22 INTRODUCTION TO IMMUNOCHEMICAL SPECIFICITY 



In the second place, the all-or-none interpretation of inhibition 

 experiments is an oversimpHfication. The experiment summarized 

 in the first Hne of Table 2-2 actually yields more information than 

 is contained in the mere statement that tubes 1 to 4 are negative and 

 tubes 5, 6, etc., are positive. The strength of the reaction in the 

 first tube to the right of the last negative tube also contributes in- 

 formation, for the reaction can vary from weak to strong. Taking 

 the simple point of view, for example, we should estimate from 

 Table 2-2 that hapten H' is only one-fourth as effective an inhibitor as 

 hapten H, for it takes four times as much to produce complete in- 

 hibition. But if we take account of the fact that the next tube after 

 complete inhibition gives a reaction of + in the case of hapten H 

 and only ± in the case of H', it is clear that H' is actually some- 

 what more than one-fourth as effective as H'. But how much more? 

 It is hard to put such things into numerical terms. It is possible to 

 invent codes for the translation of such readings into quantitative 

 terms, or appropriate numerical scores may be found by statistical 

 methods (see, for example, Fisher, 1950, pp. 289—295). In general, 

 however, such treatments of the results of inhibition tests have not 

 been found to extract enough extra information from the results 

 to justify the calculations involved. 



Recognizing, therefore, that the results of inhibition experiments 

 are only semiquantitative at best, serologists who are attempting to 

 compare the inhibitory power of two different haptens do not 

 generally attempt to make quantitative estimates, but are content 

 to say merely that hapten H is more effective, for this particular 

 serum, than H' is. Some tend to rely on the old rule of thumb, 

 which is pretty well borne out in practice, that a difference in the 

 results obtained with two haptens is significant if the difference in 

 their inhibiting capacity differs rather consistently, from one ex- 

 periment to another, by two tubes (ordinarily meaning a four- fold 

 difference in effective concentrations). If the results do not differ 

 by this much, one may suspect a difference in the effectiveness of the 

 two haptens, without venturing a confident opinion. Even such a 

 difference, however, arbitrarily judged to be "non-significant," may 

 be of value as a guide to further experiments. 



The value of statistical methods in general is of course not in 

 doubt. For a lonsj time certain biologists, and immunologists in 



