I9I5] Arthur W. Thomas 393 



The nature of the linkings between the amy logen groups in the 

 starch molecule is deduced through the fact that, in the formation 

 of soluble starch or amylodextrin, hydrolysis takes place without 

 the formation of reducing Compounds, from which it is concluded 

 that a carbinol hydrolysis has taken place; hence the amylogens 

 must be connected one to another in an ether-like or carbinol 

 manner. 



The next step was the construction of the constitutional f ormula 

 for the starch molecule and incidentally to set limits by geometrical 

 means for the possible sizes which the starch molecule may assume. 

 The fact that after i8-hydrolysis, molecules of only a single dextrin 

 of two Ci8-groups are obtained is proof that the carbinol bonds 

 contained in these molecules, and the carbinol hydroxyls of the 

 amylogen residue which bring about the union, are of equal value. 

 These three hydroxyls must occupy similar places in the amylogen 

 molecules. If we connect the middle points of these places we un- 

 consciously form an equilateral triangle on each corner of which is 

 located the dextrin carbinol bond. 



Since the three dextrin-carbinol bonds of each amylogen must 

 be equivalent to similar bonds of all other amylogens, we assume 

 that this condition will find expression in the parallel position of 

 these bonds in the starch molecule. The triangles, which we have 

 assumed to be the configurations of the amylogens, must be ar- 

 ranged in the starch molecule so that every two neighboring groups 

 will have similar, opposite, positions. 



On the ground of the above argument, in the construction of 

 the starch molecule, only so many amylogens take part as the num- 

 ber of equilateral triangles that can come together in a closed fig- 

 ure in which every two of them shall always have an adjacent angle 

 and similar opposite positions. There are only four geometrical 

 constructions, according to Synkiewski, which can accommodate 



these premises. 



Place two triangles atop each other and we have two superim- 

 posed triangles. Place a triangle at each edge, and have all tri- 

 angles possess a side in common, and we arrive at a tetrahedron. 

 Again, by similar processes we may arrive at an octahedron ; finally 

 at an icosahedron. This hypothesis allows only four alternatives 



