Abstract-Nuclear magnetic resonance proton chemical shifts may be correlated with free valence, a representative reactivity index, for fluoranthene and two novel sulphur heterocyclic analogues, acenaphtho [l,2-b]-and -[I ,2-c]thiophens. Such correlations arise fortuitously, however, and an attempt to explain them in terms of intrinsic properties of the Huckel-McWeeny theory of proton shielding is proved to be fallacious.APPARENT correlations between nuclear magnetic resonance proton chemical shifts (&), and both Hiickel MO and SCF reactivity indices, to and experimental reactivities5 in alternant polynuclear benzenoid hydrocarbons, have led to a number of applications of this approach to non-alternant hydrocarbons,6 hydrocarbon ions6 and heterocyclic molecules.'~* Thus, Clin and Salier have pointed out a relation between carcinogenicity and the sign of the deviation for K (or Krebs) region protons from a graph of 6, against superdelocalisability.* This idea is attractive from the standpoint of predicting both the reactivity towards electrophilic, nucleophilic and radical reagents, and even the carcinogenicity of novel heterocyclic analogues of fluoranthene (l), such as acenaphtho For example, Fig. 1 shows the relation, for analogous protons of 1 to 3, between 6, and free valence (F,), our chosen reactivity index, which gives predictions similar to many other indices,, and suggests correctly that electrophilic substitution (e.g. detritiation,13 bromination,14 nitrationl5 and sulphonation16) takes place preferentially at the 3 position for 1. The reactions of 2 and 3 have not apparently been studied. We believe, however, for a number of reasons that the basis of such correlations is, if not spurious, at least coincidental. We have already shown3 that if the range of protons of alternant benzenoid hydrocarbons is extended, substantial deviations from plots of 6, against F, occur which are not readily exp1ained.l' Even if these deviations are ignored, the remaining correlation arises as a fortuitous consequence of the geometry of polynuclear aromatic molecules, so that in a causative sense 6, is unrelated to F,.. Values of the latter fall into broad groups, each roughly corresponding to a different position type, e.g. relative proximity to ring centres, the values of 6, for the proton types of 4 are 6, > 6, > 6,, and for the protons of 1, 2 and 3, 6, e 6, > 6z;18*z0 apparent, but noncausatiue, relations between 6, and F, thus arise of the type shown in Fig. 1 , for which the correlation coefficient (0.87) is statisticazly significant at the 0.1 % level.* While contributions to 6, from unequal charge densities (q,, itself a reactivity index) are expected for nonalternants, hydrocarbon ions and heterocycles, 6, is still dominated by effects arising from .rr-electron circulations. A recent attemptz1 to argue a fundamental relation between a, , F, and q, depends on equating a general expression for the change in partial .rr-energy at the r'th atomic centre, (AE"), due to a general perturbation ( A m , = k,q, + k,F, + k...