On the night of October 23, 1917, the Steamboat “Duquesne” was bound down the Monongahela River with a spike tow of seven heavily laden steel barges. These barges, were placed ahead of the steamboat and were arranged in three tiers of two barges each, with the remaining barge in front. The Steamboat “Monitor” with a tow of seven empty fiats ahead was bound up stream. The night was dark. Some rain had fallen and a rain fog hung on the water. The river was about a thousand feet wide. The channel was about midway the river and piarallel with the two shores. Both steamboats with' their tows held courses in or near the channel.

The “Duquesne” in descending the river had passed a bend and had straightened out her course. When the tows of the two steamboats were from 200 to 500 yards apart, the “Monitor” sighted the starboard light on the forward barge of the “Duquesne’s” tow. Being the ascending steamer with the right under the rule (Pilot Rules, August, 1911) of selecting the passing manoeuvre, the “Monitor” promptly gave one blast of her whistle, indicating her purpose to pass port to port, to which the “Duquesne” responded by an assenting signal.

Both steamers moved their tows to starboard, but before the ma-noeuvre had been completed both captains discovered that collision was imminent. Thereupon, both reversed their engines with the result that the tow of the “Monitor” cleared the tow of the “Duquesne,” but the forward barge of the “Duquesne’s” tow rammed the “Monitor,” causing her to sink and to sustain the damages for which this libel was filed.

The District Court, finding no negligence on the part of the “Duquesne,” dismissed the libel. The libellant took this appeal.

The record discloses no pertinent question of law. The issue is solely one of fact and raises the one question: Which steamer by its negligence caused the collision?

The evidence is quite sufficient to prove that the “Monitor” had all lights set and brightly burning. This evidence is reenforced by the admission of the captain of the “Duquesne” that he saw the “Monitor” a mile or more away and kept her in sight. It is also proved that *3the “Monitor” was not slow in sighting the “Duquesne’s” tow, and that immediately upon observing its lights, she gave a passing signal; and that, considered with reference to the proximity of the two tows, their relative positions, and the brief time at her disposal, the “Monitor” did not unwisely select the signal or negligently carry out the manoeuvre.

Thus acquitting the “Monitor” of negligence, we turn to the testimony on which negligence is charged to the “Duquesne.” This charge is made upon several grounds:

(1) In violating Rule VI of the supervising inspectors, promulgated under authority of Section 4412 of the Revised Statutes, which requires a steamer navigating a river at a short bend or point, where from any cause a steamer approaching in the opposite direction cannot be seen at a distance of 600 yards, to give a signal of one long sound of the whistle as a notice to any steamer that may be approaching on the other side, and within half a mile of such bend or point.

(2) In violating the Pilot Rules of August, 1911, with respect to tow lights, which requires, when a barge is towed by a steamer ahead, that it shall have a green light on the starboard bow and a red light on the port bow.

(3) In violating the rule respecting steamer lights, presently to be mentioned.

[1] We dispose of the first charge of negligence adversely to the libellant, on 1he ground that the bend or point in the river which the “Duquesne” was passing was not such as the rule contemplates. The bend was not short at all. It was sufficiently long and fiat to enable the “Duquesne” to pick up the “Monitor” and keep her continuously in sight for a mile or more. Similarly, we dispose of the second charge of negligence on a finding that both lights on the tow of the “Duquesne” were properly set and burning.

The question of negligence resolves itself into this: Were the lights on the “Duquesne” itself burning? Of these lights there were two kinds: The guard and forecastle white lights customarily used but not required by law, and the colored stack lights required by Rule 6 of Section 4233 Revised Statutes (Comp. St. § 7948). This rule provides that:

‘■Jiiver steamers navigating waters flowing into the Gulf of Mexico, and their tributaries, shall carry the following lights, namely: One red light on the outboard side of the port smokepipe, and one green light on the outboard side of the starboard smokepipe. Such lights shall show both forward and abeam on their respective sides.”

[2, 3] All the lights with which we are now concerned were incandescent electric lights of ordinary candle power. There is much conflict in the testimony as to whether the guard lights and forecastle lights of the “Duquesne” were burning just prior to the collision, from which we find that the forecastle lights were not burning, though at least five of the ten guard lights (two on one side and three on the other) were burning. But as these lights are not required by the rules and laws of navigation (though helpful, perhaps, *4in disclosing a craft on which .they are burning), they cannot be regarded as substitutes for 'lawful lights; neither can the fact that some of them were burning exonerate the steamboat from negligence in failing to have burning the lights required by law. Therefore, we regard the issue whether the guard and forecastle lights on the “Duquesne” were burning as of no consequence if it be found that her stack lights were not burning. If her stack lights were out, then the “Duquesne” committed a positive breach of a statutory rule of navigation, promulgated to prevent just such collisions as this one. To escape the presumption of fault arising upon such a breach, she was required, under familiar principles, to show, not that her failure to obey the rule probably did not contribute to the disaster, but positively that.it could not have done so. The Pennsylvania, 86 U. S. (19 Wall.) 125, 136, 22 L. Ed. 148.

[4] The lights- of the “Duquesne’s” tow were low upon the water and were, doubtless, long concealed by the river fog from the view of the captain of the “Monitor.” The guard lights of the “Duquesne” were higher but not so high as the stack lights. Had the stack lights been burning, they probably could have been seen above the fog by the “Monitor” in time to have prevented the collision. This inference may fairly be drawn from the fact that the captain of the “Duquesne” saw the lights of the “Monitor” above the fog for at least a mile. Therefore, as we regard this case, it turns at the last on the issue of the “Duquesne’s” stack lights.

[5] On this issue it appears that no one on the “Monitor” saw the stack lights of the “Duquesne” before the collision. Six witnesses,' officers and deck hands of the “Monitor,” testified positively that immediately after the collision and for a short time following the port stack light of the “Duquesne” was not burning.- ít is a permissible inference, based on testimony that the two lights were on the same circuit, that if the port stack light was out, the starboard stack light also was out. Against this testimony one witness, the captain of the “Duquesne,” testified that he observed her stack lights burning a mile and a half above the point of collision, and three witnesses aboard the “Duquesne” testified that they saw the stack lights burning shortly after the collision. None testified that the lights were burning at the time of collision. If this were all the testimony, it would be another instance of the habit of opposing witnesses to swear by their ship and we would have difficulty in deciding where lay the truth. But in this case there was a circumstance which "lends force to the testimony of some of the witnesses and justifies the rejection of the testimony of others.

Immediately after the “Monitor” had sunk and her officers and crew had crawled into the pilot house, which remained above the water, the “Duquesne” moved down to within speaking distance. In response to a request by the men for coal and clothes, the captain of the “Duquesne” sent Anderson, the watch of the “Duquesne,” over to the “Monitor” in a yawl. While there, the captain of the “Monitor” called Anderson’s attention to the fact that there were no stack lights on the “Duquesne.” There followed conversation between thé *5two about the absence’ of these lights. This conversation was within the hearing of five witnesses whose attention was attracted by it and whose gaze was thereby directed toward the “Duquesne’s” stacks. It was this conversation which aroused the attention of these witnesses and caused them carefully to look for lights and which fixed in their minds the recollection that, on looking, they saw none.

Contrary to usual experience in such cases, Anderson, when called to testify for the “Duquesne,” admitted that the captain of the “Monitor” had called his attention to the absence of stack lights on the “Duquesne,” and testified positively that the red stack light was not burning. Being on the port side, he could not testify about the green stack light because the stacks obstructed his view. After Anderson returned to the “Duquesne,” her port stack light came on.

On this testimony, very briefly recited, we think the issue of negligence in failing to keep the stack lights burning as required by law, must be resolved against the respondent. As the respondent has not sustained the burden of showing that this breach of statutory duty could not have been the cause of the collision, the presumption, arising from the breach, that the collision was due to this fault, remains. The Pennsylvania, 86 U. S. (19 Wall.) 125, 136, 22 L. Ed. 148; The Teaser, 246 Fed. 219, 222, 158 C. C. A. 379 (C. C. A. 3d).

The decree below is, therefore, reversed with the direction that the action proceed in harmony with this opinion.