation for Hr^jHe- These relationships 

 become apparent if calculations are made from a 

 hypothetical example. Suppose that in a dog both the 

 cell label volume CI' and the plasma label volume PV 

 are found to be 500 ml; the mean circulatory hemato- 

 crit is obviously //,„ = 500/(500 -|- 500) = 0.50. If 

 blood drawn from the central circulation has the 

 hematocrit He = 0.55, the ratio H^/He = 0.50/ 

 0.55 = 0.91. Since the mean and the central hema- 

 tocrits differ by only 9 per cent, it might be supposed 

 that CVh would be fairly close to CV. That this is not 

 the case is shown by the calculation of CVh = PV X 

 He/(\ — He) = 500 X 0.55/0.45 = 610 ml, a value 

 which is 22 per cent too large. Table i gives data re- 

 ported by Chaplin et al. (38) for a group of human 

 subjects with widely varying central hematocrits, 

 showing a relatively constant ratio H„JHe at 0.910. 

 Yet if values for CVh are calculated, as has been done 

 for the table, the average overestimate of cell volume 

 is 20.5 per cent. If a constant ratio Hm/Hc is assumed, 

 and CVh calculated using HJ, as is done in the table, 

 although the average value of CVh is of necessity the 

 same as the average value of CV, the coefficient of 

 variation is 5.36 per cent. This is nearly double the 

 coefficient of variation for the ratio H^IHe- 



The average ratio Hm/Hc in normal adult humans 

 is reported to be between 0.89 and 0.94 by a large 

 number of investigators (23, 27, 38, 80, 86, 1 10, 154, 

 194, 204, 207, 233). It is reduced to 0.87 in the new- 

 born (161), and to a still lower value of 0.81 in early 

 pregnancy (33). 



The first 4 columns are data reported by Chaplin et al. 

 (38). PFis the T-1824 distribution volume, CF is the P^' cell 

 volume, He the venous (central) hematocrit. The ratio 

 Hn/Hc was obtained by equation 4 (their calculation). I 

 have calculated the remaining values from their data. CVh 

 was obtained from equation 3 and is defined in the text. 

 CP'/i' was obtained from equation 5 of the text, taking He' = 

 0.91 X He. 



There are large and unexplained differences in 

 the ratio HJHe reported by different investigators for 

 the dog. In intact dogs under barbiturate anesthesia 

 its average value is variously reported as 0.90 (118), 

 1.02 (59), and 1. 10 (191). In normal dogs without 

 anesthesia one group finds an average ratio of i .0 

 with large variations (191), while another finds an 



