In his chapter on manic-depressive insanity three years later Kraepelin[288] referred to this question again as follows:—"Under these circumstances I thought at first that the involutional depressions described as special clinical forms, melancholia in the narrower sense, which seemed to show essential differences in its general characteristics, course, and to a certain extent in the history of its development, should be separated from manic-depressive insanity. At the same time I was aware of the fact that in a considerable number of the involutional depressions, both on account of their clinical form and their association sooner or later with manic states, their connection with manic-depressive insanity could not be questioned. I therefore made an effort to establish a practical differentiation, entirely without satisfactory results. Further experience has demonstrated, as was shown in the discussion of the presenile psychoses, that they do not constitute grounds for the separation of melancholia. Deterioration is explained by the development of senile or arteriosclerotic changes. Some cases were of long duration, showing manic symptoms before recovery. The frequency of depressions in advanced years we have learned to be a legitimate development of the involutional period of life. The substitution of anxious excitement for volitional inhibition has proved to be an occurrence which is found in advancing years in those cases which had an attack of the ordinary form in the decade before (as shown in our cases 1 and 2). Hübner has, moreover, made the observation that melancholia may show retardation in one attack and not in the next. There remains, therefore, no adequate reason for differentiating the involutional depressions heretofore described as melancholia from manic-depressive insanity."