PAST estimates of the frequency of the occurrence of metastases in the central nervous system from primary growths in other organs have varied considerably. In an analysis of 12,730 necropsies performed at Basel between 1870 and 1905Krasting (1906 These widely quoted figures undoubtedly need revision because his carcinoma group includes only 15 examples in which the primary growth was in the lungs and in none of these were intracranial metastases found. It must be assumed that " mediastinal sarcoma " accounted for a large proportion of the sarcoma group, and a substitution of " carcinoma of the bronchus " for this diagnosis would substantially alter the balance between Krasting's two main groups.In a series of 68 cases in which cerebral metastases were found Gallavardin and Varay (1903) estimated that this form of dissemination occurred in one out of every 17 or 18 cases of malignant growth in other organs.More recent figures from the neurosurgical clinics of America are obviously too low, first, because patients in whom the intracranial lesions are part of a general dissemination are unlikely to be admitted to these clinics and, secondly, because cerebral metastases are sometimes found at autopsy without clinical indications of their presence. Thus the percentage of metastatic growths in all intracranial tumours is given as 4 by Grant (1926), 3 by Meagher and Eisenhardt (1931) and 3-2 by Cushing (1932) from the Peter Bent Brigham Hospital; as 5 by Dunlap (1930Dunlap ( , 1932 from the Mayo Clinic; as 7-5 by Brouwer (1932) while Boyd (1931) andBrain (1933) estimate the incidence as 3-5 per cent and 5 per cent respectively. A corresponding percentage figure of 9 is given by Elkington (1935) from analysis of a large series. On the other hand, Courville (1950), analysing the necropsy material at Los Angeles County General Hospital, found that 20*5 per cent of all intracranial growths were metastatic, a figure that accords with Krasting's (1906) earlier estimate.Yet post-mortem room figures, though more likely to be reliable, cannot be accepted as a true estimate on account of the selection of such material, and this inaccuracy is exaggerated by the modem growth of neurosurgical clinics. At