This is an appeal from a judgment in admiralty of the United States District Court, Southern District of New York, John M. Cannella, Judge, sitting without a jury, entered on November 9, 1972, holding defendant Oetker, owner of the vessel S.S. Cap Castillo, liable for damage to a cargo of 4,301 cartons of beef during that ship's voyage from Buenos Aires to Philadelphia. We reverse and remand for a new trial.

The parties agree that inadequate refrigation caused the beef to spoil at some point between its processing in Buenos Aires and its rejection by the United States Department of Agriculture (USDA) subsequent to unloading in Philadelphia. They disagree as to the cause of and the responsibility for the spoilage. This factual dispute is the principal issue raised on appeal.

Under the Carriage of Goods by Seas Act (COGSA), 46 U.S.C. 1300 et seq., the plaintiff has the burden of establishing that the cargo in question was delivered in good condition to the carrier and that it was damaged on discharge. If the plaintiff satifies this burden, the burden of proof then shifts to the defendant to establish that the damage resulted from one of the statutory exceptions of 1304(2). Section 1304(2)(q), which provides that a carrier is not liable for loss resulting from causes which arise without the fault of the carrier, is the relevant exception in this case.1 Judge Cannella concluded that plaintiff had satisfied its burden and that defendant had not shown that the spoilage was not his fault.

The fixed point by which we must steer on appeal is the clearly erroneous rule. McAllister v. United States, 348 U.S. 19, 20-21, 75 S. Ct. 6, 99 L. Ed. 20 (1954). We believe that the evidence supports the conclusion that the beef was in good condition when loaded and in bad condition when unloaded but that an erroneous assumption underlies the finding concerning defendant/appellant's diligence. We therefore reverse and remand for new trial.

In 1966, Frigorifico Anglo S.A., a subsidiary of plaintiff Tupman Thurlow, Inc., purchased cattle in local Argentine markets and slaughtered them at its plant in Buenos Aires. The meat was chilled for two or three days, boned, cut into four-inch squares, placed in a cooking vat, and cooked 200 kilos at a time for a period of one and one-half hours. It was then immersed in cold water to stop the cooking process, dipped into a fat solution, taken to a chilled room and spread out on a table until the beef reached a temperature of 0 degrees centigrade. Next, it was bagged in polyethylene bags, boxed into cartons with steel bands and taken to a freezing chamber where temperatures ranging from minus 18 degrees to minus 20 degrees centigrade were maintained.

This process continued from February to September 1966 until 4,301 cartons were accumulated in the freezing chamber for shipment to plaintiff Tupman Thurlow, Inc. During this period, Frigorifico made various bacteria checks and the Argentina Department of Agriculture (ADA) supervised the process.

On September 26, 1966 the 4,301 cartons of beef were transported from the Frigorifico plant to the pier in refrigerated trucks. The trip took approximately thirty minutes. The cargo was divided into five lots for which five clean bills of lading were issued. The bulk of the cartons were loaded in the upper tween deck of the No. 2 hatch while some cartons from two lots were loaded in the lower hold above a consignment of frozen beef destined for Norfolk, Virginia, stowed two days earlier. The ship's cargo officer testified that while the cartons were being loaded, he took the pulp temperatures of the beef, by drilling a hole into the frozen cartons. The temperatures ranged between minus 8 degrees and minus 11 degrees centigrade. He also stated that the cartons were not damaged when loaded on the S.S. Cap Castillo.

After leaving Buenos Aires on September 26, the S.S. Cap Castillo arrived in Santos, Brazil on October 4, where a shipment of shrimp was loaded and stowed in a space in the square of the hatch of the upper tween deck of the No. 2 hatch. The ship arrived in New York on October 20, where the shrimp was discharged.

On October 25, the ship arrived in Philadelphia and plaintiff/appellee's cargo of frozen cooked beef was discharged. After approximately fifteen drafts containing 600 cartons were unloaded from the ship, an inspection by the USDA revealed the presence of blood smears, tissue, and other related matter on the cartons. Several inspectors also noticed an odor, which one inspector described as similar to the smell of a kill floor of a slaughter house. The USDA ordered Tupman Thurlow, Inc. either to re-box the beef or to return it to Argentina. The beef was re-boxed in Philadelphia and allowed entry into the United States. Three of the lots remained in a Philadelphia warehouse and two were shipped by rail to Napoleon, Ohio, where Campbell Soup maintains a plant. Subsequently, all the lots were subjected to a standard defrost inspection. It was at this point that the beef was found to be spoiled, evidencing a readily detectable odor. Some of the meat was found to be not thoroughly cooked. Meanwhile, the S.S. Cap Castillo had sailed from Philadelphia to Norfolk.

Thus there were four periods of refrigeration: at the Frigorifico plant, in the delivery truck, on board the S.S. Cap Castillo, and in the warehouses or trains between the time of unloading in Philadelphia and the defrost inspections.

The court's finding that the beef was in good condition on loading in Buenos Aires on September 26, 1966, is supported by health certificates issued by the ADA, the readings of the bacteria levels in the beef when stored at the Frigorifico plant, testimony concerning the refrigeration practice at the plant and the clean bills of lading.2 Cargo Officer Bender also testified that the cartons were solid and free of stains when loaded on September 26 and that he found the pulp temperatures to be below the freezing point.

Appellant challenges the weight of the evidence as to good condition on loading. He points to the shipper's failure to produce records which might have disclosed breakdowns in the refrigeration equipment at the plant, to the shipper's failure to establish the temperature in the delivery trucks, to the inconsistency between the ADA certificates (which stated that the beef was thoroughly cooked) and the finding of the USDA that some of it was not thoroughly cooked, and to appellee's failure to produce evidence which would have dispelled the possibility that the beef spoiled during two periods when the plant's equipment was being defrosted.

The court also concluded that the beef was in bad condition upon discharge on October 25, 1966. This finding is supported by testimony of USDA inspectors that the cartons were stained and smelled bad when taken from the ship that day, and by evidence that subsequent examination revealed that the meat had high bacteria levels and a noxious odor.

Appellant disputes this, pointing out that only a small percentage of the cartons were found to be in bad condition on October 25. He notes that the beef was detained by the USDA because of the condition of the cartons, not because of beef spoilage and argues that the USDA's failure to order an immediate defrost inspection evidences the USDA's satisfaction that the beef itself was in good condition upon discharge in Philadelphia. Appellant suggests that the beef may have spoiled subsequent to its unloading, while it was being stored and transported by the USDA.

Although some of appellant's suggestions appear to be unlikely, others are plausible. The clean bills of lading and the testimony of the Cargo Officer would seem to dispel the possibility that the meat spoiled at the plant. The condition of the cartons upon discharge makes it unlikely that the meat spoiled while in the control of the USDA. While it is possible that inadequate refrigeration in the delivery trucks, compounded by the exposure of the meat to the air while it awaited loading in Buenos Aires, might have begun a thawing process which damaged the beef,3 there is no need for us to retry these issues. Appellant has not demonstrated that the finding of the court on these matters was clearly erroneous. Judge Cannella was confronted with a confusing factual situation but there is ample evidence to support his conclusion that appellee satisfied its burden of proof under COGSA.

In concluding that appellant shipowner had not shown that the spoilage was not caused by his negligence, the court below relied on the fact that the temperature chart produced by appellant did not preclude the possibility of a breakdown in the ship's refrigeration system and did not show a temperature in the relevant holds for September 26, 1966, and on the fact that appellant did not produce a book which Cargo Officer Bender suggested contained the best evidence of the refrigeration of the beef during the voyage.

We think that the court below erred in finding that the chart (Plaintiff's Exhibit 14, referred to as Defendant's Exhibit 3 in Bender's deposition) called a 'reefer chart' because of its association with the refrigeration equipment, did not contain a reading for September 26. The matter is significant, because Judge Cannella deduced from the good condition of the shrimp when unloaded in New York that the spoilage occurred between September 26 and October 4, when the shrimp was loaded in Santos. Since the loading in Buenos Aires was completed at one o'clock in the afternoon on September 26, an interpretation of the reefer chart which indicated no recordation on that date would leave fifteen hours unaccounted for by the reefer chart. And because the beef was exposed to the air while the loading progressed, these first fifteen hours could be critical.

The reefer chart itself is difficult to read. Its vertical axis consists of temperatures in centigrade; its horizontal axis consists of dates, the numbering beginning on September 27, 1966. There are two dots, or readings, 5 mm. apart, between each date, signifying that recordings were made every twelve hours. There is one dot 5 mm. to the left of the September 27 date. The ambiguity consists of this: it is unclear what time of day is signified by the date. If it is meant to signify the morning reading, then the dot to the left of September 27 would indicate a September 26 reading. If it is meant to signify the afternoon reading, then no September 26 reading appears.

The testimony concerning the reefer chart was claimed to be ambiguous. Cargo Officer Bender testified that the chart contained a September 26 reading, and at one point he stated that the chart was properly read by interpreting the dots directly above the dates to refer to four o'clock in the morning and the dots between the dates to refer to four o'clock in the afternoon. On the other hand, at two other points Bender stated that some records did not contain a September 26 reading. This latter testimony is claimed to be unclear because of some possible confusion about the labels to be given to different 'logs,' 'reports,' 'workbooks' and 'charts.' The claimed confusion seems more apparent than real. Bender was referring not to the chart, but to the deck log, and work report entries by him, and a reefer log, in which he would not make entries (App. 339a, 325a-327a).

Moreover, the testimony of the Cargo Officer is not the sole evidence. Although the chart is by no means plain, we think that close analysis reveals that Bender's testimony that the dots above the dates refer to four o'clock in the morning very probably is correct.

A comparison of the facts known about the unloading in Philadelphia on October 25 with the readings on the reefer chart on about that date indicates that the dots above the dates signify readings at four o'clock in the morning. We know that the cargo was unloaded beginning at eight o'clock in the morning and that the unloading ended at 2:30 in the afternoon. During the course of the unloading, the refrigeration equipment would be deactivated and the holds exposed to the outside air, causing an increase in the temperatures in the compartments. In fact, the reefer chart shows an increase in the temperature of over 3 degrees centigrade on about October 25, followed by an equal decrease in the temperature. We can assume that the sudden temperature decline was caused by the closing of the hatch and the reactivation of the refrigeration equipment, in preparation for the trip from Philadelphia to Norfolk. The rise and fall is shown on the chart to have occurred within the space of two dots, or twenty-four hours.

If we assume that the dots above the dates refer to four o'clock in the morning, these facts follow. At four in the morning on October 25 the temperature was minus 20 degrees in the holds. Unloading began at eight in the morning. By four in the afternoon that day, the temperature was more than minus 18 degrees. Unloading ended at 2:30. At four in the morning October 28, the temperature was back to minus 20 degrees.

There were two unloadings beside the Philadelphia unloading, in New York on October 20 and in Norfolk on October 26. A comparison of the facts known about these discharges supports the inference derived from the above analysis.

On October 20, from 1:45 in the afternoon to 2:40 in the afternoon, the shrimp was unloaded in New York. The reefer chart shows a substantial temperature increase in the holds between four in the morning of the 20th and four in the afternoon and a sharp decrease in the temperature in the next twelve hours, using the time line of the above assumption. This interpretation conforms without difficulty to the facts we know about unloading and its effects on the temperature in the holds.

On October 26, from 1:25 in the afternoon to four in the afternoon, a cargo of beef was unloaded in Norfolk. Norfolk was the last stop on the voyage, and it is reasonable to assume that the refrigeration equipment was deactivated for the duration when the Norfolk cargo began to be unloaded. The reefer chart indicates, based on our assumption, a temperature at four in the morning on October 26 of slightly over minus 20 degrees. The chart stops at that point. This is consistent with a shut-down of the equipment before the four in the afternoon reading.

One other fluctuation in the chart might be predicted, that associated with the loading of the shrimp in Santos on October 4, between 4:50 and 6:30 in the afternoon. However, no significant fluctuation appears in the chart around this time. This is explained by the fact that the four o'clock reading had been taken just before the beginning of the loading.

To summarize, fluctuations in three places in the reefer chart support the conclusion that the temperature readings above the dates on the chart refer to readings at four in the morning. This is consistent with Bender's testimony explaining the chart. Together, they comprise persuasive evidence that the reefer chart contains a temperature reading for four in the afternoon on September 26-- a reading of minus 16 degrees just three hours after the beef was loaded in Buenos Aires.

Thus, the reefer chart evidences the temperatures in the holds throughout the voyage from Buenos Aires to Philadelphia and indicates that the beef was properly refrigerated while on board the S.S. Cap Castillo. Appellant contends that under Manhattan Fruit Export Corp. v. Royal Netherlands Steamship Co., 271 F.2d 607 (2d Cir. 1959), cert. denied, 363 U.S. 812, 80 S. Ct. 1249, 4 L. Ed. 2d 1154 (1960), the chart must be accepted as prima facie evidence of his due diligence.

In Manhattan Fruit, plaintiff shipped plums under an agreement calling for refrigeration at temperatures between 35 degrees and 40 degrees or 41 degrees fahrenheit. When unloaded, the plums were spoiled, and the issue was whether the carrier had maintained the proper temperature in the hold. The carrier sought to introduce into evidence charts, constructed by recording the temperature in the compartment every two hours on scraps of paper and later averaging to obtain a daily temperature in the compartment. The district court refused to credit the charts because there was no proof as to when the temperature record was made and as to how soon they were made after the temperature was recorded on the scraps. The Court of Appeals held that it was clear error to find that the charts were unreliable, stating that they constituted prima facie evidence of the carrier's due diligence. Here, the court below dealt with Manhattan Fruit by stating that that decision 'should be confined to its facts,' but the court did not spell out why it thought the reefer chart was less reliable than the charts in Manhattan Fruit.

Two factors might make the reefer chart less credible than those in Manhattan Fruit. First, appellant took the temperature in the hold every twelve hours, whereas the carrier in that case measured the temperature every two hours. We have not been presented, however, with evidence suggesting that this difference is material. Second, there was no evidence in Manhattan Fruit that the carrier was suppressing better evidence of the temperatures in the hold. Here, there is such a claim, in the form of the mysterious Kuehltagebuch (literally, 'daily cooler book.')

Cargo Officer Bender, who was in charge of compiling the reefer chart, was the only witness to suggest the existence of the Kuehltagebuch. He stated that such a book is kept by the ship's engineer and contains, in addition to a record of temperatures in the refrigerated holds, 'information about occurrences, about the way the compressors work, about the working of the whole refrigerating system.' Bender testified that he did not know whether a Kuehltagebuch was kept on the S.S. Cap Castillo during the voyage in question but that in his experience a Kuehltagebuch is always kept. Appellant's counsel was asked to produce the Kuehltagebuch, and subsequently counsel informed the court below that no such book was among the records of the voyage. Counsel also pointed out to the court that the reefer chart had been signed and approved by the Chief Engineer of the S.S. Cap Castillo.

The Kuehltagebuch is clearly material to the issue of due diligence. It might contain an unambiguous record of the temperatures in the hold, and it might also preclude the possibility of breakdown in the refrigeration equipment.

The non-production of material evidence which is in the control of a party raises an inference that that evidence is unfavorable to that party. See, e.g., United States v. Delaware Bay and River Pilots' Ass'n, 44 F.2d 1, 4 (3d Cir.), cert. denied, 283 U.S. 838, 51 S. Ct. 486, 75 L. Ed. 1449 (1930); The Motor Launch No. 12, 65 F. Supp. 252, 253 (E.D.Pa.1946); J. Gerber & Co. v. S.S. Sabine Howaldt, 437 F.2d 580, 593 (2d Cir. 1971). This rule has been said to be particularly applicable to suits in admiralty, where records are often crucial and records and witnesses often are in the control of one party. O. F. Shearer & Sons v. Cincinnati Marine Service, Inc., 279 F.2d 68, 73 (6th Cir. 1960). The inference raised has been said to be sufficient to decide a close case. CaryDavis Tug & Barge Co. v. United States, 8 F.2d 324, 325 (9th Cir. 1925). This inference can of course be rebutted by an adequate explanation for the non-production.

The court below evidently believed that appellant's explanation for the non-production of the Kuehltagebuch was inadequate because the court stated that the non-production was evidence of appellant's failure to sustain his burden of proof. On the record before us, we cannot say that this conclusion was error.

However, although the court below was justified in taking the non-production as evidence unfavorable to appellant, we hesitate to conclude that the non-production can stand alone as the determinative factor in the case. As noted, the other evidence which the court below relied on in concluding that appellant had not demonstrated due diligence (the purported absence of a September 26 temperature entry on the chart) appears to be incorrect. The mysterious Kuehltagebuch alone remains to support that conclusion. We are not convinced that Judge Cannella intended that this evidence bear that much weight.

We think that the most appropriate course to follow in these circumstances is to remand for a new trial.4 This disposition will permit a trier of fact to sift evidence and weigh credibility in order to judge what weight the non-production should have. It will permit the parties to introduce evidence which might shed more light on the reliability of the reefer chart. The parties may, for example, illuminate the significance, if any, of the twelve-hour gap between the readings on the chart. Appellant might endeavor to produce the Chief Engineer, who would have kept the Kuehltagebuch. It may be that the reefer chart is sufficient to sustain appellant's burden; again, it may be that the non-production of the Kuehltagebuch is outcome-determinative. A thorough review of the record and due consideration of the importance of credibility to the issue of non-production convinces us that the ultimate resolution of this question cannot be achieved from that record.

Appellant objects to various evidentiary rulings by the court below. While none of these rulings requires reversal, they merit discussion because these issues will undoubtedly arise at the new trial.

Appellant objects on the ground of authenticity to the introduction of the Argentine health certificates and of the report of Inspector Drew of the USDA. However, the parties stipulated that the contents of the USDA file would be admissible, and these documents were contained in that file. Therefore, the documents are admissible. Ross v. American Export Isbrandtsen Lines, Inc., 453 F.2d 1199, 1201, n. 2 (2d Cir. 1972).

Appellant also contests the introduction of the Frigorifico bacteria charts on the ground of appellee's failure to identify the charts in the PreTrial Memorandum and Order. This objection is governed by Rule 16 of the Federal Rules of Civil Procedure, and the standard to be applied in reviewing decisions under that rule was stated in Laguna v. American Export Isbrandtsen Lines, Inc., 439 F.2d 97, 101 (2d Cir. 1971):

Rule 16 provides, in pertinent part, that the pretrial order controls the subsequent course of the action, unless modified at the trial to prevent manifest injustice. Whether to allow such a modification in a particular case poses nice questions of balancing 'the need for doing justice on the merits between the parties (in spite of the errors and oversights of their attorneys) against the need for maintaining orderly and efficient procedural arrangements.' 3 J. Moore, Federal Practice P16.20, at 1136 (3d ed. 1968). But on the whole, we have not viewed such modification with hostility. E.g., Clark v. Pennsylvania R.R., 328 F.2d 591, 594 (2d Cir.), cert. denied, 377 U.S. 1006, 84 S. Ct. 1943, 12 L. Ed. 2d 1054 (1964); Scott v. Spanjer Bros., 298 F.2d 928, 931 (2d Cir. 1962).

Judge Cannella at first refused to allow the charts to be introduced, reasoning that their surprise production worked an injustice on appellant. Later in the trial, when he became convinced that appellant was well prepared to cope with the charts the trial judge decided to hear the evidence. We think that this decision was fair to the parties and that it satisfied the interest of orderly procedure.

Finally, appellant contends that the report of an inspector of his insurance company which was compiled on September 24 in Buenos Aries and which would have demonstrated that the hold was pre-cooled to minus 15 degrees prior to the loading of the Norfolk beef should have been admitted into evidence. The court below refused to admit the report on the basis of hearsay, but appellant maintains that it is admissible as a document kept in the regular course of business under 28 U.S.C. 1732.

In the light of Cargo Officer Bender's testimony and of notations in Bender's log which evidence this pre-cooling, the exclusion of the report did not prejudice appellant to a significant degree. However, upon retrial the report may well be admitted because there is no evidence in the record to the effect that the report, which was prepared prior to notice of the accident and which is normally undertaken, was prepared with this litigation in mind. Cf. Palmer v. Hoffman, 318 U.S. 109, 63 S. Ct. 477, 87 L. Ed. 645 (1943).

The case is remanded to the district court for a new trial.