AND POLYHYDRIC ALCOHOLS 171 



Before developing the argument further, it may be men- 

 tioned that the above speculations assume the ketonic re- 

 sidues to be attached so as to form five-membered rings. So 

 far there is no rigid proof of this, but the results obtained in 

 the study of the fructose-acetones are most easily explained on 

 this assumption. Further, if we accept Fischer's view that 

 ketones condense with ft rather than with a carbon atoms a 

 series of contradictions arises. Thus the arrangement is 

 impossible in the case of glucosediacetone, and similarly the 

 ketonic residues in mannitol triacetone cannot be arranged in 

 accordance with this idea, as at least one pair of a carbon 

 atoms must participate in the condensation. The simpler 

 view is that this type of condensation is symmetrical, wherever 

 possible, so that mdnnitol triacetone becomes : 



2 



!HO 

 O 



A 



CH< 

 CHO\ 



I >' 



/ITT (\/ 

 OlloU 



If A lt A 2 , A 3 represent the acetone residues, it is seen that, 

 considering the configuration of mannitol, A 2 must of 

 necessity be a trans grouping, applying this expression in the 

 sense already indicated. On the other hand, A^ and A 3 may 

 be either cis or trans. It does not necessarily follow that in 

 each of these cases the more stable cis linking will result, as 

 the relative positions of the terminal hydroxyl groups may 

 conceivably be affected by the configuration of the remaining 

 asymmetric systems, and they may thus react preferentially 

 in one or other of the two alternative positions. Our results 

 are in fact in agreement with this view. 



