Saturday, 27 May 2017

Spinach-iron data transformations: Boussingault (1872) to Berg (1913)

For an introduction see any other entry under the spinach-iron label of this blog

In 1872, Jean-Baptiste Boussingault ("Du fer contenu dans le sang et dans les aliments." Comptes Rendus de l'Académie des Sciences, Tome 74: 1353-1359) published his results on the iron contents in the blood of various animals and in food products. This publication contains a table at pp. 1355-56 listing the contents of "Fer exprimé à l'état métallique. Dans 100 grammes de matiére." This means that the values are not for iron oxide (Fe2O3), which was the usual state in which iron contents were measured but elemental. The second statement ("Dans 100 grammes de matiére.") was specified in the text above the table: "En ce qui concerne les aliments, les dosages ont éte exécutés à l'état où ils sont consommés, c'est-à-dire avec leur eau constitutionnelle." (In the case of food, the dosages were carried out in the state in which they were consumed, that is to say with their constitutional water.) Hence, Boussingault's table gives the iron content for spinach leaves ("Feuilles d'épinards") as 0.0045g per 100 gram fresh matter.

In 1897, Emil Häusermann ("Die Assimilation des Eisens. Zeitschrift für Physiologische Chemie 23: 555-592) published a table (pages 586-588) listing the iron contents of food products per 100g dry matter. He cited Boussingault for various items, but the values vary.

Boussingault (g per 100g fresh matter)     Häusermann (mg per 100g dry matter)
Riz: 0.0015 Reis: 1.7
Haricots blancs: 0.0074 Weisse Bohnen: 8.3
Lentilles: 0.0083 Linsen: 9.5
Pommes de terre: 0.0016 Kartoffeln: 6.4
Feuilles d'epinards: 0.0045 Spinat: 39.1

Obviously, Häusermann took values for the water contents of the food stuffs from elsewhere, in order to calculate the iron contents in the dry matter from Boussingault's values for fresh matter. Apparently, these values of the water contents of the fresh matter were 11.7% in rice, 10.8% in white beans, 12.6% in lentil, 75.0% in potatoes and 88.5% in spinach leaves. Bunge (1892. "Weitere Untersuchungen über die Aufnahme des Eisens in den Organismus des Säuglings." Zeitschrift für Physiologische Chemie 16:173-186) has done the same with Boussingault's data before and given König (1889. "Chemie der menschlichen Nahrungsmittel.") as his source for water contents (see footnote ****, here). As Bunge took the water content for spinach to be 88.49%, we can conclude that König has also been Häusermann's source.

In a later edition, Josef König (1904. Chemie der menschlichen Nahrungs- und Genussmittel, Band 2, p. 353) cited the value of Häusermann properly, that is, he gave the range of values as 32.7-39.1mg per 100gr dry matter (the first value being from Bunge 1892 and the second from Häusermann 1897).

Ragnar Berg (1913. Die Nahrungs- und Genussmittel. p. 34-35) gave the contents of iron-oxide (Fe2O3) in 100g fresh matter as he explained in the introduction at page 6: "Damit nun jeder leicht umrechnen kann [...], habe ich in den folgenden Tabellen [...] den Gehalt von 100 g frischen Nahrungsmitteln an einzelnen Mineralbestandteilen in Grammen [...] aufgeführt."
     Berg (1913, p. 34-35) cited König (1904) with a value of 0.0596g Fe2O3 in 100g fresh spinach. (Berg indicated the sources by superscripts given above the values in the table. As he explained at page 11 of the introduction, the roman numeral I stands for König 1904.) If we assume that Berg took the average value (35.9mg/100g dry matter) of König's range (32.7-39.1mg/100g dry matter), then König's average iron (Fe) content in dry matter amounts to 60% of the iron-oxide (Fe2O3) content that Berg imputed to König for fresh matter.
    Berg can hardly have assumed that 60% of fresh spinach leaves were dry matter, when Häusermann had earlier taken its water content to amount to 88.5% (see above). Berg's transformation factor lies much closer to the 70% that is suggested as the correction factor needed to calculate the portion of the mass of Fe2O3 that is due to the iron in it according to the atomic weights (Fe: 55.8; O: 16). Nevertheless, a discrepancy of 10% remains. That is still not satisfying to see how Berg got from König's range of iron contents for dry matter to his imputation to König of iron-oxide content for fresh matter.

Anyway, Berg also performed his own analysis and that yielded 0.0437g Fe2O3 per 100g fresh matter. As it happens, this value was just about ten times higher than what Boussingault had started with (0.0045g Fe per 100g fresh matter) despite the fact that no decimal separator had been misplaced in any of the various data transformations.

Carl von Noorden & Hugo Salomon (1920. Handbuch der Ernährungslehre. Erster Band, p. 476) gave a range of 44-60mg Fe2O3 per 100g spinach and cited Berg (1913) as well as Hermann Schall & August Heisler (1917. Nahrungsmitteltabelle. 5. Auflage. Curt Kabitzsch Verlag, Würzburg) as sources.

The publication of Schall & Heisler (1917, not online) in turn has two values for spinach at page 41: 60mg and 44mg per 100g fresh matter. The latter value bears a footnote referring that value to "R. Berg" (sic), who had given 0.0437g/100g fresh mater as the result of his own analysis. As Berg also gave 0.0596g/100g fresh weight and referred that to König (1904), I presme that Schall & Heisler have simply taken that value from Berg as well, rounded it, but did not specifically cite König, because they have gotten it from a secondary source. The introduction of Schall & Heisler 81917) states that they collected data from "König, Rubner, Atwater und Byrant, Schwenkenbecher, Sautier, Strauss, Tischler, Leva, v. Noorden, Nauny, Magnus-Levy, Janney, Walker Hall, Brugsch, Bessau und Schmidt, Hesse, Offer und Rosenquist, Vogel, Berg, Albu-Neuberg, das "Deutsche Bäderbuch", die Angaben der Nahrungsmittelindustrie u. a. mehr." Sic! No sources, journals, publishers, years or anything else to ease retrival. The publication contains no reference list either to look up citations.