American Steamship Company, owner of the bulk carrier ship, the Buffalo, appeals from a judgment holding it liable for fifty percent of the damages from an explosion that occurred when the Buffalo passed the Jupiter, a tankship that was moored and unloading its cargo of gasoline. As the Buffalo passed, the Jupiter swung loose from its dock, rupturing both the hose transferring the gasoline to the shore and an electrical cord leading from shore to the ship. A spark from the ruptured cord ignited the spilled gasoline and caused the Jupiter 's cargo of gasoline to explode. American Steamship argues that the district court1 used a legal standard of care applicable only to passing a properly moored ship, whereas the Jupiter was improperly moored. American Steamship also claims that the district court erred in apportioning liability between American Steamship; Total Petroleum, Inc., the wharfinger; and Cleveland Tankers, Inc., owner of the Jupiter.2 Finally, American Steamship contends that the district court erred in excluding from evidence the analysis and conclusions section of the Coast Guard report on the accident. We affirm.

On September 16, 1990, at 1:45 in the morning, the Jupiter moored at the Total Petroleum facility on the Saginaw River in Bay City, Michigan to unload its cargo of 56,000 barrels of gasoline.

The Jupiter 's crew moored the ship with six lines: four wire rope cables attached to winches on the Jupiter and two polypropylene lines. The crew began unloading the gasoline to Total's shore facility by pumping it through a hose running from the ship to a pipe on shore. The Jupiter 's crew monitored the progress of the cargo transfer by means of "ullage pipes," openings from the deck to the hold that permitted the crew to gauge the contents of the hold. There was a "motor-operated valve" to shut off the pipeline on shore from the tanks behind it; the controls to the motor-operated valve were at the end of an electrical cord that was draped over a rail on the Jupiter.

During the early morning hours of September 16 another ship, the Clymer, passed the Jupiter without incident.

Later that morning, the Buffalo entered the Saginaw River from Lake Huron and started upstream toward the Jupiter. The speed at which the Buffalo was going as it approached and passed the Jupiter is the subject of a great deal of conflicting testimony, both eyewitness and expert. As the Buffalo approached the Jupiter, the water level rose about four feet, causing the Jupiter to surge forward slightly, then backward. As the Buffalo passed abeam of the Jupiter, the Jupiter surged forward so that the hindmost mooring wire (the "number four" wire) tautened. The number four wire was moored to the center piling (the "kingspile") on the "number four" cluster of pilings. When the number four wire grew too taut, the number four kingspile snapped, releasing both the number four wire and one of the polypropylene lines and leaving the Jupiter without any moorings at its stern. The stern swung out away from the dock, stretching the cargo hose until it also broke, spilling its contents (120-300 gallons of gasoline) on the pier and on the Jupiter 's deck. At the same time, the electric cord for the motor-operated valve stretching from the wharf to the ship broke and spewed sparks that ignited the spilled gasoline on the pier. Fire quickly spread from the pier along the cargo hose to the Jupiter 's deck. The fire traveled through the ullage pipes to the cargo, causing three explosions.

Several men were injured and one drowned. The Jupiter and her cargo were lost and the Total Petroleum pier was damaged in the fire.

American Steamship and Cleveland Tankers both filed for limitation of, or exoneration from, liability under 46 U.S.C.App. Sec. 183(a) (1988), the Limitation Act. Under the Limitation Act, a ship owner is entitled to exoneration if he, his vessel, and crew are found to be completely free of fault. In re Complaint of Caribbean Sea Transport, Ltd., 748 F.2d 622, 626 (11th Cir.1984), modified on other grounds, 753 F.2d 948 (11th Cir.1985). Even if not completely free from fault, the ship owner is entitled to limitation of liability if the ship owner had no knowledge of or privity to the ship's negligence or unseaworthiness. S & E Shipping Corp. v. Chesapeake & Ohio Ry. Co., 678 F.2d 636, 642 (6th Cir.1982); Matter of M/V Sunshine, II, 808 F.2d 762, 764 (11th Cir.1987). The burden of proving negligence lies on the person claiming to be injured, but once negligence is established, the vessel's owner must prove lack of knowledge or privity to the negligence. Id.

After a twenty-nine-day trial, the district court held that the Buffalo was negligent in going too fast as it passed the Jupiter and that the surge due to the excessive speed caused the accident. The district court held that Total Petroleum and Cleveland Tankers were also negligent and that their negligence contributed to the accident. Total Petroleum owned and maintained the pier. The pier's number four kingspile, which broke, was 90-95% rotten. The court specifically found that the dry rot in the kingspile was a proximate cause of the accident and that Total had unreasonably failed to discover the dry rot before the accident. The court also found the Jupiter was unseaworthy because the flame screens in the ullage pipes, which were intended to protect the cargo from fire on deck, were not fitted closely enough in the pipe to carry out their intended function. Furthermore, the court found that the Jupiter 's crew was negligent in failing to close the caps on the ullage pipes while the Buffalo passed, and that this failure contributed to the accident by providing a pathway for the flames from the deck to the cargo. The court found that neither Cleveland Tankers nor American Steamship had borne its burden of proving it had no privity to its vessel's negligence.

The district court apportioned liability on the basis of fault under United States v. Reliable Transfer Co., 421 U.S. 397, 95 S. Ct. 1708, 44 L. Ed. 2d 251 (1975). The court found the distribution of fault to be fifty percent to the Buffalo, and twenty-five percent each to the Jupiter and Total Petroleum.

American Steamship's principal argument is that the district court based its finding of negligence on a standard of care that was not pertinent to the facts of this case. It argues that the court applied a presumption from O'Donnell Transportation Co. v. M/V Maryland Trader, 228 F. Supp. 903, 909 (S.D.N.Y.1963), that a ship must proceed so as to do no harm to a properly moored vessel, whereas the defective condition of the number four kingspile made the Jupiter an improperly moored vessel. What a passing ship's duty is to an improperly moored vessel is an interesting question, as is the issue of whether a vessel is improperly moored if properly tied to a defective dock. Nevertheless, given the actual disposition of this case, those questions need not be reached. The court in this case did recite the O'Donnell test and did rely upon it in holding the Buffalo negligent. At the same time the court also found the Buffalo negligent using a completely independent and sufficient test, stating that if a vessel violates a written statutory or administrative rule of navigation designed to prevent collisions, the burden is on the vessel at fault to show that its violation could not have been a contributing cause of the collision. The Pennsylvania, 86 U.S. (19 Wall.) 125, 136, 22 L. Ed. 148 (1873). The court explicitly found that "the Buffalo passed by the Jupiter at an excessive rate of speed in violation of the Inland Navigation Rules Act, the Navigation Safety Regulations,3 the Michigan Marine Safety Act, the Bay City Code, and the ordinances of the Township of Bangor and the City of Essexville."4

The court found that the Buffalo 's violation of these rules in passing at excessive speed was "a proximate cause of the accident which occurred on September 16, 1990." Specifically, the court found that the explosion resulted from the rupture of the gasoline hose and electrical lines, which in turn resulted from the Jupiter 's response to hydrodynamic forces caused by the "speed of the Buffalo and the proximity with which she passed the Jupiter." These findings amply satisfy the Pennsylvania standard for liability, making it unnecessary to determine whether the court may have erred in applying the O'Donnell presumption to a vessel moored to a defective dock. Thus, even if the court erred in applying the O'Donnell test inappropriately, the Pennsylvania holding provides an adequate ground for its ruling. "[A] reviewing court can sustain the judgment of a lower court on any ground that finds support in the record." United States v. Anderson County, 761 F.2d 1169, 1174-75 (6th Cir.), cert. denied, 474 U.S. 919, 106 S. Ct. 248, 88 L. Ed. 2d 256 (1985).

American Steamship next contends that the district court erred in finding that the Buffalo was traveling six to nine miles per hour as it passed the Jupiter, and that it created a wake.

We may set aside a district court's findings of fact in an admiralty proceeding only if they are clearly erroneous. United States v. Paducah Towing Co., 692 F.2d 412, 421 (6th Cir.1982). "[A] finding is 'clearly erroneous' when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.... Where there are two permissible views of the evidence, the factfinder's choice between them cannot be clearly erroneous." Anderson v. City of Bessemer City, 470 U.S. 564, 573-74, 105 S. Ct. 1504, 1511, 84 L. Ed. 2d 518 (1985) (internal quotations and citation omitted).

The district court discussed at length the eyewitness and expert testimony on both sides of the hotly contested question about the speed of the Buffalo as it passed the Jupiter. The district court based its finding of excessive speed on the testimony of three eyewitnesses and two experts. The court specifically rejected the testimony of two eyewitnesses American Steamship relied on (one partly on the basis of his demeanor) and of the Buffalo 's captain, MacFalda.5 These credibility judgments are particularly the province of the district court, with only a narrow scope for appellate intervention when a district court has relied on testimony that simply cannot be believed. See Anderson, 470 U.S. at 575, 105 S.Ct. at 1512 ("[D]ocuments or objective evidence may contradict the witness' story; or the story itself may be so internally inconsistent or implausible on its face that a reasonable fact-finder would not credit it. Where such factors are present, the court of appeals may well find clear error even in a finding purportedly based on a credibility determination."). This case does not present that situation. Here, the testimony of different witnesses directly contradicted each other, and the district court made a necessary choice between two versions.

The district court gave good reasons for rejecting American Steamship's expert testimony that the Buffalo was traveling about three miles per hour. The court rejected the testimony both because it contradicted other expert testimony and because American Steamship's expert conceded that he failed to consider relevant factors and that the Jupiter would not have surged at all if the Buffalo had only been travelling 3.3 miles per hour.

American Steamship argues that the eyewitness testimony concerning excessive speed is discredited by pictures taken at the time of the accident that show no wake on the Buffalo 's bow. The pictures themselves demonstrate that they were taken when the Buffalo was well past the Jupiter. Of the series of twelve photographs, only one shows the bow of the Buffalo, partly concealed by the guard rails of a bridge. Rose Cooper, who took the pictures, testified that she and her husband had stopped their car on the bridge because the draw bridge was up. They saw the Buffalo coming toward them. Cooper got out of the car when she saw a flash out of the corner of her eye. She saw a fire had started and so began taking pictures. Cooper observed the Buffalo through the lens of her camera as it moved beyond the Jupiter, and the Buffalo seemed to be moving slowly. From studying the photographs, she concluded the Buffalo sped up after the fire started, because in the later pictures there was more water moving behind the boat. The district judge made no specific findings with reference to Cooper's testimony or the pictures. The essence of American Steamship's argument is that the photographs trump the testimony on which the district court based its findings. From viewing the photographs and considering the testimony before the district court, we are not persuaded that the photographs, taken after the passage and in which the relevant area (the Buffalo 's bow) is partially obscured, required the court to reject the eyewitness testimony concerning the wake the Buffalo created. A number of eyewitnesses testified that they saw a significant "swell," "draft," or "wake" on the Buffalo 's bow or that it was "pushing the water on [its] bow" as it passed. The pictures were not taken at the proper moment in time to require the district court to reject testimony it found to be otherwise credible.

There was ample basis for the district court's conclusion that the Buffalo was travelling six to nine miles per hour. Under the clearly erroneous standard, we may not reverse.

American Steamship next argues that the district court erred in apportioning fault among the Buffalo, the Jupiter, and Total Petroleum. American Steamship contends that Total's fault in failing to adequately inspect the pilings in its docks and the Jupiter 's fault in not closing off the ullage pipes or having better flame screens outweighed any fault of the Buffalo. American Steamship contends that, although the district court found the Jupiter negligent in leaving the ullage caps open and in using ill-fitting flame screens in its ullage pipes, the court did not recognize that the negligence of Total and the Jupiter changed the nature of the accident from a minor break-away incident to a catastrophic explosion. American Steamship also continues its attack on the district court's fact findings in rejecting American Steamship's other allegations of negligence by the Jupiter.

Key to our consideration of this point is the limited standard of review. The district court's apportionment of fault is subject only to review under the clearly erroneous standard, unless the court applied the wrong legal standard. Interstate Towing Co. v. Stissi, 717 F.2d 752, 753 (2d Cir.1983).

American Steamship makes two sorts of factual-error arguments. First, it argues that the district court erred in rejecting two of American Steamship's claims that the Jupiter was negligent. Specifically, American Steamship claims the district court should have found the Jupiter negligent in its handling of the number three winch. The court held that the Jupiter would not have prevented the accident by heaving on the number three winch when the ship was surging forward, since the number three wire was leading forward and heaving on it would have accelerated the forward motion. American Steamship argues that the court should have found negligence in the handling of the winch during the Jupiter 's movement backward, before the disastrous forward surge. American Steamship claims that the computer models of Professor Armin Troesch prove that Jupiter crewman Skinner must have released the brake on winch number three, permitting the number three wire to unwind freely so that the Jupiter moved nine feet aft. Professor Troesch's computer model is based on numerous assumptions, extrapolated from the various estimates the eyewitnesses gave. Varying the values of Troesch's assumptions (e.g., whether the Buffalo was traveling eight miles per hour or some other speed, whether the aft surge was nine feet or some other distance) would change his results. On the other hand, Skinner's own testimony is that he did not arrive at the winch controls until the ship was moving forward and the number four kingspile was bending over, on the verge of breaking. Again, in essence, American Steamship is attempting to give preemptive weight to its expert testimony. The testimony at most simply conflicts with other evidence that the district court chose to believe. We have reviewed the underlying evidence and hold the district court's finding was not clearly erroneous.

American Steamship also argues that the court erred in failing to find that the Jupiter could have prevented the fire by closing the motor-operated shoreside valve and shutting off the tanks from the transfer hose. American Steamship claims that, had the Jupiter shut off the motor-operated valve, gasoline would not have flowed back from the tanks through the hose to spill on the deck and there would not have been enough gasoline on the Jupiter 's deck to fuel the fire. This argument depends on an assumption, which Cleveland Tankers disputes, that the contents of the hose would not have spilled out without back pressure from the tanks on shore.

In its reply brief, American Steamship switches from the argument that the gasoline would not have flowed onto the deck if the motor-operated valve had been off to the argument that the gasoline would not have flowed with as great a force as it did. This argument is not inconsistent with the district court's finding that the gasoline contained in the hose itself was sufficient to fuel the fire, which was supported by testimony at trial. The Jupiter 's Captain Beckwith testified that: "[A] little bit of gasoline puts off a lot of vapor and your vapor, of course, is what burns." Similarly, First Mate Walton stated: "A little bit of gas goes a long way." The record supports the district court's conclusion that the gasoline in the cargo hose was sufficient to precipitate the explosions.

American Steamship cites Interstate Towing Co., 717 F.2d at 752, and Cement Division, National Gypsum Co. v. City of Milwaukee, 915 F.2d 1154 (7th Cir.1990), cert. denied, 499 U.S. 960, 111 S. Ct. 1583, 113 L. Ed. 2d 648 (1991), as cases in which an appellate court reversed a district court's allocation of fault. Neither is persuasive in this case. In National Gypsum, the district court erred in dividing liability according to amount of property at risk, rather than fault. 915 F.2d at 1159. In Interstate Towing, the district court based its allocation in part on a mistaken legal conclusion that one party was only vicariously liable. 717 F.2d at 757. There is no such legal error in this case.

American Steamship also contends that the court erred in allocating fault among the acts of negligence, because it argues that the subsequent negligence of Total and Cleveland Tankers radically changed the nature of the accident.6 In this case, Total was held to be negligent in maintaining a dock with a rotten piling. Total's negligence created the same danger as did the Buffalo 's speeding: both risked a breakaway by the Jupiter while transferring its flammable cargo.7 The Jupiter was held negligent for leaving the ullage pipes open and for having poorly fitted frame screens in the ullage pipes. These acts of negligence came into play only after the Jupiter had broken away, the hose had ruptured, and the gasoline had spilled on the deck and ignited. The risk of deck-fire and explosion was well on its way to fruition before the Jupiter 's negligence made its contribution to the disaster. The district court's allocation of fault is not clearly erroneous.

Finally, American Steamship argues that the district court erred in refusing to admit the analysis and conclusions section from the Coast Guard report on the accident. See In re Cleveland Tankers, 821 F. Supp. 463, 465 (E.D.Mich.1992).

The Ninth Circuit held in Huber v. United States, 838 F.2d 398, 402-03 (9th Cir.1988), that admitting the conclusions section of a Coast Guard accident report in liability proceedings would defeat the announced purpose of 46 C.F.R. Sec. 4.07-1(c) (1994)8 by discouraging Coast Guard personnel from making their most candid assessments of the causes of the accident. Section 4.07-1(c) clearly indicates that the Coast Guard accident report is to be a safety-maximizing tool, not a forensic aid. In re Complaint of Nautilus Motor Tanker Co., 862 F. Supp. 1251, 1256 (D.N.J.1994), decided not to follow Huber because the Coast Guard was not a party to the suit. That court reasoned that the rationale of Huber was not applicable where the Coast Guard's interests were not at risk in the proceeding. We read Huber more broadly, concluding that the function of the Coast Guard reports is altogether different from that of fixing liability. The Coast Guard report is, to a great extent, forward-looking, since it is meant in part to aid in developing rules to make shipping safer. See 46 C.F.R. Sec. 4.07-1(c)(2). In a lawsuit, a court must look backward to facts and rules as known to the actors at the time of the accident. Introducing the Coast Guard's conclusions may confuse the two sorts of inquiries. Huber 's reasoning is sound and we follow it in affirming the district court's exclusion of the analysis and conclusions section in this case.9

* * *

We affirm the judgment of the district court.