Dehydrating Process Co., as the owner of a tank barge and its cargo of liquid fish, brought this libel in admiralty against Boston Insurance Company and three other underwriters to recover on policies of marine insurance for damage to the barge and loss of its cargo. The only question with which we are concerned on this appeal is whether the admitted loss of the cargo and damage to the vessel were caused by 'perils of the sea,' all other questions having been disposed of by concessions or by stipulation.

The barge involved was a rectangular steel vessel 130 feet long, 9 feet 9 inches deep and 31 feet wide having raked ends and a cambered deck, i. e. its bottom sloped upward to the deck at bow and stern, and its deck was slightly higher at the center than at the sides. It had buoyancy compartments at each end which extended from the rail at the outer end toward the middle some 6 feet beyond the point were the sloping ends met the approximately flat bottom.1 The remainder of the hull was divided by transverse partitions into four cargo tanks of equal dimensions. On the stern over the after buoyancy tank there was a steel house weighing 8100 pounds and a steam pump and fittings weighing 5000 pounds so that empty the barge trimmed slightly by the stern.

Access to the buoyancy compartments and cargo tanks for the purpose of entry for inspection, cleaning, loading cargo, etc., was provided by two manholes in the steel deck over each one of the six compartments. The two manholes giving entry into each buoyancy compartment were directly opposite one another near the sides of the vessel and near the bulkhead or partition separating that compartment from the cargo tank next to it. The manholes giving access to each cargo tank were near the center line of the vessel one near each end of the tank and diagonally opposite one another. All of the manholes were provided with steel covers weighing about 50 pounds apiece which could be secured by studs.

The barge was towed to Gloucester, Massachusetts, on September 2, 1949, and tied up at a finger pier in the inner harbor about 200 yards from the libelant's plant. Soon thereafter employees of the libelant began to load the barge, and loading continued intermittently until about 11 P. M. on September 25, when the four tanks had been pumped nearly full of liquid fish. At that time the employee of the libelant in charge of loading shut off the pump which had been discharging liquid fish through a hose into tank number 4, removed the hose from the manhole through which he had been filling the tank and put the manhole cover on but did not fasten or 'dog' it down.

It was found below in part on the testimony of this employee and in part on the testimony of an expert that when loading stopped at about 11 P. M. the barge was afloat on an even keel with sufficient freeboard to remain afloat, and it was found on the testimony of the employee that she 'was right on an even keel' approximately an hour later when that witness again observed the vessel briefly in the light thrown by the headlights of his car. It is conceded that throughout the night of September 25-26 the weather was clear and the wind light.

Nevertheless, employees of the libelant about 5:00 o'clock in the morning of September 26 observed the barge to be submerged at the stern, and between 7:00 and 8:00 o'clock that morning it sank. The barge was promptly pumped out, floated, and hauled out for inspection when it was discovered that the hull was in as good condition as before the sinking except for trivial damage attributable to the stranding, consisting of minor dents and a few loose rivets which caused some insignificant leaking.

It was found below, and it is evident on the undisputed testimony as to the condition of the barge after it was raised, that the sea water which sank the barge could only have entered through the manholes in its deck. The issue is what caused the water to enter through the manholes.

Undoubtedly the libelant as the owner of the barge and its cargo has the burden of establishing by a balance of the probabilities that its loss was caused by a risk insured against, specifically in this case by a peril of the sea, for that is the only insured risk relied upon as the cause of the loss. But this does not mean that in every case an insured in the position of the libelant in the case at bar must point out the precise peril of the sea which caused the loss. It appears to be well settled in Great Britain and also in the United States, indeed the parties agree, that when a vessel sinks at a sheltered berth in calm weather without any obvious explanation, such as overloading or improper stowage of cargo, or collision, settling with the tide on a rock or other underwater obstruction, lightning, fire, explosion of cargo or other casualty, a presumption arises that the incursion of sea water which caused the sinking, and hence the loss, was due to the unseaworthiness of the vessel in some particular. But it appears to be equally well settled, and the parties also agree, that an insured can rebut the foregoing presumption by establishing that in fact the vessel concerned was seaworthy before the sinking, and was neither overloaded nor improperly loaded, and if he succeeds in doing so, the counter-presumption arises that the unexplained sinking and consequent loss was caused by some extraordinary, although unknown and unascertainable, peril of the sea. Anderson v. Morice, L.R. 10 C.P. 58, 67-68 (1874); Potter v. Suffolk Ins. Co., C.C.D.Mass.1835, 19 Fed.Cas. No. 11,339, p. 1186; The Jungshoved, 2 Cir., 1923, 290 F. 733, 735.

The libelant, relying upon the second presumption, contended at the trial, and introduced evidence expert and otherwise to prove, that its barge was seaworthy before it sank, and that it was neither overloaded, nor loaded so heavily by the stern as to submerge its afterdeck. Thus, it says, even though there is no evidence of inclement weather, exposed position, wash from other vessels passing during the night, or other known cause for water to wash over the deck, it must be presumed that the incursion of sea water into the hull through the manholes was caused by some unknown peril of the sea. The insurers, on the other hand, contended below and introduced evidence, also including that of an expert, which they say proves conclusively that the libelant's employees either overloaded the barge, or else loaded it so heavily at the stern that the afterdeck was submerged whereby water flowed under the manhole covers of the after buoyancy compartment, and then, as the barge settled by the stern, under the manhole covers into the cargo tanks beginning with tank No. 4 until enough water entered the hull to sink the vessel. Wherefore they say that the presumption relied upon by the libelant is rebutted and it must be presumed that the loss was caused by unseaworthiness due to overloading.

The Commissioner to whom the case was referred believed the libelant's evidence. He found, and the Court below adopted his findings, that the barge itself was seaworthy, and that it was not overloaded but had sufficient freeboard to remain afloat. Thus, in consequence of his finding that the barge before loading was fully seaworthy, and that there was no evidence to sustain the insurers' contention that it became unseaworthy in the course of loading by being either overloaded or improperly loaded so that its stability was endangered, he concluded that although the precise and specific cause of the sinking was unknown, the barge must have sunk alongside its pier because of 'an unknown peril of the sea.' Furthermore he found that 'what happened to the barge while moored in the harbor at Gloucester is within the purview of the definition of 'Perils of the sea' as that phrase is set forth in the policies issued by the respondents.' The court below agreed with the Commissioner's findings and conclusions and entered the final decree for the libelant from which this appeal is taken.

As already indicated, the insurers as appellants do not challenge the general principles of maritime law involved. Their attack is directed at the findings of fact made below. They say specifically and in detail that error was committed in finding that the barge was properly trimmed, had adequate freeboard, was loaded within its carrying capacity, and in view of its loading was seaworthy with respect to its manhole covers. Hence they assert that the incursion of sea water which caused the sinking must be attributed to unseaworthiness caused by so loading the barge as to submerge one or more of its manholes, which in their unsecured condition were unseaworthy in that they permitted the incursion of sea water, which is not a risk covered by the policies here in suit.

'In admiralty an appeal to the Court of Appeals is deemed to be a trial de novo.' Brooklyn Eastern District Terminal v. United States, 1932, 287 U.S. 170, 176, 53 S. Ct. 103, 105, 77 L. Ed. 240, and cases cited. But it does not follow that in admiralty we put ourselves in the shoes of the trial court and make our own independent findings of fact on the evidence in the record. The scope in admiralty of appellate review of findings of fact has been variously expressed. See Petterson Lighterage & Towing Corp. v. New York Central R.R., 2 Cir., 1942, 126 F.2d 992, and the cases cited in the appendix thereto. In this circuit in the years since the promulgation of Admiralty Rule 46 1/2 requiring the trial court to make findings of fact, it has been said that such findings 'will not be disturbed unless clearly wrong', The Berwindglen, 1 Cir., 1937, 88 F.2d 125, 126, or unless 'contrary to the preponderance of evidence.' The Josephine & Mary, 1 Cir., 1941, 120 F.2d 459, 463 citing The Parthian, C.C.D.Mass.1891, 48 F. 564.

But analysis of the above phraseology to determine its precise shade of meaning is not necessary, for this court's latest pronouncement on the subject is: 'We are not to set aside the judgment of the trial court based on his factual findings as to the cause of the accident, where, as here, there was a conflict in evidence, unless such findings are clearly contrary to the preponderance of the evidence or, what amounts to the same thing, unless such findings are 'clearly erroneous', to borrow the expression in Rule 52(a) of the Federal Rules of Civil Procedure'. Gibbons v. United States, 1 Cir., 1950, 186 F.2d 488, 489. Thus, notwithstanding the repeated statements that an appeal in admiralty is a trial de novo, we are guided in admiralty appeals so far as our review of findings of fact is concerned by the same principle which guides us in appeals in cases governed by the Federal Rules of Civil Procedure, 28 U.S.C.A.

The basic question for us is whether the finding below that the barge was not rendered unseaworthy in view of the condition of its manhole covers by being either overloaded or improperly loaded was 'clearly erroneous.' The reason for this is that any unseaworthiness of her manhole covers could not, under the circumstances of no precipitation and only light wind, have been operative to permit the incursion of sea water which caused the sinking unless the barge was so low in the water that her deck, or at least her afterdeck, was submerged, for the appellants concede, apparently because of the position of the manholes in the deck and its camber, that water could not on a still night lap up over the deck and trickle into the manholes under the covers in sufficient quantity to sink the barge in a matter of hours. We turn, therefore, to the evidence with respect to the amount and disposition of the cargo loading, and the position of the barge in the water, when loading ceased at about eleven o'clock at night on September 25, 1949.

The amount of liquid cargo loaded, and its disposition in the four tanks into which the barge was divided, was calculated from records kept by the libelant. These records consisted of loading orders given to its employees, records of the measurements of the ullage of its storage tank on shore from which it loaded the barge taken before and after specific loading operations, and records of the measurements of the ullages of the various tanks in the barge taken at different times during the loading period. From this data, the specific gravity of the liquid cargo, and the dimensions and weight of the barge itself, the expert in naval architecture called as a witness by the libelant testified that the gross amount of cargo loaded was within the carrying capacity of the barge, and that its disposition in the barge gave a trim by the stern with a calculated freeboard at the after corners of about an inch, including the thickness of the deck, and of about 5 inches, also including the thickness of the deck, at the center of the stern at the apex of the deck camber. Furthermore, this witness testified that if the barge had been overloaded to the extent of submerging its deck, water would have flowed under the unsecured manhole covers so fast that the barge would have sunk in a matter of minutes, whereas in fact it sank in a matter of hours.

In addition to this expert testimony the employee who loaded the barge the night before it sank said that when he finished loading he had no difficulty in stepping off the barge onto the pier, and that when he looked at the barge an hour or so later it was afloat in the same condition in which he had left it.

The respondents in their turn called an expert witness who supported their theory of the loss by testifying that on the basis of the libelant's loading data and his own calculations the barge was loaded so heavily by the stern that its afterdeck must have been submerged, and in that case water would have flowed into the hull under the unsecured manhole covers at an appropriate rate to cause the barge to sink in the time it admittedly took to do so. Thus there was a square conflict in the testimony.

In view of the conflicting evidence of the experts we cannot say that the findings made below as to freeboard and trim are 'clearly erroneous.' It is true that it is hard to see why the barge sank if its afterdeck was not submerged, or at least awash, when loading stopped. But on the other hand, it is equally hard to see why, if the afterdeck were submerged when loading stopped, water did not flow under the unsecured manhole covers fast enough to sink the barge, as the libelant's expert said it would, in substantially less than the eight or nine hours which elapsed between about 11 P. M. on September 25, and 7 or 8 A. M. the next morning. Furthermore, to find that the afterdeck was submerged we would have to disregard the testimony of the libelant's employee who loaded the barge whom we have not seen and did not hear testify.

The case is certainly close, and there are discrepancies and inaccuracies in the libelant's loading data used in the calculations of the opposing experts. But nevertheless on the entire evidence we cannot say that we are 'left with the definite and firm conviction that a mistake has been committed.' United States v. United States Gypsum Co., 1948, 333 U.S. 364, 395, 68 S. Ct. 525, 542, 92 L. Ed. 746. We think there was 'a choice between two permissible views', United States v. Yellow Cab Co., 1949, 338 U.S. 338, 342, 70 S. Ct. 177, 179, 94 L. Ed. 150, with respect to the findings of freeboard and trim made by the Commissioner and adopted by the court below. Thus, although the cause of the loss is a mystery, the libelant on the findings has rebutted the presumption that the cause was unseaworthiness due to overloading or improper loading, and hence it is entitled to recover under the presumption of a loss caused by some unknown and unascertainable peril of the sea.

The decree of the District Court is affirmed; the appellee recovers costs on appeal.