There were three things which caused us to delay her departure on a submarine exploration trip for a few weeks. The first was the escape of gasolene (petrol) fumes in the boat. When first built, fuel tanks were built in the hull itself and formed an integral part of the vessel. Special care was given to make these fuel tanks tight. They were tested under hydraulic pressure and found to be tight, but the fumes from gasolene (petrol) are very searching, and, after filling the fuel tanks and keeping them filled over night, gasolene fumes were found to exist in the boat the next morning to such an extent that I would not venture to make a start until a fuel tank had been built outside of the vessel, where any escape of fumes would not form an explosive mixture. I followed this practice in all our later gasolene-engined boats, which largely eliminated the danger from carrying gasolene as a fuel. A number of explosions have occurred in other types of gasolene-propelled boats, in some cases with fatal results, from gasolene fuel being carried in built-up tanks within the hull itself. The next cause of delay was due to the escape of and collection of carbon monoxide within the vessel. This developed on our first submarine run. After we had been down about two hours some of us commenced to experience a dull pain at the base of the brain and a decided feeling of lassitude. On coming to the surface a couple of our men collapsed completely, and one was very sick all night. I could not understand the cause of this, as nothing of the kind had occurred in my previous hand-propelled vessel, so we made another submerged run the following day, and after about the same period of time the pain in the head and weariness came on again. I then discovered that the engine would occasionally backfire out into the boat and that gas was escaping past the piston rings into the base of the engine and from there into the boat. To overcome this difficulty I installed what I called an induction tank, which was piped up to the air intake of the engine and also the engine base. A check valve admitted air into this induction tank. When the engine was started the check valve was automatically lifted and induced a flow of air through the tank, in which a slight vacuum was maintained, which also served to draw the gases out from the engine base. In case of a backfire, the check valve automatically closed and the gases from the backfire were caught in the induction tank, from which they were drawn out on the next stroke of the engine. This solved the difficulty, and thereafter the air was always fresh and pure when running submerged even after a submergence of several hours' duration.