On the evening of April 11, 1997, two vessels collided in the Gulf of Mexico. The chemical tanker M/V Formosa SIX ("Formosa Six") was heading outbound from the mouth of the Mississippi River, going south toward a voluntary fairway system located in an area called the Southwest Pass. The bulk carrier M/V Flora ("Flora") was heading in a northerly direction inbound. Despite the apparent expanses of available water, Flora managed to strike Formosa Six, damaging both vessels and causing Formosa Six's liquid chemical cargo, owned by Mitsubishi Corporation, to pour into the sea. The collision occurred south of a sea buoy located just south of the Southwest Pass.

In April 1997, Mitsubishi filed a claim in the Eastern District of Louisiana against Flora and Six interests1 for the cargo lost and damaged. The owner of Formosa Six, Formosa Plastics Tanker Corporation, filed a separate action against Flora interests. Flora's owners (Segesta Shipping Company, Ltd.) and managers (EFNAV Company, Ltd.) filed a petition seeking exoneration from or limitation of liability. These cases were consolidated. Flora interests later filed a counterclaim in the consolidated case against Formosa, seeking recovery of damages sustained by Flora. Mitsubishi filed a motion for partial summary judgment against Flora interests in April 1998, and the district court granted that motion, finding that Flora was at fault in the collision and was therefore obliged to fund the cargo damage claim. Flora interests filed a motion for reconsideration, which the court denied. The court bifurcated the liability and damage issues, and the liability trial was held in December 1998. The court issued an order in January 1999, finding Flora 80 percent responsible for the collision and Six 20 percent at fault. Flora again filed a motion for reconsideration, which the court denied. Following a stipulation on damages, the court entered a damages judgment in August 1999, awarding Formosa Six interests $3,343,179, plus interest and costs, and dismissing Flora's limitation action. The judgment also acknowledged settlement by the parties of the Mitsubishi claim by contributions of $220,000 from Formosa interests and $880,000 from Flora interests. Flora appeals and we affirm.

In this appeal, Flora argues that the district court erred as a matter of law in determining the proximate cause of the collision by failing to apply pertinent law. In the alternative, Flora contends that the district court's apportionment of fault was clearly erroneous. We address these arguments in turn.

Flora contends that the district court erred as a matter of law by failing to apply several arguably applicable regulations. Collisions at sea are governed by internationally accepted regulations recognized by treaty; in the United States these are codified at 33 U.S.C. foll. § 1602 ("COLREGs"). In arguing that the district court failed to properly apply these regulations, Flora apparently hopes that this court will review the findings of the trial court de novo rather than under the appropriate clearly erroneous standard. Conclusions of law are, of course, reviewed de novo. See Phillips Petroleum Co. v. Best Oilfield Servs., 48 F.3d 913, 915 (5th Cir. 1995). When reviewing mixed questions of law and fact, this court should reverse only if the findings are based on a misunderstanding of the law or a clearly erroneous view of the facts. See Bose Corp. v. Consumer's Union of U.S., Inc., 466 U.S. 485, 501 (1984). Because the proper regulations were considered and applied, as we discuss below, we find that there was no misunderstanding of the law.

Flora's argument that the district court erred in its application of the relevant regulations is grounded in a rule for finding liability in collisions at sea -the Pennsylvania rule. Under the Pennsylvania rule, if a vessel involved in a collision was violating a statutory rule intended to prevent collisions, the burden shifts to the violating vessel to show that its fault could not have been a cause of the accident. See The Pennsylvania, 86 U.S. (19 Wall.) 125 (1874); see also Garner v. Cities Serv. Tankers Corp., 456 F.2d 476, 480 (5th Cir. 1972). The rule thus creates a presumption that one who violates a regulation intended to prevent collisions will be deemed responsible; but that presumption is rebuttable. See Otto Candies, Inc. v. M/V Madeline D., 721 F.2d 1034, 1036 (5th Cir. 1983). Further, as discussed below, the Pennsylvania rule applies only to violations of statutes that delineate a clear legal duty, not regulations that require judgment and assessment of a particular circumstance.

Specifying the rules that her adversary allegedly violated, Flora first contends that the trial court erred by failing to take into account Formosa Six's violation of COLREG Rule 5 in apportioning responsibility for the collision. COLREG Rule 5 provides:

Every vessel shall at all times maintain a proper look-out by sight and hearing as well as by all available means appropriate in the prevailing circumstances and conditions so as to make a full appraisal of the situation and of the risk of collision.

33 U.S.C. foll. § 1602. Flora is asking this court to find fault with the district court's failure to acknowledge that Six's lack of a look-out was "crucial," and the proximate cause of the accident. Because Flora is here taking issue with the district court's finding of fact, we review this aspect of the district court's decision for clear error. See Bose, 466 U.S. at 501.

The trial court found that Formosa Six had not complied with the cited regulation; that is not disputed. Thus, Flora's argument involves the district court's weighing of the evidence, not its application of the law. The court erred, Flora contends, by failing to impose the Pennsylvania burden. But the district court clearly did consider the predicate facts that constitute a violation of COLREG Rule 5. The district court also found that Formosa Six violated that regulation, and that such failure was a contributing cause of the collision. See Tokio Marine & Fire Ins. Co., Ltd. v. M/V FLORA, No. CIV. A. 97-1154, 1999 WL 14000, at *14 (E.D. La. Jan. 11). The court, in fact, found that this failure "was a contributory cause, though a relatively small one, of the collision." Id. at *14. Flora argues that this was an error of law, and then cites the importance of complying with Rule 5 as well as the importance of having a look-out on the Formosa Six at this critical juncture. The importance of the regulation does not change its application, nor does it impugn the finding of fact made by the district court. The dispute Flora has is with the district court's finding of fact that Formosa Six's failure to have a look-out was not as "crucial" a factor in the collision as Flora seeks to make it. However, neither Formosa Six nor the district court clearly indicated why this failure was not to be considered a more critical cause of the accident. But because this finding is reviewed only for clear error, and Flora has not demonstrated error of that sort, we find that the district court, in its considered judgment, properly weighed Formosa Six's violation of Rule 5 in determining the proximate cause of the collision.

Flora next argues that the district court failed to impose the "requirement" embodied in COLREG Rule 8(c) on Formosa Six. Rule 8(c) states:

If there is sufficient sea room, alteration of course alone may be the most effective action to avoid a close-quarters situation provided that it is made in good time, is substantial and does not result in another close-quarters situation.

33 U.S.C. foll. § 1602. To fail to refer to this provision of the rule explicitly in its findings of fact is certainly not clear error, let alone a mistake of law. First, the district court did consider Rule 8 in making its findings. The court recognized a duty under Rule 8 to take action to avoid collision, and that both Flora and Formosa Six had failed to fulfill this duty. See Tokio Marine & Fire, 1999 WL 14000, at *12. Second, the language of Rule 8(c) is suggestive, rather than mandatory. The Pennsylvania rule applies in cases in which a "precise and clearly defined duty" is mandated by the relevant statute, not when the statute "calls for the use of interpretation and judgment." Interstate Towing Co. v. Stissi, 717 F.2d 752, 756 (2d Cir. 1983). The provision of the rules at issue here falls within the latter category rather than the former. Thus, the burden did not have to shift to Formosa Six for a violation of Rule 8(c), and the court cannot have erred in not shifting more of the onus on Formosa Six for violating it. Indeed, the Pennsylvania rule was not even mentioned in Zim Israel Navigation Co., Ltd. v. Special Carriers, Inc., M/V, 611 F. Supp. 581 (E.D. La. 1985), a case in which one party was deemed to have violated Rule 8(c). The fact that the rule had been violated was considered in determining the cause of the collision, but no presumption of fault was invoked. It is evident that the district court did not err in its application of Rule 8, nor in its failure to mention Rule 8(c).

Flora next contends that "special circumstances" justify its own violations of the COLREGs, and that these same circumstances demand a finding that Formosa Six violated COLREG Rule 2. That rule states:

Nothing in these Rules shall exonerate any vessel, or the owner, master or crew thereof, from the consequences of any neglect to comply with these Rules or the neglect of any precaution which may be required by the ordinary practice of seamen, or by the special circumstances of the case.

33 U.S.C. foll. § 1602. The "special circumstances" to which Flora refers involve an alleged custom that, in the Southwest Pass, outbound vessels must take extra care when leaving the area to avoid a collision, and inbound vessels have the right of way. But the existence of such a custom was convincingly undermined by two bar pilots' testimony.2 On the other hand, Captain Douglas Torborg, the Flora expert mariner, testified to the existence of the custom. But the district court did not take this alleged custom into account in determining fault, which Flora contends was an error of law.

However, a court does not err as a matter of law by failing to take into account a custom whose existence is successfully disproved. The court properly found that Flora's violation of a port-to-port passing agreement, which it had entered into with Six some minutes before the collision, had "no basis in maritime law or custom." Tokio Marine & Fire, 1999 WL 14000, at *12. We see no reason to conclude that an alleged custom giving inbound vessels right of way vitiated the port-to-port agreement, or required apportionment of greater responsibility for the collision to Formosa Six.

Both bar pilots testified that there was no such custom. An amicus brief filed by the Associated Branch Pilots of the Port of New Orleans asserts the same, arguing that there is no custom assigning a "right of way" in the Code of Federal Regulations or in the practice of pilots and mariners. This conclusion is buttressed by the decision of this court in Hal Antillen N.V. v. Mount Ymitos MS, 147 F.3d 447 (5th Cir. 1998), which reversed a district court's finding that in the area at issue in this case there existed a custom of passing starboard to starboard, not port to port, as the COLREGs provide. Noting that a custom that conflicts with the COLREGs would be "confusing," that this custom was not published and that there was "highly contradictory testimony" about whether the custom existed, this court held that the district court's finding that such a custom existed was clearly erroneous. See id. at 451. Flora does not dispute this, but instead argues that because this court in Hal Antillen N.V. apparently left intact the district court's finding that inbound vessels have the right of way in the Southwest Pass, that custom does exist. We believe, however, that there is sufficient evidence in the present case to support the district court's apparent conclusion that there is no right-of-way preference for inbound ships.

Even if there were such a custom, the district court found that the present vessels had entered into an agreement to pass port to port. Any such agreement would override custom. As we noted in Canal Barge Co., Inc. v. China Ocean Shipping Co., "Even when the custom prevails, pilots of approaching vessels may agree to pass in some fashion other than in the manner provided by habitual practice." 770 F.2d 1357, 1361 (5th Cir. 1985).

Flora's next line of attack is aimed at the district court's findings of fact that led to its conclusion that Flora was 80 percent at fault for the collision. We recapitulate briefly the court's findings that led to this conclusion. Formosa Six had just completed a downriver passage from the Mississippi River, and was proceeding from the Southwest Pass in the Gulf of Mexico toward the sea buoy just south of the Southwest Pass. Six was being directed by compulsory Bar Pilot Kevin Leger. Vessels departing the Southwest Pass proceed on a southerly heading into the voluntary fairway system, which looks like an inverted "Y" and bifurcates into two legs proceeding respectively in the southeast and the southwest directions. Six was being steered by hand, aided by two radars and one Automated Radar Plotting Aid (ARPA). The ARPA was equipped with a closest point of approach (CPA) alarm, set to alert navigators to any vessel that would pass within a 0.5 nautical mile radius of Six within a 20-minute time period. Flora was proceeding on a northerly course in the Gulf of Mexico, inbound and seaward of the Southwest Pass sea buoy. Flora was equipped with an ARPA and three radars. However, Captain George Gatsos, the vessel's master, testified at trial that he used only one radar and did not turn on the ARPA that evening.

Somehow, these two vessels, proceeding on a clear night in a relatively large fairway, collided. How this could happen with abundant sea room available may be mystifying to landlubbers, but navigation at night has its special challenges. What exactly happened here was a source of furious contention in the trial court. At around 2005 hours, as Six was entering the Gulf of Mexico, the Six crew observed Flora approaching and set Flora on its ARPA with a CPA of 0.5 and a CPA time of 20 minutes. Thus, the CPA alarm would sound when Flora was in a position to be within 0.5 nautical miles of Six not more than 20 minutes later. At about 2007, Bar Pilot Leger, aboard Six, conferred by radio with Flora and agreed to pass port-to-port.3 Six then turned to starboard to "make a lee" (turning to protect a small craft from the wind) to shelter Leger in departing the ship. At no time in this sequence did Six have a full-time watch. Shortly thereafter, the pilot boat that had picked up Leger asked Flora to make a lee for Bar Pilot Samuel Eddy to board Flora. Flora stated that it would turn to starboard to make a lee for the bar pilot, but instead turned to port. Flora Captain Gatsos did not notify the pilot boat or the Formosa Six that he was turning to port instead of to starboard. Thus, according to the district court, Flora made a sudden, hard and unexpected turn to port just prior to the collision. A little before this, after dropping off Pilot Leger, Formosa Six returned to a southerly course and shortly learned that it had moved into a dumpsite at the west side of the fairway. The vessel turned somewhat to port in an effort to maneuver out of the dumpsite area, still expecting to accomplish a port-to-port passage.

The location of the collision, and the speed at which the two vessels were traveling, was hotly disputed at trial. The district court ultimately found that the location of the collision was consistent with Formosa Six's version of the story: it occurred approximately at the position claimed by Six and was based on a global satellite positioning reading taken by Flora's third mate at the time of collision -on Six's side of the fairway. The court also based this finding on its acceptance of the testimony of Formosa Six's expert witness, Gary Maseuth, as well as on the location of the chemical spillage from Six on the ocean floor. The court rejected the testimony of Captain Torborg, Flora's expert mariner, who testified that no one could determine the position of the vessels at the time of collision, and that global satellite positioning data can be inaccurate. This district court finding was significant because it showed that the Flora "had moved far from the center of the safety fairway and was encroaching significantly on to the part of the fairway occupied by the FORMOSA SIX." Tokio Marine & Fire, 1999 WL 14000, at *8. The court also found that Flora was traveling at a speed of at least two knots, which was significant because, if Flora had been going much slower, as Captain Gatsos testified, it would have been powerless to avoid the accident in the last few minutes, when a collision appeared likely to occur.

This court reviews the district court's findings of fact for clear error. See Fed.R.Civ.Proc. 52(a); American River Trans Co. v. Kavo Kaliakra SS, 148 F.3d 446, 449 (5th Cir. 1998). This standard also applies to apportionment of fault. See Inland Oil & Transp. Co. v. Ark-White Towing Co., 66 F.2d 321, 325 (5th Cir. 1983), abrogated on other grounds, City of Milwaukee v. Nat'l Gypsum Co., 515 U.S. 189 (1995). The Supreme Court articulated the meaning of this standard of review:

If the district court's account of the evidence is plausible in light of the record viewed in its entirety, the court of appeals may not reverse it even though convinced that had it been sitting as the trier of fact, it would have weighed the evidence differently. 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 (1985). Findings based on the credibility of witnesses demand even greater deference. See Fed. R. Civ. P. 52(a); Anderson, 470 U.S. at 575. Flora points to nothing that resembles clear error.

Flora argues vociferously that the trial court erred by using the respective distance toward each other covered by the two vessels immediately before the crash to determine the 80-20 proportion of liability between the parties. Flora notes, correctly, that "[n]o jurisprudence supports an apportionment of fault based on distance traveled from two arbitrary points." The calculation of distance covered, Flora contends, was the sole basis for the district court's apportionment of fault. Flora claims that Formosa Six was more at fault because its turns were so subtle that Flora could not detect them and thereby come to the realization that port-to-port passing would not be feasible. Formosa Six, Flora contends, was in the best position to recognize Flora's turning movement for what it was and steer out of the way. Further, Flora argues, the fact that Six was traveling much faster than Flora meant that it could have avoided the accident more easily. All these arguments imply that, if the district court had taken them into account rather than comparative distances covered by the vessels after an arbitrary moment in time its findings as to proportion of fault would have been different. But the premise underlying this argument is incorrect.

The district court did take into account other factors in calculating the proportion of fault attributable to each party. It found that an "overwhelming percentage of the accident was caused by Flora's violation of the port-to-port agreement and lack of communication with the SIX." Tokio Marine & Fire, 1999 WL 14000, at *17. It took account of Formosa Six Captain Kuo-Hsiung Chen's lack of awareness, which it noted "may have contributed to the collision." Id. And it concluded that "the brunt of the blame" fell on Flora, because it "creat[ed] the risk of collision by executing a precipitous turn in direct violation of its port-to-port agreement with the FORMOSA SIX." Id. The court was also clearly influenced by the post-accident alterations in Flora's log book. "The unexplained alteration of a ship's record of maneuvers 'not only cast[s] suspicion on the whole case of the vessel, but creates a strong presumption that the erased matter was adverse to her contention.'" Andros Shipping Co. v. Panama Canal Co., 184 F. Supp. 246, 259 (D.C.Z. 1960) (quoting The Chicago, 94 F.2d 754, 762 (9th Cir. 1937)).

The court certainly took distance traveled into account, but that was clearly not the only factor that contributed to its conclusion as to the proportions of fault. We do note that the distance traveled toward each other during the four minutes before the collision, as the court found, was consumed about 80 percent by Flora and 20 percent by Formosa Six. It may be unusual that the trial court chose to cite their proportions in close proximity to its apportionment of fault. But using the respective distances covered in moving to the point of collision is not arbitrary or irrational. These proportions do tend to show each vessel's linear contribution to moving from a safe course to a collision course. In any event, even if we might have given different weight to different pieces of evidence than did the district court, this is not a reason to disturb that court's findings of relative responsibility, absent a showing of clear error. Here, there was no such showing.

Flora also argues that it did not unilaterally terminate the port-to-port agreement, and that therefore the trial court's finding that it had done so was clearly erroneous. It argues that "termination" implies that another type of passing was attempted, such as starboard-to-starboard. Perhaps a better term for Flora's actions is that it "violated" the port-to-port agreement rather than "terminated" it. The facts are still the same: Flora agreed to pass port-to-port and then turned sharply to port just prior to collision. There is no clear error in the district court's finding here.

In the alternative, Flora argues that it was not unilaterally responsible for the failure of the port-to-port agreement because Formosa Six had made several gradual adjustments to port prior to its last-minute turn; thus, both parties were responsible. The trial court did not find that Flora was unilaterally responsible for the failure of the agreement; the court explicitly concluded that both vessels violated the agreement. The district court took into account Formosa Six's gradual adjustments in determining Six's fault in the collision, and it relied on the bar pilots' testimony that Six's gradual turn did not contribute to the collision as heavily as Flora's sharp turn.

Last, Flora makes a half-hearted argument that there was no port-to-port agreement, because once each vessel started maneuvering to make a lee, the agreement was abandoned and the vessels were only required to adhere to the COLREG requirements. This argument is without merit. The navigators of both vessels agreed by radio that they would pass port to port, that agreement was recorded and the recording was offered into evidence at trial.4 The fact that there was such an agreement was stipulated by all parties as uncontested.

Next, Flora contends that the trial court committed clear error when it found that Formosa Six might have been able to steer clear of Flora if a look-out had been on watch. Flora objects to the use of the word "might," since Flora believes it is uncontroverted that, if Six had had a look-out, the accident could have been avoided. Captain Torborg testified that Formosa Six could have turned to starboard any time before the collision and thus avoided the collision altogether. Whether to accept Torborg's testimony is the prerogative of the trial court, which is in a position to assess the credibility of witnesses. Again, the district court considered Formosa Six's fault, and came to the conclusion that despite Six's critical error in failing to have a look-out at all times the failure was not a major factor in causing the collision. There is simply nothing in the record to demonstrate that this conclusion is erroneous.

Flora also argues that the district court's conclusion that it turned hard to port was not supported by the evidence. Flora claims that the testimony on which the district court relied that of the pilots in the pilot boat was unreliable because they were in no position to determine speed and did not see the collision. While it is true that they did not see the collision itself, they could see much of the maneuvers of the vessels before the crash. The district court carefully balanced the testimony of Pilot Eddy against the testimony of Captain Gatsos, and chose to credit Pilot Eddy's version of the story. Gatsos testified that his turn to port was not rapid or sudden but part of a long turn commenced long before the collision occurred. But Eddy, the court noted, "clearly saw the FLORA make a sudden turn hard to port about two minutes before the collision." Tokio Marine & Fire, 1999 WL 14000, at *4. Leger testified to the same effect. See id. Thus, although the two bar pilots were in no position to witness the actual collision, the district court accepted their testimony that they were able to recognize Flora's sharp turn to port shortly before the collision. We see no clear reason to reject this conclusion. We also see no reason to reject the trial court's conclusion that the pilots' testimony was credible. It was reasonable to reach such a conclusion: the pilots were disinterested observers, their account was consistent with Formosa Six's account and their account was contradicted only by that of a witness found not to be credible by the district court--Captain Gatsos.

For the foregoing reasons, we Affirm the judgment of the district court.