INTRODUCTION 2'] 



The rabbits used for titrations each receix e a single intradermal 

 injection of 0.25 c.c. of the undiluted filtrate or filtrate diluted 

 1:2 and di\ided into groups of 3. Twenty-four horns later, a 

 single intra\enous injection of the filtrate diluted in 0.85 per 

 cent NaCl solution is gi\'en to each rabbit. The dose is 1 c.c. per 

 kilo of body ^veight. Each group of 3 rabbits receives intrave- 

 nously a different dilution of the filtrate. The local reactions are 

 read foiu" to fi\e hoins after the intravenous injections. The titra- 

 tions are carried imtil the lowest dilution is found which gives 

 no reactions in the 3 rabbits tested as ^vell as the highest dihuion 

 \vhich gives reactions in one or more rabbits of the group. The 

 minimal dose of reacting factors (reacting unit) is then consid- 

 ered as lying between these two figures. If a given filtrate is em- 

 ployed for any length of time, repeated control titrations are 

 done. In these control tests the dilutions employed are both the 

 highest dilution capable of eliciting reactions and the lowest dilu- 

 tion giving no reaction. It may be of interest to point out here 

 that provided one area of the skin is prepared, the amoimt of 

 reacting factors necessary to elicit reactions in the skin may be 

 amazingly small and ^vell beyond its lethal potency. 



The percentage of positive rabbits with high dilutions depends 

 largely on the batch of filtrate employed and especially on the 

 microorganism used for its preparation. In such titrations, B. 

 typhosus "agar washings" filtrates usually yield 25 to 50 per cent 

 and meningococcus "agar washings" filtrates 25 to 75 per cent of 

 positive reactions. It may frequently happen that after a certain 

 lo\v concentration is reached, a small percentage of rabbits con- 

 tinues to react with minute amounts of toxic substances of a 

 rather wide range. The comparison of percentages of positive 

 results obtained with high dilutions in small groups of rabbits 

 may, therefore, serve as a sotnce of error. It is essential to con- 

 tinue titrations until a dilution is determined which elicits no 

 reactions in all the rabbits of the group tested. 



Once the "end-point" (one reacting unit) is foimd, several 

 multiples thereof give consistent and regular results in a high 

 percentage of rabbits tested. This is best shown in Table i. 



Pabst and Branham (1933) entirely confhnied my findings that 

 titrations to the end-point with single prepared sites give reac- 

 tions of high degree of regularity. They also state that 6 reacting 

 luiits gi\e at least 75 per cent of severe reactions. 



