The route of infection by tubercle in children is an important matter, but there are serious difficulties in the way of its determination. Children, or more especially infants, do not expectorate, and the diagnosis of tubercle in them is therefore often uncertain. The beginnings of tuberculosis are consequently not often observed. It is only when the disease is finally developed that its clinical history is a matter of notice and record. In a recent lecture, Dr. George F. Still1 considers these facts and offers some interesting statistics, based, not on clinical observation, but on post-mortem findings, for the solution of this problem. In 769 autopsies of children, tubercle was found in 269, or 35.2 per cent. Tuberculosis was the actual cause of deaths in 252, or 32.8 per cent. From these statistics, therefore, it can be roughly estimated that about one-third of the deaths in childhood are due to