delivered the opinion of the court:*
*816The decisive issue in this review of a decision of the Armed Services Board of Contract Appeals (ASBCA), Appeal of J. A. Maurer, Inc., 69-2 BCA ¶ 7884, under the standards of the Wunderlich Act, 41 U.S.C. §§ 321-22, turns chiefly on the question of impossibility of performance of an Air Force contract terminated in 1966 for default in performance. Of secondary importance (to us, but not so relegated in plaintiff’s view) are plaintiff’s other contentions based on its interpretation of the contract requirements.
Fact recitals herein are based on ASBCA findings, except as indicated. In the desert at the Air Force Missile Development Center (AFMDC) Holloman Air Force Base, 9 miles west of Alamogordo, New Mexico, the Air Force has a high speed test track, on which special sleds are propelled at great velocity while data is recorded and digested by elaborate equipment for research purposes. The track, which was extended in 1956-57 to its present length of some 35,000 feet (about 7 miles) consists of two rails embedded in a concrete girder foundation. The parallel rails run north and south, are 7 feet apart, and are seamlessly welded where the sections join. A “fiducial line” runs parallel to and 9 inches outside of the western rail. The fiducial line serves as an essential reference for alignment of the track rails, which must be flat and straight to an extreme degree of accuracy to minimize destructive vibrations that are accentuated by sled speed and track irregularity. At each of 352 locations spaced 99'8" apart along the length of the fiducial line there is inserted a bench mark by reference to which at that point the lateral and vertical positions of the rails can be compared and periodically adjusted as they get out of sync through use, or through deformations in expansion and contraction during temperature extremes, or because of foundation settlement. Vertical movements of the rails exceed lateral movements. Sometime ago the Coast and Geodetic Survey established the fiducial line bench marks to an accuracy of approximately 1 arc second by use of a conventional theodolite.
This degree of accuracy did not satisfy the Air Force, as indicated by its Technical Exhibit A dated November 20, 1964, which was later incorporated into the procurement from *817which this action derives. Technical Exhibit A, which proposed a study and report on lineal survey methods for the test track, and a determination of its linear coefficients of expansion, for the purpose of improving the accuracy of existing techniques and facilities, was attached to the Government’s Request for Proposal (RFP) issued on December 8, 1964, to seven companies, including the plaintiff at its request. In its initial form the RFP covered the features outlined in Technical Exhibit A, and was to be essentially a study program.
The plaintiff’s Technical Proposal submitted in response to the RFP and dated January 17,1965, compared several potential survey methods and of them recommended as a means interferometric and diffraction alignment techniques as opposed to electronic techniques. It particularly advocated using the Van Heel alignment apparatus to determine the lineal integrity of the fiducial line to reference the track alignment, which apparatus offered alignment accuracy reputed to be within a few tenths of an arc second. Another method, using the Axicon apparatus, was also described. Plaintiff’s proposal quoted a price of $60,782.90.
It was the only one submitted to the Government from the seven firms receiving the RFP. On February 18, 1965, Dr. Woodson, the head of plaintiff’s newly formed Optics Division and a highly respected authority in the field of interferometry (said to be second only to Dr. Van Heel of Holland), gave an oral presentation concerning plaintiff’s Technical Proposal to a group of Air Force technical representatives at AFMDC, at which he rendered a convincing explanation of the Van Heel technique which plaintiff planned to use in measuring the lineal integrity of the test track and in setting up an improved system for its continuous monitoring. He described the proposed equipment, the achievable accuracy he foresaw as possible, and the environmental problems to be solved. Because of budgetary restrictions the Government on March 2, 1965, issued Amendment No. 1 to its Technical Exhibit A, which deleted the original requirement to measure the “coefficient of expansion” of the track (i.e., its exact length under temperature fluctuations), en*818larged the lineal integrity requirement by applying it to the entire length of the track instead of merely a portion thereof, and provided that the Government should retain the survey equipment employed.
The parties then negotiated a price of $31,943 reflecting the foregoing amendments to the Government’s Technical Exhibit A, and on May 17, 1965, consummated the contract in suit which called for items 1 and 3 only. Item 1 was divided into two phases: engineering services to investigate techniques to improve lineal surveys (Phase I), and an actual survey throughout the 7-mile track (Phase II). Item 3 required plaintiff to provide the Government with “Improved Lineal Survey Equipment (ILSE), such as the Van Heel Alignment Apparatus; the Light Source (Laser) and Colli-mating Lens — fringes.” The plaintiff’s proposal culminating in the contract included an estimated allowance of $4,000 for the Yan Heel equipment. The contract intentionally omitted Phase III work of item 1 contained in the Government’s original Technical Exhibit A involving the measurement of the track length, although plaintiff had submitted a separate figure for Phase III work should the Air Force find the wherewithal.
There followed months of effort to perform the contract, time extensions, validation of the intrinsic accuracy of the interferometric procedure, but ultimate failure because of inability to develop a convection barrier which would provide sufficient shelter to laser beams traveling through it in an outdoor environment so as to duplicate the accuracy in measurements that it attained under controlled laboratory conditions indoors. In the end, when the contract was terminated for default, it appeared that accomplishment of the contract objective might be technically possible given sufficient funds and time, but practically impossible to accomplish within the monetary and time limits allowed by the fixed price contract. See, however, discussion below of ASBCA findings on this topic.
A factor of fundamental impact in the case is that alignment measurements in excess of 1 arc second of accuracy had not been theretofore achieved with the new generation of *819interferometric and diffractometer apparatus and techniques outside of controlled laboratory conditions. The prime aim of the contract — to better the existing 1 arc second of accuracy of the test track which had been obtained by use of a standard theodolite — had never been done before in an uncontrolled, outdoor environment such as that at the Hollo-man test track, and in retrospect proved to be beyond the state of the art, particularly within the short period (4 months) and contract price ($31,943) allowed for performance in this instance. That the plaintiff was fully aware of the difficulties and of the risk of failure involved is shown not only by its knowledge of paragraph 3 of the Government’s Technical Exhibit A which described the physical environment of the track, but also by plaintiff’s Technical Proposal of January 17, 1965, which recognized the atmospheric turbulence and the temperature gradients at Holloman to be a major problem, but proposed to overcome it by means of using interferometric equipment involving the projection of a laser beam through a light tube superimposed on the test track rail. The purpose of the tube was to serve as a “convection barrier” to shield the laser beam passing through it from the atmospheric effects of its desert surrounding, namely, wind, sun, temperature fluctuations, “ground boil”, “grazing rays”, etc., which had a drastic effect on the super-sensitive measurement apparatus which plaintiff intended to use. In effect, the convection barrier was aimed at duplicating laboratory conditions in an outdoor environment.
Despite Dr. Woodson’s congenital optimism over his ability to counter the atmospheric hazards in order to apply a hitherto laboratory technique in an outdoor ambience, the plaintiff’s other officials had their mental reservations and suspected the undertaking to be on the research frontier, but deferred to Dr. Woodson’s superior expertise. The contract was to become a proving ground of his theoretical calculations and scientific judgment, and in the end he conceded that the amount of research involved exceeded his expectations.
Basic to Dr. Woodson’s plan were the availability of suitable interferometric equipment and the devising of a convection barrier consisting of connected tubes which would shield *820the sensitive light source. The Van Heel apparatus was selected as preferred equipment, but when in June 1965 Dr. Woodson visited Dr. Van Heel in Holland to arrange for the equipment and to discuss prospective problems, he found that the apparatus could not be delivered for three months, at best, not soon enough to meet the 4 months’ time available for contract performance. Dr. Woodson then decided to use another apparatus recently designed by Mr. Saunders of the Bureau of Standards for alignment measurement purposes. The Saunders apparatus was an interferometer, while the unavailable Van Heel apparatus was based on interference and diffraction and was accordingly termed a diffractometer, but both of them were embraced in the field of interferometry and were designed to make accurate optical measurements in a controlled laboratory environment free of atmospheric disturbances, under which conditions they were capable of .1 arc second of accuracy or better. Dr. Woodson considered that both of the apparatuses in question performed the same function in somewhat different ways, and thought the Saunders device to be the more sensitive of the two. The Saunders equipment had never been used outdoors nor did its inventor think it could be, as Dr. Woodson understood prior to contract award. The Van Heel apparatus had been used with disappointing results in an outdoor test in Holland, yielding over a short distance an accuracy of only 1 to 14 arc second. The Government was unaware that plaintiff had decided to use the Saunders equipment instead of the Van Heel until so advised by plaintiff’s first monthly progress report for the period ending July 12, 1965. The contracting officer in acknowledging the advice stated that the Saunders apparatus “may even be superior to the Van Heel for long sighting lines.” As late as December 16,1965, the plaintiff considered that the Saunders equipment, as it was being modified by plaintiff, was “even superior to the original one contemplated [i.e., the Van Heel]”, so the unavailability of the Van Heel apparatus was not then considered by plaintiff to be the reason for its ultimate failure to achieve the desired result. The ASBCA findings imply that the substitution had no causal connection with the failure.
*821The dissimilarities in the Yan Heel and Saunders techniques were not significant, in the ASBCA’s eyes, for both were usable in making optical measurements to an extraordinary degree of refinement in a controlled laboratory environment, while as it came to pass neither was within the state of existing art to make measurements of comparable accuracy in an uncontrolled natural environment. This proved to be the obstacle plaintiff was unable to overcome. Finally, Dr. Wood-son, who was eventually dismissed by plaintiff for his inability to solve the problem which led to default termination of the contract, testified as the Government’s witness that the Saunders equipment would constitute equipment “such as” the Yan Heel in the science of interferometry. He should know.
Eeverting to the convection barrier problem, plaintiff contemplated using connected 20' lengths of tubing 7' in diameter so that the laser beams could travel in a straight line and be isolated from atmospheric disturbances. The tubing was to be mounted on the test track at night for the making of measurements in 4,000' segments at a time. Turbulence was less at night than during the day, and moreover this schedule would not interfere with the Air Force use of the test track during the day. (The plaintiff’s contention that unavailability of the test track interfered with its contract performance is without substance, as the Board also found.) At first, the plaintiff thought of using polyethylene tubing, but because of its greater rigidity elected to use aluminized waterproof cardboard tubing which was purchased and delivered about October 1965. It was intended to mount the connected tubing in some fashion 7" above the test tract in assemblies of 4,000' at a time. The 7" suspension above the track was to insulate the convection barrier from distorted readings which might result from contact with the track because of a thermogradient imbalance between the temperature on one side of the track and that on the other side caused by the night breeze in the desert blowing against one side and not the other. It was calculated that it would take about 4 hours to set up each of the 4,000' assembled segments of convection barrier, and about 3 hours to dis*822assemble it, with the actual shooting of measurements being done rather quickly once the setup was ready. If 4,000' segments were set up and measurements taken on consecutive nights, plaintiff expected to be able to survey the entire 35,000' track in nine nights when the equipment was perfected.
This did not transpire, because plaintiff was never able to develop a convection barrier that would insulate the atmosphere sufficiently to make accurate measurements outdoors. It abandoned its original plan to conduct a pilot test in Long Island during the fall of 1965 because its modifications of the Saunders apparatus were not ready in time.
After substantial delays, and a time extension to March 1, 1966, in January 1966 the plaintiff set up a 55' experimental test track inside of a structure at Holloman which provided conditions which were of less than laboratory quality but considerably better than in the open. With this experimental setup an encouraging accuracy of .012 arc seconds was obtained, which converts to an unprecedented one-fourth of a thousandth of a second in a distance of 55'. This accomplishment, which vindicated Dr. Woodson’s faith in the inherent accuracy of the modified Saunders apparatus, was superior to the 1 arc second of accuracy obtained previously with a theodolite at the test tract. But unfortunately it could not be duplicated outdoors, despite several efforts on track lengths from 75' to as much as 3,200'.
By this time it was clear that the Saunders equipment was suitable but no means had been found to use it outdoors with an acceptable degree of accuracy. At the time that plaintiff virtually ceased further performance under the contract, which was about March 1966, Dr. Woodson felt that, with an investment of another $10,000 or $20,000, and further time, a solution could be found. However, by now the plaintiff had spent a claimed $61,000 in its efforts to perform in contrast to a contract price of $31,943 and actual contract payments of $22,290, and regarded the contract to be for practical purposes impossible of performance. So it discharged Dr. Wood-son in April 1966, and thereafter until def ault termination of the contract on November 1,1966, engaged in fruitless nego*823tiations with the Air Force to either terminate the contract for convenience and reimburse the plaintiff for its losses, or convert the contract to a cost-reimbursable basis, in which case it was willing to waive its claim for losses. There was little reason for the plaintiff by that time to have had much confidence remaining in Dr. Woodson’s belief that a successful end was in sight with a modest additional expenditure, for already the performance period had far exceeded the original 4 months and plaintiff’s costs had far exceeded the contract price. In terminating the contract for default the contracting officer also demanded reimbursement of the $22,290 paid the plaintiff in progress payments. The plaintiff appealed the termination to the ASBCA on November 21, 1966, and on September 10,1969, the appeal was denied, leading to the present review proceeding.
The ASBCA decision, with much thoughtful discussion, states the conclusion as follows: “* * * The excuses of practical impossibility and ‘commercial senselessness’ are not available under the circumstances.”, terms which the ASBCA rightly considered to be synonymous. It restricted the proper application of practical impossibility to the cases where “specifications [are] furnished by the Government or otherwise warranted by it, even though impliedly.”, there being no such implied warranty found in the present case. Nor did the Government have a superior knowledge of the technology and withhold it, as in Helene Curtis Industries, Inc. v. United States, 160 Ct. Cl. 437, 312 F. 2d 774 (1963). (There is nothing to suggest any Government representative concerned in the procurement had knowledge of the technology in any way equal to Dr. Woodson’s).
The trial judge upheld the plaintiff on the “impossibility” issue, differing with the ASBCA. He apparently agreed with the ASBCA on other issues, which he did not address in any detail, deeming plaintiff’s contentions to lack substance. Since we agree with the ASBCA and differ with the trial judge as to impossibility, we find it necessary to give plaintiff’s other contentions more notice. We take them up first, leaving impossibility to the end.
*824Plaintiff’s main point is that under the requirement in Item 3 for “Improved Lineal Survey Equipment (ILSE) such as the Van Heel Alignment Apparatus * * (emphasis supplied) the Saunders equipment is too unlike to be “such as”. Therefore, when plaintiff switched to the Saunders device, it was a cardinal change. “The contract as written was impossible to perform because the Van Heel Alignment Apparatus was not available. A new contract came into existence, and the language should have been changed if defendant wanted to preserve its privilege of declaring a default.” The ASBCA analyzes the record in detail to show that plaintiff always had the option to make the switch. We agree with what it says, but a shorter answer is that in adopting the Saunders device without notice to defendant, plaintiff interpreted the contract in a sense clearly contrary to what it now urges. It did not even allow defendant the option to retain the Van Heel device at the cost of some delay, possibly less than defendant ultimately granted in time extensions anyway. A party cannot, after a controversy has arisen, arbitrarily abandon the contract interpretations it acted on to the other’s knowledge when their relations were harmonious. Gresham & Co., Inc. v. United States, 200 Ct. Cl. 97,470 F. 2d 542 (1972). It should be noted, of course, that this argument though using the word “impossible” does not concern the issue of commercial or technological impossibility, dealt with infra.
Plaintiff also urges that standards of performance were not given and therefore the contract was effectively unenforceable. We think, as did the ASBCA, that plaintiff knew what it was trying to do when it was trying to do it, and cannot now urge that the level of satisfactory performance was imprecise.
We also agree with the ASBCA that neither the short time originally allowed for performance, nor the modest contract price, establish that it did not call for research and development, when its own terms show that it did.
Turning, then, to the main legal issue, we note that the ASBCA had several things to say about the impossibility *825issue, mostly unfavorable to plaintiff, but in tbe end did not attempt to determine whether the contract was impossible to perform, in the legal as opposed to the scientific sense. A careful reading of the Decision part of its opinion that could have been labeled OonoTmions of Law, shows that its position was that the excuse of legal impossibility is only available where “under Government specifications, failure to perform is due to the circumstance that no way is known in industry how to manufacture the supplies in question according to the specified design or performance characteristics. * * * [T]he Government [therefore] is held to an implied warranty.” This is not such a case, because here the plaintiff was to supply methodology and design. In these circumstances, the parties did not intend “limitations with reference to the state of the art.” Plaintiff “knew the work tasks to be performed had not been done before”. In these circumstances, no implied warranty could be imputed to the Government. Therefore the conceptual underpinning for the impossibility defense as here invoked was missing.
The reasoning is in general in accord with our decisions and strongly supported by them. In Austin Co. v. United States, 161 Ct. Cl. 76, 314 F. 2d 518, cert. denied, 375 U.S. 830 (1963), we assumed that plaintiff’s failure to perform was solely due to the fact that it was impossible to do so under the specifications. However, the latter were proposed by the plaintiff after it had reviewed defendant’s specifications and found them unworkable. We held that plaintiff, having promised to perform under its own substituted specifications fully assumed the risk of impossibility of performance. As the Government was responsible for losses due to its own specifications if they were defective, the converse should apply to the plaintiff.
In Bethlehem Corp. v. United States, 199 Ct. Cl. 247, 462 F. 2d 1400 (1972), we followed Austin. The Bethlehem Corp. was recognized to have superior expertise. Its technical proposal was made part of the contract and defendant modified the specification, pursuant to its advice. The contract, for an environmental test chamber, was impossible to perform within the then known state of the art. Defendant’s officers *826who prepared the specifications, did not know this, but there were experts in the National Bureau of Standards who would have. We did not impute their knowledge to the contracting agency. (NOTE: the Bureau of Standards people were only aware of the difficulties, not of possible solutions, as has been incorrectly suggested here in an effort to distinguish the case.) We held that Bethlehem had assumed the risk of nonperformance notwithstanding impossibility.
In Bethlehem we cited and quoted from the leading decision of Judge Friendly in United States v. Wegematic Corp., 360 F. 2d 674 (2d Cir. 1966). There in response to an invitation for proposals for computers, Wegematie submitted a detailed proposal for a machine it characterized as “a truly revolutionary system utilizing all of the latest technical advances.” It proved impossible to deliver because of “basic engineering difficulties.” Judge Friendly dryly says, at p. 676:
* * * We see no basis for thinking that when an electronics system is promoted by its manufacturer as a revolutionary breakthrough, the risk of the revolution’s occurrence falls on the purchaser; the reasonable supposition is that it has already occurred or, at least, that the manuf acturer is assuring the purchaser that it will be found to have when the machine is assembled. * * *
In Hol-Gar Mfg. Corp. v. United States, 175 Ct. Cl. 518, 360 F. 2d 634 (1966), we enforced the implied warranty that if Government specifications are followed, a satisfactory product will result. We distinguished the Austin case, supra, on the ground that Austin had drafted substitute specifications which it had said would produce the result.
Hoi-Gar thus falls in the line of cases dealing with impossibility of performing under an assertedly defective Government specification. Others are, e.g., Natus Corp. v. United States, 178 Ct. Cl. 1, 371 F. 2d 450 (1967); Tecon Corp. v. United States, 188 Ct. Cl. 436, 411 F. 2d 1271 (1969); Tombigbee Construetors v. United States, 190 Ct. Cl. 615, 420 F. 2d 1037 (1970). In these cases the question has had to be answered: assuming absolute or utter impossibility is not the test, how impracticable, or unfeasible or commercially senseless must performance be, to excuse its not being rendered? *827It is in this context that Judge Collins in Natus, sufra, delivered his much-quoted observation that — “impossibility in its modern context has become a coat of many colors” — at p. 9, 371 F. 2d at 456. It may be that the word “impossibility” is somewhat of a semantic trap, but we do not have to fall into it here. As Judge Wright points out in Transatlantic Financing Corp. v. United States, 363 F. 2d 312, 315 (D.C. Cir. 1966), before getting into the question you first have to determine whether risk of the contingency has been allocated either by agreement or by custom. The cases say it has been, here, and therefore the ASBCA was entirely right in not deciding whether it was impossible to perform or not, in a legal sense.
It has recently been stated the Government warranty of its own specifications does not apply to “performance oriented” as distinguished from “design oriented” specifications. Patten, The Implied Warranty That Attaches To Government Furnished Design Specifications, 31 Fed. B. J. 291, 299 (1973). The rule applicable here we believe is, if the contractor, from a stance of superior expertise, asks for and obtains leave to perform according to methods defined and stated by him, he impliedly warrants that he is able to overcome the technical difficulties inherent in the project, whatever they are. As viewed in Austin, the rule is the logical and natural converse of the one respecting Government drafted specifications, the more usual case. Thus it makes no difference whether performance here was actually and literally impossible or so only in the qualified sense sanctioned in our decisions.
The “implied warranty” concept appears frequently in our oases, yet may be objected to as somewhat of a fiction: a semantic step towards a predetermined result. We may, with Judge Wright in Transatlantic Financing Corp., supra, prefer to think in terms of the reasonable expectations of the commercial community, when they have written no express provision for the contingency, and none is derivable from commercial custom. It is needless to theorize here, when the cases dictate the outcome so clearly.
The trial judge saw some conflict between Austin, and Natus both supra. He fails to mention, however, the favorable *828citation of the former in the Wegematic and Bethlehem cases, the latter more recent than Natus, and the explanation and differentiation of Austin in Hoi-Gar, phrased in a -way to raise no doubt upon its correctness. If we held for plaintiff here, we would have to overrule both Austin and Bethlehem. The trial judge’s basic error is, we think, his failure to accept the court’s position plainly stated in Austin, that even if the impossibility of performance was absolute it was no excuse for Austin’s failure to produce under specifications written by itself.
If the parties, with equal expertise, mutually agree on an impossible method of performance, the loss sometimes may be shared on a mutual mistake theory. National Presto Ind. v. United States, 167 Ct. Cl. 749, 338 F. 2d 99 (1964) cert. denied, 380 U.S. 962 (1965), (citing and distinguishing Austin). The plight of the contractor here, a small business, invites sympathy, though it escapes without any liquidated damages or reprocurement charges. Nevertheless, the mistaken business and technical judgment of Dr. Woodson emerges clearly from the ASBCA’s findings as the prime cause of the fiasco, and no compai’able errors are imputable to defendant’s officers, with their inferior expertise admitted. We think the commercial community would expect the loss herein to lie where it fell, even assuming, arguendo, they would have gagged at any kind of penalty assessment. It makes a difference, as Judge Friendly remarks in Wegematic, supra, who is suing whom for what. In that case liquidated damages and excess reprocurement were awarded, making the decision far more hardhearted than Austin, or ours herein. Our decision, contrary to the trial judge, will deter the making of only It & D contracts wherein the contractor also promises production. On the other hand, 'his conclusion would deter Government officials from ever relying on the superior expertise of a contractor.
The ASBCA decision is supported by substantial evidence, is not arbitrary or capricious, and embodies no error of law, therefore, by Wunderlich Act standards, it is binding here. Plaintiff’s motion for summary judgment is denied. Defendant’s motion for summary judgment is allowed. The petition *829is dismissed. Since defendant recovered the progress payment of $22,900 and interest thereon by withholding and setting off payments under another contract, no action with respect to such progress payment is here required.
We acknowledge the assistance rendered us by Trial Judge C. Murray Bernhardt’s able and thoughtful Rule 166(c) opinion. Agreeing with him in part only, we reach a different end result.