365. In the test sample (by comparison with the general population) deaths attributable to arthritis and other forms of rheumatism were very high (about 100 times on average); deaths attributable to clinical conditions known to be associated with ankylosing spondylitis were on average 2.9 times more common; deaths from conditions attributable to irradiation (aplastic anemia, leukaemia, cancers other than leukaemia) were about two times as common; deaths due to diseases from which the mortality could be similar to that of normal population hand a prevalence of 1.3. This latter increase regarded all conditions exami- ned and the total expierience was sufficiently large for it to be highly Significant. 366, When mortality was examined as a function of the post-irradiation period, all causes of death other than cancer and aplastic anemia were in a constant relationship to the expected mortality. In contrast, for leukaemia and aplastic anemia mortality increased within the first 5 years of observation and then fell off; deaths for cancers of heavily-irradiated sites were increased approximately two-fold at six or more years and up to 15 years after treatment. Many different types contributed to this excess, in a rough proportion to their natural incidence. Deaths from cancers originating in other lightly-irradiated tissues were not increased significantly. 367. The interesting observation in the present context is the increase in deaths due to non-specific causes and not grossly related to spondylitis or to irradiation. The authors [C28] pointed out a number of reasons that might ac- count for this finding. Firstly, non-specific deaths might contain a small proportion of rare conditions related to spondylitis (lesions of the aortic val- ves, regional enteritis, proneness to accidents}. Secondly, patients in this group carry other conditions known to be associated with spondylitis (amyloid degeneration, nephritis) that might decrease resistence to non-specific causes of death. Thirdly, the inaccuracy of the diagnoses at death, the possible ef- fect of drugs and the use of imperfect death-rate values for the calculation of the expected numbers of deaths were discussed as other possible reasons. Finally, the constancy in time of the ratio between the number of deaths observed other than those presumably related to radiation over the expected number calculated from national mortality rates suggested that the above excess mortality was likely to be dependent on the spondylitis itself and unrelated to the form of the treatment.