Anthocyanins and Anthocyanidins. 



153 



salt, but a mixture of magnesium salts of the anthocyan and flavonol. Not 

 only so, but magnesium acetate is appreciably soluble in alcohol, and is 

 precipitated from its solution by means of ether. Their extraction of the 

 precipitate with ether — to remove myricetin — would be of no avail, as the 

 myricetin would be present, not free, but combined with the metal. Exactly 

 similar results have been obtained by the present authors by mixing a trace 

 of anthocyan with a flavonol in alcoholic solution, adding magnesium acetate, 

 and precipitating and washing with ether. The green and greenish-blue 

 pigments of Shibata, Shibata and Kasiwagi can be matched in this way with 

 ease : they do not represent pure compounds, and Shibata, Shibata and 

 Kasiwagi's analytical figures are of no value. 



An important case of inconsistency arises in connection with the state- 

 ments concerning their complex compounds. They argue that the blue 

 anthocyan colours are complex metallic salts, and describe, as mentioned 

 above, the preparation of what they consider to be these complex salts, but, 

 on p. 211, they state that the products which they prepared " are soluble in 

 water as well as in alcohol," and, on pp. 217, 218, say of their green 

 pigments that they are soluble in alcohol as well as water. Despite this, 

 they say on p. 213 that the " sparing solubility of many blue or bluish- violet 

 anthocyanins in alcohol is quite consistent with the magnitude of the 

 complex molecules."" This latter statement would indicate that their 

 compounds were not the same as the natural pigments. 



The brownish red crystalline powder obtained by them (p. 218) was 

 undoubtedly a mixture of flavonol and anthocyan, for the green or blue 

 substances from which it was prepared must have contained the magnesium 

 salt of myricetin. 



It should further be noted that they formulate their complex salts as 



MeX 

 I 



O 



containmg a central group of the type : j jj whereas there is very 



/\/\ 

 OH 



considerable chemical evidence (see below) that at least in the case of the 

 salts with ferric chloride, the pyrylium ring is correctly represented thus : 



Cl-FeCIj 



I 



O 

 H 



N 2 



