432 C. Hansch and R. M. Muir 



must be aromatic in character. Despite a rather thorough search (8, 

 13,19,36,39), no aliphatic compounds, with the possible exception 

 of one or two thiocarbamates (12), have been found to be active in 

 growth promotion. The only alicyclic compounds which have con- 

 firmed activity are 1-cyclohexenyIacetic acid and its analogues (11, 19). 

 It is not unlikely that these molecules could be dehydrogenated to the 

 quite active phenylacetic acid or one of its derivatives. Such dehy- 

 drogenations are known to occur in biological systems (25). So far, 

 our attempts to establish the occurrence of this dehydrogenation with 

 Avena coleoptiles have been inconclusive. Thus, although there is 

 slight evidence to the contrary, the overwhelming preponderance of 

 evidence from thousands of carefully tested compounds indicates 

 that an aromatic ring is necessary. Most effective are compounds 

 which contain six-membered rings such as benzene, pyridine, and 

 their homologues. However, compounds having the less aromatic five- 

 membered rings, or compounds with one double bond which might 

 easily be dehydrogenated to such, have been reported active (5, 11, 32). 

 If we proceed from the assumption that an aromatic ring is neces- 

 sary, we must answer the question. What is it about such a ring that 

 is necessary? Veldstra (37) and many others have taken the view that 

 it is the lipophilic character of the ring that is important. Important 

 as this character may be, it is of little help in explaining the enor- 

 mous differences in activity obtained with the different substituted 

 phenoxyacetic and benzoic acids. For example, phenoxyacetic acid 

 is at best slightly active. The introduction of a fluorine atom in the 

 4 position increases the activity at least 200-fold. The fluorine atom 

 would have very little effect on the lipophilic character and no steric 

 hindrance; hence its effect must be electronic in nature. Again the 

 differences between chlorine and methyl groups in size or in lipophilic 

 character would be small, yet when these groups are compared in the 

 4 position of phenoxyacetic acid, the chlorine atom confers at 

 least 100 times more activity. Another striking example results from 

 the comparison of 3-methyl- and 3-trifluoromethylphenoxyacetic 

 acids. Again, both groups have about the same size so that steric effects 

 are ruled out. If anything, tiie trifhioromethyl group might make the 

 molecule less lipophilic through inductive polarization of the ring, 

 yet it is 100 times more active. Wain's group (7, 35), in a very care- 

 ful analysis of the effect of substituents on phenoxyacetic acids, showed 

 that substitution in both the 3 and 5 positions with either chlorine 

 or methyl groups gave molecules which were inactive or of very low 

 activity. That this was not due to a steric effect or lack of lipophilic 



