Respondent determined deficiencies in petitioner’s corporate income tax as follows:
Year Deficiency
1976 . $519,726
1977 . 359,422
1979 . 421,040
1980 . 1,361,312
After concessions, the issues for decision are (1) whether, during the years in issue, petitioner was entitled to use the completed-contract method of accounting for three of its divisions, and, if so, (2) whether certain contracts must be severed by delivery in order to clearly reflect income.
*342FINDINGS OF FACT
Some of the facts have been stipulated, and the facts set forth in the stipulation are incorporated in our findings by this reference. Although set forth in the present tense, facts found are determined only as of the years in issue. Petitioner Sierracin Corp. is a Delaware corporation with its principal offices and principal place of business in Sylmar, California.
Petitioner has five product divisions that are relevant to this case: the Sylmar division, the Transtech division, the Magnedyne division, the Harrison division, and the Thermal Products division. The Sylmar division (Sylmar) manufactures aircraft transparencies. The Transtech division (Transtech) manufactures laminated security glazings. The Magnedyne division (Magnedyne) manufactures electromagnetic devices, principally DC (direct current) torque motors, DC tachometers, low inertia motors, and tachometers. The Harrison division manufactures hydraulic seals and fittings. The Thermal Products division manufactures heating elements, heat controllers, and related products. Petitioner has not disputed respondent’s determination with respect to Harrison and Thermal Products.
Accounting Methods
For financial purposes, petitioner reports sales and cost of sales at the time of shipment with losses recognized at the time such losses are identified. Prior to 1979, petitioner used the same method for reporting sales and cost of sales for tax purposes as it used for financial purposes.
In 1979, petitioner applied for permission from the Internal Revenue Service (IRS) to change to the completed-contract method of accounting. Permission was granted in a letter stating that “the determination as to whether or not the taxpayer’s contracts fall within the definition of section 1.451-3 of the regulations * * * [is] subject to verification of the District Director upon examination of its return.”
On its 1979 Federal income tax return, petitioner employed the completed-contract method of accounting for long-term contracts that were entered into in, or prior to, 1979, and that had not been completed by the end of that year. Petitioner continued to use the completed-contract *343method in 1980 and 1981. The conversion from the accrual method to the completed-contract method was reflected on the Schedule M-l to petitioner’s 1979, 1980, and 1981 returns.
Sylmar Division
Petitioner’s Sylmar division manufactures advanced technology transparencies (windshields, canopies, and windows) for military and commercial aircraft. Although some aircraft transparencies are relatively simple products, the Sylmar division does not manufacture such transparencies.
In generic terms, each transparency consists of three and sometimes four major components. All transparencies made by Sylmar have structural plies, interlayers, and coatings. Some transparencies also have face plies. The structural plies determine the shape and strength of the transparency. Most transparencies have two or more structural plies. The interlayers, which have adhesive qualities, enable the structural plies to be laminated together. The coatings are applied to surfaces of the structural plies and the interlay-ers for a wide variety of reasons, including adhesion, hardness, heat conductivity, electrical conductivity, abrasion resistance, ultraviolet and infrared radiation protection, and radar deflection. The face plies are attached to the exterior of the transparency for additional strength or abrasion protection.
The transparencies manufactured by Sylmar are vital structural components of the aircraft in which they are installed. Failure of the transparency can cause the aircraft itself to fail. Accordingly, the performance requirements placed on the transparency, and hence its design, are functions of the mission of the aircraft. As the mission of the aircraft becomes more complex, so too do the design requirements of the transparency.
Each transparency within a particular aircraft performs different functions and has individual design requirements. For example, if an aircraft has a right and left forward windshield and a right and left side panel, each of the four transparencies must be individually designed. In many instances the difference in design requirements will result in *344there being different manufacturers for different transparencies within the same aircraft.
The design of a transparency begins with preliminary discussions between Sylmar and the aircraft manufacturer, proceeds through a preproduction and prototype phase, and culminates in the award of a manufacturing contract. The first formal piece of paper generated in an aircraft transparency program is a request for quote, or RFQ, directed by the aircraft manufacturer to Sylmar’s marketing department. The RFQ typically contains information regarding the specifications for the transparencies and the anticipated quantities. Upon receipt of an RFQ, Sylmar initiates a TSR (technical sales request), an internal document which is sent to all groups within the company that are affected by the potential contract. The TSR outlines the proposed transparency production program and asks all affected areas in the company to respond with information that will allow Sylmar to submit a proposal to the customer.
The RFQ submitted by the customer to Sylmar outlines the size of the total production program as defined by the customer. Sylmar’s response is based upon the customer’s definition of the size of the total program. Sylmar’s cost estimates and pricing proposals are dependent upon the size of the program and the rate of production as defined by the customer.
Sylmar and the customer then typically enter a contract that establishes the basic terms for the entire program and defines the framework for doing business for a period of years, often 5 or more. The terms of this contract may or may not contain price terms. If prices are set forth, those prices are dependent upon the customer’s definition of the entire program. Typically, delivery schedules for transparencies are arranged to accommodate the customer’s production rate for the aircraft. The price of transparencies is based on the total quantity called for in a program, and is not based upon the quantity being shipped to the customer in any specific delivery. If the price were determined on a per-shipment basis, the price for the transparencies would be substantially higher.
Concurrently with the execution of the basic contract, the customer usually releases the initial purchase order under *345the contract. This purchase order contains fixed price terms that reflect the customer’s initial definition of the size of the program and the initially estimated rate of production. Most production programs involve a single basic contract followed by the release of multiple purchase orders. Typically, purchase orders cover a time period that is shorter than the period covered by the underlying contract, because it is not practical to establish fixed prices for the entire anticipated duration of a production program.
Upon the receipt of a purchase order, Sylmar issues an internal work order that reflects the terms and conditions of the purchase order. The work order is sent to every group within the company that will be involved in the production of the transparency. The work order constitutes authorization to commence work, and upon receipt, everyone acts in reliance on it. Sylmar maintains no inventory; it does not commence production of transparencies prior to the receipt of purchase orders from customers.
Throughout a program, Sylmar is in constant contact with the customer. Subsequent purchase orders based upon these ongoing discussions and negotiations will be released by the customer on a periodic basis. These purchase orders reflect a mutually agreed upon assessment of the progress of the program. Each such purchase order contains fixed price and delivery terms and the actions taken by Sylmar are identical to those taken in response to the initial purchase order: no work is commenced prior to its receipt; upon receipt, a work order is issued; upon issuance of the work order, all affected groups in the company act in rebanee thereon. For Federal income tax purposes, Sylmar treats each work order as a contract.
The preproduction and prototype — also known as “development” — phase for a transparency of the type manufactured by Sylmar normally takes 2 or more years and involves between two and four people working full time. During the preproduction and prototype phase, the prototypes are fabricated within the engineering department. During this phase, there are no formal manufacturing instructions, no formal tooling, and no formal process definitions. Sylmar is inventing as it produces the prototype. New processes are devised and new toobng is devel*346oped. The goal of this stage is to develop a transparency that meets the performance requirements described by a particular customer. These requirements typically cut across a broad spectrum of widely disparate capabilities, and can include the ability to sustain cabin pressure, carry structural loads, withstand varying degrees of bird impact, withstand gunfire, withstand extreme heat, de-ice, de-fog, provide high optical quality despite complex, curved shaping, reduce radar cross-section, and provide resistance to a variety of military threats such as nuclear flash, lasers, and microwave weaponry. Volume — that is, the ability to produce more than a minimal number of units per month — is not a consideration.
Transparencies designed for a specific opening in a specific aircraft are generally interchangeable. Transparencies designed to fit one aircraft, however, will not fit in another. For example, a windshield in a L-1011 will not fit a 747 and vice versa.
The award of a production contract requires Sylmar to move from the preproduction and prototype phase to the production phase for a transparency. Even when Sylmar has been able to produce a transparency during the development phase, the fabrication techniques and tooling used in the development phase are unsuitable for actual production. There will have been little or no development of the skills required for production, the documentation necessary for production will be incomplete, and the tooling will be inadequate. The transition from development to production requires the formulation of detailed manufacturing instructions for each product.
The manufacturing process generally involves the following procedures: (1) Stretching, shaping, and forming the structural plies; (2) applying interlayers to the structural plies; (3) applying coatings to the surfaces of the structural plies and/or the interlayers; (4) laminating the structural plies; (5) attaching and/or coating the face ply, if required; (6) cutting the transparency to the proper size; and (7) installing the transparency in the frame that will be attached to the aircraft.
When a transparency program moves from the preproduction and prototype phase to production, Sylmar knows that *347significant production problems will be encountered but is unable to predict the nature of the problems, the severity of the problems, the frequency of the problems, whether solutions can be devised or how costly the solutions will be. The actual manufacturing process for any given transparency consists of literally hundreds of steps. Each step must be carried out in the proper: sequence with exact precision as specified in detailed manufacturing instructions. The manufacturing process and procedures are different for each design. Each design manufactured by Sylmar requires the manufacture of custom-designed forms, molds, dies, and tools. These custom-designed items cannot be used on any other design.
Because the production techniques for each transparency are specifically designed for that particular transparency, the training that production people receive is lost if production is interrupted for a significant amount of time. Because of the highly intricate fabrication process, there is no classic “learning curve” that can be applied to the manufacture of Sylmar’s aircraft transparencies. Because of outside factors over which Sylmar has no control — for example, the quality of raw materials, the environment, the temperature, the humidity, etc. — that can dramatically affect Sylmar’s costs, Sylmar cannot predict at the outset of a given contract the rate at which learning will take place, or whether learning will take place at all.
The designs of transparencies change over the life of an aircraft program. There are several reasons for such change, all of which are present throughout the life of the aircraft. First, the transparency developed during the preproduction and prototype phase will inevitably have to be changed once the transparency is placed in service, because immediately upon being placed in service, the demands imposed by actual in-flight use will demonstrate areas requiring design changes. Second, the design developed during the preproduction and prototype phase will be impractical, inefficient, or perhaps impossible to produce on a continuing basis. Third, actual production will demonstrate areas in which design changes will improve the performance of the aircraft. Fourth, as new technology is developed, the aircraft manufacturer will want to incorporate that technol*348ogy in the transparency. Fifth, long-term in-service use of the transparency will reveal areas in which, design changes are required. Finally, if the mission of the aircraft is changed, it will be necessary to change the design of the transparency.
Each change in the design of a transparency requires changes in the custom-designed tools and in the manufacturing process. Design changes also require the development of new fabrication techniques. These new procedures must be explained to the technicians responsible for that phase of the production process.
In order to manufacture transparencies, Sylmar has developed a broad range of unique proprietary products including: structural plies, interlayers, face plies, adhesives, and more them 170 different coatings. Sylmar has also developed a significant number of proprietary manufacturing procedures that enable it to manufacture transparencies that cannot be duplicated by other manufacturers. With the exception of proprietary products that have been developed by Sierracin, none of the materials used in the manufacture of transparencies has been designed for the purpose of manufacturing transparencies.
Each of the structural ply and interlayer materials used by Sylmar is a highly reactive and inherently unstable organic compound. A very slight variation in either the chemical composition of these products or the method by which these products are produced can cause a significant alteration in how the product reacts during the transparency manufacturing process, resulting in defective transparencies.
Sylmar inspects and/or tests all of the structural ply and* interlayer material purchased from unrelated vendors upon receipt. Urethane, an interlayer material manufactured by Sierracin, is inspected and tested as it is produced. Notwithstanding continuous testing and inspections, it is not possible to detect defective products until the material has become part of a transparency that is being manufactured. While some defects become apparent at an early stage in the manufacturing process (e.g., forming or initial lamination), others do not appear until at or near final inspection, *349and others are not apparent until the transparency is actually placed in service by the customer.
From start to finish, the transparency manufacturing process is subject to risks of failure which occur at random and which are not subject to Sylmar’s control. The sources of risks that can cause product failure during the manufacturing process include the following: Unpredictable reactions of the organic materials that comprise the transparency to the environment (e.g., temperature or humidity) or to each other, varying quality of raw materials or Sylmar’s inability to obtain satisfactory raw materials (e.g., a sole source supplier discontinuing production), and undetectable contaminants or pollutants. These failure-inducing elements can and do occur at all stages in the life of a program. Sylmar has never had a transparency manufacturing program that was trouble-free.
Only four other companies in the United States compete with Sylmar in the manufacture of aircraft transparencies. Most of the transparencies manufactured by Sylmar are not manufactured by any other transparency manufacturer. Although the transparencies that are manufactured by Sylmar’s competitors are interchangeable in the aircraft, it is unlikely that any two would have exactly the same cross-section. Even if the cross-sections were similar, no two would be manufactured in the same manner. Moreover, if another manufacturer had access to Sylmar’s manufacturing instructions, it would not be possible to manufacture a transparency identical to that made by Sylmar because the manufacturing process is a skilled art that requires the unique skills of particular individuals.
Transtech
Transtech is a manufacturer of security windows, known as “security glazings,” which are designed to defeat various threats that are perceived as a result of there being a window in place rather than a door or a wall. Transtech serves two primary markets, the architectural market and the vehicle market. Transtech was formed as a separate division of Sierracin in late 1979. Prior to that time, the business of Transtech was conducted by Sylmar in what was known as the commercial products department.
*350Three basic materials are used in making security glaz-ings: glass, acrylic, and polycarbonate. Each of the basic materials has its own strengths and weaknesses. Glass, for example, is particularly resistant to marring, scratching, chemicals, and fire. It is, however, a brittle material and even when strengthened can be defeated by a simple blow of a hammer or sharp instrument, such as a screwdriver. Polycarbonate, unlike glass, is highly impact resistant, but it is also more susceptible to marring, scratching, chemical attack, flame, and heat. Transtech attempts to combine the strengths of glass and polycarbonate by producing a glass-clad polycarbonate, using urethane as an interlayer.
During the years in issue, about 75 to 80 percent of Transtech’s sales were to the architectural market, with the remainder to the vehicle market. The architectural market consists of five broad submarkets which, in descending level of security, include: (1) U.S. Government and military installations (e.g., embassy safe havens, military bases, etc.); (2) retention-detention (e.g., courts, prisons, detention centers, forensic hospitals, etc.); (3) psychiatric hospitals; (4) general industrial and financial institutions (e.g., banks, machine tool guards, etc.); and (5) general corporate and retail (e.g., retail storefronts, jewelry store cases, computer centers, etc.). The vehicle market also consists of five broad submarkets: (1) Military vehicle armor; (2) financial armor (e.g., discrete vans, bullion carriers, armored trucks, etc.); (3) discrete commercial armor (e.g., limousines); (4) marine; and (5) railroads.
The threats that security glazings must protect against and the performance requirements that security glazings must meet vary widely in both nature and degree, not only between the architectural and vehicle markets and across each of the different submarkets within the architectural and vehicle markets, but also within each submarket by customer and even, for a given customer, according to the location of the security glazing within the customer’s facility or vehicle and according to the size of the glazing.
Transtech maintains sales catalogs describing generic classes of Transtech products designed to provide different levels of security protection. In ascending level of protection, these generic classes of products are Sierralite, an *351all-glass laminate, and Omniguard, Omnilite, and Omniarmor, which are all laminated composite glazings of glass and polycarbonate.
Since 1979, Transtech has produced hundreds of different cross-sections of security glazings for various applications. The cross-sections of security glazings that Transtech produces and sells can be and typically are varied or changed from the specifications set forth in its sales catalogs along a number of different parameters in order to meet the specific needs and requirements of each Transtech customer. These parameters include: (1) The order of materials (i.e., the thicknesses and number of plies of the different materials that comprise the glazing); (2) the types of materials (e.g., various types of glass and interlayer materials); (3) the process considerations and manufacturing considerations (e.g., the types of processes and equipment used to form the product will result in variations in performance capabilities of the product); and (4) the numerous special options (e.g., one-way mirrors, solar reflective and tinted glasses, hard coating, speak holes, etc.). An infinite variety of cross-sections of security glazings is available, depending upon the particular requirements of the job.
For the most part, each customer is different from another, either in the specification requirement that is in the types and levels of security that must be provided, or in the cross-sections requirement of the product. Transtech’s security glazings for the architectural market must always be custom built because the cross-sections vary from job to job with respect to their performance requirements, and the length and widths of the windows are variable in increments of one-sixteenth of an inch, creating an infinite number of window sizes.
Bonding polycarbonate to glass is very difficult to accomplish technologically because polycarbonate, at a given temperature, expands and contracts approximately 9 times as much as glass. The stress caused by these different coefficients of thermal expansion manifests itself in four different product failures. Given too much stress: (1) The product will warp; (2) the glass will break; (3) the polycarbonate will crack; and (4) the bond between the *352interlayer and the structural layer will be violated (e.g., voids will open up) and delamination will occur.
In an attempt to overcome these product failures, Transtech has developed two proprietary urethane interlay-ers of Transtech’s own formulation, design, and manufacture that are unique to Transtech. Transtech guards the proprietary nature of these urethane interlayers very closely. During the years at issue, Transtech was the only company to use urethane in the manufacture of security glazings. The role of the interlayer is very important in the performance of a glass-clad polycarbonate security glazing. The urethane that Transtech uses as the interlayer for its glass-clad polycarbonate security glazings is the most expensive component of such glazings.
The urethane that Sierracin was using at the beginning of 1979 was known as S-120. At that time, Sierracin manufactured all S-120 urethane in-house. S-120 urethane was formed through a chemical reaction known as step-growth polymerization, which is a very slow process and causes a randomness to be associated with the urethane production that causes it to be inherently impossible to make the urethane absolutely reproducible each time it is made.
During the period from 1978 up through the 1980’s, there was only one supplier for two of the three component materials necessary to manufacture urethane suitable for use in Transtech’s glass-clad polycarbonate security glaz-ings. During the period from 1978 on into the 1980’s, there was no significant way for Transtech to monitor the quality of the raw materials that it was getting for the production of urethane. During this period, Transtech encountered substantial unforeseen and uncontrollable problems with respect to the production of urethane because of defects in the raw materials.
Urethane is a volatile organic material and is very reactive both with its environment, particularly the amount of moisture in the air, and with other materials with which it comes in contact. Thus the quality of the S-120 urethane that Sierracin was able to produce varied with the weather. During the years at issue, this sensitivity to the environment and surrounding materials resulted in numerous defects occurring in the urethane at various stages of the *353production process. Many of these defects were undetectable until the urethane had been combined with other layers of material in the manufacture of security glazings.
Because of these problems, Transtech decided to change the process by which its urethane was produced from a casting process to an extrusion process, and to cease the manufacture of urethane, instead obtaining urethane from outside suppliers. The suppliers that Transtech turned to had never produced a polyurethane that was required to meet the types of optical requirements associated with the security glazings produced by Transtech. The first delivery of the product by the suppliers to Sierracin in March 1979, as well as subsequent shipments through May 1979, were completely unusable and resulted in increased costs to Transtech.
Because of the volatile way in which urethane reacts with the environment, the deliveries of urethane that Sierracin received through the balance of 1979 varied widely in quality, leading Sierracin to attempt to develop a new chemical formulation for the production of urethane. This reformulation was designated as S-121 by Sierracin. Transtech began using S-121 urethane in its production of security glazings in July of 1981. At this time, Sierracin also ceased the purchase of urethane from outside suppliers and again began producing urethane itself. Unpredictable and uncontrollable problems regarding the production of usable urethane continued through the 1980’s. These problems were so severe that Transtech had to shut down its production of security glazings in 1984.
In addition to urethane, other volatile and reactive organic materials involved in the manufacture of glass-clad polycarbonate security glazings include diamino syline primers, silicon hardcoats, and polycarbonate. Transtech has had substantial unforeseen and uncontrollable problems, and incurred additional costs because of these problems, from the years in issue up to the present as a result of the unpredictable way in which these materials react both with the environment and with each other. These problems are compounded by the limited number of suppliers of these raw materials.
*354During the years at issue, Transtech was the only company in the world to manufacture glass-clad polycarbonate security glazings. Prior to 1979, General Electric and PPG Industries attempted to supply glass-clad polycarbonate security glazings, but had abandoned the business by 1979 because of their inability to produce a satisfactory product.
Although certain similarities exist with respect to the steps involved in the process of manufacturing any particular security glazing sold by Transtech, the specific steps involved in processing and the equipment used to produce security glazings varies according to the specific product being produced. Transtech’s manufacturing process cannot be automated in any significant way because the plethora of sizes, shapes, and cross-section variables required by different customers do not permit automation. Because of the security nature of the glazing products that Transtech produces, such products cannot be cut to shape after manufacture but must be configured first and then assembled.
With respect to sales of architectural glazings, Transtech begins communication with the architect as soon as Transtech becomes aware of the construction of a facility that is likely to require security glazings. Since each facility has a different purpose and a different design influence, Transtech attempts to learn the types of different threats, as well as, for any given type of threat, the level of threat that the facility will be subjected to.
After Transtech becomes aware of the different types and degrees of performance requirements that the security glazings for a given job must meet, a series of bidding phases takes place. First, the architect will select a general contractor for construction of the project. Transtech will bid to the general contractor in support of his bid to the architect based on the perceived levels of security and approximate square footages in total of each security level required for the job as known at that time. Such a bid will specify a certain number of square feet of a particular type of glazing at a particular price per square foot; for example, 3,000 square feet of glazing A at X dollars per square foot and 4,000 square feet of glazing B at Y dollars per square *355foot. Such a quotation is known as an “area price,” and allows the general contractor to estimate his total costs.
Next, Transtech bids to a glazing contractor, who is a subcontractor of the general contractor, in support of the glazing contractor’s bid to the general contractor. The glazing contractor is responsible for the installation of the glazing system, including the frames, the installation of the glazings in the frames, etc. The glazing contractor is Transtech’s customer.
After the general contractor selects one from among the competing glazing contractors to be responsible for the glazing system, the glazing contractor resolicits bids from security glazing manufacturers like Transtech. This process typically takes approximately 1 year.
If Transtech is selected by a glazing contractor to provide glazings for a particular architectural job, a purchase order known as a “blanket PO” or a “block PO” is awarded to Transtech, specifying a quantity, in total square feet, of each type of glazing that is required, and a per-square-foot price for each type of glazing. The purchase order also lists an estimated delivery date and an estimated number of windows. The purchase order does not define the sizes of the windows that Transtech will be called upon to produce.
After the building has been constructed and settled, each individual window opening is measured, and Transtech is informed by the glazing contractor what the size of each particular window must be. Transtech then begins manufacture upon receipt of a final-size release against the purchase order.
At the time at which Transtech bids prices to supply product for a given job, it does not know either the sizes of the windows or the total volume of product that it will be required to produce for that job. Transtech orders or manufacturers many of the materials that it uses to produce the glazings for any particular job based on the glazing contractor’s estimate of the total number of square feet of product that will be required. This estimate is known as the “take-off.” Transtech’s dependence on the take-off subjects Transtech’s ability to predict the costs associated with any given job to substantial uncertainty.
*356If the glazing contractor overestimates the total volume of product that is required for a given job, Transtech’s costs are increased because Transtech has set aside manufacturing time and established a manufacturing capability to produce the volume of product estimated in the take-off in the time frame set forth on the purchase order. If that manufacturing time and capability is not fully utilized, Transtech incurs unnecessary costs. Furthermore, since Transtech orders certain materials from its vendors and itself manufacturers other materials based on the take-off, changes in the take-off increase Transtech’s material costs.
If the glazing contractor underestimates the total amount of product that is required for the job, Transtech’s costs are adversely affected because of the difficulties that Transtech encounters in providing the increased quantity of product necessary by the due dates for completion of the job. Since the products supplied by Transtech are some of the last products that go into a newly constructed building prior to doing final work on the interior of the building, such as painting and laying carpet, Transtech is always under substantial pressure to meet the due dates associated with a particular job. Neither Transtech nor the glazing contractor has any control of the completion dates that they have agreed to meet when they are awarded the business. If Transtech has reserved only enough raw materials and enough time in the factory in its schedule system for the volume indicated in the initial quotation and award, then, in order to accommodate an increased volume, Transtech must pay premiums to acquire raw materials and incur manufacturing costs associated with the inefficiencies of rescheduling production.
The commercial exigencies of supporting the glazing contractor and the general contractor require that, as a matter of practical business necessity, Transtech incur these costs if necessary in order to ensure that the job is completed on time. If Transtech fails to meet the required delivery schedule, including accommodating schedule changes and volume changes, Transtech is not only subject to a contractual penalty, but also faces the loss of substantial future business because of its failure to support the glazing contractor and the general contractor.
*357The final risk over which Transtech has no control is product rejection by the architect during his “walk-through” of the facility. An architect’s walk-through is an inspection by the architect of the glazings that have been installed. There is no objective standard governing the amount of glazings that the architect can reject on his walk-through; it is totally within the subjective discretion of the architect.
All the risks and uncertainties that affect the construction industry generally by causing schedule changes in the construction of a building affect Transtech and Transtech’s costs. Such schedule changes may take the form of unexpected delays, acceleration requirements (because of a need to complete a portion of the building or construction project sooner than originally expected), or volume changes.
Except in unusual circumstances, Transtech opens one work order for any given job, which work order covers all the product that Transtech needs to supply for that particular job. Transtech maintains no inventory. If income is received and attributable to a given work order, such income represents payments for manufactured product. As a practical matter, because of the risks inherent in the nature of Transtech’s business, Transtech cannot calculate with any degree of certainty the total costs and hence the total profit associated with any given job at any time prior to the final completion of that job. For Federal income tax purposes, Transtech treats each work order as a contract.
Magnedyne
Magnedyne produces DC torque motors and tachometers. It also has several other products: brushless DC motors, low-inertia motors, and AC (alternating current) motors. Torque motors are used for positioning things such as radar antennas, CAT scanners, ultrasound scanners, and large trackers. Tachometers are used for measuring the rate of rotation of items such as radar antennas, tracking systems, and machine tools.
Magnedyne does not maintain an inventory of products. Each motor or tachomoter is manufactured for a specific customer pursuant to a specific purchase order from that customer. Each device manufactured by Magnedyne is *358custom designed for a specific application, and each application requires a different design.
Magnedyne publishes a catalog that describes the electromagnetic devices it manufactures. The products described in the catalog are actual custom-designed products that have been produced for specific customers. Magnedyne does not maintain any products in inventory, and Magnedyne has never sold a product to a new customer that was identical to a product described in the catalog.
Magnedyne does not commence work on a product until it receives a purchase order from a customer. Upon receipt of a purchase order, a work order is opened. If another purchase order is received, another work order is opened. If the purchase order involves a new motor, a design is prepared, using the assistance of a computer program. The design is sent to the mechanical design and engineering departments which draw the unit, determine if it is a cost-effective design, and build a prototype.
The successful completion of a prototype is no assurance that a model may be produced. Production requires the design and manufacture of custom tooling, forming fixtures, test equipment, test fixtures, bonding fixtures, and molds for each different model. Because each unit is made by hand and because high performance characteristics require designs with very close tolerances, human error is the source of most problems encountered in manufacturing.
Magnedyne does not produce a sufficient quantity of a unit of a specific design to enable the manufacturing personnel to work on that unit on a continuing basis and thereby maximize and maintain the optimum level of skill required to produce the product. Instead, production personnel are constantly shifting between products and always learning new skills and techniques. The skills required to manufacture Magnedyne’s products are not skills which are otherwise available in the marketplace. As a consequence, Magnedyne must individually train all of its manufacturing personnel.
Magnedyne prices its products on a contractual basis, which is a bid based on the entire quantity ordered by the customer. Even though Magnedyne provides a per-unit-price quotation to its customers, Magnedyne does not *359expect each unit it manufactures to cost the same amount to manufacture. Magnedyne experiences large month-to-month cost variations which it expects to average out over a given order. These variations are attributable to variations in yield, variations in labor rates, and variations in skill levels of manufacturing personnel. For Federal income tax purposes, Magnedyne treats each work order as a contract.
Each model of motor or tachometer produced by Magnedyne carries a model number; for example, 402-49C, 707-01A, or 244-04N. The last two digits designate the design variant. The last letter designates the winding. The five models responsible for the largest portion of the adjustments during the years in issue were the 244-04N, 402-49C, 435-05A, 444-10, and 707-01A. Petitioner has conceded the adjustment with respect to Model 707, on the grounds that this model was manufactured by a Japanese firm and purchased and sold by Magnedyne.
Model 244-04N is a torque motor for use in a CAT scan machine. During 1979, 1980, and 1981, Magnedyne was issued three purchase orders for this model by one customer. Per-unit prices under the purchase orders increased during this period from $6,800 to $10,300. Model 244 is illustrated at page 50 of Magnedyne’s catalog.
Model 402-49C is a tachometer. It is mounted on the back of a servo motor. Magnedyne has made this tachometer since about 1971. During 1979, 1980, and 1981, Magnedyne received six purchase orders for Model 402-49C from three different customers. The per-unit prices under these six purchase orders ranged from $61 to $98.50. Model 402 is depicted at page 55 of Magnedyne’s catalog. During the years in issue, Magnedyne received purchase orders from other customers for the production of other design variants of Model 402.
Model 435-05A is a tachometer. Magnedyne has made this tachometer for more than 12 years. During 1979, 1980, and 1981, Magnedyne received at least nine separate purchase orders for Model 435-05A from three customers. Per-unit prices under the various purchase orders ranged from $72 to $175. Model 435 is depicted at page 53 of *360Magnedyne’s catalog. During the years in issue, Magnedyne manufactured variants of Model 435 for other customers.
Model 444 is a torque motor manufactured for use in a F-15 radar unit. During 1979, 1980, and 1981, petitioner received and completed three purchase orders for variants of Model 444. Per-unit prices ranged from $3,097 to $6,475. Model 444 is depicted in Magnedyne’s catalog at page 46. During the years in issue, Magnedyne manufactured variants of Model 444 for other customers.
Magnedyne uses Alnico and Samarium cobalt magnets in its products and is dependent upon unrelated third parties to supply these magnets. There is only one supplier of the Samarium cobalt magnets, a Japanese company, and only 8 to 10 vendors who can supply Alnico magnets. These magnets gire subject to significant, unpredictable price fluctuations.
Magnedyne must also rely on outside subcontractors to provide raw materials and to perform critical portions of the unit assembly process. Traditionally, payments to subcontractors have constituted about 35 percent of Magnedyne’s costs for its products. Over the years, in relying on subcontractors, Magnedyne has experienced quality control problems and delivery problems. The incidence of these problems has been random and unpredictable.
OPINION
I. Completed-Contract Accounting
A. Introduction
Section 1.451-3(d)(l), Income Tax Regs., provides that a taxpayer using the completed-contract method may defer recognition of income and deduction of expense until the relevant contract is completed and accepted. See Reco Industries, Inc. v. Commissioner, 83 T.C. 912, 921 (1984); Peninsula Steel Products & Equip. Co. v. Commissioner, 78 T.C. 1029, 1046 (1982). Only certain “long-term” contracts may be accounted for under the completed-contract method. Sec. 1.451-3(a)(1), Income Tax Regs. A manufacturing contract is a “long-term” contract if it involves the manufacture of “(A) unique items of a type which is not normally carried in the finished goods inventory of the taxpayer, or *361(B) items which normally require more than 12 calendar months to complete.” Sec. 1.451-3(b)(l)(ii), Income Tax Regs. Petitioner maintains that its products are “unique items” within the meaning of the regulation; respondent contends that they are not.
B. Uniqueness
Petitioner initially contends that items not held in finished goods inventory are for purposes of the regulation “unique.” We disagree. As a matter of grammatical construction, the phrase “which is not normally carried in the finished goods inventory of the taxpayer” modifies the singular word “type,” not the plural word “items.” Thus, as a prerequisite for completed-contract accounting, a unique item must also be of a type not normally carried in finished goods inventory. The absence of such inventories in this case is therefore not determinative of whether petitioner’s products qualify under the regulation.
Petitioner next asserts that the best meaning to be given to the word “unique,” as used in the regulation, is “unusual” or “notable.” See Webster’s Third International Dictionary 2500 (1976). Respondent argues that the word should be interpreted to mean “being the only one; sole.” See Oxford English Dictionary U-235 (1933). We disagree with both parties. Petitioner’s proposed definition is too broad and subjective, and respondent’s is too narrow and restrictive. Petitioner’s proposed definition would permit the use of the completed-contract method to account for most multi-unit manufacturing contracts. Technically unsophisticated products, such as the pet rock, hula hoop, or Frisbee, could be described as “unusual” or “notable.” Respondent’s proposed definition would bar use of the method to account for any multi-unit manufacturing contract. Numerous cases suggest that this approach is incorrect. See, e.g., Spang Industries, Inc. v. United States, 791 F.2d 906 (Fed. Cir. 1986) (electrical systems and steel bridge components); Stephens Marine, Inc. v. Commissioner, 430 F.2d 679 (9th Cir. 1970), affg. a Memorandum Opinion of this Court (three minesweepers); Reco Industries, Inc. v. Commissioner, supra (steel tanks and pipes); Peninsula Steel Products & Equip. Co. v. Commissioner, supra (pollution control equipment); *362see also section 1.451-3(b)(2)(ii)(B), Income Tax Regs, (five aircraft). Accordingly, we do not limit ourselves to any definition set forth in a particular dictionary. Instead, we focus on the meaning of the regulation and the facts and circumstances of this case in light of the justification for the completed-contract method of accounting.
Respondent admits that “a multi-unit contract may still qualify if it exhibits characteristics which justify the use of the completed contract method.” One such characteristic provides an important element of our construction of the word “unique.” Section 1.451-3(b)(l)(ii), Income Tax Regs., states as follows:
a contract to manufacture a unit of industrial machinery specifically designed for the needs of a customer and not normally carried in the taxpayer’s inventory or a contract to manufacture machinery which will require more than 12 calendar months to complete are long-term contracts within the meaning of this subparagraph; however, a contract to manufacture 15,000 folding chairs which take 3 days each to manufacture is not a long-term contract within the meaning of this subparagraph even though it takes more than 12 calendar months to manufacture all 15,000 chairs and the contract is not completed within the taxable year it is entered into.
Thus, custom-designed machinery is “unique”; folding chairs are not. The regulation suggests that this term may describe items designed for the use of a particular customer. See sec. 1.451-3(a)(l), Income Tax Regs, (reference to “special order” manufacturing contract). This approach to the word “unique” is consistent with the decided cases. In several cases, the taxpayer’s use of the. completed-contract method was not challenged where the manufactured items were designed for the needs of a specific customer. Reco Industries, Inc. v. Commissioner, supra; Peninsula Steel Products & Equip. Co. v. Commissioner, supra; Spang Industries, Inc. v. United States, supra. In at least one other case, use of the method was denied where the manufactured items were not shown to be designed for the needs of a specific customer. Schloegl v. Commissioner, T.C. Memo. 1986-440 (4,000 steel brackets). Accordingly, the degree to which petitioner’s products were custom-designed is of special importance in deciding whether they are “unique.”
*363C. Unpredictability
Petitioner argues that “completed contract accounting is based upon one very plain and simple premise, exposure to risk — risk which creates sufficient uncertainties to make it difficult to predict the outcome of the contract on an interim basis.” Respondent agrees that “risk is clearly a factor.” Respondent maintains, however, that risk “is relevant only to the extent it affects the taxpayer’s ability to reliably estimate revenues. The mere fact that a taxpayer faces the possibility of strikes, fires, and floods is of no significance if the taxpayer can accurately portray contract results on an interim basis.”
These positions of the parties are not so far apart. The parties disagree over the amount of risk, i.e., unpredictability, required for completed-contract accounting. Petitioner seems to argue that completed-contract accounting is appropriate whenever rehable estimates of future income and expense are fraught with difficulty. Respondent apparently contends that the completed-contract method is appropriate only when such estimates are impossible.
Respondent first contends that petitioner’s exposure to risk is but one of a number of factors to be considered in determining petitioner’s eligibility for completed-contract accounting. Respondent maintains that the true test of eligibility for completed-contract accounting is the degree to which the manufacturing process is “like construction.” This construction of section 1.451-3(d)(l), Income Tax Regs., is built upon an uncertain foundation. The cornerstone of respondent’s edifice, Peninsula Steel Products & Equip. Co. v. Commissioner, supra, is a case in which eligibility for completed-contract accounting was not at issue. In Peninsula Steel, the taxpayer accumulated certain costs associated with long-term contracts in work-in-process inventory accounts. The taxpayer used LIFO to value its inventories. The Commissioner determined that the taxpayer used the completed-contract method to account for income from its long-term contracts, and he argued that taxpayers using the completed-contract method could not properly account for costs of long-term contracts using inventories and, in particular, using LIFO. Although the taxpayer maintained that it relied on the accrual-shipment method of accounting, *364we held that the taxpayer failed to prove that it did not use the completed-contract method. In upholding the Commissioner’s determination that the taxpayer used the completed-contract method of accounting, we focused on whether the taxpayer in fact used that method. The dispute as to the taxpayer’s eligibility for completed-contract accounting did not involve the nature of its product. See Peninsula Steel Products & Equip. Co. v. Commissioner, 78 T.C. at 1046 n. 26.
In Peninsula Steel, the ultimate issue was whether the taxpayer’s use of LIFO-valued inventories to account for costs associated with long-term contracts clearly reflected income. After finding that the taxpayer did, in fact, use completed-contract accounting, we held that the taxpayer had properly used LIFO to value its inventories, first observing that the taxpayer’s use of the completed-contract method clearly reflected income:
Petitioner and Monotech are manufacturers of large products, typically made to conform to the customer’s specifications. Because of their enormous sizes, these products are manufactured in components and shipped in kit form. Only after assembly at the customer’s site can it be determined that the product conforms to the customer’s specifications; at times, petitioner and Monotech are required to perform additional work at the site before the customer accepts the product. A significant portion of the work performed by petitioner and Monotech during the years in issue related to long-term contracts requiring advance payments during the course of manufacturing. From an overall standpoint, it is difficult for petitioner and Monotech to accurately estimate the total costs of performing a long-term contract until the customer accepts the product. We believe (and respondent does not dispute) that the completed contract method of accounting is particularly useful for matching revenue and expenses under these circumstances. Fort Pitt Bridge Works v. Commissioner, supra. [24 B.T.A. 626 (1931).] [Peninsula Steel Products & Equip. v. Commissioner, 78 T.C. at 1047.]
From our analysis in Peninsula Steel, respondent derives five factors that arguably limit the use of long-term contract methods of accounting to those manufacturing contracts that share the characteristics of construction-type contracts.1 Our comments in Peninsula Steel, however, *365amounted to no more than a description of one situation in which completed-contract accounting was appropriate. Because we did not consider the taxpayer’s eligibility for long-term contract accounting under section 1.451-3(b)(l)(ii), Income Tax Regs., we did not attempt to set forth an exhaustive list of the characteristics of long-term manufacturing contracts. Our opinion in Peninsula Steel does not support respondent’s “like construction” test.
Each of the “factors” derived by respondent from the text of our Peninsula Steel” opinion is but one facet of a single concept — unpredictability. Each factor must, after all, bear some rational relationship to the purposes of completed-contract accounting. One such purpose is to alleviate the “bunching” of income faced by taxpayers engaged in long-term contracts. See 2 J. Mertens, Law of Federal Income Taxation, sec. 12.126, at 558 (1985 rev.). The other major purpose of the completed-contract method is to account for contracts where ultimate gain or loss cannot be accurately determined until the completion of the contract. Reco Industries, Inc. v. Commissioner, 83 T.C. at 921; Peninsula Steel Products & Equip. Co. v. Commissioner, 78 T.C. at 1046-1047; Fort Pitt Bridge Works v. Commissioner, 24 B.T.A. 626, 641 (1931), affd. on this issue 92 F.2d 825 (3d Cir. 1937). Each “factor” noted by the Court in Peninsula Steel related to the latter purpose. Peninsula Steel Products & Equip. Co. v. Commissioner, 78 T.C. at 1047. To state that ultimate gain or loss cannot be accurately determined until the contract is completed is simply to state that the contract is subject to unpredictable risks. The degree to which petitioner’s contracts were subject to such unpredictability is thus the other major factor to be considered in our determination.
D. Application
Under the facts and circumstances of this case, we conclude that the products of petitioner’s Sylmar and Transtech divisions are “unique items” within the meaning of section 1.451-3(b)(l)(ii), Income Tax Regs. In reaching this conclusion, we have placed particular emphasis on the *366degree to which petitioner’s products are designed for the use of specific customers. We have also considered the degree to which petitioner’s contracts are subject to unpredictable risks that make it difficult to account for ultimate profit or loss on an interim basis.
The items produced by Sylmar, Transtech, and Magnedyne are custom designed. The products of each division satisfied the highly specialized needs of specific customers. Each transparency produced by Sylmar is limited in use to a specific opening in a particular model of aircraft. Each glazing produced by Transtech can only be installed in a specific opening in a specific building. Many of the machines produced by Magnedyne are “one of a kind.” These products are not suitable for functions or customers other than those for which they were designed.
The contracts of Sylmar and Transtech are also subject to unpredictable risks that make it difficult to account for ultimate profit or loss on an interim basis. Sylmar’s manufacturing process is disrupted at unpredictable intervals by (1) difficulty in obtaining quality raw materials, (2) unforeseen chemical reactions of the finished materials that comprise Sylmar’s transparencies, (3) difficulty in moving from the development phase to the production phase, and (4) post-development design changes. Transtech’s manufacturing process is disrupted at unpredictable intervals by (1) difficulty in obtaining quality urethane and other materials, (2) the nonautomated nature of Transtech’s production process, and (3) Transtech’s dependence on the accuracy of the glazing contractor’s estimate, or “take-off.”
Petitioner has not shown that Magnedyne’s manufacturing process is subject to similar risks. Although production of a Magnedyne device requires raw materials that fluctuate in price, custom tooling, skilled craftmanship, and rehable subcontractors, we cannot conclude on this record that Magnedyne was unable to make reasonably accurate predictions of ultimate profit or loss. The record suggests otherwise. Many of the design variants produced by Magnedyne were only nominally “unique”; Magnedyne manufactured units sharing a basic design for many customers over a period of many years. Although Magnedyne expected month-to-month fluctuations in the costs of pro*367ducing a particular design variant, the division’s long experience in the production of the basic design led it to expect costs to average out over a given order. Nothing in the record suggests that Magnedyne’s attempts to estimate overall contract costs were unsuccessful or even particularly difficult.
We hold that petitioner properly used the completed-contract method to account for the contracts entered into by its Sylmar and Transtech divisions. Completed contract accounting was not appropriate for petitioner’s Magnedyne contracts.
II. Severance
In the alternative, respondent would limit petitioner’s use of completed-contract accounting by requiring that gain or loss be reported as each unit or combination of units is delivered by the Sylmar or Transtech division. As in effect during the years in issue, section 1.451-3(e), Income Tax Regs., provided in pertinent part as follows:
(e) Severing and aggregating contracts. (1) For the purpose of clearly reflecting income, it may be necessary in some instances either to treat one agreement as several contracts or to treat several agreements as one contract. Whether an agreement should be so severed or several agreements so aggregated will depend on all the facts and circumstances. Generally, one agreement will not be treated as several contracts unless such agreement contemplates separate delivery or separate acceptance of portions of the subject matter of the contract or unless there is no business purpose for entering into one agreement rather than several agreements. However, separate delivery or separate acceptance of portions of the subject matter of a contract does not necessarily require severing of a contract. Several agreements will not generally be aggregated unless the several agreements would be treated as one contract under customary commercial practice in a taxpayer’s trade or business or unless there is no business purpose for entering into several agreements rather than one agreement. An example of a factor which is evidence that two contracts entered into between the same parties should be aggregated is that one of the contracts would not have been entered into containing the terms agreed upon but for the entering into of the other contract.
Petitioner argues that its contracts may not be severed by delivery because it treated work orders as contractual units for valid business reasons. Respondent concedes that petitioner’s treatment of work orders was motivated by valid *368business reasons, but argues that petitioner’s contracts must nevertheless be severed by delivery in order to clearly reflect income.
Section 446(b)2 and sections 1.451-3(e), 1.446-l(a)(2), and 1.446-l(b)(l), Income Tax Regs., vest respondent with broad discretion in determining whether a taxpayer’s contracts should be severed so as to clearly reflect income. “Since the Commissioner has ‘[m]uch latitude for discretion,’ his interpretation of the statute’s clear reflection standard ‘should not be interfered with unless clearly unlawful.’ ” Thor Power Tool Co. v. Commissioner, 439 U.S. 522, 532 (1979), quoting Lucas v. American Code Co., 280 U.S. 445, 449 (1930). To overcome respondent’s determination, petitioner must establish that respondent was plainly arbitrary in severing petitioner’s contracts by delivery. See Reco Industries, Inc. v. Commissioner, 83 T.C. at 920; Peninsula Steel Products & Equip. Co. v. Commissioner, 78 T.C. at 1046. Petitioner has persuaded us that respondent’s determination was arbitrary.
Section 1.451-3(e), Income Tax Regs., was promulgated in 1976, and there was no comparable provision under the prior regulations. No case reported since adoption of the regulation has dealt with the severance of a contract accounted for under the completed-contract method.
Petitioner relies on two cases in which courts upheld the taxpayer’s aggregation of several nominally separate contracts as a single contract. In National Contracting Co. v. Commissioner, 25 B.T.A. 407 (1932), affd. 69 F.2d 252 (8th Cir. 1934), the taxpayer had entered into eight contracts for the construction of two school buildings. The Board of Tax Appeals affirmed the taxpayer’s treatment of the eight contracts as a single undertaking, pointing to the fact that the taxpayer had always treated the contracts as a single “job.” National Contracting Co. v. Commissioner, 25 B.T.A. at 411. In Cameron, Joyce & Co. v. United States, an unreported opinion (S.D. Iowa 1937, 22 AFTR 2d 1229, 38-1 USTC par. 9618), the court upheld the taxpayer’s treatment of six road-building contracts as two contracts. In each case, the different contracts related to a single undertaking *369or “job,” and in each case, it would have been difficult to allocate the profit from the “job” among the various contracts. Of particular significance here, the nominally separate contracts in the above cases do not appear to have been independently priced.
Of all the facts and circumstances to be considered on this issue, special emphasis should be placed on independent pricing. An example set forth in the regulation as in effect during the years in issue confirms this:
Example (2). Y, a calendar year shipbuilder using a long-term contract method, enters into two contracts at about the same time during 1972 with M. These contracts are the product of a single negotiation. Under each contract, the taxpayer is to construct for M a submarine of the same class. Although the specifications for each submarine are similar, it is anticipated that, since the taxpayer has never constructed this class of submarine before, the costs incurred in constructing the first submarine (to be delivered in 1973) will be substantially greater than the costs incurred in constructing the second submarine (to be delivered in 1974). If the contracts are treated as separate contracts, it is estimated that the first contract would result in little or no gain, while the second contract would result in substantial profits. It is unlikely that Y would have entered into the contract to construct the first submarine for the price specified without entering into the contract to construct the second submarine. In these circumstances, the two contracts must be treated as one contract for purposes of applying Y’s long-term contract method. [Sec. 1.451-3(e)(2), Income Tax Regs.; emphasis supplied.]
A subsequent revision of section 1.451-3(e), Income Tax Regs., underscores the importance of independent pricing. In 1982, Congress directed the Secretary of the Treasury to “clarify” when long-term contracts are to be severed or aggregated. Sec. 229, Tax Equity and Fiscal Responsibility Act of 1982 (TEFRA), Pub. L. 97-248, 96 Stat. 493. Although TEFRA’s legislative history and the preamble to the subsequent Treasury Decision state that the revised regulations are effective for years ending after December 31, 1982, neither document suggests that the “clarification” of the regulations was intended to be a substantive change. H. Rept. 97-760, at 549 (1982), 1982-2 C.B. 638; T.D. 8067, 1986-1 C.B. 218, 220. As amended, section 1.451-3(e)(l) provides as follows:
(ii) Whether an agreement should be so severed or several agreements so aggregated will depend on all the facts and circumstances.' Such facts *370and circumstances may include whether there is separate delivery or separate acceptance of units representing a portion of the subject matter of the contract, whether such units are independently priced, whether there is no business purpose for one agreement rather them several agreements or several agreements rather than one agreement, and such other factors as customary commercial practice, the dealings between parties to the contract, the nature of the subject matter of the contract, the total number of units to be constructed, manufactured, or installed under the contract, and the contemplated time between the completion of each unit.
(iv) One agreement may be severed, or several agreements may be aggregated, based upon the pricing formula of such agreements. For example, in the case of a multi-unit agreement for several similar items, if the price to be paid for similar units is determined under different terms or formulas (for example, if some units are priced under a cost-plus incentive fee arrangement, and later units Eire to be priced under a fixed-price arrangement), then the difference in the pricing terms or formulas may indicate that the agreement should be treated as several contracts.
Examples set forth in the amended regulation confirm the importance of truly “independent” pricing:
Example (6). T, a calendar year taxpayer engaged in the business of manufacturing aircraft and related equipment, enters into an agreement in 1982 with the B government to manufacture 10 military aircraft for delivery in 1984. It is anticipated at the time the agreement entered into that B may enter into an agreement with T for the production and sale of as many as 300 of these aircraft over the next 20 years. In negotiating the price for the agreement, B and T take into account the expected total cost of manufacturing the 10 aircraft, the risks and the opportunities associated with the agreement, and all other factors that the parties consider relevant, in such a manner that T would have entered into the agreement with the terms agreed upon whether or not T would actually enter into one or more additional production agreements. However, it is unlikely that T would have entered into the agreement but for the expectation that T and B would enter into additional production. In 1984, the 10 aircraft are completed by T and accepted by B. In 1984, T also enters into an agreement with B to manufacture 20 aircraft of the same type for delivery in 1986. In negotiating the price for these 20 aircraft, B and T take into account the fact that the expected unit costs for this production of 20 will be different than the unit costs of the 10 aircraft completed in 1984, but also that the expected unit costs of this production of 20 will be substantially higher than the costs of future production. Because the price awarded for each of the two agreements takes into account the expected total costs and the risks expected for each agreement standing alone, the terms agreed upon for any one of the *371agreements are independent of the terms agreed upon for the other agreements. Under the facts of this example, the two agreements may not be aggregated into one contract for purposes of applying T’s long-term contract method.
Example (7). R, a calendar year taxpayer engaged in the manufacture of industrial machinery, enters into one agreement in 1982 with Z to manufacture five specialized machines and to manufacture spare and replacement parts for the machines. The machines are to be delivered in 1982 and 1983, and the spare and replacement parts are to be delivered in 1983 through 1985. The portion of the total contract price attributable to the five machines and to the spare and replacement parts reasonably can be determined. The portion of the total contract price reasonably attributable to the spare and replacement parts is more than an insignificant amount of the total contract price. Assume that, under all the facts and circumstances, it is determined that the portion of the agreement attributable to the five machines need not be severed as between the machines. In these circumstances, because the agreement contemplates separate delivery of the machines and the parts, because more than an insignificant amount of the total contract price is allocable to the spare and replacement parts, and because spare or replacement parts are items different than an entire machine, it may be necessary for the Commissioner to sever the agreement, treating the agreement to manufacture the five machines as a separate contract and the agreement to manufacture the spare and replacement parts as another separate contract (or as several separate contracts depending on the facts and circumstances) for purposes of applying R’s long-term contract method.
[Emphasis supplied.]
The circumstances surrounding the Sylmar and Transtech contracts are similar to the facts of example (2) as set forth in both versions of the regulation and example (6) of the regulation as amended. Because many of Sylmar’s products involve novel technology that changes over the lifetime of a program, the price of each transparency is based on the total quantity called for in a program, not the quantity shipped to the customer in any specific delivery. Sylmar’s delivery schedules are arranged to accommodate its customer’s production rates for aircraft. If prices were determined on a per-shipment basis, they would be substantially higher. Transtech’s deliveries are also not independently priced. As a practical matter, Transtech cannot calculate with any certainty the total costs and hence the total profit associated with any job prior to the completion of that job. Because petitioner’s Sylmar and Transtech contracts could not be independently priced by delivery, and in consideration of all other facts and circumstances, we conclude that *372these contracts need not. be severed as proposed by respondent.
III. Conclusion
We have carefully considered additional arguments of the parties set forth in their briefs. We find these arguments unpersuasive. Petitioner properly used the completed-contract method to account for income from its Sylmar and Transtech contracts. Petitioner should not have relied on the completed-contract method to account for income from its Magnedyne contracts. Finally, petitioner’s Sylmar and Transtech contracts need not be severed by delivery. To reflect the foregoing,
Decision will be entered under Rule 155.