Abstract
Many of the chronic diseases are remittent in nature and indefinite as to onset, therefore incidence and point prevalence are inadequate as estimates of their frequency. It is suggested that the frequency distribution of persons according to proportion of time in episode is the appropriate measure for this situation. This frequency distribution can be obtained from a series of examinations by the method of moments. This type of frequency distribution is applied to rheumatoid arthritis and is used to demonstrate that the apparent increase in frequency of the disease with age is more due to a marked increase in the proportion of persons who spend a large part of their time in episode than it is to an increase in the total number of persons affected for any proportion of their time.