STUDIES ON HEMOGLOBIN.

I. The Advantage of Alkaline Solutions for Colorimetric

Determination of Hemoglobin.BY

HSIEN WU.

(Fromthe Laboratory
of Physiolugbeal
Chemistry,
Pelainy
UnionMedicalCollege,
Pekinq.)
(Receivedfor publication,October10, 1922.)
Methods for the determination of hemoglobin based on the com
parison of the color of oxvhemoglobin (Gowers, 1879), caiboxy
hemoglobin (Hoppe-Seyler, 1892; Haldane, 1900-1; Palmer,
1918), methemoglobin (MclEllroy, 1920), cyanheuloglobin (Stadie,
1920) and acid hematin (Sahli 1909, Cohen and Smith, 1919;
Robseheit, 1920) have all been proposed. Of these the first
named method, on account of the difficulty of preparing a suitable
and stable standard, has long passed into disuse. The carboxy
hemoglobin method and the acid hematin method have the prestige
of long continued use, and in the form recommended by Palmer
(1918) and Cohen and Smith (1919) respectively they seem to meet
ordinary needs of clinical hematometry. The methemoglobin and
cyanhemoglobin methods were proposed only recently for the de
termination of total hemoglobin in bloods containing methemoglobin
and they do not seem to have been put to the test by a general
use.
In our study of the peroxidase activity of the blood* we had to
determine the total hemoglobin. For this purpose, the otherwise
excellent method of Palmer cannot be used, as methemoglobin is
not changed to carboxyhemoglobin by carbon monoxide. Since he
matin can be obtained from all forms of hemoglobin, the acid lie
* See the followingpaper.
174 Hsien Wu:

matin method is theoretically the most general method for hemo

globins. Stadie (1920) states, however, that the methemoglobin
does not form with acid a product which can be colorimetrically
compared with that formed from oxyhemoglobin. We have found
also that the acid hematin solutions prepared from equivalent
amounts of oxyhemoglobin, carboxyhemoglobin, methemoglobin and
cyanhemoglobin give difforent color readings. We are not prepared
to say whether the difference in reading is due to a chemical dif
ference in the acid hematius or due t) other causes, but it is certain
that the acid hematin method cannot be used for the determination
of tot,ul hemoglobin. The acid hematin method even in oxyhemoglo
bin determinations is subject to several sources of error, which we
wish ti point out in this connection.
Acid hematin solutions freshly prepared from blood are crystal
clear, but the standard prepared from the concentrated stock solu
tion is always turbid. This turbidity renders the reading difficult
and may cause :in error of from several per cent to 10 per cent
or more depending on the visual habit of the reader. Terrill
(1922) in a recent paper proposed the u4e of acid hematin protein
powder and highly concentrate l solutions kept in ampules which
except for their being cumbersome would seem to considerably lessen
the chance of error in the method.
In applying the acid hematin method to pure hemoglobin so
lutions we canoe to observe another source of error. When the acid
hematin color prepared from blood is compared with that of a. so
lution prepared from an equal amount of pure hemoglobin the latter
is found to be about 20 per cent too weak. The color of this so
lution can be increased by adding plasma, and when 1 cc. of plasma
is added the color of the solution matches with that prepared from
the blood. When a washed corpuscle solution is used instead of
pure hemoglobin solution the result is just the reverse, the color
being descreased by the addition of plasma. The following ex
periment may be cited.
Washed sheep corpuscles were laked with 8 volumes of water
and 1 volume of 5 per cent alumina cream was added. After
 Studies on Hemoglobin. 175

thorough shaking the mixture was filtered. The filtrate was per

fectly clear, it should contain no protein other than the hemoglo

bin. 5 cc. portions of this solution were measured into a series

of 100 cc. flasks, different amounts of plasma were added, the so

lutions diluted to about 90 cc., 1 cc. of conc. HC1 added, made up

to volume, mixed, and the color compared after 40 minutes, with

the following results:

Pure hemoglobin, as standard, set at 20mm.

Same +025 cc. sheep plasma reads 17.6

•d +0.50 16.5

•d +1.00.•d•d•d•d 16.1

•d +1.50•d•d•d•d16.0

•d +20)•d•d•d•d 15.9

The same proce lunre with corpuscle solution which was nut

treated with alumna cream gave on the contrary the following

results

Corpuscle solution. as standard, set at

Same +0.25 cc, sheep plasma reads 30.4

•d +0.50 •d•d•d•d20.8

•d +1.00•d•d•d•d2l 2

It is thus seen that the color of the acid hematin solution is

greatly influenced by the plasma and by the stroma of the cor

puscles, that is by the proteins and lipoids of these components of

the blood. Since hematin is insoluble in N/10 HCl, the solution

prepared from hemoglobin must be colloidal and it is a matter of

common knowledge that the state of aggregation of a colloid, and
consequently its color, is greatly influenced by proteins and lipoids.
The color of an acid hematin solution prepared from the blood would
thus depend on the interplay of the protein and lipoid content of
the plasma and the corpuscles.
We have found that the addition of 0.5cc. of plasma to 1 cc.
of blood does not appreciably affect the color of the acid 1lematin
solution, and probably in all normal bloods the variation of the
176 Hsien Wu:

protein and lipoid contents is not sufficient to cause an error.. But

with bloods of abnormal composition, the error may be considera
ble. Cob en and Smith reported some instances where they found
a wide variation between the figures obtained by the acid hematin
method and those obtained by the oxygen capacity method. We
would suggest the variation of the protein and lipoid contents as
a probable explanation of that discrepancy.
Now the two sources of error inherent in the acid hematin
method are both removed if the color comparison is made in alkaline
solution. By making the solution alkaline the turbidity of the old
standard solution disappears. At the same time the plasma and
strorna proteins and lipoids cease to have any appreciable influence
on the color. If in the above mentioned experiment 50 cc. of the
acid hematin solutions containing different amounts of plasma were
made alkaline by adding 5 cc. of 10 per cent sodium hydroxide,
the colors of the solutions became identical.
As already mentioned above, the colors of the acid hematin
solutions prepared from different hemoglobins are quantitatively
different, but this difference also disappears when the solutions are
made alkaline by the addition of sodium hydroxide. The following
experiment bears out this point.
1 cc. portions of sheep blood were measured into a series of four
100 cc. volumetric flasks. To the first flask 20 cc. of 4 per cent
ammonia water were added, the solution was saturated with coal
gas, made up to volume with water and mixed. Ammoniacal
methemoglobin and cyanhemoglobin solutions were prepared in the
second and third flasks as described below. To the fourth flask 20
cc. of ammonia water were added, the solution was made up to
volume and mixed. 1 cc. of concentrated hydrochloric acid was
then added to each flask, mixed, and the acid hematin color was read
after 40 minutes. 50 cc. portions of each solution were measured
aut, 5 cc. of 10 per cent sodium hydroxide were adde I and the color
of the alkaline hematin was compared. The results are shown in
Table I.
 Studies on Hemoglobin. 177

TABLE I.
Color Value of Acid and Alkaline Hematin solutions prepared from
different forms of Hemog'obin.

From the above findings a new method based on the com

parison of alkaline hematin has emerged. When sodium hydroxide
is added to a blood solution the hemoglobin is first changed to
methemoglobin and the alkaline hematin is formed only after a
long time. An alkaline hematin solution is readily prepared,
however, by making an acid hematin solution alkaline. The
method is simply as follows: To 1cc, of blood in a 100cc.
volumetric flask add 80cc: water and 1 cc. clue nitrated hydro
chuloric acid. Allow to stand 30-40 minutes, add 10 cc. 10 per
cent sodium hydroxide, m!rke up to volume and mix.
The standard solution can be prepared from the stock acid
hematin solution prepared according to C o lien and Smith. Mea
sure 5 cc. of the 20 per cent blood solution into a 100 cc. flask,
dilute to about 80 cc., add 1 cc. concentrated hydrochloric acid,
10 cc. of 10 per cent sodium hydroxide, make up to volume and
mix.
The methemoglobin method.-This method in the form proposed
by McEllroy does not yield satisfactory results. The neutral
methemoglobin solution prepared from blood by adding ferricyanide
is turbid, even when it is fresh. The color is also too light, so
that a 2 per cent solution of blood has to be used instead of the 1
per cent solution employed in other methods. By making the
178 Hsien Wu:

solution unmoniacal, the color becomes much darker so that 1 cc.

of blood can be diluted to 100 cc. At the same time the turbidity
disappears, and accurate color readings can be made. The am
moniacal methemoglobin solution is prepared as follows. To 1 cc.
of blood in a 100 cc. volumetric flask add 20 cc. of 4 per cent
ammonia water, and 0.25 cc. of 4 per cent potassium ferricyanide.
Allow to stand 30 minutes, make up to volume, mix and read.
The cyanhemoglobin mehod. -This method in the form proposed
by Stadie is open to the same objection as McEllroy's method.
The turbidity of the freshly prepared solution is only slight, but
it grows on standing and within a week or two a sediment may
T,e formed. The addition of ammonia keeps the solution clear. The
ammoniacal cyanhemoglobin solution is prepared as follows : To
Ice. of blood in a 100 cc. volumetric flask add 20 cc. 4 per cent
amtconia water and 0.25 cc. 4 per cent potassium ferricyanide.
After 20 minutes add 2.5 cc. of 0.1 per cent potassium cyanide,
make up to volume and mix. The reading may be taken after
10 minutes.
We have found the three methods just described to give
entirely satisfactory results. The solutions being crystal clear, the
color reading can be made with the maximum degree of accuracy
w whichcolorirnetry permits.* The colors of solutions prepared from
pure hemoglobin solution and from blood of the same hemoglobin
content are the same, and it need hardly be stated that the readings
are strictly proportionate to the concentration of the hemoglobin
when the standard and the unknown are not more than 30 per cent
apart.
The ammoniacal methemoglobin solution keeps only for 1-2
weeks but the ammoniacal cyanhemoglobin solution keeps for at
least a month at room temperature. By adding sodium fluoride
the solnteILa may lie preserved for longer periods but this is hardly
necessary as they can be readly prepared fresh from stock standard
It shouldperhaps be mentionedthat the ammoniacalearboxyhemoglobin
solution in Palmer's method is also clear, but, as alreadypointedout, this
methodcannot be used for the determinationof total hemoglobin.
 Studies on Hemoglobin. 179

solutions. The alkaline hematin solution does not keep as the

hematin is changed to hematoporphyrin, and it should be prepar
ed fresh from a stock standard solution. For general use we
prefer the ammoniacal cyanhemoglobin method and others may
prefer to read the ammoniacal metheinoglohin. Individual aptitude
for color reading must decide the choice. But the alkaline
hematin method is the only suitable method for the determination
of total hemoglobin in chicken blood. The sodium hydroxide
dissolves the stroma of the chicken corpuscles which would give
so much turbidity with other methods that the solution cannot be
read.
Sbck Standard Solutions.-In the course of this study we have
tried to preserve hemoglobin solutions with various organic and
inorganic reagents and the only substance which answers our
purpose is sodium fluoride. A 10 per cent solution of blood or a
corresponding solution of pure hemoglobin can be preserved with
4 per cent sodium fluoride for at least one month at room tempera
ture and probably much longer is the cold. The sodium fluoride
does not interfere with the procedure of the determination by any
of the three above methods. The stock standard solution is pre
pared as follows: Measure into a liter volumetric flask 100 cc. of
blood the hemoglobin content of which has been determined by
the oxygen capacity method. Add 800 cc. of water and 40 grams
of sodium fluoride, shake until all the fluoride is dissolved, warming
to 35-40° if necessary. Cool, make up to volume, mix and filter
if necessary. In a few days all hemoglobin is changed to metlie
moglobin in the fluoride solution, but this is of consequence. A
2 per cent solution of blood prepared according to Palmcr 1 inay
also be used as a stock standard solution as the carboxhemoglobiu
is readily changed to methemoglobin, cyanhemoglobin and leuuatim.
To prepare a standard solution for use meeuore 10cc of the 10
per cent or 5cc. of the 20 per cent blood elution into is 100cc.
volumetric flask and proceed as already deseribed.
180 Hsien Wu.

SUMMARY.

Sources of error of the acid hematin, methemoglobin and

cvanhewoglobin methods for the determination of hemoglobin are
pointed out. New methods based on the colorimetric comparison
of alkaline hematin, ammoniacal methemoglobin and ammoniacal
cyanhemoglobin are proposed.

REFERENCES.
Cohen, B., and Smith, A.H. (1919): J. Biol. 39, 489.G
owers. W. R. (1879): Tr. Clin. Soc. London, 12, 64.
Haldane, J. (1900-01): J. Physiol., 26, 497.
Hoppe-Seyler, F. (1892): Z. physiol. 16, 505.
McEllroy, W. S. (1920): .J. Biol. Chem., 42, 297.
Pa1mer, W. E. (1918): J. Biol. Chem., 33, 119.
Robscheit, F. S. (1920): J. Biol. Chem., 41, 209.
Sahli, H. (1909): Klinische Untersuchungsmethoden, Leipsic and Vienns 5th,
Edition, 845.
Stadie, W. C. (1920): J. Biol. Chem. 41, 237.
Terri1, E. H. (1922): J. Biol. Chem., 53, 179.