Ueber die Grösse des Hämoglobinmolecüls

by O. Zinoffsky

Hoppe-Seyler's Zeitschrift für Physiologische Chemie 10, 16 - 34 (1886)

Translated by Cathy Saltern (Feb. 1993) and revised by H. B. White and A. Reichert (Jan. 1994)

Background material on Zinoffsky's article


On the Size of the Hemoglobin Molecule

by Dr. O. Zinoffsky

(accepted for publication 29 July 1885)


The previous analyses of hemoglobin have led to the surprising result, that this compound of one iron atom and 600 carbon atoms, consequently contains at least 600 carbon atoms in one molecule. Hemoglobin is therefore incredibly more complicated than all the previously known chemical compounds. The given size is but a minimum. It is calculated on the assumption that hemoglobin contains only one iron atom. This assumption is discarded as soon as one proves that there is no clear cut stoichiometry between the iron content and the amounts of other elements. Should it be known that one iron atom is bound not to one sulfur atom, but a whole number and a fraction of a sulfur atom, the size of the hemoglobin molecule must be multiplied by the denominator of the fraction. We would be forced to think of this molecule as being much bigger. It is therefore important to determine the exact amount of iron and sulfur in hemoglobin. We would thus be given a glimpse of the size and complexity of protein molecules. Until now the iron and sulfur content have given very different results and no simple ratio between the two elements has been found. Due to these findings, it has been presumed that hemoglobin is not a chemical entity at all. Instead the iron-containing hematin simply is carried along mechanically with the crystallizing proteins.

If it is discovered through more exact analysis that there is a stoichiometry between sulfur and iron, this concern would be rejected and the chemical uniqueness of hemoglobin established. For these reasons an exact determination of these elements is important.

 In Table 1, I compile the results of previous determinations. These ratios are such that there are two possibilities;

1. The preparations used by the authors could be impure, or

2. The analytical methods were a source of error and the amount of substance used for the determinations were insufficient.

I decided I had to:

1. Make a preparation that everyone must accept as pure, and

2. After careful inspection of the available methods of analysis, to analyze these preparations.


Table 1. Previous Determinations of the Sulfur and Iron Content of Hemoglobin



Percent Percent  

Mean Value

Amount of

Incinerated Hemoglobin

  Sulfur Iron Sulfur/Iron For Sulfur For Iron
Dog Blood % %   g g
Schmidt 0.66 0.43 2.686 1.6637 1.6637
Hoppe-Seyler 0.375 0.45 1.60 1.9972 1.7062
" 0.448 0.42 - 1.4033 2.5422
" 0.359 0.42 - 1.8915 1.7855
Horse Blood          
Bucheler 0.6532 0.4670 - 1.8481 0.9430
" 0.6443 0.47238 2.427 1.8132 "
" - 0.46720 - - -
Kossel 0.65 0.47 2.42 - -
Otto 0.67 0.45 2.60 2.1876 1.0866


Purification of Hemoglobin

Hoppe-Seyler prescribes: To purify hemoglobin one must first separate the blood corpuscles from the serum. To this end the defibrinated blood was filtered and left standing for a long time so that the corpuscles can sink in the serum. The liquid above the corpuscles was decanted and replaced with a sodium chloride solution which should free the corpuscles from the serum proteins. Through letting it stand and decanting, the major part of the salt solution can be removed. Hoppe-Seyler says in this paper, "the washing of blood corpuscles using this method is not the easiest of jobs. The corpuscles, which sink easily in the serum, do not sink as well in the salt solution." In fact, the corpuscles need a long time to settle to the bottom of the container. It takes 3 to 5 days for the sinking to be somewhat complete, but after 3 days, inspite of the fact the work was being done in the cold, the corpuscles take on a venous color and there is a danger of decomposition. Therefore, it appears desirable to avoid washing the corpuscles. It is possible that a portion of the serum proteins can remain with the corpuscles without hindering the disruption of the corpuscles, the crystallization of oxyhemoglobin and the cleaning of the crystals. The following experiment answers this.

One volume of serum, three volumes of water, and one volume of alcohol barely become cloudy. Thus the serum contents remain in solution. From this I conclude that unwashed corpuscles can be used in this work. Oxyhemoglobin can then be separated later from the serum constituents by repeated crystallization. As it will be shown later, this modification does not affect the purity of the preparation.

The separation of hemoglobin from the stroma follows; here as well, experimentation had to show the proper path. Dilute the blood corpuscle solution with three volumes of distilled water and heat to 35oC. The solution will be varnish-colored, an indication that the oxyhemoglobin dissolved. The stroma are now suspended in the liquid as colorless discs. As one now knows, these discs cannot be removed by filtration because they obstinately adhere to hemoglobin crystals. So I set about to dissolve them.

Treat the varnish-colored solution which has been warmed to 35oC with a small amount of ammonia. The stroma dissolve. If one now mixes a sample of the liquid with a concentrated solution of sodium sulfate and some sulfuric acid, one no longer sees stroma in the microscope. The ammonia, whose amount must be known exactly, will then be neutralized with an exact amount of dilute hydrochloric acid because hemoglobin will not crystallize out of an alkaline solution.

The same result can be achieved with ethyl ether. My experiences show that smaller amounts than Hoppe-Seyler gives in his article are enough. Thirty-nine cubic centimeters of ether are enough to dissolve the stroma from nine liters of blood.

After this manipulation we have an impure watery solution of oxyhemoglobin. We have to crystallize oxyhemoglobin from this solution. For this purpose, cool the solution to near 0oC, add one quarter volume of absolute alcohol and allow to crystallize at 0oC. After about 24 hours crystals begin to form. It is desirable to allow the solution to stand for 48 hours because the crystals increase in size and do not clog the filter used to separate the crystals from their mother liquor. The crystals must now be washed with a mixture of one part alcohol and four parts water cooled to 0oC. After many trials on my part, I suggest that a three-fold volume of the alcohol-water mixture be added to the crystal paste, mixed, allowed to stand, and then decanted. For further cleaning, the crystal paste should be dissolved in water at 35oC, the solution filtered, and recrystallized after the addition of alcohol as described in the manner above.

How often should one wash the crystals and recrystallize in order to obtain an absolutely pure preparation? In order to answer this question many experiments were done.

As described above, an impure oxyhemoglobin solution was produced and the crystalline product was recrystallized twice. From each crystallization product and its mother liquor, a portion was analyzed. As a check of the purity, the iron content of both the crystals and the mother liquor were determined.

The percent iron content of the mother liquid and the crystals differed by 0.07925 in the first preparation, by 0.0376 in the second, and in the third the mother liquor contained almost the same amount of iron as the crystals. I give the calculations for this in the analytical section of this article.

After I had found a method to clean the crystals, I had to find a way to dry them. Hoppe-Seyler suggests the crystal mass be dried in a vacuum at 0oC. I attempted to dry them at the lowest possible temperature. The drying of 50 grams of hemoglobin took two months. This attempt taught me to place the crystal paste very thinly in flat containers so that it can dry in 8 hours at 18 to 20oC without decomposing.

Oxyhemoglobin dried in this manner dissolves clear in water. The solution gives a pure oxyhemoglobin spectrum and does not precipitate when basic lead acetate is added - an indication that no methemoglobin had formed. If the preparation had been dried at 40oC, it does not dissolve easily and completely in water.

These preparative studies showed the way to purify oxyhemoglobin. Now it is possible to proceed to the description of several preparations. I have completed three in all, for which I give the method for each. It should be mentioned that I used horse blood for all preparations because the blood corpuscles sink more readily, it is easier to get in large quantities, and the hemoglobin crystallizes more easily.


Preparation I

Twenty liters of horse blood (blood from two horses) were defibrinated to 10 liters and placed in the cold where after three hours the corpuscles have sunk to 1/4 of the volume. The serum was decanted, the corpuscle slurry mixed with 8 volumes of 2% sodium chloride solution and set aside. After three days, the corpuscles settled to the lower 1/4 of the container. The salt solution was decanted and the corpuscle slurry mixed with 3 volumes of distilled water and heated to 35oC. The not yet varnish-colored solution was then stirred with 16 cc of titrated 1/10 Normal ammonia solution. After 5 minutes the solution is varnished-colored. The ammonia was then neutralized by a corresponding amount of very dilute hydrochloric acid and the mixture cooled as quickly as possible. (Experience taught me that by cooling slowly methemoglobin forms.) To the near 0oC chilled liquid, absolute alcohol chilled to 0oC was added - one volume of alcohol to 4 volumes of solution - and everything put into a mixture of ice and salt. After three days the crystals were separated from the mother liquor, washed two times with a 0oC cooled mixture of 4 volumes of water and 1 volume of alcohol, then dissolved in 3 times the volume of water at 35oC, the solution filtered, cooled, alcohol added, and again put in the cold mixture. The resulting crystals were put through the same cleaning process twice and then dried under vacuum.

 Yield: about 200 g.


Preparation II

Ten liters of horse blood was used. The corpuscles obtained after defibrination, allowing the blood to stand, and decanting, were treated, not with salt solution, but immediately with 3 volumes of water at 35oC, and then treated like preparation I to gain pure crystals.

 Yield: about 520 g.

Preparation I yielded only 200 grams of hemoglobin from 20 liters of blood. This preparation gave 520 grams from 10 liters. I believe that washing with salt solution lessens the yield.


Preparation III

The blood corpuscles from 9 liters of horse blood were freed as in the former preparation from the serum without using the salt solution. The blood corpuscle solution was placed in 3 volumes of water, heated to 35oC, cooled, and 30 cc of ether added. One quarter volume of alcohol was then added to the clear solution. The resulting crystals were cleaned as before by recrystallization twice. Samples of each of the crystallization products and their mother liquors were analyzed for their iron content.

This preparation proved to be the purest of the three. The ashes contained almost no trace of chlorine. After incineration of about 2 grams with sodium carbonate, addition of silver nitrate to the sodium nitrate solution of the ashes produced a barely perceptible opalescence. Alkalis could not be found in the ashes. Phosphorus was certainly present but it could not be measured quantitatively in the ashes from 23.6 g. Hemoglobin contains only an immeasurable amount of magnesium pyrophosphate. Attempting to measure its mass yielded 0.0021 g. Calcium and magnesium oxide were found in even smaller traces.

Preparation I contained definite traces of chlorine, no alkalis, and calcium and magnesium could only be determined in the ashes from 60.8006 g. Somewhat higher amounts of phosphorus were found: the ashes from 60.8006 g yielded 0.0223 g of magnesium pyrophosphate which corresponds to 0.0143 g of pyrophosphate giving 0.0235% P2O5.

Preparation II was the least pure. It contained 0.0401% P2O5, 0.0097% CaO, and 0.0434% MgO. All preparations were examined microscopically while still moist. In this manner, no impurities were found. The contours of the crystals were sharp and straight. They were not frayed or pitted as is the case when stroma are present.

 Calculation of Water Content

After the prescribed drying at 40oC (See page 3), the preparations were put in an air stream with a Bunsen-Kemp'schem regulator at 115oC and dried until after 24 hours of drying there was no more weight loss.

Methods for Iron Determination

The methods of iron determination belong to the most reliable and reproducible which we possess. Despite that, I did not find it superfluous to check the exactness of the execution of the methods based on titration with potassium permanganate solution and compare it to gravimetric analyses. My honored teacher Dr. G. Bunge was kind enough to carry out the gravimetric analyses.

For each analysis, a large amount of oxyhemoglobin (in one case even 60 g) was used. The errors then must be very small. The reagents were completely iron-free.

For the titration, the hemoglobin was incinerated in a platinum crucible, the ashes dissolved in hydrochloric acid, the solution evaporated almost to dryness, taken up with sulfuric acid, reduced with zinc in the known method, and then titrated. The titrant was standardized with metallic iron.

Gravimetric analysis proceeded as follows: incineration of hemoglobin, dissolving the ashes in hydrochloric acid; this was neutralized with ammonia, then ammonium acetate added (the solution would become slightly acidic), heated until iron acetate and iron phosphate were no longer solid, filtered hot, washed with hot ammonium acetate, ignited the filtrate until it glowed, and weighed. The measured weight corresponds to the amount of iron oxide and iron phosphate. Dissolve the weighed amount in hydrochloric acid, add tartaric acid, then ammonia, precipitate the iron with ammonium sulfate, transform the iron sulfate to iron oxide in the normal way, and weigh the precipitate. From these ashes, calcium, magnesium, and phosphate were determined. From the filtrate of the iron sulfate, the phosphate and magnesium were precipitated. From the filtrate of the iron acetate and iron phosphate, calcium can be precipitated through saturation with ammonia and adding ammonium oxalate; the filtrate of calcium oxalate was evaporated, the ammonium salts removed, the small amount which was left dissolved in hydrochloric acid, and from this solution, the magnesium precipitated using ammonia and ammonium phosphate.

As the reader can see (Table 2), the numbers agree with each other, but not at all with those of other authors (Table 1). That preparations I and III are unquestionable pure, I believe I have proved

 Table 2. Results of the Iron Determination #

  Percent composition in iron of Crystals and mother liquor
Method Preparation I


Preparation II


Preparation III*


Preparation III*

Mother Liquor

Titrimetric a) 0.336% a) 0.327% a) 0.309% a) 0.230%
  b) 0.337% b) 0.327%+ b) 0.334% b) 0.2966%
  c) 0.334%   c)a ) 0.334% c) 0.347%
      b ) 0.338%  
      g ) 0.337%  
Gravimetric 0.330%+ 0.325%+ 0.334%+  

* Samples from Preparation III correspond to impure crystals(a), first crystallization product(b), and

second crystallization product(c).

+ Determined by Dr. Bunge.

# Note: These data were not presented in tabular form in the original article.

 through the documentation of their production and characteristics. The accuracy is guaranteed by noting the fact that Dr. Bunge and I came to the same results, and that the titrations gave the same results as the gravimetric analysis. Therefore, I have every reason to believe that mine are the correct numbers.



Table 3. Water Content of Oxyhemoglobin Preparations

  "Wet" Weight H2O H2O
  grams grams percent
Preparation I 2.0282






Preparation II 2.0139






Preparation III a) 2.1953

b) 2.0060

c) 1.9801







Preparation III

Mother Liquor

a) 2.1775


b) Water Free

c) 1.9958










 Table 4. Iron Content of Oxyhemoglobin #



Calc. Dry


Vol. of











Prep. I


a) 10.3058

b) 10.0430

c) 21.5152












mg Fe/cc




















Prep. II* a) 22.7538

b) 23.4728




















Prep. III a) 22.5587

b) 20.5787

c)a ) 11.0430

b ) 13.2016

g ) 9.9213






































Prep. III



a) 22.1770

b) 20.3522

c) 5.2869



















* Other gravimetric determinations on preparation II by Dr. Bunge include:

0.0290 g Mg2P2O7 = 0.0186 g P2O5 = 0.0409% P2O5

0.0165 g Mg2P2O7 = 0.0059 g MgO = 0.031% MgO

0.0107 g CaSO4 = 0.0044 g CaO = 0.011% CaO

# Note: Data not in tabular form in the original article.


Method for Sulfur Determination

When one looks at the determination of sulfur, especially those done on casein; when one sees how far apart the individual researcher's numbers are, claiming the others have used impure preparations, etc.; it is immediately clear that the differences stem not from the inexactness of the execution but rather the methods for sulfur determination probably have inherent errors. First and foremost, one must destroy the organic substance so that all of the sulfur is bound to the ashes. This has been attempted in many different ways, but one always comes back to an old method, fusing salt peter and caustic potash. From earlier analyses, we find that hemoglobin contains 0.65% sulfur and that in order to obtain a precipitate with enough sulfur to measure, we must use a larger quantity in our work. The calculation showed that 10.0 grams of sample would be enough.

The hemoglobin must be destroyed. How much caustic potash and salt peter one should use had to be determined by experiment. After multiple experiments, it was discovered that a mixture of one part hemoglobin, 7 parts salt peter and 3 parts caustic potash should be fused gently without flames and burned completely. The fused mixture was white with a touch of yellow, the color came from iron. It became apparent that the best way to go about it was to dissolve the oxyhemoglobin into the caustic potash solution, then add the salt peter, put it in a silver dish in a steam bath to evaporate and fuse. I must report a practice that everyone does when igniting such substances, but which is never reported in the literature. While heating the mixture of salt peter, caustic potash, and protein, I discovered that powerful noxious gases are produced before the igniting. It might be thought that a portion of the sulfur goes into the air with these gases. This is not the case. I placed a mixture of oxyhemoglobin and caustic potash in a flask, heated it, trapped the distillation product in aqua regia and analyzed it; there was no sulfur present.

The result of the fusing was a small amount of potassium sulfate in more than 100 grams of salt, which was composed of salt peter, potassium chloride, and potassium carbonate. The large amount of salt was quite unexpected. It was predicted that the precipitation of sulfur as barium sulfate must be done in at least one liter of liquid.

If it is done as Hammarsten did it with casein, oxyhemoglobin heated with nitric acid, a portion will be destroyed. Carbonic acid is given off, the remaining material dissolved in the acid to give a clear yellowish liquid, which, when evaporated, leaves behind a yellow substance. This substance can be melted with smaller amounts of salt peter and caustic potash. This results in a smaller amount of salt.

From the solution of the fused salt the sulfur should be precipitated as barium sulfate, but first the liquid to do that must be prepared. If one dissolves the fused substance, one can remove the iron in suspension by filtration. It must be done, because it carries over into further operations, it precipitates with barium sulfate and increases the weight of the precipitate. The alkali must be transformed to the chloride, and one must be absolutely sure the nitrate salts are destroyed because the barium precipitate brings the salt along with it. For this reason, the salt mass must be treated with concentrated hydrochloric acid 3 to 5 times. I have convinced myself of the complete decomposition of the salts in each analysis, in that I acidified small amounts of the saltmass with sulfuric acid and, after adding potassium-iodide-starch paste, treated it with zinc. If the starch turned a weak bluish color, the evaporation with hydrochloric acid had to be repeated. This resulted in an amount of potassium chloride in which a small amount of potassium sulfate could be formed. It remained to be investigated what influences a potassium chloride solution of known concentration could have on the results of the analysis. 0.1980 grams of sodium sulfate, 5 cc of 25% hydrochloric acid, and 50 grams of potassium chloride were dissolved in 500 cc of water, a hot solution of 0.62 g of barium chloride added, precipitated in the normal way, washed, heated to dryness, and weighed. This experiment was repeated twice.

 I weighed out barium sulfate 0.3220 g

II weighed out barium sulfate 0.3210 g

 That calculates to

 I 0.04422 g of sulfur

II 0.04450 g of sulfur

The theoretical calculated amount was 0.04458 g of sulfur.

As in these experiments, I have in all later experiments used a known amount of BaCl2. Also the amount of HCl was known. I chose known amounts of both reagents because too much BaCl2 precipitates while large amounts of HCl dissolve in the BaSO4. Potassium chloride can hardly hurt the experiment as we have seen. The only other point to be added, is that neither digestion of the heated precipitate nor the dissolving and repeated precipitation affected the weight greatly. The reagents used were sulfur-free. Large quantities of hydrochloric acid and nitric acid were evaporated and there was no evidence of sulfuric acid in the residue. Ten grams of caustic potash were fused with 10 g of saltpeter, the fused substance dissolved, made acidic with hydrochloric acid, and placed in a warm place after BaCl2 was added. There was only an unweighable trace of barium sulfate the next day.


Results of Sulfur Determination

Preparation I gave: 0.3902% sulfur

0.3916% sulfur

 Preparation II gave 0.3583% sulfur

0.3658% sulfur

 Preparation III gave 0.3899% sulfur

0.3881% sulfur



Table 5. Sulfur Content of Oxyhemoglobin #

  H2O Weight Calculated

Dry Weight





  % grams grams grams %
Preparation I 4.305










Preparation II 4.64










Preparation III 7.34










# Note: These data were not presented in tabular form in the original article.

 Carbon, hydrogen, nitrogen were also determined on Preparation III, which was the purest so that it could be compared with the preparations of other authors and a chemical formula calculated.


Determination of Carbon and Hydrogen

The preparation, dried in an air stream at 115oC, was ignited, as usual, in a stream of oxygen in the presence of lead chromate and copper chromate in a combustion tube and having a 25 cm long clean glowing hot copper spiral condenser.

It gave: I Carbon 51.2909% II Carbon 51.01%

Hydrogen 6.727% Hydrogen 6.79%


I 0.5500 g of oxyhemoglobin gave 1.0345 g of CO2 and 0.3330 g H2O = 51.2909% C and 6.727% H

II 0.5952 g of oxyhemoglobin gave 1.1133 g of CO2 and 0.3638 g of H2O = 51.02%C and 6.79%H


Determination of Nitrogen

 The method of Will-Warrentrapp'schen was followed and gave:

 I Nitrogen = 18.02% II Nitrogen = 17.87%



I 0.4683 g of oxyhemoglobin gave 0.526 g Pt = 18.02%N

II 0.2751 g of oxyhemoglobin gave 0.3452 g Pt = 17.87%N



 In the following summary tables the results from preparation II are omitted because, as mentioned, this preparation was least pure.

 Table 6. Summary of the Sulfur and Iron Content of Oxyhemoglobin

Results of Different Analyses on Preparation I Average
S 0.3902% 0.3916%       0.3909%
Fe     0.336% 0.337% 0.330% 0.334%


 Table 7. Summary of the Elemental Analyses of Oxyhemoglobin

  Results of Various Analyses of Preparation III, Percent Composition
C 51.29 51.01                 51.15
H 6.73 6.79                 6.76
N     18.02 17.87             17.94
S         0.3899 0.3881         0.3890
Fe             0.334 0.338 0.337 0.3344 0.3358


Average of both preparations.

 C 51.15%, H 6.76%, N 17.94%, S 0.3899%, Fe 0.335%, O 23.4251%

 We now calculate the number of sulfur atoms that come with an atom of iron. We find for preparation I:

 0.3909/0.334 = N x 32/56 = N x 4/7; N = 2.01`


For Preparation III:

 0.3890/0.3358 = N x 4/7; N = 2.03

 The average for all determinations:

 0.3899/0.3350 = N x 4/7; N = 2.03


From these numbers it is obvious that for every atom of Fe there are exactly 2 atoms of S in oxyhemoglobin and that oxyhemoglobin is a distinct chemical compound.

 Unfortunately, for other reasons it was not possible for me to calculate the amount of loosely bound oxygens, in order to decide if this element also has a simple stoichiometric relationship to Fe.

 I give finally the chemical formula for oxyhemoglobin resulting from my values.

 C712 H1130 N214 S2 Fe O245

When oxyhemoglobin decomposes, the hematin splits off with 34 carbon atoms. The rest can break into at most two protein molecules, because there are only 2 sulfur atoms. Each protein (globulin) molecule would contain one atom of sulfur and 339 atoms of carbon.

In fact the protein can be peptonized and the peptones still contain sulfur.

Let us assume that the peptones be cleaved products of the protein, then we must assume that the protein molecule is in two parts and to be sure one sulfur atom and 339 carbon atoms for every peptone molecule. If one assumes, however, that the protein contains at least two sulfur atoms bound in different ways and a portion of this tends to split off as H2S, then one is forced to account for at least 4 atoms of sulfur and 1356 carbon atoms in the protein molecule.

 I thank Prof. Dr. G. Bunge for the suggestion to do this work, his kindness and his support of me. I ask the professor to accept my warmest and deepest thanks for his efforts and inspiration.

Return to Hal White's Home Page, Course Home Page, or Departmental Home Page.

Last updated: 19 January 1999 by Hal White

Copyright 1999, Department of Chemistry and Biochemistry, University of Delaware