At Hampton Roads, on one occasion, after a submarine run, we came up under a small launch and picked her up bodily on the deck. We had not seen the boat until we heard her bump against the conning tower and heard some of the ladies scream. We submerged quickly and lowered them into the water again. Another time we came up under a large barge, but all the damage incurred was a broken flagstaff. The best mode of procedure at such times is to bring the vessel to rest while submerged and stop all machinery, then listen for the sound of the machinery of surface vessels. These noises can be heard for a considerable distance under water. If no sound is heard it is then safe to come up. Even in this case there is some possibility of coming up under or just in front of a sailing vessel. One has to take some chances, and I do not consider this taking any greater chance than is taken by the navigator of a surface vessel in running in a fog or in a snow storm. The question of air supply was at one time one of the most difficult problems to solve on paper with which early experimenters with submarines had to contend. There was no exception in my case. I thought it would be possible to remain submerged only a short time unless I provided some sort of apparatus to extract the carbonic acid gas and restore oxygen to the air after breathing and exhaling the air in an enclosed space like a submerged vessel. I took up the question with various physicians and with a professor of physiology at Johns Hopkins University, and, according to their information and text-books, it would be a very difficult matter to carry sufficient air to remain submerged without change of air except for a very short time. Their text-books stated the quantity of free air that should be allowed per individual. This varied from fifteen hundred to three thousand cubic feet of air per individual per hour. It would be impossible to provide this amount of air in a submarine. What it was essential to discover was _how little_ air a man could live on without suffering ill effects. I then built a box containing twenty-seven cubic feet of air space. I entered this and was hermetically sealed within it. At fifteen-minute intervals I lighted matches to note how freely they would burn. At the expiration of three-quarters of an hour the matches still burned brilliantly at the top of the box, but went out when lowered to about the level of my waist. This indicated that about one-half of the oxygen had been consumed and converted into carbonic acid gas. I was surprised to find how distinctly the line was drawn between the air containing oxygen and that containing the heavier carbonic acid gas. I concluded from this experiment that from fifteen to twenty cubic feet of air per individual per hour was sufficient to maintain life for short periods of time without injury.