To the Editor: The article by Bantle et al. in the July 7 issue* is a serious example of the dangers involved in making broad clinical recommendations on the basis of a very narrowly defined experiment. The most serious deficiency of the study is that patients with uncontrolled diabetes were studied. An average fasting blood sugar level of over 180 mg per deciliter and two-hour postprandial blood sugar levels of over 340 mg per deciliter indicate control of insulinrequiring diabetes that is quite poor by any standard. Many of the group who supposedly had non-insulin-requiring diabetes should have been receiving.