In this proceeding any similarity between the libellant's version of the facts and the respondent's version seems to be purely accidental and not because they both relate to the defendant's tug Felicity Anne, or its tow, having struck the libellant's barge on the evening of September 9, 1956, the accident that generated this litigation.

Girod Motor Company, Inc., libellant-appellee, is engaged in the business of marketing sand and gravel dredged from the Mississippi River. In the operation of its business, Girod employs barges, tow boats, a crew boat, draglines, trucks, and other equipment. Its chief customer in September 1956 was the Fordyce Construction Company, a company engaged in manufacturing concrete mats for revetment work on the Mississippi River. On September 9, 1956, the date of the accident, Girod's plant was operating twenty-four hours a day, dredging, cleaning, classifying, marketing, and stockpiling sand and gravel.

Girod filed a libel against the tug Felicity Anne, owned by Hanson Dreijer Marine Enterprises, and asked for the recovery of damages for: (a) the total loss by sinking of a wood-hull composite barge, allegedly struck and damaged by the Felicity Anne; (b) loss of equipment on board the barge, including two trucks, the existence of which is denied by the respondent; (c) damage to a crew boat which respondent's witnesses say they did not see; (d) loss of a gravel barge (missing and never found), which respondent's witnesses say they did not see; (e) salvage costs; and (f) loss of production in the nature of demurrage. The district court gave judgment, in part, for the libellant. Only the respondent appeals.

The unusual feature of the case is that the actual sinking of the composite barge took place October 17, 1956, thirty-nine days after the day the Felicity Anne caused the damage allegedly responsible for the sinking. In the meantime the September 9 damage was surveyed and the barge was repaired and put back in operation.

On September 9, 1956, Girod's composite barge was moored in Delta Chute, a small cut of the Mississippi River, at Delta, Louisiana. According to the libellant, fifty feet upstream from this barge was a crew boat, and 400 to 600 feet upstream a steel deck gravel barge (No. 20). The respondent denies their existence. The trial judge found that the composite barge, the crew boat, and No. 20 barge were in existence, properly lighted, and properly moored. At dusk the Felicity Anne arrived light at Delta to fetch two empty cement barges from the facilities of Fordyce Construction Company at Delta, and to return them to Baton Rouge. These cement barges were slightly downstream from the composite barge. The Felicity Anne's master planned to form a single unit of tug and barges, when made up stern-log to stern-log, a difficult maneuver to make at that particular place. The Felicity Anne's bow was to be secured to one of the cement barges so that the barges could be towed out into the chute where the water was deep enough and where there was sufficient room to arrange and secure the tow. As the crew of the Felicity Anne tell the story, they attached a line to the libelant's composite barge (then 400 to 600 feet upstream from the cement barges) in order to hold position while securing the wires of their tow; at this point, the cables of the composite barge, which were insecurely attached, slipped loose allowing the barge to drift toward the center of the stream. By Girod's version, which the trial judge accepted, the composite barge was securely cabled to the shore; the Felicity Anne maneuvered negligently, and it is a fair inference that the tug shoved the cement barges upstream so that they struck the composite barge tearing it loose from its moorings and away from the bank. Under either explanation, the composite barge, the Felicity Anne, and the other two barges formed a flotilla and drifted to the opposite bank. Two thirty-foot, steel I-beams, hinged to the deck of the composite barge and used to support a ramp to the bank, were dragged through the water. These dug into the river bottom, spudding down the barge and pinning the flotilla against the opposite bank. Shortly thereafter, Girod's tug, Betty, came to the rescue, pulled the I-beams free, and returned the composite barge to its moorings. The crew boat was completely crushed; she sank at her mooring. The steel deck barge (Barge No. 20) was never found.

On September 12, 1956, representatives of the interested parties conducted a joint survey "to ascertain [the] nature and extent of damage [and] recommended repairs". The barge was repaired and placed back in operation two days after the survey. From time to time a check was made, and although an inspection showed that she was taking some water this did not appear to be serious.

About the first of October, Girod moved the composite barge about twenty miles up the river for use in operations at Brunswick Landing. It stayed there, in use and apparently in good condition, until October 17. That night, without warning, it sank. Girod conducted salvage operations during the next two weeks, but with little success; it recovered a dragline but was unable to raise the barge which was finally abandoned to the United States Engineers. Dreijer contends that the salvage efforts showed a reckless waste of time and money in an effort to salvage an ancient wood-hull barge worth $5000 at the most, that from the beginning was obviously a total loss. The trial judge allowed $6,240 for salvage costs.

The district court entered a decree in favor of Girod for $26,304.20.1 For purposes of this appeal, we note that the trial judge found: (1) the evidence was insufficient to show that the Felicity Anne or its tow ever struck or came in contact with the steel barge Number 20, 400-600 feet upstream, or that the negligent maneuvers of the Felicity Anne caused Barge Number 20 to break loose from its moorings; (2) the libellant owned the composite barge and was entitled to recover for its total loss; (3) the accident on September 9 was caused by the negligence of the Felicity Anne and her crew; (4) the composite barge was securely moored on September 9 and the accident that night did not result from Girod's negligence; (5) the evidence establishes that a dragline, pipe and other equipment, including two trucks, were on the deck of the barge and were lost overboard as a result of the accident September 9; (6) the libellant is entitled to recover the value of certain lengths of pipe, lost overboard September 9; (7) the sinking of the barge October 17 at Brunswick thirty-nine days after the initial accident was caused by that accident.

A. Appellant contends first that Girod failed to show its ownership of the composite barge; that an oral lease-purchase agreement, on which libellant relies for proof of ownership, is contrary to the Statute of Frauds. By express terms, the Mississippi Statute of Frauds does not apply if "the buyer shall receive part of the personal property, goods, wares and merchandise, or shall actually pay or secure the purchase-money, or part thereof * * *." Mississippi Code of 1942, Section 268. Here, the evidence shows that in accordance with an oral lease-purchase agreement the buyer received the barge and paid part of the purchase price. This case falls within the exception to the statute.

B. There is a direct conflict in the evidence on the question of the Felicity Anne's negligence September 9. Witnesses for Girod testified that the barge was securely held to two trees on the bank by ¾ inch cables, that the cables were inspected many times, that the clamps held tight and did not cut into the wire, and that there were lights on the barges. The only eye-witnesses were the members of the Felicity Anne's crew. There was, however, one witness near the scene. He testified that he had watched the Felicity Anne maneuvering in the river and that shortly after it approached the barge there was a loud noise as if a truck were backing into a wall, and then crashing, snapping, and popping sounds. A photograph showing where the I-beams had been sunk into the river bank indicates that they were wrenched out by a powerful force pushing in an upstream direction. On the other hand, the crew of the Felicity Anne testified that the cables of the composite barge were spliced, "frazzled", "in bad shape", and were inadequately secured to small willow trees; that there were no lights on the barge; that there was no contact at all between the barge and any of libellant's vessels until after the barge had broken away from the bank; that the depth of the water prevented the Felicity Anne from coming closer than eight or ten feet to the vessel; that the upstream mooring of the barge slipped through its clamps and parted, and the downstream wire slipped over the small tree to which it was fastened; that there was no loud noise at any time; and that the I-beams had slipped straight out from the bank. Resolution of the conflict is the province of the trial judge who has the opportunity to observe the witnesses. Sufficient evidence supports his finding of negligence.2 The libellant is aided somewhat by the fact that the Felicity Anne or its tow, moving vessels, struck an anchored vessel.3

C. The defendant admitted that a dragline and miscellaneous equipment were on the deck of the composite barge but denied that there were any trucks. Four witnesses aboard the Felicity Anne testified categorically that they saw no trucks aboard the composite barge. The captain of the Felicity Anne stated that although he had come within two feet of the composite barge and had played the tugboat's floodlight over it, he did not see any trucks. But, Mr. Kilgore, Girod's superintendent, testified that the trucks were on the barge, and the day after the accident Girod recovered two trucks from the river. The marine surveyors stated that the trucks had not been in the river for any length of time. Again the evidence would support a finding either way, and on this point, too, we affirm the holding of the trial judge.

D. The appellant contends that the trial judge allowed damages in the amount of $1282.81 for certain lengths of 16" and 18" pipe lost overboard, an amount $134.49 more than the value agreed to in the parties' stipulation. The record shows that the court dictated the stipulation listing the agreed value of all equipment allegedly lost September 9.4 The lengths of pipe are conspicuous by their absence from this list, and the record shows that the appellant failed to comment on this omission. We note also that Mr. Girod was asked on direct examination: "Now, going back to September 9, 1956, and the materials on board the composite barge. There are stipulations on the value of all the material except the pipe, 16 and 18 inch pipe." Mr. Girod's testimony on the value of the pipe covers five full pages in the record. The damages awarded for the loss of this pipe are slightly less than the value ascribed to the pipe by Mr. Girod. We hold that the finding as to these damages was proper and should stand.

The trial judge awarded the libellant damages for losses suffered when the composite barge sank October 17. This award was based on the court's conclusion that "the collision and strain to which she was placed on September 9 weakened her timbers and seams to the extent that it ultimately sank." We hold that there is not sufficient evidence to support this finding.

The trial judge has the duty of sifting the evidence and choosing among conflicting factual inferences. As a fact finder he is authorized to draw reasonable inferences, and he may disbelieve testimony contradicting inferential evidence. An inference, however, must be supported by evidence of sufficient probative value to form a rational basis for the court's conclusion; it must be drawn by reason from the facts on which it purports to rest. An inference is not a surmise or conjecture. Kentwood Lumber Co. v. Illinois Cent. R. Co., 5 Cir., 1933, 65 F.2d 663; Pearson v. Louisiana & Arkansas Railway Co., 1954, 226 La. 834, 77 So. 2d 411.

The causal relation between the wrongdoing on September 9 and the sinking on October 17 is not clear enough to bring this case within the doctrine that when a ship is damaged but not sunk in a collision, and later is injured or lost, the presumption is that the collision caused the later injury or sinking.5 As we see it, the time interval between the initial accident and the sinking is so long and the range of alternative possible explanations is so great that a finding of causation does not necessarily follow as a matter of logical deduction from the known circumstantial evidence. The assumption that the sinking resulted from the September 9 accident springs not from any indication by prior events that the sinking was likely to occur as a consequence of the accident but rather from the absence of another ready explanation for the sinking. This is admitted by the surveyor, L. M. Fellows, who came to the same conclusion as the trial judge. It is entirely possible, however, that the sinking was a result of some other cause, not the least of which might be the intervening negligence of the owner in failing to take proper care of his vessel.

In asserting that the libellant was negligent, the respondent is at a disadvantage: the pertinent facts are in the possession of the libellant and are not readily discoverable by an opposing party. The evidence however does admit the inference that the shipowner's intervening negligence is a reasonable explanation of the sinking.

To begin with, there is evidence that the timbers suffered no harm from the collision. The stipulated damages for repairs was only $600 and neither Girod nor the surveyors regarded the barge as seriously damaged. Mr. Girod himself told the surveyors that it was not necessary to examine the interior of the hull. Nevertheless, his superintendent inspected the hull and reported that it was sound. The loss of the loading ramp constituted the major damage; this would not affect the structural integrity of the vessel. The only hull damage was a torn wrapper plate on the downstream inshore corner of the vessel well above the water-level, the position of which might indicate that the plate was damaged before the collision. With the benefit of a survey, all apparent injury was repaired. During the thirty-nine day period before the barge sank, Girod used the barge for the last thirty-two days. Although it was necessary to pump water from the barge on occasions, the libellant's witnesses stated that there were no serious leaks. On this evidence, the barge was in satisfactory condition and had sustained no such damage as would account for her sinking — without some intervening cause.

There was no water in the barge at the time of the September 12 survey. Later there was water, but it was pumped out for three or four days and, according to Mr. Girod, "all of a sudden it quit leaking and didn't leak anymore." But the vessel was never hauled out or otherwise examined to ascertain the cause of the leaking, and no precautions were taken except that Kilgore, Girod's superintendent, was instructed to watch it closely; Girod said that at Delta he and Kilgore inspected the bilges every day. Girod continued to use the barge at Delta for sixteen days after discovery of the leak, then towed it to Brunswick where it was used for sixteen more days. Fellows, Girod's surveyor, testified that any wood barge is liable to work while under tow in the open river "and you are bound, when the hull works, to develop a little leak here and there"; accordingly the bilges must be watched closely and checked at least once a day. Kilgore, charged with maintaining Girod's marine equipment, had no marine experience. Although he had checked the barge at Delta, he testified that he had never checked the barge while it was at Brunswick Landing. One of Girod's laborers testified that he checked the barge regularly and had checked it the day before it sank; he found "just a little water", not enough to pump out. After Wall's inspection at 5:30 p. m., the barge was left unattended overnight. In the morning it was found submerged on the revetment.

The barge was forty years old and was a vessel with wood-hull. One should expect the seams to open small leaks when she was towed upstream. If the barge ran aground while being moved or was struck by heavy debris floating down-stream, the seams would have opened. In October 1956 the river was falling about a foot a day; if the barge was not moved in time to prevent its being raised on one side by the effect of the river falling, its timbers were subjected to undue strain.

This case strongly tempts a court to follow the primrose path to post hoc ergo procter hoc reasoning. The district court resisted the temptation when exposed to it by the libellant's claim for steel barge No. 20. Yet, the circumstantial evidence and inferences pointing to the cause of the composite barge's sinking are as few and airy as those pointing to the cause of barge No. 20's disappearance shortly after the September 9 accident. The difficulty is the sufficiency of proof of the cause. When the facts are unascertainable or so meager as they are here and an effect may follow from one of several causes it is a false generalization to suppose that because a sinking follows a collision thirty-nine days later, the sinking necessarily was an effect of the collision. In reaching our decision on the question of causation we have given weight to the slight damage the barge suffered in the September 9 accident, the apparent lack of harm to its timbers, the effect of the survey, the fact that all repairs were made in accordance with the survey, and the later use of the barge for thirty-two days, indicating that foreseeable damage had been repaired. We have considered the age of the barge, her wood-hull, the likelihood that her seams would leak under any conditions and the likelihood of serious leakage on exposure to stress and unusual conditions. We have also considered as significant factors the owner's possession and control of the vessel, his knowledge of the leaks, his awareness of the exposure of the barge to leakage during and after the time she was towed to Brunswick, and the relative lack of care given the vessel at Brunswick. We conclude that it is as plausible to infer that the vessel sank because of an intervening cause, perhaps unknown to the owner, or, that she sank because of the owner's intervening negligence in failing to attend to a vessel known to leak, as it is to infer that the barge sank because of the collision on September 9.

Where the evidence of the party on whom rests the burden of proof is equally consistent with several different hypotheses no single one is proved. Mutual Life Ins. Co. of New York v. Hess, 5 Cir., 1947, 161 F.2d 1, motion to retax denied 5 Cir., 1951, 191 F.2d 817; One 1941 Oldsmobile Sedan v. United States, 5 Cir., 1947, 161 F.2d 348. "Where the facts proven show that there are several probable causes of an injury, for one or more of which the defendant was not responsible, and it is just as reasonable and probable that the injury was the result of one cause or the other, plaintiff cannot have a recovery, since he has failed to prove that the negligence of the defendant caused the injury." Ingersoll v. Liberty Bank of Buffalo, 1938, 278 N.Y. 1, 14 N.E.2d 828, note 4. The burden of proof requires that the party's evidence indicate the facts on which his recovery depends actually did occur, not merely that it is probable that they occurred. See 2 Harper and James, Law of Torts, § 20.2, pp. 1110-1120.

We find that libellant failed to carry the burden of proof: the evidence is too thin to support the inference that the damage occurring October 17 was caused by the negligence of the Felicity Anne September 9. The award of damages for losses incurred in this accident must be reversed.

The judgment is reversed in part and affirmed in part.