This litigation grew out of a collision between the Brazilian steamship Mandu and the German steamship, Denderah, which *362occurred in the harbor of Santos, Brazil, in 1929. Certain aspects of the litigation were before this court on a prior appeal, reported as The Mandu, 2 Cir., 102 F.2d 459. The present appeal is from an interlocutory decree that grants limitation of liability to the appellant, owner of the Man-du, but holds both vessels at fault for the collision and awards recovery of half damages to Great American Insurance Company, which filed claim in the limitation proceeding as an underwriter and assignee of underwriters of cargo carried on the Denderah. The district judge wrote an opinion (not officially reported) and also made extensive findings of fact. It is the appellant’s contention that on the facts as found the Mandu should be exonerated from liability, or at least should not be held equally culpable with the Denderah. The latter contention is material because the Brussels Convention of 1910 governs recovery in the case at bar and Article 4 thereof provides that if two vessels are at fault, the liability of each shall be “in proportion to the degree of the faults respectively committed” — with a proviso that if it is impossible to establish the degree of the respective faults, the liability shall be apportioned equally.
The collision occurred at 7:27 P. M. (Denderah time) on July 31, 1929. • It was dark but conditions of visibility were otherwise favorable and lights could be seen at a distance of several miles. There was no wind or tide to impede navigation, the incoming tide being less than a mile an hour. The Mandu was outward bound and her course lay westerly through a narrow channel the northerly side of which was marked by three black buoys, with flashing white lights, spaced about 1800 -meters apart. The most westerly of these is referred to in the briefs as the “corner” buoy or buoy No. 3. There the channel turns in a south-westerly direction, its westerly edge being marked by buoy No. 1, the “waiting” buoy. Buoy No. 1 is southwest from buoy No. 3 and about a mile due west, from the red flashing buoy (No. 4) off Pt. Limoes. This buoy (No. 4) is about three-quarters of a mile southeast of buoy No. 3. It marks the easterly and southerly line of the channel, and an. incoming vessel must make a sharp turn to the east to round it. The Mandu stopped her engines and dropped her pilot at Pt. da Praia. Shortly thereafter she sighted the incoming Denderah, then southwest of buoy No. 1 and displaying a blue light as a signal for a pilot. The Mandu sounded a one blast signal and directed her course .slightly to starboard. The vessels were then a mile apart; because of the turn in the channel, it was a meeting situation and concededly the signal was a proper one under the International Rules applicable to those waters. Those in the Denderah were inattentive and did not hear it. The Mandu blew a second one blast signal, which likewise evoked no response. About a quarter of a mile behind the Mandu was another outbound vessel, the West Calumb. She blew a two blast signal to the Den-derah and received an answer of two blasts. In so responding the navigators of the Denderah supposed that the two blast signal had been sounded by the Mandu as they failed to observe the West Calumb following her out. The navigators of the Mandu, in the belief that the Denderah had answered the West Calumb, then sounded a third one blast signal, which the Denderah immediately answered with one blast. The Mandu and the Denderah were about one-half mile apart — “less than a half a mile”, the district judge says; but it could not have been much less, as four minutes elapsed before the collision and during part of that. time both vessels were reversing at full speed. Moreover, the district court found that the Denderah, if properly navigated, could have passed port to port with the Mandu and starboard to starboard with the West Calumb (Finding 36). The collision occurred “at the turn in the channel” and on the Mandu’s side of the channel. The vessels came to-, gether at right angles, and the stem of the Mandu cut a hole 18 feet deep into the starboard side of the Denderah just forward of the bridge. While the Denderah was not actually stationary at the moment of collision, her speed was not such as to twist the bow of the Mandu.
The faults of the Denderah are numerous and glaring; indeed, although disputed below, they are now admitted. She entered a harbor of known danger, without a pilot and in charge of a captain unfamiliar with the waters, which he had only once navigated,' and that twenty-four years before. She had no lookout, and her navigators were so inattentive as not to hear the first two signals of the Mandu and to mistake the West Calumb’s signal for one sounded by the Mandu. She violated the narrow channel rule and for some unexplained rea*363son brought herself broadside across the course of the Mandu, which had continuously maintained her course on her own starboard side of the channel. It is significant that the Denderah’s captain did not testify and that her owners, although she sank and was a total loss, never made any claim against the Mandu.
Despite these glaring faults, no longer disputed, the appellee argues that the Mandu was also at fault in not stopping her way in the presence of danger. The district judge made only two specific findings as to the Mandu’s speed. Finding 31 says that “she reduced her speed somewhat” when she sounded her second one blast signal; Finding 40 states: “However, it would appear that the Mandu stopped to drop the pilot, then proceeded for’ several minutes full speed ahead, then went at half speed, and, just before the collision, full speed astern.” From the “conclusions of law” (Nos. 17, 18 and 19) it appears that the district court was of opinion that the Mandu should not have proceeded at all after giving her third one blast signal; that she should have stopped, backed and blown three blasts “giving those on the Denderah a chance to extricate themselves from their own carelessness”; and that it was careless for the Mandu to proceed at what, in the circumstances, was excessive speed.
We cannot agree that the Mandu should have sensed danger when she sounded her third one blast signal. This was immediately answered by the Denderah, indicating that she was altering her course to starboard and that her navigators believed a port to port passage was possible, for otherwise they would have sounded the backing signal. They are supposed to know the maneuvering possibilities of their vessel, and their responses to the signal of the West Calumb and then to the signal of the Mandu naturally indicated to the latter that the Denderah expected to pass between the two. Indeed, there is a finding that she could have done so, if properly navigated. Hence the Mandu was under no duty immediately to stop and back. She could not know that the Denderah had mistakenly thought the West Calumb’s signal to have been given by the Mandu. The only carelessness of the Denderah of which the navigators of the Mandu had knowledge was that she was entering without a pilot and was on the wrong side of the channel, but they could reasonably believe that she could get over to the right side, since her one blast signal indicated that she thought so. All of the eye witnesses testified that there appeared to be no danger when the one blast signals were exchanged. The court’s conclusion that the Mandu was negligent in proceeding at all after the exchange and that she should immediately have stopped and backed cannot be supported, in our opinion.
The question remains whether she proceeded at what, in the circumstances, was excessive speed. Her testimony is that she reduced from full to half speed when her first one blast signal was sounded, and from half speed to slow when her second signal was given; after her third signal she continued with her engines at slow speed until the Denderah, whose course after her answering one blast signal had been altered to starboard so that her green light was entirely shut out, suddenly sheered to port across the Mandu’s bow; the Mandu’s engines were then put full speed astern and a backing signal sounded, which the Den-derah answered with a backing signal. The district judge did not specifically discuss the testimony supporting the Mandu’s version of the collision. Finding No. 40, however, seems to indicate that he rejected the testimony that she was proceeding at slow speed when her third one blast signal was sounded; and the depth of the hole cut in the Denderah’s hull tends to discredit her testimony as to speed. An injury so extensive indicates a very considerable speed by the Mandu at the time of impact and justifies an inference either that she was proceeding too fast when danger first became apparent, or else that she should have discovered the danger and reversed earlier. Such inferential faults we cannot regard as of equal gravity with the gross faults of the Denderah. Indeed, if the district judge had seen fit to apply the rule that where fault on the part of one vessel ,is clear and of itself sufficient to account for the collision, doubts as to the management of the other vessel should be resolved in her favor, we should have hesitated to overrule him. But he expressly rejected application of the rule, gave careful consideration to the navigation of the Mandu and found her at fault for excessive speed. In view of the extent of the injury she inflicted, we cannot say the finding is unsupported by evidence.
We cannot agree, however, that her fault was equal in degree to the faults of the *364Denderah. Article 4 of the Brussels Convention requires the liability of each vessel to be apportioned according to the degree of their respective faults, unless this cannot be established. Having regard to all the circumstances we believe that the faults .of the Denderah were four times as serious in degree as. those of the Mandu. The decree will therefore be modified to provide that only one fifth of the damages may be recovered from the Mandu. Full appellate costs are awarded the appellant.
Decree modified.