PHYSICO-CHEMICAL PROPERTIES 79, 



negati\e reactions \vere obtained. In calculating the percentage 

 of reco\ery of activity, the maximum error of titration \vas used. 

 For example, when injected intra\enously, a certain filtrate (be- 

 fore dialysis) elicited positi\e reactions in prepared skin sites at 

 a dilution 1:100 and negative reactions at a dilution 1:200. The 

 reacting titer assigned to this filtrate was, therefore, 150 ± 50 

 units per c.c. The limits of the error thus were =*= 50/150 or =t 33 

 per cent. After dialysis, the filtrate gave positive results at dilu- 

 tions 1:91 and negative results at dilutions 1:140. The titer com- 

 puted was then 1:116 ± 25 reacting units per c.c. In this in- 

 stance, the limits of error ^\ere then "^ 25 116 or ±22 per cent. 

 During the dialysis of this filtrate, the volume had increased from 

 36 to 79 c.c. The percentage of reco\ery, therefore, was: 



(116) (79) (100) 



-7 : — i—Fi ^ 170 per cent =^ 57. The error =•= 57, ap- 



(150) (36) ' ^ ^' .'/I 



plied to the 170 per cent reco\ery, ^vas obtained by taking 33 per 



cent of the latter, i.e., the maximum of the titration errors before 



and after dialysis. It is realized that this does not represent the 



maximum possible error. 



Before dialysis, the same filtrate preparation was used for both 

 the skin-preparatory and intra\enous injections. After dialysis, the 

 fluid remo\'ed from the bag Avas titrated to the end-point for re- 

 acting potency in rabbits prepared by single intradermal injec- 

 tions of the original non-dialyzed filtrate. Since the titer of filtrates 

 may change after long periods of time, care was taken to ascertain 

 their potency just previous to dialysis and to titrate the dialyzed 

 solutions as soon as possible thereafter. 



It ^vas observed that after dialysis the B. txpJwsus and B. coli 

 solutions remained clear, ^\•hereas the meningococcus prepara- 

 tions sometimes formed slight precipitates. The decrease of pH, 

 Avhich practically ahvays occurred, suggested the presence of sub- 

 stances whose isoelectric points were on the acid side. The biuret 

 and Molisch tests were usually positi\e. After remo\ ing an aliquot 

 sample for analysis the solutions ^vere adjusted ^vith saline to 

 approximate normality (physiological) and stored in the ice 

 chest. Even after six months they had remained clear. 



Inasmuch as the active principles of the phenomenon were con- 

 sistently retained inside of "Cellophane" bags, the extent of 

 jjurification attained by means of dialysis was significant. The pmi- 

 iication thus eff^ected amounted, on the average, to about a three- 



