The length of the butt was yet to be ascertained; and this, though apparently a simple operation, cost me a good deal of consideration, before I could get at it with any degree of exactness. You may fancy that it would have been easy enough to get at the length, by just placing the stick parallel to the cask, and notching it square with the ends of the latter? And so it might be easy enough, with plenty of light around you to see when it _was_ square, and a level surface upon which to rest your measure. But as I had the advantage neither of light nor level ground, I encountered great difficulty in this operation. I could not tell when the ends lay even with each other, merely by the touch. I had to pass my fingers from one to the other, and could not grasp both at one time--that is, the rim of the cask and the end of the rod--since they must needs be several inches apart. The stick, too, lay unsteady, and by the feel I could not be sure when its end was exactly "flush" with the head of the cask. The mistake of an inch--it might be several--would falsify all my computations, and render them of no use. It would not do to proceed upon such a conjectural basis, and for a while I was puzzled, and had to pause. This was an unexpected obstacle, for I had from the first regarded the diameters as the only difficulty; about the possibility of obtaining the length, I had never entertained a doubt. But my wits again came to the rescue, and I soon discovered a plan that would effect the end in view. I had to make another rod--by splicing two more lengths split from the board--and with this I was able to determine the point. I managed the matter thus: The old rod I pushed along the head of the cask quite beyond its outer edge, so that it rested at both ends against the projecting rim. Thus placed, it was exactly parallel with the plane of the barrel's head, while a foot or more projected outward and towards me. Holding the end of the second rod against this projecting part, and at right angles, I gave it a direction along the side of the cask, and I was able to mark the point, where the middle part of the swell came in contact with the second rod. This, of course, after deducting the depth of the rim and the presumed thickness of the head, gave me half the length of the interior of the cask, and that was all I wanted, since two halves make one whole. I was now in possession of the _data_ of my problem; it only remained for me to seek the solution. CHAPTER THIRTY ONE. "QUOD ERAT FACIENDUM."