I was feeling remarkably happy when my thesis was written, because I felt I had made an important advance, and then, disaster! Maerker and Roberts published a paper (J.A.C.S. 1966, 88, 1742-1759) that asserted that the cyclopropyl ring also stabilized adjacent negative charge. If this were correct, the cyclopropane ring did conjugate with adjacent charge, and my polarization explanation, and my PhD thesis, were just plain wrong. The reason is, of course, that a polarization field will stabilize one charge, but must destabilize the other, because the force between like charge is repulsive. My first response was deep despair; my second was, perhaps I had better read this paper carefully.
 
There are three major complications. The first is that if the lack of stability is indicated through rearrangement, that only means something else is more stable. Thus a Grignard reagent made from cyclopropylmethyl chloride leads to a ring-opened rearranged "carbanion". As it happens, the cation made from cyclopropylmethyl chloride, or cyclopropylmethyl alcohol, or a tosylate, is also unstable and promptly rearranges. (This is the famous bicyclobutonium "non-classical" carbenium ion.) Rearrangements of the carbenium ion are inhibited by bulky substituents, but the stabilizing effect of the cyclopropyl ring is easily shown by considering the rates of formation of the ion, or the energy of the species by mass spectrometry. However, at the time neither of these techniques were available for the anion. The second is solvation. Carbanions tend to be generated in solvents such as ether or petrol, which almost forces ion pairing, whereas the carbenium ions tend to be made in acids more acidic than concentrated sulphuric acid, and hence have very high dielectric constants and strong solvating properties. Thus the cyclopropycarbinyl carbenium ion made in solution is never stabilized to anywhere near the extent as is found by mass spectrometry. The third is that the polarization field is not a simple field. Four orbitals move towards a substituent, but at the corner of the cyclopropane ring, there is weak positive polarization field, due to the movement of the three orbitals about that atom, which, being very close, over-ride the stronger effect of the more distant movement. Further, while the cyclopropyl anion receives some localized stabilization together with the expected destabilization, the associated cation in the ion pair is strongly stabilized, and overall, the "anion" appears to be slightly stabilized. This effect is most strongly seen in calcium carbide, which, of course cannot be stabilized by conjugation without violating the Exclusion Principle. Further, according to the polarization interpretation, the "bare anion" formed on a carbon atom adjacent to the cyclopropyl ring should be destabilized, but by less than half as much as the cation is stabilized (because the charge is more distant, and by applying the virial theorem). In solution, solvation becomes an issue, as does the location of the cation.
 
So what was the evidence Roberts found relating to the conjugative stabilization of the anion? Some of the evidence, in my opinion, falsified the conjugative explanation because the anion refused to form when it should have. Such failures included: treating with butyl lithium (which meant that the protons were less acidic than those of butane), refluxing for 46 hr with phenylpotassium in heptane, treating with pentylpotassium (which reacts smoothly with ethylbenzene), stirring at 80^o in heptane with potassium and sodium monoxide. My theory might be still alive!
 
However, evidence for conjugation was claimed when the phenylmethylcyclopropylcarbinyl anion formed with potassium as the counterion, but it rearranged to the corresponding allylcarbinyl anion with any tendency towards covalent character. Roberts argued (almost certainly correctly) that when something like lithium was the counterion, the lithium would get close to the anionic centre and partial covalent binding would occur. Potassium was big enough that the bulky substituents forced it away. However, it was here that we differed in our interpretation. Roberts claimed that the extra stability with potassium was due to the fact that the "pure anion" was formed, and the cyclopropyl ring provided conjugative stabilization. My interpretation was, the potassium formed the "pure anion", which permitted delocalization, which in turn permitted the anionic charge to be delocalized by the benzene ring, out of range of the cyclopropane ring. Any attempt at localizing the negative charge on the carbinyl carbon led to repulsive interactions from the cyclopropane ring, and hence rearrangement.
 
There were two problems with that explanation. The first is, is it convincing? You, the reader, can judge. The second was, there was no way to publish it. The problem with a scientific paper was, once something was asserted, that explanation stood. Falsification with independent evidence was required, not a simple assertion. Nevertheless there is another lesson here. Just because somebody asserts that something has happened, that does not make it so. Read the evidence carefully!