This appeal presents the question whether a shipowner is liable to cargo owners for cargo loss caused by the deliberate scuttling of the ship by a member of the ship's crew. The district court concluded that because the shipowner had shown itself independently free of fault in respect of the crew member's act, it was entitled to exoneration from liability under applicable provisions of the Carriage of Goods by Sea Act of 1936, 46 U.S.C. §§ 1300-1315 (COGSA), and gave judgment accordingly. For errors going to the question of COGSA's application to the facts stipulated and found in the district court, we vacate the judgment and remand for further proceedings.

This litigation to decide whether the economic loss resulting from destruction of a ship's cargo shall be borne by the cargo owners or the shipowner grew out of a stark tragedy involving the destruction as well of several human lives.

The M/V MIMI sailed from Miami, Florida, at around 10:30 p. m., October 9, 1975, bound for ports of call in Venezuela and Guyana. Her cargo consisted of bagged fertilizer and other general cargo stowed below and containers of general cargo stowed on deck. A small vessel, just under 500 gross registered tons, the MIMI was owned at the time by the appellee-appellant International Properties Management, S. A., a Panamanian corporation (Shipowner) and was under time charter to two joint venturers, All-Caribbean, Inc. and High Watch Shipping Company, Ltd. (Charterers). She carried a complement of four German officers, four Indonesian seamen, and a Filipino cook. Among the seamen was one Gun Gun Supardi who had signed aboard five months before in Hamburg, Germany, and had since served continuously as a member of the crew.

During loading operations just before departure from Miami, Gun Gun Supardi was cut above one eye when a cable snapped and lashed him. He was taken to a hospital after the MIMI had left the dock, given four stitches and then returned to the vessel. The captain sent him to his cabin to rest and Supardi stayed there for about twenty-four hours. Shortly before midnight on October 10, having left his quarters he encountered the Chief Engineer in another part of the vessel. Words were exchanged, Supardi struck the officer with what he later described as an "iron," then knifed him. When another officer came to the rescue, Supardi knifed him. The berserk seaman then proceeded to seek out and knife in turn each of the other two officers. Those who survived the knife wounds were bludgeoned into submission. Supardi then awakened the sleeping crew members and forced them at knife-point to provision and lower a lifeboat. At some point in this process he shut down the engines and opened the sea-valves in the engine room. At about 4:00 a. m., October 11, 1975, while Supardi stood by in the lifeboat with his captive fellow crew members, the MIMI sank, carrying with it all its cargo and its already dead or incapacitated officers. Later that morning the lifeboat occupants were picked up by a passing ship, and at this point the author of the catastrophe passes from this factual account of the litigation's background.

In due course Shipowner petitioned in the United States District Court for the Eastern District of Virginia for exoneration from or limitation of liability under 46 U.S.C. §§ 181-189. Various owners of cargo and containers (Cargo) then filed claims to recover damages for the loss of their goods, and a variety of cross and counterclaims were eventually added to the litigation.1 After extensive pre-trial proceedings and a trial confined to liability issues, the district court entered a judgment that, Inter alia, exonerated Shipowner from liability in respect of all of Cargo's claims. Cargo appealed, challenging by various assignments of error the district court's judgment insofar as it exonerated shipowner from liability.2

At the outset, we are confronted with the question whether a necessary factual predicate for COGSA's applicability to the question of Shipowner's liability to Cargo has been established. Cargo's claims were made under COGSA theories of liability and the district court applied COGSA in adjudicating the claims, but the Shipowner unsuccessfully urged in the district court and now argues on this appeal that the statute has not been shown to apply to the claims made. This failure of proof, says Shipowner, defeats any right of recovery by Cargo, whose claims were entirely predicated upon COGSA's applicability. In consequence, the argument runs, this provides an alternative basis for affirming the district court's judgment which was entered on the basis that COGSA did apply but that under its terms and on the facts found, Shipowner was entitled to exoneration.

Because we do decide, for reasons developed in Part III, that if COGSA applies, Shipowner is not entitled under its terms to exoneration so that the district court erred in its contrary conclusion, we must consider this alternative basis for affirmance. For reasons that follow, we conclude that the issue of COGSA's applicability must be remanded for determination by the district court on a reopened record.

The factual predicate not established according to Shipowner was that Shipowner had entered into a contract of carriage with Cargo. Liability for cargo loss is imposed under COGSA only on those charterers and shipowners who meet the definition of "carrier" contained in 46 U.S.C. § 1301(a), that is, those "who enter( ) into a contract of carriage with the shipper." In prosecuting a claim for cargo loss under COGSA, the burden of proof to establish this predicate to its coverage is upon the cargo claimant. Associated Metals & Minerals Corp. v. S. S. Portoria, 484 F.2d 460, 462 (5th Cir. 1973); See Yeramex International v. S. S. Tendo, 595 F.2d 943 (4th Cir. 1979). Here Cargo did not offer any proof of the existence of a contract between Cargo and Shipowner.3 Nevertheless the district court found the statute applicable on alternative grounds: (1) that it applied by its terms even though no contract existed; and (2) that Shipowner was estopped by its conduct in the litigation to deny its applicability. There is no basis in law for the first ground.4 The second presents a more difficult problem, but a review of the whole record persuades us that the district court exceeded a proper exercise of its discretion in holding Shipowner estopped on this factual issue.

In its memorandum opinion, the district court identified three factors that justified its holding Shipowner estopped: that the issue was first raised by Shipowner "during final oral argument," and "without prior notice to the claimants"; that Shipowner "had submitted to the COGSA provisions in the final pretrial order"; and that it had similarly submitted to COGSA "in its brief opposing the motion of Cargo to increase bond." Shipowner contends against this that its position throughout the pretrial and trial proceedings was perfectly and openly consistent with a legitimate reliance upon two alternative positions: that COGSA was not applicable; but that if it were, Shipowner was by its terms entitled to exoneration. Shipowner points out that on the first position the burden was upon Cargo, and argues that this completely justified Shipowner's raising the deficiency in proof for the first time at the close of all the evidence, that it was under no duty to alert its adversary to the possible deficiency at any earlier time. Further, says Shipowner, any reference to COGSA in its brief on the bond increase issue and in the pretrial order is similarly consistent with the legitimate maintenance of these alternative tactical positions. In summary, Shipowner's contention is that a fair appraisal of its conduct in the factual and legal context of this litigation does not support the implication of "sandbagging" that is reflected in the district court's estoppel holding. While we find the matter troublesome, we are persuaded on balance to Shipowner's position.

It clearly lies within a trial court's informed discretion to hold a litigant estopped by its litigation conduct to have an issue considered by the trier of fact. See, e. g., Case v. Abrams, 352 F.2d 193 (10th Cir. 1965). Particularly may this be justified when a party has failed to identify an issue for inclusion in a pretrial order while under a clear duty to do so. Id. at 196. We consider this possible basis of estoppel first. Rule 16 of the Federal Rules of Civil Procedure plainly contemplates exactly this consequence and mandates an order "which limits the issues for trial to those not disposed of by admissions or agreements of counsel." Further, it provides that "Such an order when entered controls the subsequent course of the action, unless modified at the trial to prevent manifest injustice." Fed.R.Civ.P. 16 (emphasis added). But "such" an order, within the most obvious meaning of Rule 16, is one in which a Judicial determination of the issues remaining for trial has been made. See Life Music, Inc. v. Edelstein, 309 F.2d 242, 243 (2d Cir. 1962) (per curiam).

Without attempting exhaustive analysis of the whole document here, we can only say that the pretrial order entered in this case is not such an order. It contains no specification of the issues for trial as determined by the pretrial judge. Presumably following the local rule format, it does contain a recitation of "the triable issues as contended by (the respective parties)." While stating the issues in this way as party contentions may sometimes fairly serve the superseding function contemplated by Rule 16 for a Judicial determination of The issues,5 the hazards of ambiguity and resulting surprise are great. The statements of opposing contentions and the resulting difficulty in this case perfectly illustrate the hazards. A "ships passing in the night" allusion is irresistible. If one looks only to the "triable issues as contended by the (Shipowner)," one does not find listed any issue of COGSA's applicability nor, more precisely, of the existence of a contract of carriage between Shipowner and Cargo. But when one turns to Cargo's contentions, there appears this: "(6) Does the Carriage of Goods by Sea Act, 46 U.S.C. 1300, et seq., apply to this case." If issue definition by pretrial order is to be attempted in this questionable way, each party must be entitled to rely not only upon its own statement of "contentions" as to the "triable issues" but upon those of the opponent. Whatever significance Cargo attached to this statement of its own "contention," in total context we think it must be held to wash out the pretrial order as a decisive source of estoppel on the issue.

Neither do we find the other sources of estoppel relied upon by the district court sufficiently compelling, either alone or in combination, to justify the drastic sanction. As indicated, the fact that in its brief resisting bond increase and in its statement of "Factual contentions" in the pretrial order Shipowner adverted to COGSA as if it were applicable is consistent with its maintenance of alternative defensive positions. Similarly, estoppel to deny COGSA's applicability cannot be based simply on Shipowner's act of bringing the limitation proceeding, nor on the prayer for "exoneration from liability" as well as limitation. Finally, though not directed by any argument of the parties, we have sought without success to find any express or implied judicial admission on the issue in the pleadings and discovery materials.

Though we conclude that the district court acted beyond its discretion in finding estoppel here, we do not think that the appropriate way to correct the error is now to hold Cargo to the resulting failure of proof. The ambiguity that infected the pretrial issue-defining endeavors operated as well upon Cargo. At the time the failure of proof was first suggested in the district court an appropriate solution would have been to allow Cargo to reopen its case to attempt proof, possibly with a continuance. Cf. Rule 15(b), Fed.R.Civ.P. In view of the remand necessitated by our disposition, that solution may now, though belatedly, be put into effect.

For the guidance of the district court on remand of this issue, we offer these observations. The burden of proof on the issue is, as indicated, upon Cargo, and must by it be carried, assuming that the issue remains in genuine dispute. Without suggesting any dissimulation on Shipowner's part, we note the obvious possibility that it actually has no way of disputing the issue, having relied so far simply upon Cargo's failure of proof. That very question might well be addressed to Shipowner by the district court upon remand, as it could with profit have been addressed at the pretrial conference. If inquiry reveals the issue to be disputed, as conceivably it could be, See note 3 Supra, Cargo should be given a reasonable opportunity to conduct any discovery required to address the issue.

We turn now to the district court's conclusion that under COGSA the Shipowner was entitled to exoneration from liability. As indicated above, we find this conclusion erroneous. We hold to the contrary that on the facts found and not in dispute in the district court, if COGSA applies, the Shipowner is by its terms not entitled to exoneration.

Discussion of our reasons requires a brief preliminary analysis of the relevant substantive provisions of COGSA and of a segment of the important proof scheme that controls its application in litigation. COGSA provides the substantive framework for determining liability for damage to cargo carried in international trade to or from ports of the United States. As this nation's codification of the international Hague Rules, its applicability to a particular loss depends upon two essential factual predicates: that the damage occurred after loading was commenced and before unloading was completed, See 46 U.S.C. §§ 1300, 1311;6 and that, as discussed in Part II of this opinion, the charterer or owner sought to be charged with the loss was a "carrier" as defined in 46 U.S.C. § 1301(a).7 Assuming its applicability, COGSA in effect then provides two distinct substantive frameworks within which liability for a particular loss may be established. These build essentially upon the two fundamental pre-statutory obligations of public carriers of goods by sea that the courts had established as "overriding" in respect of all common law or contractual exceptions to the original strict liability of carriers as insurers or warrantors of safe arrival and of seaworthiness. These are: the duty to use due care with respect to the cargo, and the duty to use due diligence to furnish a seaworthy vessel. See generally G. Gilmore & C. Black, The Law of Admiralty §§ 3-23, -26 (2d ed. 1975).

COGSA defines the substantive scope of these two obligations in their now codified form in two sets of paired sections each of which defines the general duty and sets out its limits or qualifications. Sections 3(1) and 4(1) together impose (actually in mirror image statement of duty and immunity) an obligation to exercise due diligence to provide a seaworthy ship "at the beginning of the voyage."8 Sections 3(2) and 4(2) together define the "due care" obligation by posting in § 3(2) a general duty to "properly and carefully load, handle, stow, carry, keep, care for, and discharge the goods carried," then laying out in § 4(2) sixteen specific, (a)-(p), and one catch-all, (q), causes of loss or damage to cargo that serve to limit the scope of the duty as generally defined in § 3(2).9

The practical effect of these four sections is to give a cargo claimant two substantive theories upon which he may be able to establish a carrier's liability for a particular loss. Each is clearly a fault theory, as the key duty-defining language "due diligence," "carefully," "fault or neglect" makes plain. This of course imposes critical proof problems for any shipper, and COGSA accommodates to these by a proof scheme that is of supreme importance in its application to particular disputes.

The primary stage of this proof scheme, partly integral to COGSA, and partly the result of judicial gloss, is decisive here and must be summarized. As at common law and under the predecessor Harter Act, a claimant makes out a Prima facie case of liability for loss or damage to his goods10 by the traditional means given to bailors: proof that he delivered the goods to the carrier and that they were not returned, or that he delivered them in good condition and they were damaged upon return. See Mamiye Bros. v. Barber Steamship Lines, Inc., 241 F.Supp. 99, 109 (S.D.N.Y.1965), Aff'd, 360 F.2d 774 (2d Cir. 1966). On this state of proof standing alone, the unseaworthiness theory has not been directly put in play, and it is possible to take the view that the bailor's Prima facie case goes only to the "due care to prevent loss or damage" theory or that it goes as well to the unseaworthiness theory, with corresponding proof burdens placed upon the carrier to avoid the Prima facie case of liability. Compare Maxine Footwear Co. v. Canadian Government Merchant Marine, Ltd., (1959) A.C. 589, 602-03 (Can.P.C.); Toronto Elevators Ltd. v. Colonial Steamships Ltd., (1950) Can.Exch. 371, 375 (proof of both required), With Firestone Synthetic Fibers Co. v. M/S Black Heron, 324 F.2d 835, 836-37 (2d Cir. 1963) (per curiam); Isbrandtsen Co. v. Federal Insurance Co., 113 F.Supp. 357 (S.D.N.Y.1952), Aff'd per curiam, 205 F.2d 679 (2d Cir. 1953) (only proof of § 4(2) cause required). See generally W. Tetley, Marine Cargo Claims 153-54 (2d ed. 1978). Under either view it is clear that faced with a bailor's Prima facie case, whether or not it is accompanied by direct evidence of unseaworthiness causing or contributing to loss, a carrier must at least establish one of the exculpating causes of § 4(2) to avoid the fault that is otherwise imputed to him in respect of the obligation imposed by § 3(2). If he cannot do this, any proof related to the unseaworthiness theory that may have been adduced by either party is simply irrelevant to the determination of liability. Because we conclude that liability in this case was established at this stage of the proof scheme, we need not explore here the further ramifications of that scheme,11 and turn now to the proof in this case.

Except for the coverage predicate element discussed in Part II of this opinion, Cargo's Prima facie case of delivery and failure to return was stipulated. Similarly, the actual cause of loss was plain and undisputed: the ship was willfully scuttled by a member of its crew. Cargo adduced a great deal of evidence designed in various ways to establish that Shipowner had breached its obligation under § 3(1) to provide a seaworthy ship and that these breaches proximately caused the loss. This evidence went essentially to failures "properly (to) man" the ship, 46 U.S.C. § 1303(3)(1)(b), by pointing to carelessness in employing Gun Gun Supardi and in providing a crew adequate to have restrained him or otherwise prevented the scuttling. Shipowner countered with evidence designed to establish its § 4(1) "immunity" under the unseaworthiness theory by showing its "due diligence" in manning the ship in all the many respects challenged by Cargo and the lack of proximate causation between any of its acts or omissions respecting its "proper manning" obligation and the loss. The evidence of record supporting these opposing contentions respecting the unseaworthiness theory is voluminous, but its details need not be further described, for decision turns on whether or not shipowner successfully overcame by proof the Prima facie imputation to it of fault in failing under § 3(2) "properly and carefully (to) carry, keep, care for, and discharge the goods carried."

As indicated in our general analysis of the proof scheme, this could only have been done by proof that established the cause of loss as one of those specifically excepted in § 4(2)(a)-(p), or as "(a)ny other cause arising without the actual fault . . . of the carrier and without the fault or neglect of the agents or servants of the carrier," under § 4(2)(q) (the "Q-clause"). In attempting to carry this burden, the Shipowner was of course constrained by indisputable evidence that the direct cause was the willful act of its seaman employee.

In summary, this was the proof laid before the district court and upon which it concluded that the Shipowner was entitled to exoneration. From that court's memorandum opinion it appears that although the parties argued and the court was advertent to the possible application of both theories of recovery, the court's legal conclusion of exoneration was related only to the unseaworthiness theory.

In attempting to exculpate itself from the breach of its general duty of care under § 3(2) that had been established Prima facie, Shipowner expressly relied upon both § 4(2)(a) (errors in management) and § 4(2)(q) (the omnibus clause). The district court rightly rejected the "errors in management" exception and Shipowner has abandoned any reliance upon it on appeal. When the court then turned to the "Q-clause" omnibus exception, it revealed its critical misapprehension of the proper application of COGSA to the facts of the case, saying:

The exception provided in § 1304(2)(q) appears to provide the same burden given in 46 U.S.C. § 1304(1), namely, that (Shipowner) prove that its fault, if any, did not contribute to or proximately cause the loss complained of. Thus, the one question before this Court is whether (Shipowner) has successfully proven that the cargo loss sustained from the sinking of the MIMI was not proximately caused by its negligence, if any."

The court then proceeded to analyze the evidence as it tended to prove or disprove proximate causal connection between the specific breaches of the seaworthiness obligation urged by Cargo and the loss. Finding no proximate cause, it held the Shipowner exonerated.

In this analysis, the district court wrongly equated the defensive proof requirements of §§ 4(1) and 4(2). While both do clearly relate to the question of Shipowner's fault, they are addressed, as our general analysis of COGSA has shown, to two distinctly different theories of recovery, and consequently have different substantive content. Specifically, the court's equating of the two led it completely to disregard the "Q-clause" requirement that the carrier prove not only that the identified cause of loss arose "without the actual fault and privity of the carrier" but also "without the fault or neglect Of the agents or servants of the carrier" (emphasis supplied). These dual requirements are plainly stated conjunctively in COGSA's "Q-clause."

Rightly applied to the undisputed facts of the case, COGSA posed a narrow question of law that the district court failed to identify as decisive. It is whether the loss of the MIMI's cargo by Supardi's act resulted from a cause "arising without the fault or neglect of the agents or servants of the carrier" within the meaning of the Q-clause, 46 U.S.C. § 1304(2)(q). Stated more abstractly, the question, apparently one of first impression under COGSA, is whether under the "Q-clause" a seaman's willful act of scuttling a ship is an excepted cause of cargo loss that exculpates the carrier/employer from liability.

Unless the plain language of the statute is to be disregarded, the answer seems equally plain: that it is not. Only one narrow interpretive problem may be thought left open by the quoted language. Applied to this case, it is whether Supardi, whose fault clearly caused the loss, was at the time of the scuttling an agent or servant of the Shipowner. That he was at the time employed by Shipowner is undisputed. He was therefore a "servant" of Shipowner unless there is to be read into the statute a "scope of employment" qualification outside whose reach his causative act fell.

Recognizing the problem posed for it by the literal language of this critical provision, Shipowner on this appeal argues that such a "scope of employment" qualification should be read into the Q-clause, citing an English case, Leesh River Tea Co. v. British India Steam Navigation Co., (1966) 2 Q.B. 250 (C.A.), as persuasive authority for the proposition. Because the question is novel and because of the general policy favoring uniformity of interpretation of the various national counterparts to the Hague Rules, See Robert C. Herd & Co. v. Krawill Machinery Corp., 359 U.S. 297, 301, 79 S.Ct. 766, 3 L.Ed.2d 820 (1959), we have given careful consideration to the Leesh River decision.

We need neither adopt nor reject the Leesh River rule as one of general application in order to decide that Supardi's act in this case was that of a servant under the Q-clause, for his act lay within the scope of employment contemplated by Leesh River in any event. In Leesh River a carrier was sued for damage to cargo caused by seawater entering the hold of its ship. Stevedores employed by the carrier to transfer cargo had removed and stolen a brass plate from the hull, leaving a hole through which the water entered after the voyage resumed. Applying the English counterpart to pertinent provisions of COGSA, the Leesh River court held the carrier exculpated from liability under the Q-clause. It reached this result by applying the land-based bailment rule of Morris v. C. W. Martin & Sons, Ltd., (1966) 1 Q.B. 716 (C.A.1965). Within that rule, the Leesh River carrier would have been liable had the stevedores stolen the goods, but they did not steal them, rather they damaged the ship. Since their employment pertained only to the handling of goods, their act of damaging the ship lay outside the scope of their employment per the Morris rule as interpreted in Leesh River and read into the Q-clause.

The Leesh River decision has been criticized by a Canadian commentator, W. Tetley, Supra at 250, for reading Any scope of employment requirement into the statute when none is explicit. While we find the criticism persuasive, we need not in this case reject the Leesh River interpretation out of hand in order to find Supardi's act not an exculpating cause; for that conclusion would follow here even if the Leesh River rule were applied.

Obviously the very terms "agent and servant" imply at the minimum a temporal coincidence of general employment with causative act, and any rational interpretation of the Q-clause must start with that interpretation. Leesh River reads in only a modest further qualification: that the causative act shall have occurred at a place and time, and in respect of an object related to the general duties of the employment. It does not read in any further qualification related to the servant's culpability or state of mind, a qualification of the Q-clause that Shipowner here has contended for. That this last is so is shown by considering the land-based bailment rule of the Morris case from which the Leesh River rule was derived.

Morris resolved an anomaly in English law regarding the liability of English bailees for misconduct of their servants. Prior to Morris a bailee was liable for theft by a third party if it occurred through the negligence of the bailee's employee, Abraham v. Bullock, 86 L.T.R. (n. s.) 796 (C.A.1902), but not if the employee himself was the thief, Cheshire v. Bailey, (1905) 1 K.B. 237 (C.A.1904). The Morris resolution simply shifted focus from the degree and kind of servant culpability to the relationship between act and employment duties. A fur was left with a furrier for cleaning. Since the furrier did no cleaning, the fur was sent with the bailor's permission to a person who did, and an employee of that person stole the fur. The court held the bailee liable, and from the seriatim opinion an important distinction emerged: had the thief been a maintenance employee of the cleaner, no liability of the bailee would have resulted; but because the thief was an employee specifically assigned to clean the fur, liability resulted. See also Lloyd v. Grace, Smith & Co., (1912) A.C. 716. Were this rule to be applied to the instant case, Supardi's act would similarly have fallen within any scope of employment requirement implied as a condition in the Q-clause. Consequently, even assuming the Leesh River rule as a qualification of the statute's literal language, we conclude that under the Q-clause the cause of loss here was not shown to be one arising "without the fault of a servant" of the Shipowner.

We are reinforced in this construction of the Q-clause by two other considerations than the plain meaning analysis. First, we think the result dictated by this construction conforms to the general policy that must underlie COGSA's careful allocation of proof burdens in respect of loss of goods at sea. In the proof scheme earlier analyzed appears a frequently used procedural device to force a preferred substantive result where proof is in equipoise or unavailable, or where only very specific exceptions to generally assumed liability are to be recognized. Where such a device is used it ordinarily reflects an intention that in the case where both parties are without direct fault but one must suffer the loss, it is ordinarily fairer that the loss shall fall upon the party in the better position to have controlled, and to produce evidence of, the operative facts. The device used is the evidentiary presumption (Prima facie case, Res ipsa loquitur ) that casts the burden, hence the loss, on that disfavored party. That such a policy may well also cast the loss upon the disfavored party in some situations not precisely contemplated nor adequately described in a statute is itself likely to be an intended consequence. As between carrier and shipper of goods by sea, carrier is obviously the party more likely to be in control both of events and of evidence of events. So, we think is the result dictated by our construction of the Q-clause likely to conform to congressional policy in enacting COGSA.

Finally, the construction is suggested by considering Supardi's act as one of classic barratry. It meets the definition. See, e. g., Carver's Carriage of Goods by Sea § 185, at 159 (12th ed. R. Colinvaux 1971); And see National Union Fire Insurance Co. v. Republic of China, 254 F.2d 177, 182-84 (4th Cir. 1958) (insurance coverage). Before cargo damage law was codified, barratry was one of the exceptions to liability traditionally listed by the carrier in bills of lading. Many of these exceptions were carried into the specific exceptions in § 4(2) of COGSA. Barratry was not; and as perhaps the most obvious conceivable example of "fault" of a seaman servant, its intended inclusion within the general Q-clause reference to servant fault seems a construction compelled by any common sense reading. From this, it would appear that barratry was simply not intended to be an exculpating cause of loss under COGSA. See Scrutton on Charter Parties art. 113, at 239 (18th ed. A. Mocatta, M. Mustill & S. Boyd 1974).

The judgment of the district court exonerating Shipowner from liability and dismissing Shipowner's claim against Charterers for indemnification is vacated, and the case is remanded for further proceedings consistent with this opinion. If the factual predicate of a contract of carriage is established in accordance with Part II of this opinion, the district court should proceed to determine whether Shipowner is entitled to limitation of liability. If a contract of carriage is not established, judgment dismissing all Cargo claims against Shipowner and all claims derivative of Shipowner's liability should be entered.

On this appeal, Shipowner has suggested that the claims for loss of cargo containers are in no event cognizable under COGSA. Because of the district court's dismissal of all claims, it was not necessary for it directly to address this separate issue. On remand, it should be addressed by that court in the first instance if COGSA is found applicable to the case.

VACATED AND REMANDED FOR FURTHER PROCEEDINGS.