dissenting:
Cynthia Sue (“Cindy”) Reber, the wife of Boyd Philip Reber, and Courtney Brianne Reber, his daughter, through her mother as guardian ad litem, brought this action against the United States. They sued under the Public Vessels Act, 46 U.S.C.App. § 781, charging that ships of the United States Navy had negligently caused Boyd Reber’s death. A man who fished for his living, Boyd Reber had been killed at sea. His wife and daughter contended that he had been killed by live ordnance of the Navy. In their view he, his crewman Frank Germano, and his boat the Cindy Fay had been blown up by a naval bomb.
The judgment of the district court must be upheld, although we would have weighed the evidence differently, if the error is not clear. Anderson v. City of Bessemer City, 470 U.S. 564, 573, 105 S.Ct. 1504, 1511, 84 L.Ed.2d 518 (1985). If there are “two permissible views of the evidence,” the factfinder’s choice cannot be clearly erroneous. Id. If the district court’s view is “plausible in light of the record viewed in its entirety,” and not contradicted by extrinsic evidence, we may not reverse. Id.
Measured by this generous standard, the district court’s finding as to causation is clearly erroneous. Its view of the evidence is not uncontradicted by extrinsic evidence. In the light of the record in its entirety its account is not plausible.
The Last Trip of the Cindy Fay. Boyd Reber, 38 years old, had been a commercial fisherman for twenty years. He had a reputation for competence and for care as to safety. On February 21, 1986 he sailed from San Pedro, California in the Cindy Fay.
The Cindy Fay was his boat, custom built in 1979 by Knight and Carver for him. Her hull was of molded fiberglass of good design and heavy construction. Her equipment included a six-man inflatable life raft, three adult life preservers, three survival suits, a flare kit and an Emergency Position Indicating Radio Beacon (EPIRB). The safety equipment was in operating condition. The Cindy Fay was designed for gillnetting, a form of fishing in which nets are lowered and hauled in with their catch the next day. Her net was “one-thousand fathom” with a mesh of 92' by 20". Five months before the disaster the boat underwent a survey by American Marine Surveyors Inc. The boat was pronounced to be well-maintained and well-suited for her intended use.
Reber, with 21 year old Germano as his assistant, was bound for San Clemente Island. Its waters were attractive to fisher*980men. He had fished there since he had begun to fish. At the same time there was danger there. For forty years the Navy had conducted training exercises in bombardment, shelling portions of the island. The public was warned not to enter areas where the Navy was conducting bombardments. The whole island and its waters were not, however, off bounds to civilians. Unexploded bombs existed in the waters that had not been recovered or identified by warnings.
Boyd Reber had an agreement with his wife to call her every two or three days to assure her of his safety. On Sunday evening, February 23, 1985 he called her to say that all was going well and that he and Frank were even then cleaning fish they had hauled in. He was never heard from again.
The Wreck of the Cindy Fay. On March 2, 1986, 1,300 yards south of Mail Point on San Clemente Island, in a cove lined by cliffs, about 500 yards offshore entangled in kelp, Reber’s lifeless body was found. He was dressed in a sweatshirt and Levis and wore no life preserver. The time of his death was estimated by the coroner to be between February 23 and February 26 and was fixed by the Coast Guard as “most probably” 11:30 a.m., February 25, 1986. The lack of any communication with his wife since February 23 converged with this estimate. The cause of death was found by the coroner to be drowning. Reber also had a broken nose and injuries from blunt impacts on his head and torso.
Reber’s crewman, Germano, also died in the disaster. His body, clothed in jeans and T-shirt, was found among wreckage on the shore. The cause of his death was lacerations of the lung and liver due to blunt impacts upon the torso and head.
Portions of the Cindy Fay were found on shore, portions at sea. On the shore of the cove was the main diesel engine with its tailshift and propeller still attached; the anchor and anchor chain; parts of the hull; the EPIRB; the gillnet spool with part of the net still filled with fish, and various other items. In deep water, 300 to 400 yards off the coast, on the ocean floor under 35 feet of water, over two weeks after the disaster, divers found a variety of heavy metal objects that had been on the stern of the boat — the stern roller; the stern transom; the radar reflector mounted on the stern; the two support pins of a roller; and the pocket for a net part. Also found at sea was half of the Cindy Fay’s net, also containing fish and attached to the net anchor. One item from the forward part of the boat, the stabilizer, was also recovered in this area.
In the light of this evidence it was not disputed by the parties that the wreck of the Cindy Fay must have occurred in deep water, not on the rocks. No other explanation would account for the objects found in water far from shore. It was equally undisputed that once damaged the Cindy Fay would have been washed up on the rocks and pounded by surf. What was at issue was what caused the wreck.
The Plaintiffs’ Case. The plaintiffs contended that the Cindy Fay must have been wrecked either by a wave or by human activity; that they could exclude a wave; and that it was more likely than not that the human activity was negligent naval activity resulting in an underwater explosion.
It was not contested that in 35 feet of water where the wreck had occurred a wave that could have broken on the Cindy Fay would have had to have been at least 27 feet high. There was undisputed evidence that weather conditions in the relevant days of February would not have allowed a wave 27 feet high to form. The Navy’s weather logs indicated ripple waves, swells of 5 to 6 feet, winds of 10 to 16 knots. The Coast Guard reported the sea calm. Timothy Barnett, an oceanographer and expert on waves, testified that it would have been physically impossible for a 27 foot wave to form in such seas with such winds.
The exclusion of a wave as cause of the disaster was buttressed by the testimony of John Knight, the builder of the Cindy Fay. He testified that the boat could not have been demolished even by a 27 foot wave; that the boat was built to sustain *981pressures greater than those encountered in Southern California waters; and that he had never heard of a boat suffering from wave action alone the structural damage sustained by the Cindy Fay.
Final confirmation of this part of the plaintiffs’ case was the casual dress of Reber and Germano. They were not dressed for stormy weather. They had seen no need to take safety precautions.
With surf excluded because the disaster had occurred in deep water and with a large wave shown to have been at least unlikely, the plaintiffs developed the case that there had been an underwater explosion. There was uncontested evidence that the Navy had been conducting bombardments with targets on the island for forty years. That such bombardments would occasionally leave unexploded shells was a matter of common knowledge. That a fishing net catching an unexploded piece of ordnance could set off an explosion with dire consequences to the boat and skipper had been established in the Atlantic Ocean by the fate of the scalloper Snoopy, whose net caught a torpedo that exploded killing eight persons on the vessel. Petition of Trowler Snoopy, Inc., 268 F.Supp. 951 (D.Me.1967).
That the explosion had not been in the air was shown by the absence of burns on the recovered wreckage and the absence of shrapnel that an explosion in the air would have generated. The plaintiffs’ expert, Joseph Hrzina, was an engineer with extensive experience in calculating the force of explosions. His testimony was that only an underwater explosion would have produced the effects observed. In support of this testimony his detailed analysis was admitted as an exhibit in evidence. This analysis showed the forces at work if a 500 pound bomb had been detonated in 35 feet of water at a net angle of 45 degrees about 42 feet astern of the Cindy Fay.
The net angle was chosen as the angle at which a gillnet comes in when it is retrieved. The bomb size was that of a “small-bomb,” similar to one that some evidence indicated had been retrieved by a fisherman’s net off San Clemente Island. The location of the bomb aft of the stern was based on the fact that far greater damage was done the stern than the forward part of the boat. The analysis took the position that “the stern was literally blown off.”
Hrzina’s analysis of the physical pressures at work received corroboration from the testimony of two experienced gillnet fishermen, Tim Houshar and David Tib-betts. Houshar testified that when a wave broke over the stern of his boat, the gillnet would “just stretch” and not break. He added, “[T]hey hang in there, they’re built to do it.” Tibbetts testified that “on many occasions” a wave had broken over the stern of his boat and the net had just stretched “a little bit”. Neither fisherman had ever heard of a wave severing the net. In contrast, the net of the Cindy Fay had been severed and one part was on the shore, the other in the deep water.
Hrzina’s testimony and analysis strongly supported the view that an unexploded bomb caught in the net was the source of the injury. Nonetheless, a good part of the plaintiffs’ case was devoted to an alternative effort to show that live bombardment exercises on February 25, 1986 had led to overshooting the target and the consequent fall of a bomb in the water near the Cindy Fay. The plaintiff’s expert, Craig Ploss, supported this approach. The government effectively rebutted the evidence of overshooting; Hrzina’s analysis undermined Ploss; and the district court found this theory only remotely possible. It had a dramatic appeal but was not proved and is not worth stating in detail here. The Re-bers’ case turned on whether Hrzina’s testimony and analysis were credible.
The Defendant’s Case. The United States relied on a single witness, Robert D. Short. Short had worked for 26 years for the United States Navy. He was an engineer with special expertise in underwater explosions and in the protection of ships from the effects of such explosions. The district court found him to be an impressive witness.
Short testified at length to the consequences of an underwater explosion from a *982bomb of the character likely to have been in the area upon a boat like the Cindy Fay. He explained the extraordinarily short life above surface of the force of such an explosion and the vertical movement of the boat that the explosion would produce. He agreed that it was possible that a bomb had caused the wreck of the Cindy Fay as described by the plaintiffs’ expert, Hrzina. But he listed four “discrepancies” between the evidence and this possibility.
The first was the lack of injuries to the lower portions of the bodies of Reber and Germano. Having heard the coroner’s testimony that he did not look for injuries there, Short withdrew his reliance on this point.
The second was the degree of bending in the net pins. In Short’s opinion an underwater explosion of the kind described would have not have had the force to produce such bends but a 27 foot wave would have. On cross-examination, however, he admitted that his opinion was based on three assumptions, none of which were verified: (1) that the roller cage was on the pins; (2) that laboratory tests had correctly assessed the strain necessary to produce the observed bends; (3) that a 27 foot wave had struck the Cindy Fay. In fact, no one knew if the roller cage was on the pins, the testimony being that the cage was sometimes removed; (2) there was uncertainty as to how the laboratory tests had been performed and what they had ascertained; and (3) the possibility of a 27 foot wave was excluded by the evidence.
The third “discrepancy” was the bending in the propeller blades. Again Short doubted that the described explosion could have produced the bends. At the same time Short had no explanation of how the propeller damage could have occurred: “I cannot relate the damage to the propeller to any of the phenomena.” The propeller damage had happened some time, somehow. That it did not occur because of an underwater explosion had no probative force to show that such an explosion did not take place.
Finally, Short pointed to the distribution of the parts of the boat found in deep water. The maximum distance between the objects recovered was 530 feet. In Short’s opinion an underwater explosion would not have distributed the objects so widely. The vertical impact would have tossed objects in the air and they would have either fallen on the boat or near the boat; there would not have been a wide scattering. On cross-examination, however, Short conceded that objects that fell back on the damaged boat could have slid from its deck as it was borne by waves after the explosion. There was no certainty that the location in the ocean of the objects was fixed immediately after the disaster. Kelp covered a portion of the bottom: Reber’s body was caught in it. Other objects from the boat could have been similarly caught and moved with the movement of the kelp. In short, the pattern of dispersal was far from dispositive as to the cause of dispersal.
The District Court’s Findings of Fact. Ruling orally from the bench, the district court found that the parties agreed that the loss of the Cindy Fay was not caused by an explosion in the air either aboard the Cindy Fay or outside it. The court found the possibility of a shell from the naval bombardment exercises to be “extremely remote” and unproven. The court discussed the likelihood of an unexploded bomb being in the water near the Cindy Fay but made no finding. The court turned to consideration of the three experts who testified on underwater explosions. It found that the Rebers’ two experts “did not prove by a preponderance of the evidence that an underwater explosion caused the loss of the Cindy Fay,” while “Dr. Short convinced the court that an underwater explosion did not cause the loss of the Cindy Fay for the several reasons he expressed.” The district court listed these reasons as (1) the damage to the pins; (2) the lack of injury to the decedents’ lower extremities; and (3) the dispersal of the offshore objects.
The court found none of the experts helpful or convincing on the impact of waves. As to whether waves could have reached 27 feet, the court noted Barnett’s testimony that the maximum possible wave was 13.7 *983feet, but the court did not find this statement to be evidence of a fact. The court also noted, without making a finding, Paul Donahoe’s testimony that the surf had been “enormous” the weeks of the accident. The court declined to decide whether waves or surf had caused the disaster. The court concluded that the plaintiffs had only shown a possibility of causation by underwater explosion and, “as we all know, a possibility is not proof by a preponderance of the evidence.”
ANALYSIS
The Rebers offered uncontradicted evidence that the weather was such during the week of the wreck that a 27 foot wave would not have occurred, and they offered the uncontradicted evidence of the builder of the Cindy Fay that even a wave as high as 27 feet would not have caused the Cindy Fay to capsize or split into pieces. The district court failed to take into account this testimony. In so failing, the court failed in its duty to weigh all the evidence and was in clear error.
Without making a finding based upon it, the district court appeared to give credence to Donahoe’s testimony on the dangerousness of the surf. The district court could not rely on this testimony without making a finding. Even if the district court had made a finding, the court would have had to have explained why the dangerousness of the surf was relevant when uncontra-dicted evidence indicated that the Cindy Fay had been wrecked in deep water. The court’s half-use of Donahoe’s testimony was in clear error.
The court’s reliance on Short to find that an underwater explosion did not cause the wreck was equally erroneous. The court relied on evidence as to the bodies of Reber and Germano which Short expressly rejected as a basis for his opinion. The court relied on the dispersal of the objects without acknowledging that Short himself had only testified that the initial explosion would not have caused such dispersal while admitting that later events could have caused the spread. The court relied on Short’s testimony on the net pins without acknowledging that the opinion was based on unverified assumptions. In accepting Short’s testimony as excluding an underwater explosion, the court went far beyond Short’s own stated opinion which left the possibility of such an explosion open. The court’s reliance on Short’s reasons and the court’s mischaracterization of Short’s actual opinion were clear error.
The court’s oral findings express the conclusion that the plaintiffs’ experts did not establish that the cause was an underwater explosion. This conclusion is a conclusion of law. It cannot be sustained without findings of fact to support it. There are no such findings. The court engaged in a desultory review of the qualifications of the plaintiffs’ two experts on the explosions. While it is possible to see why the court did not have confidence in Ploss, the basis for its scepticism towards Hrzina is unexplained. Hrzina, an authority on aerodynamics, had offered an elaborate analysis of the forces in play after an explosion. Weighing of all the evidence required the court to consider this testimony. The court’s perfunctory conclusion, without support in findings of fact as to Hrzina’s expertise or as to his analysis, was error. The district court also failed to examine and evaluate the testimony of Houshar and of Tibbetts on the elastic characteristic of gillnets and so overlooked the significance of the severance of the gillnet of the Cindy Fay. Again, the court’s failure to weigh all the evidence was clear error. Finally, in clear error, the court failed to examine and evaluate the evidence that the dress of the two dead fishermen showed no expectation of stormy weather.
The court took the position that it did not have to determine the cause of the wreck. Abstractly stated, the court’s position is, of course, irrefutable. A plaintiff must prove his case. Here, causation was part of the plaintiffs’ case, which they had to prove. It was their job to show that an underwater explosion by an unexploded bomb was more likely than not.
The plaintiffs did show by clear and convincing evidence that the Cindy Fay was wrecked in deep water, not surf, and that the weather was such that a large wave capable of breaking over the boat and wrecking it was highly improbable. Any*984one who reviewed the evidence would have to conclude that it was more likely than not that the Cindy Fay had not been destroyed by a big wave.
What was left? On old maps of the world there used to be a terra incognito sometimes marked, “There be monsters here.” If we still believed in sea monsters, we could ascribe the wreck of the Cindy Fay to these beasts. But we don’t and we can’t.
Similarly we cannot suppose that Martians or pirates descended upon the boat, taking nothing but breaking it apart.
The possibility that the Cindy Fay was wrecked by an unexploded Navy bomb is the only possibility that remains. Once it is the only possibility that remains, it becomes more likely than not that it is what happened. We know the wreck occurred. We know that every other explanation does not hold up. The possibility converts itself into a probability.
Mindful as we must be of the deference owed the trier of fact, we are not here substituting one weighing of the evidence for another weighing of the evidence. The trial court made no findings of fact as to Knight, the builder’s, evidence, or as to Hrzina’s analysis, or as to the testimony of the two fishermen, Houshar and Tibbetts, or as to the dress of the deceased. I must conclude that it did not weigh this evidence. The trial court relied on a part of Short’s reasoning that Short had withdrawn and a part of Short’s reasoning that was based on unproven assumptions, and the court overstated Short’s conclusion, all in manifest error. The court failed to consider the evidence showing that the boat was outside the surf and the evidence showing that the wind and waves were quiet. The failure to consider this “extrinsic evidence” made erroneous the court’s refusal to recognize that only one explanation of the wreck was likely.
We do not and cannot satisfy our minds beyond a reasonable doubt as to what happened to the Cindy Fay. Such certainty is not sought in a civil trial, which deals only with probabilities. When a fact has been shown “to be more likely to be true than not,” it has been proved for purposes of the trial. Anderson v. City of Bessemer City, 470 U.S. at 580, 105 S.Ct. at 1514-15. Agnosticism is inappropriate. The defendant cannot take refuge in the only other possibility offered that evidence has shown to be highly improbable.
Death is always a mystery, even when it comes to the very old or the very sick. When it strikes down a man in his prime, it is especially mysterious. Why Boyd Reber, 38, a fisherman of long experience, should have been struck so quickly and so terribly will always remain a mystery. But the mystery of who lives and who dies is not the mystery of the loss of the Cindy Fay. Not beyond a shadow of a doubt but as considerably more likely than not, the cause of her loss may be determinable. The job must be done by the trial court. It is not the task of this court to weigh the evidence. But we must reverse the trier of fact when the evidence has not been weighed and when the evidence as a whole leaves us with the definite and firm impression that the conclusion of the trial court was in error. I dissent from the opinion of this court.