These appeals and a cross-appeal from a judgment of the District Court for the Southern District of New York awarding damages for cargo loss and damage again confront us with the application to containers furnished by the carrier of § 4(5) of the Carriage of Goods by Sea Act (COGSA), 46 U.S.C. § 1304(5). This provides in pertinent part:

Neither the carrier nor the ship shall in any event be or become liable for any loss or damage to or in connection with the transportation of goods in an amount exceeding $500 per package lawful money of the United States, or in case of goods not shipped in packages, per customary freight unit, ... unless the nature and value of such goods have been declared by the shipper before shipment and inserted in the bill of lading....

By agreement between the carrier, master, or agent of the carrier, and the shipper another maximum amount than that mentioned in this paragraph may be fixed: Provided, That such maximum shall not be less than the figure above named. In no event shall the carrier be liable for more than the amount of damage actually sustained.1

The Facts and the Proceedings in the District Court

The shipments here at issue consisted of 1834 tin ingots shipped to plaintiffs Mitsui & Co., Ltd. and Ataka & Co., Ltd. (hereinafter collectively referred to as Mitsui)2 in Japan and 1705 rolls of floor covering shipped by plaintiffs Armstrong Cork Canada, Ltd. and Armstrong Cork Company to ABC Trading Co., Ltd., also in Japan (hereinafter collectively referred to as Armstrong).3 The shipments were carried by a vessel, the S.S. Red Jacket, owned and operated by defendant American Export Lines, Inc. (AEL). The Red Jacket sailed from New York to Japan on December 26, 1973. The loss occurred on January 10, 1974, when a stow of 50 containers on the weather deck of the Red Jacket collapsed during a storm in the North Pacific, sending 43 containers overboard and damaging the remainder. In Houlden & Co., Ltd. v. S.S. Red Jacket, 1977 A.M.C. 1382 (S.D.N.Y.), aff'd by order, 582 F.2d 1271 (2 Cir. 1978), cert. denied, 439 U.S. 1128, 99 S. Ct. 1045, 59 L. Ed. 2d 88 (1979), rehearing denied, 440 U.S. 968, 99 S. Ct. 1521, 59 L. Ed. 2d 785 (1979), AEL was found liable for the loss. All damage claims were then settled save for the two here at issue.

Trials with respect to the amount of damages recoverable by Mitsui and Armstrong were held before the district court. Mitsui claimed that the ingots were not shipped in packages and that AEL's maximum liability was therefore $500 per ingot under a clause in the bills of lading fixing liability at "$500 per package or per shipping unit", thereby permitting recovery of the full value of the ingots since this was less than the $917,000 produced by multiplying the number of ingots by $500. AEL contended that its liability for the ingots was limited to $2500 on the ground that the shipment consisted of 5 packages, namely, the containers in which the ingots, piled into 124 stacks, had been shipped. Armstrong claimed damages of $357,946.19, the C.I.F. value of the goods, on the basis that each of the 1705 rolls of floor covering was a package and that the package limitation thus did not reduce its recovery; AEL sought to limit liability to $6,500 on the ground that the packages within the meaning of § 4(5) of COGSA were the 13 containers in which the rolls had been placed.

The S.S. Red Jacket was a container ship. The containers in both cases were large metal boxes, those in the Mitsui case being 8' high, 8' wide, and 20' or 40' long, which were packed and sealed by the shippers at their own premises and were intended to be forwarded, unopened, to the consignees at their places of business in Japan.

With respect to the ingots, Judge Motley made the following findings of fact, which are not questioned:

Upon the trial of the issue of damages with respect to the packaging of the ingots, the parties relied upon the testimony developed at the liability trial and the court's findings with respect to such packaging. Each ingot was found to be approximately 4 inches wide, 5 inches in depth and 18 inches long. Each ingot weighed 75 pounds. Each ingot had a 2-inch lip extending out from each end of the top side of the ingot so that the ingot could be lifted by forklift or other device. In November 1973 the shipper, through its agents, requested AEL to provide it with 9 containers to be used for the shipment of 200 long tons of tin ingots. The ingots were loaded into the containers at the United States Navy Construction Battalion Center at Davisville, Rhode Island on a loose, "as is", basis by the employees of the General Services Administration, from which agency the ingots had been purchased by the shipper. The evidence disclosed that the ingots were, in fact, loaded into the containers in stacks consisting of 15 ingots three across and five high. When placed in these stacks of fifteen, they were referred to as "bundles". These bundles of 15 were the way in which the ingots had been stacked on the grounds of the federal installation. When the bundles were loaded onto the floors of the containers, they were arranged three across, a space between each, starting at the front end of the container. This pattern was then repeated. Next the bundles were placed four across, a space between each. This pattern continued to the rear of the container. The bundles were placed in the container in such a way as to have them abut each other at the after end of each bundle, from the front of the container to the rear. These bundles were not banded or strapped together as the use of the word bundle might suggest. They were not secured in any way. There was no chocking, dunnage, or other securing devices employed to keep the bundles from moving or coming apart. The shipper was of the view that on placing the bundles in the containers, it was not necessary to band or strap the bundles since the weight of the ingot (sic) and the manner in which they were stacked made them self securing. The evidence disclosed, however, that during the course of transit, which included a severe storm in the North Pacific, the bundles crumbled and the ingots fell to the floor of the containers in disarray. One of the containers was off-loaded in New York from a feeder vessel which had carried it to the S.S. RED JACKET from Boston to New York, since that container had sustained damage described in the court's opinion on liability. When the ingots in that particular container were reloaded by AEL agents in New York into another container, the evidence disclosed that they were loaded on an entire floor of the container, two high, and a wooden platform placed over the ingots and braced to secure them on the floor of the container.

Other evidence on the damage trial and the liability trial disclosed that with respect to break bulk shipments prior to containerization, ingots were always bundled and strapped or banded together with metal bands and placed in the hold of the vessel. (footnote omitted)

The bills of lading were furnished by the carrier but filled out by the shipper. A typical one for the shipments here at issue listed the contents as follows:

---------------------------------------------------------------------------
                                                   PARTICULARS
         CARRIER'S RECEIPT                    FURNISHED BY SHIPPER
---------------------------------------------------------------------------
  Marks and    No. of Cont.           Description                  Gross
   Numbers    or other Pkgs.           of Goods    Measurement     Weight
---------------------------------------------------------------------------
TWO (twenty foot CONTAINERS                                       HOUSE
  s.t.c. as follows:                                            TO HOUSE
OMLU 122590  30 bundles of 438 pcs.)  GRADE "A"                    33703 #
NICB 4327    30 bundles of 438 pcs.)  TIN INGOT                    33540 #
---------------------------------------------------------------------------
                                                                   67243 #
                                                                30.019 L.T.
                           STOWED IN CONTAINERS
---------------------------------------------------------------------------

The bills of lading contained spaces designed to permit the shipper to declare the nature and value of the goods, but these were left blank. Paragraph 16 of the bills of lading read as follows:

In case of any loss or damage to or in connection with goods exceeding in actual value the equivalent of $500 lawful money of the United States, per package, or in case of goods not shipped in packages, per shipping unit, the value of the goods shall be deemed to be $500 per package or per shipping unit. The Carrier's liability if any, shall be determined on the basis of a value of $500 per package or per shipping unit or pro rata in case of partial loss or damage, unless the nature of the goods and a valuation higher than $500 per package or shipping unit shall have been declared in writing by the Shipper upon delivery to the Carrier and inserted in the bill of lading and extra charge paid. In such case if the actual value of the goods per package or per shipping unit shall exceed such declared value, the value shall nevertheless be deemed to be declared value and the Carrier's liability, if any, shall not exceed the declared value and any partial loss or damage shall be adjusted pro rata on the basis of such declared value. The words "shipping unit" shall mean each physical unit or piece of cargo not shipped in a package, including articles or things of any description whatsoever, except goods shipped in bulk, and irrespective of the weight or measurement unit employed in calculating freight charges.

With respect to the floor covering, the evidence demonstrated that here also the shipper requested the containers to be delivered to its plant for loading and sealing, and that the containers were intended to be delivered unopened to the consignee's place of business in Japan. A typical roll of floor covering was roughly 6' long, contained approximately 60 square yards of material and weighed between 250 and 350 pounds. The court found that each roll was covered with two or three turns of Kraft paper; that there was a disc consisting of several pieces of fibre at the bottom of each roll; that the top was a single disc consisting of a single piece of fibre; that the bottom was covered with a burlap cloth to keep the disc in place; and that as to the shipments by the American Armstrong company, there was a hollow cardboard roll around which the floor covering was wrapped. The bills of lading characterized the shipments as a certain number of containers "shippers load and count said to contain" specified numbers of rolls, the weight and measurement of which were indicated. Paragraph 16 of the bills of lading was identical to that quoted with respect to the ingots.

The district judge attempted to apply to this evidence the "functional packing unit test" first announced in Royal Typewriter Co. v. M/V Kulmerland, 483 F.2d 645, 649 (2 Cir. 1973), and elaborated in Cameco, Inc. v. S.S. American Legion, 514 F.2d 1291 (2 Cir. 1974), with Judge (now Chief Judge) Feinberg filing a separate concurrence. In the Mitsui case, although finding that "prior to containerization, ingots were carried on board vessels in bundles and were always banded or strapped ... and that such bundles are the customary way in which ingots are transported", Judge Motley nevertheless concluded, apparently because the stacking itself was useful in the loading and unloading process, that the unbanded and unstrapped stacks of ingots were packages within the meaning of § 4(5) of COGSA, reducing plaintiffs' recovery to $62,000. Both sides appeal. Mitsui contends that the ingots were not "shipped in packages" and that it should recover $369,404.40, the full value of the ingots, under paragraph 16 of the bill of lading, which states that the carrier's liability shall be $500 "per package or per shipping unit",4 since that figure is less than the $917,000 produced by multiplying 1834 ingots by $500. AEL takes the position that the containers should be regarded as the packages. In the Armstrong case the court found that each roll was a package and, since multiplication of the 1705 rolls by $500 produced a sum exceeding the value of the goods, awarded judgment in the amount of the latter. AEL has appealed, again contending that the container was the package and that its liability therefore should be limited to $6,500. The difficulties experienced by the district judge, the appeals engendered by her decision (with both sides appealing in the Mitsui case), and the problems we have encountered on review make it appropriate for us to make still another endeavor to deal with the "package" problem, particularly as it arises in the context of large carrier-furnished containers.

We begin our inquiry by examining the language of § 4(5), the intent of that provision, and the underlying purposes of the broader statutory scheme of which it is a part. The language of § 4(5) reveals two important points. First, the provision establishes a clear distinction between goods shipped in packages and "goods not shipped in packages"; the package limitation applies only when cargo is in fact shipped in packages.5 Second, § 4(5) does not include any definition of the term "package". As the Ninth Circuit has observed,

"(L)egislation when not expressed in technical terms is addressed to the common run of men and is therefore to be understood according to the sense of the thing, as the ordinary man has a right to rely on ordinary words addressed to him." Addison v. Holly Fruit Products, Inc., 322 U.S. 607, 618, 64 S. Ct. 1215, 1221, 88 L. Ed. 1488 (1944). Since no specialized or technical meaning was ascribed to the word "package," we must assume that Congress had none in mind and intended that this word be given its plain, ordinary meaning. Malat v. Riddell, 383 U.S. 569, 571, 86 S. Ct. 1030, 1032, 16 L. Ed. 2d 102 (1966); Bruhn's Freezer Meats v. United States Department of Agriculture, 438 F.2d 1332, 1338 (8th Cir. 1971). See generally 2A Sutherland on Statutory Construction § 47.31 at 155-56 (4th ed. 1973).

The dictionary definitions of "package," though alone insufficient, provide at least a starting point in this inquiry. Webster's Third New International Dictionary 1617 (1966) defines a package as follows: "a small or moderate sized pack: bundle, parcel ... a commodity in its container ... a covering wrapper or container ... a protective unit for storing or shipping a commodity." The word "package" is defined in Black's Law Dictionary 1262 (rev. 4th ed. 1968) as: "a bundle put up for transportation or commercial handling; a thing in form to become, as such, an article of merchandise or delivery from hand to hand .... As ordinarily understood in the commercial world, it means a shipping package."

Hartford Fire Ins. Co. v. Pacific Far East Lines, Inc., 9 Cir., 491 F.2d 960, 963, cert. denied, 419 U.S. 873, 95 S. Ct. 134, 42 L. Ed. 2d 112 (1974). See Nichimen Co. v. M.V. Farland, 462 F.2d 319, 334 (2 Cir. 1972) (dictionary definitions of the word "package" are "not to be wholly disregarded").

In addition to the guidance supplied by the ordinary meaning of the word "package", further illumination is supplied by the purposes of the package limitation. Those purposes must be understood in terms of the dual function § 4(5) serves: it both limits the carrier's liability to "$500 per package ... , or in case of goods not shipped in packages, per customary freight unit" and makes null and void any agreement reducing the carrier's liability below that level.6 Although the legislative history of COGSA does not make clear why Congress created a ceiling on liability, the drafters of § 4(5) must have contemplated that a fixed limit on liability applicable in the absence of an agreement setting a higher limit would permit the parties to "ascertain at the time of contract when additional coverage was needed, place the risk of additional loss upon one or the other, and thus avoid the pains of litigation."7 Standard Electrica, S.A. v. Hamburg Sudamerikanische Dampfschifffahrts-Gesellschaft, 375 F.2d 943, 945 (2 Cir.), cert. denied, 389 U.S. 831, 88 S. Ct. 97, 19 L. Ed. 2d 89 (1967). As for the prohibition of agreements reducing the carrier's liability below $500 per package or customary freight unit, the legislative history of COGSA demonstrates that the creation of this minimum level of liability was intended to prevent carriers from using their superior bargaining power to compel shippers to agree to provisions reducing their liability to insignificant amounts. The Congress that enacted COGSA was familiar with such provisions. It was noted in the Senate that fixing a substantial level of liability was "important when we recall that present limits are usually $100 or $250". Hearings before the Senate Committee on Commerce on S. 1152, 74th Cong., 1st Sess. (1935), p. 47, quoted in Jones v. The Flying Clipper, 116 F. Supp. 386, 388 n.10 (S.D.N.Y.1953). See Hearings, supra, at 38-39 (referring to Reid v. Fargo, 241 U.S. 544, 36 S. Ct. 712, 60 L. Ed. 1156 (1916), in which the bill of lading limited the carrier's liability for an automobile worth roughly $3,900 to $100).

Finally, we must look to the broader purposes underlying the statutory scheme that Congress enacted in 1936. See Mastro Plastics Corp. v. NLRB, 350 U.S. 270, 285, 76 S. Ct. 349, 359, 100 L. Ed. 309 (1956) (" 'In expounding a statute, we must ... look to the provisions of the whole law, and to its object and policy.' ") (quoting United States v. Boisdore's Heirs, 49 U.S. (8 How.) 113, 121, 12 L. Ed. 1009 (1850)), and generally Cox, Judge Learned Hand and the Interpretation of Statutes, 60 Harv.L.Rev. 370, 375-79 (1947). As the Supreme Court observed in Herd & Co. v. Krawill Machinery Corp., 359 U.S. 297, 301, 79 S. Ct. 766, 769, 3 L. Ed. 2d 820 (1959), "(t)he legislative history of the Act (COGSA) shows that it was lifted almost bodily from the Hague Rules of 1921, as amended by the Brussels Convention of 1924".8 That history leaves no room for doubt that the two dominant objectives of Congress were to ensure uniformity in the basic rights and responsibilities arising out of bills of lading and most important for our present inquiry to fix "an irreducible minimum of immunity of the carrier from liability". H.R.Rep.No. 2218, 74th Cong., 2d Sess., 1 (1936). This latter objective underlies both the carrier's duties to furnish a seaworthy ship and to care properly for the cargo, 46 U.S.C. §§ 1303(1) and 1303(2), and the minimum level of liability established by § 4(5). To fulfill Congress' ultimate objective, both aspects of the statute must be given effect; the carrier's duties would be meaningless if their violation entailed no significant liability. In interpreting § 4(5), courts must therefore take a critical look at any proposed construction of that section that would reduce a carrier's liability below reasonable limits. A similar conclusion is dictated under Lord Diplock's thesis that the carrier must be subject to "a liability for loss or damage to the goods sufficient to provide him with a commercial inducement to undertake precautions (in addition to those required for protection of the vessel), the cost of which would be economically justified by the risk of loss or damage to the goods." Conventions and Morals Limitation Clauses in International Maritime Conventions, 1 Journal of Maritime Law & Commerce (hereinafter JML&C) 525, 527 (1970).9

For many years the controversy over what was a package mainly concerned claims by carriers that large pieces of machinery, which admittedly would not be packages if shipped without some attachment, became such because of the affixing of protective covering for part of the machine or of skids or other devices useful in loading and unloading. The courts reached varying results, relying to a significant extent on the language of the bill of lading and other evidence of the intention of the parties. See Gulf Italia Company v. American Export Lines, Inc., 263 F.2d 135 (2 Cir.), cert. denied, 360 U.S. 902, 79 S. Ct. 1285, 3 L. Ed. 2d 1254 (1959); Aluminios Pozuelo Ltd. v. S.S. Navigator, 407 F.2d 152 (2 Cir. 1968); and Hartford Fire Ins. Co., supra, 491 F.2d at 964-65, and cases therein cited. Another recurring question was the proper categorization of cartons stacked by shippers on their own pallets. In Standard Electrica, supra, 375 F.2d 943, a majority, relying to a considerable extent on shipping documents and correspondence between the parties, held, over a strong dissent by Judge Feinberg, that each pallet rather than each of the six corrugated fibreboard cartons placed upon it constituted the "package" for purposes of § 4(5) of COGSA.

The container revolution added a new dimension to the problem. See generally Schmeltzer & Peavy, Prospects and Problems of the Container Revolution, 1 JML&C 203 (1970). In contrast to the wooden pallets in Standard Electrica, which were 39 in length, 33 in width and 42 in height and carried only six cartons each, containers are large metal boxes resembling truck trailers save for the absence of wheels, roughly 8' high, 8' wide and with lengths up to 40', see Simon, The Law of Shipping Containers, 5 JML&C 507, 510 (1974), capable of carrying hundreds of packages in the normal sense of that term. Unlike the pallets in Standard Electrica, which were provided by the shipper, containers are typically supplied by the carrier, must be returned to the carrier by the consignee, and are used and reused hundreds of times. Many ships, including the S.S. Red Jacket, are so constructed that shipments must be made in containers. The shipper normally pays for the weight of the pallet but not for that of the container. Taking account of all this, we have characterized a container as "functionally a part of the ship", Leather's Best, Inc. v. S.S. Mormaclynx, 451 F.2d 800, 815 (2 Cir. 1971). The Supreme Court has observed, in a different but not unrelated context, that "the container is a modern substitute for the hold of the vessel". Northeast Marine Terminal Co., Inc. v. Caputo, 432 U.S. 249, 270, 97 S. Ct. 2348, 2360, 53 L. Ed. 2d 320 (1977).

It did not take long for the carriers (or their P. and I. insurers) to realize that if they could persuade the courts to consider a container rather than smaller units stowed inside to be the "package" for purposes of § 4(5) of COGSA, they would thereby practically nullify that section, since in today's world a suit for $500 or even, when there were a number of containers, several times that sum would not warrant the cost, see Simon, Latest Developments in the Law of Shipping Containers, 4 JML&C 441, 442 (1973).

The first time that a federal appellate court considered such an effort was our decision in Leather's Best, supra, 451 F.2d at 815-16. The case concerned a shipment of leather packed in 99 cartons which were girded with steel straps. The cartons were shipped in a container furnished by the carrier and loaded by the shipper at its plant. They were shipped under a bill of lading made out by the carrier which described the shipment as "1 container s. t. c. 99 bales of leather" and specifically limited liability to $500 for the entire contents of the container. While acknowledging that treating the containers as packages would promote uniformity and predictability,10 we unanimously held that this result was precluded by the underlying purpose of § 4(5), saying:

(W)e cannot escape the belief that the purpose of § 4(5) of COGSA was to set a reasonable figure below which the carrier should not be permitted to limit his liability and that "package" is thus more sensibly related to the unit in which the shipper packed the goods and described them than to a large metal object, functionally a part of the ship, in which the carrier caused them to be "contained."

Id. at 815 (footnote omitted). The provision in the bill of lading limiting the carrier's liability to $500 was therefore invalid. While we left open the possibility that there might be some instances where a container might be the package, e. g., when the shipping documents, like those in Standard Electrica, gave the carrier no information as to the contents, see 451 F.2d at 815 and n.17, the clear holding of the opinion, as the lower courts and commentators recognized, see du Pont de Nemours International S.A. v. S.S. Mormacvega, 367 F. Supp. 793, 796 (S.D.N.Y.1972), aff'd, 493 F.2d 97 (2 Cir. 1974); Rosenbruch v. American Export Isbrandtsen Lines, Inc., 357 F. Supp. 982, 983-84 (S.D.N.Y.1973), aff'd, 543 F.2d 967 (2 Cir.), cert. denied, 429 U.S. 939, 97 S. Ct. 353, 50 L. Ed. 2d 308 (1976); Sperry Rand Corp. v. Norddeutscher Lloyd, 1973 A.M.C. 1392, 1398 (S.D.N.Y.); Case Comment, 38 Bklyn.L.Rev. 909, 914 (1972), was that at least when what would ordinarily be considered packages are shipped in a container supplied by the carrier and the number of such units is disclosed in the shipping documents, each of those units and not the container constitutes the "package" referred to in § 4(5).

Some 21 months later, in Royal Typewriter Co. v. M/V Kulmerland, supra, 483 F.2d 645, this court was confronted with a question Leather's Best had left open the proper treatment under § 4(5) of a container that is not furnished by and whose contents are not disclosed to the carrier. In that case the shipper had directed his freight forwarder to stow cartons of adding machines in a container furnished by the latter. The bill of lading described the goods as "1 Container Said to Contain Machinery" and did not refer either to the number of cartons or to the nature of the machinery. Noting that the case therefore presented an issue Leather's Best had reserved, the district court distinguished that case and held that the container constituted a COGSA package. 346 F. Supp. 1019, 1024 (S.D.N.Y.1972). Although Leather's Best clearly could have been distinguished on these grounds, the panel that affirmed the decision went further, stating that "underlying Leather's Best ... is the concept that the 'bales' there could have been shipped individually rather than in the container ultimately held not to be a 'package'." Leather's Best was thus to be viewed "as holding that, where the shipper's own packing units are functional, a presumption is created that a container is not a 'package' which must be overcome by evidence supplied by the carrier that the parties intended to treat it as such." The Kulmerland panel then declared that where the shipper's units would not have been suitable for breakbulk shipment, as it held to be true in the case before it, the container was presumptively a COGSA package and the shipper could avoid this result only if he could "show by other evidence that his units are themselves 'packages.' " 483 F.2d at 649. Thus the so-called functional economics test was born.

If the functional economics test were consistent with Leather's Best, we would be bound to follow it here despite the criticism it has received from commentators and other courts, see the discussion infra, and our own conviction that it is at odds with the language of § 4(5) and the purposes of that provision and of COGSA as a whole.11 In fact, however, the Kulmerland test is basically inconsistent with the holding of Leather's Best. Although the Kulmerland panel asserted that the critical fact in Leather's Best was that the bales of leather could have been shipped breakbulk, nothing in the opinion suggested that the decision had proceeded on that basis, there was no evidence on the point one way or the other, and the shipper had advanced no such argument. As the lower courts and commentators recognized, the clear holding of Leather's Best was that carrier-furnished containers whose contents are fully disclosed are not COGSA packages. Under the functional economics test, however, many such containers would be COGSA packages, since few shippers would incur the wasteful expense of supplying packaging unnecessary for container shipments simply to avoid having the container deemed the package.12 Since the case before us involves carrier-furnished containers whose contents were disclosed to the respective carriers, we must apply the square holding of Leather's Best unless that decision can fairly be distinguished and not the inconsistent test announced in Kulmerland.

We therefore do not consider ourselves bound by the Kulmerland test, and do not regard its analysis as helpful to the case before us. Although we have little difficulty with the suggestion that cartons, crates and other units that were treated as COGSA packages when they were shipped breakbulk should ordinarily continue to be so treated when they are shipped in containers, we do not discern a proper basis for the other half of the Kulmerland test, i. e., the rule that the container is presumptively the package where the units inside are not suitable for breakbulk shipment. Even if this might tend to show that each of those units is not a package a conclusion that is by no means ineluctable it does not at all follow that the container is. It could just as reasonably, indeed far more reasonably, be the case that the goods are "not shipped in packages" at all a class of cargo specifically provided for in § 4(5). Kulmerland does not explain why carrier-furnished containers stowed with fully described units unsuitable for breakbulk shipment constitute packages despite our earlier recognition in Leather's Best, since reinforced by the Supreme Court in Northeast Marine Terminal Co., Inc. v. Caputo, supra, 432 U.S. at 270, 97 S. Ct. at 2360, that a container is "functionally a part of the ship", 451 F.2d at 815.13 The Kulmerland test simply does not take into account the important possibility that goods shipped in such containers with packaging insufficient for breakbulk shipment might be "goods not shipped in packages".14

Even counsel representing carrier interests, who might have been expected to welcome Kulmerland, recognized that the shipper who uses carrier-furnished containers may have reason to complain of a rule whereby he "can avoid the 'package' limitation (i. e., a ruling that the container is the package) when using containers only if he ships goods packed in such a way that they need not be shipped in containers" and thereby incurs significant economic waste unless such packaging would be required for subsequent distribution, which is by no means always the case. DeOrchis, The Container and the Package Limitation The Search for Predictability, 5 JML&C 251, 257 (1974). Not surprisingly, counsel representing cargo interests put this point even more forcefully, Simon, supra, 5 JML&C at 522. Both critics agree that the hope expressed in the summation of the Kulmerland opinion that it would provide a " 'common sense test' under which all parties concerned can allocate responsibility for loss at the time of contract, purchase additional insurance if necessary, and thus 'avoid the pains of litigation' ", 483 F.2d at 649, citing Standard Electrica, supra, 375 F.2d at 945, is illusory. Only a few reasons need be stated. How is the shipper to know in advance whether his packages will pass the "functional" test? How particularly is the carrier to know since, unless he has engaged in the waste of sending a representative to the shipper's plant, he will not even know what the packages are? Finally, what basis is there for thinking that the shipper will decide to "purchase additional insurance" depending on his guess that his package may not survive the functional test? In fact, as noted above, he will have almost always purchased full insurance already.

In Cameco, Inc. v. S.S. American Legion, 514 F.2d 1291 (2 Cir. 1974), where Judge Feinberg, concurring specially, expressed doubts about the Kulmerland test, the author of Kulmerland recognized some of these criticisms in an opinion joined by one other judge but did not fully answer them. The result in Cameco is entirely consistent with Leather's Best, since the court refused to apply the package limitation to a container which was supplied by the carrier and whose contents were fully disclosed in the bill of lading, holding that the limitation instead applied to the cartons of canned hams, some of which were shipped on pallets.15

Recent district court opinions in other circuits indicate that the likelihood of general acceptance of the functional economics test is small. The outstanding such opinion is that of Judge Beeks, an experienced admiralty lawyer before his appointment to the bench, in Matsushita Electric Corp. v. S.S. Aegis Spirit, 414 F. Supp. 894 (1976), in the Western District of Washington, a district with many maritime cases. After observing that our decisions in Leather's Best, supra, and Shinko Boeki Co. v. S.S. "Pioneer Moon", 507 F.2d 342 (2 Cir. 1974), on the one hand, and Kulmerland and Cameco, supra, on the other, are irreconcilable, the court concluded, on the basis of a wealth of reasoning, see 414 F. Supp. at 904, that the "functional economics" test "is an unsatisfactory guide to decision-making" because "a test for determining whether a container is a package must reflect the realities of the maritime industry of today while remaining faithful to the express language and legislative policy embodied in the pertinent COGSA provisions", id. at 903-04. After quoting the statement in Leather's Best, supra, 451 F.2d at 815, that treating a container as a package is inconsistent with the congressional purpose of establishing a reasonable minimum level of liability, Judge Beeks wrote, 414 F. Supp. at 907 (footnotes omitted):

Although this approach has not completely escaped criticism, there is, nonetheless, much to commend it. It gives needed recognition to the responsibility of the courts to construe and apply the statute as enacted, however great might be the temptation to "modernize" or reconstitute it by artful judicial gloss. If COGSA's package limitation scheme suffers from internal illness, Congress alone must undertake the surgery. There is, in this regard, obvious wisdom in the Ninth Circuit's conclusion in Hartford that technological advancements, whether or not forseeable by the COGSA promulgators, do not warrant a distortion or artificial construction of the statutory term "package". A ruling that these large reusable metal pieces of transport equipment qualify as COGSA packages at least where, as here, they were carrier-owned and supplied would amount to just such a distortion.

Certainly, if the individual crates or cartons prepared by the shipper and containing his goods can rightly be considered "packages" standing by themselves, they do not suddenly lose that character upon being stowed in a carrier's container. I would liken these containers to detachable stowage compartments of the ship. They simply serve to divide the ship's overall cargo stowage space into smaller, more serviceable loci. Shippers' packages are quite literally "stowed" in the containers utilizing stevedoring practices and materials analogous to those employed in traditional on board stowage.

In Yeramex International v. S.S. Tendo, 1977 A.M.C. 1807 (E.D.Va.), rev'd on other grounds, 595 F.2d 943 (4 Cir. 1979), another district with many maritime cases followed Judge Beeks' reasoning in Matsushita and similarly rejected the functional economics test. Judge Kellam held that when rolls of polyester goods are packed into cardboard cartons which are then placed in containers, the cartons and not the containers are the packages. In another decision in the District Court for the Eastern District of Virginia, Complaint of Norfolk, Baltimore & Carolina Line, Inc., supra, 478 F.Supp. at 392, Judge Clarke concluded that the functional economics test presents "too narrow a test for determining whether the containers in this case are COGSA packages", since "(t)he implementation of Congress' purpose cannot rest on so nebulous a factor as the durability of the shipper's original container". He listed twelve criteria that should be applied in determining whether the shipper's "packages" or the containers should be regarded as the packages for purposes of § 4(5) of COGSA and held that, particularly because of the reference in the shipping documents to the container and not the contents, the containers were the COGSA packages in the case before him.

Developments with respect to the 1968 Brussels Protocol, supra, to which we referred in both Leather's Best and Kulmerland, reinforce the conclusion suggested by the language and purposes of COGSA. The Protocol alters Art. IV(5) of the 1924 Convention in two respects here relevant. One is that the limits of liability are $622 per package or unit or $.90 per pound of gross weight of the goods lost or damaged, whichever is higher. The second change is the explicit treatment of the container problem. Where a container, pallet or similar "article of transport" is used to consolidate goods, the number of packages or units enumerated in the bill of lading as packed in such articles of transport shall be deemed to be the number of packages or units; if, on the other hand, the bill of lading does not show how many separate packages there are, then each "article of transport" shall be deemed a package or unit. Following acceptance by 10 contracting states, the Protocol, which had been signed by 20 countries including the United States, came into effect as among them on June 23, 1977, see 1977 Lloyd's Maritime and Commercial Law Quarterly 512. The Protocol has now been ratified or acceded to by some 17 countries.16 Even if the language and purposes of COGSA left us in doubt as to whether carrier-furnished containers whose contents are disclosed should be treated as packages, the interest in securing international uniformity would thus suggest that they should not be so treated. Cf. Restatement of Foreign Relations Law of the United States (Revised) § 134 (Tent. Draft No. 1 1980). Clearly the goal of international uniformity is better served by the approach in Leather's Best that generally a container supplied by the carrier is not a COGSA package if its contents and the number of packages or units are disclosed, than by the functional economics test of Kulmerland.17 For all these reasons this panel respectfully declines to follow the functional economics test set forth in Kulmerland.18

In light of our decision not to follow the functional economics test,* we have little difficulty in affirming the decision of the district court in Armstrong. The rolls here at issue sufficiently conformed to the dictionary definitions cited in the Hartford Fire Insurance case, supra, 491 F.2d at 963. While the raw rolls of floor covering would not themselves have been packages, here the shipper had wrapped them (whether sufficiently for breakbulk shipment or not is immaterial), had inserted fibre discs at the bottom and the top to protect the rolls, and had wrapped the bottom with a burlap cloth to hold the discs in place. Moreover, it had done everything possible in the bills of lading to put AEL on notice that it considered it was shipping rolls of floor covering and not just loaded containers.

The Mitsui case is more difficult. The stacks of ingots, though described as "bundles", did not conform to the ordinary meaning of that term. The shipper had done nothing to hold them together, although that had been the custom even after containerization. While piling the ingots in stacks reduced the ground area on which they were stored prior to shipment and facilitated loading and unloading by permitting the stacks of ingots to be raised and lowered as such rather than individually, we do not see how these stacks could be regarded as packages in the ordinary meaning of language, and we hold that the district court's contrary conclusion was erroneous.

However, it does not at all follow, as urged by AEL, that the containers become the packages. If the ingots were not shipped in packages, and we hold they were not, then, in the absence of the circumstance discussed below, the $500 limit would apply "per customary freight unit". Nothing in § 4(5) suggests that simply because the goods were not sufficiently wrapped, bundled, or tied to be COGSA packages, the container inexorably becomes the package. The considerations foreclosing such a conclusion, recited earlier in this opinion, are fully applicable here.

On that view, again in the absence of the circumstance discussed below, we would next have to consider what the customary freight unit was in this case. If we were to look only to the language of § 4(5), there would be little doubt that on this record the customary freight unit would be the long ton, on the basis of which the freight was calculated. See Brazil Oiticica v. The Bill, 55 F. Supp. 780 (D.Md.) (Chesnut, J.), aff'd per curiam sub nom. Lorentzen v. Brazil Oiticica, Inc., 145 F.2d 470 (4 Cir. 1944); Waterman S.S. Corp. v. S. R. & M. Co., 155 F.2d 687 (5 Cir.), cert. denied, 329 U.S. 761, 67 S. Ct. 115, 91 L. Ed. 656 (1946); Petition of Isbrandtsen Company, Inc., 201 F.2d 281, 286 (2 Cir. 1953); Gulf Italia Company, supra, 263 F.2d 135; Freedman & Slater, Inc. v. M.V. Tofevo, 222 F. Supp. 964 (S.D.N.Y.1963); India Supply Mission v. S.S. Overseas Joyce, 246 F. Supp. 536 (S.D.N.Y.1965); General Motors Corp. v. Moore-McCormack Lines, Inc., 451 F.2d 24 (2 Cir.), aff'g per curiam 327 F. Supp. 666 (S.D.N.Y.1971); Eaton Corp. v. S.S. Galeona, 474 F. Supp. 819 (S.D.N.Y.1979).19 On this basis Mitsui would recover some $31,000, roughly half of the sum allowed by the district judge but over twelve times the $2500 argued for by AEL.

However, Mitsui persuasively argues that a limitation based on the long ton must be rejected because of the last sentence of Clause 16 of the bill of lading, which reads:

The words "shipping unit" shall mean each physical unit or piece of cargo not shipped in a package, including articles or things of any description whatsoever, except goods shipped in bulk, and irrespective of the weight or measurement unit employed in calculating freight charges.

Section 4(5) explicitly permits the parties to set a level of liability greater than $500 per package or customary freight unit. AEL is clearly wrong in asserting that the ingots come within the exception for "goods shipped in bulk" a term reserved for such items as liquids, grains, or goods not separated into individual units, see, e. g., Standard Oil Co. v. Commonwealth, 119 Ky. 75, 82 S.W. 1020, 1022 (1904); Naftalin v. John Wood Co., 263 Minn. 135, 116 N.W.2d 91, 98 (1962), not for tin ingots which are discrete objects susceptible of separate shipment. AEL argues, with somewhat greater force, that Clause 16 cannot increase its liability under COGSA since Clause 1 states in pertinent part that:

This bill of lading shall have effect subject to the provisions of the Carriage of Goods By Sea Act of the United States of America, approved April 16, 1936, which shall be deemed to be incorporated herein, and nothing herein contained shall be deemed a surrender by the Carrier of any of its rights or immunities or an increase of its responsibilities or liabilities under said Act....

Granting that the bill of lading is thus ambiguous, we would resolve the ambiguity against AEL in light of three principles that ocean bills of lading are contracts of adhesion ambiguities in which must be resolved against the carrier, see Encyclopaedia Britannica, supra, 422 F.2d at 15; the rule of construction contra proferentem, see, e. g., Associated Metals & Minerals Corp. v. M/V Vishva Shobha, 530 F.2d 714, 718 (6 Cir. 1976); and the rule that a specific provision should prevail over a general one. Since, as developed in the statement of facts, multiplication of the number of ingots by $500 would greatly exceed Mitsui's loss of $369,404.80, Mitsui would be entitled to recover the latter.

Nevertheless, there is one consideration that precludes this result and supports that reached by the district court. The typewritten material in the bill of lading, found to have been included by the shipper, represented that the contents of the containers consisted of "bundles" of ingots. If the ingots had in fact been bundled, each bundle would have been a package. See Primary Industries Corp. v. Barber Lines A/S, 78 Misc. 2d 603, 357 N.Y.S.2d 375 (1974); cf. Nichimen Co. v. M/V Farland, supra, 462 F.2d at 334. When the S.S. Red Jacket sailed from New York, AEL had no way of knowing that the shipper's representation was false.20 If the action here were by the shipper, this would seem a classic case for the application of estoppel. Mitsui and its insurer can stand no better; the shipper must be regarded as their agent in preparing the shipping documents.21

We come finally to AEL's objections to the district court's allowance of prejudgment interest. It objects to any allowance of interest, and, if this be overruled, to the running of interest from the dates of shipment in December, 1973, rather than from the expected date of delivery, January 19, 1974, or the even later dates when the insurers paid the consignees.

Although the allowance of prejudgment interest in admiralty is said to be a matter committed to the trial court's discretion, United States Willow Furniture Co. v. La Compagnie Generale Transatlantique, 271 F. 184, 186-87 (2 Cir. 1921); O'Donnell Transportation Co. v. City of New York, 215 F.2d 92, 94-95 (2 Cir. 1954), it should be granted in the absence of exceptional circumstances. The Wright, 109 F.2d 699, 702 (2 Cir. 1940); Moore-McCormack Lines, Inc. v. Richardson, 295 F.2d 583, 592-93 (2 Cir. 1961), cert. denied, 368 U.S. 989, 82 S. Ct. 606, 7 L. Ed. 2d 526 (1962). AEL's arguments against the allowance of any prejudgment interest focus primarily on the complexity of the litigation, the doubt with respect to its liability, and the length of time since the accident a consideration that cuts both ways. Such arguments are properly addressed to the trial court; we find no abuse in its rejecting them in favor of following the general rule.

AEL is likewise mistaken in asserting that it was an abuse of discretion not to defer the accrual of interest until payment by the cargo insurers on whose behalf these actions are being prosecuted. Two of the cases cited, Lytle v. Freedom International Carriers, S. A., 519 F.2d 129 (6 Cir. 1975); and Welded Tube Co. of America v. Hartford Fire Insurance Co., 1973 A.M.C. 555 (E.D.Pa.1972), are not in point.22 In the third, Sprague & Rhodes Commodity Corp. v. S. S. Toronto, 1977 A.M.C. 758, 770 n.1 (S.D.N.Y.), Judge Conner, acting sua sponte, did start the running of interest from the date of payment of the loss by the plaintiff's insurer, reasoning that to allow anything more would constitute a windfall to the insurer on the facts there sub judice. Nevertheless, the general rule remains that:

Although the suit is brought for the use of the insurer, and it is the sole party beneficially interested, yet its rights are to be worked out through the cause of action which the insured has against the common carrier. The legal title is in the insured, and the carrier is bound to respond for all the damages sustained by the breach of his contract. If only part of the loss has been paid by the insurer, the insured is entitled to the residue. How the money recovered is to be divided between the insured and the insurer is a question which interests them alone, and in which the common carrier is not concerned.

Mobile & Montgomery Ry. Co. v. Jurey, 111 U.S. 584, 593-94, 4 S. Ct. 566, 570-71, 28 L. Ed. 527 (1884). See 16 Couch, Cyclopedia of Insurance Law § 16:332 (2d ed. 1966).

This leaves only AEL's claim that it was an abuse of discretion to start the running of interest at the date of shipment rather than at the expected date of delivery. The case law is overwhelmingly to the effect that, both in land and in sea carriage, prejudgment interest is ordinarily awarded from the time when destroyed or lost goods should have been delivered by the carrier. Mobile & Montgomery Ry. Co. v. Jurey, supra, 111 U.S. at 596, 4 S. Ct. at 571; New York, Lake Erie & Western R. R. Co. v. Estill, 147 U.S. 591, 622, 13 S. Ct. 444, 456, 37 L. Ed. 292 (1893); The Gold Hunter, 10 Fed.Cas. 554, 556 (D.C.S.D.N.Y.1832) (No. 5,513); Bazin v. Steamship Co., 2 Fed.Cas. 1097, 1101 (C.C.E.D.Pa.1857) (No. 1,152); The Patrick Henry, 18 Fed.Cas. 1302, 1303 (D.C.S.D.N.Y.1867) (No. 10,805); The Eroe, 8 Fed.Cas. 775, 776 (C.C.S.D.N.Y.1879) (No. 4,522); The Nith, 36 F. 86, 96 (D.C.D.Or.), aff'd, 36 F. 383 (C.C.D.Or.1888); The Arctic Bird, 109 F. 167, 175 (N.D.Cal.1901); Northern Commercial Co. v. Lindblom, 162 F. 250, 255 (9 Cir. 1908); United S. S. Co. v. Haskins, 181 F. 962, 965 (9 Cir. 1910); The Cabo Villano, 18 F.2d 220, 221 (2 Cir. 1927) (interest from date of misdelivery, apparently same date when goods should have been delivered); Lehigh Valley R. Co. v. Russia, 21 F.2d 396, 406 (2 Cir.), cert. denied, 275 U.S. 571, 48 S. Ct. 159, 72 L. Ed. 432 (1927); American Smelting & Refining Co. v. Black Diamond Steamship Corp., 188 F. Supp. 790 (S.D.N.Y.1960); Interstate Steel Corp. v. S. S. "Crystal Gem", 317 F. Supp. 112, 122-24 (S.D.N.Y.1970); Trans-Amazonica Iquitos, S. A. v. Georgia Steamship Co., 335 F. Supp. 935, 942 (S.D.Ga.1971); Iligan International Corp. v. S. S. John Weyerhaeuser, 372 F. Supp. 859, 869 (S.D.N.Y.), aff'd, 507 F.2d 68 (2 Cir. 1974), cert. denied, 421 U.S. 965, 95 S. Ct. 1954, 44 L. Ed. 2d 452 (1975); Jamaica Nutrition Holdings, Ltd. v. Great Circle Shipping, Inc., 433 F. Supp. 1067, 1071 (S.D.Ala.1977). See 96 A.L.R. 18, 46-50 (1935); 36 A.L.R. 2d 337, 374-76 (1954), 34 A.L.R.Fed. 126, 213-24 (1977). The theory underlying these cases is that interest on damages is a form of compensation intended to make the injured party whole, see Miller v. Robertson, 266 U.S. 243, 258, 45 S. Ct. 73, 78, 69 L. Ed. 265 (1924); Lekas & Drivas, Inc. v. Goulandris, 306 F.2d 426, 429-30 (2 Cir. 1962), and that the plaintiff has not suffered any loss until the time when the goods should have been, but were not, delivered. To start the interest running at a prior date arguably would give the plaintiff more then he would have had if the contract had been performed, a result prohibited by a reasonable reading of the provision in § 4(5) of COGSA that "(i)n no event shall the carrier be liable for more than the amount of damage actually sustained." Only a few dated and probably distinguishable cases support award of prejudgment interest from a date earlier than that of anticipated delivery. See Jackson v. The Julia Smith, 13 Fed.Cas. 215, 217 (D.C.D.Mich.1855) (No. 7,136); The Ocean Queen, 18 Fed.Cas. 556 (C.C.S.D.N.Y.1867) (No. 10,410) (date of collision); The Elvaston, 279 F. 935, on rehearing, 279 F. 940, (5 Cir. 1922) (date of destruction of goods). In the absence of unusual circumstances not developed here, any departure from the principle that in cases of this sort prejudgment interest should run only from the date of anticipated delivery must be deemed a "misuse", see Pearson v. Dennison, 353 F.2d 24, 28 n.6 (9 Cir. 1965), of discretion. Cf. Noonan v. Cunard S. S. Co., 375 F.2d 69, 71 (2 Cir. 1967).

The judgment is therefore modified so that prejudgment interest shall run from January 19, 1974, the date on which AEL failed to deliver the goods, rather than the dates of shipment in December, 1973, and is affirmed as so modified. Armstrong may recover its costs from AEL. No costs as between the Mitsui plaintiffs and AEL.

As the author of the unanimous panel opinion in Royal Typewriter Co. v. M/V Kulmerland, 483 F.2d 645 (2d Cir. 1973), as well as of the majority panel opinion in Cameco, Inc. v. S. S. American Legion, 514 F.2d 1291 (2d Cir. 1974), I have been aware that the presumptive "functional economics" test of Kulmerland had limitations. It was not intended as the last word on a problem that as Judge Friendly said in Leather's Best, Inc. v. S. S. Mormaclynx, 451 F.2d 800, 814-15 (2d Cir. 1971), "demands a solution better than the courts can afford," dealing as it does with "a statutory provision that has become ill-suited to present conditions." Id. Kulmerland was an attempt to devise "a 'common sense test' under which all parties concerned could allocate responsibility for loss at the time of contract, purchase additional insurance if necessary, and thus 'avoid the pains of litigation.' " See Kulmerland, 483 F.2d at 649, quoting Standard Electrica, S. A. v. Hamburg Sudamerikanische Dampfschifffahrts-Gesellschaft, 375 F.2d 943, 945 (2d Cir.), cert. denied, 389 U.S. 831, 88 S. Ct. 97, 19 L. Ed. 2d 89 (1967). And as the Cameco opinion indicates, 514 F.2d at 1300, the criticism of the Kulmerland test on the part of counsel in the Journal of Maritime Law & Commerce was by no means totally persuasive, since from an economic point of view it seemed evident that many shippers would have to package their goods for shipment after the goods left the ship's container.

But I have at all times, see Cameco, 514 F.2d at 1300, been aware of Judge Feinberg's statement in his dissent in Standard Electrica, 375 F.2d at 948, that "certainty at the expense of legislative policy and equity is undesirable and often turns out to be ephemeral." And Judge Friendly's most perceptive opinion in this case, coupled with that of Judge Beeks in Matsushita Electric Corp. v. S. S. Aegis Spirit, 414 F. Supp. 894, 903-07 (W.D.Wash.1976), referred to and quoted at length by Judge Friendly, have persuaded me that the "functional economics" test of Kulmerland does not function well and had better be abandoned. In the realm of container shipping, where the bill of lading specifies the contents, the ship's container should not be deemed a package even presumptively only irrespective of how the goods within it are packed. I therefore am joining in the abandonment of the Kulmerland-Cameco test, noting only that the results of neither case would be changed by virtue of today's decision.