Most adults do well on about 30 calories per kilogram of body weight; children of four years need 75 calories per kilogram, children of eight years need 60, and children of twelve years need 50. If sugar appears in the urine during the process of raising the diet, we drop back to a lower diet, and if this is unavailing, start another starvation day, and raise the diet more slowly. But it will be found, if the diet is raised very slowly, sugar will not appear. It is not well to push the average case; if the patient is taking a fair diet, say protein 50, carbohydrate 50 and fat 150, and is doing well, without any glycosuria, it is not desirable to raise the diet any further. The caloric intake may seem rather low in some of these diets, but it is surprising to see how well most patients do on 1500 or 2000 calories. It will be seen that the treatment can be divided into three stages: (1) The stage of starvation, when the patient is becoming sugar-free. (2) The stage of gradually working up the diet to the limit of tolerance. During the first two stages a daily weight record should be kept, and the urine should be examined every day. The patient should, of course, be under the immediate supervision of the physician during these two stages. It is always well to discharge a patient on a diet somewhat under his tolerance, if possible. (3) The stationary stage, when the diet is kept at a constant level. The patient is at home and going about his business. Most patients may be taught to test their own urine, and they should do this every other day. If there is sugar in the urine, the patient should go back to a lower diet, and if he cannot be made sugar-free this way, he should be starved again. A semi-starvation day of 150 grams of vegetables, once a week, whether or no the urine contains sugar, is of value for the purpose of keeping well within the margin of safety and of reminding the patient that he is on a strict diet. _It is very important for a diabetic to take a considerable amount of exercise: he can utilize his carbohydrate better, if he does._ If this treatment is to be successful, it is absolutely necessary for the patient to adhere very strictly to the diets, and to measure out everything very carefully; the meat especially should be weighed. It will be noticed that in some cases the calories in the diets do not tally exactly with the protein, fat and carbohydrate values. The reason for this is that for the sake of convenience the calories have been given in round numbers--5 or ten calories one way or the other makes no difference. The essential points brought out by Allen's treatment are as follows: (1) It is not dangerous to starve a diabetic, and two or three days of starvation almost always make a patient sugar-free, thus saving a good deal of time, as contrasted with the old treatment of gradually cutting down the carbohydrate.