Journal of

APPLIED

PHYSIOLOGY
VOLUME 4 Jzlne 19~2 NUMBER 12

Rebztionship of Oxygen Tension to Hemoglobin Oxygen

Shtwution in the Arteridl Blood
of Normud Men’
CHRISTIAN J. LAMBERTSEN,2 PAUL L. BUNCE, DAVID L.
DRABKIN AND CARL F. SCHMIDT. From the Laboratory of
Pharmacology, University of Pennsylvania School of Medicine and
Department of Physiological Chemistry, University of Pennsylvania
Graduate School of Medicine, Philadelphia, Pennsylvania

F EW OF THE MANY FACTORS

theoretical and practical
Its shape determines
involved in the respiratory
importance than the dissociation
the rate at which the alveolar to pulmonary
gas exchange have greater
curve of oxyhno&$in.
capillary p02
gradient changes as oxygen is taken up by blood in the lungs. Standard curves of this
type are still in common usage as the basis for estimation of blood ~02 from measure-
ments of percentage oxygen saturation and PH. Moreover, the reverse of this proced-
ure has recently been employed in the process of calculating venous admixture in
the lungs, or the degree to which blood exposed to high alveolar oxygen tensions is
diluted with less well oxygenated blood from anatomical shunts and poorly venti-
lated alveoli (2, 3). Yet in spite of widespread application of oxyhemolgobin dissocia-
tion curves to these and other purposes in respiratory and circulatory studies, their
accuracy as physiological standards appears open to question for the following rea-
sons: a) there are two dissociation curves in common use (4, 5) which differ appreci-
ably, even though they were derived from the blood of the same person; b) it is
possible that there may be occasional wide individual departures from the saturation-
tension relationships depicted in the standard dissociation curves; c) there is no as-
surance that the behavior of hemoglobin exposed to alveolar air in the lungs is ac-
curately portrayed by that of blood removed from the body and equilibrated later
and in an abnormal environment; d) the previously unchallenged procedure of de-
termining the oxygen saturation of blood gasometrically has recently been shown
Received for publication August 13, 195 I.
l This investigation was supported by a research contract with the Office of Naval Research
(Medical Sciences Division). A report dealing with certain aspects of this work was presented at the
Sir Joseph Barcroft Memorial Conference on Hemoglobin, Cambridge, England, June 1948 (I).
2 John and Mary R. Markle Scholar in Medical Science.

873

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 874 LAMBERTSEN, BUNCE, DRABKIN AND SCHMIDT Volume 4

to contain inherent errors (6) leading to systematic under-estimation of the true

saturation; and e) the oxygen saturation determined by an accurate spectrophoto-
metric procedure (7) is consistently higher than that derived gasometrically.
The validity of several of these objections to the arbitrary use of one or the
other of the available in vitro dissociation curves could be evaluated by obtaining ob-
servations of the oxygen tension-saturation relationships in the arterial blood of a
large number of individuals subjected to varying degreesof anoxia, of others by im-
provements in analytical technique. A major obstacle to such studies has been the
lack of a convenient method for accurately determining the oxygen tension in blood
as it is drawn from the body. The microtonometric procedure introduced by Krogh
(8), in which ~02 is estimated by analysis of a small gas bubble brought to tension
equilibrium with the blood, promised to solve this problem but after a tentative
trial by Barcroft and Nagahashi (9) on the venous blood of one subject the method
was little used until recently (10-13). Riley, Proemmel and Franke (12) greatly
simplified the procedure by devising a technique which employed the r-ml. Rough-
ton-Scholander syringe analyzer (14) both for equilibration and subsequentanalysis
of the gasbubble.
After applying the microtonometric method to an investigation of oxyhemo-
globin dissociation in I I subjects, Riley, Lilienthal, Proemmel and Franke (IS> con-
cluded that the oxygen dissociation curve in vivo is practically identical with the one
constructed from data obtained by Dill et al. (5, 16) who employed macrotonometry
in vitro and that differences between individuals are too small to be detected by the
methods used. These conclusions are rather surprising for they were based upon
measurementsof both venous and arterial blood during rest, exercise, carbon monox-
ide inhalation, anoxia and hyperventilation, and at serum PH values ranging from
7.29 to 7.61, whereas Dill’s curve (5) is referred to a serum PH constant at 7.40.
Furthermore, 42 of the 55 observations made by Riley et al. were upon only three
personswhoseindividual characteristics must have had a dominant effect.
For these reasons,as well as those enumerated previously, we set out to explore
further the relationships of oxygen tension and oxygen saturation in the arterial
blood of normal men. While similar in principle to the experiments of Riley et al.,
our studies differed in the application of refinements of the microtonometric proced-
ure that provided greater accuracy in determining blood gas tensions, in the use of
both the spectrophotometric and the gasometric determination of percentage hemo-
globin saturation, in the use of only arterial blood collected under resting conditions,
of inhalation of oxygen-nitrogen mixtures to change arterial oxygen tension, and of
samplesfrom many different individuals rather than repeated specimensfrom a few.

METHODS

Subjects
Fifteen healthy men served as subjects. They varied in age from 21 to 29 years,
with the exception of one (CFS), who was 54. All were nonsmokers, a precaution
taken to minimize the possiblepresenceof significant amounts of carboxyhemoglobin
in the circulating blood. The subjects reported between 8:oo and ro:oo A.M. without
having had breakfast, then rested in a supine position for at least I hour before the
experiments were begun. During the first half of this period all the necessaryprep-
arations, including arterial puncture, were completed. Thus there was an undisturbed
rest period of at least 30 minutes prior to withdrawing the first blood samples.

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 June Ig5.z ARTERIAL OXYGEN TENSION-SATURATION IN VIVO 875

Gas Administration
After normal values were obtained during the breathing of air, arterial anoxemia
of varying degree was produced by inhalation of gas mixtures containing 4 to 18 per
cent oxygen in nitrogen administered by means of a breathing system consisting of
a mouthpiece, noseclip, breathing valves and a demand regulator attached to a
cylinder of the desired gas mixture. Except for the 4 per cent oxygen, each gas mix-
ture was breathed at least 15 minutes before withdrawing the blood samples.

Blood Sampliq
Following perivascular infiltration with 3 ml. of I per cent procaine solution, a
ro-gauge spinal needle was inserted into a femoral artery and connected to a heparin-
filled manifold of three-way metal stopcocks (17). Blood was collected anaerobically
in syringes or tonometers attached to the manifold for ease and speed of sampling.
During the breathing of each gas mixture blood samples were drawn over a total
period of z minutes without the subject’s knowledge. The syringes were sealed with-
out trapping air bubbles and all samples except those for spectrophotometric de-
termination of percentage hemoglobin saturation were immediately placed in beakers
of water containing cracked ice.
Blood Analyses
Percentage Hemoglobin Saturation. This was determined in every instance by
both the gasometric and the spectrophotometric method.
Gasometric method. The carbon dioxide content, oxygen content and oxygen
capacity of the whole blood were measured by the manometric method of Van Slyke
and Neil1 (18, 19), the only modification being the use of the same acid-saponin-ferri-
cyanide reagent for oxygen content and for oxygen capacity (20, 21). The ro-ml.
blood samples for these determinations were drawn into glass syringes, the dead
space (average about 0.2 ml.) of which had been filled with commercial heparin
solution to prevent clotting and exclude air bubbles. For determination of the oxygen
capacity a J-ml. portion of blood was rotated at room temperature for 60 minutes
in an air-filled tonometer of roe-ml. capacity. Evaporation of water from the blood
sample was minimized by closing the ends of the tonometer with rubber stoppers
equipped with 22-gauge hypodermic needles permitting periodic flushing with fresh,
humidified air as well as maintenance of a constant pressure. The analyses were per-
formed in duplicate upon r-ml. aliquots of blood and were completed within 2 hours
of withdrawal of the blood samples. The precautions listed by Van Slyke and Neil1
(18) were carefully observed. Percentage hemoglobin saturation was calculated from
the analytical data corrected for the known solubility of oxygen in whole blood (22)
by the standard formula:
(02 content - dissolved 0,)
Percentage saturation = - x 100
02 capacity - dissolved O2
Spectrophotometric method. Determinations of the concentrations of oxyhemo-
globin and total pigment were made upon anaerobically hemolyzed blood samples,
essentially as described by Drabkin and Schmidt (7). The Drabkin and Austin
cuvette of o.oo7-cm. depth (23) was employed and the following equations were used
in the treatment of the spectrophotometric data:
I) Fraction
=P where the term Zp is the summation
of Hb, r = -- of the partial
ZT’

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 LAMBERTSEN, BUNCE, DRABKIN AND SCHMIDT Volume 4

changes in fractional molar extinction coefficients (E) at wave lengths 600, 578, 562
and 542 rnp, and CT is 16.45, the total changeat the same wave lengths between the
extinctions of HbOt and Hb. At the stated wave lengths the C(C = I mM/liter; d =
I cm.) for HbOn equals 0.75, 15.34, 8.50 and 14.61 and for Hb equals 4.05, 9.96,
12.75 and I I .09 (24) where c is the concentration referred to an equivalent weight
of 16,700, and d is the depth in centimeters. The summation of the A c HbOz --+ Hb
at each wave length yields CT = 16.45. Total pigment was determined independently
as cyanmethemoglobin (24) which is equivalent to estimation of hemoglobin iron.
2) Fraction of HbOz = I - r.

HbO:! X IOO
3) Percent age hemoglobin saturation =
Total pigment
Although this was the method employed in obtaining the reported data, an
alternative, somewhat less accurate procedure may be used (25). It depends upon
the ratio of densities at two wave lengths only and has the advantage of ready calcu-
lation of percentage HbOn directly from the density readings, D, as follows:
0.780
4) Fraction of HbO2 = where 0.780 equals the ratio of E for
0.333 y + 0.423’
Hb at wave lengths 578 and 652 rnp or =75/9*96* Measurements were performed
in duplicate at room temperature ( 2 5O-2 7OC.), and w .ere completed within 4 minutes
of the time of collecting the blood sample. The random error of the method is ho.5
per
- cent of total pigment (7).
Oximeter. During the inhalation of room air and low oxygen mixtures the Milli-
kan oximeter (26) was used routinely to indicate the attainment of a stable state of
arterial oxygen saturation. The relationships between these values and those ob-
tained manometrically were similar to what has been reported by others and so are
not bresented here.
Blood gas tensions. A microtonom etric or bubble-equil ibration technique modi-
fied from the r-ml. Roughton-Sch .oland .er syringe method of Riley, Proemmel and
Franke (I 2) w ‘as used. As a result of a search for sources of error entailing several
thousand analyses, substantial reductions in both the systematic and the absolute
error of the original method have been achieved (27 ). The more important d .eviations
from the basic method i nclude : a) a decrease in the reading error from the eq” ivalent
of ab lout +2.0 mm. Hg to about 4~0.5 mm. Hg bY using a 200-mm. capill .ary- gradu- -
ated in I-mm. units in place of the Ioo-mm. capillary graduated in 2-mm. units,
b) a decrease in the original bubble-blood ratio from about I: 75 to I: I 60 by increas-
ing the blood sample used for a single determination from I to 5 ml. while only ap-
proximately doubling the gas bubble volume, c) a decrease in undesirable exchange
of gas with reagents by pre-equilibration of the latter with gases having tensions
approximating those encountered du .ring the various stages of analysis of bubbles
equilibrated
- with arterial or venous blood, d) removal of the blood film from the
analytical syringe capillary prior to beginning a bubble analysis, and e) the achieve-
ment of a more uniform filming of reagents on the wall of the syringe capillary during
determinations by modifying reagents and by the use of a mechanical syringe holder
with a screw adjustment to provide smooth control of the bubble being analyzed

A major advantage of microtonometric procedures is that their accuracy can

always be checked by analysis of blood samples of known gas tensions prepared by

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 June rgp ARTERIAL OXYGEN TENSION-SATURATION IN VIVO 877

a suitable macrotonometric technique. In this manner we learned that the error not
only varies between investigators but even with the same investigator at different
times. This indicates that the accuracy of each analyst must be repeatedly confirmed
and that general application of the stated error of the originators of the method is
not justified (12, 27). We carried out such determinations by rotating ao-ml. samples
of blood (bovine or human) for I hour in contact with the gas phase of a rooo-ml.
tonometer at 37OC. and atmospheric pressure, determining the ~02 and pCOz in the
blood by the microtonometric procedure, in the air by a suitable gas analyzer, and
then comparing the results. For the gas analyses we used the Scholander 0.5-cc. gas
analyzer (28) which in our hands has an accuracy of ~ko.02 per cent for both COn
and Oz. The gas phase of the macrotonometer was considered the standard of reference
in estimating the error of the bubble equilibration method.
A summary of three separate series of such macrotonometer experiments to de-
termine the error of the analysts who used the modified microtonometric method in

Fig. I. ROUGHTON-SCHOLANDER ANALYTICAL SYRINGE as modified for increased accuracyinthe

microtonometric estimation of blood gas tensions by the method of Riley et al. (I I). The s-ml.
syringe is shown with its zoo-mm. capillary and the mechanical manipulator with a coarse adjust-
ment clamp for filling and a fine adjustment screw for bubble analysis.3 INSERT shows the technique
of filling the syringe analyzer from a sampling syringe (I 2).

this work is presented in table I. Each series consists of consecutive observations, re-
taining gross errors, and has been statistically analyzed as a group of single observa-
tions. It can be seen that in only one instance did a significant systematic error exist
and that the standard deviation of individual measurements is approximately & 2.0
mm. Hg for ~02, somewhat less for pCOz. Correction factors were therefore not
applied.
Direct determinations of blood gas tensions by the modified method were per-
formed in duplicate at 37OC. upon blood collected in heparinized glass syringes and
were completed within 15 minutes of withdrawal. Since a series of experiments had
previously indicated that the rate of fall of ~02 in blood kept at 37’C. is considerably
slowed by the addition of cyanide, the sampling syringes at first contained not only
sodium fluoride to decrease glycolysis but potassium cyanide to inhibit blood oxygen

3 The glass syringe analyzer was made by Mr. J. D. Graham, Department of Physiology,
University of Pennsylvania Medical School, Philadelphia, Pa. and the manipulator by Mr. Otto
Heble, Biology Department, Swarthmore College, Swarthmore, Pa.

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 878 LAMBERTSEN, BUNCE, DRABKIN AND SCHMIDT Volume 4

metabolism and sodium oxalate to buffer the potassium cyanide. The final concen-
trations of these three agents in the blood sample were .OI N, .OOI N and .OOI N, re-
spectively. Further studiesunexpectedlyrevealed that theeffect of cyanide was not due
to inhibition of blood oxygen consumption (as we had supposed) but to a slow forma-
tion of cyanmethemoglobin. The reaction HbOa G HHbCN + O2 apparently is
shifted to the right by the addition of KCN and the slower fall of ~02 in the blood,
while real, is due to the concomitant liberation of oxygen from hemoglobin nearly
balancing the oxygen uptake by the system, which actually is undiminished. More-
over since at 37OC. even I per cent sodium fluoride did not effectively prevent glycol-
ysis, we now add only heparin to the blood samples. That the change has not affected
the accuracy of the tension determination is evident in table I in which the first
and second series were carried out with the addition of the fluoride-cyanide-oxalate
mixture, the third using heparin alone.

TABLE I. ACCURACY OF DIRECT BLOOD GAS TENSION MEASUREMENTS

ACCURACY

NO. OF
RANGE
ETER
OF
GAS
MACROTONOM-
TENSIONS
-~
peon
ANALYST ANAL-
.~
YSES

PC02 PO2 PI Pl

mm. Hg mm. Hg mm. Hg mm. Hg

CJL II 30-35 130-150 -1.3 4x2.0 ’ -0s
PLB 16 45-P 64-74 -0.6 4x2.1 <*3, >.2
GLE 20 37-44 90-100 0.0 I rt1.6 B-9
I
Mean error and standard deviation were calculated for a consecutive series of single determin-
ations upon blood samples prepared in macrotonometer, the gas phase of the macrotonometer serving
as the standard of reference. Determinations grossly in error were retained.
'P = probability of obtaining t values as large (or larger) than observed by chance alone when
mean error is not significantly different from zero.

Blood PH. Blood pH measurements were performed anaerobically with a glass

electrode of the MacInnes-Belcher type maintained at 37OC. by means of a thermo-
statically controlled water jacket. The Cambridge Model R PH meter used, although
it permits reproducibility of duplicate determinations to within +.OI u, has an
over-all accuracy of only A.02 u. Two phosphate buffer solutions, of PH 7.36 and
PH 6.98 at 37’C. were used as standards. Repeated checks of one buffer against the
other were made during the period when blood samples were being analyzed and the
pH meter was standardized before and after each blood measurement. The p14 de-
terminations were performed in duplicate and completed within I hour of the time
the blood samples were collected.

RESULTS

The results of analyses performed upon arterial blood samples collected from 15
healthy, basal subjects breathing room air and gas mixtures with lower percentages
of oxygen are presented in table 2. Since it was desirable to restrict this investigation
to conditions existing in arterial blood during life, no attempt was made to extend the
observations below the z-mm. Hg oxygen tension encountered during the brief
inhalation of 4 per cent oxygen in nitrogen. The data illustrate the composite effects
of increasingly severe anoxia uncorrected for the concomitant hyperventilation, hypo-
capnia and rise in PH.

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 June 1952 ARTERIAL OXYGEN TENSION-SATURATION IN VIVO 879

TABLE 2. ARTERIAL BLOOD GAS DATA OBTAINED UPON NONSMOKERS

SPECTROPHOTOMETRIC 1 MANOMETRIC
/ BAR
02
TRIAL SUB JECt IN- PH PRES-

I )
02 02 02 02
SPIRED SURE
Con Capac- % Con- Capac- sFt
Sat. tent ity .
fent ity
- -~~-~-~
% vol. y* vol. 70 vol. y* vol. 70 mm. Hg’mm. Hg
I AMC 20.9 19.8 20. I 98.4~ 19.2 =9-o 96-s 9= 39 7-37~ 754-I
2 AMCl =9*7 20.0' 98.4 18.8 =9-s 96.4 9= 40 7.38 754-I
3 JPM 21.51 21.8 98.7 20.6 21.3 96.7 92 36 7.38 762.0
4 MB 19.0 19.3 98-4 18.9 =9*9 95*o 99 40 7-40 767-I
5 MS 20.6 20.8 99.2 =9*7 20.0 98.5 98 38 7.38~ 758.8
6 LS 19.6 19.9 98.4 19.4 =9*9 97-S 93 38 7.39 738-I
7 WK 20.8 21.2 97.9 20.2 21.2 95.3 92 40 7-35 753-9
8 DR 18.5 18.7 98.8 18.4 18-9 97-4 94 36 7-39 762-4
9 LS’ 19.6 20.0 97.9 19.3 19.6 98-S 94 39 7*4o, 761.2
IO REM IO-4 =9*7 98.5 18.5 18.8 98.4 96 41 7*38# 760. I
II CFS 18.3 18.7 97.6 17.8 18.2 97.8 93 38 7.43 76r.5
Average’ =9-7 20.0 98.4 19.2 19.8 97.0 94.2 38.4 7.39
S.D. &I.1 &I.1 zto.6 zto.9 *1.0&1.3 rt2.8 hr.7 ko.02
--- 4-
-- --_c__-
I2 EMcC 18 20.7 21.4 96.7 19.8 20.9 94.8 85 38 7.41 770.6
I3 RS 20.7 21.8 94.8 19.8 21.3 93.0 70 36 7*38 77o* 7
I4 JJJ 18.3 19.c 96.6 18.7 =9*4 93-8 73 33 7*4r 758. I
I5 RDW 19.2 20.1 95.5 19.2 20.9 92.3 72 33 7.41 768.5
16 MHP =7-s 18.2 96.2 17.2 18.0 95.6 83 38 7-40 758.5
I7 CFS 18.1 18.8 96.5 16.5 I8.I 93.4 77 39 7.41 76r.s
--- --- ------
I8 JPM 16 =9-7 21.5 91.8 21.2 93.7 62 33 7-40 762 .o
I9 MB 18.2 19.8 91.6 20.2 88.6 62 40 7.39 767.1
20 MS 18.8 20.3 92.8 21.6 91.2 63 36 7*4= 758.8
21 MHP 17.2 18.4 93-2 18.2 91.8 63 42 7.42 758-s
22 CFS 17.4 18.8 92.6 18.6 90.3 62 41 7.43 761.5
-- -- -- P-P- --
23 EMcC I4 18.0 20.7 86.7 18.~ 21.2 84.9 49 38 7.40 770-6
24 RS 18.2 21.8 83.6 I7.5 21.9 81.8 46 37 7.38 770.7
25 RDW =7*9 20.2 88.4 18.5 20.5 90.2 53 32 7.45 768.5
26 SW 19. I 21.1 90.8 18.5 2r*I 87.7 59 37 7.36 762*5
-e-- -- -- -- -- P-P--
27 LS 12 r4*7 20.6 7r.4 15.c 2=*0 7=*5 35 33 7.40 738. r
28 WLK 16.0 20.5 78.1 15.2 20.2 75.8 40 37 7.40 753.9
29 DR 14.5 19.2 75.5 15.1 19.6 77.0 36 35 7-43 762.4
-- -- -- ----_ -- --
30 RDW IO 13.2 204 63.1 14.1 21.1 66.8 30 32 7.46 768-S
31 LS 13.0 20.5 63.2 13.2 20.4 65.2 32 37 7.43 761.2
32 REM 12.0 19.5 61.7 13.2 18.9 70.4 31 40 7.41 760. I
33 SW 13.8 21.4 64.8 14.t 7.40 762-S
-- -- --

34 LS 8 IO.5 20.7 1I.C 7.48 761.2

35 REM 10.7 19.8 12.4 7.47 760. I
SW 11.4 21.2 11.1 7-46 762-S
--
I
37 1 MHP 1 4 1 8.5’ 18.8 45.0i 7.6 r7.8’ 42.7; 22 1 29 1 7dr/ 758.5

1 In calculating the averages of values obtained during air breathing trials 2 and p were omitted
since these were second determinations upon subjects already studied at this inspired ~02. The
averages are therefore based upon single measurements in each of nine different subjects.

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 880 LAMBERTSEN, BUNCE, DRABKIN AND SCHMIDT Volume 4

In figure 2 the observed oxygen tensions are plotted against the spectrophoto-
metrically determined percentage hemoglobin saturations at the observed PH. For
purposes of comparison the findings are superimposed on the two most widely used
ifz vitro dissociation curves, those of Bock, Field and Adair (4) at a pCO2 of 40 mm.
Hg, and of Dill (5) at a serum pH of 7.40. Both of these curves were constructed from
the results of experiments upon the blood of one man, A. V. Bock (29). The second,
derived from later data used for establishing a curve at an erythrocyte PH of 7.10
(16), has been considered the more reliable because of the availability of superior
methods at the time the experiments were performed.
Our uncorrected in vie10 findings with the methods in which we have the most
confidence (the refined microtonometric estimation of arterial oxygen tension, the
spectrophotometric procedure of Drabkin and Schmidt (7) for arterial oxygen satura-
tion), as presented in figure 2, agree remarkably well with the in vitro curve of Bock,
somewhat less well with that of Dill, and indicate that any differences among in-
dividuals are small. Thus one of the major conclusions derived by Riley, Lilienthal,
Proemmel and Franke (IS) from their similar study is in general confirmed. However,
data obtained under circumstances producing alterations in pCOz or hydrogen ion
concentration in the blood cannot be directly compared with curves based upon
constant PH. It was therefore necessary to reduce the observed percentage oxygen
saturations to their corresponding values at PH 7.40 before the relationships could
be finally evaluated. This was done in accordance with the method of Dill et al. (16)
by converting the individual values and those of the oxygen dissociation curves at
PH 7.2, 7.4 and 7.6 in the Handbook of Respiratory Data in Aviation (5) to log pOz
and log Hb/HbOz in order to perform the necessary interpolation with greater ac-
curacy. The results of plotting the observed oxygen tensions against the oxygen
saturations reduced to serum PH 7.40 in our experiments and, for comparison, in
those of Riley et al., are shown in figures 3, 4 and 5. It is evident that the direction
and degree of scatter of the observations vary considerably depending upon the
methods used. It is also to be seen that in each instance the variability of percentage
hemoglobin saturation values is greatest in the steep portion of the dissociation curve,
possibly due in part to the greater influence in this region of small errors in the esti-
mation of ~02. The data obtained spectrophotometrically have the least scatter and
the highest position, being for the most part in close agreement with or even higher
than the in vitro curve of Bock. Although more of the manometric values lie above
than below the curve of Dill, the trend is less distinct than with the spectrophoto-
metric data. Finally, the microgasometric observations of Riley and his associates
show a definite tendency to fall below the in vitro curves. It is clear that studies thus
far do not permit a direct choice of one or the other of the in vitro oxyhemoglobin
dissociation curves as more representative of conditions in vivo. Nevertheless, we
tend to favor the higher curve for reasons which will now be discussed.

DISCUSSION
In the course of these experiments we made simultaneous measurements of
arterial oxygen saturation by the spectrophotometric and manometric methods in
nine different, resting subjects breathing air at sea level without respiratory appara-
tus. In table 2 it is shown that the average spectrophotometric value was 98.4 Z!Z0.6,
practically identical with the 98.6 per cent found by Drabkin and Schmidt (7), while
the manometric measurements averaged 97.0 & 1.3, only slightly lower than the
values of 97.4 & 2.1 obtained by Comroe and Walker (30) and Wood’s (31) finding

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 June 1952 ARTERIAL OXYGEN TENSION-SATURATION IN VIVO 881

of 97.9 =I=1.9. In each instance the indicated variability is the standard deviation of
the individual values. In our experiments the smaller standard deviation of the spec-
trophotometric data should indicate the maximum variability due to differences

I I
pO,,MM HG
0 I I
In 20 30 40 50 60 7A

100
Fig- 5
90

IpOI,MM HGI TOTAL EFFECTIVE GAS TENSION = pOe l M, CO YU HG

I I 0 I I
IO 20 30 40 50 60 70 80 90 100 IO 20 30 40 50 60 70 80 90 100

Fig. 2. RELATIONSHIPS OF SPECTROPHOTOMETRIC percentage oxygen saturation data at ob-

served pHand ~02 to the in vitro curves of Bock and Dill.
Fig. 3. RELATIONSHIPS OF SPECTROPHOTOMETRIC percentage oxygen saturation data at serum
PH 7.40 and observed ~02 to the in vitro curves of Bock and Dill.
Fig. 4. RELATIONSHIPS OF MANOMETRIC percentage hemoglobin saturation data at serum PH
7.40 and observed ~02 to in vitro curves of Bock and Dill.
Fig. 5. RELATIONSHIPS OF PERCENTAGE oxygen-carbon monoxide saturation data of Riley
et al. at serum pH 7.40 and observed ‘total effective gas tension’ (~02 + pC0) to in vitro curves of
Bock and Dill.

in the nature of hemoglobin in the group of nine normal men. Similarly, the difference
between the standard deviations of the percentage hemoglobin saturations obtained
simultaneously by the manometric and by the spectrophotometric methods should
indicate only the larger random error of manometric analysis and its associated
blood handling techniques.

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 882 LAMBERTSEN, BUNCE, DRABKIN AND SCHMIDT Volume 4

For the purposes of this study we intentionally took no special precautions to

reduce the errors described by Roughton (6) as inherent in estimating percentage
saturation by the gasometric method. Nevertheless the average of our manometric
observations during air breathing is nearly as high as obtained by Comroe and
Walker and by Wood who carefully minimized known sources of error. At our ob-
served average ~02 the average percentage hemoglobin saturations reported by these
independent workers fall above Dill’s dissociation curve. The most recent gasometric
studies thus appear to give saturation values consistently higher than the 95 to 96
per cent formerly in general acceptance while the spectrophotometric data average
0.7 to 1.6 per cent higher than the gasometric. In our experiments the difference
between the two averages (98.4 and 97.0) is significant and is greater than any of the
individual differences revealed by the spectrophotometric method during air breath-
ing. The cause of this discrepancy cannot be determined from our data. If, however,
the criticisms by Roughton (6) of the gasometric procedure are valid, our manometric
value should be considered about 2 per cent lower than the actual average percentage

TABLE 3. EFFECT OF HEMOLYSIS UPON BLOOD GAS TENSIONS

BLOOD
MEASUREMENT TRIAL Nonhemolyzed Hemolyzed
PC02 I 48 57
2 37 44
3 32 41
4 36 44

PH I 7-32 7.21
2 7.38 7.18
3 7.41 7.33
4 7-39 7-33

PO2 I 26 22
2 2s 2T
-7
8s
;: 93

saturation. There would then no longer be a significant difference between the

measurements obtained simultaneously by the spectrophotometric and gasometric
techniques, and the tendency of the latter values to fall above the curve of Dill would
be more definite.
That the spectrophotometric observations may be too high deserves serious
consideration, and three major possibilities suggest themselves. First, the spectro-
photometric measurements were carried out at room temperature (~5~-27~C.) rather
than at 37OC. The proportion of oxyhemoglobin to reduced hemoglobin should be
greater at the lower temperature than at the higher but the evidence now available
indicates that this is a negligible factor. Repeated spectrophotometric measurements
on the same blood at the high level of 98.4 per cent oxyhemoglobin showed no de-
tectable differences on changing the temperature from 38” to 28°C.; at a lower level
of saturation a similar change in temperature produced a shift from only 74.6 to
74.8 per cent oxyhemoglobin. This empirical evidence supports the concept that
fairly marked changes in temperature can have little effect upon the percentage
saturation of hemoglobin when the blood sample is kept from contact with a gas
phase.
Second, the use of hemolyzed blood for the spectrophotometric analyses m .aY
lead to an increase in the oxyhemoglobin content over that prevailing in the blood as

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 June I 952 ARTERIAL OXYGEN TENSION-SATURATION IN VIVO 883

drawn. Direct measurements showed that when arterial blood samples were hemo-
lyzed anaerobically in the manner employed in these experiments there occurred a
decrease in ~02, an increase in pCO2 and an acid shift in PH (table 3). Oxygen there-
fore must have been taken from physical solution, lowering the ~02 and raising the
percentage hemoglobin saturation in the sealed vessel. This apparent inconsistency
is referable to the fact that, after hemolysis, hemoglobin is exposed to a fluid environ-
ment more alkaline than that prevailing inside normal erythrocytes but more acid
than that which previously existed in the plasma (16). The actual shift therefore is
to a less acid environment in which the affinity of hemoglobin for oxygen would be
increased and the observed changes can be explained on this basis. Calculation of the
rise in saturation to be expected in a closed system from combination with an amount
of physically dissolved oxygen corresponding with a drop of about IO mm. Hg in
~0~ (table 3) indicates an increase of only 0.1 per cent in oxygen saturation. This
factor, like the preceding, seems quantitatively negligible. Whether the combined
effects of lower temperature and change in PH on hemolysis operate together to mag-
nify such changes in oxygen saturation was not tested, primarily because spectro-
photometric measurements on whole blood do not provide the necessary accuracy.
Finally, it has been suggested (32, 33) that blood may contain hemoglobin
molecules that behave like oxyhemoglobin in the spectrophotometer but are func-
tionally inactive. So far as we are aware, adequate proof of the existence of such sub-
stances has not been furnished, but of course this does not exclude the possibility.
On the other hand, such an explanation is at present merely a reaffirmation of com-
plete confidence in the gasometric method when it disagrees with the spectrophoto-
metric. Protagonists of the latter method may properly point out that the major
errors of the gasometric technique are due to a) the need for exact measurement
and delivery of the blood sample into the analytical apparatus, and b) the necessity
for performing separate determinations of blood oxygen content and capacity. Each
of these factors is eliminated in the spectrophotometric method.
The possibility thus remains that in the spectrophotometric procedure a number
of individually insignificant errors may act consensually to produce a systematic
over-valuation of the actual hemoglobin saturation. This would be analogous to the
gasometric method, though here the several errors cooperate to cause under-estima-
tion of the saturation (6). The errors have been identified in the gasometric but not
in the spectrophotometric procedure.
Our data up to this point indicate a very narrow range of values for normal ar-
terial hemoglobin saturation and oxygen tension in healthy young men at rest. It
should now be re-emphasized that only nonsmokers were chosen for the studies
described. Subsequent observations upon a separate group of subjects, each of whom
smoked approximately 20 cigarettes daily, reveal a very different state of affairs.
In table 4 are shown the data obtained from seven normal, male students (ages 19-30)
who regarded themselves as moderate smokers. While these measurements were per-
formed several years later than our original study of hemoglobin dissociation charac-
teristics, the methods of collection and analysis of blood samples were identical ex-
cept for estimation of oxygen capacity, which was determined spectrophotometrically
as total pigment by the cyanmethemoglobin method (24.) On the basis that 1.34 ml.
of oxygen combines with each gram of hemoglobin (34) percentage hemoglobin
saturation was calculated by the formula,
02 content - dissolved 02
Percentage saturation =
Total pigment (gm/Ioo ml.) X 1.34 ml. O/gm. Hb ’

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 884 LAMBERTSEN, BUNCE, DRABKIN AND SCHMIDT Volume 4

The data in table 4 s low that both the percentage hemoglobin saturation and
the ~02 of arterial blood of the smokers breathing air at rest are significantly lower
than found in the nonsmokers under the same conditions. Furthermore, the average
percentage hemoglobin saturation of 93.6 is lower by at least 2.5 per cent than that
corresponding to the in vitro curve of Dill for the observed average ~02 of 87 mm.
Hg. It is evident that in this group of subjects the percentage hemoglobin saturation
of arterial blood falls below generally accepted normal limits for two reasons, one of
which is the low arterial pOz; the other is uncertain but may be related to the presence
of excessive amounts of inactive blood pigments which were not measured. The
existence of such pigments in normal blood has recently been demonstrated by Van
Slyke and his collaborators (35). While an interference with oxygen uptake (low
~02) and oxygen transport (low saturation) exists in these men as compared with
the group of nonsmokers all had been selected as healthy subjects. Whether or not
there exists a direct relationship to smoking per se, we have no particular basis for
considering these men abnormal and any concept of a narrow range of normal values
for percentage oxygen saturation and oxygen tension appears unjustified.

02 02
TRIAL SUBJECT
CONTENT CAPACITY
y. SAT. PO2 PC02

_ -~ ~
vol. 70 vol. 70 mm. Wg mm. Hg
HR =7*7 18.7 93-S 87 45 7*36
AA 17.8 18.9 93-I 90 40 7.38
WS 18.9 19.3 93.3 84 37 7.38
JT 17.1 17.8 94.3 87 47 7-38
CS 19.0 19.9 94.2 91 37 7-40
BR 19.2 20.4 92.7 90 41 7-38
JD 18.2 19.3 94.3 80 38 7.33
Average 18.3 19.2 93.6 87 41 7.37
S.D. ho.8 ho.8 ho.7 zt4.0 zt4.0 zto.02

SUMMARY

To determine whether the characteristics of oxyhemoglobin dissociation in blood

equilibrated with alveolar air in the lungs are accurately portrayed by existing in
vitro curves, and whether major differences exist in the blood of different individuals,
the relationships of percentage hemoglobin saturation, pOz, pCO* and PH were de-
termined in 15 normal subjects by analysis of arterial blood collected during the
inhalation of gas mixtures ranging from 21 to 4 per cent oxygen in nitrogen. Per-
centage oxygen saturation was measured by both the spectrophotometric and the
manometric method in an attempt to evaluate apparent discrepancies between the
results obtained by previous investigators using one or the other of these techniques.
Improvements in the direct microtonometric estimation of blood gas tensions re-
sulted in elimination of the systematic error and diminution of the random error for
both pCOz and ~02.
No differences in the dissociation characteristics of oxyhemoglobin could be de-
tected in the 15 subjects of these experiments. Simultaneous measurements of per-
centage oxygen saturation and ~02 indicate closer agreement with the in vitro dis-
sociation cu .rve of Bock than with that of Dill . Our findings are supported by values
for normal, rest ing arterial percentage oxygen saturation recently reported by other

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.
 June 1952 ARTERIAL OXYGEN TENSION-SATURATION IN VIVO 885

investigators which are higher than can be explained by the dissociation curves of
Dill or by the data of Riley et al. Percentage oxygen saturation values obtained by
the spectrophotometric method were significantly higher and had a smaller random
error than those obtained by the manometric method upon blood sampled simultane-
ously. The previously reported large discrepancy between the spectrophotometric
and the manometric techniques has been reduced materially. The remaining disagree-
ment may be related to sources of consistent error in either or both of these methods.
The arterial blood of smokers was found to be significantly lower in both ~0, and
percentage oxygen saturation than that of nonsmokers.
The authors gratefully acknowledge the technical advice of Dr. William C. Stadie and Dr.
David B. Dill and the assistance of Dr. J. Harold Austin in the statistical evaluation of the data.
Permission to make use of his data was kindly granted by Dr. Richard L. Riley.

REFERENCES
I. DRABKIN, D. L. Haemoglobin (1st ed.). New York: Interscience, 1949, pp. 35-52.
2. LILIENTHAL, J. L., JR., R. L. RILEY, D. D. PROEMMEL AND R. E. FRANKE. Am. J. Physiol.
147: 199, 1946.
3. RILEY, R. L. AND A. COURNAND.J. Applied Physiol. I: 825, 1949.
4. BOCK, A.V.,H. FIELD, JR. AND G.S.ADAIR. J.Biol.Chem. 59:353,1924.
5. Handbook of Respiratory Data in Aviation. Washington: National Research Council, 1944.
6. ROUGHTON,F. J. W., R. C. DARLING AND W. S. ROOT. Am.J. Physiol. 142: 708, 1944.
7. DRABKIN, D.L. AND C. F. SCHMIDT. J.Biol. Chem. 157: 69, 1945.
8. KROGH, A. Skandinav. Arch. f. Physiol. 20: 259, 1907-08.
9. BARCROFT, J. AND M. NAGAHASHI. J. Physiol. 55: 339, 1921.
IO. FERGUSON,J. K. W. J. Biol. Chem. 95: 301, 1932.
II. COMROE, J. H., JR. AND R. D. DRIPPS., JR. Am. J. Physiol. 142: 700, 1944.
12. RILEY, R. L., D. D. PROEMMEL AND R. E. FRANKE. J. Biol. Chem. 161: 621, 1945.
13. ROOS, A. AND H. BLACK. Am. J. Physiol. 160: 163, 1950.
14. ROUGHTON,F.J. W. AND P. F. SCHOLANDER.J.BioZ. Chem. 148: 541, 1943.
15. RILEY, R. L., J. L. LILIENTHAL, D. D. PROEMMEL AND R. E. FRANKE. J. Clin. Investigation 25:
139, 1946.
16. DILL, D. B.,H.T. EDWARDS,M. FLORKIN AND R. W. CAMPBELL. J.BioZ. Chem. 95: IQ, 1932.
I S. S. AND C. F. SCHMIDT. J. Chz. Investigation
7. KETY, 27: 476, q48.
18. VAN SLYKE, D. D. AND J. M. NEILL. J. Biol. Chem. 61: 523, 1924.
19. PETERS, J. P. AND D. D. VAN SLYKE. Quantitative Clinical Chemistry (1st ed.). Balti-
more: Williams & Wilkins, 1932, vol. 2, p. 321.
20. CUMMINGS, A. J., J. K. CLARK, A. P. CROSLEY AND H. G. BARKER. J. Lab. & Clin. Med. 35:
164, 1950.
2 I.COMROE, J. H., JR. Methods in Medical Research. Chicago: Yr. Bk. Pub., 1950, vol. 2, p. 143.
22. SENDROY,J.,R.T. DILLON AND D.D. VANSLYKE. J.Biol.Chem. 105: 598,193~.
23. DRABKIN, D. L. AND J. H. AUSTIN. J. Biol. Chem. 112: 105, 1935.
24. DRABKIN, D. L. AND J. H. AUSTIN. J. Biol. Chem. 112: 51, 1935-36.
25. DRABKIN, D. L. In: Medical Physics, edited by 0. GLASSER. Chicago: Yr. Bk. Pub., 1950, vol.

26. MILLIKAN, G. A. Rev. Scient. Instruments 13: 434, 1942.

27. COMROE,J. H., JR. Methods in Medical Research. Chicago: Yr. Bk. Pub. 1950, vol. 2, p. 163.
28. SCHOLANDER,P. F. J. Biol. Chem. 167: 235, 1947.
29. DILL, D. B. Personal communication.
30. COMROE,J. H., JR. AND P. WALKER. Am. J. Physiol. 152: 365, 1948.
31. WOOD, E. H. J. Applied Physiol. I: 567, 1949.
32. AMMUNDSEN, E. .%ienCe 90: 372, 1939.
33. AMMUNDSEN, E. AND M. TRIER. Acta. med. Skandinav. IOI: 451, 1939.
34. PETERS, J. P. AND D. D. VAN SLYKE. Quantitative Clinical Chemistry. Baltimore: Williams &
Wilkins, vol. I, p. 524.
~cp,
35. VAN SLYKE, D. D., A. HILLER, J. R. WEISIGER AND W. 0. CRUZ.J.BioZ. Chem. IOO: 121, 1945.

Downloaded from www.physiology.org/journal/jappl at Lunds Univ Medicinska Fak Biblio (130.235.066.010) on February 14, 2019.