872 CHEMOTHERAPY 



3. Factors Related to the Medicament. We are strongly inclined to 

 believe that in the interpretation of the causes of reactions following 

 the administration of salvarsan (and we are employing this term to 

 indicate dioxydiamino-arsenobenzol, irrespective of the trade name 

 under which it is marketed), too great a stress has been laid upon errors 

 of technic and upon personal factors, and too little attention has been 

 paid to a study of causes related to the compound itself. In order to 

 intelligently discuss this aspect of the question it is necessary to refer 

 to the chemistry of the substance under consideration. Salvarsan is an 

 extremely complex substance, perhaps the most complex drug employed 

 in medicine. Before its final elaboration numerous intermediate products 

 must be prepared. The purity of the final compound will depend to a 

 certain degree upon the character of some of the precedent intermediate 

 substances. Even if these be in a pure state, certain impurities are apt 

 to develop during the process of reduction with sodium hydrosulphite. 

 Salvarsan is precipitated by ether out of a hydrochloric acid methyl 

 alcohol solution of the base; other substances in minute quantities may 

 be precipitated in addition to salvarsan. Inasmuch as salvarsan cannot 

 be purified by repeated crystallization, we cannot regard salvarsan as an 

 absolutely pure compound. A number of analyses made by us corrob- 

 orate this statement. Salvarsan does not give arsenic values quite up 

 to the theoretic amount, and, furthermore, there are elements present 

 which have no place in the chemical formula. We have found in our 

 analyses that salvarsan contains on an average 1 to 2 per cent, of 

 sulphur; our own product, arsenobenzol, likewise contains sulphur. The 

 sulphur doubtless becomes attached to the arsenic during the process 

 of reduction. 



Inasmuch as salvarsan is not an absolutely pure chemical substance, 

 the drug in powder form will vary to a slight extent in different lots. 

 Indeed, we doubt whether any two batches can be prepared which are 

 identical in all respects. We have already stated that different serial 

 products vary in their oxidizability. From numerous experiments 

 which we have carried out we are convinced that they also vary to a cer- 

 tain extent in therapeutic properties and to a greater degree in toxicity. 

 Unfortunately, these variations cannot at the present time be determined 

 by ultimate chemical analysis, but only by the biologic effects of the 

 drug. 



With these prefatory remarks in mind, let us discuss the relation of 

 variations in the drug to reactive phenomena following its use. Ehrlich 

 and his associates have repeatedly called attention to the liability of 

 dioxydiamino-arsenobenzol on exposure to air to undergo oxidation 



