BOUDIN, Circuit Judge.

Over the course of three years, beginning in early 1987, Knapp Shoes, Inc. (“Knapp”) purchased nearly 300,000 pairs of shoes from Sylvania Shoe Manufacturing Corp. (“Sylva-nia”). The relationship between the two companies underwent strains during its final year, and broke off early in 1990. Knapp filed suit in April 1990 claiming that Sylvania had manufactured defective shoes; Sylvania counterclaimed for unpaid bills. In March 1995, the magistrate judge awarded net damages of less than $65,000 in favor of Sylvania. Knapp appeals. We affirm in part, reverse in part, and remand for the entry of a new judgment as specified in this opinion.

I. THE UNDERLYING FACTS

Knapp, a Massachusetts corporation, manufactures and distributes work shoes. In addition to selling and distributing shoes that it manufactures, Knapp also sells and distributes shoes manufactured by other shoe companies under the Knapp logo. Sylvania, a Pennsylvania corporation, was one such supplier to Knapp.

In late 1986, Jack Esser, then Knapp’s vice president for merchandising and manufacturing, told Knapp personnel to contact Sylvania to arrange for the manufacture of shoes Knapp was selling to the U.S. Postal Service. Sylvania delivered over 10,000 pairs of two styles of shoes — models 1249 and 1250 — by mid-February 1987. By all accounts, there were few problems with these shoes, nor were there problems with over 5,000 pairs of 1249s delivered between September 1987 and May 1988.

Thus encouraged, Sylvania and Knapp expanded their collaboration, and by early 1988 Sylvania had made or was making over two dozen models of shoes for Knapp. These later models all differed in construction from the 1249s and 1250s. While the latter in each case consisted of a leather upper cemented to a polyurethane sole, the new models were constructed of three parts: a rubber outsole, an ethyl vinyl acetate (EVA) midsole, and a leather upper. Among Knapp’s various problems with Sylvania shoes, the most serious complaint was that the leather upper and the EVA midsole tended to fall apart.

The bulk of Knapp’s purchases were in three categories. The first, style 1251, ac *194 counted for nearly 25,000 pairs. These shoes were athletic-style postal shoes. The second category was the 2600 series, which accounted for over 140,000 pairs. These shoes were steel-toed shoes, intended for use in industrial settings where OSHA regulations required protective footwear. The final category was the 2800 and 2900 series of non-steel-toe shoes, of which perhaps 70,000 pairs were sold. A number of models that Sylvania produced for Knapp are not implicated in this litigation.

Quality control problems with shoes in these three lines appeared almost immediately and continued throughout the history of the two companies’ relationship. In mid-1987, Knapp found that the toe bumpers of style 2600 were improperly bonded to the shoe and could be peeled off; these shoes were returned to Sylvania for repair before being shipped to Knapp’s customers. A further problem — this time with sole adhesion— appeared soon afterwards, affecting white shoes in the 2600 and 2800 lines. Sylvania, on the advice of its cement company, had in late 1987 added white pigment to the cement for cosmetic reasons, and this seemed to affect the bond. This difficulty led to the recall of thousands of shoes in early 1988.

By summer 1988, the sole separation problems had spread to black shoes. In a letter sent by John Sprague, the individual at Knapp charged with quality control and product development, to Colin Elliot, a vice president at Sylvania, Sprague wrote that the problems reported with the black shoes “scare[ ] the hell out of us” and reported also that a “[r]ash of telephone calls” had complained of sole separations on style 1251. Following these complaints, a number of shoes were sent to the Footwear Institute of America for pull testing. 1 These tests indicated that some sole adhesion problems were caused by improper manufacture.

Esser later testified that he concluded at the time that the problems were minimal and he authorized continued purchases from Syl-vania. However, both Sprague and Esser remained in almost daily contact with Elliot in an effort to correct the defects. In addition, Knapp began to place a legend on the bottom of some of its purchase orders, “ORDER PENDING CORRECTION OF SOLE SEPARATION PROBLEM.”

Throughout this period, the evidence indicates that Sylvania and Knapp worked together to attempt to solve the problems that were affecting the shoes. Various design changes were suggested by Knapp, and Syl-vania implemented many of them. The suggestions included adding toe bumpers to models that lacked them and substituting a polyurethane midsole for the EVA midsole in certain models. In addition, Sylvania changed cement companies in mid-1988.

In 1989, the relations between the two companies began to deteriorate. Sylvania blames this deterioration on Knapp’s then-parlous financial state. Knapp established a cash committee in February 1989 (of which John Esser was a member); Richard Ned-der, Knapp’s president, was replaced by Joel Murray in April. Knapp fell behind on its account with Sylvania, prompting Sylvania president Robert Pearlstein to send letters of complaint in the summer of 1989.

Knapp insists that defects in Sylvania shoes were jeopardizing some of its most important accounts and it offered evidence that failures in models 2810 and 2930 were of particular concern. In the summer of 1989 Knapp attempted to return 1000 pairs of 2810’s that were produced with the EVA midsole; Sylvania refused to accept the returns. At trial, Sylvania offered evidence, credited by the magistrate judge, that many of these 2810 and 2930 shoes were not manufactured by Sylvania, but were instead imported by Knapp from Taiwan. There was also evidence that sole adhesion problems affected some models that were never produced by Sylvania.

In an effort to work out a payment schedule, Pearlstein met with Murray and others at Knapp twice in the summer and fall of 1989. At the latter meeting on October 17, *195 Knapp sought to demonstrate that there were quality problems with Sylvania shoes by twisting the soles and uppers apart manually. This was done, but the ease, and the significance, of the demonstration were disputed. At that meeting, Knapp agreed to pay $40,000 for every $35,000 of product sent by Sylvania, the extra $5,000 being part of Knapp’s attempt to repay earlier amounts owed to Sylvania.

Records of both companies show that shipments continued in October and November 1989. Payments were also made by Knapp against its outstanding balance in January 1990. On December 1, 1989, Dick Sebastiao joined Knapp as executive vice president with the understanding that he would become president in February 1990. By the end of 1989, Sprague had been fired and Esser, who had been on the board of directors of the company, also had left.

A final shipment of Sylvania shoes was delivered in February 1990, after Knapp made an advance payment. James Crabtree, a Knapp employee, testified that when the shoes were inspected, he was able to pull them apart with his bare hands; he alerted Sebastiao, who called Pearlstein at Sylvania. Pearlstein asked for a sample to inspect, and disputed Crabtree’s findings. The magistrate judge later found Crabtree’s testimony incredible and credited Pearlstein’s assertion that the shoes in the February 1990 shipment were not defective, although two Sylva-nia employees — Elliot and John Cartwright (Sylvania’s manufacturing supervisor) — admitted that they had been able manually to separate the soles on some of the shoes.

Crabtree further testified that he then began to test Knapp’s existing inventory of Sylvania shoes and found that it was “95% defective.” Again, this testimony was not credited by the magistrate judge. Knapp also performed various tests on the shoes from the February shipment and on shoes in inventory after the start of litigation; its evidence at trial was that these tests consistently found problems "with the bonding of the shoes. This evidence was also not credited by the magistrate judge.

II. PRIOR PROCEEDINGS

On April 10,1990, Knapp filed this diversity action against Sylvania under Massachusetts law for breach of contract (count 1), breach of express warranty and implied warranties of merchantability and fitness for a particular purpose (counts 2-4), breach of the duty of good faith and fair dealing (count 5), fraud and negligent misrepresentation (counts 6 and 7), and violation of Mass.Gen.L. ch. 93A (count 8). 2 Sylvania also counterclaimed, seeking $277,000 for unpaid bills, plus multiple damages and attorneys fees under Chapter 93A.

Both parties consented to proceed before a magistrate judge without a jury, and the magistrate judge bifurcated the trial into a liability phase and a damages phase. Evidence in the liability phase was completed on January 31, 1991, after nine days of testimony. That same day, the magistrate judge entered a four-page memorandum and order that devoted one paragraph each to five of Knapp’s eight counts, without discussing Syl-vania’s counterclaims.

In this decision, the magistrate judge ruled that Sylvania had breached its warranties only insofar as Knapp had shown, or could show, that shoes delivered to it were defective. The magistrate judge also found that Knapp had failed to prove fraud, negligent misrepresentation or — “except to the extent that plaintiff has shown,, or can show,” a refusal by Sylvania to credit returned defective shoes — breach of the duty of good faith and fair dealing. As for Knapp’s chapter 93A claim, the decision said that Sylvania had not been shown to have engaged in unscrupulous conduct; it noted, but did not decide; the question whether some payments might still be due Knapp under chapter 93A based on a regulation of the state’s attorney general.

In May 1991, prior to the damages phase of the trial, Knapp learned that Sylvania was *196 going out of business and liquidating its assets. Fearful that Sylvania would soon be judgment-proof, Knapp obtained a temporary restraining order precluding Sylvania from dissipating assets in the amount of $3,775,-657.22 — the amount of damages that Knapp hoped to prove in the damage phase of the trial. The magistrate judge modified this order on June 10, 1991, converting it into a preliminary injunction and amending it to allow Sylvania to make limited payments to its creditors and lawyers.

The damages phase of the trial took place over five days in June 1991; at Sylvania’s behest, an additional day of evidence was heard on November 25,1991. Proposed findings of fact and conclusions of law were filed by the parties in March 1992. Then, in March 1993, the magistrate judge issued an order proposing to certify certain questions to the Massachusetts Supreme Judicial Court. Both sides opposed certification, but on April 8, 1993, the magistrate judge certified two questions, both relating to the possible application of Chapter 93A to “a simple breach of warranty.”

The magistrate judge prefaced the certified questions with a ten-page statement. In it, he first repeated the rulings on the five counts contained in the January 31, 1991, order. Then, he determined for the first time that the parties had agreed by express negotiations, express understandings and express course of dealings that in the event of defects, “Knapp’s remedy, and sole remedy, would be the replacement of [or credit for] those shoes shown to be defective and returned' — nothing more, and nothing less” (footnote omitted; brackets in the original). The certification also said, in a footnote, that less than three percent of the Sylvania shoes delivered to Knapp were defective.

On Sylvania’s motion, the magistrate judge dissolved the preliminary injunction on May 5,1993, concluding that Knapp now had little hope of a substantial recovery. On Knapp’s appeal, this court stayed and then vacated the order dissolving the preliminary injunction. Knapp Shoes, Inc. v. Sylvania Shoe Manufacturing Corp., 15 F.3d 1222 (1st Cir.1994). We held that Sylvania had waived the affirmative defense of limitation of remedies by failing to raise it in a timely fashion. Since the issue of limitation of remedies had never been litigated by the parties, the waiver could not be avoided by amending the pleadings to conform to the proof. Cf. Fed.R.Civ.P. 15(b). Accordingly, we said that “the limitation of remedies defense is out of the case and cannot support the order vacating the injunction.” Id. at 1227.

We also rejected Sylvania’s alternative argument that the termination of the injunction could be supported by the magistrate judge’s footnote finding, in the certification, that the percentage of defects was very small. Our opinion pointed out that the magistrate judge had not set forth findings or analysis to support this conclusion as to the quantity of defects, so we were “unable to make a reasoned judgment whether, on this critical issue of defects, the magistrate judge’s finding was or was not ‘clearly erroneous’” under Fed.R.Civ.P. 52(a). Id. at 1228-29.

On October 13, 1994, the Supreme Judicial Court answered the certified questions that had been submitted by the magistrate judge. Knapp Shoes, Inc. v. Sylvania Shoe Manufacturing Corp., 418 Mass. 737, 640 N.E.2d 1101 (1994). The Court held that 940 Code Mass.Regs. § 3.08, which provides in part that “[i]t shall be an unfair and deceptive act or practice to fail to perform or fulfill any promises or obligations arising under a warranty,” was not meant “to encompass a contract dispute between businessmen based on a breach of the implied warranty of merchantability.” Id. 640 N.E.2d at 1105.

Thus informed, the magistrate judge issued his final decision on March 1, 1995. With respect to counts 3 and 4 (breach of warranties of merchantability and fitness), he again held that Sylvania had breached its warranties to the extent that particular shoes were defective for whatever reason. On count 1 (breach of contract), the magistrate judge said again that there was no violation beyond the breaches of warranty covered by counts 3 and 4. He also reaffirmed that Knapp had failed to establish its claims in count 2 (breach of express warranties), count 5 (duty of good faith and fair dealing), count 6 (common law fraud), count 7 (common law *197 negligent misrepresentation) and count 8 (Mass.Gen.L. ch. 93A violation).

On Sylvania’s counterclaims, the magistrate judge found that Knapp was liable for the outstanding balance due for shoes that Sylvania had delivered to Knapp. The magistrate judge found that Sylvania was not entitled to recover the contract price of additional shoes it had manufactured for Knapp but not yet delivered; the reason was that Sylvania had failed to make a reasonable effort to resell the shoes. He also rejected Sylvania’s own chapter 93A claim against Knapp. Sylvania does not challenge these rulings on appeal.

In computing damages, the magistrate judge held that Knapp should receive credit only for defective shoes still in its inventory — which he determined to be between 3 and 4 percent of the total — and for specified incidental damages. He ruled that Knapp had no right to revoke acceptance of any non-defective goods in inventory, nor to reject the final shipment of shoes sent in February 1990, nor to recover for lost profits. In the final tally, Knapp was awarded damages of $160,062.74, and Sylvania was awarded damages of $223,626.47. Finally, the preliminary injunction was dissolved. We stayed the judgment pending disposition of this appeal.

III. LAW OF THE CASE AND STANDARD OF REVIEW

Our review of the facts found by the trial judge is normally deferential; findings are not to be set aside unless clearly erroneous, with “due regard ... given to the. opportunity of the trial court to judge of the credibility of the witnesses.” Fed.R.Civ.P. 52(a); see Williams v. Poulos, 11 F.3d 271, 278 (1st Cir.1993). Knapp argues that our review should be less deferential in this case, drawing attention to paragraph 49 of the magistrate judge’s findings of fact and its accompanying footnote.

[Paragraph 49] With respect to all lines of shoes which Knapp requested be manufactured by Sylvania, one finding is unmistakenably [sic] clear: By their express negotiations, by their express understandings, by their demonstrated conduct, ’ and by their express course of dealings, Sylvania promised Knapp that, in the event that shoes were defectively manufactured, Knapp’s remedy, and sole remedy, would be the replacement of [or credit for] those shoes shown to be defective and returned — nothing more, and nothing less.

[Footnote] This finding is not made and reported to suggest that Knapp waived its rights under Sections 2-601, 2-608, and 2-609 of the Uniform Commercial Code (M.G.L., eh. 106, §§ 2-601, 2-608 and 2-609). Although it was and is clear to this court — then and now — that that was the sole understanding of the parties in terms of remedies and relief, and that it was clear (then and now) to all' parties that the defense of waiver loomed throughout, and that plaintiff could not — then or now— establish any legal prejudice (that is, plaintiff could not — then or now — make any showing that the case would have been tried differently), our Court of Appeals, in the context of an interlocutory appeal relating to the dissolution of an injunction, has concluded otherwise.

Nevertheless, this finding and conclusion bear heavily on the credibility of the witnesses called by Knapp, and other issues to be determined herein.

Knapp argues that this discussion violates the law of the case by contradicting our holding in Knapp that Sylvania had waived any such limitation of remedies defense, that the issue of limitation of remedies had not been litigated at trial, and that it was now “out of the case.” 15 F.3d at 1227. Knapp further insists that this error by the magistrate judge tainted his other factual findings, requiring us to subject these findings to heightened scrutiny. Paragraph 49 has certainly complicated matters on this appeal, but we conclude that Knapp has somewhat exaggerated its import and consequences.

The law of the case doctrine has more than one dimension and certain complexities, but as applied to the problem before us, the doctrine provides that when a court of appeals makes a ruling of law, whether on appeal of a final judgment or in an interlocutory appeal, that ruling becomes the law of *198 the case in any subsequent proceedings in the trial court. Elias v. Ford Motor Co., 734 F.2d 463 (1st Cir.1984). Such a ruling is “[a] mandate [that] is completely controlling as to all matters before the appellate court and disposed of by its decree.” Id. at 465.

We agree with Sylvania that the magistrate judge did not contradict our legal ruling in Knapp that the defense of limitation of remedies was waived and now out of the ease. He did not rest any of his own legal rulings on the proposition that Knapp had limited its remedies by contract to credit for returned shoes; indeed, he awarded Knapp damages for defective shoes still in its possession (although he found the number of such shoes to be very small). To that extent, Knapp’s law-of-the-case claim is something of a diversion.

At the same time, on an intermediate proposition of fact a direct conflict exists between our earlier opinion and the most recent decision of the magistrate judge. The magistrate judge repeats in the footnote to paragraph 49, quoted above in text, his earlier conclusion that Knapp and Sylvania did agree in fact to limit Knapp’s remedies; and he recognizes that this court on the prior appeal “concluded otherwise.” Our actual conclusion was slightly narrower — we said that the parties had not purported to litigate the issue and we could find no evidence of such an agreement — but the fact remains that the magistrate judge has reasserted his view that such agreement has been proved.

All this might matter little if the magistrate judge’s disagreement with us played no role in his decision, but he goes on to say that “this finding and conclusion bears heavily on the credibility of the witnesses called by Knapp, and other issues to be determined herein.” No resort to law of the case doctrine is required for us to determine that his “finding and conclusion” that such an agreement existed limiting remedies is clearly erroneous. The very same defect — the lack of evidence to show such an agreement — identified in our earlier decision remains, utterly unaltered.

We explained in our earlier opinion why we were not persuaded of such an agreement by the magistrate judge’s reliance on Esser’s testimony that a return remedy existed— testimony that did not even purport to address the exclusivity of the remedy — and also why we saw no course of dealing by the parties that could prove such a limitation. 15 F.3d at 1226-27. No new evidence was taken on remand to prove this limitation; no additional support for it is mustered by the magistrate judge out of the pre-existing record.

About the best we can do in this disturbing situation is to defer to the usual extent as to those findings of the magistrate judge that we are confident have not been infected by his belief in the supposed agreement limiting remedies; and, on all other findings, to consider them in the knowledge that the magistrate judge has credited or discredited certain witnesses based (at least in part) on a premise that we have already held to be mistaken. Sorting out findings in this way, and deciding how to treat infected findings, is something of a task but better than an outright remand on all issues to a new judge.

IV. THE MERITS

Our analysis of the merits is divided in three parts. Knapp raises some rather halfhearted objections to the magistrate judge’s rulings that reject most of its counts; we find that these rulings are sustainable on this record. Knapp’s next objections concern the magistrate judge’s rulings on remedies; here we find that the magistrate judge’s analysis of the remedies available to Knapp was correct on one issue (Knapp’s purported revocation of acceptance of all shipments) but flawed on another (Knapp’s rejection of the first February 1990 shipment). Finally, we find that elements of the magistrate judge’s determinations on damages were clearly erroneous but that the necessary corrections can be made on the existing record.

A. Liability

Knapp’s complaint alleged eight causes of action. The magistrate judge ruled that Knapp had failed to persuade on all except for counts 3 and 4, breaches of implied warranties of merchantability and fitness; these warranties were held to be breached only *199 with respect to those shoes that were actually defective. Sylvania does not appeal that limited finding of liability; Knapp appeals from the decision rejecting its other counts, but its arguments are without merit.

Breach of Contract. With respect to count 1, breach of contract, Knapp objects that the magistrate judge in rejecting this count relied upon his footnote 49 finding that the parties had previously agreed to limit remedies. But Knapp’s own count 1 alleged only that Sylvania failed to deliver shoes of promised quality; and it has not shown how this claim exceeds the breach of warranty claims that the magistrate judge allowed under counts 3 and 4. We therefore affirm the magistrate judge’s disposition of count 1.

Breach of Express Warranty. On count 2, the breach of express warranty, Knapp argues that Sylvania warranted that it would produce a defect-free shoe and that the high defect rate caused each entire shipment to be in breach of Sylvania’s express warranty. Knapp points to two letters sent to Knapp by Sylvania in mid-1988, when sole bond problems were affecting shoes in the 2600 line. In the first, Elliot wrote, “We know the industrial consumer wants this product and it is up to us to give it to him or her without defect.” In the second, Elliot wrote “As always, we stand behind our product and fully warrant the product against manufacturing defect.”

If Knapp means to suggest that Sylvania had warranted that each and every shoe in a given shipment would be free of defects and that a shipment could be rejected based on any such defect, that suggestion is not reasonable. At trial witnesses for both parties agreed that no one in the shoe industry expects that any shipment of shoes will be entirely free of defects. Sylvania’s quoted statements can at most be read to mean that it was capable of producing shoes that met industry standards as to the percentage of defects.

Of course, to the extent that Sylvania breached its various commitments, it may be hable — under certain circumstances — to rejection of more than the particular shoes shown to be defective and to consequential damages that exceed the price paid for the particular defective shoes. See IV(B) and (C), below. But so far as- Knapp’s claim depends on the notion that Sylvania promised no defects at all, we think no such warranty was made.

Breach of Duty of Good Faith and Fair,Dealing. Count 5 alleged that Sylvania breached its duty of good faith and fair dealing. 3 In his January 31,1991, memorandum, the magistrate judge found that Knapp had failed to carry its burden on this claim, ruling that “the defendant acted in the utmost good faith in an attempt to correct a problem not reasonably foreseen by any of the parties to the relationship.” Knapp insists that Sylvania’s failure to correct its manufacturing processes, despite its assurances to Knapp that the quality of its shoes would improve, breached its duty.

Whether conduct violates the duty of good faith and fair dealing is necessarily a fact-specific inquiry, see E. Allan Farnsworth, Contracts § 7.17 (2d ed. 1990), and the magistrate judge’s finding here is not clearly erroneous. There was ample untainted record evidence that Sylvania strove to improve the quality of the shoes it delivered; it changed its cement suppliers and implemented various design changes suggested by Knapp. There is no- evidence that Sylvania’s failure to improve its manufacturing process was motivated by bad faith rather than by a simple failure to root out and remedy all of the problems.

Fraud and Negligent Misrepresentation. Counts 6 and 7 charged fraudulent and negligent misrepresentation, claims that in Massachusetts require a false representation of a material fact, knowledge of falsity or carelessness on the part of the *200 defendant, and reasonable reliance by the plaintiff. 4

The magistrate judge found, in his memorandum of January 31, 1991, that Knapp’s misrepresentation claims failed because Syl-vania had done what it said it would do — use its best efforts to produce shoes free of defects. On appeal, Knapp insists that Sylva-nia had repeatedly and falsely assured Knapp that it would correct any problems with the shoes, and that Knapp reasonably relied on these representations in continuing its purchases.

We need not decide whether such reliance would be reasonable, but cf. IV(C) below, because we agree that no false representation has been shown. The evidence at trial indicated that Knapp and Sylvania remained in regular contact regarding proposed design and manufacturing changes, and Sylvania never misrepresented the changes it proposed to carry out. In this context, Sylva-nia’s conclusory statements that its quality would improve — and Knapp calls our attention to no more specific alleged misstatement — are no more than an “expectation, estimate, opinion, or judgment.” See Powell v. Rasmussen, 355 Mass. 117, 243 N.E.2d 167, 168 (1969).

Chapter 93A. On count 8— Knapp’s ch. 93A claim — we again affirm the magistrate judge. For Knapp to prevail, it must prove that Sylvania’s conduct included an “unfair or deceptive act,” a phrase that the Massachusetts courts read as requiring a showing of “rascality;” the showing is especially difficult where the case involves arm’s-length transactions between sophisticated business entities. Anthony’s Pier Four, Inc. v. HBC Assocs., 411 Mass. 451, 583 N.E.2d 806, 821 (1991).

The crux of Knapp’s argument is that Syl-vania was aware that there were problems with its manufacturing process and failed to make the needed changes. But as the magistrate judge held, there was no evidence at trial that Sylvania’s failure to improve the quality of its shoes was deliberate or willful; on the contrary, it made repeated efforts on its own and in consultation with Knapp to correct perceived problems. Knapp’s claim for multiple damages and attorneys fees was therefore properly denied.

B. Revocation and Rejection Remedies

Even if liability is based only on counts 3 and 4, Knapp argues on appeal that the magistrate judge nevertheless erred in determining the remedies available to it. It says that he improperly denied to Knapp remedies of “revocation of acceptance” (with respect to all prior shipments) and “rejection” (with respect to the shipment of February 1990) and that these errors led him into a further error, namely, to direct his damage assessment only to those shoes actually shown to be defective.

Revocation of Acceptance as to All Shoes. Knapp first argues that it is entitled to revoke acceptance of all inventory purchased from Sylvania — including non-defective shoes — because the rate of defects was so high as to make it impossible for Knapp to sell shoes from that inventory. The magistrate judge rejected this remedy because he found a very low rate of defects. Although this finding is not adequately supported, see IV(C) below, we agree with the result because Knapp failed to make an effective revocation of its acceptance under UCC § 2-608.

When a buyer “accepts” goods, as defined in UCC § 2-606, the buyer forfeits the right to “reject” the tender. UCC § 2-607. However, a buyer may be able to “revoke acceptance” under UCC § 2-608 as follows:

(1) The buyer may revoke his acceptance of a lot or commercial unit whose nonconformity substantially impairs its value to him if he has accepted it

(a) on the reasonable assumption that its non-conformity would be cured and it has not been seasonably cured; or

(b) without discovery of such non-conformity if his acceptance was reasonably induced either by the difficulty of discovery before acceptance or by the seller’s assurances.

*201 (2) Revocation of acceptance must occur within a reasonable time after the buyer discovers or should have discovered the ground for it and before any substantial change in condition of the goods which is not caused by their own defects. It is not effective until the buyer notifies the seller of it.

The buyer who revokes has the same rights and duties as if the buyer had rejected the goods. UCC § 2-608.

Knapp argues that this case falls squarely within section 2 — 608(l)(b). It says that the sole-bond problems were hard to discover because they could not be detected by visible inspection-and that it reasonably relied upon Sylvania’s assurances that it had corrected the bond problem. Knapp cites to S & R Metals, Inc. v. C. Itoh & Co. (America), 859 F.2d 814 (9th Cir.1988), as authority for the view that a buyer can revoke acceptance when the prior acceptance was made without knowledge of the defect and the defect was “latent and difficult to discover.”

The difficulty with Knapp’s position is that by mid-1988 it had ample knowledge that customers were complaining about separation, and its own experience confirmed that Sylvania was not successfully solving the underlying problems. Knapp itself points to pull-tests done in 1988 which, it argues, prove that Sylvania’s manufacturing process was flawed. Yet not until spring 1990 — the precise date is disputed — did Knapp purport to revoke acceptance for all shoes delivered to it over the entire period.

This is not a revocation occurring within “a reasonable time” after the buyer discovered the ground for rejection, and the contrast between the present facts and those of S & R Metals underscores the point. In that case, the buyer had no reason to doubt the quality of the steel until complaints were received; advance testing would have been expensive and destroyed the valuable product tested; and the revocation of acceptance was made within nine days after the defect had been discovered and confirmed. S & R Metals, 859 F.2d at 817. Nothing in this decision, or any other cited by Knapp, 5 suggests that a buyer can accept deliveries of a vast number of items over a period of a year and a half and then suddenly revoke the acceptance of all of them based on defects whose presence was known or suspected during the entire period.

Rejection of the February 1990 Shipment. A much narrower and stronger claim by Knapp is that it properly rejected the single shipment of shoes received in February 1990. Section 2-601 of the UCC provides that if goods or tender fail to “conform” to the parties’ contract, the buyer may reject the whole delivery; 6 but this rejection must be within a reasonable time, and the buyer must seasonably notify the seller of the rejection. UCC § 2-602(1). The magistrate judge found that Knapp failed to prove that the February shipment was nonconforming, and further found that Knapp’s rejection was untimely.

Knapp insists that the shipment was nonconforming because all of the shoes in the shipment were defective inasmuch as they could be pulled apart manually. Crabtree, Knapp’s key witness on this issue, so testified at trial. The magistrate judge made clear that he did not accept Crabtree’s testimony, but this appraisal may well have been affected by the magistrate judge’s mistaken finding on limitation of remedies: Crabtree also testified to the fact that the procedure of return of customer defects was established only in mid-1988 and that before that time customer returns were discarded, a fact that the magistrate judge found surprising in *202 view of the “clear understanding between the parties.”

Knapp did carry out a prompt inspection of the shoes that arrived in the February 1990 shipment, complained immediately to Pearl-stein that the shoes were 100 percent defective, and sent him a case for his own inspection. When the formal “rejection” occurred is disputed — the magistrate judge found that it did not occur until May — but Knapp’s actions were certainly consistent with its claim of substantial defects and represented steps toward rejection. It immediately placed the seller on notice that the defects were pervasive and began to negotiate the seller’s response.

The magistrate judge chose to credit fully the testimony of Pearlstein, Sylvania’s president, to the effect that no shoes from the sample case sent by Knapp to Sylvania from the February shipment separated. But quite apart from other Knapp witnesses who supported Crabtree, other Sylvania employees — Elliott and Cartwright — both agreed that at least some of the shoes could be pulled apart by hand. Thus we find it hard to accept the magistrate judge’s finding that “plaintiff has failed to establish by a preponderance of the evidence that any of the shoes — much less the whole of the lot — failed to conform.” 7

This court finds clear error only where, “on the whole of the record, we form a strong unyielding belief that a mistake has been made.” Cumpiano v. Banco Santander P.R., 902 F.2d 148, 152 (1st Cir.1990) (citations omitted). In this instance, we conclude that the magistrate judge was clearly in error in finding that the February 1990 shipment was free of defects and conformed to industry standard. The question whether Knapp gave prompt notice is more difficult, but we need not resolve the issue because whether Knapp rightfully rejected the nonconforming shipment has no practical impact on the damages to which it is entitled. With the magistrate judge’s finding on defects set to one side, Knapp’s ordinary damages as to the February shipment produce essentially the same amount as it would receive under the “rejection” remedy. See IV(C) below.

C. Damages

Knapp’s damage claims were based upon its allegation that Sylvania shoes suffered from a very substantial rate of defects. In his final damages calculation, the magistrate judge found that the rate of defects in the inventory shoes involved in this litigation was less than four percent. Knapp argues on appeal that the magistrate judge’s defect-rate finding was clearly erroneous, and insists that the evidence established a defect rate of at least 40 percent for the shoes remaining in inventory at Knapp. We think that the magistrate judge’s finding is clearly erroneous and that on this record the 40 percent figure is the only alternative choice.

Knapp’s allegations of substantial defect rates in the shoes remaining in inventory were supported by detailed testimony by Crabtree; by corroborating testimony of other Knapp personnel; by evidence of substantial customer returns, dissatisfaction and can-celled relationships to which the customers testified; by testimony that the number of defective shoes held by Knapp plus the number of prior returns acknowledged by Sylva-nia was nearly twice the magistrate’s 3-to^t percent finding; by evidence that both Knapp and customers threw away additional defective shoes; and, finally, by a fairly detailed sampling study that appeared to establish a defect rate of at least 41.7 percent. 8

In adopting the 3-to-4 percent defect rate figure, the magistrate judge accepted the *203 testimony of Esser and Sprague (both disaffected former Knapp employees) and of Syl-vania personnel, to the effect that there was never a substantial problem with Sylvania shoes. He disregarded Crabtree as a liar; and he dismissed customer testimony as not necessarily relating to Sylvania • shoes but rather to shoes manufactured in Taiwan. He found the pull tests inconclusive, insisting that Knapp had failed to present evidence that a pull-test failure at under 20 pounds showed that a shoe was defective. He refused to believe Knapp’s accounting of the number of defective shoes in inventory.

Under the Cumpiano, standard, we reject as clearly erroneous the magistrate judge’s assessment of the percentage of defects in the Sylvania shoes. We have already explained why his related appraisal of testimony regarding quality of the February 1990 shipment is unpersuasive, see TV(B), and this is equally so on the broader question of the remaining inventory. Knapp’s claims of defects came not only from Knapp but from customers disinterested in the litigation, from the presence in inventory of defective shoes actually returned to Knapp, and from the pull-tests conducted by independent experts — the last of which sampled the entire remaining Knapp inventory in the contested models.

The notion that any of these defective shoes came from foreign sources rests on doubtful evidence; but, given the poor quality of the February 1990 shipment admittedly from Sylvania, it is obvious that foreign sources do not explain away the problem. From the testimony on pull-tests, it appears that the 20-pound figure that Knapp asked the expert to employ was not only plausible but conservative. Knapp’s accounting of the number of defective shoes in its inventory was essentially uncontradicted. In sum, the evidence is overwhelming that the defect rate was considerably higher than the rate adopted by the magistrate judge.

Each side had its chance to present evidence and neither side is entitled to introduce further evidence on this issue. On the present record the stark choice is between Sylvania’s claim that defects were no greater than normal and Knapp’s evidence to the contrary. The Knapp evidence was that .it had 6,045 known defective pairs in inventory, comprising shoes returned by customers, the 1,422 pairs torn-apart from the February 1990 shipment and other shoes pulled apart during inspection of inventory, and — in addition — that it had in inventory 21,010 new shoes of the models involved in this litigation, at least 41.7 percent of which the statistical sampling showed to be defective.

The evidence as to raw numbers in inventory was essentially uncontradicted even though the magistrate judge declined to accept the numbers. As for the statistical study, the 20-pound pull-test was (as already noted) a conservative standard based on the evidence; and the 41.7 percent figure even more so. Sylvania has chosen not to contest its liability for defective shoes no matter how long in inventory. “Where, as here, the record is sufficiently developed that we can apply the law to the facts before us ... that route is available to us.” Lipsett v. Blanco, 975 F.2d 934, 943 (1st Cir.1992). We take it here and find that Knapp has proved the raw figures and defect rate just discussed, any other conclusion on this record being clearly erroneous.

As for the cost of the shoes, we agree with the magistrate judge that the most accurate figure is the average price of the shoes purchased by Knapp, rather than Knapp’s “Fifo average cost.” 9 Using the magistrate judge’s price figures for each model and applying that price to the larger number of defective shoes that we have found to be present in Knapp’s inventory (14,806), we have calculated Knapp’s damages for defective shoes as $338,138.31. The calculations, by model number, are set forth in Appendix A.

Knapp argues that regardless of the precise percentage of defective shoes in its inventory, all of the shoes are useless to it *204 because the number of defects is too high to allow Knapp to deliver the shoes to customers without fear that they will fall apart on their feet. Knapp is essentially attempting to invoke through the back door the revocation of acceptance remedy that Knapp failed ■to invoke in a timely manner. Whatever decrease Knapp has suffered in value of the non-defective shoes, Knapp inflicted this decrease on itself when it accepted the deliveries and failed to revoke that acceptance within a reasonable time.

Knapp also claimed a detailed list of incidental damages, some of which were granted by the magistrate judge and are not challenged by Sylvania on appeal. Knapp insists that it is also entitled to reimbursement for a credit it issued to Federal Express and for storage costs beyond those granted by the magistrate judge. The magistrate judge found that Knapp had failed to prove that it had not received credit from Sylvania or replacements for the defective shoes for which Federal Express demanded and received credit.

We agree that Knapp failed to prove that this credit represented actual damages. 10 As to the storage charges, Knapp will be credited only for the amount attributable to the actually defective shoes, as the magistrate judge held, but that amount is increased to $4,146 to account for the enlarged number of defective shoes determined on this appeal. The result is achieved by using the magistrate judge’s own formula set forth in his March 1, 1995, decision.

Knapp further claims that it is entitled to lost profits from the shoes in inventory it could not sell. To recover lost profits, the plaintiff must show by a preponderance of the evidence that the actionable breach caused the loss and that the loss was foreseeable and calculable with reasonable certainty. Matsushita Electric Corp. v. Sonus Corp., 362 Mass. 246, 284 N.E.2d 880, 890 (1972). For the pre-February 1990 shipments, Knapp’s claim is hopeless. Knapp did not show that it had an urgent need for any of that inventory, nor justify its failure to obtain alternative sources given its knowledge of persisting problems. See UCC § 2-715(2)(a) (consequential damages include only those losses “which could not reasonably be prevented by cover or otherwise”).

With respect to the February 1990 shipment, the issue is closer, because Knapp established at trial that it had an urgent need for that inventory; the quality of that shipment was even lower than earlier shipments; and Sebastiao testified that Knapp lost orders because it could not deliver these shoes. On the other hand, Knapp was on notice that Sylvania shoes suffered from a high rate of defects, so its failure to switch to other, more reliable suppliers at an earlier date is hard to justify. Knapp knew or should have known that it might well get defective shoes in the February shipment, and it chose to take that risk. We affirm the magistrate judge’s finding that lost profits are not appropriately awarded.

Knapp also presses on appeal its broader claim for consequential damages. It presented at trial expert testimony that because of the defective shoes Knapp lost future profits of $2,895,326 that it would have enjoyed on other sales to customers who left Knapp out of dissatisfaction with its product. Its evidence certainly showed that its relationships with important customers were irretrievably damaged. But it is equally clear that Knapp could have minimized the damage through reasonable diligence by increasing its inspections and, to the extent needed, finding alternative suppliers.

Although Sylvania was at fault for supplying defective goods and must reimburse Knapp for those goods proved to be defective, Knapp cannot recover for any larger harm done to its own reputation and customer relations. The general principle is well settled that a party cannot recover for harms that its own reasonable precautions *205 would have avoided. Columbia Novelty Co. v. Leslie Dawn, Inc., 6 U.C.C.Rep.Serv. (Callaghan) 679, 679, 1969 WL 11074 (N.Y.App. Term 1969); UCC § 2-715(2)(a). To permit such a recovery on this record would simply reward Knapp for its own lack of attention, poor quality control, and bad judgment.

CONCLUSION

In summary, Knapp is entitled to damages as follows:

Cost of Defective Inventory: $338,138.31

Storage Costs: $ 4,146.00

Unpaid Credits: $ 92,472.97

Credit to Miami Hilton: $ 9,250.00

Refund to Marriott: $ 3,194.54

Price concessions to Federal

Express: $ 11,096.25

Replacement shoes supplied

to Hilton: $ 6,401.20

Increased inspection costs: $ 4,167.30

Freight charges: $ 6,877.60

Travel expenses: $ 1,793.92

Total $477,538.09

The first two entries are adjusted to reflect the corrections explained in the opinion; the others are as determined by the magistrate judge in rulings that Sylvania has not appealed. The award against Knapp in favor of Sylvania for unpaid bills, in the amount of $223,626.47, has also not been challenged on appeal and so stands as previously entered.

On remand, the judgment should be modified to award $477,538.09 to Knapp and $223,626.47 to Sylvania, so that the net award is now in Knapp’s favor in the amount of $253,911.62. In addition, the magistrate judge’s order terminating the injunction against Sylvania, as entered on May 31,1991 and amended on June 10, 1991, is vacated; the magistrate judge is free to reduce the amount embargoed to the net award plus anticipated costs and interest.

It is so ordered.

Notes

. Pull tests, standard in the industry, allow for measurement of the strength of shoe construction. In addition to measuring the pounds of pressure that can be applied to a particular area of the shoe before it will fail, the tests indicate whether the failure was caused by material failure — a tearing of the upper or of the sole — or by a bond failure.

. Chapter 93A outlaws "[u]nfair methods of competition and unfair or deceptive acts or practices in the conduct of any trade or commerce," and permits awards of multiple damages and attorneys’ fees.

. In Massachusetts, ”‘[e]very contract implies good faith and fair dealing between the parties to it.” Warner Ins. Co. v. Commissioner of Ins., 406 Mass. 354, 548 N.E.2d 188, 193 n. 9 (1990) (quoting Kerrigan v. Boston, 361 Mass. 24, 278 N.E.2d 387, 393 (1972)). Similarly, Mass.Gen.L. ch. 106 ("UCC”) § 1-203 imposes an “obligation of good faith."

. VMark Software, Inc. v. EMC Corp., 37 Mass.App.Ct. 610, 642 N.E.2d 587, 593 n. 8 (1994); Zimmerman v. Kent, 31 Mass.App.Ct. 72, 575 N.E.2d 70, 77 (1991).

. In Fortin v. Ox-Bow Marina, Inc., 408 Mass. 310, 557 N.E.2d 1157 (1990), also cited by Knapp, the court allowed the buyer to revoke acceptance of a boat delivered four months earlier when the seller had provided repeated assurances that it would cure the identified defects in the boat. The case involved a different problem — a promise to repair a specific item already delivered — and is governed by a different provision (section 2-608(1)(a)).

. To permit rejection of the entire shipment for nonconformity, Knapp had to show not just that there were defective shoes but that the defect rate was higher than agreed upon or, lacking specific agreement, than the standard in the industry. See Agoos Kid. Co. v. Blumenthal Import Corp., 282 Mass. 1, 184 N.E. 279, 281 (1932).

. The magistrate judge suggested that the "hand” pull test is not standard in the industry and that evidence that shoes could be pulled apart by hand did not necessarily prove them defective. Common sense, buttressed by ample evidence at trial, confirms that an industrial work shoe is defective where it can readily be pulled apart by hand.

. A biostatistician analyzed the results of pull-tests performed on randomly selected pairs of unused shoes in inventory. Using a 20-pound figure for the pull-tests, 60 percent of the shoes failed. Given the sample size tested and the number of shoes in inventory, the expert concluded that "we have a 95 percent level of confidence that the actual proportion of defective pairs in the inventory ... is at least 41.7 percent."

. Even if we were to accept (as the magistrate judge did not) that Knapp maintained a strict Fife inventory procedure, some of the new shoes in Knapp's Brockton warehouse were returned from field warehouses; some may have been from early shipments, and the average price figure - therefore appears to us to be the better measure.

. It is apparently undisputed that Knapp gave credit to its customer Federal Express to resolve complaints about defective shoes. Nevertheless, the evidence left it uncertain whether some of the shoes reflected in this figure had already been returned by Knapp to Sylvania or were among shoes in Knapp's inventory, damages for which are already reflected in the calculations in Appendix A. The burden was on Knapp to show actual damages without double counting, and it failed to do so.

*205 Appendix A

x 41.7% Total Ave. Cost Cost Style Detects 5 <

1244 150 199 8233 18.20 4.240.60

1245 63 189 7142 23.75 3,372.50

1251 493 509 212 705 21.65 15.263.25

279 26.25 7.323.75

1,233 24.25 29.900.25 2600 914 CO

72 25.75 1,854.00 2601 47

714 27.90 19.920.60 2670 464 W LO

734 27.90 20.478.60 2675 566 M ^

1,531 21.91 33,544.21 2810 514 ^ c<f

668 18.75 12,525.00 2814 53 ^ r-T

598 22.10 13,215.80 2815 51 Ü1 CO t-T

62 22.10 1,370.20 2816 26

354 27.40 9.699.60 2840 125 no

777 27.30 21,212.10 2845 136 10

149 18.75 2.793.75 2916 39 CO

754 19.90 15,004.60 2935 211 CO r-T

17 18.75 318.75 2950 1

_25 18.75 468.75 2955 25

21,010 8,761 14,806 338,138.31 Total 6,045

The first column refers to the Knapp shoe style number; the second column sets out the number of customer returns plus shoes pulled apart by Knapp personnel in Knapp’s inventory; the third column sets out the number of new shoes in inventory; the fourth column gives the number of defective shoes among the new shoes, based on the statistical survey; the fifth column gives the total number of defective shoes in Knapp’s inventory; the sixth column gives the average price for each style; the last column gives the price paid by Knapp for defective shoes still in inventory.