THE VALENCY OF ANTIBODY 251 



an essential part of the lattice hypothesis, and on p. 220 we have indicated a 

 number of objections to the lattice hypothesis which imply the rejection of a 

 multivalent antibody. Nevertheless, variations in "avidity" and precipitating 

 power of antibody obtained during a prolonged course of immunization are con- 

 veniently explained in terms of variations in valency. But, as we should expect 

 from our ignorance of globulin metabolism in the animal body, there are no impres- 

 sive data to support the speculation. The most cogent objection to the theory 

 of multivalence (see Hooker and Boyd 1942) lies in the fact that immunization 

 with a known mixture of distinct antigens does not result in polyspecific antibody 

 (see, for example, Hektoen and Boor 1931, Dean, Taylor and Adair 1935). It 

 is argued that, if during the synthesis of antibody in the animal there was in the 

 growing antibody globulin molecule more than one region that could be influenced 

 by antigen, there is no good reason why two or more regions on one molecule 

 should not be influenced by separate antigens, and thus give rise at least to a 

 dispecific antibody. 



The conception of dispecific antibody offers a simple explanation of the results of 

 antibody response to certain natural complex antigens (Meyer 1936, Miles 1939), but since 

 the antigenic constituents of the mixtures were not fuUy defined, the success of the hypo- 

 thesis is not good evidence of its validity. The difficulty Hiay in part be resolved by 

 assuming that monovalent Svedberg units, of molecular weight 35,000, are in fact formed 

 in the body, and that these are later united into larger molecules, which are multivalent. 

 But here again, unless we postulate that the units are always built on to one another 

 in conjunction with a single antigen particle, there is no good reason for assuming that 

 Svedberg units of diverse specificity should not be joined to form a polyspecific antibody. 



Molecules '^f antigen having on the surface two {intigenic determinants may give rise 

 to apparently dispecific antibodies. For example, Heidelberger and Kendall (1934), 

 immunizing rabbits with ovalbumin modified by an azo-dye, produced antibodies to 

 ovalljumin alone and to azo-protein alone, and also antibody that reacted equally well 

 with azo-protein or ovalbumin (see also Singer 1942). Landsteiner and van der Scheer 

 (1938), on the ether hand, could find no evidence of a double specificity even in antibodies 

 prepared against a single azo-antigen having two haptenic groups. They prepared antigens 

 with aminosuccinanihc acid (S), with aminophenvlarsenic acid (A) and with a compound 

 of the.two (SA). Antisera to SA precipitated with SA, A and S, but sera absorbed with 

 A precipitated only with S, and sera absorbed with S, only with A. The various antisera 

 contamed one or two kinds of specific antibody, but never dispecific antibodies. Haurowitz 

 and Schwerin (1943) oljtained analogous results by immunizing rabbits with a globulin 

 antigen containing both ^-azo-phenylarsonic (A) and m-phenylsulphonic groups (S) or 

 with a globuhn containing both A groups and di-iodotyrosine (T) groups. The resulting 

 antisera contained antibodies against A and S, and against A and T respectively, but 

 there was no evidence of dispecific antibodies against (A + S) or (A -1- T). 



With regard to the apparent variations in combining power displayed by 

 monospecific antibody, Hershey (19416) points out that it is not necessary to 

 invoke differences in valency to explain the properties of " low grade " or of highly 

 avid antibody. The " affinity " of a unit reactive patch on the antigen molecule 

 for unit reactive patch on the antibody may be described in terms of a dissociation 

 constant k, and Hershey has ingeniously subsumed most of the observed individual 

 variations of antibody under variations in k. For example, a small k explains 

 large maximal antigen-antibody ratios, high avidity, and broadened reactivity 

 in sera taken late in immunization. Heidelberger's univalent antibody, and non- 

 precipitable antibody left after absorption of precipitable antibody, would have 

 a large k. Variations in k would be independent of the nature of the immune 



