[3] In illustration it may be noted that as soon as a man has just transgressed the physiological limit which divides stimulation from narcosis, he is liable to throw overboard all prudential considerations and drink until he is completely drunk. This is one of the chief dangers of convivial after-dinner drinking. [4] For the physiology of this pupil-change, not uncommon in various kinds of acute narcosis, see the Appendix to Anstie. One apparent exception to this generalization needs only to be explained in order to confirm the rule. Memory, which usually figures as a high intellectual faculty, is often, even in deep drunkenness, capable of performing marvellous feats. While in college we once heard a tipsy fellow-student repeat _verbatim_ the whole of that satire of Horace which begins "Unde et quo, Catius?"--which he had read over the same day before going to recitation, but which, as we felt sure, he could never designedly have committed to memory. It appeared, however, that, in the literal though not in the idiomatic sense of the phrase, he had "committed it to memory" to some purpose, for as we, struck with amazement, took down our Horace and followed him, we found that he made not the slightest verbal error. This performance on his part was almost immediately followed by heavy comatose slumber. On afterward questioning him, it appeared that he remembered nothing either of the Satire or of his remarkable feat. Several analogous cases are cited by Dr. Anstie.[5] [5] _Stimulants and Narcotics_, pp. 174-178.