A collision between two ships on the Willamette River resulted in the filing of a libel and cross-libel to fix the responsibility for the accident. The district court found both vessels negligent and decreed an equal division of damages. Both appeal, each protesting its own innocence of fault.

The Barry K was a stern wheel river boat owned by the Western Transportation Company. She was 190 feet long, extended about 35 feet above the water line, drew less than 3 feet of water. In the early morning of February 1, 1941, a clear night with little wind, she was proceeding at about 6 miles an hour up the eastern or left side of the Willamette River. Two barges, each drawing as loaded 4 to 5 feet, were lashed one to her bow and one alongside. At the same time the Pennsylvanian, owned by the American-Hawaiian Steamship Company, was making her way downstream at about 8 miles an hour. She was an ocean-going steamer 430 feet long overall and drew as loaded about -9 feet.

The vessels sighted each other, when approximately a' mile apart, in the vicinity of the Post Office ¿Bar, where the river makes a long easy curve. At that point the river is 1200 feet wide with a 30-foot dredged ship-canal 800 feet wide and a 35-foot channel 700 feet wide. The Barry K gave a signal of two whistle blasts, signifying a desire for a starboard-to-starboard passage. Those on the bridge of the Pennsylvanian heard only a single blast, the port-to-port passage signal, agreed to the same by answering with one blast, and swung to the starboard, or to the eastern side of the river. The ships were then about three-quarters of a mile from each other.

The Barry K was showing her red side light to the Pennsylvanian, but after the exchange of signals she showed a green light. Those navigating the Pennsylvanian attributed the fact to a temporary sheer and thought that the Barry K would right herself immediately. They asserted, although the testimony was conflicting, that as a precaution the engines were put to half speed and stop almost immediately after the appearance of the green light. The Barry K blew the danger signal, four or more rapid blasts. Thereupon, according to the Pennsylvanian’s officers, the ship’s engines were put to full speed astern, the three-blast whistle indicating the same given, and the helm ordered hard right.

In the meantime, after her two-blast signal, the Barry K was bearing to port, or easterly, in an effort to reach the (shoal water on the eastern side of the river where there was sufficient depth only for shallow draught vessels. When the Pennsylvanian answered his two-blast signal with one blast, the pilot of the Barry K twice blew the danger signal alternated with the two-blast whistle for a starboard passage. He thought the other ship would understand his two-whistle blast and would pull back into her regular course. Consequently, he continued toward the eastern shore at full speed. After his second danger signal, he stopped the engines, and reversed them when he received the Pennsylvanian’s three-blast signal. The ships collided about 200 feet from the eastern bank of the river and at the eastern edge of the 30-foot channel.

The district court found that both vessels negligently contributed to the collision in failing to stop their engines when the Barry K blew the first danger signal.

The Pennsylvanian claims that the sole cause of the accident was the Barry K’s disobedience of Article 25 of the inland water rules, 33 U.S.C.A. § 210 : “In narrow channels [the Willamette River channel must be regarded as a narrow one] every steam vessel [both the vessels concerned are steam propelled] shall, when it is safe and practicable, keep to that side of the fairway or mid-channel which lies on the starboard side of such vessel.” The *480 Barry K, admitting that it did not “keep to the right” in 'accordance with the rule, contends that the exception to the rule governed the situation as it was unsafe and impracticable for river boats and their tows to proceed up the western, or.right, side of the river. '

The evidence establishes that it was a common practice for river boats to use the eastern side of the river. However, proof of custom or convenience alone will not justify an infringement of the narrow channel rule. The Richard J. Barnes, 2 Cir., 111 F.2d 294 ; The Priscilla, 1 Cir., 55 F.2d 32 ; Lehigh Coal & Nav. Co. v. Compagnie Generale Transatlantique, 2 Cir., 12 F.2d 337 ; Occidental & O. S. S. Co. v, Smith, 9 Cir., 74 F. 261 . The standard set by Article 25 to govern departures from the general rule is one of safety and practicability. The Felix Taussig, 9 Cir., 5 F.2d 612 ; The Transfer No. 21, 2 Cir., 248 F. 459 .

The use of the eastern part of the channel by river boats on their upstream voyages seems to have grown up for the following reasons: Light draught boats with high superstructures, such as tugs and barges, are affected by cross winds. The prevailing winds in the area with which we are concerned are westerly in direction, and therefore vessels rounding the outside of the long curve in which the collision occurred, are subject to sideslipping toward the western bank of the river. A sand bank topped with trees on the easterly side shields a vessel hugging this side from the full influence of the wind, a situation of greater importance because a dike constructed of piling of irregular height stretches along the westerly side of the river and is likely to puncture the bottoms of river boats thrown against it. River boats in common use would have to keep their bows at least 200 feet. from this dike safely to avoid being carried upon it by the prevailing wind. Displacement swells from ocean-going ships and kick water from their propellers tend to move river boats toward the dike, when a port-to-port passage is being made. If a vessel is driven upon the sand bank of the easterly river side, serious damage is not so likely to occur as when driven upon the piling of the westerly side. In addition to these facts the current of the river on the west side is swifter than it is on the east side. Between the eastern bank and the eastern edge of the channel there are several hundred feet of shallow water in which river boats can travel entirely off the dredged-channel course used by ocean-going ships. The deep draught of ocean-going ships makes them far less susceptible to sideslippage than the shallow-bottomed river boats and barges. Tugs with tows generally must stop when they meet other ships to prevent breakage of their lines; while stopped, they are particularly subject to the force of the current, wind, swells, and kick water. These facts were the basis of river pilots’ testimony to the effect that there is a long-standing custom to take river boats up the easterly side of the river generally and particularly if any large ship is moving in the vicinity. It is argued that these facts entirely justify the Barry K’s position and its intention to pass the Pennsylvanian starboard-to-starboard.

Both vessels involved herein were under the pilotage of men long experienced in the use of the Willamette River channel. Both would be well versed in the safe and practical navigation of the channel. It seems to us, and we hold, that the evidence establishes the fact that it was normally unsafe for the upstream river boat traffic to use the western portion of the channel and therefore impractical, while using the easterly course to pass down stream oceangoing vessels using the dredged channel, to “keep to that side of the fairway or mid-channel which lies on the starboard [right] side of such vessel.” It follows that the Barry K was not negligent in navigating the easterly portion of the channel. It also follows that the Pennsylvanian was bound to direct it» course so as to accommodate the upstream easterly course of river boats and to pass them starboard-to-starboard.

There appears to be considerable merit in the suggestion made during the oral argument that a long standing dispute between two groups of pilots may have been the underlying cause of the collision. Nothing could be more obvious than that one rule of a positive and unequivocal nature should be provided by law for the passing of ships in the river and that compliance with such law should be strictly required of all pilots.

While we find the Barry K free from fault in proceeding on the easterly side of the channel and in choosing to pass the Pennsylvanian on the left, we find her at fault in another respect. Her pilot was aware of some difficulty when he sounded *481 the danger signal after liis starboard passage signal had been answered by a port passage whistle from the Pennsylvanian. It is the duty of a navigator of a ship, when he realizes that his signals have been misunderstood, to stop and reverse the engines until an agreement is reached, and he is negligent if he disregards this duty. The Johnson, 9 Wall. 146 , 19 L.Ed. 610 ; The Brandon, 4 Cir., 273 F. 176 ; The John D. Rockefeller, 4 Cir., 272 F. 67 . The reason for this rule is apparent in this cáse as there was reasonable ground for the pilot of each vessel to anticipate collision. Attempting to avoid collision by running clear of the dredged channel through which the Pennsylvanian would have to proceed, the Barry K proceeded full speed ahead toward the shoal waters on the eastern side of the river after her danger signal had been blown. That the pilot thought the Pennsylvanian would soon understand the starboard passage whistle is immaterial, for he had no reason under the circumstances to assume that the other ship would accede to his request.

The Pennsylvanian was also at fault, for she too failed to stop and reverse her engines when she heard the danger signal from the Barry K. The evidence with respect to the procedure followed aboard the Pennsylvanian just before the accident is conflicting. The captain and third officer, who were on the bridge during the crucial period, asserted that her engines were stopped when the Barry K first showed her green light and before the danger signal was blown, and that after the danger signal her engines were immediately reversed. The same facts were included in a report made on the day of the collision by the captain and the pilot to the United States Steamboat Inspectors. The testimony of the pilot at the trial was confused. He declared, first, that his ship slowed when the Barry K showed her red light, stopped at the first danger signal, and reversed at the second danger signal. Next, he said that he slowed down at the first danger signal, stopped later, and reversed when the vessels were 800 feet apart. On cross-examination he reiterated the fact that he merely slowed down at the first group of blasts. Finally, he changed his story to agree with that of the captain and third officer. The district judge resolved the conflicting testimony and found that the engines were not stopped immediately after the first danger signal. Our consideration of the record compels us to accept his finding in accordance with the principle announced in Thomas v. Pacific SS Lines, Ltd., 9 Cir., 84 F.2d 506 , 507: “ * * * the correct rule is that, where a substantial part of the evidence was heard in open court where the trial judge had opportunity to see the witnesses, hear their testimony, and observe their demeanor while testifying, with ample opportunity to pass upon their credibility and accuracy, such findings [of the trial court] are accompanied with a rebuttable presumption of correctness * *

As to the negligence of the Barry K, it cannot be ignored on the ground that the accident would have occurred even had she stopped and reversed at the first hint of danger. It has not been proved beyond a reasonable doubt that the collision could not have been avoided by the exercise of such precautions, the showing necessary to bring the case within the exception to the general rule decreeing fault. The Richard J. Barnes, 2 Cir., 111 F.2d 294, 296 ; The Fulton, 2 Cir., 54 F.2d 467, 469 ; The Komiles, D.C.N.Y., 35 F.Supp. 194, 197 . On the contrary the district court in its memorandum decision expressed the belief that: “Had either or both pilots stopped engines when it became apparent that a dangerous situation had arisen, it is unlikely that the vessels would have come together.” The record supports this view, which is also conclusive as to the issue of proximate cause, for if the damage might have been prevented by the stopping of both vessels, their failure to decrease speed was a factor directly contributing to the collision.

Affirmed.