The determination of the blood and red cell volumes is important in clinical, sports medicine, and athlete contexts (22). We recently conducted a meta-analysis of the common techniques to estimate blood volume, red cell volume, and hemoglobin mass (Hbmass) and concluded that carbon monoxide (CO) rebreathing has low error (~2.2%) comparable to that for radioactive labeled 51Cr (~2.8%) and lower than that for Evans blue dye (~6.7%) (12).
The use of CO-rebreathing to determine Hbmass dates from at least 1899, when Haldane and Smith (15) used this approach. Recently, Schmidt and Prommer (23) have proposed that a 2-min test of CO-rebreathing for Hbmass has good reliability (typical error = 1.7%) and also satisfactory validity compared with Hbmass predicted after removal of a known volume of blood. Following the multicompartment model of Bruce and Bruce (7), Schmidt and Prommer concluded that capillary blood sampled 5 min after starting to rebreathe CO would yield accurate values of percent carboxyhemoglobin (%HbCO) and hence Hbmass. However, to develop their model, Bruce and Bruce (7) relied on data from studies (5,8) that used venous rather than capillary blood. An additional consideration is that the time for circulatory mixing has been estimated at 7.5 min (9) and 8.9 min (19). Thus, it is possible that sampling capillary blood (23) only 5 min after inhalation of CO may be too premature in some instances. In the case of incomplete mixing of the inspired CO bolus within the circulation, there would be a nonuniform distribution of HbCO in all vascular compartments as a function of time. Accordingly, we hypothesized that initially there would be higher %HbCO values in capillary than venous blood, but that these would converge over time consistent with the previous work of Hutler et al. (17) during a longer rebreathing period.
The aim of this study was to compare capillary and venous samples of blood collected simultaneously to determine whether the site of blood sampling alters the measured %HbCO and hence the calculated Hbmass. To assess the validity of the 2-min CO-rebreathing procedure, we compared it against the 20-min CO-rebreathing procedure that we have used extensively (2-4,11,13,20). Given the convenience of capillary compared with venous blood for subjects, a second aim was to determine the optimal time for capillary blood samples to reflect adequate circulatory mixing of rebreathed CO.
This study comprised three experiments to examine the reliability of CO-rebreathing and compare the accuracy of venous with capillary blood samples.
Experiment 1 was designed to assess independently the reliability of a new 2-min CO-rebreathing procedure described by Schmidt and Prommer (23). Unlike Schmidt and Prommer, we sampled blood concurrently from both an antecubital vein at the elbow and from a fingertip. We also compared the accuracy of the 2-min procedure against a 20-min procedure with which we have had extensive experience (2-4,10,11,13,20,21).
Experiment 2 was a comparison of the time course of %HbCO in capillary and venous blood samples after inhalation of a bolus of CO.
Experiment 3 was determination of the reliability of a modified version of the 2-min CO-rebreathing procedure using capillary blood samples.
Fifteen healthy, recreationally active volunteers (10 men and 5 women) agreed to participate in this study after being informed of the potential risks associated with participation. These potential risks were bruising or infection at the site of venepuncture or lancet puncture. CO poisoning was not a risk because the CO dose was strictly limited to < 100 mL (the syringe maximum volume) with the 2-min CO-rebreathing procedure; a dose of approximately 1000 mL for an adult male athlete would be required to induce life-threatening symptoms (18). Each subject provided written informed consent, and all protocols were approved by the Australian Institute of Sport ethics committee. Twelve subjects completed experiment 1, and six of them completed experiment 2. A total of six subjects completed experiment 3, three of whom completed experiment 1 and three who had completed neither experiment 1 nor 2. The subject characteristics are contained in Table 1 and their fitness status during the previous 3 months was estimated as "aerobically fit" using self-report against the American College of Sports Medicine criteria (1) of at least three sessions per week using large muscle groups, for at least 15 min per session.
Two tests were performed 1-8 d apart (average = 3 d) using the 2-min CO-rebreathing method. This test was recently described in detail (23). Briefly the procedure comprised inhalation of a bolus of 99.5% chemically pure CO (BOC Gases, Sydney, Australia; men: aerobically fit 1.0 mL·kg−1, unfit 0.7 mL·kg−1; women: aerobically fit 0.8 mL·kg−1, unfit 0.6 mL·kg−1) delivered via a 100-mL plastic syringe (Omnifix®, B|Braun, Melsungen, Germany) that was rebreathed for 2 min in a 3- to 3.5-L anesthetic bag connected to a handmade glass spirometer (University of Bayreuth, Germany; to purchase contact email@example.com) with measurement of blood %HbCO before, as well as 4 and 6 min after starting to rebreathe. After completing the 2 min of rebreathing, measures were made with a handheld parts-per-million CO analyzer (model 220, Fluke, Canada) to quantify the volume of CO that was not taken up by the body, which was calculated as the product of the rebreathing bag volume and the concentration of CO in it (Fig. 1). An additional correction was made to account for the CO volume that was exhaled after disconnecting from the spirometer until the time midway between the final two blood samples. This extra correction was calculated at minute 5 after starting the rebreathing, as the measured end-tidal CO concentration (CO analyzer, model 220, Fluke) multiplied by the estimated alveolar ventilation (5.25 L·min−1 (25) between minutes 2 and 5. This estimate is quite robust because a 40% error in the alveolar ventilation (e.g., 7.35 vs 5.25 L·min−1) would alter the calculated Hbmass by only 0.3%. The subjects remained semirecumbent throughout the procedure. They were instructed not to exercise for 1 h prior to the CO-rebreathing test but otherwise were not requested to control their diet or hydration status.
A third test was performed by each subject using a 20-min rebreathing procedure based on the protocol of Burge and Skinner (8), which was preceded by a 5-min period of breathing 100% O2 to flush nitrogen from the airways (8). The procedure was also modified to include two doses of CO (99.5% chemically pure, BOC Gases, Sydney, Australia) administered into the rebreathing circuit via a 60-mL plastic syringe (model SS-60L, Terumo Corporation Australian, New South Wales). The first dose was 15 mL for women and 25 mL for men. The second dose was based on fitness and body mass; unfit women: 1 mL·kg−1, aerobically fit women and untrained men: 1.25 mL·kg−1; aerobically fit men: 1.5 mL·kg−1. Both doses of CO were rebreathed for 10 min with the subject semirecumbent throughout. Blood was sampled for determination of %HbCO at minutes 10 and 20, corresponding to the end of the first and second 10-min rebreathing periods. This protocol takes 26 min to administer because of the 5-min period of breathing 100% O2 that precedes the first 10 min of rebreathing CO, and there is an additional 1 min required to introduce the second dose of CO (11). For simplicity, the blood sampling will be referred to as minute 10 and minute 20 to better indicate the timing of the blood samples relative to the duration of CO-rebreathing.
The order of the three tests (2 × 2 min and 1 × 20 min) was counterbalanced among the first 12 subjects. All three tests were completed within 8 d (mean and SD = 4 ± 2 d).
After puncturing a fingertip with a lancet (Unilet, Owen Mumford, Oxford, England), arterialized blood samples (~200 μL) were taken from a hand prewarmed for 3-5 min in approximately 15 L of water (~45°C). Blood samples were collected in glass preheparinized capillary tubes (Clinitubes, Radiometer, Copenhagen, Denmark) that were filled in approximately 20 s, commencing 10 s before the target sampling time. Two 200-μL tubes were taken at baseline, one was taken at minute 4, and one was taken at minute 6. Each sample was gently mixed, capped, and stored in a refrigerator (4°C) until analysis, which usually occurred within 15 min of collection. Each sample was analyzed a minimum of five times for %HbCO via a CO-oximeter (OSM3, Radiometer, Denmark), and the target range of the replicates was ≤ 0.3%. The average of the replicates was used in calculations of the change in %HbCO and Hbmass.
Venous blood samples (~2.5 mL) were taken via a cannula (22G Optiva ®, Johnson and Johnson Medical, Sydney, Australia) kept patent with several heparinized saline flushes (10 IU·5 mL−1, Pharmacia, Perth, Australia). A three-way stopcock (Terumo, Elkton, MD) was used for sampling into sterile preheparinized (5000 IU·mL−1, Pharmacia, Perth, Australia), 5-mL ground-glass syringes (Enterna-Matic model, Sanitex, Switzerland). Blood sampling from the antecubital vein was matched to the timing of the capillary blood sample, but only one 2.5-mL sample was taken at baseline. Each sample was analyzed a minimum of five times for %HbCO with the same CO-oximeter (OSM3, Radiometer, Denmark) as used for the capillary tube samples.
The total hemoglobin mass for both the capillary and venous blood samples was calculated for minute 5 as follows:
* K = (ambient barometric pressure mm Hg × 273°K)/(760 mm Hg × ambient temperature °K);
* MCO = CO volume administered into the system minus CO volume not bound to hemoglobin (calculated as the sum of CO volume remaining in the spirometer and the lung, as well as CO volume exhaled during the time between disconnecting the subject from the spirometer and the final blood sample), which was then multiplied by 0.99 to correct for 1% loss of the CO dose to myoglobin (7).
* Δ%HbCO = difference between %HbCO at baseline and %HbCO in the blood samples 5 min after CO administration. The minute 5 concentration was calculated as the mean of the minute 4 and minute 6 blood samples.
* 1.39 = Hüfner's count for the CO-binding capacity of hemoglobin (1.39 mL CO·g−1 Hb) (14,24).
The differences in Hbmass calculated from capillary or venous blood only derive from the Δ%HbCO factor in Equation 1. All other factors in this equation influence equally the calculations for both capillary and venous blood.
The same calculation for Hbmass was used for the 20-min procedure, but in this case Δ%HbCO = difference between HbCO after the first and second doses of CO. Also in this case, MCO was calculated as the volume of CO administered minus the volume of CO remaining in the rebreathing system, which was estimated as 2.2% of the dose (8).
The second experiment compared the time course of %HbCO in both capillary and venous blood. The 2-min procedure used in experiment 1 was repeated in six of the subjects after an interval of 1 month with simultaneous sampling of blood from an antecubital vein and a fingertip. However, unlike experiment 1, blood was sampled serially; that is, before as well as 4, 6, 8, 10, and 12 min after commencing CO-rebreathing.
The final experiment on six subjects sought to quantify the reliability of a modified version of the 2-min CO-rebreathing procedure, with approximately 200 μL of capillary blood sampled from a fingertip before as well as 8 and 10 min after commencing CO-rebreathing. This experiment was conducted 6 wk after experiment 2, with two trials completed within 3 d (mean and SD = 2 ± 1 d). The same calculation for Hbmass was used as in experiment 1, except that both MCO and Δ%HbCO were determined at minute 9 after CO administration. The minute 9 concentration of %HbCO was calculated as the mean of the minute 8 and minute 10 blood samples.
The reliability of duplicate tests for experiments 1 and 3 were expressed as the typical error, which is calculated as the standard deviation of difference scores divided by √2 (16). When expressed as a percent of the grand mean, the percent typical error (%TE) is obtained. Bland and Altman (6) plots were used to compare the %HbCO and Hbmass from the two 2-min tests conducted as part of experiment 1, and between the venous and capillary blood samples within both of the 2-min tests and the 20-min procedure. Repeated-measures ANOVA with main effects of type of blood (capillary vs venous) and time (0, 4, 6, 8, 10, and 12 min) were used to compare the change in %HbCO and Hbmass for experiment 2 using Statistica (version 6.0, Statsoft, Tulsa, OK). A repeated-measures ANOVA for type of blood and time (5, 7, 9, and 11 min) was also used to compare the resultant Hbmass calculated from mean data of minutes 4 and 6, 6 and 8, 8 and 10, and 10 and 12 for experiment 2. Tukey HSD post hoc tests were used to identify differences between cell means. Values are reported as mean ± SD unless indicated otherwise.
The %TE (and 90% confidence limits) for Hbmass from the 2-min procedure using capillary blood was 1.1% (0.9-1.8%), and the corresponding value using venous blood was substantially larger 3.4% (2.5-5.4%). Figure 2 contains the individual data of the differences in Hbmass between the first and second of the 2-min procedures. The mean difference between %HbCO determined from the capillary and venous blood for the first 2-min procedure was −0.02% before rebreathing, but 0.29 and 0.28% at minutes 4 and 6, respectively. Similarly, for the second 2-min procedure, the corresponding values were −0.01% before rebreathing but 0.30 and 0.31% at minutes 4 and 6, respectively. For the 20-min procedure, the mean difference between %HbCO of the capillary and venous samples was −0.0.2% at minute 10 and 0.05% at minute 20. Figure 3 contains the individual data for the aforementioned comparisons between capillary and venous blood. The difference between the Hbmass calculated from the capillary versus the venous blood for the first 2-min test was −28 ± 48 g or −3.2 ± 5.5%, and for the second 2-min test was −53 ± 79 g or −5.5 ± 7.7% (Fig. 4). The difference between the Hbmass calculated from the capillary versus the venous blood for the 20-min procedure was −10 ± 25 g or 1.0 ± 2.5% of the mean venous value of 870 g (Fig. 4). Compared with the venous blood from the 20-min procedure, Hbmass for the average of the capillary blood for two 2-min methods was biased low by −4.8 ± 3.3% (−39 ± 26 g).
Overall, the Hbmass calculated from capillary and venous blood was different between the type of blood and time of sampling, (F(4,20) = 4.4, P = 0.01); and significantly lower at minute 4 for capillary (735 ± 247 g) than venous (762 ± 219 g) blood (Fig. 5). The %HbCO for the capillary (5.94 ± 0.95) and venous (5.65 ± 0.90) blood samples were significantly different only at minute 4 for the 2-min procedure, F(5,25) = 3.1, P = 0.03 (Fig. 5). Before rebreathing, the capillary-venous difference was 0.04 ± 0.06%, and at minutes 4, 6, 8, 10, and 12, the differences were 0.29 ± 0.33, 0.16 ± 0.17, 0.08 ± 0.07, 0.01 ± 0.09, and 0.02 ± 0.06%, respectively. The convergence of the capillary and venous curves together with the small standard deviation is consistent with complete circulatory mixing at minute 10 for all subjects. The Hbmass for capillary blood at minute 5 (740 ± 249 g) was significantly lower than those at minute 9 (761 ± 254 g) and minute 11 (773 ± 259 g). The Hbmass for venous blood at minute 5 (761 ± 230 g) was not significantly different from those at minutes 7, 9, or 11. The Hbmass calculated for capillary blood at minute 5 was significantly less than that from venous blood.
The %TE for Hbmass was 1.2% (0.8-2.5%) for the 2-min procedure with capillary blood sampled at minutes 8 and 10.
When we replicated the 2-min CO-rebreathing procedure of Schmidt and Prommer (23) using capillary blood, the typical error for Hbmass in our hands was 1.1%. Taking into account the confidence limits of our error estimate, it is not different from the typical error of 1.7% reported by Schmidt and Prommer (23), who pioneered this technique. However, Hbmass was consistently lower (3-6%) when calculated from capillary than from venous blood, which likely reflects incomplete circulatory mixing of the inhaled CO. Our time series of both capillary and venous blood suggests that, in most healthy resting subjects, capillary blood should be sampled at 8 and 10 min after CO-rebreathing commenced. This contrasts with the sample times recommended by Schmidt and Prommer (23), namely 4 and 6 min after CO-rebreathing commenced.
The low typical error obtained by both ourselves and Schmidt and Prommer (23) is better than that previously obtained by virtually all investigators using CO-rebreathing to estimate Hbmass, except Burge and Skinner (8), who achieved a value of just 0.8%. The method of Burge and Skinner assumes that 2.2% of the CO is not taken up by the body. The small typical error of the 2-min method of Schmidt and Prommer (23) is likely due to the individual corrections for the residual CO in the rebreathing bag and that exhaled during the few minutes seated between the end of rebreathing and the midpoint of the two post-rebreathing blood samples. Schmidt and Prommer's method requires measurement of these amounts of CO via a handheld CO analyzer.
Our conclusion that circulatory mixing of CO is incomplete in some subjects 4-6 min after commencing the CO-rebreathing is supported by the result that the capillary and venous blood samples taken at rest were virtually identical for %HbCO for both the 2-min tests and the 20-min procedure (Fig. 3). However, at both minutes 4 and 6 for the 2-min tests, the capillary values for %HbCO were up to approximately 0.8% higher than the corresponding venous values, particularly for two or three subjects. In comparison, the venous and capillary samples were within a mean of 0.05% of each other throughout the 20-min procedure. The difference in %HbCO for the capillary blood from the 2-min method (average of tests 1 and 2) versus venous blood from the 20-min procedure equated to a 4.8% underestimate of the calculated Hbmass. Part of the discrepancy between the 2- and 20-min procedures for the calculated Hbmass could be due to our assumption that there was loss of CO to myoglobin, which was different for the two methods. Following the recommendation of Schmidt and Prommer (23), 1% loss was assumed for the 2-min test when blood was sampled at minutes 4 and 6, whereas Burge and Skinner (8) assumed no loss of CO to myoglobin for the 20-min procedure. When no correction factors were applied in both cases, the calculated Hbmass was −3.8 ± 3.3% lower for the 2-min test compared with the 20-min procedure. However, the time for CO diffusion is substantially different for the two methods, being 5 and 10 min for the 2- and 20-min procedures, respectively. If the appropriate corrections for the time of CO diffusion were applied (1% per 5 min (7,23)), the calculated Hbmass was −2.9 ± 3.5% lower for the 2-min versus the 20-min method. When sampling blood at minute 8 and −10 after 2 min of CO-rebreathing, this small difference becomes negligible (~1%) as can be derived from experiment 2 (Fig. 5). We therefore recommend to use minutes 8 and 10 for capillary blood sampling for the 2-min test and to apply at 2% correction for loss of CO to myoglobin.
From our second experiment, we conclude that, in resting subjects, capillary blood should be sampled at 8 and 10 min after CO-rebreathing commences, because the difference between capillary and venous concentrations of %HbCO was very close to zero at both minute 8 and minute 10 after beginning to rebreathe the CO bolus. Moreover, at these times, the standard deviation of the difference between capillary and venous concentrations of %HbCO was approximately 0.1%, which indicates that circulatory mixing was complete for all of our subjects, unlike at minutes 4 and 6. Our results are consistent with those of Hütler et al. (17), who concluded that the peaks in earlobe-capillary and antecubital-venous fractions of HbCO were after 5 and 8.5 min, respectively, of CO-rebreathing. However, Hütler and coworkers used traditional CO-rebreathing of a CO dose mixed within a reservoir of O2. Our data, using the 2-min bolus CO-rebreathing test, had the highest concentration of venous HbCO at minute 6 (5.70%), and the values at minute 8 (5.68%) and minute 10 (5.62%) were slightly lower (Fig. 5) than the zenith. This diminution of HbCO between minutes 6 and 10 (0.08%) will make a difference of 2 g or 0.2% to the average Hbmass of 870 g for our group of subjects. Consequently, we are confident that minutes 8 and 10 represent optimal times to sample capillary fingertip blood of most healthy subjects after 2 min of rebreathing a bolus of CO. Our results on the time course of %HbCO also indicate that for the 2-min CO bolus test, venous blood should also be sampled at minutes 8 and 10 for those who prefer not to use capillary samples. Our conclusion about the timing of capillary blood sampling does not negate the excellent reliability for Hbmass obtained by using capillary blood at minutes 4 and 6 after CO administration, but in our opinion capillary blood sampling at minutes 8 and 10 yield results that are more accurate compared with the criterion 20-min CO-rebreathing procedure. Therefore, we recommend this small but important variation to the 2-min CO-rebreathing technique first developed by Schmidt and Prommer (23) to allow for complete circulatory mixing of CO. Furthermore, in clinical settings, for instance, with patients with peripheral vascular disease, the appropriate timing of blood sampling after using the 2-min CO-rebreathing method warrants careful investigation.
Using our modification of the Schmidt and Prommer method, we obtained a typical error of 1.2% for blood sampled 8 and 10 min after starting CO-rebreathing, which was not different from the value of 1.1% obtained after sampling blood at minutes 4 and 6. Thus, capillary blood sampling at both minutes 4 and 6 or minutes 8 and 10 have low measurement error, which opens the way to use this technique widely to follow longitudinal changes in athletes, patients, or even adolescents.
In conclusion, the 2-min bolus CO-rebreathing procedure of Schmidt and Prommer (23) using capillary blood to estimate Hbmass has very low typical error (~1%), and is thus suitable for use in multiple experimental settings where researchers may want to track small changes in Hbmass over time. We conclude that, for healthy adults, capillary fingertip and venous blood should be sampled 8 and 10 min after CO-rebreathing commences to allow for adequate circulatory mixing.
This study was funded by the National Sport Sciences Quality Assurance Program of the Australian Institute of Sport.
1. American College of Sports Medicine. Guidelines for graded exercise testing and exercise prescription
, 2nd edition, Philadelphia: Lea and Febiger, 1980.
2. Ashenden, M. J., C. J. Gore, C. M. Burge, et al. Skin-prick blood samples are reliable for estimating Hb mass with the CO-dilution technique. Eur. J. Appl. Physiol.
3. Ashenden, M. J., C. J. Gore, G. P. Dobson, and A. G. Hahn. "Live high, train low" does not change the total haemoglobin mass of male endurance athletes sleeping at a simulated altitude of 3000 m for 23 nights. Eur. J. Appl. Physiol.
4. Ashenden, M. J., C. J. Gore, D. T. Martin, G. P. Dobson, and A. G. Hahn. Effects of a 12-day "live high, train low" camp on reticulocyte production and haemoglobin mass in elite female road cyclists. Eur. J. Appl. Physiol.
5. Benignus, V. A., M. J. Hazucha, M. V. Smith, and P. A. Bromberg. Prediction of carboxyhemoglobin formation due to transient exposure to carbon monoxide. J. Appl. Physiol
6. Bland, J. M., and D. G. Altman. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet
7. Bruce, E. N., and M. C. Bruce. A multicompartment model of carboxyhemoglobin and carboxymyoglobin responses to inhalation of carbon monoxide. J. Appl. Physiol.
8. Burge, C. M., and S. L. Skinner. Determination of hemoglobin mass and blood volume with CO: evaluation and application of a method. J. Appl. Physiol.
9. Gibson, J.G., and W. A. J. Evans. Clinical studies of the blood volume. I. Clinicalapplication of a method employing the azo dye "Evans blue" and the spectrophotometer. J. Clin. Invest.
10. Gore, C., N. Craig, A. Hahn, et al. Altitude training at 2690m does not increase total haemoglobin mass or sea level VO2max
in world champion track cyclists. J. Sci. Med. Sport
11. Gore, C. J., A. G. Hahn, C. M. Burge, and R. D. Telford. VO2max
and haemoglobin mass of trained athletes during high intensity training. Int. J. Sports Med.
12. Gore, C. J., W. G. Hopkins, and C. M. Burge. Errors of measurement for blood volume parameters: a meta-analysis. J. Appl. Physiol
. 99:1745-1758, 2005.
13. Gore, C. J., J. Stray-Gundersen, F. A. Rodríguez, M. J. Truijens, N. E. Townsend, and B. D. Levine. Comparison of blood volume via CO re-breathing and Evans blue dye. Med. Sci. Sports Exerc.
14. Gorelov, V. Theoretical value of Hufner's constant. Anaesthesia
15. Haldane, J., and J. L. Smith. The mass and oxygen capacity of the blood in man. J. Physiol.
16. Hopkins, W. G. Measures of reliability in sports medicine and science. Sports Med.
17. Hütler, M., R. Beneke, and D. Böning. Determination of circulating hemoglobin mass and related quantities by using capillary blood. Med. Sci. Sports Exerc.
18. Killick, E. M. Carbon monoxide anoxemia. Physiol. Rev.
19. Noble, R. P., and M. I. Gregersen. Blood volume in clinical shock. I. Mixing time and disappearance rate of T-1824 in normal subjects and in patients in shock; determination of plasma volume in man from 10-minute sample. J. Clin. Invest.
20. Parisotto, R., C. J. Gore, K. R. Emslie, et al. A novel method utilising markers of altered erythropoiesis for the detection of recombinant human erythropoietin abuse in athletes. Haematologica
21. Saunders, P. U., R. D. Telford, D. B. Pyne, et al. Improved running economy in elite runners after 20 days of moderate simulated altitude exposure. J. Appl. Physiol.
22. Sawka, M. N., V. A. Convertino, E. R. Eichner, S. M. Schnieder, and A. J. Young. Blood volume: importance and adaptations to exercise training, environmental stresses, and trauma/sickness. Med. Sci. Sports Exerc.
23. Schmidt, W., and N. Prommer. The optimized CO-rebreathing method. A new tool to determine total hemoglobin mass routinely. Eur. J. Appl. Physiol.
24. Shimizu, S., Y. Enoki, H. Kohzuki, Y. Ohga, and S. Sakata. Determination of Hufner's factor and inactive hemoglobins in human, canine, and murine blood. Jpn. J. Physiol.
25. West, J. B. Respiratory Physiology - the essentials
, 5th edition, Baltimore, MD: Williams & Wilkins, 1995.