OAKES, Circuit Judge:

A barge owner (Seaboard) and a tug owner (Moran) are before the court on cross appeals, each blaming the other for the loss of the lives of two bargemen, whose survivors have already been compensated by settlement. The parties here reserved their rights against each other under the settlement. The court below held, 320 F. Supp. 229, in this exceptionally well-briefed case, that neither the negligence of the tug owner, Moran, the unseaworthiness of the barge, the negligence of Seaboard, nor the negligence of the bargemen had been established by a preponderance of the evidence. Thus, the parties were left as the court had found them, and their respective proceedings for exoneration from or limitation of liability were dismissed.

The deaths of Arthur Naess and Charles Thompson occurred in November 1966 when they were lost from the barge Oil Transfer 32 (O.T. 32) during a wild storm on Lake Michigan while the barge was under tow by the tug Margot Moran (Margot). Captain Naess's body was found, clad in foul weather gear, on November 20, 1966, on the east shore of the lake. His death was attributed to drowning. Mate Thompson's body was never recovered. When the tug's mate boarded the O.T. 32 at 0600 hours on November 4, after the storm had subsided, he found the aft deck of the barge awash about one foot at its lowest point just forward of the cabin, and the cabin filled with water to the sill, though its horizontal door was tightly closed and its vertical door ajar only about one inch. He also found both lifeboat and liferaft missing, the deck in a shambles with a broken port boom, cargo hoses over the side, cowl ventilators missing and pump and engine rooms full. The hull was later found sound by a diver. Moran and Seaboard, below and here, each attempted to reconstruct the events in a light most favorable to itself. We find them both at fault.

The O.T. 32 was a typical inland waterway steel barge, 1278 gross and net tons, 229.4 feet long by 42.9 feet wide and 14.5 feet deep. She had no propulsion power, but carried a crew of two certified tankermen. On October 30, 1966, she loaded a cargo of benzoil at Lemont, Illinois, bound for Bay City, Michigan. She was pushed from Lemont to Chicago by the Margot, a tug of 141 gross and 96 net tons, 84.8 feet long with a 24-foot beam and 9.6-foot draft, powered by 1280 HP twin diesels. The O.T. 32's vocaline radio, which would have permitted communication with the Margot, had been out of order for several months, and its other means of communication, a sheet or a towel, floodlight, foghorn and flares, were not effective in the severe snowstorm that was brewing. The O.T. 32's lifeboat had been damaged in September 1966 (a Seaboard employee thought it, and the trial court found it, unseaworthy), and a canister containing an inflatable liferaft had been placed on board, but the canister was simply put in the useless lifeboat rather than on a proper cradle near the side of the barge. To compound the prescription for the disaster which ensued, the barge was loaded to a draft of 11 feet 1 inch fore and aft, her Intermediate Marks giving a freeboard of 3 feet 5 inches. This was fine for the day she was loaded, October 30, but the season for Intermediate Marks expired at 2400 hours the following day, October 31, and for the winter season following her maximum permissible drafts were 2¾ inches less than that permitted by the Intermediate Marks. Thus she was overloaded for the tow.

The Margot, like O.T. 32, was not fully equipped. Despite her direction finders and weather receiving radio equipment, the Margot's barometer was not calibrated for the Great Lakes; it was calibrated for the New York area and consequently the Margot's captain never bothered to check it.

After an uneventful trip from Lemont, the flotilla remained at Chicago from the afternoon of October 30 to 0830 hours November 1, because gale warnings were up and the October 31 forecast was for 34 to 40 knot winds. With gale warnings up at the Coast Guard dock and still being broadcast on October 31 and November 1, the tug captain nevertheless left, telling the bargemen the trip might be "a little sloppy." About fifteen minutes after departure, when the lock between the Coast Guard dock and the harbor was cleared, the 1200-foot long hawser, with bridle and bitts, was placed, so that the barge was approximately 1300 feet or a quarter-mile astern of the tug. Because of the weather, instead of heading straight to Point Betsie (sometimes spelled Betsy in the record) Light across the lake, as would ordinarily have been done, after clearing the breakwater the flotilla went up the west shore of Lake Michigan. At 10:10 AM on November 1, the wind was northwest to north, 25-35 knots, the seas were heavy according to the Margot's log, and the forecast was for winds increasing to 30-38 knots for the rest of November 1 and for November 2, with some snow flurries.

The Margot with the O.T. 32 in tow continued up the west shore of the lake, slowing down at Wilmette from a full 650 RPM's to 575. The gods of weather gave the flotilla a little respite in the late afternoon: the gale warnings were reduced to small craft warnings and the northerly winds prediction was reduced to 24-32 knots. But at midnight the predicted snow flurries turned into a heavy snowstorm off Milwaukee, totally obstructing the Margot's radar as well as a visual fix on Wisconsin's largest city. The Margot slowed to 475 RPM's, but proceeded onward. During this time and for the remainder of the storm there was no radio contact with any other vessel on the lake. For another 9½ hours on Wednesday, November 2, the flotilla continued up the west side of the lake in heavy seas that were building up gradually from 5 feet and running from north to north-northeast, with intermittent heavy snow. Weather forecasts at 4:10 AM and 10:10 AM on November 2 carried small craft warnings and predicted snow flurries. Not insignificantly, the 10:10 AM November 2 forecast predicted higher winds (22-32 knots) than had the 4:10 AM forecast (20-27 knots).

Off Sheboygan at 9:30 AM on November 2, the venturesome Margot made a sharp course change, heading east-northeast across the lake toward Big Sable Bull Point on the east shore. The sea was now running at 7 feet, the flotilla was 11 miles off Sheboygan and about 50 miles from Big Sable.

That afternoon at 1:30 a Special Lakes Warning was broadcast changing small craft warnings back to gale warnings for that portion of the lake south of a line between Manitowoc on the western shore and Ludington on the eastern shore. The Margot was below the line. On toward Big Sable across the lake she pushed, however, reducing speed to 425 RPM's, at midnight, November 2, with the seas increasing, snow falling and no communication, visual or electronic, with the barge.

The 4:10 PM Wednesday, November 2, weather and the 10:10 PM forecasts had told in stark terms of the doom awaiting Naess and Thompson:

4.10 p.m. * * * Gale warnings in effect Manitowoc Ludington southward and change to gale warnings 7 p.m. EST Wednesday north of Manitowoc Ludington. Northerly winds 35 to 45 knots tonight becoming northwesterly 35 to 45 knots Thursday. Snow.

10.10 p.m. * * * Gale warnings in effect. Northerly winds 40 to 50 knots tonight becoming northwesterly Thursday morning and diminishing Thursday afternoon and night. Snow or snow flurries.

When asked to explain why he continued on course toward Big Sable despite the 1:30 PM special warning of November 2, when the flotilla could have turned back and made haven at Sheboygan (as the trial court found), the tug captain said, "We kept going in the same direction to get lee on the east shore." By 0600 hours on November 3, 10 miles off Big Sable Point, course was changed again and the flotilla ran north or northwest, into the wind, away from shore and head-on into one of the worst storms on the lake for years: "the worst one I have had," said a Moran deckhand with experience towing oil barges on the Great Lakes since 1949. The gyro on the tug "kicked out" and then the rack on the Margot's stern came loose in the rough weather. The wind was gusting 60-65 knots, steady 50, by 1500 hours on November 3, and the seas were running at least 15 feet. During all the night of November 2 and morning of November 3 there had been no running lights visible on the barge but a white light showed through a porthole in the cabin, seen during lulls in the storm. At 1830 hours on November 3 the storm cleared, and the Margot could see Point Betsie and the other fixes on the lake's east shore. It was 0645 hours on Friday morning the 4th, however, when the Margot's master noted that even the barge's cabin light wasn't lit and the barge was down, that is, its bow was high. The captain put full ahead to near shore until 0800 hours, when the hawser was pulled, and the tug's mate made his inspection trip, which revealed that Naess and Thompson were missing and the deck of the barge in the shambles partially described above. The barge's running light stanchions were bent over, part of the lifeboat cradle missing, the starboard pump room hatch dogs loose, three of the four stern ventilators gone, the after half of the engine room hatch open, cargo hoses tangled up and flopping over, engine room, pump room and cabin all flooded.

In Cleary v. United States Lines Co., 411 F.2d 1009 (2d Cir. 1969), this court recently reaffirmed its long-held view that the "clearly erroneous" test of Rule 52(a), Fed.R.Civ.P., does not apply to findings of negligence (or lack of it) although the conclusion of the district judge after a careful examination of the facts is entitled to great weight. We must therefore examine the trial court's ultimate determination in the light of its findings and the evidence. Esso Standard Oil Co. v. S.S. Gasbras Sul, 387 F.2d 573 (2d Cir.), cert. denied, 391 U.S. 914, 88 S.Ct. 1808, 20 L.Ed.2d 653 (1968).

Doing so, we deny exoneration to Seaboard, the barge owner. In three material respects the O.T. 32 was unseaworthy and, because the unseaworthiness was known or should have been known to Seaboard, in each respect Seaboard was also negligent. Its radio was out of order, with the antenna rusted away, and had been for the whole season. Since September 1966 its lifeboat had had a dent in the bottom one foot by two feet, and three to six inches deep, which Ode Odegaard, the barge's former captain — replaced for this voyage by Captain Naess — thought would make the lifeboat leak; and the inflatable liferaft installed in its place was improperly stowed. Beyond this, Captain Odegaard had supervised the loading of the O.T. 32 to her Intermediate Marks, and it was evident that the flotilla could never have made Bay City by 2400 hours October 31, the time the season for Intermediate Marks expired. Cf. Ionian Steamship Co. of Athens v. United Distillers of America, Inc., 236 F.2d 78, 81 (5th Cir. 1956); The Vestris, 60 F.2d 273 (S.D.N.Y. 1932).

Seaboard argues, however, that there was no evidence that any of the barge's defects or Seaboard's negligent acts probably caused the deaths, and that therefore there is a failure of proof. We need go no further here than to say, however, that given the statutory violations alone (the liferaft stowage and the overloading), Seaboard is not entitled to exoneration. The Pennsylvania, 86 U.S. (19 Wall.) 125, 136, 22 L.Ed. 148 (1873), makes it clear that the burden is on a ship in violation of a safety statute — in this case the barge — to show "not merely that her fault might not have been one of the causes, or that it probably was not, but that it could not have been." Seaboard is wrong in its contention that admiralty applies this rule only in collision cases. Kernan v. American Dredging Co., 355 U.S. 426, 78 S.Ct. 394, 2 L.Ed.2d 382 (1958) (fire on a tug caused by open-flame kerosene lamp carried on scow in statutory violation); The Denali, 112 F.2d 952 (9th Cir. 1940) (stranding case). It is true that a panel of this court has recently held the rule of The Pennsylvania inapplicable in the case of a seaman's death from a heart attack after a violation of statutory working hour limitations. Wilkins v. American Export Isbrandtsen Lines, Inc., 446 F.2d 480 (2d Cir. June 21, 1971). But another panel has specifically reached the same result as we do here in a Load Line Act [46 U.S.C. § 85(b), (c)] violation case. Petition of Long, 439 F.2d 109 (2d Cir. 1971). See also The New York Marine No. 10, 109 F.2d 564, 566 (2d Cir. 1940) (collision case).

Moran likewise was not entitled to exoneration. The Margot's master knew that he did not have a properly calibrated barometer, which might have given him quicker warning of the erratic track of the storm than he received from the periodic Great Lakes radio bulletins. He knew he lacked radio communication with the barge so that if the O.T. 32 were in trouble the only way he would know was by a sheet or a towel or a floodlight signal which "the biggest part of the time you would not have been able to see" on this fatal November voyage. He had admittedly been informed of the damage to the barge's lifeboat. The barge bore Plimsoll marks on her side, and it may be inferred that the Margot's master knew how deeply the barge was loaded, for the Margot pushed the O.T. 32, it will be recalled, until both vessels cleared the Chicago breakwater. Knowing these defects in the tug and tow and in their intercommunication system, the Margot's master nevertheless set out after gale warnings had been issued and continued when he might readily have made safe haven at Sheboygan, even after the upward change in wind forecast at 10:10 AM on November 2, and the Special Lakes Warning at 1:30 PM for south of the Manitowoc-Ludington line. What he did after he was practically across the lake, heading north into the storm, may have been in reaction to an emergency, if not in extremis, cf. The Oregon, 158 U.S. 186, 204, 15 S.Ct. 804, 39 L.Ed. 943 (1895), and we do not fault him on that. But, with the flotilla's conditions which he knew, to attempt to cross Lake Michigan in early November and to do so in the face of increasingly bad weather reports we think negligent. To be sure, as Moran points out, the Margot stayed close to the westerly shore until Sheboygan because her master was concerned about winds and sea from the northwest, as then predicted. But when he set out for the middle of the lake he had an increasing swell from the north-northeast, he had had no communication with any vessels on the lake, he had intermittent heavy snow and northeast to north winds predicted by both the 4:10 AM and the 10:10 AM broadcasts. Under any theory of the case — and the parties have advanced several — as to how the men were actually lost, we think the Margot should have made for safe haven, rather than risk the northeaster, only to be exposed inevitably to the wind shift to northwest once the Margot made Big Sable and the eastern shore. This was not a voyage of exploration seeking a route around Cape Horn or through the Northwest Passage, where a venturesome admiral might risk life and vessel for a taste of glory; it was a tow across a big, tough lake to deliver a barge load of benzoil to Bay City, Michigan. If foreseeability of harm is the test of negligent conduct, setting out for Big Sable was negligent. Petition of Kinsman Transit, 338 F.2d 708, 723-724 (2d Cir. 1964), cert. denied, Continental Grain Co. v. City of Buffalo, 380 U.S. 944, 85 S.Ct. 1026, 13 L.Ed.2d 963 (1965).

No matter what actually happened on that barge in the storm, we think it must be inferred that the negligence of the Margot in crossing the lake easterly was a contributing cause of the seamen's deaths, and coincidentally to the barge's damage. We do not quarrel with the district judge's finding that "Plainly, the cause of the unfortunate loss of the two bargemen was the extremely violent winds and seas which were encountered during the final northwesterly leg of the voyage." The element of causation may, of course, be proven by circumstantial evidence. Johnson v. Griffiths S.S. Co., 150 F.2d 224, 226 (9th Cir. 1945). See The Black Diamond, 273 F. 811, 812 (2d Cir. 1921); and 2 Harper and James, Torts § 20.2 (1956). But without the negligence of the tug in heading easterly with a mute and helpless tow into the approaching whole gale, this tragedy would not have occurred.

What legal consequences flow from the ultimate findings, then, that both Moran and Seaboard were at fault and that such fault legally caused the deaths of Naess and Thompson (and the damage to the barge)? Seaboard argues that it is entitled to indemnity from Moran, because even though the tow may have been unseaworthy, the tug brought that unseaworthy condition into play by the negligent manner in which it performed its towing operation, relying upon Dunbar v. Henry DuBois' Sons Co., 275 F.2d 304 (2d Cir. 1960). Moran argues that it is entitled to indemnity from Seaboard because it is "Hornbook law" that the tow impliedly warrants its seaworthiness to the tug, relying upon, e.g., South, Inc. v. Moran Towing and Transportation Co., 360 F.2d 1002 (2d Cir. 1966). Here, however, each was at fault, and the deaths and tow damage were the result of mutual wrongdoing, each contributing to the end result, where the acts of one alone might not have been sufficient to cause any of the damage. We believe that contribution lies, despite Halcyon Lines v. Haenn Ship Ceiling Refitting Corp., 342 U.S. 282, 284, 72 S.Ct. 277, 96 L.Ed. 318 (1952). Halcyon held that there was no right of contribution between a shipowner and a shoreside contractor who were joint tort-feasors in a case involving injuries to an employee of the contractor while engaged in repair work on the ship. The Court said:

Where two vessels collide due to the fault of both, it is established admiralty doctrine that the mutual wrongdoers shall share equally the damages sustained by each, as well as personal injury and property damage inflicted on innocent third parties. This maritime rule is of ancient origin and has been applied in many cases, but this Court has never expressly applied it to non-collision cases. 342 U.S. at 284, 72 S.Ct. at 279 (footnotes omitted).

The Court expressly referred in its footnote 5 at 284 to American Stevedores, Inc. v. Porello, 330 U.S. 446, 67 S.Ct. 847, 91 L.Ed. 1011 (1947), where the Court had seemed to approve contribution in non-collision cases, but referred to the holding there as "only incidental." In Halcyon the Court did not mention another non-collision case, White Oak Transportation Co. v. Boston, Cape Cod New York Canal Co., 258 U.S. 341, 42 S.Ct. 338, 66 L.Ed. 649 (1922). There a vessel sank in a canal due to the negligence of both shipowner and canal company. In an opinion by Mr. Justice Holmes, the Court found proper an equal division of all damages arising from the loss of the vessel and its cargo and injury to the canal. But White Oak did not involve, as, significantly, did Halcyon, an area of the law — harbor workers' personal injuries — in which Congress had adopted legislation (the Longshoremen's and Harbor Workers' Compensation Act, 33 U.S.C. § 901-950).

In so holding we are by no means sailing uncharted waters. That busy admiralty court, the Fifth Circuit Court of Appeals, has only recently held Halcyon inapplicable where a barge capsized as a result of its own unseaworthiness and the jointly concurring negligence of tug and tow, causing the death of a deckhand from the tug who had boarded the barge to retrieve tug equipment. Horton Horton, Inc. v. T/S J.E. Dyer, 428 F.2d 1131 (5th Cir. 1970), cert. denied, 400 U.S. 993, 91 S.Ct. 461, 27 L.Ed.2d 441 (1971). In a carefully considered opinion by Judge Wisdom, the Fifth Circuit has followed its Horton case in Watz v. Zapata Off-Shore Co., 431 F.2d 100, 120 (5th Cir. 1970). See also G. Staring, Contribution and Division of Damages in Admiralty and Maritime Cases, 45 Calif.L.Rev. 304 (1957), for an extensive discussion of the history of contribution in admiralty law.

Here, as in Horton, had Seaboard and Moran not settled with the personal representatives of Captain Naess and Mr. Thompson, both would have been responsible in damages, unlike the situation in Halcyon where the injured employee was restricted by the limited tort liability provisions of the Harbor Workers' Act and thus precluded from suing his employer once he had gone against the shipowner. Cf. Luigi Serra, Inc. v. S.S. Francesco C, 379 F.2d 540 (2d Cir. 1967). We believe Halcyon to be inapplicable in cases where the joint tort-feasor against whom contribution is sought is not immune from tort liability by statute.

We are further of the view that contribution in a tug-tow case involving fault of both produces the most equitable result. See, e.g., Gregory, Contribution Among Joint Tortfeasors: A Defense, 54 Harv.L.Rev. 1170 (1941). There is also a certain "utility and economy" in spreading the loss. 1 Harper and James, Torts § 10.2 (1956).

Insofar as White Oak, supra, has never been overruled, it controls here on the question of contribution and the extent of it.

Reversed and remanded for an equal division of the damages paid in death settlements and the damage to the O.T. 32.