W. EUGENE DAVIS, Circuit Judge:
While loading a ship, a loading arm fell from a grain elevator tower causing the ship damage. The district court found the elevator operator, the designer of the tower, and the ship’s charterer liable for that damage. The various parties have filed multiple appeals and cross-appeals raising several issues. We affirm in part, reverse in part, and remand the case to the district court.
I. FACTS
In March 1984, a steel loading arm fell from a grain elevator on the Mississippi River onto the deck of the M/V TREBI-ZOND which was loading cargo in the berth below. This incident damaged the ship and delayed its departure. The shipowners, Orduna, S.A. and Transglobal Maritime Corporation (collectively Orduna), sued its voyage charterers, Euro-Fracht-kontor G.m.b.H. (Euro); the owner and operator of the grain elevator, Zen-Noh Grain Corporation (Zen-Noh); the design engineering firm of the collapsed structure, F & P Engineers, Inc. (collectively, with its insurers, F & P); and the loading arm [1152]*1152manufacturer, Buhler-Miag, Inc. (Buhler). These defendants filed a flurry of cross-claims and third-party claims among themselves.
Following a bench trial, the district court found Zen-Noh and F & P jointly and severally liable to Orduna. The district court awarded Orduna $378,528.75 in damages with interest running from date of judgment. The court also found Euro liable to Orduna, but granted Euro full indemnification from Zen-Noh and F & P. The court apportioned fault between Zen-Noh and F & P at one-third and two-thirds respectively-
The parties raise several issues on appeal. Zen-Noh argues that the trial court erred in finding it negligent in not inspecting and maintaining its grain elevator properly. Zen-Noh also argues that the trial court should have found that the exculpatory clause of Zen-Noh’s dock tariff relieved it from liability. Zen-Noh further argues that the trial court erred in failing to allow it to submit rebuttal testimony, and it questions the damages awarded to plaintiff.
F & P argues that because its inadequate design of the tower did not cause the accident, the judgment against it cannot stand. Euro argues that it should not be liable to Orduna based solely on the safe berth clause in the charter party. Orduna also challenges the district court’s denial of prejudgment interest. We will address all these issues in turn.
II. ZEN-NOH’S APPEAL
A.
Zen-Noh argues first that the trial court erred in finding it negligent in not properly inspecting and maintaining its grain elevator.2 We have said that a wharfinger, such as Zen-Noh, has the
“duty to exercise reasonable diligence to furnish a safe berth and to avoid damage to the vessel. This includes the duty to ascertain the condition of the berth, to make it safe or warn the ship of any hidden hazard or deficiency known to the wharfinger or which, in the exercise of reasonable care and inspection, should be known to him and not reasonably known to the ship owner.”
Bunge Corp. v. M/V FURNESS BRIDGE, 558 F.2d 790, 795 (5th Cir.1977) (quoting Trade Banner Line v. Caribbean S.S. Co., S.A., 521 F.2d 229, 230 (5th Cir.1975)), cert. denied, 435 U.S. 924, 98 S.Ct. 1488, 55 L.Ed.2d 518 (1978). We will review the trial court’s findings that Zen-Noh was negligent in its inspection and maintenance procedures under the clearly erroneous standard. Noritake Co. v. M/V HELLENIC CHAMPION, 627 F.2d 724, 728 (5th Cir. Unit A 1980).
Zen-Noh installed the loading arm at issue in this case in the spring of 1982, and the casualty occurred in March of 1984. The district court found that “[n]o formal inspection was ever made by Zen-Noh of structural members of the loading arms” during this approximately two-year period. The testimony of Zen-Noh’s own plant manager supports this finding. The plant manager further testified that before the casualty Zen-Noh never inspected the upper block assembly of the loading towers, even though such an inspection was feasible. F & P’s expert testified that one should inspect yearly such structural members on lifting devices.
The district court further found that from a walkway one could inspect the tower support visually from a distance of about ten feet. Testimony that the pre-cas-ualty cracks in the steel supporting members would “have been seen on visual inspection by a knowledgeable person with a trained eye” supports this finding. Zen-Noh’s plant manager testified that after the casualty Zen-Noh installed hinges on the grating of these walkways so it could inspect more easily. This evidence supports the trial court’s conclusions that Zen-Noh did not adequately inspect the struc[1153]*1153tural members of its loading tower for almost two years and that this failure to inspect was a proximate cause of the casualty.
As the district court explained, the 75-ton steel loading arm was suspended almost 100 feet above vessels and longshoremen working below. The district court was justified in holding Zen-Noh to a high standard of care and inspection when the risk involved substantial property damage and personal injury. We cannot say that the trial court’s findings that the elevator owner was negligent were clearly erroneous.
B.
Zen-Noh next argues that an exculpatory clause in its dock tariff should have relieved it from all liability to Orduna. Zen-Noh claims Orduna’s berth application incorporated by reference the indemnity provision in Zen-Noh’s dock tariff.3 Before enforcing an indemnification clause for an indemnitee’s own negligence, a court must be firmly convinced that the exculpatory provision reflects the intention of the parties. See United States v. Seckinger, 397 U.S. 203, 211-12, 90 S.Ct. 880, 885-86, 25 L.Ed.2d 224 (1970) (4-3 decision) (courts are reluctant “to cast the burden of negligent action upon those who were not actually at fault”); see also Branch v. Fidelity & Casualty Co., 783 F.2d 1289, 1294 (5th Cir.1986) (“To secure indemnification from one’s own negligence ... the intent of the parties must be expressly and specifically manifested.”).
In the instant case, the shipowner’s agent signed a berth application which he agreed was “in compliance with and subject to all applicable tariffs, rules, and regulations of Zen-Noh Grain Corporation, The South Louisiana Port Commission, and the Federal Grain Inspection Service.” No excerpts of the pertinent texts of these documents or even a list of the documents appeared on the berth application. Thus the berth application did not identify by either date or number the specific dock tariff that purported to exculpate Zen-Noh from its own negligence. To be enforceable, clauses exempting one from one’s own negligence must be “specific and conspicuous.” Restatement (Second) of Contracts § 195 comment b (1981) (emphasis added).4
Zen-Noh’s port coordinator testified that when Zen-Noh began operating he mailed the dock tariff to local shipping agents and to the South Louisiana Port Commission. Although he stated that he kept a list of agents to which he mailed the tariff and that he had return receipts indicating delivery, the port coordinator produced neither the list nor the receipts at trial. Conversely, the shipping agent’s general manager testified that when the casualty occurred the shipping agent had not received a copy of Zen-Noh’s tariff and therefore was unfamiliar with its terms when applying for berth. The district court found that the [1154]*1154plaintiff • never received or consented to Zen-Noh’s tariff. The trial court saw and heard both witnesses. We cannot say his factual determination was clearly erroneous. Given this factual finding, the district court’s conclusion follows: Zen-Noh did not carry its burden of demonstrating that Orduna intended to exculpate Zen-Noh from Zen-Noh’s own negligence.
Zen-Noh cites this court’s opinion in Farmers Export Company v. M/V GEORGIS PROIS, 799 F.2d 159 (5th Cir.1986), in support of its position. This reliance is misplaced. In Farmers Export, we upheld enforcement of specific provisions of a dock tariff that were contained in a berth application. Unlike the instant case, however, the issue in Farmers Export was not the effective incorporation of an exculpatory clause by reference. In that case the clauses at issue were damage provisions, and they were specifically included in the berthing application itself. Id. at 161. Moreover, the captain also signed additional documents which included the relevant provisions. Id. The district court did not err in rejecting Zen-Noh’s defense predicated on the exculpatory clause in its dock tariff.
C.
Zen-Noh argues next that the trial court erred in failing to permit it to submit testimony to rebut F & P’s evidence that Zen-Noh should have inspected the structural members connecting the loading arm to the tower. Zen-Noh presented evidence on this identical issue as part of its case in defense of the main demand. F & P’s evidence raised no new issue. “ ‘[Qjuestions as to order of proof are committed to the sound discretion of the trial judge.’ ” Rodriguez v. Olin Corp., 780 F.2d 491, 494 (5th Cir.1986) (quoting McVey v. Phillips Petroleum Co., 288 F.2d 53, 54 (5th Cir.1961)). The district court’s refusal to allow testimony by rebuttal witnesses will be upheld unless such refusal was an abuse of discretion. Id. The district court did not abuse its broad discretion in ruling that Zen-Noh could not submit this testimony.
III. F & P’S APPEAL
While admitting responsibility for a defect in the design of the loading towers, F & P vigorously argues that this “insufficiency” did not cause the accident and, therefore, it should be exonerated on appeal. Thus F & P does not challenge the district court’s finding that it was negligent. It only challenges the court’s finding that its negligence was a proximate cause of the accident. We review that finding by the trial court under the familiar clearly erroneous standard. Noritake, 627 F.2d at 728.
F & P maintains that the collapse occurred not because of its design defect which admittedly caused cracking in the loading towers but because of a sudden shock overload. It points to the testimony of its expert in supporting this theory. F & P argues that the district court was compelled to accept its expert’s testimony on this theory because of his prominence in the field. The trial court found that the evidence introduced to support F & P’s sudden shock overload theory was “insufficient.” The court specifically found credible the testimony of the designer of the loading arm, which refuted this theory.
This trial was a classic battle of the experts. The credibility determination of witnesses, including experts, ' is peculiarly within the province of the district court. We cannot say the district court committed clear error in accepting the testimony of other experts over that of F & P’s no matter how eminent or learned F & P’s expert was.
The trial court found that F & P’s faulty design together with Zen-Noh’s negligent inspection of its premises caused this accident. The court apportioned one-third of the fault to Zen-Noh and two-thirds to F & P. The trial court’s allocation of fault was not clearly erroneous.
IV. ORDUNA’S DAMAGES
Zen-Noh also disputes the computation of Orduna’s damages. In obtaining a figure that would fairly represent lost prof[1155]*1155its for days lost in the voyage at issue, the district court averaged profits for three voyages: the voyage immediately before the casualty; the casualty voyage itself; and the voyage immediately after the casualty voyage. Zen-Noh argues that the district court should have based its calculations on at least six voyages: three voyages before the casualty and three voyages after the casualty.
Zen-Noh argues that we required the six-voyage computation method in Delta Steamship Lines v. Avondale Shipyards, Inc., 747 F.2d 995 (5th Cir.1984), modified, 753 F.2d 378 (5th Cir.1985). However, in Delta, while approving the trial court’s choice of six voyages, we did not mandate this computation method as the exclusive way of calculating lost profits. Id. at 1001 n. 12; see also Todd Shipyards Corp. v. Turbine Serv., Inc., 674 F.2d 401, 414 (5th Cir.) (approving district court’s calculation of average net profit of vessel by using two years before the casualty), certs. denied, 459 U.S. 1036, 103 S.Ct. 447, 448, 74 L.Ed.2d 602, 603 (1982). To the contrary, we specifically noted that some courts, including the Second Circuit, use the “three voyage rule” which the district court used in the instant case. Delta, 141 F.2d at 1001 n. 12 (citing Moore-McCormack Lines v. The ESSO CAMDEN, 244 F.2d 198, 201 (2d Cir.), cert. denied, 355 U.S. 822, 78 S.Ct. 29, 2 L.Ed.2d 37 (1957)); see also id. at 1001 (“in maritime cases ... a proper method of determining lost detention profits is to seek a fair average based on a number of voyages before and after”).
The district court was not required to average profits from six voyages instead of three to obtain a representative profit. The district court’s methodology permitted it to arrive at Orduna’s damages with “reasonable certainty.” No more is required. The CONQUEROR, 166 U.S. 110, 125, 17 5.Ct. 510, 516, 41 L.Ed. 937 (1897).
Although the district court did not err by choosing to use three voyages to determine lost profits, it did err mathematically in its computations. As shown in more detail in the margin, total profit for the three voyages was $817,096.52 or $8,337.72 per day.5 Multiplying the number of days lost, 17.-396, by the representative profit per day of $8,337.72 results in a total loss of profits of $145,042.98, not $156,477.00 as determined by the trial court.
The other damages that Zen-Noh disputes include purchase of a hatch tent to protect the cargo; expenses of the owner’s representative while attending the loss; multiple survey fees; fuel and oil expenses due to detention; and crew overtime wages. The district court’s determination that these expenses flowed from the casualty and were necessary and reasonable is not clearly erroneous.
V. EURO’S APPEAL
Euro, the vessel charterer, argues that the district court erred in imposing liability against it in the face of the finding that its conduct was unrelated to the tower’s collapse. The district court predicated liability against Euro on the charter party between Euro and Orduna which contained a standard “safe berth” clause.6 The trial court held that the clause made Euro the warrantor of the safety of the berth it selected for the vessel. Euro argues that at most this clause imposed upon it the [1156]*1156obligation to use due diligence to select a safe berth.
In imposing liability against Euro, the trial court relied on three Second Circuit cases. These cases hold that a charterer warrants the safety of the berth it selects when the charter party includes a safe berth clause. See Venore Transp. Co. v. Oswego Shipping Corp., 498 F.2d 469, 472-73 (2d Cir.), cert. denied, 419 U.S. 998, 95 S.Ct. 313, 42 L.Ed.2d 272 (1974); Ore Carriers of Liberia, Inc. v. Navigen Co., 435 F.2d 549, 550 (2d Cir.1970) (per curiam); Paragon Oil Co. v. Republic Tankers, S.A., 310 F.2d 169 (2d Cir.1962), cert. denied, 372 U.S. 967, 83 S.Ct. 1092, 10 L.Ed.2d 130 (1963). But see Hastorf v. O’Brien, 173 F. 346, 347 (2d Cir.1909) (charterer held to reasonable man standard); accord The TERNE, 64 F.2d 502 (2d Cir.) (charterer not liable when ship caught in ice and damaged), cert. denied, 290 U.S. 635, 54 S.Ct. 53, 78 L.Ed. 552 (1933).
The commentators have strongly criticized these Second Circuit decisions which impose liability without fault on the charterer under a safe berth clause. Gilmore and Black argue that “these authorities go too far” for several reasons. G. Gilmore & C. Black, The Law of Admiralty § 4-4, at 204-06 (2d ed. 1975). First, the master on the scene, rather than a distant charterer, is in a better position to judge the safety of a particular berth. The master is an expert in navigation, knows the draft and trim of his vessel, and is on the spot. Conversely, the charterer, who is usually a merchant, may know nothing about navigation or the vessel and is ordinarily far from the scene. Moreover, the charterer customarily chooses ports and berths based on commercial as opposed to nautical grounds. Id.
Second, requiring negligence as a predicate for the charterer’s liability does not increase the risk that the vessel will be exposed to an unsafe berth. The standard safe berth clause does not compel the master to take a vessel into an unsafe berth. Id. Because courts have interpreted safe berth clauses to free the master from any obligation to enter an unsafe port or berth, “it is by no means necessary that they be given the quite different meaning of creating an affirmative liability of charterer to ship, in case of mishap.” Id. at 205; see also J. Ramberg, Unsafe Ports and Berths § 17, at 664 (1967) (“[A]n evaluation of the arguments pro et contra does not lead to a strict liability for the charterer.”); Smith, Time and Voyage Charters: Safe Port/Safe Berth, 49 Tul.L.Rev. 860, 868 (1975) (“[C]ourts are placing an undeserved burden on the charterer in holding him to a warrantor’s liability.”).
Moreover, Euro and the commentators (see G. Gilmore & C. Black, supra, § 4-4, at 205-06; Smith, supra, at 862-63) argue that a Supreme Court decision which has never been overruled or weakened supports their position. In Atkins v. Fibre Disintegrating Co., 2 F.Cas. 78 (E.D.N.Y.1868) (No. 601), aff'd, 85 U.S. (18 Wall.) 272, 21 L.Ed. 841 (1873), the district court confronted a claim for damages to a vessel that resulted when she entered an unsafe port. Judge Benedict held:
[I]t is said that the master was induced to accept [the port] as within the terms of the contract by the representations of the charterers’ agent that it was a safe port; and that his acceptance was a qualified acceptance, given upon representations which amounted to warranty.
... I do not think it could be justly held ... that any thing said or done by the master was calculated to lead the charterers’ agent to' suppose that [the port] was not within the privilege given by the charter party, or to inform him that the charterer was to be held responsible in case the vessel received injury in using that port.
Id. at 79.
On appeal of a jurisdictional issue, the Supreme Court affirmed Judge Benedict’s decision on the merits:
In regard to the merits — after a careful examination of the record — we have found no reason to dissent from the views of the learned district judge by whom the case was heard. However full might be our discussion, we should announce the same conclusions. They are clearly expressed and ably vindicated in [1157]*1157his opinion. To go again through the process by which they were reached would be a matter rather of form than substance.
85 U.S. (18 Wall.) at 299 (citation omitted). Several district courts have at least in dicta expressed a preference for this view. See In re Jubilant Voyager Corp. of Pan. (The COCKROW), 1984 AMC 1725, 1726 (E.D.Va.1983) (citing Gilmore and Black) (the proposition that a safe berth clause holds a charterer liable regardless of fault “may not be a sound one”); California through Dep’t of Transp. v. The S/T NORFOLK, 435 F.Supp. 1039, 1047-49 (N.D.Cal.1977) (charterer not liable to shipowner under safe berth clause when charterer exercised due diligence); National Marine Serv., Inc. v. Gulf Oil Co., 433 F.Supp. 913, 917-18 (E.D.La.1977) (quoting Gilmore and Black) (charterer not liable to shipowner under safe berth clause for shipowner’s own negligence), aff'd mem. per curiam, 608 F.2d 522 (5th Cir.1979); see also International Tank Terminals, Ltd. v. M/V ACADIA FOREST, 579 F.2d 964, 967 (5th Cir.1978) (citing Gilmore and Black) (noting that use of safe berth clause “to impose liability upon the charter party who directs the vessel to a certain port ... has been under attack recently”).
We agree with the commentators cited above that no legitimate legal or social policy is furthered by making the charterer warrant the safety of the berth it selects. Such a warranty could discourage the master on the scene from using his best judgment in determining the safety of the berth. Moreover, avoiding strict liability does not increase risks because the safe berth clause itself gives the master the freedom not to take his vessel into an unsafe port.
In conclusion, we hold that a charter party’s safe berth clause does not make a charterer the warrantor of the safety of a berth. Instead the safe berth clause imposes upon the charterer a duty of due diligence to select a safe berth. Thus because the district court did not find Euro negligent, we must vacate the district court’s judgment in favor of Orduna and against Euro.7
VI. ORDUNA’S APPEAL
Orduna argues that the district court erroneously denied it prejudgment interest. “Under the general maritime law, an award of prejudgment interest is the rule rather than the exception.” In re P & E Boat Rentals, Inc., 872 F.2d 642, 654 (5th Cir.1989); see also Nunley v. M/V DAUNTLESS COLOCOTRONIS, 863 F.2d 1190, 1204 (5th Cir.1989); Albany Ins. Co. v. Bengal Marine, Inc., 857 F.2d 250, 253-54 (5th Cir.1988). The district court may deny prejudgment interest only when peculiar circumstances would make such an award inequitable. Reeled Tubing, Inc. v. M/V CHAD G, 794 F.2d 1026, 1028 (5th Cir.1986); Ryan Walsh Stevedoring Co. v. James Marine Servs., Inc., 792 F.2d 489, 492 (5th Cir.1986). While the district court’s finding of peculiar circumstances is reviewed under the clearly erroneous standard (Noritake Co. v. M/V HELLENIC CHAMPION, 627 F.2d 724, 729 (5th Cir. Unit A 1980)), denial of prejudgment interest based on those peculiar circumstances “is reviewable under an ‘abuse of discretion’ standard.” P & E Boat Rentals, 872 F.2d at 655 (citing Inland Oil & Transp. Co. v. Ark-White Towing, 696 F.2d 321, 327 (5th Cir.1983)).
In Reeled Tubing, we outlined some of the circumstances that justify denial of prejudgment interest: (1) an improper delay of the action caused by the plaintiff; (2) a genuine dispute over a good faith claim in a mutual fault setting; (3) equitable considerations which caution against an award; and (4) a damage award substantially less than [1158]*1158that claimed by the plaintiff. 794 F.2d at 1028.
In the instant case, the district court found that the following circumstances justified denial of prejudgment interest: (1) a genuine dispute as to liability between the parties, including applicability of the dock tariff; (2) complex issues of fact as to engineering and design criteria which the plaintiff introduced at trial; and (3) a judgment substantially less than the amount originally claimed.
First, although a genuine liability dispute between the parties may have existed, it certainly was not a dispute that the plaintiff’s fault contributed to the loss. See Reeled Tubing, 794 F.2d at 1029 (limiting the genuine dispute exception to a mutual fault setting). Fault of at least one of the defendants was highly probable. Like in Reeled Tubing, “no more evidence of a ‘good faith’ dispute exists here than would exist in any nonfrivolous admiralty suit.” Id. Moreover, Orduna ultimately prevailed on all issues, including the nonapplicability of the dock tariff.
Defendants further argue that in the past we often have denied prejudgment interest to plaintiffs without any showing of fault on their part. For example in P & E Boat Rentals, no one suggested that the plaintiffs who were passengers in a collision between two crew boats were at fault, yet they were denied prejudgment interest. Defendants overlook, however, that unlike the instant case the district court in P & E Boat Rentals found that the plaintiffs improperly delayed the trial by taking several delaying procedural steps. 872 F.2d at 655.
Second, the district court points to complex issues of fact as a peculiar circumstance counseling denial of prejudgment interest. In a similar case, we rejected complexity as a sufficient exceptional circumstance to deny prejudgment interest. Socony Mobil Oil Co. v. Texas Coastal & Int'l, Inc., 559 F.2d 1008 (5th Cir.1977). In Socony Mobil, we reversed the trial court’s denial of prejudgment interest which it based in part on the “complex nature of [the] consolidated cases.” Id. at 1014. We said that the plaintiffs “cannot be charged with ... the complexity of the case” and that the complexity was “insufficient ... to justify the denial of prejudgment interest, even under the abuse of discretion standard of review.” Id. Moreover, in this case, the dispute between the defendants Zen-Noh and F & P was the source of much of the factual complexity. From the outset, the innocent shipowner had an excellent chance of recovery from one of the defendants. The battle of the experts was almost entirely between these two defendants with each defendant seeking to shift liability to the other.
Third, the district court based its denial of prejudgment interest on its finding that the plaintiff recovered substantially less than it originally claimed. The plaintiff sought damages of $395,000 in its original complaint; the district court awarded the plaintiff damages of $378,528.75. The amount ultimately awarded, therefore, is 96 percent of the amount originally claimed. At trial Orduna sought to recover $454,183.37. The award of $378,528.75 is still 83 percent of that amount. We cannot say that this judgment is substantially less than that claimed. See Mecom v. Levingston Shipbuilding Co., 622 F.2d 1209 (5th Cir.1980) (award of 36 percent of amount claimed one factor justifying denial of prejudgment interest).
We conclude that the district court abused its discretion in declining to award Orduna prejudgment interest. The underlying casualty occurred in March 1984; judgment was not rendered until February 1989. None of this delay is attributable to Orduna. The defendants had the use of plaintiff’s funds for almost five years before judgment. No particular circumstances exist to permit this inequitable result. On remand the district court shall calculate interest on Orduna’s amended award from the date of loss. Reeled Tubing, 794 F.2d at 1028 (“Prejudgment interest is usually awarded to the date of loss_”); cf. Ryan Walsh Stevedoring, 792 F.2d at 492-93.
VII. CONCLUSION
In summary, we affirm the district court’s findings of negligence by Zen-Noh [1159]*1159and F & P and its apportionment of liability between them. We reverse the judgment in favor of Orduna and against Euro. We affirm the remainder of the judgment except to modify it to correct the court’s calculation of detention damages and to award prejudgment interest. Accordingly, we remand the case to the district court with instructions to modify the judgment consistent with this opinion.
AFFIRMED IN PART, REVERSED IN PART, AND REMANDED.