This appeal presents solely the question of how much damage was done to the libel-ant’s steamship W. L. Steed when it was in collision with the tank barge Portsmouth in tow of the Shipping Board’s steamtug Hulver in the Panuco River, Mexico, on May 25, 1920.

The W. L. Steed, a single screw vessel four hundred thirty feet long, was loaded alongside the Huasteca Dock on the Panuco river below Tampico, Mexico, with a full cargo of petroleum products. Her bow was then up stream, and it was necessary for her to turn to put out to sea. At that time the Panuco, swollen by freshets, was running with a current velocity of about six miles per hour. *570Its navigable channel'for a boat of her size above the Huasteca Dock to the point where the Steed began to turn was about one thousand feet wide. She east off and steamed up stream about a mile and a half with the intention of putting her stern into a canal there and so to obtain added turning room. Finding this impossible because the mouth of the canal was obstructed by a ship, she started to turn to starboard in the current by going ahead from near the left bank under a hard right rudder as far as safety permitted, then backing with rudder amidships to a point as near the left bank as it was thought she could go without touching, and by repeating the maneuver until she was headed down stream. She went forward and back until she succeeded without difficulty in turning athwart the stream, but then found herself unable to complete the turn and drifted with the current in an unmanageable condition for about half a mile. While doing so she was drifting broadside over the bottom of the river at a speed, of approximately five hundred feet a minute. The weather was clear with hardly any wind. While she was so drifting down, the steam tug Hulver with her tow came up the river. She was one hundred fifty-five feet long with a beam of thirty feet. The Portsmouth was "a steel oil tank barge then empty. She was two hundred fifty feet in length, had a beam of thirty-five feet, and was made fast to the starboard side of the tug. As the Steed seemed to be nearer to the left than to the right bank of the river, the Hulver, blowing one blast with the purpose of indicating her intent to pass between the stem of the Steed and the right bank, was turning to starboard when the Steed signaled and started ahead. Thereupon the Hulver blew again, the Steed let go her anchors, the tug put her helm hard astarboard and tried to go under her stem. She was too close, however, to avoid fouling the starboard rigging of the Portsmouth’s mainmast and one of her lifeboats on the Steed’s stem flagpole and railing. No damage was done to the hull of the Portsmouth. She was towed on to her loading point up river, where she took aboard her load and carried it to Galveston, Tex., without any repairs. The Steed was dragging her anchors when the tug and tow went by, but soon swung to them and was let down stream to a pier, where her cargo was discharged and her damage surveyed. It was found that, 'in addition to the minor damage to her flagpole and railings, her rudder and rudder stock was twisted to the right about seventy-five degrees. The only matter in dispute is whether or not this damage was caused by the collision with the Portsmouth'. The matter of damages was heard by a commissioner, who came to the conclusion that the libelant had failed to prove that the Portsmouth damaged the rudder and so reported. The District Court declined to follow the report, and entered a decree for the libelant for the rudder damage as well as that to the flagpole and railings.

With the burden of proof on the libel-ant, to prove affirmatively, that the Portsmouth hit and caused the damage to the rudder, California Navigation & Improvement Co. v. Union Transportation Co. (C. C. A.) 176 F. 533, it is necessary to determine whether this burden has been discharged. In doing so we do not overlook the fact that the libelant once had in its possession positive evidence which would have made it a very simple matter to show that the Portsmouth did injure the rudder, if such was the fact, and that it failed to produce it. This is evidence to show where the force which did the damage was applied to the rudder blade. All agree that a vessel of the size and draft of the Portsmouth could have struck the Steed’s rudder only at one place near the top of the blade. Consequently it was incumbent upon the libelant to prove affirmatively that the rudder was struck at that point in order to charge the Portsmouth with having done the damage. We have no means of knowing why this most important evidence was not produced, and, in the absence of any explanation, cannot help but notice its unexplained nonproduction. Without it, the libelant comes dangerously close to relying upon mere speculation to prove a fact absolutely essential to its ease. To understand why this is so, it is necessary to know what parts of the rudder are meant by the terms used in whatever description of the damage is to be found in the record. The plate is the blade; it is made of steel one and one-eighth inches thick. The stock is the shaft running up into the ship from the palm at the extreme top of the rudder blade, and through it the turning force is transmitted to the rudder from the steering mechanism. It is of steel several feet long and twelve inches in diameter. The main body piece is at the edge of the blade nearest the ship and extends downward from the palm, to which the stock is attached, the entire length of that edge of the rudder from top to bottom. It serves to strengthen that side of the blade and to take up and distribute to all parts of the blade the turning energy coming into it from the stock through the palm. The arms are re-enforcing pieces extending outward at right angles from the *571main body piece along tbe sides of the blade and also projecting inward from the main piece to act as a portion of the hinges on which the rudder turns. There were four such arms, and only one was above the only possible point of contact the Portsmouth could have made with the rudder blade.

To show how the rudder was damaged, we have the preliminary survey report made at Tampico, Mexico, on May 23,1920, which describes the damage as follows: “On examination of this vessel on the day named I found that the rudder blade is bent very badly to port and that the upper part of the rudder stock is twisted heavily to port. As now standing when the rudder itself is hard-a-port the helm is also hard-a-port showing a distortion of about seventy-five degrees.” After temporary repairs were made at Tampico, the Steed went to New York for permanent repairs, and we have the report of the survey made at the dry dock of the Morse Dry Dock & Repair Company on and following July 3, 1920, which shows: “The upper stock of rudder badly twisted also the main body piece of rudder badly bent and twisted. Arms on rudder more or less bent. Rudder blade bent.”

On July 7, 1920, a survey was made for Lloyd’s Register showing: “Rudder badly set over to port and stock and main piece badly twisted. Rudder plate bent.”

In addition to this, we find in the deposition of Chief Officer Pharr of the Steed, who testified that he made an examination, that “the rudder post was twisted, when the wheel on the bridge was amidships the rudder was hard over and when the wheel was hard over the rudder was amidships. The tiller engine, the rudder was twisted to an angle of about 90 degrees.”

In the deposition of Olaf Berg, the master of the Steed, appears the following question and answer, which from its setting must refer to a time soon after the collision: “Q. You didn’t notice any fault in her steering? A. We noticed a fault in the steering, yes. When we were headed up the river we had to work around and the pilot says, ‘There is something wrong.’ I looked over the stern and saw a little dent on top of the rudder, and I saw the quadrant hard stove to one side of the rudder nearly amidships.”

The libelant also introduced the deposition of its superintending engineer at Tampico, who testified that he examined the rudder after the collision and that “the condition of the rudder show.ed a twist between the palm of the rudder stock and the rudder head resulting in the rudder head being thrown out of the normal center position.” This is all the evidence we have been able to find in the record which can in any way be said to throw light on whether the rudder blade was struck where the Portsmouth would have hit it or lower down where a rock or other obstruction at or near the river bank would have been more likely to have hit it.

As we view this evidence, it has no tendency to show that the force was applied near the top of the rudder blade, except what can be made out of the testimony of the captain, that he saw “a little dent on top of the rudder.” It is impossible to believe that he could have seen and described as “a little dent” any spot hit by the Portsmouth hard enough to apply sufficient force to twist the rudder and its steel stock twelve inches in diameter seventy-five degrees out of alignment.

Looking now to the evidence making against the claim that the Portsmouth hit the rudder, we find that the Steed, before the collision, was unable to turn further than athwart the river during the half mile she traveled down in that position. This was either because she was prevented from turning by the current in some way not easy to understand or by her rudder having already been disabled. No one knows which, but, as she was moving over the bottom at the rate of about five hundred feet a minute, it is possible that her rudder had struck a rock or other obstruction when she was near the left bank and had received the damage before the Portsmouth arrived. Since the Steed was steered from the bridge by a telemotor, this could have been so and not have affected the operation of the telemotor itself, so that those on the bridge, except as they were led to believe that something was wrong by the failure of the ship to respond, would not have known of the rudder injury. Then, too, the twist' was given in the Tampico survey as seventy-five degrees and in the captain’s deposition as ninety degrees. The free play of the rudder from straight amidships was only thirty-eight degrees on each side. As it is obviously impossible for the twist to have been as much as the captain testified, the only fair inference to be drawn is that the rudder was twisted from its extreme right position to hard left. To do this, the force must have been first applied when the rudder was hard right, and it would presumably have been in this position when the Steed started ahead after backing as near the left bank of the river as it could have gone with apparent safety. Moving down stream as it then was in the current, the damage done is what might have *572been expected to result from her hitting a rock in that position; but all this is mere conjecture and goes no farther than to show that the rudder might have been damaged before the collision with the Portsmouth. Its only usefulness is to make it apparent that the damage discovered after the collision was not necessarily done at the time of the collision.

The expert testimony was both ways on the possibility of sneh a twist as was found having been imparted by force applied to the only point it could have been by the Portsmouth. Some of this evidence tends to show that it could have been, and some that force sufficient to create a twist of as much as seventy degrees from that point to the top of the rudder stock would have been in excess of eight hundred tons, would have been about eight times the tensile strength of the material of which the stock was made, and would surely have resulted in a fracture rather than a twist. This evidence indicates that part of the distortion must have been in the rudder blade below that point, and of course could be there only if the point of contact was below. There is evidence also which tends to show that, applied near the bottom 'of the rudder blade, a force of about one hundred eighty-five tons would have been sufficient to do the damage, and a twist from extreme left rudder to extreme right rudder at the bottom would have been absorbed by the rudder blade in such part that the stock would not have been broken in being twisted to the extent necessary to make up the total distortion found. Nor were these experts by any means in agreement as to whether the Portsmouth could have withstood the reacting force of the rudder blade without receiving a gash in her side or springing a serious leak. Furthermore, we should notice in passing, for whatever it may be worth, that the captain of the Hulver testified that he saw the Steed’s rudder was hard over upstream or starboard helm before the collision.

From the foregoing it is apparent that the evidence is so conflicting, not only as to the basic facts, but as to the inferences which should be drawn from them, that, without the evidence the libelant once had and should have preserved in some way and produced at the hearing, to show at what point the rudder blade was struck, it is impossible for us to reach the conclusion that the libelant has discharged the burden of proving by a preponderance of the evidence that the Portsmouth struck the rudder. Accordingly we agree with the commissioner that this fact has not been proved.

Decree reversed.