"First and last, I tried a goodly variety of 'fire' experiments," D---- continued, snuggling down for a moment with both arms under the blankets, "and I don't mind admitting that I'd like to have a few of 'em, smoke and all, flaming up all over this refrigerator right now. The thing I finally decided to try consisted of nothing more than a light, shallow tank of ordinary kerosene--paraffin oil, I believe they call it here--made fast to a small, roughly built raft. The _modus operandi_ was as simple as the contrivance itself. As soon as a U-boat was sighted, the raft was to be launched on the _opposite_ side, and kept about thirty feet out by means of a light boom. The next move was to be up to Fritz, and it was fairly certain he would do one of two things--submerge and make off, or remain on the surface and begin to shell us. In the latter case we were to start firing in reply, of course; but that was only incidental to the main plan. This was to wait until we were hit, or, preferably, until he fired an 'over,' the fall of which, on account of his low platform, he could not spot accurately, and then to fire the tank of kerosene. A line to a trigger, rigged to explode a percussion-cap, made it possible to do this from the rail. As the flames, besides giving off a lot of smoke, would themselves leap high enough to be seen from the other side, it was reasonable to suppose that Fritz would be deluded into thinking we were burning up, and make his approach a good deal more carelessly than otherwise. If he persisted in closing us on the surface, there would be nothing to it but to make what fight we could with our fo'c'sl' gun, and try to make it so hot for him that he would have to go down before his heavier shells had done for us. But if, following his usual procedure, he made his approach submerged, then there were two or three other little optical and aural illusions prepared for his benefit. I will tell you of these in describing how we actually used them."