On the plausibility of homeopathic 'similitude'