This suit in admiralty against the United States, as owner and operator of the motor-ship Sweepstakes, involves a claim of damage by sea water to a shipment of flour. The trial court found that 3,087 bags out of a total of 10,500 contained in the shipment were in an extensively damaged condition from sea water when the flour was delivered to appellee as consignee at Rio de Ja-neiro. The question we shall consider is whether in the condition of the proof this finding is justified.

The parties are agreed that the problem is for determination do novo, the evidence bearing on the point being entirely •by deposition or by exhibits. The Ernest H. Meyer, 9 Cir., 84 F.2d 496 .

The flour was loaded at New York aboard the Sweepstakes on January 31, 1946, in apparent good order and condition. The Sweepstakes is a Wilmington-type C-2, a modern fully-welded cargo vessel of five holds, with forced-draft ventilation from the top of the king posts, and steel pontoon hatch covers on the weather decks. The testimony is that at the time of loading three tiers of criss-crossed dunnage were laid on the floor of each hold where this and certain other flour was stowed, bottom tiers being laid fore and aft to allow drainage of water into the scuppers. Heavy dunnage paper was laid over this, and over exposed metal and wood, for the protection of the sacks. Cargo battens covered all longitudinals and stringers. The flour was given top stowage for quick discharge as priority cargo, being stowed in Nos. 1, 2, 3, and 4 ’tween decks, and No. 4 lower ’tween deck. In addition to appellee’s shipment, the flour cargo included 13,357 sacks consigned to others.

There was no rain during loading. The vessel’s chief officer testified that on January 31, at which time less than 2,000 bags of the flour had .been loaded, he inspected all the hatches, examining each hold personally. Everything appeared in good order, and he found no indication of moisture or dampness or water in the hatches. When the loading was completed the vessel’s weather hatches were battened down, three new tarpaulins, each inspected and found without tears or defects, were laid properly over each hatch cover and fastened with strongbacks, and ventilation blowers were put in operation. Captain Lane, master of the Sweepstakes, described the weather experienced during the voyage to Rio as “quiet, moderate breeze, moderate seas.” Nothing appears in the log to contradict this characterization of the trip. The ship was light-loaded, about two feet above her marks. The only port of call en route was Trinidad, where the vessel discharged mail from the No. 5 hatch, took bunkers, and shortly departed. The hatches containing the flour were not opened.

The vessel arrived at Rio February 19 and the following day she docked at a deep-water berth. Discharge of the cargo commenced on February 21. While the log indicates that the unloading of all Rio cargo was completed on the morning of the 25th, it definitely appears that the unloading of the flour was completed on the 23d. There were occasional rains during unloading, but as required by customs and is in accord with general practice the work was stopped and the hatches covered during these periods. Lighters were not used. The flour destined for appellee and others was discharged under mixed marks and appellet’s flour was loaded directly on open flat cars, sometimes referred to as wagons, provided for appellee by the docks administration. There appears to have been no detailed inspection, but it is undisputed that the bags were tallied by customs men, Moore-McCormack tally clerks, and dock tally clerks. So far as the record shows, nobody engaged in the work observed any of the bags to be wet or discolored. The flour, being priority cargo, did not pass through customs, but there is competent and undisputed testimony that observed damage would nevertheless normally be called to the attention of the customs officials and noted by them to avoid future doubts with reference to the duties. Customs’ records show no damage of any kind to this shipment.

Several persons whose testimony was taken saw portions of the flour as it was unloaded from the vessel. Chief Officer Parsons testified that he was on duty when the hatches were opened up and that he “sighted” them and looked at the general condition of the cargo, his purpose being to see if there was any sweat damage. He observed none. During the course of discharge the officer frequently entered the compartments to see if there was any damage. After discharge he made a further inspection to see if any cargo had been overlooked. He observed no water-damaged flour, received no report of any, and saw no evidence of moisture or water in the holds. He testified further that he saw no possibility of the flour having come into contact with salt water during the voyage. The master, too, received no report of any water damage to the cargo. The local claims agent for the Moore-McCormack people, who was on the scene much of the time during the unloading, testified to having no knowledge of any damage to the flour either by water or in other visible form at the time of discharge, and he says that “such damage would have to come to my attention through the records in the customs house register and by verbal advice from stevedores or our own employees.” He further thought it “extremely unlikely” that any extensive water damage could have escaped his attention and that of the stevedores, tally clerks and dock and customs personnel.

So much for the evidence bearing on the condition of the consignment as far as observed at the time of its delivery to the consignee. For the discharge of the latter’s burden of proving damage existent as of that time, it relies, as it must, on an inspection and tests made after the flour had reached its premises in a manner now to be stated.

As already observed, all the flour was off the ship by the night of February 23. It was to be taken by rail in the flat cars direct to consignee’s warehouse or mill in Rio, some 1800 meters (a little more than a mile) from the dock. Luiz, a checker for the port administration, testified that as soon as loaded the cars were covered by tarpaulins in good condition and securely tied with a rope. However, other portions of Luiz’ testimony render this observation of little value. He said that he played a very small part in the unloading, and Supervised the removal of perhaps one carload. Asked specifically whether each carload was securely covered with a tarpaulin, he said, “I saw only one wagon.” This witness stated that the cars “probably left the dock the same day they were loaded.” The testimony of the consignee’s employees is that the flour began to arrive at the warehouse on February 22 and continued to arrive in unspecified quantities up to and including March 1. No explanation is proffered for what would seem inordinate delays en route and no information given as to conditions encountered in the interim, or even as to conditions existent when the cars arrived and were unloaded. On inspection by an office clerk at the warehouse numerous of the sacks were found to be wet. On what date or dates the wet flour arrived is not shown. In describing the damaged sacks the clerk states “some bags were wet on top others on the bottom or sides, and others completely wetted.”

A surveyor engaged by consignee’s insurer inspected the flour on March 6 and 8 and forrad 3,087 bags damaged. At this time the sacks were dried and spotted, the flour had formed a hard crust, and the damage was in patches of various sizes on the ends or sides. No pattern or system could be noted. The surveyor reached no conclusion of his own as to the nature of the damage, that is, whether or not it was from salt water. He took samples selected at random from six sacks and delivered the same to a chemist, no sample being preserved. The chemist shortly reported to consignee’s insurer that he had analyzed two samples of wheat flour and had reached the conclusion that the contaminant was salt water. In his deposition given much later he testified that he had made an analysis upon six separate samples and said that salt water was the contaminant.

Appellant attacks the adequacy of this sampling. An experienced chemist testifying on its behalf gave it as his opinion that the sampling of six bags out of 3,087 (about 0.2%) “would be most inadequate.” He stated that the accepted method is to take samples from a number of bags approximately equal to the square root of the total, with the provision that not less than 10 bags be sampled. Application of this method here would have required the sampling of 55 bags.

We suppose that the object of sampling in a situation like this is to establish to a high degree of probability that the samples are representative of the whole. One can readily conceive of circumstances in which the sampling of six bags out of a total of 3,000 might afford a fair basis upon which to predicate a finding as to the condition of the whole lot as, of the critical date; but we think the circumstances would have to be more stable or more fully known than they were here. Obviously some sacks may have encountered much different conditions than others during a week-long sojourn in open cars. Counsel for appellant argues that salty tarpaulins, rain, spray from the seaside, the wooden flat cars perhaps fresh from carrying salt hides or casks, or any number of possible combinations, might account for the salt reaction obtained by the chemist. Appellee replies that the argument is pure speculation, unsupported by evidence. But the speculations are not inconsistent with the proven facts, nor do they seem more infirm than the course of reasoning followed by the trial court. The latter, unable to put a finger on any faulty condition or defect in the vessel or in the care given the cargo, thought it “possible that from some cause unknown during the voyage the sea water was permitted to contact the goods.”

Appellee makes no claim that the carrier is liable for damage to the flour oc curring subsequent to its discharge from the holds, at which time the carrier’s custody of the goods terminated. The burden was accordingly on the libelant to prove by the greater weight of the evidence that the bags were wet when unloaded from the ship. This it undertook to do circumstantially, by a showing that sea water was the culprit. Its case must rest on its analysis of the handful of samples as being sufficient to show that the whole number of damaged bags were contaminated by salt water. In this respect its evidence is not strong even when considered in isolation, and when weighed against the countervailing circumstances, the absence of salt water damage to any other cargo, and the impressive showing of seaworthiness of the vessel it fails to preponderate. Compare Compania Naviera Limitada v. Black, 9 Cir., 183 F.2d 388 , involving an analogous principle. Cf. also The Monte Iciar, D.C., 67 F.Supp. 201 .

Appellee relies on our decision in United States v. Apex Fish Co., 9 Cir., 177 F.2d 364 , involving a question of the sufficiency of the libelant’s proof that a shipment of fish was in good order and condition at the time of loading. We said, 177 F.2d at page 368 , that while the libelant “did not show examination and inspection of all barrels at the very moment of loading, we think to require such proof would have been unreasonable and impractical.” The two cases do not present situations at all parallel.

Appellee cites, also, numerous cases, including The Medea, 9 Cir., 179 F. 781 , Armco International Corporation v. Rederi A/B Disa, 2 Cir., 151 F.2d 5 , The Folmina, 212 U.S. 354 , 29 S.Ct. 363 , 53 L.Ed. 546 , and Schnell v. The Vallescura, 293 U.S. 296 , 55 S.Ct. 194 , 79 L.Ed. 373 , all of which presuppose actual damage appearing on delivery of cargo. The principle they stand for is that, in such state of facts, the ■burden rests on the carrier to bring himself within some exception in the bill of lading whereby he is relieved of liability. These cases are not helpful.

Reversed.

. In addition to the showing outlined above, other factors in evidence indicate the unlikelihood of leakage. One of these is the small and fairly constant bilge soundings as recorded in the ship’s log, the average reading, according to the master, being three to four inches. There were no deep tanks in the ship, and three, at least, of the holds containing the flour were dear of pipe lines. There were no cowl type ventilators. . That customs tallied the sacks as they were unloaded seems fairly established. The master of the Sweepstakes so testified saying “Customs tallied the cargo as it went into the open flat cars. That I observed myself.” . On March 2 appellee wrote Moore-Mc-Cormack, S.A., stating in reference to the consignment that “having noted a certain number of torn sacks with loss of contents, and as many more wet, on the unloading of the flour * * * we are advising you that we hold the steamer’s owners responsible” etc. Appellant fil-sists that this was insufficient notice of the huge damage later claimed to afford appellant opportunity to make a survey. We do not decide the point, but it appears seriously debatable whether the notice was adequate under the circumstances.