Plaintiff-appellant ITT Corporation (“ITT”) appeals from a judgment of the district court dismissing, after a bench trial, its contract action against LTX Corporation (“LTX”). 732 F.Supp. 1225. This is a diversity ease applying Massachusetts contract law.
I.
The alleged contract on which the present action was brought was made on May 6, 1987, in settlement of a prior lawsuit. That lawsuit resulted from LTX’s refusal to pay for certain specialized electrical equipment called “cable assemblies” ordered by LTX from ITT Cannon, a division of ITT. The cable assemblies — each consisting of six-foot cables containing more than 155 parallel small insulated wires, with associated connectors, back-shells, and strain relief clamps — were meant for use in LTX semiconductor testing equipment. As produced by ITT, the cable assemblies proved to be commercially unusable due to the fact that some of the wires within a cable would break (especially near the connectors) when the cable assembly was subjected to the considerable bending and flexing in commercial use by LTX’s customers. When ITT still had on hand some 479 of the expensive assemblies which it had manufactured specially for LTX, the latter refused to accept further delivery of them, complaining of their lack of functionality, and blaming ITT. Insisting that the products had been built to LTX’s order utilizing Gore cable specified by LTX, and that ITT bore no responsibility for whatever design deficiencies were causing the wires in the cable assembly to fracture when the cable was repeatedly flexed in commercial use, ITT sued LTX, seeking to force LTX to take delivery of, and pay ITT for, its remaining inventory of cable assemblies. That litigation was eventually dismissed by stipulation pursuant to the parties’ letter agreement dated May 6, *12601987. Thereafter, the present dispute arose as to whether that letter agreement required LTX to accept ITT’s remaining inventory of cable assemblies. It is uncontested that design changes specified therein were, or would be, made by ITT (“Revision A” changes), but these did not cure the wire-breakage problem. The present dispute, based on the interpretation of the settlement agreement, carries on the parties’ original disagreement over who should bear responsibility for the continuing failure to solve the breakage problem.
II.
We need not repeat for present purposes the detailed findings of the district court based upon the parties’ stipulations and upon documentary evidence and testimony from the parties’ officers and employees. The disputed settlement agreement of May 6, 1987, is reproduced at the end of this opinion. It was written by ITT’s general counsel at the behest of Mr. Sekeres, an ITT executive, who on the previous day had talked to Mr. Boltrus, a top official of LTX. Both men were trying to resolve ITT’s previous lawsuit against LTX. Both testified that at the conference they, in effect, maintained their conflicting positions, which were, respectively, that ITT denied any responsibility for the malfunctioning of the cable assemblies, while LTX said it would not accept them unless usable. While Sek-eres could remember no discussion of testing, Boltrus testified that they had discussed testing the cable assemblies after they were altered to meet the so-called Revision A criteria. No one claims, however, that any specific test procedures were then determined.
The two men apparently believed they had reached an agreement, because the next day Sekeres submitted the now disputed letter agreement to Boltrus, who signed it for LTX.1 The letter provided, in essence, that LTX would immediately buy back a certain large stock of Gore cable which ITT had purchased for the project,2 and that, later, within the next eight to ten months, on specified dates, LTX would take delivery of the existing 479 cable assemblies and would pay ITT for them at prices stated in the letter. It was also agreed that ITT would modify the cable assemblies to conform to so-called Revision A specifications. These consisted of two changes to the existing cable assemblies proposed by LTX in hopes they would cure the wire-breakage problem: to wit, (1) replacement of the Gore cable outer jacket with a new elastic jacket and (2) addition of a rubber boot extending beyond a certain strain relief mechanism, already used, so as to increase the cable’s bending radius. ITT promised in the letter agreement to make these modifications for $124 per unit, payable by LTX, or else, if LTX desired, LTX could make any modification at its own expense. However, the agreement stated flatly: “Regardless of the alternative selected, LTX shall accept delivery of the cable assemblies and pay for them in accordance with the schedule above.” It was *1261agreed that the pending litigation would be dismissed. See footnote 2, supra.
The letter agreement also provided for three “acceptance test criteria” to which, at the time of shipment, the cable assemblies would have to conform. The third criterion (hereinafter “criterion three”) is the basis of the present dispute. It provides:
At the time of shipment the cable assemblies will conform to the following acceptance test criteria:
s): }¡< j}¡ jfc
100% of cable assemblies will be tested for continuity while subjected to flexing stress. The cable is exercised in a circular pattern while observing 156 indicator lights on a special test fixture. This test insures that no wires are broken and that there are no intermittent conditions.
The parties dispute whether criterion three provides for a test of commercial durability designed to ensure that each cable assembly will withstand the rigors of future commercial use (LTX’s position), or, alternatively, simply a test of electrical continuity designed to ensure that the myriad wires within each cable assembly were unbroken when the cable assembly was shipped (ITT’s position).3
The district court, after determining first that criterion three was ambiguous, concluded in light of the evidence of previous dealing and negotiation between the parties, that the reasonable interpretation of this provision was that it was intended to ensure that the cable assemblies performed under the flexing stress which both parties knew would occur in use. On that reading, the court upheld the right of LTX under the letter agreement to refuse to accept delivery of, and pay for, the revised cable assemblies. It was undisputed that following execution of the compromise, LTX had received from ITT, and had tested, ten prototype cable assemblies, revised to meet the Revision A specifications, using a test of LTX’s own devising that flexed the assemblies until the wires broke. The test data revealed that the wires in the assemblies would continue to break in commercial use—a finding ITT does not contest. On appeal, ITT challenges both the court’s finding that criterion three is ambiguous and its construction that criterion three calls for something more than a simple test of a cable assembly’s soundness upon delivery.
We do not agree with the district court that criterion three is ambiguous. For the reasons given below, we believe the court erred in finding that criterion three was capable of being read to provide for a durability test such as the court subsequently found. Rather, we hold that the clear language of criterion three calls for a continuity test, as ITT maintains. Since criterion three was not ambiguous, the contract was binding on both signatories according to its written terms.
We also reject the district court’s finding that the contract contained an implied warranty of fitness.
We therefore reverse and remand for further proceedings.
III. No Ambiguity
The determination of whether a contract provision is ambiguous is a question of law subject to plenary review. Fashion House, Inc. v. K Mart Corp., 892 F.2d 1076, 1083 (1st Cir.1989); RCI Northeast Services Division v. Boston Edison Co., 822 F.2d 199, 202 (1st Cir.1987). Under Massachusetts law, parol evidence may not be admitted to contradict the clear terms of an agreement, or to create ambiguity where none otherwise exists. Governor Apartments Inc., v. Carney, 342 Mass. 351, 354, 173 N.E.2d 287, 289 (1961); Tup-*1262per v. Hancock, 319 Mass. 105, 108, 64 N.E.2d 441 (1946); Farber v. Mutual Life Ins. Co. of New York, 250 Mass. 250, 253, 145 N.E. 535 (1924). See also In re Navigation Technology, 880 F.2d 1491, 1495 (1st Cir.1989) (“Contractual language is considered ambiguous where the contracting parties reasonably differ as to its meaning.”); Fashion House, 892 F.2d at 1083 (“Contract language is usually considered ambiguous where an agreement’s terms are inconsistent on their face or where the phraseology can support reasonable difference of opinion as to the meaning of the words employed and obligations undertaken.”).
ITT argues that criterion three is unambiguous in that it simply requires a test of electrical continuity, i.e., a test to establish that none of the 156 wires within the cable assembly were broken at the time of shipment. ITT submits that the record clearly shows, and the district court found, that it drafted criterion three to describe such a minimal continuity test, one which it had, in fact, regularly conducted for several years on the items in issue. ITT argues that the “flexing stress” contemplated was only that necessary to separate already broken wires. Were the cables tested without being flexed at all, the contiguous broken wires would continue to carry electricity and thereby escape detection. Nothing more, it says, was intended or can be inferred from the language of criterion three.
LTX responds, and the district court held, that criterion three was nevertheless ambiguous. The words “while subjected to flexing stress,” set no standard for how often, in what manner and with what force the cable assemblies are to be flexed and stressed during the test, nor is the “special test fixture” described. LTX argues that the language reasonably supports a flexing durability test, in which the amount of flexing stress, although unstated, is whatever amount is required to simulate future conditions of commercial usage to which the cable assemblies will ultimately be subjected. The court thus found that, in light of the incomplete language, criterion three as applied was ambiguous as to whether the flexing stress contemplated was that amount sufficient only to verify electrical continuity at the time of the shipment, or alternatively, was the amount sufficient to guarantee the cable assemblies’ commercial durability during its future use.
We think the district court reads more into criterion three than the language permits. In our view, the language is not ambiguous.
Criterion three states that, “[a]t the time of shipment ... 100% of cable assemblies mil be tested for continuity while subjected to flexing stress.” (Emphasis supplied.) The announced purpose of the test is merely to test the 156 wires contained in the cable for their “continuity ” at the time of shipment and testing. Nothing is said about testing their durability to withstand the rigors of expected future usage. To be sure, wire continuity is to be tested “while subjected to flexing stress.” This forms the basis for LTX’s argument that the court should infer that the aim of testing is to establish durability. As the amount of “flexing stress” is unspecified, so the argument goes, it is possible that the cable assembly will be flexed 1,000 or 10,000 or even more times, turning the test into one for durability. There are several answers to this argument, however.
First, the most logical inference from the complete absence of any reference in criterion three to how many flexes or how much force will be used is that the “flexing stress” contemplated is minimal. Unless the amount of flexing were quantified—or at least, related to some stated durability objective—it is impossible to translate what is written into a meaningful test of durability. How could anyone test for durability without knowing how much stress to impose or, at very least, without being told that the assessment of future durability was an object of the test?
Second, the final sentence of criterion three specifically states the aims of the test—and these aims have nothing whatever to do with future durability. The final sentence says:
*1263This test insures that no wires are broken and that there are no intermittent conditions.” (Emphases supplied.)
This sentence, which is in the present not the future tense, states two purposes, and only two purposes, for the test: It insures (1) that no wires are broken; and (2) that there are no “intermittent” conditions. “Intermittent” means “coming and going at intervals; not continuous.” Webster’s Third Int’l Dictionary (1971). Similarly, “intermittent current” is defined in the dictionary as “an electric current that flows and ceases to flow at regular or irregular intervals.... ” Id. Thus, criterion three says the test insures (present tense) against broken wires and conditions of in-termittency, i.e., conditions where a given wire will sometimes conduct electricity and sometimes not, most likely due to a break which interrupts the flow of current and can be best discovered when the wire is bent so as to cause the broken ends to stop touching. If criterion three were meant to insure against future failure to meet the stresses and strains of commercial usage, one would expect the sentence to say, at very least, “this test insures that no wires will become broken during the commercial usage expected in the future,” or the like. No such test purpose related to future use is announced. Rather, the language of the third sentence speaks merely to insuring the cable assembly’s present soundness, i.e., at the time of shipment.
Finally, the requirement that “100% of cable assemblies will be tested” strongly militates against a durability test interpretation. To impose severe flexing stress on each and every cable assembly, in order to see how much punishment it could take, raises obvious problems. Such a test conducted on 100% of all units would either destroy them, or would consume much of the unit’s useful life. When LTX tested cable assemblies altered under the Revision A concepts, it did so by rigorously flexing ten prototype units until they were destroyed. Common sense suggests that a test of this nature would be limited to a few sample items, as was in fact done, not applied to 100% of the cable assemblies. To be sure, one could perhaps imagine testing in this manner where the need was to detect flawed components or construction errors that would manifest themselves very quickly in use. A new diesel engine might, for example, be run for 20 hours to make sure its parts were all sound. But this record does not indicate that the problem here was due to latent defects in some components that could be discovered in advance by a good shaking up.4 Rather, there was a design problem common to all the cable assemblies that allowed too much strain to be placed on the fragile wires within the cable when the cable was flexed in use. A test for durability applied to a few sample units would be sufficient to ascertain whether this design defect had been corrected. Where a design defect limiting commercial durability is at issue, subjecting all units to intense punishment makes little sense.
We conclude that criterion three cannot reasonably be read to describe a durability test. The district court erred, therefore, in holding it to be ambiguous. To find the possibility of a durability test in language that so clearly limits the purpose of the test to determining “continuity,” “broken wires” and “intermittent conditions” at the time of shipment goes beyond the permissible. LTX, whatever its hopes and expectations, is not entitled to insist that the provision be tortured into something totally different from what it says. Pahlavi v. Palandjian, 809 F.2d 938, 945 (1st Cir.1987) (“[Cjontracting parties are bound by objective manifestations and expressions, not subjective expectations.”); Liberty Mutual Ins. Co. v. Gibbs, 773 F.2d 15, 17 (1st Cir.1985) (“Under Massachusetts law, ‘[wjhere the wording of the contract is unambiguous, the contract must be enforced according to its terms.’ ”) (citation omitted).
*1264We recognize under Massachusetts law that, in order to utilize extrinsic evidence of the parties’-intent, a court need not invariably find facial ambiguity. The Massachusetts courts have said:
When the written agreement, as applied to the subject matter, is in any respects uncertain or equivocal in meaning, all the circumstances of the parties leading to its execution may be shown for the purpose of elucidating, but not of contradicting or changing its terms.
Keating v. Stadium Management Corp., 24 Mass.App.Ct. 246, 249-50, 508 N.E.2d 121, 123 (1987), rev. denied, 400 Mass. 1104, 511 N.E.2d 620 (1987) (quoting Robert Industries, Inc. v. Spence, 362 Mass. 751, 753-54, 291 N.E.2d 407 (1973)). (Emphasis supplied.) In Keating, the court considered extrinsic evidence to aid in deciding whether a clause of general application relating to the availability of commissions applied under certain, very specific, circumstances. 508 N.E.2d at 123. The instant situation is not, however, analogous. Given — as we have just concluded— that the clause is facially un ambiguous concerning the kind of test described, the only use of extrinsic evidence would be the impermissible one of contradicting the very terms of the agreement. Id.
We can stop here, but we go on to add that the extrinsic evidence of record does little to actually contradict the unambiguous language of criterion three. It is undisputed, as the court found, that criterion three was specifically drafted to describe the simple continuity test which ITT had for some time been applying to these cable assemblies. LTX does not claim otherwise. To be sure, the court found that LTX believed at the time of the May 6, 1987 agreement, “that it would have an opportunity to test and, if not satisfactory, modify the new cable assemblies.” The court went on to find,
Although not specifically set forth in the agreement, the parties agreed that ITT Cannon would initially provide LTX with samples of new cable assemblies modified to the Revision A specifications. LTX would then evaluate the new Revision A cable assemblies using its own testing.
(Emphasis supplied.) But while this finding supports the existence of some kind of understanding dehors the agreement about testing, it specifically rejects the notion that the agreement, including criterion three, provides for such a test.5 In fact, the testing referred to by the court could not possibly have been the test described in criterion three. As stipulated, ITT was simply to provide LTX with samples of cable assemblies modified to the Revision A specifications for evaluation by LTX before modifications were made to the balance of the existing cable assemblies in inventory. This arrangement, the parties themselves stipulated, was not specifically set forth in the May 6, 1987 agreement. As noted, criterion three speaks of testing 100% of cable assemblies, not of testing just samples.6
It is true that Boltrus of LTX had apparently never previously encountered criterion three before seeing it in the May 6, 1987, draft that he signed. But even so, no specific evidence exists as to the clause’s role in his decision to sign nor even that he was especially aware of it at the time. The agreement Boltrus signed is explicit as to LTX’s duty to accept ITT’s remaining inventory subject only to the Revision A *1265modifications (which LTX would pay extra for, if done by ITT) or to LTX’s right to make its own modifications. The agreement states,
Regardless of the alternative selected, LTX shall accept delivery of the cable assemblies and pay for them in accordance with the schedule above.
Boltrus could have consulted counsel or inquired of ITT had he wished to determine the meaning of any provision, including criterion three, before he signed the agreement — although, as we say, there is no evidence whatever that criterion three was then an inducing factor in his mind. For all that appears, Boltrus’s and LTX’s present construction of criterion three did not emerge until litigation arose. Putting aside LTX’s problematic reading of criterion three, the executed agreement is clear as to LTX’s duty to take back the cable assemblies without regard to their functionality.
Criterion three, for reasons already pointed out, requires extraordinary optimism to believe that it established a criterion for durability testing. Even a casual reader would have sought some statement therein that, besides merely testing for “continuity,” “broken wires,” and “intermittent conditions,” at the time of shipment, the test was meant to ascertain whether the product would stand up to LTX’s commercial durability requirements in the future. There is not even a hint in the language of criterion three that such was the object of the test or how it was to be accomplished.
There is, moreover, unrebutted evidence that shortly prior to the agreement, ITT had sent to LTX drafts of proposed agreements containing clauses identical to criterion three. These drafts were explicit that ITT would take no responsibility for the performance characteristics of the cable assemblies — making it obvious to anyone who paid attention that ITT did not regard criterion three as a test of durability. We accept, to be sure, as the court found and Boltrus testified, that the latter did not see these papers, which ITT had addressed to another person at LTX. But ITT cannot be assumed to have known this, and the fact they were sent at all undercuts LTX’s argument that we must infer that ITT inserted criterion three simply to mislead LTX.
In any case, we hold that criterion three is not ambiguous: it cannot be read reasonably as a test of durability rather than simply of continuity. Had LTX wanted an agreement conditioning its duty to buy back the cable assemblies upon their meeting certain commercial durability standards, it should have declined to sign the agreement as written, since no such clause or statement appears within the contract’s four corners.
The district court erred, therefore, in holding criterion three to be ambiguous and then construing it in light of extrinsic evidence as a durability test.7 Even, in*1266deed, if extrinsic evidence is considered for any purpose, we reach the same result. The criterion clearly called for no more than a test of wire continuity at the time of delivery, not a test of the product’s ability to stand up against the stresses and strains to which it later would be subjected in the particular use intended by LTX.
IV. Application of UCC, Article Two
The district court held that the May 6, 1987 agreement was a transaction for the sale of goods under article two of the Uniform Commercial Code (UCC), as adopted in Massachusetts, Mass.Gen.L. ch. 106, § 2-102 (“[T]his Article applies to transactions in goods_”); see also Mass.Gen.L. ch. 106, § 2-105 (“ ‘Goods’ means all things (including specially manufactured goods) which are movable at the time of identification to the contract for sale_”). Article two is inapplicable to a contract for the rendition of services. White v. Peabody Construction Co., Inc., 386 Mass. 121, 132, 434 N.E.2d 1015 (1982); Cumberland Farms, Inc. v. Drehmann Paving & Flooring Co., 25 Mass.App.Ct. 530, 534, 520 N.E.2d 1321 (1988). It is, further, well settled in Massachusetts that the test for applicability where the contract is for both goods and services intertwined "is whether the predominant factor, thrust, or purpose of the contract is (1) ‘the rendition of service, with goods incidentally involved, ... or is [instead (2)] a transaction of sale, with labor incidentally involved....’” Id. at 534, 520 N.E.2d 1321 (quoting Bonebrake v. Cox, 499 F.2d 951, 960 (8th Cir.1974)). See also White, 386 Mass. at 132, 434 N.E.2d 1015; Zapatha v. Dairy Mart, Inc., 381 Mass. 284, 290 n. 8, 408 N.E.2d 1370 (1980).
Having concluded that the UCC applies to the May 6 agreement here, the district court held that, pursuant to the UCC, an implied warranty of merchantability attached to the agreement. See Mass.Gen.L. ch. 106, § 2-314(1) (“Unless excluded or modified by section 2-316, a warranty that the goods shall be merchantable is implied in a contract for their sale if the seller is a merchant with respect to goods of that kind.”); Mass.Gen.L. ch. 106, § 2-314(2) (“Goods to be merchantable must at least be such as are fit for the ordinary purposes for which such goods are used_”). The court also found that, because “ITT Cannon clearly had reason to know the particular purpose for the goods, and furthermore LTX was relying on ITT Cannon’s expertise to draw specifications and to modify the cable assemblies for such use,” ITT breached an implied warranty of fitness for a particular purpose under the UCC. See Mass.Gen.L. ch. 106, § 2-315.
We think the district court erred in its application of the UCC here. As both parties were fully aware, the primary purpose of the May 6 agreement was to settle litigation regarding precisely the issue of whether ITT had an obligation to provide cable assemblies which were fit for commercial use pursuant to an earlier contract for the sale of these goods. ITT had argued to the contrary based, inter alia, upon an express provision in its original contract repudiating such implied warranties. In light of the parties’ aim to settle this very issue, we doubt the May 6 agreement was a sale of goods within the purview of article two from which ITT could only exclude implied warranties by a conspicuous writing which mentions merchantability. See Mass.Gen.L. ch. 106, § 2-316(2). However, even assuming that the UCC applied to the May 6 settlement agreement as a contract for the sale of goods, any implied warranty of merchantability was excluded by the course of dealing between ITT and LTX under the UCC. Mass.Gen.L. ch. 106, § 2-316(3)(c); see also Mass.Gen.L. ch. 106, § 1-205(1).
A. Applicability of the UCC Article Two
The district court found that- the UCC applies here because the sale of goods dimension of the May 6 agreement predom-*1267mated over the services dimension. In so ruling, the court distinguished the case of New England Power Co. v. Riley Stoker Corp., 20 Mass.App.Ct. 25, 34-35, 477 N.E.2d 1054, 1060-61, rev. denied, 395 Mass. 1103, 481 N.E.2d 197 (1985). In Riley Stoker, the Massachusetts appeals court refused to treat an agreement settling a dispute over the sale of defective boilers as one for the sale of goods. The district court distinguished Riley Stoker on the grounds that, as the boilers had already been delivered and installed, the settlement agreement was really an agreement for services, repairs to the boilers forming the crux of the agreement.
It is questionable whether Riley Stoker can be so easily distinguished. We think the district court overlooked the emphasis in Riley Stoker on the fact that the contract was not a modification of the original sales contract, but rather was a settlement agreement whereby alterations would be made to the boiler system in exchange for release of withheld payments.8 In the present case, as in Riley Stoker, the buyer and seller in an underlying transaction for the sale of goods settled their dispute; an undertaking was made to alter the goods together with the buyer’s agreement to make payment on all of the goods originally sold. In both cases the underlying sales were not complete — in Riley Stoker the boilers were not paid for, and here all the cable assemblies were neither fully delivered and accepted nor paid for. The settlement agreements both involved modifications of the goods, surrender of claims, and cash payments. The distinction is that the underlying dispute arose in Riley Stoker after full delivery of the goods, whereas here the underlying dispute arose only after a portion of the cable assemblies had been delivered. We cannot say, however, that actual delivery is the touchstone of a sale of goods. We think it likely, therefore, that a Massachusetts court would apply Riley Stoker to hold that the May 6 agreement was not a contract for the sale of goods.
B. Exclusion of Implied Warranties
Quite apart from Riley Stoker, however, and even assuming that the May 6 agreement was a contract for the sale of goods within the purview of article two, we believe that any implied warranties of merchantability and fitness for a particular use must be excluded by the course of dealings between ITT and LTX. See supra, Mass. Gen.L. ch. 106, § 2-316(3)(c) (“[A]n implied warranty can also be excluded or modified by course of dealing_”); Mass.Gen.L. ch. 106, § 1-205(1) (“A course of dealing is a sequence of previous conduct between the parties to a particular transaction which is fairly to be regarded as establishing a common basis of understanding for interpreting their expressions and other conduct.”). ITT and LTX had disputed— since at least the time when LTX first refused to accept delivery of cable assemblies in October, 1985 — their respective responsibility for the cable assemblies’ break*1268age when subjected to commercial use. This was the point of contention throughout the negotiations leading up to the May 6, 1987, settlement agreement. ITT consistently maintained that, under its original contract and relationship, it was not responsible for producing cable assemblies that stood up under the conditions of usage encountered by LTX. It consistently refused to accept such responsibility and, in effect, warrant their fitness for LTX’s intended use. LTX consistently maintained that ITT was at fault and that LTX had no obligation to accept unusable cable assemblies. If nothing more, the parties certainly were aware that the bone of their contention was exactly the question of responsibility for the goods’ merchantability and fitness for a particular purpose.
The conduct and negotiations between the parties up to the May 6 settlement agreement establish a common basis of understanding that the responsibility for the failure of the cable assemblies in actual use was the central point the agreement would resolve. By virtue of this understanding, the parties could not possibly have expected that the settlement agreement would resolve this point by operation of an implied warranty. The parties could only fairly assume that the allocation of responsibility for the cable assemblies’ merchantability and fitness for actual use — the point any settlement would have to resolve — would be reflected in the terms of the settlement agreement. We think that LTX, in light of the parties’ course of conduct in negotiating this agreement, could not have reasonably expected that ITT’s failure to comply with ordinary UCC requirements (that implied warranties be disclaimed in conspicuous writing with explicit reference to merchantability) would result in ITT’s acceptance, by legal implication, of a responsibility that it had so long emphatically and so importantly denied.9
To be sure, application of the UCC provision for disclaimer of implied warranties by course of dealings has most commonly arisen where the buyer had agreed in prior transactions to the seller’s express disclaimer of implied warranties, leading both parties to anticipate continuation of the same disclaimer. See, e.g., Hartwig Farms, Inc. v. Pacific Gamble Robinson Co., 28 Wash.App. 539, 625 P.2d 171, 175 (1981) (though buyer had accepted invoices containing disclaimer of warranties for 15 years, there was no evidence that, on prior occasions, buyer had agreed to disclaimer at time of ordering, and no course of dealing from which to infer assent to disclaimer); Wilson v. Marquette Electronics, Inc., 630 F.2d 575, 581 (8th Cir.1980); cf. Bowdoin v. Showell Growers, Inc., 817 F.2d 1543, 1545 n. 18 (11th Cir.1987). Nevertheless, our construction is not only fully consistent with the course of dealings exception but with the overall rationale of UCC § 2-316:
“This section is designed principally to deal with those frequent clauses in sales contracts which seek to exclude ‘all warranties, express or implied.’ It seeks to protect a buyer from unexpected and unbargained language of disclaimer by denying effect to such language when inconsistent with language of express warranty and permitting the exclusion of implied warranties only by conspicuous language or other circumstances which protect the buyer from surprise.”
(Emphasis supplied.) Fairchild Industries v. Maritime Air Service, Ltd., 274 Md. 181, 333 A.2d 313, 316 (1975) (quoting Official Comment 1); see also Geo. C. Christopher & Son, Inc. v. Kansas Paint & Color Co., Inc., 215 Kan. 185, 523 P.2d 709, 716, modified on other grounds and rehg. denied, 215 Kan. 510, 525 P.2d 626 (1974) (“The very purpose of the statutory requirement [of conspicuous disclaimers] is that any limitation be brought to the attention of the buyer at the time of the contract.” Thus *1269course of dealing disclaimer based on post-contract conduct was ineffective.). There is no question here, however, that LTX was aware at the time of contracting that ITT had been disclaiming responsibility for the fitness of the cable assemblies just as LTX had been insisting upon ITT’s responsibility, and thus LTX had no basis on which to believe that such responsibility would be allocated to ITT by implication. To the contrary, LTX, in signing the settlement agreement, would have expected the dispute to be resolved within its four corners. This expectation, growing out of the prior course of dealings, was sufficient to protect LTX from any “surprise.” Indeed, the only surprise would be if implied warranty terms modified the agreement’s meaning.
We hold, therefore, that LTX, in rejecting the cable assemblies, was not entitled to rely on ITT's breach of an implied warranty of fitness.
V.
We accordingly reverse the district court’s determination that LTX was entitled to refuse to accept the cable assemblies (1) on grounds of the court’s particular reading of the meaning of criterion three and (2) on breach of warranty grounds. The judgment below is reversed and the case remanded for further proceedings consistent herewith.
So ordered. Costs for appellant ITT.
APPENDIX A
May 6, 1987
Mr. Richard Boltrus
Vice President, Administration
LTX Corporation
LTX Park at University Avenue
Westwood, Massachusetts 02090-2306
Dear Dick:
This is to confirm in writing the agreement we reached yesterday. What follows is my understanding of that agreement. Your signature will reflect that our understandings coincide and this letter will constitute our agreement.
The inventory at Cannon of cable and cable assemblies that are the subject of the current litigation between us is as follows:
Item Cannon P/N Quantity Description
1 970-0008-478 1,183 CORE Cable P/N NR RCN4982-1
2 3 970-0008-479 111620-0008 1,532 CORE Cable P/N RCN4983-1 313 TS88 Assem. DL156 to DL156
4 111620-0009 166 TS88 Assem. DL156 to DL96
5 111620-0007 2 TS88 Single Cable Assembly
Except for Item 5 which Cannon will keep and which is not a subject of this agreement, LTX will take delivery of and pay for the inventoried cables and assemblies in accordance with the following schedule:
Item Description Quantity Shipped Price Extended Each Price Date Payment Terms
$185,888 5/8/87 COD/Bank Trf. 1 & 2 970-0008-478 rH CO cn CD CO
185,888 5/8/87 COD/Bank Trf. 970-0008-479 t — ( CO or CD CO
1184 $371,776 Subtotal
$190,302 8/5/87 Net 30 days 1 & 2 970-0008-478 CO t — 1 Oí to
$190,302 8/5/87 Net 30 days 970-0008-479 C\1 CO tH Oí io
1182 $380,604 Subtotal
*1270Item Description Quantity Price Extended Shipped Each Price Date Payment Terms
$192,168 11/5/87 Net 30 days 3 & 4 111 620-0008 (M i — 1 CJI —3
101,592 11/5/87 Net 30 days 111 620-0009 eg . OO CO
$293,760 Subtotal O
3 & 4 111 620-0008 156 $1,272 $198,432 2/5/88 Net 30 days
111 620-0009 83 $1,272 105.576 2/5/88 Net 30 days
Subtotal 239 $304,008
In addition, LTX will pay $18,100 non-recurring tooling and tooling use charge for Revision “A” modification net 30 days after invoice from Cannon.
Part numbers used in the above schedule are the part numbers currently in use. Mutually agreed upon modifications to these assemblies may result in part number changes. Cannon will inform LTX of any part number changes.
Prices used in above schedule include $124 each for Cannon incurred costs to modify existing cable assemblies to conform to “Revision A” specifications as presented by LTX (see Cannon Drawing No’s R 86002-000 and R 86002-1000) or to design currently being produced for LTX by Wire Tech Corp. of Florida. Price changes for any other modification proposed by LTX will be subject to mutual agreement. LTX will have the right to perform any modification at their facility or any facility they designate, in which case the prices for the assemblies included in the above schedule will be reduced by $124 each. Regardless of the alternative selected by LTX, LTX shall accept delivery of the cable assemblies and pay for them in accordance with the schedule above.
At the time of shipment the cable assemblies will conform to the following acceptance test criteria:
(1) 100% of cable assemblies will have continuity and short testing using Cox Digitrace Equipment.
(2) 100% of cable assemblies will have hi-pot testing with each conductor to all other conductors and shield, 200 VRMS, 60 HZ using ITT Cannon’s custom hi-pot test system.
(3) 100% of cable assemblies will be tested for continuity while subjected to flexing stress. The cable is exercised in a circular pattern while observing 156 indicator lights on a special test fixture. This test insures that no wires are broken and that there are no intermittent conditions.
All cable assemblies modified to LTX’s “Rev. A” requirements, to Wire Tech design or to other LTX requirements shall be subject to the standard Cannon terms and conditions of sale.
If all this is mutually acceptable, our attorneys will work out a dismissal of the pending litigation between us.
If this letter reflects your understanding of our agreement, please sign and return the duplicate copy of this letter.
ITT Cannon, a division of
ITT Corporation
s/n
W.P. Sekeres
Vice President, General Manager
Components Division
Accepted this of day
LTX Corporation
s/n
Richard Boltrus
Vice President Administration
. Sekeres and Boltrus had conducted their negotiations outside the presence of their lawyers, thinking to avoid legalistic haggling. Each apparently emerged with a different understanding: Sekeres had the view that LTX had agreed to buy ITT’s inventory of cable assemblies whether or not the Revision A changes {see infra) solved' LTX’s problems, subject only to Sekeres's assurance that ITT’s engineers would, in the future, be as helpful as possible in providing assistance toward curing the problems of the cable assemblies. Boltrus, on the other hand, while conceding that Sekeres denied responsibility, testified that he understood that LTX need not take the cable assemblies should future planned durability tests of the revised cable assemblies indicate that they were still unable to stand up in use. Boltrus did not, however, testify that any specific test procedure was discussed or that language such as that found in the disputed criterion three, infra, was ever mentioned in discussions between himself and Sekeres. Indeed, there is no testimony as to what Boltrus made of that provision when he signed the letter agreement, much less that he relied upon it or believed that it insured against the wire breakage plaguing the project.
. This was in fact done, as was the stipulated dismissal of ITT’s contract action against LTX. Because the cable assemblies, which are the focus of this dispute, were not to be purchased by LTX until eight to ten months after it signed the agreement, the present controversy was still well in the future when these preliminaries (i.e., purchase of the cable and dismissal of the suit) took place.
. It is undisputed that ITT drafted criterion three to reflect a simple continuity test ITT had previously been performing on the unusable cable assemblies. The object of that test was to see that every wire within the cable was unbroken; it was not a test to assess future endurance. Each cable assembly was exercised in a circular pattern while under the stress of a flexed position to make sure that the ends of any broken wire were separated and hence the break would be revealed by the failure of the attached light bulb. A broken wire may carry current if the ends touch. Flexing, however, would cause the wire to open and close, causing the light to blink.
. As a simple manufacturing test to guard against defective wires, however, it makes good sense to test every item. Flawed component wires broken in or before manufacturing would be detected in any unit. Testing 100% of the units logically comports with this manufacturing defect interpretation.
. LTX places great emphasis upon the evidence that it never backed down from its original refusal to accept unusable cable assemblies. But while this suggests that (from LTX’s perspective) the May 6, 1987 agreement should have had a provision protective of that position, it does nothing to elucidate the intended meaning of criterion three.
. As already noted, LTX's durability flex tester was destructive; the cable assembly was flexed until wires broke and the number of flexes was then counted. If such a test were applied to “100% of cable assemblies,” as the criterion three test must be applied, one wonders how it could be used in practice, since if adopted literally, it might ruin every one of the assemblies. Presumably, some level of flexing would have to be established short of destruction, which, when reached, would cause the product to "pass," but this would seem to use up a significant portion of the cable assembly's useful life. Neither the district court nor LTX addresses this perplexing aspect of the problem.
. Were we to find the clause ambiguous, and to accept the court’s finding that it should be reasonably construed to require the cable assemblies to undergo successfully, at delivery, the flexing stress to which both parties knew that the cables would be subjected in commercial usage, we could still not affirm. In an ordinary case of ambiguity, the court’s construction of the written contract’s reasonable meaning would, unless clearly erroneous, resolve the case legally. Clark v. State St. Trust Co., 270 Mass. 140, 169 N.E. 897 (1930). See Corbin, On Contracts, § 536 (1963). Here, however, the court expressly found that ITT’s understanding was that criterion three described its own modest in-house continuity test whereas LTX, unaware of that test, read it as meaning something else altogether. These polar differences in the parties’ actual understanding of the contract's meaning left no occasion for the court to divine a "reasonable" interpretation that would reflect a single probable intent to which both parties subscribed. Thus, if we assume ambiguity, there was plainly no "meeting of the minds" here — giving rise to the inescapable question whether, on the signing of the agreement, any binding contract was made at all. Raffles v. Wichelhaus, 2 Hurl. & C. 906 (1864); Kyle v. Kavanagh, 103 Mass. 356 (1869) (if the parties gave materially different meaning to the contract language, there is ordinarily no meeting of the minds and no contract). See also 1 A. Corbin, On Contracts, § 104 (1963). LTX recognizes the problem, but wants us to find a contract notwithstanding the absence of a meeting of the minds under the exception existing where one party knows or has reason to know that the other party understood the contract to mean something different from what the first party intended. Chelsea Industries, Inc. v. Accuray Leasing Corp., 699 F.2d 58, 60-61 (1st Cir.1983). *12663 A. Corbin, On Contracts, §§ 104, 538 (1963). See also 1 S. Williston, Contracts, § 95 (1957). The district court, however, made no finding that ITT knew or had reason to know that LTX read criterion three as establishing a durability test. We ourselves, at the appellate level, are in no position to make such an assumption and would decline to do so.
. The Riley Stoker court stated:
As we read that [settlement] agreement, however, it is not a contract for the sale of goods, nor does it modify the original contract, thereby bringing it under the umbrella of any warranties therein given.
The 1974 document is labelled a "Settlement Agreement,” not a modification agreement as [the buyer] characterizes it. The October amendment specifically refers to amendment of the July “Settlement Agreement." The preamble to the substantive clauses of the July agreement recites that "disputes have arisen between [buyer] and [seller] relating to the contract" between them for the Salem boiler and that “the parties hereto are desirous of compromising and settling said dispute." In exchange for [buyer’s] release of the withheld payments to [seller], [seller] agreed to "fabricate, free of charge," the stringer support system. If that system were to exhibit any defects within one year following "start-up after installation,” then Riley "will provide free of charge engineering services as necessary to correct said defects." [Buyer] makes no argument that [seller] failed to provide any necessary engineering services under this agreement. As we construe the [settlement] agreement, as amended, it was not a contract for the sale of goods, and it did not give rise to any implied warranties. See Stafford v. International Harvester Co, 668 F.2d 142, 147 (2d Cir.1981); Consolidated Edison Co. of New York v. Westinghouse Elec. Corp., 567 F.Supp. 358, 362 (S.D.N.Y.1983).
Riley Stoker, 477 N.E.2d at 1061 (citations omitted).
. ITT underscores, as evidence of its emphatic denial of responsibility: (1) the parties’ stipulation that ITT denied responsibility in the prior litigation; (2) testimony that ITT denied responsibility at the December 11, 1986 meeting; (3) February, 1987 correspondence reiterating this position; (4) March 24, 1987 correspondence enclosing ITT's standard terms and conditions (which include a disclaimer of implied warranties); (5) and testimony of LTX’s Boltrus conceding that, at the May 5, 1987 meeting, ITT’s Sekeres "wouldn’t guarantee anything."