2. Relapses are more frequent--probably considerably more frequent--under treatment by salicylates than under other methods. Thus, the average of relapses in eight different tables of cases treated by the salicyl remedies ranged from 16.6 per cent. to 35 per cent., giving a general average of 26 per cent.;[138] while under other methods in three different tables the average ranged from 5.4 per cent. to 27.6 (this last under the full alkaline), giving a general average of 16 per cent.[139] Relapses appeared to recur less frequently in those cases which yielded to the salicylates within five days than in those which took from six to eleven days to yield, in the ratio, according to Fagge's figures, of 26.6 per cent. for the first, and 29.4 per cent. for the second day; and, according to Hood's, as 18.4 per cent. to 24.4 per cent. There does not appear to be any regularity in the order of occurrence or recurrence of relapses, nor is Southey's definite statement that in "relapsing cases the temperature is nearly or quite normal on the eighth evening, and a slight relapse occurs on the thirteenth morning," borne out by the statistics produced at the London Medical Society. Moreover, W. Carter's cases[140] have not confirmed Southey's precise statement respecting the gradual remission of the temperature on the eighth and ninth days of illness in the continued or non-relapsing, uncomplicated forms. Irregularity and inconstancy are the typical features of articular rheumatism. The relapses under the treatment by the salicylates have been referred to the premature disuse of those remedies, but they do occur notwithstanding {53} the continued employment of them. It is a general opinion that exposure to cold, errors in diet, and an early return to work are frequent causes of relapse; and Broadbent refers the increased liability to relapse under the salicyl compounds to the rapidity with which those remedies relieve the acute symptoms of articular rheumatism, in consequence of which sufficient care is not observed either by the patients or their nurses, and they are exposed to some of the above exciting causes of relapse. All the above causes do probably play their part so long as the materies morbi (if that really exist either as a chemical principle or as a germ) has not been wholly eliminated or destroyed. Indeed, the short intervals which frequently obtain between the primary invasion of the so-called relapses, and the failure of the salicyl compounds to prevent peri- and endocarditis, render it probable that what are commonly spoken of as relapses are not due to a new infection, as in the case of the relapse of typhoid fever, but to the recrudescences of a disease not yet terminated, but over some of the manifestations of which--the articular inflammation and the pyrexia--the salicylates exercise some control.