Perkin-Elmer Corp. v. Commissioner

PERKIN-ELMER CORPORATION AND SUBSIDIARIES, Petitioner v. COMMISSIONER OF INTERNAL REVENUE, Respondent Perkin-Elmer Corp. v. Commissioner Docket No. 28860-89 United States Tax Court T.C. Memo 1993-414; 1993 Tax Ct. Memo LEXIS 424; 66 T.C.M. (CCH) 634; September 8, 1993, Filed *424 For petitioner: Robert J. Cunningham, James M. O'Brien, Mark A. Oates, Fred G. D'Amato, and Debra F. Novack. For respondent: Victoria W. Fernandez, Christine Halphen, George H. Soba, and William J. Gregg. TANNENWALD TANNENWALD MEMORANDUM FINDINGS OF FACT AND OPINION TANNENWALD, Judge: Respondent determined deficiencies in the Federal income taxes of petitioner, The Perkin-Elmer Corp. (P-E) and consolidated subsidiaries, as follows: Taxable Year Ended Deficiency July 31, 1975 $ 1,277,967 July 31, 1976 1,439,257 July 31, 1977 2,334,572 July 31, 1978 1,970,301 July 31, 1979 3,001,785 July 31, 1980 4,248,794 July 31, 1981 11,864,674 The issues relating to section 482, the subject of this opinion, have been severed from certain other issues. 1 The issues presently before us for decision are: (1) Whether respondent's allocations of gross income to P-E under section 482 were arbitrary, capricious, or unreasonable; (2) whether the prices P-E paid for finished products to a wholly owned subsidiary operating in Puerto Rico were arm's-length amounts; (3) whether the prices the subsidiary paid to P-E for parts that went into the finished products were arm's-length amounts; *425 (4) whether the royalties the subsidiary paid to P-E on sales of the finished products to P-E were arm's-length amounts; and (5) for prices or royalties that were not arm's length, what the arm's-length amounts are. In addition to finished products, which needed no further manufacturing work, the subsidiary sold so-called component products to P-E. The parties settled the section 482 implications of these component products prior to trial. FINDINGS OF FACT Some of the facts have been stipulated. The seven stipulations of fact, together with their accompanying exhibits, are incorporated by this reference. I. Introductory Background A. The P-E Organization The principal office of P-E was located at all relevant times in Norwalk, Connecticut. P-E and affiliated subsidiaries filed consolidated Federal income *426 tax returns. Perkin-Elmer Caribbean Corp. (PECC), a wholly owned subsidiary of P-E operating in Mayaguez, Puerto Rico, did not join in the consolidated returns. On its returns for the taxable years ended July 31, 1975 and 1976, PECC excluded gross income by invoking section 931. On its 1977 return, PECC elected under section 936(e) to claim a credit against its income tax liability, which election was in effect through 1981. During the years in issue, 1975 through 1981, the worldwide operations of P-E and its subsidiaries were organized into five operating groups, each of which was responsible for the research, manufacturing, sales, and servicing of its products. 2 The five product areas were analytical instruments, optical systems, computer systems, flame spray equipment and materials, and military avionics. The Instrument Group, responsible for analytical instruments, included the following units with manufacturing operations: (1) The P-E Instrument Division, of which PECC was a part for management and accounting purposes, headquartered in Norwalk, Connecticut; (2) the Coleman Instruments Division, located in Illinois; (3) Perkin-Elmer Ltd. (P-E Ltd.), a United Kingdom corporation; *427 and (4) Bodenseewerk Perkin-Elmer & Co., GmbH (Bodenseewerk), a West German corporation. P-E Ltd. and Bodenseewerk each had its own purchasing, engineering, manufacturing, marketing, and sales departments. B. The Analytical Instrument Industry 1. Products Analytical instruments are used to analyze samples of solids, liquids, and gases to obtain quantitative or qualitative information, including details of the chemical structures of the samples. An instrument called a spectrophotometer, with atomic absorption (AA) and infrared (IR) as common varieties, *428 uses light as the means of analysis. An AA spectrophotometer is generally used to determine the concentrations of metallic elements in liquid samples. Typically, the sample burns in a flame and converts into atoms that float in the flame. A light source, usually a hollow cathode lamp (HC lamp), emits the characteristic wavelength of the element being measured, copper for example. The copper atoms in the sample, and only the copper atoms, absorb the copper light, and the remainder of the light strikes a detector. The proportion of the copper light absorbed by the atomized sample reveals the concentration of copper atoms in the sample. An HC lamp is cylindrical in shape with a plastic plug-in base. Much of its length is glass, which sometimes mates at the end with a quartz window. The glass bulb contains a filler gas, usually neon. Although this lamp typically has only a single-element cathode useful in measuring that element alone, multi-element HC lamps have existed since the 1960s. The useful life of an HC lamp, which eventually burns out and needs replacement in regular use, is substantially shorter than the spectrophotometer itself, causing HC lamps to be referred to *429 sometimes as consumables. An alternative light source for some elements is an electrodeless discharge lamp (ED lamp), which offers greater intensity and longer life and sometimes offers better sensitivity. Among the major accessories for a flame-burner AA spectrophotometer is a furnace atomizer, which greatly enhances the analytical sensitivity of the instrument by atomizing samples in a furnace rather than a flame. Another AA accessory, a background corrector, minimizes the effects of undesirable background absorption and scatter; it is generally sold already installed in the instrument rather than as a later add-on. A third major AA accessory, an ED lamp power supply, is a separate power source for ED lamps. An IR spectrophotometer identifies molecules in a sample and determines their structure and concentration. A beam of invisible, IR light containing a spectrum of wavelengths passes through the sample, which absorbs the many wavelengths in varying degrees. A chart recorder graphically depicts the amount of energy absorbed at the several wavelengths, beam intensity versus wavelength. This graphical "fingerprint" is then compared to those of standard samples with known compositions. *430 2. Instrument Users The primary users of analytical instruments during the period in issue were industrial, university, and governmental laboratories engaged in sophisticated analyses in areas such as research and product testing. A low price was normally not the main consideration for an instrument purchaser. Instead, factors such as a company's technical reputation in the product line, its servicing reputation, the instrument specifications, and product reliability were usually more important than price. However, because of the functional and design similarities among competitors' products in the low-priced realm, purchasers of low-priced instruments were primarily concerned with price and reliability. Overall, instrument users purchased more than half of their nonconsumable accessories from the same manufacturer that had supplied them with the associated instrument. 3 A customer generally wanted to deal with a supplier that could provide both the desired type of instrument and the associated accessories, thus entailing a single purchase order and a single warranty. In addition to convenience, other important considerations to accessory customers were quality and compatibility. *431 Instrument users preferred to purchase consumables in general from supply houses, with little worry about compatibility, in part because they perceived supply houses as considerably less expensive than instrument companies. However, instrument users were much more likely than not to purchase lamps from the same instrument company that had supplied the associated instrument. When purchasing accessories or consumables, customers tended to value a low price somewhat less than they valued quality and delivery time, but price was not unimportant. A typical laboratory with one or more AA spectrophotometers might have several working HC lamps covering different elements, and users changed HC lamps themselves, without the need for service assistance. As with lamps*432 in general, an instrument user had a strong tendency to buy HC lamps from the instrument supplier, both at the time of instrument purchase and subsequently. Nonetheless, various sources of competition existed in the form of supply houses, other instrument manufacturers, and HC lamp makers that did not also manufacture instruments. Price was ordinarily not the main consideration for an HC lamp purchaser, but major price differences caused shifts to lower priced suppliers. HC lamps of different sizes among competitors were not a major problem for purchasers because instrument manufacturers and supply houses sold lamp adaptors that accommodated the differences. Worldwide sales of analytical instruments, which were approximately $ 1.25 billion in calendar 1977, grew at a compounded annual rate exceeding 10 percent from the late 1960s to the late 1970s. By 1975, the basic AA and IR technologies were mature and widely known. 3. P-E There were a few full-line manufacturers in the analytical instrument industry and many small firms with strengths in selected product areas. P-E, one of the full-line manufacturers, was a leader in the industry in the United States and worldwide. *433 Principal reasons for this position were its broad line of instruments, its continual introduction of new products and incorporation of new technology, and its strong customer support and service organization. P-E had a reputation for high quality products, and its desired "high quality, high technology" image discouraged it from seeking to meet the often lower instrument prices of competitors. In the spectroscopy area, which included the AA and IR product lines, instrument users in 1981 rated P-E the best of four major manufacturers on many nonprice factors, including technical reputation, completeness of product lines, product reliability, and service speed. P-E was the worldwide leader in the AA instrument market, holding about a 40-percent market share of units sold overall and about a 60-percent share of units sold in the United States. Its two main competitors taken together did not closely approach the P-E market share. Within the AA market in 1975, however, the P-E position was relatively much weaker in the low-priced market segment (below a $ 6,000 price) than in any of the higher priced segments. Despite its strong worldwide position in the AA area, P-E had a market*434 share below 5 percent in Japan, a country with at least 10 percent of the worldwide AA market. The AA product line was the most successful and profitable for the Instrument Division at P-E. Instrument Division annual unit sales of AA instruments were about 25 percent higher in 1980 than in fiscal 1974, and dollar sales in the AA product line had nearly doubled. The nature of the industry was such that the P-E AA instruments often reached their sales peaks only 2 or 3 years after introduction, followed by a gradual decline in the absence of significant improvements or modifications. P-E's generally high pricing relative to its competitors sometimes created pricing "umbrellas" under which competitors could sell comparable AA instruments profitably. Nonetheless, the P-E reaction to heightened competition, in the AA and other product lines, was more likely to take the form of increased marketing efforts or accelerated product improvements than published price reductions or systematic discounting. At the low-priced end of the AA product line, the P-E pricing was generally close to that of competitors but still often higher. Marketed under the "Intensitron" trademark, P-EHC lamps*435 were physically larger than those of competitors, with capacity for more filler gas and potential for longer life. P-E experienced quality problems with some of its HC lamps in about 1976, which competitors exploited. Specifically, unwanted hydrogen was appearing in the lamps because of a failure of the epoxy that sealed the quartz window to the bulb, leading P-E to switch to an improved, yet more expensive, graded seal. P-E completed the conversion in early calendar 1980. P-E also sold ED lamps, which were technologically superior to competitors' products in the mid-1970s. A P-E ED lamp, which was directly interchangeable with Intensitron HC lamps, required a separate power supply, and P-E voided the ED lamp warranty if the user operated the lamp with other than a P-E power supply. Between 1976 and 1980, P-E's dollar volume of AA lamp sales, meaning HC lamps and ED lamps combined, increased nearly 50 percent. P-E management always viewed AA lamp sales as very profitable. P-E, with about a 50-percent market share, was also the worldwide leader in the IR instrument market and, by 1977, dominated both the domestic and export markets for instruments priced over $ 7,000. Market*436 surveys by the periodical Analytical Chemistry in 1980 and 1981, as with AA instruments, showed P-E well ahead of its competitors in terms of IR quality, delivery/availability, and servicing. The weakest P-E market segment, as in the AA product line, was at the low-priced end. II. The P-E Instrument Division A. Product Departments The Product Departments, each headed by a product manager and staffed in part with specialized chemists, were responsible for the technical marketing of the products manufactured, or purchased and resold, by the Instrument Division. Each Product Department handled a distinct product line, such as AA, IR, gas chromatography, or fluorescence, and within the AA product line, management regarded spectrophotometers, lamps, furnace atomizers, and accessories as separate sublines. The Product Departments performed much of the customer training that took place in Connecticut and also developed marketing support materials, including periodic newsletters and occasional studies on the analysis of samples. Successful product development depended on constant interaction between the Product Departments and the sales organization, the latter often being*437 more immediately conscious of customer needs. The Product Departments also interacted with the service organization, which, in addition to identifying problems and recommending changes in the products, made suggestions for the improvement of instruction manuals and other user documentation. Product managers held periodic product support meetings at which the participants discussed engineering, manufacturing, quality, and field problems. Annually, each Product Department contributed to the preparation of a detailed written "Summary of Project Status and Plans" (SP2 report), which described by product line, both for the recent past and as projected, financial results, products, markets, competitors, and marketing strategies. The SP2 reports sometimes noted the tendency of an HC lamp user to purchase from its AA instrument supplier, while also noting that pricing and other factors, such as quality and delivery time, could override this inclination. The SP2 report prepared in 1977 described P-E ED lamps as "definitely superior in quality" to the competition, while Intensitron lamps, as a technical matter, were "at least the equivalent of any competitive" HC lamps. B. Engineering *438 The Engineering department had two principal responsibilities: product development research, which included feasibility studies and product improvements, and so-called continued engineering. Engineering determined the cost and time frame for proposed products and generally sent a product, whether an instrument or an accessory, through several development phases. After exploratory research, Engineering prepared a working "breadboard" that the Product Department evaluated on a preliminary basis. Engineering then built a few prototypes for testing by Engineering itself and perhaps a customer or two. Later, Engineering released and then worked with Manufacturing to refine detail, assembly, and test drawings, and at some point Engineering and the Manufacturing department prepared technical documentation that reflected the original design of the product and manufacturing know-how. The continued engineering function, which supported a product after it entered commercial production, dealt with design problems that arose during either manufacture or customer use. One of the duties of a continued engineer was involvement in approving engineering change notices, which revised or updated*439 specifications shown in blueprints and other engineering documents. During 1975 through 1981, the Instrument Division spent over $ 40 million on research and development, with approximately one-third of the total relating to the AA and IR product lines. C. Manufacturing The Manufacturing department was responsible for the building and final testing of instruments, accessories, and lamps. A P-E internal report prepared sometime before July 1980 described the relationship between parts and the finished instruments: From a manufacturing point of view, the products can be characterized as follows. There are 2,000 to 3,000 parts per instrument. The [Instrument] Division sells many models, all with fairly low volume; for instance, the highest-volume product has sales of 750 units per year. Due to the existence of 7 different product lines, there is low commonality of parts. Thus, total parts-on-file number roughly 96,000. Of these, 60% are purchased (mostly electronic components) and 40% are manufactured in-house. The functional sections within Manufacturing included sheet metal, the machine shop, raw board, painting, electrical assembly, preassembly, final assembly*440 and test, lamps, and packing and shipping. The raw board section made printed circuit boards, and the machine shop and sheet metal sections produced various types of other parts with the help of computer-controlled machines. The production process for a P-E instrument typically involved the combining of several subassemblies that were themselves each made up of many parts. The final assembly and test section was important largely because of its position at the end of the manufacturing process. In this section, workers assembled optical, electrical, and mechanical parts and subassemblies into completed products, which were then tested to ensure that they met specifications. The final test technicians, who were among the most skilled employees in Manufacturing, had to have an understanding of the complete instrument and an ability to detect and correct problems. HC lamp manufacturing required precise control of materials and conditions, with an emphasis on avoiding contamination. During the manufacturing process, the lamps underwent parts inspection, cathode fabrication, glass lathing, electrical assembly, preassembly, final assembly, vacuum pumping, and testing. This process, *441 which involved relatively few parts but certain specialized equipment, was much different than the manufacturing process for instruments. During the years in issue, the Instrument Division manufactured about 30 HC lamp models in Norwalk, and the manufacturing equipment maintained in Norwalk was capable of producing any HC lamp model that P-E sold. The Instrument Division did not purchase any finished HC lamps from unrelated suppliers. D. Quality Assurance Quality Assurance was independent of Manufacturing, in part to avoid conflicts of interest, and the Quality Assurance manager reported to the general manager of the Instrument Division. Quality Assurance involvement was widespread. This department assisted in the selection of parts suppliers and also communicated with and visited suppliers if problems arose in the field. The inspection section inspected samples of purchased materials and parts and also sampled work-in-process from Manufacturing sections. Quality Assurance ran tests during the preproduction phases of new product development, and it audited manufactured finished products and those purchased from others for resale. Because assembly and test specifications*442 did not cover all possible problems, Quality Assurance performed tests that differed from those performed by the Manufacturing final test technicians. As a general rule, defective instruments and accessories were not scrapped in their entirety; defective parts within instruments and accessories were sometimes reworked rather than scrapped. E. Manufacturing Staff Functions 1. Production Planning The Instrument Division planned its production on a model-by-model basis. Each Product Department prepared an annual sales forecast, subject to monthly revision, based on input from sales personnel. These annual forecasts served as the foundations for manufacturing and operating budgets. Forecasting and production planning were especially important because the cumulative production lead time for a typical instrument was much longer than the desired delivery time of 4 to 8 weeks following a customer order. In 1964, the Instrument Division began using the Material Requirements Planning system (MRP system), a software program and computer system designed to improve operational control of the manufacturing process, especially inventories. This was an ordering and scheduling system*443 that constructed a manufacturing plan from sales data, production history, and shipping schedules. Cataloged in the system was a wide variety of information, including sales and cost data, lead times, purchasing information, bills of materials, job status reports, and complete histories of products. As specific output, the MRP system, which included in its coverage HC lamps, generated at least monthly suggestions to production planners for issuing work orders to the factory or purchase orders to vendors. The production planners could adjust the suggestions based on their reviews of forecasts and inventories. In one way or another, the MRP system interacted with virtually all areas of Instrument Division activities, including the Purchasing department, Engineering, the service organization, and the final assembly and test section of Manufacturing. Analytical instrument manufacturers ordinarily built some portion of their production in direct response to customer orders rather than as inventory. The Instrument Division used this technique in Norwalk for some of its AA products that had many variations, but did not use it for the more standardized IR product line. 2. Purchasing *444 The Instrument Division maintained a Purchasing department, with 12 or 13 employees, to ensure timely supplies of materials and parts. The expenses associated with Purchasing, not including amounts paid for purchased items, totaled approximately $ 3.1 million from 1975 through 1981. Purchasing ultimately selected almost all of the Instrument Division vendors, although the selection process could involve Engineering, Quality Assurance, and manufacturing engineers. Engineering was familiar with specialized vendors and sometimes specified them for certain parts. Quality Assurance surveyed potential vendors, as did the manufacturing engineers, and also kept files and history cards as a record of vendor quality. For mechanical parts, as contrasted with electronics, Purchasing usually acted alone in selecting vendors. Purchasing maintained a qualified vendor list, which was proprietary information but not unlike what other instrument manufacturers compiled, and an "approved component list" of the most reliable sources for basic electronic parts. The Purchasing buyers, most of whom had college degrees, usually served at least a few years as parts expediters before becoming buyers. *445 Initial training for buyers consisted of a 12-week purchasing and procurement course, and training available thereafter included technical seminars conducted by vendors, internal monthly meetings, negotiation seminars, and trade school enrollment. Some Purchasing personnel read technical bulletins and trade journals and maintained contacts with account executives of major vendors. The more senior buyers in Purchasing were responsible for purchasing hard-to-locate items and so-called complex and highly complex items, which were made to P-E specifications. For the complex and highly complex parts, the buyer provided the vendor with specialized engineering drawings that the buyer had reviewed with Engineering. Parts that were difficult to manufacture usually necessitated in-person negotiations, which the buyer often attended with representatives from Quality Assurance and the manufacturing engineering function. The two supervisory purchasing agents in Purchasing became directly involved in production planning by attending planning/shortage meetings twice a week. By 1975, the MRP system enabled Purchasing to be much more effective in scheduling and rescheduling delivery dates on*446 purchase orders. Interfacing with the MRP system was a computerized purchasing system, with a vendor historical file base, that generated reports and purchase orders. The purchasing process for mechanical parts began when Purchasing received a requisition form from the production people specifying a part number, quantity, and required delivery date. If P-E had at some previous time manufactured the part itself, the requisition form would indicate the P-E manufacturing standard cost. Assuming adequate lead time, which was not always the case, Purchasing was supposed to pay an outside vendor no more than this standard cost. These outside purchases of parts that had also been manufactured by P-E were recorded in inventory account 2403 of the P-E general ledger. The electronics specialists in Purchasing, unlike the mechanical specialists, rarely used account 2403. Petitioner engaged the public accounting firm of Arthur Andersen & Co. (Arthur Andersen) to make calculations and render opinions concerning various accounting matters. After adjusting for items such as consignment transactions and purchases requiring additional work by P-E, Arthur Andersen determined that the account*447 2403 data from March 1978 through July 1980 showed aggregate actual purchased costs below aggregate manufacturing standard costs by 19.6 percent in 1978 and 2.1 percent in 1979, and purchased costs above standard costs by 8.2 percent in 1980; the summed purchased costs over the full period were 11.5 percent below the summed standard costs. The costs recorded in account 2403 during this period represented less than 1 percent of the Instrument Division cost of sales. III. P-E Distribution Activities The P-E marketing and sales organization for analytical instruments was superior to most or all of its competitors for the AA and IR product lines. The Instrument Division sold its products to end users in the United States through the U.S. Sales and Service Division (USSD) of the Instrument Group, which had offices throughout the country. For foreign distribution, apart from some direct sales to unrelated foreign customers, the Instrument Division sold either to foreign subsidiaries of P-E or to Perkin-Elmer International, Inc. (P-E International), both of which resold to unrelated foreign distributors or dealers. P-E International, a domestic corporation wholly owned by P-E, *448 joined in the P-E consolidated income tax returns. The Instrument Division used these same channels to distribute the products purchased on an original equipment manufacturer (OEM) basis. OEM products were those manufactured by others, then purchased by P-E and resold under the P-E name and trademarks. The USSD sales organization had sales engineers and product specialists among its over 300 employees. A product specialist was a chemist who provided technical assistance to users in a particular product line. Located in regional offices and at the Norwalk headquarters, the product specialists conducted most of the training for Instrument Division customers, and a training course was free with the purchase of an AA spectrophotometer. Sales laboratories located around the United States enabled customers to learn how to operate instruments while testing their own samples. The P-E marketing and sales approach for analytical instrumentation involved customer visits and technical literature, and management thought it important to maintain a full range of products in order to be able to satisfy all of a customer's needs. USSD salespeople, in visits with existing and prospective customers, *449 determined their analytical requirements, helped them prepare product proposals, sometimes arranged for demonstrations at sales laboratories, prepared product quotations, and ultimately took orders. Compensated by commission, the salespeople generally exerted more effort on the higher priced products, which coincided with management's philosophy to avoid overselling low-priced instruments to the detriment of higher priced units. Technical literature, normally prepared by the Product Departments, was either product specific or more general in nature. Product-specific literature consisted mainly of data sheets, which described detailed specifications, and product brochures. The data sheets and brochures were more elaborate for the more complex and expensive products in the AA and IR lines, and brochures were generally not prepared specifically for accessories or HC lamps. Major accessories, however, often had a short data sheet. More general technical literature, such as customer newsletters and original or reprinted journal articles, illustrated various applications of analytical techniques, sometimes with the use of a specific product. All AA instruments sold by P-E came with*450 a 1-year newsletter subscription and a "cookbook" describing over 400 analyses. There was a similar cookbook for the furnace atomizer AA accessory. The advertising department in USSD placed ads in professional publications such as technical and trade journals, prepared mailings, and organized exhibitions and workshops. Some ads covered specific instrument models, others covered a product line, and still others promoted the P-E corporate image. Trade show exhibitions were an important promotional method in the analytical instrument industry, and P-E was a regular participant. The Instrument Division usually displayed its newest products at trade shows, but sometimes used extra space to exhibit best sellers; both categories at times included lower to mid-priced instruments. P-E often marketed its instruments and accessories together. An instrument brochure frequently included a section highlighting available accessories, and at times P-E offered price incentives on AA and IR accessories for a package instrument/accessory deal. In the AA product line overall, P-E did not widely advertise its accessories. The AA product line included some accessories that customers avoided because*451 of perceived high prices, and, because of development resources diverted elsewhere in the mid-1970s, some accessories lost their competitive edges technologically. At least some of the AA accessories, however, faced no outside competition. The IR accessory business for P-E, which tended to follow the IR instrument business, was very profitable in the mid-1970s. This encouraged competition and prompted P-E to offer discounts and other customer incentives such as generous trade-in allowances. For most of the period in issue, the sale of HC lamps was little more than an order-taking activity for P-E salespeople, like that of catalog salespeople. Although the annual SP2 reports sometimes urged or predicted increased marketing efforts, P-E did not attempt aggressive promotion and advertising until about 1980. P-E informational brochures, however, did mention that Intensitron HC lamps were "particularly notable for brightness, stability, and long life" and were "designed specifically for best atomic absorption performance, and combine long, usable lifetimes with high intensity and stability." The catalog of a large supply house that sold HC lamps manufactured by Westinghouse described*452 these lamps as "Spectral tubes designed for high spectral output, excellent stability, low noise and long operating life; providing the analyst with higher precision, sensitivity and detectability." Also a part of USSD was the service department, with personnel spread among field offices, regional offices, and the Norwalk headquarters. Service engineers installed all AA and IR instruments sold by USSD, and they also performed warranty and maintenance work. An instrument user could easily install HC lamps, however, without service assistance. Although outside service contracts were an option for users, P-E tended to service its own instruments, and, for its nonwarranty service work, its activities were generally profitable. A 1-year warranty covered the instruments and accessories sold by the Instrument Division. HC lamps had a 2-year spectra-emission warranty and a 6-month intensity warranty. Use of competitors' HC lamps in a P-E instrument did not void the P-E instrument warranty. IV. PECC A. Origin During 1968 and 1969, Instrument Group management was exploring the possibility of adding a Puerto Rican operation, in part because of capacity constraints in Norwalk. *453 In May of 1970, Gaynor Kelley (Kelley), the head of Manufacturing, made a presentation to the P-E board of directors recommending that P-E establish a manufacturing operation in a leased facility in Mayaguez, Puerto Rico. According to Kelley's written proposal, the Instrument Division intended the new facility to perform assembly operations on specified finished products and component products, all using parts, supplies, and subassemblies provided by the Instrument Division. The proposal stated that the Instrument Division would transfer all the necessary parts at standard cost plus 10 percent and would purchase all the finished products at a 25-percent discount off list price. At least for the early years, Kelley envisioned the new operation selling its entire output to P-E. The proposal emphasized various tax exemptions and income tax savings available in Puerto Rico. Among the other favorable factors Kelley mentioned were a high unemployment rate relative to the United States and a low wage rate relative to Norwalk. He assumed a 25-percent increase in productivity over comparable Instrument Division operations in the United States. Following Kelley's presentation, the board*454 of directors approved a manufacturing operation in Puerto Rico and authorized the creation of a wholly owned subsidiary. PECC filed its certificate of incorporation with the State of Delaware in June 1970, and operations commenced at the Mayaguez facility in early October. In December 1970, the government of Puerto Rico granted applicant PECC a 15-year industrial tax exemption applicable to various specified products. In its application dated in July 1970, PECC described its marketing outlets as follows: "Sole purchaser of products manufactured by Applicant will be The Perkin-Elmer Corporation of Norwalk, Connecticut, which enjoys a worldwide market." B. Licensing and Distribution Agreements P-E and PECC entered into a General Licensing Agreement dated as of October 1, 1970, by the terms of which P-E granted PECC an exclusive right to manufacture in Puerto Rico and a nonexclusive right to use and sell worldwide the instruments and accessories to be identified in specific licenses. P-E also agreed to furnish PECC with all design and manufacturing information, including any then still to be developed, associated with any licensed products. PECC agreed to pay royalties on*455 the products based upon the "Net Sales Price", defined as "the net amount billed and payable for * * * [licensed products] excluding import duties, insurance, transportation costs, taxes which are separately billed and normal trade discounts." In practice, P-E and PECC interpreted this definition to mean the amount PECC billed P-E rather than the amount P-E billed upon resale. The specified term of this agreement was until the expiration of the last license entered into pursuant to the agreement. The General Licensing Agreement was very similar in form and content to agreements of July 1966 between P-E and subsidiaries P-E Ltd. and Bodenseewerk, except that the 1966 agreements were cross-licensing agreements covering P-E's role as both a licensor and a licensee. As part of the P-E agreements with P-E Ltd. and Bodenseewerk, the parties licensed specific AA and IR instruments for a 6-percent royalty and a Model F-11 gas chromatography instrument for a 7-percent royalty. In its "intersite" licensing arrangements with subsidiaries, the P-E policy was to reduce the royalty rate by half if, for a particular licensed product, the parts sold by the licensor to the licensee represented*456 more than 35 percent of the licensor's manufacturing cost for the finished product. A 1968 internal P-E memorandum described the 35-percent crossover point for a so-called partial versus a so-called complete transfer as based on an "arbitrary assumption". Although neither the General Licensing Agreement nor the specific licenses with PECC manifested this intersite policy, P-E incorporated it into the 1966 agreements with P-E Ltd. and Bodenseewerk by means of 1968 amendments. The General Licensing Agreement did state that P-E would reduce the royalty if PECC caused a "major redesign" of a product, but PECC never made any changes that met this classification. The first license that P-E and PECC entered into pursuant to the General Licensing Agreement was also dated as of October 1, 1970, and had a term of 10 years. This license identified HC lamps, the Model 700 series IR spectrophotometers, and four models of AA spectrophotometers as licensed products subject to a royalty rate of 3 percent. It also identified a Model F-11 gas chromatography instrument as subject to a royalty rate of 3.5 percent. The parties intended the PECC royalty payments to cover engineering technical assistance*457 but not other technical assistance. P-E and PECC entered into no more written licenses until one dated as of October 1, 1980, again for a term of 10 years. This license identified various finished products as subject to a 3-percent royalty. In May of 1971, representatives of P-E and PECC executed a Distributor Agreement in which P-E agreed to act as a worldwide distributor for PECC. The agreement did not explicitly state that P-E would be an exclusive distributor. P-E was obligated to purchase from PECC, f.o.b. Mayaguez, only the quantities it ordered and at the prices specified in then current price lists and schedules. PECC was to sell and deliver products subject to the terms and conditions of its standard sales order. The agreement did not specify ordering lead time, and payment terms were to be as "mutually agreed". The agreement was to continue in effect for at least 1 year and thereafter indefinitely until either party gave 90 days' written notice of termination. C. Products and Sales PECC never sold or otherwise distributed its finished products to any unrelated party. Although alternative distribution arrangements, at least on a small scale, were considered, *458 PECC sold and shipped all of these products to P-E. The PECC finished products were AA spectrophotometers and accessories, HC lamps, IR spectrophotometers and accessories, and gas chromatography accessories. The AA and IR instruments were generally low priced, high volume, and easy to use relative to the other P-E instruments in these two product lines. The PECC finished product sales totaled $ 74,012,133 from 1975 through 1981. 4 As percentages of this total, instrument sales (including autobalances and furnace atomizers) were approximately 45 percent, HC lamp sales were 30 percent, and accessory sales were 25 percent. 5*459 PECC recorded gross profit margins of 30.2 percent for instruments, 62.7 percent for HC lamps, and 55.4 percent for accessories. 6 PECC generally produced the lowest priced AA spectrophotometers then offered for sale by the Instrument Division, plus one or two models from adjoining low-priced or mid-priced segments. Except for Models 560 and 2380, which had customer list prices in the neighborhood of $ 12,000, all of the PECC AA spectrophotometers had published prices between $ 4,000 and $ 10,000. PECC also produced heated graphite furnace atomizers, Models HGA 2100 and HGA 2200, with list prices between $ 4,000 and $ 5,000. Another major AA accessory that PECC produced was the autobalance, Models AD-2 and AM-2, used to weigh solid samples to be analyzed with a furnace atomizer. The list price of the Model AD-2 *460 averaged about $ 3,500. PECC recorded sales of AA instruments, HGA furnaces, and autobalances as follows: AA Product Years Produced Sales Model 103 1975-1977 $ 453,661 Model 107 1975-1977 797,700 HGA furnace 1975-1979 6,348,729 Model 360 [1976] 293,513 Model 370 1976-1977 784,125 AM-2 autobalance 1976-1978 98,409 AD-2 autobalance 1976-1979 975,326 Model 372 1977-1979 755,507 Model 272 1978-1980 1,348,200 Model 280 1979-1980 1,163,246 Model 380 1979-1980 3,301,340 Model 560 [1980] 534,660 Model 2280 1980-1981 1,885,261 Model 2380 1980-1981 5,964,078 Total $ 24,703,755 In 1975, the P-E market share in units for its low-priced AA instruments, Models 103 and 107, was only 26 percent in the United States and, because of P-E's relatively high overseas pricing, 13 percent worldwide. Orders for these models were already in a state of decline because of the competition, including that from P-E's own lower mid-priced instruments. The Models 103 and 107 lacked reliability and the key features offered by competitors, rendering them unattractive to both users and the sales force. P-E discontinued these models in 1977. The Models 360 and 370, introduced in 1974*461 and 1975, respectively, were in the lower mid-priced market segment in 1975, where P-E had a 56-percent U.S. market share and a 40-percent worldwide share. Management forecast a drop of about 12 percent in unit volume for 1976, but, mainly because of a desirable upward shift in the P-E product mix, the actual drop was 33 percent. In 1977, when P-E discontinued these models, the drop in unit volume was again 33 percent. Because of pricing pressure from competitors, the Model 372, introduced in 1977, was priced overseas at or below domestic prices. Management viewed this model as offering more for the user's money than competitors' products. Customer response was excellent, and unit orders in 1978 exceeded the forecast. In 1979, when P-E introduced the replacement Model 380, unit sales decreased substantially from 1978, primarily because of severe competition from the new products offered by Varian/Techtron. In 1979, the Model 560 also confronted Varian/Techtron competition, which prompted P-E to offer a free background corrector valued at about $ 1,500 with the instrument, which in turn greatly increased sales. The Model 2380 replaced the Models 380 and 560 in 1980, and, according*462 to the SP2 report covering that year, "demonstrated itself to be a real winner." The Model 272, which replaced the Models 103 and 107, was P-E's only single-beam AA instrument in 1978. Although management viewed it as a very good value for its low price, USSD sales engineers preferred to push higher priced models, and it sold overseas at domestic prices or less. P-E replaced the Model 272 with the Model 280 in 1979, which was in turn replaced in 1980 by the Model 2280, still a single-beam instrument. The Model 2280 encountered severe price competition in Europe. Nonetheless, unit sales of P-E single-beam instruments increased 25 percent in 1980. P-E modified the HGA 2100 to create the HGA 2200 in March 1977. The market for furnace atomizers was very competitive and price sensitive, but P-E began to take control in the late 1970s because of superior products, especially the replacement models for the PECC-produced furnaces. P-E replaced the HGA-2200 in March 1979. In the medium-priced market for nonrecording electronic microbalances, which the P-E autobalances shared with only one major competitor, P-E had over half of the market through 1976. P-E salespeople, however, showed*463 little interest in selling autobalances. The first IR spectrophotometer produced by PECC was the Model 700, beginning in 1971. From 1975 through 1981, PECC produced successor products in the Model 700 series, which generally had customer list prices between $ 4,000 and $ 8,000. In 1981, PECC began production of the Model 1300 series, with list prices between $ 8,000 and $ 13,000. PECC recorded sales of IR instruments as follows: IR Instrument Years Produced Sales Models 710, 710B 1975-1981 $ 2,584,804 Models 727, 727B 1975-1981 3,198,855 Models 735, 735B 1975-1981 1,906,588 Model 1310 [1981] 113,530 Model 1320 [1981] 152,950 Model 1330 [1981] 483,000 Total $ 8,439,727 The Model 700 series instruments, P-E's lowest priced IR spectrophotometers, had become technically reliable by 1975, and the sales force was comfortable selling them. However, the P-E market share did not approach that of Beckman Instrument Co. in this low-priced segment, and the SP2 report for 1976 characterized these products as "extremely old" and the "only weak link in the [IR instrument] chain". P-E introduced the Model 1300 series in 1981 and within a few months discontinued selling the instruments*464 replaced, Models 727B and 735B. The Model 710B remained as P-E's lowest priced IR spectrophotometer. Even after the exclusion of autobalances and HGA furnaces, see supra note 5 and p. 31, most of the PECC accessory dollar volume, which was nearly $ 19 million in total, was in the AA product line. The highest dollar volume accessory was the ED lamp power supply, with sales of $ 4.8 million. Other PECC AA accessories generated another $ 7.9 million in sales. The PECC sales of IR and gas chromatography accessories were $ 5.1 million. In 1980 alone, PECC produced at least 30 accessories with differing part numbers. The P-E weighted average resale price of a PECC accessory sold to an unrelated party, after all applicable discounts, was $ 294 in 1980 and $ 341 in 1981. Only a few of these accessories had unit selling prices well below $ 200, but the range was from less than $ 40 for AA burner chambers and lamp brackets to over $ 2,600 for a sophisticated ED lamp power supply. PECC also built over 50 types of HC lamps with differing cathodes, some of them multi-element. The customer list price for most models was between $ 125 and $ 300, and P-E offered discounts of 10 and *465 20 percent for volume purchases of 7 and 12 lamps, respectively. The HC lamp adaptor sold by P-E, which accommodated the smaller HC lamps of competitors, had an average list price of about $ 100; a similar adaptor was available from a large supply house in 1981 for $ 70. D. Operations PECC, which conducted its operations in a facility leased from the Puerto Rican government, owned the machinery, equipment, and tooling, most of which it purchased from P-E. PECC grew from 7 employees in October 1970 to 93 in 1980. Most employees had high school educations, with many having electronics or other technical training but none having an engineering degree. PECC experienced low employee turnover and, generally, was at least as efficient as Norwalk in terms of labor hours for transferred products. The general manager at PECC reported directly to the head of Manufacturing. In the early 1970s, when P-E sent parts to PECC for a newly transferred finished product, it also sent various types of documentation, including a set of blueprints and detailed part drawings. At least one person at PECC, the general manager, was capable of interpreting this technical documentation, and the blueprints*466 were sometimes used to troubleshoot products during the final assembly and test stage. Later parts shipments for an already transferred finished product occasionally included engineering change notices and documents affected by the changes. Sometime during the years in issue, PECC began updating its prints based on the change notices instead of continuing to receive new prints. Other documentation sent to PECC by P-E included assembly drawings, assembly and test specifications, and operations sheets describing in some detail how to assemble the product. The Instrument Assembly and Test department at PECC used a progressive production line, consisting of various stations, to combine subassemblies and other parts into instruments and accessories. In addition to the final assembly operations, completion of an instrument involved optical alignment, final testing, and quality checks. Optical alignment took 2 to 4 hours, and final testing averaged about 15 hours. Production of accessories was similar to, but generally less complex than, production of instruments. For the instruments and accessories worked on by PECC, the final assembly and test function was similar to that at P-E. *467 PECC established an HC lamp production area in 1971. Cathode fabrication involved either the mixing and compressing of metal powders, followed by oven heating, or the induction heating of pellets; the resulting metal slug was then lathed into shape. Later, glass lathing joined the outside bulb to the stem and assembly. The lamp at some point was filled with an inert gas and then purged of the gas, with several repetitions, and a vacuum system came into play during the final stages of production. As a test of the completed HC lamp, its intensity was checked while in use in an AA instrument. PECC incurred both scrap and rework costs during production. Although production problems usually did not result in the scrapping of an entire instrument or accessory, defective HC lamps were scrapped. If Quality Assurance in Norwalk discovered major defects in PECC finished products, they were reworked in Norwalk; P-E charged the associated labor costs to PECC. The PECC production volume was neither constant nor assured, which at least once caused PECC to lay off some employees. PECC production scheduling depended heavily on in-hand customer orders at P-E, and, as noted, customer demand*468 for a given instrument model commonly peaked 2 to 3 years after introduction and then declined. Another influence on an uncertain production volume was the refusal of Instrument Division management to transfer production to PECC if the move would cause a loss of jobs in Norwalk. A mitigating factor for PECC, however, was the requirement of the Puerto Rican government that PECC maintain a specified minimum level of employment, with which management strived to comply. In addition to uncertain production volume, PECC had no guarantee of a minimum gross margin on a product. For example, the PECC >costs of sales for HGA furnaces exceeded net sales revenue in both 1978 and 1979, resulting in negative gross margins. E. Parts and Parts Kits PECC purchased from P-E substantially all of the parts required for its finished products. These parts included electrical items such as capacitors, resistors, transistors, and transformers, and mechanical items such as sheet metal parts, castings, machined parts, plastic parts, and optics. Essentially all of these individual parts were either standardized, off-the-shelf items readily available from third parties, often by catalog order, or*469 they could be custom-made by, for example, machine shops or sheet metal shops. P-E sold many parts to PECC that had been purchased from suppliers and then shipped without any application of P-E direct labor. Although PECC itself sometimes produced electrical or mechanical subassemblies that went into its finished products, in other cases P-E produced subassemblies before shipping parts to PECC. For the parts to which P-E applied direct labor manufacturing and finishing operations, examples of such operations were: (1) Manufacturing of printed circuit boards; (2) machining, including lathe work on steel tubing and copper rods; (3) injection molding of plastic parts; (4) sheet metal work; (5) manufacturing of optical parts; and (6) for HC lamps, manufacturing of graded seal bulbs, cathode holders, and lamp bases. The P-E standard costs for these types of parts often approximated or exceeded the cost if purchased from outside suppliers. However, P-E rarely contracted out to others the specific parts it manufactured and sold to PECC; very few of the parts that Arthur Andersen analyzed in account 2403, see supra pp. 20-21, were identical to any of those that P-E sold to PECC. *470 For each of its finished products, P-E maintained a computerized listing of the parts necessary for its manufacture. For cost accounting, invoicing, and shipping purposes, certain listed groupings of parts, known as "parts kits", were each assigned a unique part number with the prefix "031". In addition to meaning a listed grouping of parts, "parts kit" could refer to the physical parts themselves. In this latter sense, however, "kit" was a misnomer because the parts in a given parts kit were not packaged in an individual grouping for shipment. There was no market for physical parts kits as such, and P-E did not sell them in the ordinary course of business to unrelated parties. A complicated product could have several parts kits in its comprehensive parts listing, and often a parts kit was one of the listed parts in a higher level parts kit. The dual ED lamp power supply, for example, had seven parts kit listings, some of which were embedded in one or more of the others. A parts kit also sometimes included parts without an 031 prefix that, like parts kits, were composites of several other parts. The parts kit system enabled PECC to streamline its ordering of parts for a particular*471 finished product, and, when there were embedded parts kits, PECC ordered the high-level one. When P-E billed PECC for the parts, the invoice listed only the part number and price of the high-level parts kit. The invoice did not list embedded parts kits and other parts that comprised the high-level parts kit. For ease of inventory retrieval and shipping, P-E first resorted and consolidated the listings of constituent parts in the parts kits ordered by PECC, in part number order, on an "8080" listing. The parts were then drawn from inventory against the 8080 listing. When the parts arrived at the PECC receiving area, in up to 40 boxes, the PECC receiving personnel unpacked the shipping crates, visually inspected and counted the parts, documented any damages, and then routed the parts to the appropriate work centers. Missing and backordered parts were included in subsequent shipments. PECC did not keep an inventory of spare parts and thus was dependent upon P-E to supply replacements for defective parts discovered during production. This lack of inventory could lead to the stoppage of a production line while awaiting the replacement parts. Arthur Andersen calculated the P-E*472 inventory carrying cost for parts sold to PECC as $ 1.8 million during the period in issue, based on a cost of capital of 37 percent. The bank prime interest rate during this time, which averaged 10.58 percent, closely approximated the P-E cost of short-term borrowing. F. Relationship and Interaction with P-E 1. Sales and Royalty Transactions For the parts it sold to PECC, P-E charged its standard cost plus 21 percent through 1978 and standard cost plus 13.12 percent thereafter. The P-E standard cost included incoming freight and inspection. For the 7-year period in issue, PECC bought parts for finished products at $ 4.6 million over the P-E standard cost of $ 27.8 million, for an average markup of over 16 percent. The P-E invoices for parts stated that the parts were f.o.b. P-E, and PECC paid the freight to Mayaguez. Through August 1977, PECC sold its finished products to P-E at a discount of 25 percent off the P-EU.S. list price on the date P-E issued its purchase order. Beginning in September 1977 and continuing through 1981, the discount was 30 percent. 7PECC invoiced P-E at the time of shipment to P-E. *473 PECC paid royalties to P-E based on the prices PECC charged P-E for the finished products, which, as noted, were 25 or 30 percent below the P-EU.S. list prices. The royalty rate was 3 percent. 8 From 1975 through 1980, PECC paid royalties on several products, including all accessories it sold to P-E, that were not identified in the 1970 licensing agreements with P-E. 2. Intercompany Account P-E and PECC used an intercompany account to record transactions between them. Amounts owed by PECC for administrative services, parts purchases, royalties, and other intercompany charges were netted against amounts owed to PECC by P-E. P-E, which was the net debtor in this arrangement, made cash payments to PECC to reduce its intercompany payable balance. As determined by Arthur Andersen, the P-E intercompany payables were outstanding for an average of 77 days. The PECC invoices, however, *474 called for payment by P-E on finished products within 30 days. 3. Norwalk Support Group The Norwalk Support Group, organized in August 1974 and based in Connecticut, consisted of two to five employees who monitored the PECC activity and provided continual manufacturing engineering, production planning and scheduling, and other support services. The salaries, employee benefits, and allocable overhead expenses of the Norwalk Support Group members were included in the PECC general and administrative overhead. The manufacturing engineering function in the Norwalk Support Group identified finished products with potential for transfer to PECC, collected and sometimes prepared necessary technical documentation, developed training programs for the PECC employees, assisted in the acquisition or repair of capital equipment, and monitored the quality of PECC instruments. The production planners in the Norwalk Support Group used the MRP system in scheduling the PECC production. See supra pp. 17-18. Other specific responsibilities included preparing the purchase orders and sales orders for parts and finished product transfers between P-E and PECC, determining the parts necessary*475 for the manufacture of new products, and expediting parts through Purchasing. With the assistance of the MRP system, the production planners released a parts kit to PECC by issuing a PECC purchase order to P-E; simultaneously, they issued a P-E purchase order to PECC for the purchase of the associated finished product. PECC production generally corresponded closely to customer orders, in the sense that a customer order for the finished product was in hand at P-E over 80 percent of the time when PECC placed a purchase order for parts. An order was almost always in hand at the time PECC completed a finished product for shipment to P-E. 4. Other As part of its regular activities and with no special effort, Purchasing in Norwalk acquired the outside parts needed for the PECC finished products. A small scale purchasing operation at PECC, which would have required three or four people, would have been largely a duplication of Norwalk cost and effort. An on-site PECC purchasing department, however, obviously would have been closer to parts suppliers, machine shops, and sheet metal shops that were available in Puerto Rico. Although PECC sometimes recommended process changes to*476 P-E that could affect the operations sheets, PECC had no research and development facilities or personnel and no engineering department. Quality Assurance in Norwalk audited PECC finished products and performed tests that differed from those performed by the PECC final test personnel. PECC also lacked marketing, sales, and service functions, and among the P-E activities with which PECC had no involvement were preparation of application studies, customer training and support, and customer billing. P-E also handled all of the registration requirements for PECC products with Federal regulators. Training for PECC employees often involved Norwalk personnel, whether the Norwalk Support Group or others. Sometimes a PECC employee spent training time in Norwalk, and sometimes a Norwalk technician or engineer traveled to PECC. P-E charged PECC for at least some training expenses every year, but the charges did not include training costs incurred by Engineering personnel. P-E performed several functions in distributing PECC finished products: marketing and sales, installation, customer training and support, and service. PECC products were never advertised under that name. Depending *477 on its nature, a PECC finished product was the technical marketing responsibility of the AA, the IR, or the gas chromatography Product Department. Although P-E provided the same high level of customer support and service for the AA and IR instruments produced by PECC, the USSD salespeople typically did little except respond to telephone inquiries and take telephone orders. Nonetheless, the salespeople were familiar with the characteristics and specifications of these instruments, partly through memoranda disseminated by the Product Departments. PECC did not provide a warranty on the finished products sold to P-E, nor did it reimburse P-E for product defects discovered by users. For the period 1976 through 1981, Arthur Andersen determined that USSD installation and warranty expense was 3.1 percent of P-E's total net sales of PECC finished products to unrelated parties, while billable and contract service income net of associated expenses was 2.7 percent of those sales. 9 *478 5. P-E Resale of PECC Finished Products The Instrument Division resold the PECC finished products under the P-E name and trademarks through the same sales channels it used for its other products. Like Instrument Division products in general, a large portion of PECC finished products went into foreign markets. Arthur Andersen, which assumed a 30-percent discount from domestic list prices for all years, see supra note 7, concluded that during the period in issue P-E earned an average 31.1 - percent gross margin on resales to unrelated parties of the PECC finished products. The high was 32.4 percent in 1980 and 1981, and the low was 29.7 percent in 1979. By type of finished product, the average gross margins were 31.5 percent for instruments (ranging from 29.7 to 33.9 annually), 32.0 percent for accessories (ranging from 29.9 to 35.6 annually), and 29.5 percent for HC lamps (ranging from 27.8 to 30.4 annually). In 1980 and 1981, the spread between the highest and lowest margin product groups was over five percentage points. According to these calculations, resales in foreign markets were over one-third of the total resales to unrelated parties. For the PECC finished*479 products that P-E resold to unrelated parties in 1980 and 1981, the consolidated gross margins by product category were as follows: [1981] Instruments 61.4% 62.4% HC lamps 70.2% 71.0% Accessories 73.8% 72.8% These gross margin figures are a function of the P-E net sales revenue and the combined standard costs of P-E and PECC. Higher gross margins for HC lamps and accessories were typical even when P-E alone manufactured and sold these products. As determined by Arthur Andersen, P-E held the PECC finished products in inventory for an average of 27 days before shipment to customers. The holding period was 20 days for instruments and 33 days for HC lamps and accessories. V. Potential Comparable Transactions A. Finished Products 1. Spectra-Physics and P-E An integrator is an electronic data handling accessory that analyzes measurements made by a gas chromatography instrument. During the years in issue, the Instrument Division purchased and resold integrators manufactured by Spectra-Physics, Inc. (Spectra-Physics). The annual agreements establishing the terms and conditions treated P-E as a nonexclusive, worldwide OEM dealer (see supra p. *480 21 for "OEM" definition) for integrators, associated accessories, replacement parts, and supplies. The agreements sometimes set forth an annual delivery schedule listing the number of units per quarter and always included a provision requiring 90 days' notice for P-E schedule changes or shipping requirements. For fiscal 1976, as specified in the agreement dated July 31, 1975, the price charged by Spectra-Physics to P-E was to be a 32-percent discount from the P-E domestic list price if P-E accepted delivery of at least 300 units. The lowest discount was 26 percent for delivery of 100 to 249 units. The discount for supplies and accessories coincided with the applicable integrator discount. The express payment terms were net 30 days from the date of the Spectra-Physics invoice. The agreement also stated that Spectra-Physics would make personnel available to provide product or service training, upon request, up to 10 days in each of Europe and the United States, with both parties bearing their own expenses. The parties later amended the July 1975 agreement, effective for shipments made after October 28, 1975. Under the revised agreement, the maximum discount, again for 300 units*481 or more, was 30.5 percent off the Spectra-Physics domestic list price. The agreement for fiscal 1977 was substantially the same as that of the previous year as amended. There was, however, no provision covering training. In addition, the new agreement included a warranty that Spectra-Physics would repair free of charge all integrators rejected by P-E within 30 days of the shipment date. The parties also agreed on the applicable charge for nonwarranty repair work performed by Spectra-Physics. The fiscal 1978 agreement lowered the delivered quantity to 250 to qualify for the maximum 30.5-percent discount, but was otherwise substantially the same as the preceding agreement. The agreement covering June 1980 through May 1981 specified: (1) A 200-unit delivered quantity to qualify for the maximum discount of 28.5 percent; (2) exact prices rather than the normal integrator discount for several parts, accessories, and supplies; and (3) an intricate method for determining the Spectra-Physics repair charge for nonwarranty work. Spectra-Physics painted the integrators in P-E colors, which enabled P-E to resell them without further manufacture. P-E resold them as the Model 1 computing*482 integrator and the Model 2 calculating integrator, at list prices ranging from $ 2,300 to over $ 5,000. Under the nonexclusive annual agreements, P-E sold these products in competition with those sold by Spectra-Physics through other channels. A 1976 internal P-E memorandum referred to the "awful experience" of customers rejecting attempts to sell at higher prices than those of Spectra-Physics, and P-E generally sold at prices no higher than what Spectra-Physics charged users. Spectra-Physics provided some service training and documentation to P-E but only limited marketing support, such as brochures. In 1975, the gas chromatography Product Department conducted seminars for the product specialists and prepared and distributed its own brochures. The Model 1 product brochure stated that the integrator would be "installed and working in minutes", and neither integrator product brochure mentioned customer training, a newsletter, or an applications cookbook. In accordance with the annual agreements, Spectra-Physics provided a free operator's instruction manual with each unit shipped. P-E resales for both models peaked in 1977 when they were becoming obsolete, and by this time P-E*483 had already begun to institute major price cuts. For those Spectra-Physics integrators resold to unrelated parties, Arthur Andersen calculated P-E's average gross margin on over $ 6 million in sales as 34.1 percent. Arthur Andersen also concluded that: (1) P-E paid Spectra-Physics an average of 50 days after the invoice date during 1978 through 1981; (2) P-E held the integrators in inventory an average of 34 days from date of receipt until shipment to customers; and (3) during 1976 through 1981, USSD installation and warranty expenses for the integrators averaged 4.5 percent of net sales to unrelated parties, with billable and contract service income net of associated expenses equal to 3.0 percent. 2. Hitachi and P-E P-E began selling fluorescence instruments in 1968. The largest selling instrument of one of P-E's major competitors had been in existence for over 10 years by 1975 when the worldwide fluorescence market outside of Japan was about $ 14 million in sales. Over $ 10 million of this market was instruments priced below $ 9,000, a segment in which the P-E market share was less than 10 percent. However, because its market share exceeded 50 percent for instruments*484 priced above $ 9,000, P-E had over 30 percent of the market in the $ 3,000-to-$ 16,000 price range. Although the P-E share of the mid-priced market did not increase in succeeding years, its annual sales in the fluorescence product line more than doubled between 1975 and 1979. The geographical breakdown for the entire fluorescence market in 1979, which was very similar to that of AA spectrophotometers, was 45 percent United States, 35 percent Europe, 10 percent Japan, and 10 percent other. On August 1, 1971, P-E and Hitachi, Ltd. (Hitachi), a Japanese corporation, entered into a sales agreement under which P-E agreed to use its best efforts to distribute various scientific instruments, related accessories, supplies, and spare parts manufactured by Hitachi, including fluorescence products. As amended and extended from time to time, the agreement was in effect throughout the years in issue. The P-E distribution rights for fluorescence instruments, which generally covered North and South America and western Europe, were with minor exceptions exclusive under the agreement. The Hitachi prices to P-E, f.o.b. Japan, were to be determined in advance by mutual agreement, subject to changes*485 by Hitachi after appropriate notice to P-E. P-E was obligated to make payments within 30 days after receipt of the products. The sales agreement, which did not specify an ordering lead time, stated that Hitachi would sell and ship products subject to "the terms and conditions of its standard sales order". P-E was to use its best efforts to advertise the products, and Hitachi agreed to provide literature, specifications, and manuals at P-E's request to assist in the preparation of advertising. P-E agreed to submit its advertising material for preapproval by Hitachi if performance or operation was mentioned. Hitachi agreed to submit its own planned advertising copy to P-E for comment. Hitachi warranted that its products were free from defects in material and workmanship and met applicable specifications for the shorter of 12 months after delivery to P-E or 16 months after shipment from the Hitachi factory. For product defects covered by the warranty, if repairable by P-E, Hitachi was to send free replacement parts for installation at P-E expense. The fluorescence Product Department bore responsibility for the Hitachi fluorescence instruments, which were sold under the P-E name*486 and trademarks by P-E salespeople, installed by P-E service technicians, and backed by a P-E warranty. These instruments, which had customer prices generally between $ 3,000 and $ 15,000, were similar in technical complexity to the instruments produced by PECC. P-E used essentially the same marketing and selling techniques for fluorescence products as it did for the AA and IR product lines. The P-E standard cost for Hitachi instruments included freight from Japan and U.S. customs duties. As determined by Arthur Andersen, P-E purchased over $ 3 million of fluorescence instruments from Hitachi during the years in issue, earning an average gross margin on resale to unrelated parties of 36.2 percent. At times, actual margins fell short of projections because of unfavorable currency exchange rates and unanticipated increased costs. Arthur Andersen further concluded that in 1979 P-E paid Hitachi on average 19 days prior to receipt of an instrument, and, after receipt, held the instrument in inventory for an average of 76 days before shipping to the customer. 3. Analect and P-E During calendar 1981, P-E entered into an OEM agreement (see supra p. 21 for "OEM" definition) *487 with the Analect Instruments Division of Laser Precision Corp. (Analect), covering the P-E purchase, for nonexclusive worldwide resale, of the Analect Model FX-6200 spectrophotometer and accessories. The FX-6200 was a Fourier Transform IR (FTIR) instrument, nondispersive in nature as opposed to the dispersive technology in P-E and PECCIR instruments. 10 In the United States, the FTIR market was over 20 percent of the entire IR market. As sold to users by Analect, the FX-6200 had a relatively low price for an FTIR instrument, in the range of $ 30,000 to $ 35,000. Analect held about a 20-percent share of the low-priced FTIR market. At the time the parties negotiated the agreement, P-E was selling no IR products employing the nondispersive technology. Management saw purchasing the FX-6200 as a quick entry into the low-priced segment of the FTIR market while awaiting the completed*488 development of a P-E product. Management also believed that gaining this foothold would facilitate the later marketing of the product developed by P-E. The P-E product development process had begun a few years earlier. By the end of calendar 1979, P-E management was well aware of the fundamental advantages of the FTIR nondispersive technology, which had the potential to improve cost effectiveness and operating convenience, partly through a tremendous reduction in analysis time. A P-E study group recommended the initiation of a project to develop an FTIR instrument, to be ready for shipping in January 1983. Most in the industry believed that FTIR was an emerging dominant technology that would render dispersive techniques largely obsolete. The failure of P-E to offer an FTIR product to customers, which caused it to lose IR market share during 1980, threatened its reputation as the world leader in the IR area according to the SP2 reports prepared in 1980 and 1981. Upon receipt of the FX-6200 from Analect, P-E performed no additional manufacturing work on it. However, as provided in the OEM agreement, P-E did not purchase the Analect control and display terminal for the instrument. *489 Instead, P-E combined the FX-6200 with its own data station, essentially a computer, for sale under the P-E name and trademarks as the Model 1500 or Model 1550 system, depending on the data station. The P-E data stations, which had substantially greater capabilities than the relatively primitive Analect control and display terminal, required special applications software, which P-E provided. Analect intended the integrated P-E system to enter a higher priced market segment, thus bypassing direct competition with Analect. The OEM agreement stated that the price per unit to P-E would be 30 percent off the Analect domestic list price at the time of ordering. Analect did not have a separate list price for its control and display terminal, but, for purposes of its discounting arrangement with P-E, it adjusted the list price of the complete FX-6200 system (based on proportionate standard costs) to exclude the terminal. Accessories and spare parts were subject to the same 30-percent discount, and the accessories that P-E purchased included items that had Analect list prices of $ 3,500 and up. The agreement allowed P-E 45 days from delivery to pay for the products. Analect agreed *490 to provide both software assistance for the merging of the P-E data station and the FX-6200 and "reasonable technical support" for up to 12 months. In actuality, Analect provided marketing assistance, complete service documentation and training, user's manuals, and the contemplated assistance in interfacing software. The technical support, for which there was no separate compensation, also included visits by Analect personnel to the P-E facilities and visits by P-E personnel to the Analect facilities. The IR Product Department was responsible for the technical marketing of the Models 1500 and 1550, including the preparation of brochures and data sheets, the conducting of seminars, and the performance of application studies. Analect warranted that its products would be free of defects for 1 year from shipment. The recourse for P-E was to send a defective product to the Analect plant, freight prepaid, where Analect would decide whether to repair or replace it. Analect was to bear the return freight charges. In accordance with their agreement, P-E paid Analect $ 250,000 up front for an option to enter into a license in January 1982. By the terms of this license, P-E was to have*491 the exclusive right, except for Analect itself, to use Analect manufacturing information in making and selling the FX-6200 worldwide. P-E exercised the option in January 1982, at which time it paid an agreed upon royalty of $ 1,000,000. P-E never used this licensed technology to manufacture the FX-6200. P-E did not make FX-6200 purchases after calendar 1984. From fiscal 1982 through fiscal 1985, as calculated by Arthur Andersen, P-E sold 70 of the FTIR systems to unrelated parties for over $ 2.7 million, earning an average gross margin of 25.2 percent. The P-E list prices ranged from about $ 35,000 for a basic version of the Model 1500 to over $ 80,000 for a sophisticated version of the Model 1550. For the three largest selling versions of these systems, the list prices of the P-E additions to the FX-6200 (i.e., data station, software applications package, and sometimes a printer) ranged from one-fourth to one-half of the P-E list price for the entire integrated system. According to Arthur Andersen: (1) P-E paid Analect, on average, 35 days after the Analect invoice date; (2) P-E held the FX-6200 in inventory for an average of 39 days after receipt until shipment to a customer; *492 and (3) USSD incurred warranty and installation expenses that amounted to 8.7 percent of net sales to unrelated parties, while billable and contract service income net of associated expenses was 1.5 percent. 4. P-E and Scientific Products The Scientific Products Division of American Hospital Supply Corp. (Scientific Products) was a large, nationwide distributor of laboratory instruments, equipment, and supplies. Its catalog in the early 1980s listed over 70,000 items from over 900 manufacturers. During the period in issue, Scientific Products purchased for resale approximately $ 23.3 million of instruments, accessories, and repair parts from the Coleman Instruments Division of P-E (hereinafter Coleman), reselling them nonexclusively under the Coleman name and trademark. Most of Coleman's remaining $ 12.2 million in sales went to other unrelated U.S. dealers, but some went through USSD. Among the analytical instruments handled by Scientific Products were Coleman ultraviolet-visible (UV-VIS) spectrophotometers. The Coleman products sold to Scientific Products were less sophisticated than the PECC finished products, thus entailing less technical support by the distributor. *493 Scientific Products employed over 400 sales representatives, over 200 service people, and about 35 technical specialists. Although both Scientific Products and USSD salespeople visited customers, a Scientific Products salesperson generally had responsibility for a broader range of products and had less specialized training. The technical specialists, who were similar in background to P-E product specialists, were trained and salaried technicians or chemists who presented technical information to the customers and provided demonstrations of the more complicated Scientific Products instrumentation, which included the Coleman instruments. Coleman trained the technical specialists responsible for its products and provided telephone assistance thereafter. Coleman also trained and conducted seminars for users of its instruments. Among its marketing activities, Scientific Products participated in trade shows and seminars, advertised in journals, and prepared selling materials such as catalogs and mailings. Coleman, in addition to providing Scientific Products with marketing information such as applications brochures, data sheets, and instrument photographs, reviewed advertising copy*494 prepared by Scientific Products. P-E personnel also sometimes exhibited Coleman products at trade shows displaying Instrument Group products. The service personnel at Scientific Products did some installation work but spent most of their time performing warranty or billable service. Coleman trained those responsible for maintaining and servicing its products. Coleman also supplied instruction manuals and maintenance manuals. Scientific Products generally took responsibility for the relatively high labor portion of warranty costs, and Coleman agreed to supply free replacements for defective parts. Scientific Products contracted to purchase Coleman instruments and accessories at a discount of 33 percent off the Coleman list prices, provided that the total list price of the order was at least $ 10,000. Smaller orders had a lower discount, but Scientific Products usually attained the maximum 33 percent. Scientific Products recorded an average resale margin of 23.6 percent on Coleman products for calendar years 1975 through 1978. 5. P-E and Foreign Distributors The Instrument Division sold all of its product types, including those in the AA, IR, and gas chromatography product*495 lines, to P-E International, which in turn sold to over 100 independent foreign distributors or dealers. The discounts provided to the foreign distributors generally ranged from 10 to 20 percent off the P-E export list prices, which were higher than the domestic list prices. A given foreign distributor received the same discount on instruments and ancillary products within a product line, except that some distributors received a special 33-percent discount on the autobalance accessory. 6. Accessory Resales P-E purchased a variety of products for resale from unrelated parties, including accessories, consumable supplies, and replacement and repair parts. In the AA and IR lines, P-E generally did not actively market these products. The consumable supplies and replacement parts consisted of items such as chart paper, recorder pens, and graphite tubes. An inch-long graphite tube, designed to hold a sample while heating in a graphite furnace, sometimes needed replacement daily in heavy use. P-E's AA price lists offered several types. P-E neither sold nor quoted prices for individual tubes, and the price lists showed sets of 5, 10, 20, and 50 per box, depending on the specific*496 tube type. During 1980 and 1981, the published prices for 50-count boxes were from $ 450 to $ 970; for 20-count boxes, from $ 315 to $ 425; for 10-count boxes, from $ 120 to $ 205; and for 5-count boxes, from $ 70 to $ 84. During these years, P-E purchased from unrelated parties and resold to unrelated parties only one type of graphite tube for which the boxed price, in this case a box of 50, exceeded $ 200; the list price, $ 500 in September 1979, rose to $ 675 by June 1981. Graphite tube vendors did not aggressively pursue sales directly to instrument users until after 1981, when the tubes became widely available through distributors and catalogs. The P-E computerized sales database for 1980 and 1981 included detailed information by product line on P-E resales to unrelated parties of noninstrument products purchased from many vendors around the world. The weighted average gross margins on those products with an average unit selling price above $ 200 were as follows: Product Line 1980 Gross Margin 1981 Gross Margin AA 58.5% 62.4% IR 37.8% 43.9% Gas chromatography 36.8% 37.3% Graphite tubes comprised over 85 percent of the net sales volume of the over-$ *497 200 products in the AA line. Because of their relatively high gross margins, factoring out the graphite tubes would lower the AA gross margins to 41.1 percent in 1980 and 49.5 percent in 1981. For the products other than graphite tubes, taken collectively without product line distinction, the 1980 gross margin would be 37.7 percent and the 1981 gross margin would be 41.5 percent. 7. Deuterium Arc Lamps A deuterium arc lamp, an essential part of a deuterium background corrector for AA spectrophotometers, provides a source of light that passes through the sample area alternately with the HC lamp or ED lamp. For more sophisticated AA instruments, a UV-VIS background corrector has both a deuterium lamp and a tungsten-halide lamp. P-E offered deuterium background correctors for most of its AA instruments at list prices between $ 1,000 and $ 2,000, and UV-VIS background correctors were priced $ 350 to $ 500 higher. Deuterium lamps for use in background correctors had list prices between $ 200 and $ 400. P-E exerted no marketing and selling efforts to speak of for AA deuterium lamps. Deuterium lamps are also a source of ultraviolet radiation in UV-VIS spectrophotometers, which*498 were marketed by the fluorescence Product Department and Coleman. Initially installed at the factory, the lamp is an integral part of the UV-VIS instrument rather than part of an accessory. In this setting, a tungsten lamp provides the visible light source, and the instrument design incorporates a lever that conveniently moves one or the other lamp, tungsten or deuterium, into the light path without danger of radiation to the user. A February 1980 P-E price list indicated prices for replacement deuterium lamps in UV-VIS spectrophotometers ranging from $ 235 to $ 305. For both AA and UV-VIS instruments from 1975 through 1981, a deuterium lamp was not changed or removed unless it failed. The lamp was considered a consumable item because it had a finite life and needed replacement when it burned out, typically every 1 or 2 years for an AA instrument. Replacement installation required a partial disassembly of the instrument, electrical hook-up, and optical adjustment of the new lamp. Optical adjustment, performed with the instrument turned on, was potentially dangerous because of the heat and ultraviolet light emitted. There were also possible electrical hazards involved. Replacement*499 of a deuterium lamp took about 20 minutes, compared to less than 5 minutes for an HC lamp. An instrument user normally did not buy additional deuterium lamps when purchasing the associated AA instrument. The P-E price lists recommended service installation, which cost extra, for the deuterium lamps used in deuterium background correctors. The price lists did not mention service installation for either deuterium lamps used in UV-VIS background correctors or deuterium lamp assemblies used in certain deuterium background correctors. In fact, a 1975 pamphlet on the Model 360 and 370 spectrophotometers touted "Easy access" to the deuterium lamp assembly in the optional deuterium background corrector, with a lamp mount "designed to provide for easy interchange and quick, accurate alignment." Service installation was the general recommendation for deuterium lamps in UV-VIS spectrophotometers. Instrument manufacturers were available to replace deuterium lamps, as were independent service companies, but deuterium lamp manufacturers rarely offered service installation. Sometimes a user did not know specifically that a deuterium lamp needed replacing, instead recognizing only that the *500 instrument was not performing properly. When this situation arose, the user tended to buy a replacement lamp from a service person, who often worked for the instrument manufacturer, and to have that service person install the lamp during the service call. P-E did not manufacture deuterium lamps and instead purchased them from unrelated parties, mostly or exclusively GTE-Sylvania (Sylvania). The SP2 report prepared in 1978 stated: "We are considering shifting over to the engineering of our own deuterium arc lamp. These lamps are sold in tremendous quantities and are no end of trouble to ourselves and to our customers." P-E had purchased lamps from Sylvania for many years with no formal agreement and continued to do so during the years in issue based on purchase orders. The payment terms were net 30 days. The Sylvania lamps, which came to P-E with a warranty of 500 hours, bore the P-E name and part number. They were standard lamps that Sylvania adapted to P-E specifications communicated through Purchasing or Engineering. P-E did not provide its own warranty to customers. P-E purchased over 1,000 deuterium lamps from Sylvania in a typical month, including those for Coleman. *501 For 1979, Arthur Andersen concluded that 48 percent of the P-E purchases went into OEM production, 39 percent went directly to customers, and the remainder went to the service organization. Sylvania considered P-E an OEM customer, which was Sylvania's target market for these lamps, rather than a distributor. Although Sylvania maintained a distributor price list, it sold perhaps 20 lamps annually to any particular distributor. This low distributor volume was intentional because Sylvania had concerns about product liability arising from ultraviolet radiation and electrical hazards. The OEM prices charged by Sylvania to P-E were up to 50 percent below the list prices to distributors. P-E earned an average gross margin of 63.1 percent on the resale to unrelated parties of replacement deuterium lamps and lamp assemblies used in AA background correctors. The net sales of those items for which the price lists recommended service installation were 48 percent of the total in 1980 and 41 percent in 1981. The net sales of AA deuterium lamps and assemblies for 1980 and 1981 combined were less than 4 percent of the P-E net sales of PECCHC lamps. Arthur Andersen determined that in 1979, *502 for those AA deuterium lamps and assemblies that were sold to customers outside of the service organization, P-E held them in inventory for an average of 113 days. B. Parts On August 22, 1977, Coleman granted a license to Sao Paulo Laboratorio, Ltda. (Sao Paulo), an unrelated Brazilian dealer, to assemble and sell the Coleman Model 295 and Junior III spectrophotometers in Brazil. Because of a 100-percent Brazilian import duty at the time on finished instruments, Coleman had not been able to compete with local manufacturers. To avoid the import duties, Coleman and Sao Paulo intended the Coleman instruments to be assembled sufficiently in Brazil to warrant an "Assembled in Brazil" label. The license agreement required Sao Paulo to purchase all of the subassemblies from Coleman and to sell the instruments under the P-E name. Coleman was to provide, at no charge, assembly drawings, specifications, testing and calibration data, and training at P-E facilities. Sao Paulo was not obligated to pay a royalty. The two licensed models, which were less sophisticated than the instruments assembled by PECC, were small instruments that Coleman sold in the United States through Scientific*503 Products and other dealers. Upon arrival in Brazil, the subassemblies were almost completed instruments in terms of the finishing work left. Thus, the Sao Paulo final assembly and test operations, covering 2 to 4 hours per instrument, were relatively simple and straightforward compared to what PECC did. In setting the prices for the subassemblies sold to Sao Paulo, Coleman began with the export list price for the finished instrument, then subtracted a 33-percent dealer discount, and finally reduced the price for labor to be performed on the subassemblies. Based on 1981 sales and cost data, the subassembly price to Sao Paulo represented a markup over Coleman standard cost of 59 percent for the Junior III and 44 percent for the Model 295; for sales of finished instruments by Coleman to customers, independent of Sao Paulo, the markup over standard cost was 62 percent for the Junior III and 50 percent for the Model 295. Sao Paulo was able to sell the Coleman instruments at unusually high prices because of the import duties that deterred outside competition. C. Royalties 1. In General From the mid-1950s through 1981, P-E entered into over 100 licensing arrangements with*504 unrelated parties involving analytical instruments and associated technology. Hitachi manufactured, sold, and serviced analytical instruments while also engaging in research and development. See supra p. 50. In 1960, P-E and Hitachi entered into a 10-year agreement for the organization and operation of a jointly owned Japanese corporation called Hitachi Perkin-Elmer, Ltd. (HIPE). HIPE was to be an intermediary for the exchange of product and technical information and for the planning and coordination of engineering, manufacturing, and sales. The agreement stated that P-E and Hitachi would enter into license and technical assistance agreements with HIPE to provide for the manufacture of each other's products. In the ensuing years, the three parties entered into various technical exchange and licensing agreements. In connection with these agreements, P-E and Hitachi agreed to provide each other with all available technical information relating to the subject products. While the agreements recognized Hitachi as the progenitor and licensor of certain technological principles, such as liquid chromatography, they recognized P-E as the progenitor and licensor of AA technology. *505 In 1965, the parties entered into a supplemental technical exchange agreement that specified ranges of reasonable royalties. The royalty was: (1) 6 to 7.5 percent for a licensed product specifically designed by the licensor; (2) 2 to 4 percent for a licensed product designed by the licensee but employing concepts, techniques, or technical information of the licensor; or (3) for jointly developed products, some point in between ranges (1) and (2). In order to permit more effective direct communication between P-E and Hitachi, HIPE ceased acting as a liaison in 1970. In 1971, P-E and Hitachi directly entered into a comprehensive license agreement, whereby P-E granted Hitachi the exclusive right to use P-E technical information and patents to manufacture in Japan and a few nearby areas products from P-E's AA, IR, and gas chromatography product lines. The agreement required P-E to convey or make available all design and manufacturing information relating to the licensed products and, upon request, to send a technical expert to Japan for up to a cumulative 30 days during the term of the agreement, with travel expenses payable by Hitachi. P-E also granted Hitachi the nonexclusive *506 right to sell these products in Japan and several other countries in Europe and Asia. The products covered by the license included all of the instruments, accessories, and HC lamps produced under the licensing arrangement between P-E and PECC. Hitachi, which did not purchase parts kits from P-E for the instruments licensed, agreed to pay a 7.5-percent royalty on most of the licensed products, computed on the Hitachi "Net Sales Price". The "Net Sales Price" was defined generally as the price charged to "customers", with this exception: In the case of sales by * * * [Hitachi] to a subsidiary thereof or to an individual, corporation, partnership or association which is so closely associated to * * * [Hitachi] as to prevent arms length [sic] bargaining, the Net Sales Price shall be deemed to be that charged by * * * [Hitachi] for similar products sold in the same period to customers not thus closely associated. In October 1977, P-E licensed the manufacture and sale of its "diff3" instrument, a blood analyzer, to Coulter Electronics, Inc. (Coulter). The agreed royalty percentage of 7 percent, which decreased over time, was based on the "Net Selling Price", defined as "either*507 the actual price at which the article is sold by Licensee, f.o.b. Licensee's plant and shall not include taxes, freight and insurance, or the fair market value of the article when otherwise shipped for placement at a customer's site." The agreement, which explicitly defined "Licensee" to include any affiliate of Coulter, stated that only one royalty was payable for a particular instrument. P-E and Coulter subsequently disagreed about the royalty computation base and exchanged correspondence over several months. Coulter contended that it could sell the diff3 to affiliated resellers at a price less than that applicable to the end users, yet pay one royalty based on the lesser price. P-E maintained that Coulter was liable for royalties based on the price received by Coulter or an affiliate for the first sale in the chain to an unrelated party. In an internal memorandum assessing the P-E litigating position in February 1982, P-E counsel mentioned without elaboration or specific support that it was "customary practice in patent licensing" to give effect to third-party sales in royalty computations while disregarding intermediate sales between affiliates. However, counsel also pointed*508 out in the memorandum that, regardless of custom, a licensor and a licensee could agree on some other computation base. The Coulter position ultimately prevailed. 2. HC Lamps By written agreement dated August 18, 1977, S & J Juniper & Co. (Juniper), an English manufacturer of HC lamps with a primary market in Europe, granted a license to manufacture and sell HC lamps to Instrument Laboratory, Inc. (Instrument Lab), an unrelated U.S. instrument manufacturer. Instrument Lab acquired the exclusive right to manufacture and sell the licensed HC lamps in the United States, except that Juniper retained the right to sell through its existing agents and dealers. Instrument Lab was also authorized to sell the licensed HC lamps elsewhere worldwide, nonexclusively. Prior to this licensing arrangement, Instrument Lab bought finished HC lamps from Westinghouse, earning a 35- to 40-percent gross margin on resales. Juniper agreed to and did provide Instrument Lab with complete HC lamp technical documentation and know-how. Juniper agreed to train Instrument Lab employees in England at the expense of Instrument Lab and to provide any other assistance requested by Instrument Lab, including*509 visits by Juniper personnel, again generally at the expense of Instrument Lab. Juniper actually did provide training over a 1-year period to Instrument Lab employees and also assisted Instrument Lab in the acquisition and construction of the special equipment necessary to the manufacturing of HC lamps. Instrument Lab agreed to pay a 2-percent royalty for 10 years based on the invoice price of the first sale following manufacture, whether or not the purchaser was related to Instrument Lab. It also agreed to pay # 25,000 as technical assistance fees and # 35,000 as nonrefundable advance royalties, the latter being in addition to the 2-percent royalty. At the end of 10 years, Instrument Lab was entitled to continue using the Juniper know-how without any further consideration. Instrument Lab purchased its sheet metal, machine shop, and plastic base parts for HC lamp production from unrelated vendors, but fabricated a few parts itself. It did not purchase anything resembling parts kits from Juniper. The HC lamps covered by the license, which were usable in P-E instruments with the help of an adaptor, were similar to those produced by PECC. D. ARC From 1973 through the years*510 in issue, the P-E Optical Group manufactured microlithography equipment called the Micralign projection mask alignment system. This Micralign system, a revolutionary innovation for the production of semiconductor chips, used optics to project and etch a circuitry design onto the light-sensitive surface of a silicon wafer. The light source was a mercury capillary lamp specifically designed for the Micralign system. When introduced in 1973, the Micralign system was priced at $ 95,000, and, early on, P-E had trouble meeting the heavy demand for the system. Later models, in the early 1980s, were priced at about $ 500,000. When P-E was developing its Micralign equipment, management sought out a source for curved mercury capillary lamps because the common straight lamp, readily available, was unworkable. Advanced Radiation Corp. (ARC) had been organized in 1971 as a two-person enterprise. The president of ARC, Raymond Paquette, who eventually accumulated 10 or 11 patents relating to arc lamps, had a reputation for producing quality mercury capillary lamps largely from his prior work at another company. Hearing of the P-E need, he sent a few unsolicited sample lamps to P-E that incorporated*511 a technique for heating and uniformly bending straight capillary tubing. After talking to several companies, P-E settled on ARC as its supplier of unbased lamps, in part because of anticipated flexibility, cooperation, and low prices. ARC disclosed its heating and bending technique to P-E during the P-E vendor selection process, after which P-E improved on a part of the technique and provided ARC with equipment. Beginning in about 1972, P-E began purchasing curved unbased lamps from ARC by means of purchase orders. After a P-E employee spent a few weeks learning the manufacturing process at ARC, P-E set up its own lamp manufacturing facility. P-E nonetheless continued to purchase unbased lamps from ARC at the request of a major customer who wanted P-E to maintain a second source. P-E occasionally asked ARC personnel for advice concerning the manufacture of mercury lamps. By 1975, P-E held a patent on its mercury lamp mounting base, which precisely positioned the lamp upon installation and allowed for quick, easy, and precise replacement. Effective in May 1977, P-E and ARC entered into a Patent and Technology License Agreement. Under this agreement, P-E licensed both its patent*512 and its engineering and manufacturing technology so that ARC could manufacture, assemble, and sell based lamps worldwide on a nonexclusive basis directly to Micralign users. P-E also agreed to provide ARC employees with free training at mutually agreeable times. During the course of its fiscal years ending on January 31 in 1976 through 1982, ARC sold unbased Micralign lamps to P-E from 1976 to 1980, based Micralign lamps to P-E from 1979 on, and based Micralign lamps to others from 1978 on. From 1976 to 1981, the ARC annual unit sales of the lamps increased from 2,050 to 44,234, and annual net sales after adjustment for discounts increased from under $ 50,000 to almost $ 3 million. Between April 1975 and January 1982, ARC had Micralign lamp net sales, based and unbased, of nearly $ 9 million. ARC, which did not maintain a finished goods inventory and instead built lamps to specific customer orders, provided a warranty on the based lamps it sold to P-E and to others. It performed its own purchasing and manufacturing functions, including the testing of lamps it sold to P-E and to third parties. Although ARC had commissioned salespeople, they did not concentrate on the Micralign*513 lamps, which were marketed partly by word-of-mouth among Micralign system users. VI. Notice of Deficiency As the foundation for a section 482 allocation of income in the notice of deficiency, respondent treated PECC as a consignment "contract assembler" for P-E. Respondent applied a cost-plus method by setting the arm's-length PECC profits equal to a fixed percentage of its adjusted operating costs, which were in turn equal to: (1) the PECC recorded costs, less (2) the PECC recorded costs for royalties and materials paid to P-E or Bodenseewerk, plus (3) the PECC estimated location savings. The notice of deficiency, as it related specifically to the PECC finished products, allocated income of $ 29,299,836 to P-E for the 7-year period in issue. Prior to trial, respondent formally abandoned the contract assembler theory and its associated cost-plus methodology. OPINION I. Introduction From 1975 through 1981, PECC, a wholly owned subsidiary of P-E located in Mayaguez, Puerto Rico, built analytical instruments, HC lamps, and accessories. P-E sold PECC substantially all of the parts for the products at a price equal to the P-E standard cost for the parts plus a markup, *514 21 percent through 1978 and 13.12 percent thereafter. PECC sold the finished products to P-E at a 25-percent discount from the P-E domestic list prices through August 1977 and a 30-percent discount thereafter. PECC also paid royalties to P-E representing 3 percent of the amounts billed to P-E for the finished products. Section 482 authorizes respondent to allocate gross income, deductions, credits, or allowances between two related corporations if the allocations are necessary either to prevent evasion of taxes or clearly to reflect the income of the corporations. The applicable standard is arm's-length dealing between taxpayers unrelated by ownership or control. Sec. 1.482-1(b)(1), Income Tax Regs. A business purpose for an arrangement or a set of transactions does not by itself insulate a taxpayer from a section 482 allocation. Bausch & Lomb, Inc. v. Commissioner, 92 T.C. 525, 582 (1989), affd. 933 F.2d 1084 (2d Cir. 1991). Petitioner has conceded that respondent is entitled to make income allocations from PECC to P-E to conform to a 30-percent discount from domestic list prices for all years. See supra note *515 7. After this adjustment, which affects the intercompany finished product sales and the royalties derived therefrom, petitioner contends that the three transactions between P-E and PECC -- sale of parts, sale of finished products, and royalties -- were arm's length in amount, thereby precluding additional adjustments. Respondent's primary trial position is that none of the three transactions was arm's length within the meaning of section 482, thus necessitating income allocations to P-E of $ 6,221,911 on the sale of parts and $ 17,131,821 on the sale of finished products. After an allocation of $ 1,110,341 away from P-E for royalties, see infra p. 152, respondent's total income allocation to P-E is $ 22,243,391. Respondent's alternative trial position, which differs only in the amount of the parts adjustment, leads to a total income allocation to P-E of $ 29,290,304. Although the period in issue spans 7 taxable years and the PECC product mix varied considerably from year to year, the parties generally have been content to work with categories of finished products no narrower than instruments, accessories, and HC lamps, and to avoid making distinctions from year to year. To*516 the extent feasible, we have sought to respect the parties' simplified approach. II. Preliminary Matters Rule 151(e)(3) requires that the parties' briefs set forth proposed findings of fact based on the evidence. Petitioner's proposed findings include many drawn directly from the conclusions on various accounting matters appearing in the Arthur Andersen written reports. In her reply brief, respondent objects to many of these proposed findings on the ground that the underlying business records are not in evidence, a point repeated in respondent's proposed findings and argument. The four Arthur Andersen written reports are similar in format. After describing the questions posed by petitioner's counsel, they detail the assumptions, information sources (including business records), and procedures used to arrive at the conclusions. Petitioner does not dispute that much of the underlying information is not in evidence. We find unpersuasive respondent's approach to our use of the Arthur Andersen reports. Neither at trial, nor on brief, has respondent cast her objections in the form of a challenge to the admissibility of the reports into evidence -- reports which had been *517 furnished to respondent weeks before trial. Nor did respondent object, at trial, to any particulars of the testimony of the Arthur Andersen representatives, Robert G. Kutsenda and Gary E. Holdren, on the ground which now forms the basis of her objections. Consequently, we see no need to analyze the impact of rules 103(a)(1) (dealing with objections to admissibility) and 703 (dealing with the permissible bases of expert opinions) of the Federal Rules of Evidence, which are applicable to proceedings in this Court (sec. 7453 and Rule 143(a)), although it would appear that those rules of evidence offer little support for respondent's position. Under the foregoing circumstances, respondent's wholesale attack on the reports is unwarranted. III. Arbitrary, Capricious, or Unreasonable When the Commissioner has determined deficiencies based on section 482, the taxpayer bears the burden of showing that the allocations are arbitrary, capricious, or unreasonable. Sundstrand Corp. v. Commissioner, 96 T.C. 226, 353 (1991); Eli Lilly & Co. v. Commissioner, 84 T.C. 996, 1131 (1985), affd. on this issue, revd. in part, and remanded*518 856 F.2d 855, 860 (7th Cir. 1988). We must sustain the allocations absent a showing of respondent's abuse of discretion. Bausch & Lomb, Inc. v. Commissioner, 92 T.C. 525, 582 (1989), affd. 933 F.2d 1084 (2d Cir. 1991); G.D. Searle & Co. v. Commissioner, 88 T.C. 252, 358 (1987). Respondent based the notice of deficiency on a characterization of PECC as a consignment contract assembler. See supra p. 73. Prior to trial, respondent formally abandoned the contract assembler theory and its associated cost-plus methodology, adopting instead a position that respects the PECC status as a licensed manufacturer. Petitioner, while acknowledging that alternative positions do not result in per se arbitrariness of a deficiency notice determination, nonetheless contends that respondent's abandonment of both the factual and legal bases underlying the notice causes the determinations therein to be arbitrary, capricious, and unreasonable. In contrast to petitioner's position that the resolution of this matter was a foregone conclusion before trial, respondent argues that we *519 must take account of her experts' testimony presented at trial in making our determination. Here, respondent abandoned a theory before trial that, in similar cases, we have rejected after a trial on the merits. E.g., Sundstrand Corp. v. Commissioner, supra at 357-358; Bausch & Lomb, Inc. v. Commissioner, 92 T.C. at 583-584; Eli Lilly & Co. v. Commissioner, 84 T.C. at 1132-1133. In these circumstances, we hold that petitioner has met its burden of showing respondent's allocations to be arbitrary, capricious, or unreasonable. This holding, however, does not relieve petitioner of its burden of proving that the transactions between P-E and PECC meet the arm's-length standard; if it fails to do so, the Court must decide the proper allocations of income between P-E and PECC. Sundstrand Corp. v. Commissioner, supra at 354; Eli Lilly & Co. v. Commissioner, 856 F.2d at 860, 872. The task thus thrust upon us is, to put it mildly, frustrating. In this case, as in other significant section 482 cases before the Court in recent years, each party has spent most of the time attacking*520 the other party's allocation formula rather than establishing the soundness of its or her own formula. In pursuing this path, an unduly long and unnecessarily complicated trial record has been created, replete with bickerings between counsel over unimportant and often irrelevant evidentiary questions and a continuance of the "play your cards close to the chest" attitude of petitioner's counsel and the lack of focus on the part of respondent's counsel exhibited during the discovery process. Both of these elements seemingly reflect a strategy of telling the judge as little as possible. Thus, in the final analysis, the Court is left to its own devices without the usual anchor of decision, namely that, if the taxpayer fails to carry its burden of proof, the deficiency is sustained. The Court must find a formula, without the benefit of sufficient help from the parties as to what that formula might be. In a section 482 case, this task usually requires the Court to find a middle ground -- a task which it has disavowed, in other contexts, see Buffalo Tool & Die Mfg. Co. v. Commissioner, 74 T.C. 441 (1980), and one which is most difficult to perform in light*521 of the Court's inevitable lack of knowledge of the realities of the workings of a specific industry and of the business world generally, including particularly the international competitive atmosphere which those realities reflect. It is against this background that we now turn to finding precise answers based on an imprecise record. IV. PECC Sales of Finished Products A. Background Section 1.482-2(e), Income Tax Regs., describes allocation methods intended to reflect an arm's-length price for tangible property sales between a related buyer and seller. Such sales between related parties are called "controlled sales". Sec. 1.482-2(e)(1)(i), Income Tax Regs. "Uncontrolled sales", used for comparison purposes, are sales in which the buyer and seller are not related. Sec. 1.482-2(e)(2)(ii), Income Tax Regs. The three allocation methods described in detail are the comparable uncontrolled price method, the resale price method, and the cost plus method, in that order of priority. Sec. 1.482-2(e)(2), (3), and (4), Income Tax Regs. Under the comparable uncontrolled price method, the arm's-length price of a controlled sale between related parties is equal to the price *522 paid in comparable uncontrolled sales, as adjusted. Sec. 1.482-2(e)(2)(i), Income Tax Regs. Uncontrolled sales are comparable, and adjustments are permissible, if the physical property and circumstances involved are so nearly identical to the controlled sale that any differences both can be reflected by a reasonable number of adjustments to the price of the uncontrolled sales and have a definite and reasonably ascertainable effect on price. Sec. 1.482-2(e)(2)(ii), Income Tax Regs. Petitioner and respondent agree that there are no comparable uncontrolled sales available for applying the comparable uncontrolled price method to finished product sales from PECC to P-E. We will return to this method, however, in our consideration of the parts transactions. See infra p. 145. The next method in priority under the regulations is the resale price method, which generally applies only if there are no comparable uncontrolled sales and if: (1) An applicable resale price is available; (2) the reseller has not added more than an insubstantial amount of value by physically altering the property before resale; and (3) the reseller has not added more than an insubstantial amount of value *523 by the use of intangible property. Sec. 1.482-2(e)(3)(ii) and (iii), Income Tax Regs. Under the resale price method, the arm's-length price of a controlled sale is equal to the applicable resale price reduced by an appropriate markup percentage, with adjustments. Sec. 1.482-2(e)(3)(i), Income Tax Regs. The applicable resale price is ordinarily the price at which the buyer resells the property in an uncontrolled sale. Sec. 1.482-2(e)(3)(iv), Income Tax Regs. The appropriate markup percentage is the gross profit percentage, relative to sales, earned by a reseller on property that is both purchased and resold in an uncontrolled transaction, if the resale is sufficiently similar to the resale that follows the controlled sale. Sec. 1.482-2(e)(3)(vi), Income Tax Regs.Close physical similarity of the products in the compared sales is not required, but characteristics important to the similarity of resales are: (1) The type of product; (2) the functions performed by the reseller; (3) the intangible property, such as patents and trademarks, used by the reseller; and (4) the geographic market in which the reseller performs its functions. Id. Adjustments should reflect material *524 differences between the uncontrolled resale used to calculate the appropriate markup percentage and the resale that follows the controlled sale. Sec. 1.482-2(e)(3)(ix), Income Tax Regs. These are differences in functions or circumstances that have a definite and reasonably ascertainable effect on price. Id. The parties agree that the resale price method applies to the sale of finished products by PECC to P-E. As petitioner applies it, this method establishes that the transfer price between PECC and P-E, 30 percent off P-E domestic list prices after petitioner's concession (see supra note 7), was arm's length; this would entail no adjustment to P-E income other than the concession. Respondent, on the other hand, maintains that of her total income allocation to P-E under section 482, $ 17.1 million results from application of this method to the finished products. In reaching this adjustment, respondent uses purportedly arm's-length gross margins for P-E resales of 32.7 percent for instruments, 57.2 percent for HC lamps, and 57.6 percent for accessories. B. Resale Price Method and Product Aggregation Although the parties agree on the applicability of the resale price*525 method, they have a fundamental disagreement about whether that method should be applied to the finished products as a whole or to instruments, accessories, and HC lamps separately. Petitioner contends that the finished products should be aggregated, characterizing respondent's contrary position as a "hypothetical and theoretical disaggregation". Petitioner emphasizes that manufacturers generally granted uniform resale discounts (i.e., uniform discounts from list prices) to unrelated distributors and that a particular resale discount translated into a similar gross margin upon resale by the distributor. As part of the agreement between P-E and Spectra-Physics, the same contractual discount generally applied to the primary product, which was an integrator, and the associated accessories. In the agreement between P-E and Analect, the same discount applied to the FX-6200 and its accessories. The discount that Coleman provided to Scientific Products varied by order volume but did not vary between instruments and accessories. Finally, for products distributed in foreign markets, P-E International normally granted a uniform discount to a particular foreign distributor on instruments, *526 accessories, and HC lamps. Given these discount congruences, petitioner argues that we should not attempt to determine three separate margin percentages for the finished product categories. We disagree. One obvious difficulty with petitioner's position is that Spectra-Physics and Analect did not sell HC lamps to P-E, nor apparently did Coleman sell them to Scientific Products, yet HC lamps were 30 percent of the PECC output. The relationship of the accessories to the instruments in these transactions causes at least three further obstacles to our adopting an aggregation approach. First, the accessories involved in the Spectra-Physics and Analect transactions were only those directly associated with the primary product. Walter Doyle (Doyle), the president of Laser Precision Corp. who testified about the Analect transaction, stated that there was no reason for different discounts because the systems were typically sold as "packages" consisting of an instrument and one or more accessories. There was similar testimony about the Coleman products sold to Scientific Products. PECC, in contrast, sold accessories to P-E that were not limited to use with the PECC instruments. As one*527 example, PECC sold gas chromatography accessories but did not sell any gas chromatography instruments. Second, since the anticipated accessory or replacement part volume was relatively small for the Spectra-Physics and Analect transactions, there would be little apparent incentive to bother with differing discounts. It does not necessarily follow, however, that uniform discounts in this type of situation would stay uniform as the situation changed to one in which the secondary products were a more important part of the deal. The secondary products that PECC sold to P-E, accessories and HC lamps, were over half of the PECC sales volume. Third, the only two accessories that Doyle could specifically recall selling to P-E had Analect list prices of $ 3,500 and up, which places them near or in the PECC instrument price range. That these accessories had the same discount as the FX-6200 instrument says little about whether the PECC accessories and HC lamps, with average prices to users below $ 400, should have the same discount as the PECC instruments. For the foreign distributors, the generally uniform discounts provided by P-E were in the 10- to 20-percent range, well below the 30*528 percent advocated by petitioner as arm's length. In addition, petitioner's principal expert witness in economics, Dr. William J. Baumol, emphasized in his initial report that the foreign distributor transactions were not directly comparable to those between PECC and P-E. In these circumstances, we do not find the discount uniformity meaningful. Even if we agreed with petitioner that PECC would have granted an arm's-length distributor a uniform discount on instruments, accessories, and HC lamps, that point is merely a first step. The applicable regulations revolve around the appropriate markup percentage earned by the distributor, not the percentage discount arrangement for transfers between manufacturer and distributor. Sec. 1.482-2(e)(3)(i), (vi), Income Tax Regs. Petitioner attempts to bridge this gap by arguing that, in actual market transactions, a given discount translates into a similar margin upon resale by the distributor, at least if there are competitive products in the market. The P-E resales of PECC finished products arguably support this relationship in view of the small difference between the 30-percent discount to P-E and its overall 31.1-percent resale margin*529 as calculated by Arthur Andersen. See supra pp. 45-46. Nonetheless, the connection between discounts and corresponding resale margins is typically not an especially close one, judging by the transactions in the record. The P-E discount for Spectra-Physics products did not exceed 30.5 percent, except for 32 percent during the initial 3 months of the first year, yet Arthur Andersen determined a P-E overall resale margin of 34.1 percent. For Coleman sales to Scientific Products, which usually entailed a discount of 33 percent, the recorded resale margin was only 23.6 percent. Even the seemingly close connection for PECC finished products does not hold up under closer scrutiny. Over the course of the years in issue, as calculated by Arthur Andersen, P-E's annual resale margin for instruments varied by over four percentage points and for accessories by over five. Within a particular year, the spread between product categories sometimes exceeded five percentage points. See supra p. 46. There are several reasons why, for a specific sale and subsequent resale, the distributor's resale margin could differ from the resale discount provided by the manufacturer. For one, the*530 distributor's resale list price could differ from the manufacturer's list price, which served as the discount reference point in the Spectra-Physics, Analect, and Scientific Products transactions. As a factor applicable to P-E and PECC, their discount was computed against the P-E domestic list price for the product. The export price list differed from the domestic price list, and P-E resold many PECC products in foreign markets. The peculiarities of the foreign markets also sometimes necessitated special pricing, such as that for the Model 372 and 272 AA instruments assembled by PECC. Furthermore, P-E sales into foreign markets through P-E International were not directly to users; this different level of the market imparts further strain on the connection between resale discounts and resale margins. Even for a domestic sale, volume discounts and other customer incentives such as package deals were not unheard of. Any of these factors, or others, could have had a disproportionate effect on the resale margins of the three types of finished products. On a different tack, petitioner argues that imposing separate resale margins by product category treats P-E as something more than*531 a mere distributor of PECC finished products, citing Bausch & Lomb, Inc. v. Commissioner, 92 T.C. 525 (1989), affd. 933 F.2d 1084 (2d Cir. 1991). In Bausch & Lomb, we described the assumptions used by the Commissioner's economics expert in his determination of an arm's-length price for contact lenses purchased by a domestic corporation from an Irish subsidiary. The expert reduced his estimate of the arm's-length price in part because the domestic corporation had sales and research functions to support, which would require it to pay a lesser amount for the lenses than a distributor not so encumbered with other functions. In rejecting the expert's conclusions, we did not find it significant that the domestic corporation engaged in functions other than distribution. Bausch & Lomb, Inc. v. Commissioner, 92 T.C. at 589. Petitioner, who acknowledges that manufacturers earned varying profit margins on instruments, accessories, and HC lamps, appears to argue that separate treatment of the finished product categories is tantamount to injecting the P-E manufacturing function into an analysis that*532 should be limited to distribution activities. In Bausch & Lomb, however, we did not confront a similar product aggregation issue. We instead declined to consider activities other than distribution in our derivation of a comparable uncontrolled price for soft contact lenses, which were considered a fungible commodity. Id. Here, our resolution of the aggregation issue is independent of how the P-E manufacturing activities or margins differed across product categories. We also do not agree with petitioner that disaggregation is grounded in abstract economic principles, nor that it creates an unreasonable degree of economic sophistication that precludes practical application. Instruments, accessories, and HC lamps are fundamentally different functionally, with the potential for differing and predictable influences on their respective resale margins, including the "captive" tendencies we discuss below. See infra p. 113. Indeed, within the AA product line that comprised 80 percent of the PECC output, P-E management considered spectrophotometers, lamps, furnace atomizers, and accessories to be distinct sublines. Petitioner itself acknowledges on brief, in a different *533 context, that "The SP2 Reports regarded HCLs [HC lamps] as a major product line." The P-E accounting systems and records were sophisticated enough to track the three product categories separately, and the voluminous stipulated exhibits in the record include many financial documents with precisely this tripartite breakdown, including the original PECC financial statements. Dr. Baumol, petitioner's expert in economics, did not aid petitioner's cause on this matter. Although Dr. Thomas Horst, respondent's expert in economics, clearly stated his preference for separate product categories in his original report, Dr. Baumol did not comment specifically on this point in his rebuttal report. At trial, Dr. Baumol testified that generally "the more one can disaggregate, the better" and proceeded to explain why the resale margins on HC lamps and accessories might very well differ from an instrument margin. Moreover, although petitioner on brief invokes case law speaking of abstract economic principles, degrees of economic sophistication, and perfectly competitive markets in the economic sense, counsel for petitioner never asked Dr. Baumol about these concepts at trial, nor do they appear *534 in his written reports. We conclude that the resale price method is appropriately applied not to the PECC finished products as a whole, but instead to the separate categories of instruments, accessories, and HC lamps. C. Instruments Of respondent's proposed income allocation to P-E of $ 17.1 million derived from the resale price method as applied to finished products, less than $ 400,000 is attributable to instruments. This relatively small amount follows from a coupling of Dr. Horst's conclusions that P-E earned an actual resale margin on PECC instruments of 31.9 percent for the period in issue and that P-E at arm's length should have earned a margin of 32.7 percent. Respondent's choice as the most reliable comparable transaction for instruments under the resale price method is the P-E resale of Hitachi fluorescence instruments. Petitioner recommends instead, for the finished products as a whole, the P-E transactions with Spectra-Physics, Analect, and Scientific Products, in that order. The parties themselves recognize that none of the proposed comparable transactions is close to an ideal fit. An expert adjusted the resale margin of his proposed comparable upward if he*535 quantified a circumstance either favorable to the distributor or unfavorable to the manufacturer compared to the P-E dealings with PECC. Similarly, he adjusted the resale margin downward if he quantified a circumstance either unfavorable to the distributor or favorable to the manufacturer compared to the P-E dealings with PECC. Generally, an expert supporting the comparability of a particular uncontrolled resale mentioned some differences without attempted quantification, often on the rationale that quantification was not feasible or that the associated adjustments were in a favorable direction and thus unnecessary to quantify. The other side often countered with more unquantified adjustments. Because the overall similarity sought in the resale price method relates to the probable effect upon the markup percentage of any differences in characteristics, sec. 1.482-2(e)(3)(vi), Income Tax Regs., these circumstances undermined the proposed comparables. Another general shortcoming in the proposed comparable transactions is that AA (atomic absorption) instruments dominated the PECC production, yet neither party offers a comparable from this product line. Each of the Instrument Division*536 product lines had its own distinct Product Department, and, despite similar marketing programs among the Product Departments, the markets themselves were not the same. As illustrated by the annual SP2 reports, see supra p. 13, product lines at any given time could differ in several respects, including total market size, rate of market growth, P-E market share, P-E product and customer mix, completeness of P-E line, and number and identities of major competitors. Because of the parties' mutually broad approaches to the issues in this case and the record as conformingly developed, we cannot easily evaluate and even less easily quantify product line differences. Based on the evidentiary record and the parties' arguments, we discuss the proposed comparable transactions in descending order of value, in our view, to our ultimate determination of the appropriate resale margin for instruments. 1. Hitachi and P-E Similarities between P-E's resale of PECC instruments and resale of Hitachi fluorescence instruments include the price range and technical complexity of the instruments, the use of the P-E name and trademarks upon resale, installation and warranty provided by P-E, an*537 established analytical technology, and exclusive distribution of the products by P-E. See supra pp. 50-52. Petitioner and respondent agree, coincidentally in view of their differing computation methods, that P-E earned an average margin of 36.2 percent on resales of the Hitachi instruments to unrelated parties. From this starting point, Dr. Horst made a financing adjustment for differences in payment practices between the Hitachi/P-E and PECC/P-E transactions. The period relevant to this adjustment was the time between the P-E cash outflow (payment to the manufacturer) and the P-E cash inflow (payment received from the customer). 11 Dr. Horst accepted the Arthur Andersen conclusion that a detrimental*538 95 days elapsed between P-E's prepayment to Hitachi and P-E's shipment of the instrument to a customer. See supra p. 52. He computed the comparable period for the PECC transaction, using as inputs a 4-day shipping period from PECC to P-E, the 20-day inventory holding period at P-E determined by Arthur Andersen, see supra p. 46, and a 50-day period between PECC's shipment to P-E and P-E's payment to PECC. The resulting time between P-E's shipment to its customer and its later payment to PECC was a beneficial 26 days (i.e., 50 minus 20 minus 4). Combining P-E's 95-day detriment in the Hitachi transaction, where P-E was in effect a lender, with its 26-day benefit in the PECC transaction, where P-E was in effect a borrower, meant that the Hitachi transaction was 121 days worse for P-E from a cash flow standpoint. Dr. Horst applied the P-E short-term cost of borrowing, 10.58 percent, to this 121 days, which caused him to reduce the unadjusted Hitachi resale margin of 36.2 percent by 3.5 percentage points. 12 This left an adjusted margin of 32.7 percent. *539 Petitioner disapproves of Dr. Horst's use of the P-E short-term borrowing rate of 10.58 percent rather than the 37-percent estimate of the P-E weighted average cost of capital offered by Dr. Sanford J. Grossman, petitioner's other economics expert. According to Dr. Grossman: (1) The annual cost of capital can be defined as the before-tax earnings level that a firm must earn to provide its shareholders and creditors with their required rates of return; (2) P-E must offer a rate of return that rewards the shareholders of its common stock for the risk associated with fluctuations in P-E's value, and, on a before-tax basis, this rate of return ranged from 34 percent in 1975 to over 50 percent in 1981; and (3) the required return on debt for the P-E creditors, with the long-term component based on the AAA-rated bond yield, ranged from under 7 percent in 1977 to over 13 percent in 1981. Weighting the equity and debt required returns by their respective proportions of the P-E capitalization, with the equity generally comprising over 90 percent, Dr. Grossman arrived at the 37-percent overall figure. We do not question the accuracy of this computation, but we do have reservations about*540 its applicability to the circumstances surrounding the P-E and PECC dealings. In light of our ultimate conclusion regarding the appropriate resale margin for instruments, see infra pp. 110-111, we find it unnecessary to resolve this point definitively. As described, Dr. Horst assumed a period of 50 days between PECC's shipment to P-E and P-E's payment to PECC, rather than the 77 days calculated by Arthur Andersen. See supra p. 42. Without disputing the Arthur Andersen calculation, Dr. Horst justified his use of 50 days as "a reasonable estimate of payment practices when parties are dealing at arm's length". Incorporating the 77-day period into Dr. Horst's calculations would result in an adjusted resale margin of 31.9 percent. Respondent asserts that this adjustment for credit terms should be limited either to the 50-day estimate of Dr. Horst or, better yet, to the contractual terms between P-E and PECC. Because the Distributor Agreement lacked a provision specifying payment terms, respondent derives the contractual terms from PECC invoices calling for P-E to pay within 30 days of the PECC shipment. In Eli Lilly & Co. v. Commissioner, 84 T.C. 996, 1171, 1179 (1985),*541 affd. in part, revd. in part, and remanded 856 F.2d 855 (7th Cir. 1988), we made an adjustment for contractual payment terms as part of our application of the comparable uncontrolled price method. We did not, however, refer to any extended payment practices, in fact noting that the taxpayer in practice used the full contractual payment period. Eli Lilly & Co. v. Commissioner, 84 T.C. at 1171. Even assuming that the PECC invoices set the contractual payment terms, P-E's actual payment practice with PECC strikes us as the appropriate benchmark for this adjustment. A focus on actual practice is consistent with the treatment of another point arising out of the distribution arrangement, the exclusivity of the P-E distribution rights. Although the Distributor Agreement did not state that P-E would be PECC's exclusive distributor, respondent, petitioner, and the experts have treated P-E as such for comparability purposes. As another example of deference to actual practices, respondent has not sought to apply a different legal analysis under section 482 for those many PECC finished products omitted from the 1970 licensing agreements. *542 See supra pp. 41-42. Further, we are not impressed with respondent's argument that section 936, relating to Puerto Rico and the possessions tax credit, caused P-E to prefer that late payments to PECC appear in adjustments to the transfer price rather than as a separate interest charge. Cf. Eli Lilly & Co. v. Commissioner, 84 T.C. at 1179. Petitioner suggests another downward adjustment to the P-E resale margin on Hitachi instruments, this time to account for P-E's short lead time for ordering from PECC as compared to a 6-month lead time in dealing with Hitachi. In this regard, petitioner's industry expert, Herbert L. Kahn, testified that a distributor would accept a reduction of two percentage points or more in its resale margin for the relative flexibility afforded by PECC. The sales agreement between P-E and Hitachi was vague about ordering, however, stating that Hitachi would sell and ship products subject to the terms and conditions of its standard sales order. The record does not contain such a sales order, and petitioner relies instead on a single transaction for which the P-E purchase order showed an order date of December 4, 1978, *543 and a "ship to arrive" date of June 30, 1979. Because we are not persuaded that Hitachi required a 6-month lead time, no adjustment therefor is warranted. We turn now to the purported differences between the fluorescence and AA product lines. Dr. Baumol characterized the fluorescence market as relatively new and small, requiring the P-E salesperson to sell both the technique and the P-E product and to seek correspondingly higher compensation from the manufacturer in the form of a higher discount. He contrasted this situation with PECC products that had supposedly captured a major share of the low-cost instrument market by 1975. However, Dr. Baumol expressly disclaimed expertise concerning the analytical instrument industry, and the record does not support his vague descriptions of the fluorescence market and the PECC market share. Nor does it necessarily follow that a relatively new and small market requires more selling effort, especially if the growth rate is substantial, as it apparently was in the fluorescence area. In this regard, petitioner's industry expert, Mr. Kahn, characterized fluorescence as "a very rapidly moving market". Interestingly, respondent makes an argument*544 similar to that of Dr. Baumol about P-E's relative market dominance in the AA and IR (infrared) areas as compared to fluorescence. Instead of support for a higher discount, however, respondent sees this as indicative of lesser competition in the AA and IR markets, enabling P-E to achieve better margins than in the more competitive fluorescence market. Like Dr. Baumol, however, respondent overlooks the pertinent point of comparison in emphasizing P-E's overall positions in the AA and IR markets, rather than its much less dominant position in the low-priced segment occupied by the PECC instruments. Finally, respondent on brief suggests several other circumstances that supposedly require upward adjustment to the P-E resale margin on the Hitachi instruments in order to create true comparability. Dr. Horst did not specifically quantify any of these upward adjustments, although he mentioned some, nor does respondent on brief. Although we cannot readily quantify the effect, nor say for certain that P-E availed itself of these benefits, we do agree with respondent that Hitachi agreed to provide P-E with promotional support and free replacement parts under warranty, neither of which PECC*545 provided to P-E. 2. Spectra-Physics and P-E Although the resale price method is not always suited to situations in which the reseller adds product value by means of intangible property, petitioner maintains that P-E added the same such value to the products of PECC, Spectra-Physics, and Analect, thereby negating any comparability problem. See sec. 1.482-2(e)(3)(ii)(d), (iii), Income Tax Regs. Dr. Horst took a similar position in support of the Hitachi transaction as a resale comparable. Like the PECC finished products, P-E sold Hitachi, Spectra-Physics, and Analect products under the P-E name and trademarks through the P-E marketing and sales organization. Petitioner's first preference as a comparable resale transaction is Spectra-Physics, whose integrators, according to Arthur Andersen, generated a 34.1-percent margin to P-E on resales to unrelated parties. See supra pp. 47-50. Dr. Baumol pointed out several differences from the PECC arrangement with P-E, all of which he saw as favorable to Spectra-Physics relative to PECC: (1) Spectra-Physics granted its maximum discount only if P-E reached a threshold quantity level; (2) P-E had to commit to an annual delivery *546 schedule; (3) P-E had to notify Spectra-Physics of scheduling changes 90 days before delivery; (4) P-E paid its invoices after only 50 days rather than the 77 days for PECC; (5) Spectra-Physics could compete with P-E on sales of the integrators; (6) the Spectra-Physics integrators entailed higher P-E net service costs; and (7) Spectra-Physics was not required to purchase parts from P-E, at a substantial markup, for inclusion in the finished products. Dr. Baumol believed that all of these differences, for comparability purposes, exerted downward pressure on the 34.1-percent Spectra-Physics margin, but he quantified only two. First, using Dr. Sanford's 37-percent estimate of the P-E weighted average cost of capital, Dr. Baumol concluded that the 27-day difference in payment practices translated into a reduction of 1.9 percentage points in the resale margin. 13 For Dr. Baumol's second quantified adjustment, he adopted the Arthur Andersen calculation concerning the P-E net service costs, which resulted in a further margin reduction of 1.1 percentage points. 14 Dr. Horst agreed with Dr. Baumol that an adjustment for differences in service costs was appropriate. After accounting for*547 these two factors, Dr. Baumol's adjusted resale margin for Spectra-Physics integrators stood at 31.1 percent. Substituting the P-E cost of short-term borrowing, 10.58 percent, into Dr. Baumol's calculation of the difference in payment practices results in an adjusted resale margin of 32.5 percent rather than 31.1 percent. Concerning one of Dr. Baumol's unquantified factors, the P-E status as a nonexclusive distributor for Spectra-Physics, both he and Dr. Horst recognized that P-E was in effect an exclusive distributor for PECC, but they disagreed about whether exclusivity tends to increase or decrease a resale margin. In Dr. Baumol's view, a distributor would accept a smaller discount*548 from the manufacturer if the manufacturer agreed not to sell the same product to others, the rationale being that the distributor, protected from direct competition, could make sales with less effort. The smaller discount under this theory leads to a smaller resale margin. In Dr. Horst's view, on the other hand, a lack of direct competition enables the distributor to sell at a higher price, thereby earning a higher resale margin. Mr. Kahn agreed with Dr. Baumol's point about reduced selling effort and Dr. Horst's point about more distributor price control, concluding that these were benefits to the distributor for which it would sacrifice approximately three percentage points of discount. In our view, however, the resale margin effect of an exclusive arrangement versus a nonexclusive arrangement, in the circumstances before us, is unsuited to generalizations like those advanced by the experts. Based on the P-E attempt to price the integrators higher than Spectra-Physics, which was unsuccessful because of the latter's competing presence, we recognize that P-E probably would have priced the integrators higher in an exclusive arrangement. This does not by itself, however, prove*549 Dr. Horst's point. Even he acknowledged as a general proposition that the presence of third-party competitors' products could limit pricing freedom. Because the low-priced market segments that included the PECC instruments were price sensitive and not dominated by P-E, we do not believe that the Spectra-Physics pricing implications transfer readily to the PECC instruments. Even if we assume that P-E had some measure of pricing freedom with the PECC instruments, Dr. Horst's position faces another obstacle. In attempting to reconcile his view with that of Dr. Baumol, Dr. Horst agreed that a manufacturer would probably seek and get a higher price from the distributor in an exclusive arrangement. He recognized that, if this were so, the price charged by the distributor to the user would have to increase even more (in terms of the percent increase) than the price charged by the manufacturer to the distributor, in order for a higher resale margin percentage to result. Thus, even if we accept that a distributor would charge a higher price in an exclusive arrangement, it is by no means certain that a higher resale margin percentage would result. We are also not persuaded that, as Dr. *550 Baumol suggests, a rational manufacturer would necessarily seek a smaller discount percentage off the distributor's price in a state of lessened competition from an exclusive arrangement. If the distributor set a higher price while the discount percentage stayed the same, the manufacturer would still reap a proportionate share of the increased price. For example, at an unchanging discount of 30 percent, 70 percent of the distributor's price increase would be additional revenue to the manufacturer. Considering the effects in the opposite direction, moving from exclusive to nonexclusive, Dr. Baumol testified that if PECC had sold some of its products directly to customers, then P-E, if at arm's length, would have demanded a higher discount to compensate for the more difficult selling task it faced. In actuality, however, P-E exerted relatively little effort in selling the PECC instruments, with salespeople often acting merely as passive order takers. Given that P-E did not try to exploit its exclusivity with aggressive selling, it is not as clear to us, as it was to Dr. Baumol, that a nonexclusive, arm's-length P-E would increase its selling efforts and thus need or demand more*551 compensation. In sum, we do not believe that the resale margin effect of exclusive versus nonexclusive distribution arrangements is as straightforward as suggested by the experts, nor do we believe that they have adequately accounted for the particular circumstances of this case. We therefore agree with Dr. Horst's more general proposition that the uncertainty surrounding this factor undermines the comparability of the Spectra-Physics transaction. Of the remaining four factors that Dr. Baumol cited as relatively favorable to Spectra-Physics, we see none that necessarily affects the P-E resale margin on PECC instruments. As to a required minimum order level to reach the maximum discount percentage, a stipulated exhibit indicates that P-E regularly exceeded that level, suggesting that the contracting parties contemplated high order levels that rendered the lower discounts largely meaningless. 15 The agreed annual delivery schedule, which the allowance for 90-day rescheduling largely mitigated, in context appears to have been merely a guideline, not a binding obligation that Spectra-Physics would have cared to enforce. Concerning the Spectra-Physics freedom to purchase parts from*552 other than P-E, our conclusion below that PECC did not pay more than an arm's-length amount for its parts negates this purported distinction. See infra p. 152. Finally, the access of the Norwalk Support Group to the sophisticated MRP system, coupled with P-E's provision of necessary parts to PECC, potentially offsets to a great extent what would otherwise be a risk disparity with Spectra-Physics, which had 90 days' notice of P-E scheduling requirements. Respondent argues that the Spectra-Physics transaction is not comparable, or at least its resale margin must be adjusted upward to become comparable, because the integrators exhibited price sensitivity, showed a clear obsolescence trend, and needed limited sales support. These circumstances, according to respondent, contrast with the price premiums commanded by P-E's quality AA and IR*553 products. However, although in general P-E maintained a high pricing policy and enjoyed a reputation for high quality, the pricing policy became somewhat muted for the relatively low-priced PECC instruments. Because of functional similarity among competitors, price was more important to users at the low-priced end than for higher priced instruments, and the P-E prices on PECC instruments were usually at least close to those of competitors. Admittedly, the Spectra-Physics integrators had obsolescence problems that adversely affected margins, but obsolescence was also not uncommon among the PECC instruments. Sales of an AA instrument model at P-E generally peaked after only 2 or 3 years, and PECC did not always begin production of an instrument model at its marketable inception. See supra pp. 31-34. Spectra-Physics did agree to provide some benefits to P-E that PECC did not provide, such as service training, but respondent's suggested differences in marketing support and in the P-E distribution activities seem insignificant, largely because PECC instruments themselves entailed little in the way of distribution activities. 3. Analect and P-E Petitioner's second choice*554 as a comparable transaction under the resale price method is the P-E resale of Analect FX-6200 FTIR instruments, which P-E sold in combination with its own data stations as the Models 1500 and 1550. See supra pp. 52-56. P-E purchased the instruments during fiscal 1982 through 1985 at a 30-percent discount from the Analect list price, and Arthur Andersen concluded that P-E earned a resale margin of 25.2 percent on the Model 1500 and 1550 systems. Petitioner contends that any necessary adjustments to arrive at true comparability would reduce the unadjusted resale margin and, accordingly, that the already low margin is corroborative evidence that the P-E resale margin on PECC instruments as calculated by Arthur Andersen, 31.5 percent, was arm's length. The only such adjustment quantified by Dr. Baumol, of three percentage points, was for the 42-day difference between P-E payments to PECC (77 days after invoice) and P-E payments to Analect (35 days). As with the Spectra-Physics integrators, P-E sold the FX-6200 instruments while the manufacturer continued to sell them itself. The record, however, is not clear about the actual degree of direct competition. On the one hand, Gaynor*555 Kelley of P-E testified that P-E faced some bothersome competition from Analect. On the other hand, Analect intended the integrated P-E system to enter a higher priced market segment, thus avoiding direct competition. Assuming that the Analect intentions played out in practice, thus creating in effect a largely exclusive arrangement like that between P-E and PECC, several other factors cumulatively cause us to conclude that this is not an appropriate comparable transaction. The Analect FX-6200 became part of P-E's IR product line, making this transaction the first we have discussed that involved a product line for which PECC produced instruments. However, the P-E list prices for the Models 1500 and 1550, beginning at $ 35,000, were well above the mainly under-$ 10,000 prices for the PECC instruments. This huge price discrepancy is indicative of the relative complexity of the instruments, and what follows from more complexity is usually more costly technical support provided by the distributor to the user. Petitioner's industry expert, Mr. Kahn, attempted to downplay the complexity difference, but his testimony on this point was vague and unpersuasive. Dr. Baumol did not mention*556 prices in his discussion of the Analect instruments in his original written report, but, elsewhere in that report, he acknowledged the relationship between complexity and technical support. At trial, Dr. Baumol stated generally that price is a legitimate comparability factor. More to the point, he admitted that the Analect price disparity "certainly casts very severe doubt on the comparability". Furthermore, at the time P-E entered into the agreement with Analect, management, which was well aware of the advantages of the nondispersive technology, had already authorized development of a P-E product. Lack of a nondispersive product had caused a loss of P-E market share in 1980 and had threatened its status as the world leader in the IR area. Management perceived purchasing the FX-6200 from Analect as an expeditious way to enter the market, which would provide a foothold for the later marketing of the P-E product under development. These considerations may have caused P-E to sell at a lower price than it otherwise would have, which would afford a degree of reconciliation of Gaynor Kelley's testimony and the Analect intentions regarding competition. Cf. sec. 1.482-2(e)(2)(iv), *557 Income Tax Regs. (relevant circumstance in comparable uncontrolled price method is low pricing for primary purpose of establishing or maintaining a market for a product). Analect also provided P-E with marketing assistance, service documentation and training, other technical support, and warranty coverage, none of which PECC provided to P-E. Dr. Baumol admitted at trial that the presence of technical assistance and warranty coverage would affect the conclusions he drew in his original report, but he did not attempt to estimate the effect. He also did not raise the Arthur Andersen net service figures as a possible offset. See supra p. 56. We are left with several possible adjustments moving in different directions, with little in the way of quantitative guidance. Finally, the Model 1500 and 1550 systems included P-E components, such as data stations and applications software, that comprised significant portions of the total value, thus possibly distorting the resale margin. Petitioner points out that certain PECC finished products were also packaged together with P-E products for resale, but petitioner has not given us a basis upon which we could conclude that this practice*558 was widespread, as it was with the Analect instruments. To overcome the possible margin-distorting effects of combined P-E and Analect components, Arthur Andersen computed a P-E resale margin of 17 percent that supposedly isolated the Analect-sourced products. Nonetheless, we need not address respondent's arguments concerning possible methodological errors in this calculation. Petitioner does not refer to this calculation on brief, and we deem it abandoned. 4. Coleman and Scientific Products Coleman sold various items to Scientific Products, usually at a 33-percent discount from Coleman list prices, for which Scientific Products recorded an average resale margin of 23.6 percent. See supra pp. 56-58. This margin lies well below the resale margin that Arthur Andersen concluded P-E earned on PECC instruments, 31.5 percent, which petitioner maintains was arm's length. Unlike the PECC relationship with P-E and the three potential comparable transactions already discussed, the distributor in this transaction, Scientific Products, resold the products under the manufacturer's name and trademarks rather than its own. The P-E name and trademarks were valuable marketing intangibles*559 that should be accounted for somehow under the resale price method as applied to the PECC instruments. See sec. 1.482-2(e)(3)(vi)(c), Income Tax Regs. We have no way on this record, however, to quantify the value of these intangibles. The uncertainty on this important matter alone greatly undermines any purported comparability. Cf. sec. 1.482-2(e)(2)(ii), Example (2), Income Tax Regs. (difference in trademarks under comparable uncontrolled price method normally renders uncontrolled sale noncomparable). Moreover, although P-E and Scientific Products had similar distribution organizations structurally, and Scientific Products performed similar distribution functions to those of P-E, we are not satisfied that Scientific Products performed them as well or to the same extent. Indeed, Dr. Baumol suggested two factors relating to reseller functions as exerting upward pressure on the Scientific Products resale margin for comparability purposes. See sec. 1.482-2(e)(3)(vi)(b), Income Tax Regs. First, the relative simplicity of the Coleman instruments compared to the PECC instruments meant less costly technical support. Second, Coleman bore a portion of the warranty *560 costs for the instruments it sold to Scientific Products. Other Coleman assistance that relieved Scientific Products of some measure of distribution functions included: (1) Training and seminars for the instrument users; (2) technical training and ongoing assistance for Scientific Products personnel; and (3) detailed product and marketing information. Understandably, the resale price method regulations strongly encourage the use of comparable transactions that involve the same reseller as in the related-party transaction. Sec. 1.482-2(e)(3)(vii), Income Tax Regs. Petitioner here, however, does not suggest how we might attempt to quantify necessary adjustments to the Scientific Products resale margin. With such notable differences to account for, and no discernible way to account for them, we attach minimal significance to this transaction. 5. Conclusion Neither party has provided a proposed instrument comparable sufficiently similar to the P-E resale of PECC instruments, in and of itself, to satisfy the resale price method in section 1.482-2(e)(3), Income Tax Regs. Consequently, we must undertake an evaluation of the arm's-length price for PECC instruments sold to P-E *561 based on the entire record in respect of the various transactions offered as comparables by the parties. We continue to focus on what the regulations call an appropriate markup percentage and what we have often referred to as a resale margin. See sec. 1.482-2(e)(3)(i), (vi), Income Tax Regs. Respondent has not specifically contested the accuracy or reliability of the Arthur Andersen determination that P-E earned a resale margin of 31.5 percent on PECC instruments sold to unrelated parties, so we accept that figure as our reference point. See supra p. 46. Accordingly, the question before us is whether 31.5 percent was an arm's-length resale margin. With primary reliance on the Hitachi transaction, which was the least controversial proposed comparable, and somewhat lesser reliance on the Spectra-Physics transaction, we are satisfied that the P-E resale margin of 31.5 percent on PECC instruments was not lower than arm's length. After considering the more readily quantifiable adjustments surrounding the Hitachi and Spectra-Physics transactions, we were left with adjusted resale margins of 31.9 percent and 32.5 percent, respectively, even after incorporating respondent's preference*562 for the short-term borrowing rate in the time-value-of-money aspects. See supra pp. 94, 99. We do not mean to suggest that further adjustments to the Hitachi and Spectra-Physics margins would not have been appropriate had we been furnished with more complete information or more easily quantified factors. However, on the record before us, and in light of the parties' broad approach in terms of product groupings and combined taxable years, we are satisfied that the prices charged for instruments by PECC to P-E fall within the range of permissible levels under section 482. Cf. U.S. Steel Corp. v. Commissioner, 617 F.2d 942, 950 (2d Cir. 1980), revg. T.C. Memo. 1977-140 and T.C. Memo. 1977-290; Bausch & Lomb, Inc. v. Commissioner, 92 T.C. 525, 592-593 (1989), affd. 933 F.2d 1084 (2d Cir. 1991). Accordingly, we hold that respondent may not allocate income to P-E based on the prices it paid PECC for instruments, except for the stipulated allocation to account for petitioner's concession that a 30-percent discount should have applied in all*563 years. See supra note 7. D. HC Lamps Petitioner advanced the Spectra-Physics, Analect, and Scientific Products transactions as comparables under the resale price method for PECC finished products as a whole. Because these three transactions have a closer connection to instruments than to HC lamps or accessories, and because we did not find them comparable even for instruments, we will not consider them again in our analysis of the noninstrument PECC products. Respondent contends that the P-E resale of Sylvania deuterium lamps, after appropriate downward adjustment to the resale margin, was comparable to the P-E resale of HC lamps produced by PECC. See supra pp. 61-64. Under respondent's primary trial position, $ 9.0 million of the total income allocation to P-E of $ 22.2 million is attributable to HC lamps. Petitioner disputes the comparability of the deuterium lamps on the ground that identifiable differences caused P-E systematically to earn higher resale margins on deuterium lamps than on HC lamps. 16 *564 According to Dr. Horst's calculations, P-E earned a margin of 63.1 percent on its resale of deuterium lamps, a figure that petitioner does not contest. Much like he did for the P-E resale of Hitachi fluorescence instruments, Dr. Horst made a financing adjustment to account for the Arthur Andersen conclusions that P-E held deuterium lamps in inventory for 113 days and HC lamps for only 33 days. See supra p. 46. Applying the P-E cost of short-term borrowing of 10.58 percent, he reduced the deuterium lamp margin by 2.3 percentage points. He further reduced the margin, by 3.6 percentage points, to account for the volume discounts customers could get for purchasing several HC lamps at a time. His adjusted resale margin for deuterium lamps, which respondent argues should apply to the HC lamps produced by PECC, was 57.2 percent. The parties argue about the comparability of deuterium lamps in three principal areas. One is the technical and functional characteristics of each type of lamp. A second area is the nature and extent of the functions performed by the distributor, P-E. Finally, the parties contest the "captive" nature, or lack thereof, of HC lamps and deuterium lamps*565 as they relate to instruments. Dr. Horst raised this last topic in his initial written report, describing a captive product (e.g., razor blades) as one that a customer tends to buy, largely because of the value of a company's name or goodwill, from the same company that sold the customer the principal product (e.g., a razor). This tendency, as described by Dr. Horst, frequently leads to high markups on the captive product. Dr. Baumol agreed with Dr. Horst's general description of the principle. 17 Under the resale price method, one of the important characteristics in determining which resales are most similar for comparability purposes is the type of property involved. Sec. *566 1.482-2(e)(3)(vi)(a), Income Tax Regs. On the subject of technical and functional characteristics, respondent's industry expert, Dr. Nelson L. Alpert, observed in his initial report that both types of lamps are radiation sources of limited useful life in instruments, that they are of comparable complexity, and that P-E sold both in moderately high quantities. There is no question that both lamps were radiation sources of limited useful life that P-E resold at similar prices. Further, petitioner's industry expert, Mr. Kahn, who responded in his rebuttal report to Dr. Alpert, did not rebut Dr. Alpert's statement that the lamps were of similar technical complexity, instead merely stating that HC lamps for various elements differed in technical respects from one another. Nonetheless, it is beyond dispute that the specific functions of the two lamps differed. Neither party suggests that a deuterium lamp could substitute for an HC lamp in an AA spectrophotometer, nor that an HC lamp could substitute for a deuterium lamp in an AA background corrector or in a UV-VIS spectrophotometer. In addition, a notable difference in use was that an instrument user might temporarily remove and*567 reinstall various functioning HC lamps in order to measure different elements, yet a deuterium lamp would only be removed upon failure after some necessary disassembly of the instrument. Section 1.482-2(e)(3)(vi)(a), Income Tax Regs., provides examples of acceptable product categories for comparability purposes, listing specifically machine tools, men's furnishings, and small household appliances. From a technical, functional, and use standpoint, HC lamps and deuterium lamps appear at least as closely related as common items within these categories. Because petitioner has not described the probable effect upon the markup percentage of the differences in technical specifications, functions, and use characteristics, we are not prepared to disqualify deuterium lamps as a comparable based on these differences alone. The second broad point of contention between the parties is the nature and extent of the functions performed by P-E in distributing the two types of lamps. See sec. 1.482-2(e)(3)(vi)(b), Income Tax Regs. As we view the evidence, the P-E distribution activities were not substantial for either lamp, and we see no distribution factor that we would expect to have*568 a material effect on a resale margin, with the possible exception of warranty costs. P-E bore the entire warranty expense on HC lamps; the deuterium lamps had more limited warranty exposure because of the Sylvania 500-hour warranty to P-E and because P-E did not give its own warranty to customers. HC lamps required little P-E effort and expense in the areas of installation, customer training, and servicing. For deuterium lamps, there was no such thing as initial installation for the replacement market, nor is there evidence that P-E provided any customer training to users already accustomed to the instruments and accessories that contained those lamps. P-E did perform service installation for some portion of the replacement deuterium lamps, but there was a separate charge for this to the customer. In the marketing and sales area, advertising was not a factor for HC lamps most of the time, and the P-E salespeople acted for the most part as mere order takers. The deuterium lamps also did not involve marketing and sales efforts by P-E personnel. P-E held deuterium lamps in inventory over three times as long as it held HC lamps, and we agree with Dr. Baumol that, other things being*569 equal, a distributor would require a higher resale margin as compensation for this extra burden. However, Dr. Baumol did not estimate the quantitative effect of this factor, and it is therefore unclear to what extent he believed that Dr. Horst's financing adjustment based on inventory holding periods failed to address this point. See supra p. 112. Another important characteristic in determining which resales are most similar for comparability purposes is the effect on price of any intangible property used by the reseller. Sec. 1.482-2(e)(3)(vi)(c), Income Tax Regs. Respondent, who considers the captive product phenomenon a marketing intangible in this regard, argues that HC lamps and deuterium lamps were equally captive in relation to their associated P-E instruments. Petitioner counters that deuterium lamps were captive, but that HC lamps were much less so, if at all. As sometimes noted in the P-E SP2 reports, a customer tended to purchase HC lamps from its instrument supplier. An Instrument Lab employee testified about Instrument Lab customers to the same effect. More generally, a P-E market study conducted in 1981 confirmed the existence of this tendency for lamps*570 as a whole, that is, not just P-E lamps and not just HC lamps. Respondent, while recognizing that P-E HC lamps were not captive products in the strict sense of one-of-a-kind replacement parts, offers several explanations for their generally captive nature in order to give us a sense of the similarity to deuterium lamps. It does makes sense, as respondent suggests, that convenience and compatibility concerns would push a customer toward purchasing HC lamps from the instrument supplier at the time of the instrument purchase, especially in view of the relatively low HC lamp prices compared to AA spectrophotometer prices. Convenience and compatibility, however, go only so far. The convenience rationale at the time of the instrument purchase would not apply to the same extent for replacement lamps purchased later, and the presence of inexpensive and widely available lamp adaptors would seemingly allay compatibility concerns. P-E itself apparently did not try to exploit these concerns. P-E included lamp adaptors in its published price lists, and, when customers used competitors' HC lamps, P-E did not void its instrument warranty. In terms of the competitive environment, respondent*571 argues that customer loyalty to the instrument supplier derived in part from limited alternative sources for HC lamps. We are confident from this record, however, that someone seeking an HC lamp had several alternative suppliers from which to choose, including other instrument manufacturers, and that a lack of suppliers was not a major contributing factor to customer loyalty. Moreover, someone seeking an HC lamp sometimes had another lamp option. P-E itself successfully sold ED lamps that were superior to the corresponding HC lamps for certain elements and applications. We also disagree with respondent that P-E cultivated strong customer loyalty through promotion of a superior product. We read the P-E promotion language for HC lamps as standard product puffing; it differs little from how a supply house catalog described WestinghouseHC lamps. See supra pp. 24-25. As to actual quality, many Intensitron HC lamps suffered quality problems from unwanted hydrogen for much of the period in issue. Further, the SP2 report prepared in 1977, without regard to the hydrogen problem, described those lamps as no more than "at least the equivalent" of competitive products. The same *572 report immodestly described the P-E ED lamps as "definitely superior". In any event, superior P-E quality would not explain why the captive tendencies of HC lamps apparently extended across the industry; clearly, not everyone was offering superior quality. In addition to suggesting reasons for the captive nature of HC lamps, respondent rationalizes the higher pricing assumed to follow from a captive product by arguing that price was relatively unimportant to consumables purchasers when compared with factors such as quality and delivery time. Although price may have been subordinate to other considerations for consumables in general, customers did not consider price unimportant and, further, consumables include products other than HC lamps. See supra p. 8 and note 3. Moreover, for HC lamps in particular, P-E management noted in the SP2 reports that customer preferences shifted to alternative sources if pricing varied too much between suppliers. Management demonstrated its recognition of customer price consciousness by offering published volume discounts on HC lamps. Petitioner seeks to distance deuterium lamps from HC lamps by arguing that the former were much more captive*573 than the latter. Unlike HC lamps, which were often purchased separately either with AA instruments or as later add-ons, customers purchased deuterium lamps only as replacements for failed lamps that originated as integral parts of AA background correctors or UV-VIS spectrophotometers. Although replacement parts have something of a captive aura about them to begin with, petitioner emphasizes the circumstances surrounding replacement, with reliance on its industry expert, Mr. Kahn. In his rebuttal report, Mr. Kahn concluded that "as a practical matter, the overwhelming majority of customers have no choice but to purchase necessary replacement * * * [deuterium] lamps from the instrument manufacturer and to have the instrument manufacturer's service people install them", a situation ripe for high markups. The major premises leading up to this conclusion were that, first, the complexity and danger of installing the lamps discouraged all but a very few users from doing it themselves, and second, deuterium lamp manufacturers almost never provided installation services. Mr. Kahn added in his trial testimony that a user typically would not even know that a deuterium lamp had failed until*574 so informed by a service person diagnosing an unidentified problem. Mr. Kahn's view that instrument users rarely installed their own replacement deuterium lamps strikes us as an exaggeration lacking independent support in the record. Even assuming that the reluctance of Sylvania to sell to distributors extended to other lamp manufacturers, this would not preclude users from purchasing the lamps from instrument manufacturers, which was the Sylvania target market, for self-installation or for service company installation. Indeed, this was apparently the case at P-E in the AA product line. In analyzing deuterium lamps from the P-E records, Arthur Andersen determined that three-fourths of the lamps that did not become components in the production of instruments and accessories went directly to users rather than to the service organization. See supra pp. 63-64. P-E also promoted at least one spectrophotometer model as having easy access to the deuterium lamp in the background corrector, and P-E's AA price lists, which recommended service installation for only some deuterium lamps and assemblies, required service installation for none. From a competition standpoint, petitioner*575 claims that an absence of market competition characterized deuterium lamps, thus enabling P-E to earn an abnormally high resale margin. The deuterium lamp manufacturers, says petitioner, declined to compete with instrument manufacturers because of product liability exposure, the small size of the market, and the high cost of maintaining a service organization. Given the inadequate picture of the competitive setting in the record before us, we find none of these points persuasive. We further note that petitioner has not suggested that only P-E deuterium lamps were compatible with P-E instruments. Although Sylvania was indeed concerned about its product liability exposure, this feeling was apparently not universal. P-E, for example, sold some models of deuterium lamps without even recommending service installation. As to the size of the market, it was certainly small compared to the HC lamp market but not small in an absolute sense. In fact, P-E at one time considered engineering its own deuterium lamps, in part because they sold "in tremendous quantities". Concerning the purported high cost of a service organization, we are not convinced that a service organization was necessary. *576 Although some of petitioner's arguments have not impressed us, we are nonetheless unwilling to conclude, as respondent would have us do, that the P-E resale of deuterium lamps was comparable to the P-E resale of HC lamps purchased from PECC. Both lamps exhibited captive tendencies, but the differing and uncertain circumstances preclude us from making quantifiable adjustments to the P-E deuterium lamp resale margin. Differences suggesting that the arm's-length resale margin for deuterium lamps should have been higher than the arm's-length HC lamp margin include: (1) Customers bought HC lamps at the time of the instrument purchase and later both to expand the elements analyzed and to replace failed lamps, while deuterium lamps as sold by P-E were only replacement products; (2) the HC lamp market had many competitors, while competition in the deuterium lamp market is unclear from this record; (3) service people did not install or replace HC lamps, while they at least sometimes replaced deuterium lamps; and (4) a typical instrument user purchased HC lamps much more frequently, and at a higher aggregate cost, than deuterium lamps. We therefore do not adopt respondent's recommendation*577 of a 57.2-percent resale margin. Instead, we turn to the entire record to determine the arm's-length price for HC lamps sold to P-E by PECC. Regardless of the specific reasons, which remain somewhat elusive to us, customers had a definite inclination to purchase HC lamps from their instrument suppliers, who we believe would reasonably seek to exploit the tendency through higher resale margins than they earned on the instruments. Consequently, the starting point for the HC lamp resale margin is 31.5 percent, which we have accepted as an arm's-length margin for the P-E resale of PECC instruments. See supra pp. 110-111. Frank Boes, a division general manager for Instrument Lab during the years in issue, testified that for the HC lamps Instrument Lab purchased from Westinghouse prior to the agreement with Juniper, see supra p. 69, Instrument Lab earned a margin on resale "in the range of 35 or 40 percent". Respondent elicited this testimony on cross-examination, and it appears to be the foundation of petitioner's argument on brief that dealings between Westinghouse and Instrument Lab offer an appropriate resale margin for HC lamps. However, petitioner's counsel did not *578 ask the witness to elaborate on the circumstances surrounding this arrangement, nor was he asked for a more definitive percentage. Thus, we are left with no explanation of how the circumstances of this arrangement differed from those of the PECC and P-E dealings. We assign the adverse consequences of this gap in the record to petitioner, whose inexactitude is of its own making. See Sundstrand Corp. v. Commissioner, 96 T.C. 226, 375 (1991); Eli Lilly & Co. v. Commissioner, 84 T.C. 996, 1148, 1167, 1180 (1985), affd. in part, revd. in part, and remanded 856 F.2d 855 (7th Cir. 1988); Cohan v. Commissioner, 39 F.2d 540, 543-544 (2d Cir. 1930). Under the foregoing circumstances, we are not prepared to adopt a resale margin below 40 percent. At the same time, we find no basis in the record for adopting a figure in excess of 40 percent but less than the 57.2 percent advocated by respondent, which we have rejected. In short, using our best judgment on the record before us, we conclude that 40 percent constitutes an arm's-length resale margin for HC lamps. The starting*579 point for the corresponding income allocation to P-E is the PECC net sales of HC lamps to P-E, after accounting for petitioner's concession regarding a constant 30-percent discount. See supra note 7. The arm's-length price for these sales is equal to the applicable resale price reduced by the appropriate resale margin. Sec. 1.482-2(e)(3)(i), (iv), (vi), Income Tax Regs. The applicable resale price shall be calculated by the parties for each year by increasing the PECC net sales base by the actual P-E resale margin for that year on sales to unrelated parties, as determined by Arthur Andersen. See supra pp. 45-46. This applicable resale price, which will represent total P-E resales to unrelated and related parties, shall then be reduced by the arm's-length resale margin determined by the Court, 40 percent. To the extent that the actual PECC net sales exceed the arm's-length price, gross income is allocated to P-E. E. Accessories Respondent proposes an income allocation to P-E of $ 7.8 million to account for the arm's-length pricing of accessories between PECC and P-E. See supra pp. 59-61. Respondent bases this allocation on a purported comparable resale margin*580 of 57.6 percent, which Dr. Horst derived largely from the P-E 1980 and 1981 computerized sales database. In his original report, Dr. Horst had calculated a lower comparable resale margin. As the most comparable resale transactions under the resale price method, he had targeted the P-E sales of all accessories purchased from unrelated suppliers in each of the three product lines covered by PECC accessories. He had found the gross margins for those sales, by product line and by year, in the P-E accounting records under categories called "Resale Hitachi" and "Resale Other". After weighting the resale margins by product line and by year based on PECC standard costs, he had arrived at a weighted average resale margin of 53.2 percent applicable to all PECC-assembled accessories. In his subsequent supplemental report, Dr. Horst adopted the recommendations of petitioner's accounting experts that the 1980 and 1981 sales database was preferable to the accounting records he had used originally. He also agreed to factor out those products that P-E resold at unit prices not exceeding $ 200. Weighting the percentages from the table in part V.A.6. of our findings, see supra p. 60, by *581 the P-E resales of PECC accessories to unrelated parties in each product line, he computed separate resale margins for 1980 and 1981, made adjustments to account for 1975 through 1979, and ultimately arrived at his proposed comparable resale margin of 57.6 percent for the entire period in issue. Petitioner's principal objection to the Horst methodology is his failure to factor out the substantial effect of the graphite tubes. Because the graphite tubes comprised over 85 percent of the over-$ 200 net sales for AA accessories resold by P-E in 1980 and 1981, and because Dr. Horst's weighting factor favored the AA product line, the graphite tubes went a long way toward determining Dr. Horst's 57.6-percent resale margin. Indeed, applying Dr. Horst's methodology without the graphite tubes results in a 44-percent resale margin. Petitioner's rationale for excluding the purportedly captive graphite tubes is that, with a price per tube below $ 25, this was a relatively inexpensive, consumable item that was similar to many items within the low-priced ($ 200 or less) accessory category that Dr. Horst voluntarily disregarded in reaching his 57.6-percent margin. P-E never sold graphite tubes*582 singly, however, leading respondent to argue that a 50-count box is a legitimate over-$ 200 product. As support for inclusion of the graphite tubes and for the overall comparability of the Resale Other and Resale Hitachi categories, respondent cites section 1.482-2(e)(1)(iv), Income Tax Regs. This provision permits the application of an appropriate arm's-length pricing method to product lines or other groupings, including those derived from reasonable statistical sampling techniques, if it is impractical to apply the method to each product. Because PECC assembled accessories with 30 different part numbers in 1980 alone, some sort of grouping method presumably cannot be avoided. Respondent maintains that Dr. Horst applied a reasonable statistical sampling technique and that excluding graphite tubes is inconsistent with the statistical principle that the "law of numbers" eliminates distortions caused by specific items. Further, argues respondent, venturing into an evaluation of one product in the group undermines the very reason the regulations permit grouping, which is the impracticality of examining every product and sale. We disagree. Dr. Horst merely adopted categories established*583 by the P-E accounting records and then eliminated those products priced at $ 200 or less. Respondent does not attempt to describe the specific factors that made the Resale Other and Resale Hitachi transactions comparable to the P-E resale of PECC accessories. Instead, respondent states broadly that these "Resale" transactions involved the P-E resale of accessories in the same three product lines as the accessories produced by PECC. We do not view these circumstances as reflecting the application of a statistical sampling method. Admittedly, a so-called law of numbers, which we interpret as grounded in a large sample size, may tend to even out distortions among items when no one item is weighted disproportionately. We fail to see, however, how any such principle could operate effectively in a weighted-average situation where one item comprises 85 percent of the total in the heaviest of three weighted categories. Moreover, respondent's argument that delving into the details of the sample undermines the grouping rationale is out of place. Dr. Horst himself willingly eliminated from the sample those products selling for under $ 200, even though PECC sold some products in that price*584 range to P-E. Respondent does not contend that PECC itself manufactured graphite tubes, nor that graphite tubes were comparable to any specific accessory that PECC produced, nor that PECC packaged any of its products for resale in similar bulk quantities. Given these circumstances and the extreme influence that the graphite tubes had on the derivation of Dr. Horst's 57.6-percent margin, we do not believe that the graphite tubes are an appropriate element in the P-E accessory grouping. After elimination of the graphite tubes, we still cannot conclude that Dr. Horst's sample is comparable to the P-E resale of PECC accessories. In this regard, section 1.482-2(e)(3)(vi) and (ix), Income Tax Regs., requires consideration of factors such as reseller functions and product types, with adjustments made for material differences between the controlled and uncontrolled transactions. Certainly, section 1.482-2(e)(1)(iv), Income Tax Regs., which authorizes product groupings, implicitly relaxes some of the normally strict constraints on comparability. Nonetheless, we do not believe that respondent's vague comparability rationale here passes muster. From the record as it stands, however, we*585 do think that a captive product premium, which Dr. Horst took into account, applies to the PECC accessories resold by P-E. Generally, because of concerns with quality and compatibility, for example, an instrument customer would more likely than not purchase an accessory from the instrument manufacturer without great concern about price. P-E encouraged the tendency by often marketing instruments and accessories together. As with HC lamps, however, P-E's potential pricing flexibility had its limits. The SP2 reports disclosed that customers avoided some AA accessories because of high prices and that others lost their competitive edges technologically in the mid-1970s. P-E also sometimes offered package deals and other customer incentives. We know little specifically about the facts and circumstances surrounding the resale of PECC accessories. Respondent mentions the ED lamp power supply, PECC's largest volume accessory, as one that faced no competition and thus had a high degree of captivity. 18 However, although this product probably faced little competition (because of the strict ED lamp warranty, see supra p. 11) in the minds of P-E's ED lamp owners, P-E was not the only*586 ED lamp seller. Because P-E's ED lamps were directly interchangeable with Intensitron HC lamps, P-E instrument customers presumably could buy their ED lamps and power supplies elsewhere, just as they could their HC lamps. In the circumstances confronting us, our best judgment leads us back to the 1980 and 1981 sales database used by Dr. Horst, which involved a wide variety of vendors. For adjustments, we will ignore the products with unit prices not exceeding $ 200 because both parties accepted this restriction, and we will disregard the graphite tubes. In addition, we will not weight the product line margins by the PECC accessory output, in part because elimination*587 of the graphite tubes makes the AA accessory figures from the sales database much smaller, and associated margins potentially less reliable, than the other two product lines. Combining the sales and cost of sales for all three product lines in the 1980 and 1981 sales database, before computing the resale margin, results in a resale margin of approximately 40 percent. Again using our best judgment, we hold that 40 percent is the arm's-length margin for the P-E resale of the accessories that PECC produced during the years in issue. The starting point for the corresponding income allocation to P-E is the PECC net sales of accessories to P-E, after accounting for petitioner's concession regarding a constant 30-percent discount. See supra note 7. The applicable resale price shall be calculated by the parties for each year by increasing the PECC net sales base by the actual P-E resale margin for that year on sales to unrelated parties, as determined by Arthur Andersen. See supra pp. 45-46. This applicable resale price, which will represent total P-E resales to unrelated and related parties, shall then be reduced by the arm's-length resale margin determined by the Court, 40*588 percent. To the extent that the actual PECC net sales exceed the arm's-length price, gross income is allocated to P-E. V. P-E Sales of Parts For the parts it sold to PECC, P-E charged its standard cost plus 21 percent through 1978 and standard cost plus 13.12 percent thereafter. Of the total income allocation to P-E of $ 22.2 million under respondent's primary trial position, $ 6.2 million is attributable to a parts adjustment based on markups over standard cost of 40.2 percent for instrument parts, 50.2 percent for HC lamp parts, and 28.5 percent for accessory parts. The parties' point of divergence on how to calculate an arm's-length price for the parts is whether the parts are properly analyzed as individual parts or as parts kits. Respondent takes the position that a parts kit is a "single economic unit" in the form of an unassembled and untested finished product rather than a random collection of unrelated parts. A parts kit is not susceptible to analysis under the comparable uncontrolled price method, according to respondent, because the only uncontrolled sale of parts kits in the record was from Coleman to Sao Paulo, and those spectrophotometers were less complicated*589 than the instruments produced by PECC. See supra pp. 64-66. Nor does the resale price method apply, says respondent, leaving the next method in priority under the regulations, the cost plus method. Petitioner, in contrast, views a parts kit not as a single economic unit but as a collection of individual parts, each of which fits neatly into a generalized view of the comparable uncontrolled price method. A. Respondent's Position Under the cost plus method, the arm's-length price of a controlled sale is the cost of producing the property as increased by the appropriate gross profit percentage, plus or minus adjustments. Sec. 1.482-2(e)(4)(i), Income Tax Regs. The appropriate gross profit percentage is the gross profit percentage relative to costs earned by a seller on the uncontrolled sales of property that are most similar to the controlled sale. Sec. 1.482-2(e)(4)(iii), Income Tax Regs. Characteristics important to the similarity of the sales include the functions performed by the seller, examples being contract manufacturing, product assembly, selling activity, processing, servicing, and delivering. Sec. 1.482-2(e)(4)(iii)(b), Income Tax Regs. When the most similar*590 sales used to derive the appropriate gross profit percentage differ in any material respect from the controlled sale, the arm's-length price must be adjusted to reflect the differences. Sec. 1.482-2(e)(4)(v), Income Tax Regs. The differences must have a definite and reasonably ascertainable effect on price. Id. Respondent's recommended approach to the pricing of parts kits, which follows Dr. Horst's analysis, is novel and complex. According to respondent, the appropriate P-E markup on a parts kit is the corresponding markup that P-E achieved on sales of the associated finished product when produced in Norwalk. The foundation of this position is the purported comparability of a parts kit and the finished product into which the parts kit was assembled. A parts kit and a finished product were obviously much different in terms of both physical appearance and functional readiness, but respondent believes that a parts kit embodied the same manufacturing intangibles, such as product design and manufacturing know-how, as the finished product itself. 19 *591 Although we here describe only the Horst methodology for pricing parts that went into instruments, the methodology was similar for accessories and HC lamps. Dr. Horst's first step was to calculate a representative gross margin for select Norwalk instruments. Because P-E usually stopped manufacturing an instrument after its transfer to PECC for production, he and Dr. Alpert, respondent's industry expert, selected from the instruments that P-E continued to manufacture in Norwalk the ones judged most closely comparable to the instruments produced by PECC. Dr. Horst then computed the gross margins that P-E earned on the Norwalk instruments, grouped the instruments by product line and year, and ultimately arrived at a weighted average gross margin of 52.0 percent. In keeping with the nature of these initial steps, he described this as the Comparable Norwalk Products method. Dr. Horst treated the gross margin of an instrument manufactured in Norwalk as a summed combination of a distribution or resale margin and a so-called manufacturing margin. To isolate the manufacturing margin, he subtracted his previously determined resale margin of 32.7 percent (the Hitachi fluorescence comparable, *592 see supra pp. 92-93) from the total gross margin of 52.0 percent, leaving a manufacturing margin of 19.3 percent. Because the cost plus method involves a markup over cost, he converted the manufacturing margin as a percentage of sales to a 40.2-percent markup over manufacturing costs. 20 Finally, he applied the 40.2-percent manufacturing markup applicable to Norwalk instruments to the P-E standard cost of the instrument parts kits it sold to PECC. Stated summarily, Dr. Horst attributed the P-E manufacturing markup on a finished product manufactured in Norwalk to the parts that P-E sold to PECC for the production of a similar product in Mayaguez. If the PECC production costs were the same as the P-E production costs for the similar product, this*593 method in effect also attributed the P-E manufacturing markup to the PECC labor and overhead, a situation that causes petitioner to characterize this as an impermissible formulary apportionment of total manufacturing profit based on respective P-E and PECC direct costs. If the PECC production costs exceeded those of P-E, however, then the PECC profit on its labor and overhead would be a smaller percentage than the P-E manufacturing markup applicable to the finished products and the parts. Similarly, if the PECC production costs were less than those of P-E, then the PECC profit on its labor and overhead would be a larger percentage than the P-E manufacturing markup applicable to the finished products and parts. Dr. Horst advanced an alternative approach to parts kit pricing denominated the Proportionate Profits method, which is the foundation of respondent's secondary trial position. For certain models of AA spectrophotometers built at both Norwalk and PECC during the same year, or built at Norwalk one year and at PECC the next, Dr. Horst determined that the total gross margins were approximately the same for the two locations. 21 This led him to conclude that the P-E and PECC*594 consolidated gross margin on a PECC-produced instrument was a reliable estimate of what the P-E gross margin would have been had it continued to manufacture the instrument in Norwalk. To implement the Proportionate Profits analysis through a simple procedure, Dr. Horst began with the consolidated gross margins of the instruments actually produced by PECC and in effect split the consolidated manufacturing profit in direct proportion to the respective P-E and PECC manufacturing costs, with the PECC costs including only labor and overhead. For example, P-E in 1980 made resales of PECC instruments totaling $ 11,367,863, according to Dr. Horst. He derived this amount by grossing up PECC's actual sales to P-E, $ 7,570,997, to account for the 33.4-percent resale margin he drew from the 1980 sales database. The P-E standard cost for the instrument*595 parts it sold to PECC, including all purchased material, was $ 3,738,902, and the PECC standard cost for its labor and overhead was $ 445,965, for a cost total of $ 4,184,867. Dr. Horst calculated the consolidated gross margin as the difference between resales and cost, $ 7,182,996, divided by resales, $ 11,367,863, or 63.2 percent. This 63.2 percent became the total gross margin that corresponded to the 52.0 percent used in the Comparable Norwalk Products method. From this point on, beginning with the subtraction of the 32.7-percent resale margin based on the Hitachi proposed comparable, the computational steps in the two methods were the same. Unlike the Comparable Norwalk Products approach, in which the derived manufacturing markup on costs was constant from year to year in each finished product category, these percentages varied in the Proportionate Profits method. In addition, for the 7-year period overall, the percentages were higher under the Proportionate Profits method: 67.3 percent for instruments, 78.2 percent for HC lamps, and 46.4 percent for accessories. Compare supra p. 130. This method leads respondent to a P-E income allocation for all PECC finished product*596 parts of $ 13.3 million, more than double the $ 6.2 million resulting from the Comparable Norwalk Products analysis. Despite the higher Proportionate Profits income allocation, respondent treats it as a secondary trial position on the rationale that it effects a profit split that is subordinate under the section 482 regulations to the Comparable Norwalk Products cost plus approach. See sec. 1.482-2(e)(1)(iii), Income Tax Regs. A key underpinning of respondent's parts analyses, as earlier noted, is that a parts kit was a single economic unit, in the form of an unassembled and untested finished product, rather than a random collection of unrelated parts. From a theoretical perspective, we cannot fault respondent for attempting to incorporate the "coherent grouping" of a parts kit into an analysis of the appropriate arm's-length price for parts. After all, except for occasional backordered or replacement parts, the parts kit system was how P-E and PECC communicated. With their distinctive part number prefixes, the parts kits streamlined both the PECC ordering and the P-E billing. Concerning another of respondent's premises, we do not disagree that a parts kit, when viewed as such, *597 embodied what respondent has characterized as manufacturing intangibles. During the product development process, in designing and bringing to production a particular instrument, P-E did not start with a box of random parts and seek to build the best possible product from those parts. Instead, P-E sought to design an instrument with desired characteristics or specifications, and only after completion of the finished product did P-E know and fully document the composition and specifications of the necessary parts. When viewed this way, the parts were an output of the product development process rather than an input. Because PECC received the parts in the form of a parts kit that matched exactly the composition and specifications of the parts needed to produce the finished product, PECC indeed benefited from the so-called manufacturing intangibles of the finished product. Despite our acceptance of some of Dr. Horst's important conceptual and theoretical premises, we nonetheless decline to adopt either of his approaches to the arm's-length pricing of parts. Because no outside market existed for the parts kits as such, determining the appropriate markup under the Comparable Norwalk*598 Products cost plus analysis involves practical difficulties and unsupported assumptions that we cannot ignore. The Proportionate Profits method generally has the same shortcomings. As Dr. Horst openly acknowledged at trial, he backed into the appropriate percentage markup on parts (i.e., his manufacturing markup) rather than deriving that markup directly from a comparable transaction. The percentage he did compute directly under the Comparable Norwalk Products approach was the total gross margin, including a distribution or resale margin, on the P-E sale of a Norwalk finished product. See supra p. 133. A concern we have with this indirect approach is its extreme sensitivity, leaving little room for error; small differences in the directly computed percentage become large differences in the "appropriate gross profit percentage" that serves as the "plus" in the cost plus method. See sec. 1.482-2(e)(4)(i), (iii), Income Tax Regs. For example, Dr. Horst began the Comparable Norwalk Products method for instrument parts with P-E's total gross margin of 52.0 percent on finished instruments, which became a 40.2-percent markup over manufacturing costs. See supra pp- 133-134. *599 The Proportionate Profits approach to instrument parts began with a total gross margin of 59.6 percent for the 7 years taken together, which evolved, under the same computational steps as the Comparable Norwalk Products method, into a 67.3-percent manufacturing markup. Thus, less than 8 percentage points of difference in total margin blossomed into over 27 percentage points of difference in manufacturing markup. For HC lamps, similarly, less than 5 percentage points of difference in total margin became 28 percentage points of difference in manufacturing markup. These amplified differences caused respondent's income allocation to P-E for all parts to more than double, to the tune of an additional $ 7 million, under the Proportionate Profits analysis. Another problem with both of the Horst parts pricing methods is the seemingly simplistic allocation of the P-E manufacturing intangibles. Assuming that the PECC production costs were equal to the P-E production costs for the compared similar instruments, then the Comparable Norwalk Products method would have three profit percentages coinciding: the P-E manufacturing markup on the sale of the finished Norwalk instruments, the P-E *600 markup on the instrument parts kits sold to PECC, and the PECC profit on the labor and overhead costs it applied to the parts kits. In effect, this scenario assumes, as does the Proportionate Profits method, that the amount of manufacturing intangibles utilized was a direct function of an entity's manufacturing costs. Apart from simplicity and convenience, we see no meaningful justification for this assumed direct relationship, and respondent's broad and vague definition of manufacturing intangibles does not aid our understanding. It is true, as argued by respondent, that if the PECC production costs were different than those of P-E, then the Comparable Norwalk Products method would in effect assign a higher or lower profit percentage to the PECC labor and overhead costs than the derived P-E markup percentage on the parts. See supra p. 134. While this imparts a degree of flexibility to the Comparable Norwalk Products method, which counters petitioner's "formularly apportionment" argument, it does not solve the valuation problem for the manufacturing intangibles. Respondent has failed to justify the mathematical relationships, or even the implicit connection, between PECC*601 efficiency and the use of manufacturing intangibles. If respective manufacturing costs are indeed an appropriate means of allocating manufacturing intangibles, then we would still be troubled by how respondent proposes to define the respective manufacturing costs. Under both of the Horst approaches, the P-E parts kit cost base, prior to markup, includes the cost of value added by P-E before shipment to PECC and the entire purchased material cost, which was not insubstantial. PECC was left with its value-added cost, which consisted of only labor and overhead. Respondent, in effect, treats PECC like a consignment contract assembler in this regard -- a position coinciding with that on which respondent based the deficiency notice and then abandoned before trial. See supra p. 73. Dr. Horst explained his rationale in his initial report: P-E incurred the costs of identifying qualified suppliers, negotiating purchase prices, inspecting incoming shipments, stocking and releasing inventory, and gathering parts into parts kits. That process required P-E to invest in parts inventory, inventory storage facilities and the MRP system. Because the process of acquiring and inventorying*602 purchased parts and materials was anything but routine, we conclude that P-E's costs of purchased parts and materials should be combined with its other manufacturing costs in applying the cost plus method. While citing examples such as the intercompany account, see supra p. 42, respondent adds that this total material allocation to P-E accurately reflects PECC's lack of risk. Unlike respondent, we see no risk implications in PECC's use of the intercompany account for its parts kit payments, largely because P-E was the net debtor in this arrangement. We agree with respondent, however, that PECC benefited in terms of reduced risk by committing to purchase a parts kit only when P-E simultaneously committed to purchase the associated finished product. This arrangement effectively relieved PECC from the burdens of maintaining a parts inventory and of perhaps buying parts that would never find their way into sold finished products. Nonetheless, PECC did retain some risk in connection with its parts kit purchases. For example, P-E invoiced the parts f.o.b. P-E, and PECC incurred both scrap and rework costs during production. On balance, we do not believe that the circumstances*603 warrant a 100-percent allocation of material costs to P-E for purposes of the cost plus method, even after consideration of the cost-oriented factors emphasized by Dr. Horst. Respondent urges us to consider one other proposed comparable transaction under the cost plus method if we reject the Comparable Norwalk Products approach. We obligingly, but briefly, address the Coleman transaction with Sao Paulo, in which Coleman marked up the parts it sold to Sao Paulo by an average 51.5 percent. 22 See supra pp. 64-66. *604 Respondent concedes that the Coleman spectrophotometers sold by Sao Paulo were simpler than those assembled by PECC, thus rendering the Coleman parts kits noncomparable under the comparable uncontrolled price method. Although the cost plus method generally accommodates physically distinguishable products more readily than does the comparable uncontrolled price method, we cannot disregard meaningful distinctions even under the cost plus method. See sec. 1.482-2(e)(4)(iii), Income Tax Regs. Because respondent treats parts kits as imbued with intangibles such as technology, know-how, and an experienced engineering staff, the level of sophistication of the finished products makes a difference. Dr. Horst admitted that he lacked reliable information to quantify the effects of either technical complexity or Sao Paulo's right to use the P-E name in Brazil. We therefore decline to adopt this transaction as a cost plus comparable for parts. Because of the limited data and the circumstances detailed in our findings, we also do not find relevant the similarity between the Coleman markup on its parts kits and the markup on the associated finished instrument. See supra pp. 65-66. As*605 our final point concerning respondent's parts pricing adjustments, we note that respondent has failed to explain persuasively why royalties cannot adequately and more simply account for the manufacturing intangibles inherent in the parts kits. On this matter, respondent points to the P-E intersite licensing arrangements with subsidiaries, in which the P-E policy was to reduce the royalty rate by half if, for a particular licensed product, the parts sold by the licensor to the licensee represented more than 35 percent of the licensor's manufacturing cost for the finished product. See supra p. 28. Dr. Horst, who admitted that the royalty rate could in theory reflect the value of manufacturing intangibles embodied in the parts kits, thought the intersite royalty policy was consistent with his parts methodologies because the policy recognized a partial transfer of technology to the licensee rather than a complete transfer. He sought in his transfer pricing analysis to mirror the intent of P-E and PECC in entering into their licensing arrangement, which intent was, according to Dr. Horst and respondent, to reflect the value of manufacturing intangibles in the prices of the parts*606 kits. We do not attach the same importance that respondent does to the intersite royalty policy. Ordinarily, in an arm's-length analysis under section 482, we disregard transactions between the taxpayer and a related entity. Sundstrand Corp. v. Commissioner, 96 T.C. 226, 371-372 (1991); Eli Lilly & Co. v. Commissioner, 84 T.C. 996, 1143-1145 (1985), affd. in part, revd. in part, and remanded 856 F.2d 855 (7th Cir. 1988). As far as we know, and as suggested by the name, the intersite policy only applies to such related entities. Even for Dr. Horst's seemingly limited purpose of setting the basic structure of a licensing arrangement, the intersite policy comes up short. A 1968 internal P-E memorandum described the 35-percent crossover point for a partial versus a complete transfer as based on an "arbitrary assumption". Consistent with this characterization, examples in the same document showed a 34-percent cost figure not affecting the royalty rate and a 36-percent figure causing a halving of that rate. More importantly, Dr. Horst did not reconcile his parts pricing analyses with the apparently*607 nonexistent parts pricing implications in the intersite royalty policy. The markup percentage on the transferred parts in the memorandum examples was a constant 10 percent before and after the royalty rate dropped. A 1969 internal memorandum on intersite transactions again illustrated a halved royalty rate accompanied by an unchanging parts markup. B. Petitioner's Position Petitioner suggests a relatively straightforward approach to the arm's-length pricing of parts, which it argues comports with the comparable uncontrolled price method in section 1.482-2(e)(2), Income Tax Regs. See supra p. 80. Petitioner maintains that the arm's-length price of a parts kit is equal to the sum of the arm's-length prices of the individual parts, plus an upward level-of-the-market adjustment for the P-E purchasing, inventory, and freight forwarding activities. Even after a level-of-the-market adjustment, asserts petitioner, the markup PECC paid P-E over the P-E standard cost more than covered that adjustment. Petitioner accordingly argues that no section 482 income allocation attributable to parts is warranted. We begin with what we will call "manufactured" parts, to which, unlike*608 "purchased" parts, P-E applied at least some direct labor before shipment to PECC. The record is clear that all of the manufactured parts, including those with P-E specifications, were available from unrelated suppliers. Because the General Licensing Agreement granted PECC access to all of the technical documentation relating to its finished products, PECC in theory could have availed itself of these sources. 23 As to the cost of the manufactured parts, there was both general testimony and some more specifically directed toward account 2403, see supra pp. 20-21, supporting petitioner's position that these parts were available from outside suppliers at or below the P-E standard cost of manufacture. Section 1.482-2(e)(1)(iv), Income Tax Regs., permits the grouping of products in the determination of arm's-length prices in order to avoid the impracticalities*609 associated with a large number of products or transactions. Certainly, with the many thousands of parts implicated in this case, some form of grouping is necessary to apply the comparable uncontrolled price method. Moreover, the similarity in physical property and circumstances required by section 1.482-2(e)(2)(ii), Income Tax Regs., must not be read so strictly as to "threaten effective nullification of the * * * method of choice for testing transfers of tangible property between commonly controlled entities." Bausch & Lomb, Inc. v. Commissioner, 933 F.2d 1084, 1091 (2d Cir. 1991), affg. 92 T.C. 525 (1989). Nonetheless, we conclude that petitioner has not met the requirements of the comparable uncontrolled price method. The regulations contemplate comparisons with actual transactions in the form of uncontrolled sales. See sec. 1.482-2(e)(2)(ii), Income Tax Regs. Only with actual transactions can the similarity of the "physical property and circumstances" be evaluated to determine both comparability and any necessary adjustments to the uncontrolled price. Id. Petitioner emphasizes the physical similarity of the general*610 types of parts P-E sold to PECC as compared to the general types of parts that witnesses testified were available outside at less than the P-E standard cost. However, petitioner fails to mention the word "circumstances" in this portion of its brief. Section 1.482-2(e)(2)(ii) and (iii), Income Tax Regs., lists as possible differences having an effect on price the quality of the product, the terms of sale, delivery, and payment, and the level of the market and the geographic market in which the sales take place. Petitioner's only venture into this area is to assert that the level of the market was not a differentiating factor: a parts manufacturer sold to an instrument manufacturer in both the controlled (P-E to PECC) and uncontrolled (unrelated supplier to P-E) sales. As another possible circumstance, however, one of P-E's experienced purchasing agents testified that the length of lead time affected the outside price to P-E. PECC could adjust to a relatively short lead time, and benefited in this regard from the P-E parts inventory, because it normally did not order parts until a customer order was in hand at P-E. Disregarding the circumstances, which perhaps could have been *611 accounted for through statistical sampling of actual PECC parts or through more specific testimony, causes petitioner, in our view, to fail to come within the comparable uncontrolled price method. Without regard to differing circumstances, respondent challenges petitioner's assertion that the PECC parts were available from the outside at or below the P-E standard cost of manufacture, primarily by attacking account 2403. Recorded in account 2403 were outside purchases of parts that had also been manufactured by P-E at some time previously. See supra pp. 20-21. We agree that the conclusions drawn from this account are of limited value. Only the mechanical parts specialists in Purchasing regularly used account 2403, the purchases in the account represented only a small fraction of the Instrument Division cost of sales, and very few of the parts in the account were identical to parts that P-E sold to PECC. Respondent also correctly notes that P-E manufacturing reports on costs and hours showed P-E at least sometimes beginning to manufacture a part in-house with anticipated or resulting cost savings over its outside purchase price. Still, the evidence cited by respondent does*612 not effectively rebut or qualify the testimony cited by petitioner that manufactured parts of the types sold to PECC generally cost no more on the outside than the P-E standard cost. This testimony does not necessarily conflict with P-E's continuing to manufacture parts in circumstances such as when its variable manufacturing cost (i.e., excluding fixed overhead) was at or below the outside price; other circumstances might include desires to ensure availability when needed and to monitor quality more easily. Respondent has not demonstrated that potential suppliers were lacking for any major part category, nor has respondent given us other reason to think that petitioner's witnesses were wrong to a material degree in their generalizations. For purchased parts, which involved no P-E direct labor prior to shipment to PECC, petitioner again seeks to apply the comparable uncontrolled price method in section 1.482-2(e)(2), Income Tax Regs. See supra p. 80. Petitioner this time refers to more specific uncontrolled sales: before selling purchased parts to PECC, P-E acquired the very same parts from unrelated suppliers in arm's-length transactions. The only material difference between*613 the controlled and uncontrolled sales, according to petitioner, is that in the former P-E performed purchasing, inventory, and freight forwarding activities for PECC, suggesting the existence of different levels of the market. Petitioner contends that the $ 4.6-million excess over P-E standard cost that PECC paid for parts, see supra p. 41, far exceeded the market level price adjustment to which unrelated parties would have agreed. Despite the clear physical identity of the products involved in petitioner's controlled and uncontrolled sales, we must again reject petitioner's comparable uncontrolled price approach as ignoring potential differences in circumstances other than the level of the market. Petitioner, who did not offer third-party comparables at trial for the profit element in the services P-E performed for PECC, argues that P-E's 100-percent or more return on the cost of the services made the purchased parts prices to PECC more than arm's length. 24 Petitioner derives the cost of the P-E services by: (1) Allocating to PECC, based on sales, $ 424,000 of the Purchasing department expenses of $ 3.1 million, see supra p. 18; (2) adopting $ 1.8 million as P-E's maximum*614 inventory carrying cost for parts, based on an Arthur Andersen calculation, see supra p. 40; and (3) disregarding as de minimis the P-E costs to pull the parts from inventory and pack them for shipment. With a total P-E cost of less than $ 2.3 million under this analysis, the remaining $ 2.3 million of the excess that PECC paid over P-E standard cost was supposedly profit to P-E on these parts services. Respondent does not directly dispute petitioner's three cost premises, *615 instead arguing that petitioner deceptively minimizes P-E's "valuable and complex" purchasing function, which allegedly included many intangibles. Respondent does not attempt to quantify this value, however, either in total or element by element. Some of the specific points mentioned by respondent are well-educated and highly trained buyers, continual interaction with other departments, a proprietary qualified vendor list, and a computerized purchasing system that interacted with the MRP system. See supra pp. 18-21. In our view, factors such as these were not material in effect. The education and training of the Purchasing buyers, for example, were presumably inherent to some extent (through salaries and training expenses) in the Purchasing expenses of $ 3.1 million, which petitioner allocated based on PECC sales. Further, although Purchasing did indeed interact with Quality Assurance, Engineering, and manufacturing engineers, the extent of that interaction does not appear from the evidence to have been substantial. Even the qualified vendor list, although proprietary in some sense, was apparently nothing special in the industry. C. Conclusion Section 1.482-2(e)(1)(i), *616 Income Tax Regs., provides that an arm's-length price normally involves a profit to the seller. Although petitioner ignored this principle in applying its purported comparable uncontrolled price method to manufactured parts, we, having rejected the comparable uncontrolled price method for manufactured parts, will not. Accordingly, if we are to conclude that the parts prices overall were arm's length, the $ 4.6-million total markup over P-E standard cost, which petitioner associated entirely with purchased parts, now must cover the profit element to P-E on both manufactured and purchased parts. We lack a specific indication of respondent's position, however, concerning how much profit an arm's-length P-E would have earned on parts not imbued with so-called manufacturing intangibles. After consideration of the entire record, which is wanting on this matter through the fault of both parties, we conclude that the amounts PECC paid P-E for parts that went into finished products were arm's length. In so concluding, we reject petitioner's position that P-E received an excess over an arm's-length amount, which excess petitioner attempts to use as an offset to respondent's income allocations*617 in the finished product and royalty transactions. See supra p. 103; infra p. 162. VI. Royalties Pursuant to a licensing arrangement with P-E, PECC paid royalties on its finished product sales equal to 3 percent of the prices paid by P-E, which prices were either 25 or 30 percent below the P-E domestic list prices. See supra p. 41. Respondent's proposed royalty adjustment is actually a reverse income allocation, away from P-E, of $ 1.1 million. The royalty rates leading respondent to this allocation are 1.6 percent for instruments, 1.2 percent for HC lamps, and 3.8 percent for accessories. The royalty issue as presented to the Court for resolution is in an unusual posture. Petitioner maintains that the 3-percent royalty rate is equal to or higher than an arm's-length rate, yet it opposes respondent's use of Dr. Horst's overall lower rate. Respondent initially argues on brief that the 3-percent rate is "below market", then continues with reasons why it should be higher, and then adopts Dr. Horst's adjustments, which for instruments and HC lamps reduce the already "below market" rate. This strange situation, with petitioner appearing to favor a higher royalty*618 than respondent, is largely a result of the Horst parts pricing methods. Because respondent's parts and royalty allocations are interrelated, petitioner's arguments against respondent's downward royalty adjustments are, in effect, arguments against the Comparable Norwalk Products and Proportionate Profits methods for pricing parts. In the determination of an arm's-length royalty on the transfer of intangible property, the applicable standard is the amount that would have been paid by an unrelated party for the same intangible property under the same circumstances. Sec. 1.482-2(d)(2)(ii), Income Tax Regs. The best indicators of this amount are generally transfers by the same transferor to unrelated parties involving the same or similar intangible property under the same or similar circumstances. Id. If a sufficiently similar transaction involving an unrelated party is unavailable, the arm's-length amount is determined by evaluating various facts and circumstances. Sec. 1.482-2(d)(2)(iii), Income Tax Regs. The parties agree that the P-E license agreement with Hitachi from 1971 is sufficiently similar to the P-E and PECC licensing arrangement to serve as a comparable transaction*619 for instruments and accessories. See supra pp. 66-68. They also agree that the Juniper licensing arrangement with Instrument Lab is the best available comparable transaction for HC lamps. See supra pp. 69-70. They do not agree, however, about how the 7.5-percent rate paid by Hitachi and the 2-percent rate paid by Instrument Lab should be adjusted to account for differences from the P-E/PECC circumstances. We consider the HC lamps first. On the subject of the PECC royalty paid on HC lamps, Dr. Baumol stated in his original report that, under the terms of the Hitachi licensing agreement with P-E, HC lamps were subject to a royalty of 4 percent at most. He also described the overall comparability of the Juniper/Instrument Lab transaction. He calculated that adjusting the 2-percent royalty to account for the # 60,000 paid by Instrument Lab for advance royalties and technical assistance fees still left the resulting royalty below 3 percent. Petitioner, who adopts Dr. Baumol's one adjustment as the only material difference between the P-E/PECC license and the Juniper/Instrument Lab license, consequently argues that the 3-percent royalty paid by PECC on HC lamp sales was*620 an arm's-length amount. Respondent, in contrast, recommends a 4-percent royalty on HC lamps, for the reason that both Dr. Baumol and Dr. Horst supported this figure. Dr. Baumol, however, did not unequivocally adopt the 4-percent rate; indeed, his report stated that this was an "extremely conservative" ceiling amount, an assessment he repeated at trial. Moreover, although Dr. Baumol did not associate a 4-percent rate with the Juniper/Instrument Lab license, Dr. Horst in his supplemental report mistakenly "accepted Dr. Baumol's recommendation that the * * * [HC lamp] royalty be set at 4 percent, based on the Juniper License Agreement." Respondent does not quantify any upward adjustments to the Juniper/Instrument Lab royalty rate of 2 percent and does not attempt to rebut Dr. Baumol's calculated effect for the # 60,000 payment. Further, contrary to respondent's thinking, we do not believe that Juniper necessarily had difficulty penetrating the U.S. market, nor do we believe that P-E necessarily provided substantially more ongoing HC lamp technical assistance to PECC than Juniper provided to Instrument Lab. In these circumstances, given respondent's acceptance of the Juniper/Instrument*621 Lab arrangement as a comparable transaction, we conclude that the arm's-length royalty rate for PECC sales of HC lamps is tentatively 3 percent, subject only to our consideration of Dr. Horst's technology adjustment below. As noted, the parties agree that the P-E license agreement with Hitachi is a comparable royalty transaction for instruments and accessories. In this agreement, P-E granted Hitachi the exclusive right, with geographical restrictions, to use P-E technical information and patents to manufacture products from P-E's AA, IR, and gas chromatography product lines. P-E also granted Hitachi the nonexclusive right to sell the products, again in a limited area. Dr. Horst quantified two adjustments to this 7.5-percent rate, one upward and one downward. Adopting the upward adjustment, respondent argues that the Hitachi royalty of 7.5 percent is not directly comparable to the PECC royalty situation because the Hitachi royalty was based on the arm's-length "retail price" charged to "ultimate customers", while PECC paid royalties to P-E based on the transfer price between them. To compensate, Dr. Horst increased the 7.5-percent royalty to account for his calculated arm's-length*622 margins on P-E's finished product resales, which he had derived separately for instruments (32.7 percent) and accessories (57.6 percent) as part of his resale price method analysis discussed earlier. 25 See supra pp. 92-93, 124-125. Petitioner does not agree that the Hitachi royalty percentage applied to prices charged to ultimate customers. Respondent relies heavily upon the definition of "Net Sales Price" in the Hitachi licensing agreement. This was, as a general rule, the price charged to "customers", except that -- In the case of sales by * * * [Hitachi] to a subsidiary thereof or to an individual, corporation, partnership or association which is so closely associated to * * * [Hitachi] as to prevent arms length [sic] bargaining, the Net Sales Price*623 shall be deemed to be that charged by * * * [Hitachi] for similar products sold in the same period to customers not thus closely associated. Respondent's position appears to boil down to reading the word "customers" to mean instrument or accessory users. The agreement, however, nowhere defined this key term, and we do not have sufficient information to conclude that Hitachi planned to sell predominately or exclusively to users directly. On balance, we believe that "customers" reasonably could have a broader intended meaning that encompassed all those to whom Hitachi first sold the products, including intermediate distributors. 26 Both respondent and petitioner refer to other licensing transactions as evidence of industry practice and thus what Hitachi and P-E intended in their agreement. Respondent emphasizes P-E's position in the dispute with*624 Coulter over the blood-analyzer royalty, in particular the memorandum statement of counsel that giving effect to the first third-party sale while disregarding intermediate sales between affiliates was "customary practice in patent licensing." See supra pp. 68-69. The memorandum, however, did not specifically support the proposition, nor was the proposition consistent with the Juniper/Instrument Lab license, which gave effect to sales to related parties. In addition, the memorandum acknowledged that an agreed arrangement could differ from custom. Finally, we note that respondent's interpretation was apparently not "customary" enough to avoid a prolonged and ultimately unsuccessful dispute with Coulter. In sum, we are not prepared to interpret "customers" in the Hitachi agreement as restrictively as respondent urges, and we have not been directed to evidence concerning the nature of the Hitachi distribution system for the P-E products. Without clear guidance from the other licensing transactions in the record, we reject respondent's proposed adjustment to the P-E/PECC royalty premised on differing royalty computation bases. Dr. Horst intended his second adjustment to the Hitachi*625 7.5-percent royalty, this one downward, to account for the relatively limited manufacturing technology used by PECC in its operations. P-E supplied parts kits to PECC, and, as already detailed (see supra pp. 131-145), Dr. Horst adjusted the prices of those parts kits to reflect the value of the P-E manufacturing intangibles therein. His specific methodology for the royalty technology adjustment, which he viewed as consistent with his parts kit pricing analyses, was to reduce the royalty by the percentage of the P-E and PECC consolidated standard cost represented by the P-E standard cost of parts kits. For sample year 1980, the P-E standard cost for instrument parts sold to PECC was $ 3,738,902 (or 89.3 percent) of the consolidated standard cost of $ 4,184,867, and the PECC standard cost for labor and overhead applied to those parts was $ 445,965 (or 10.7 percent). See supra pp. 135-136. Dr. Horst multiplied 10.7 percent by the 7.5-percent Hitachi rate to arrive at an adjusted royalty of 0.8 percent for instruments. Respondent adopts Dr. Horst's downward technology adjustment to the royalty rate only contingently, subject to our acceptance of one of the Horst parts pricing*626 analyses. Because we have declined to adopt either the Comparable Norwalk Products method or the Proportionate Profits method, and because petitioner does not approve of the Horst technology adjustment to the Hitachi royalty, we will disregard this downward adjustment. As an aside on this matter, given the nature and state of progress of the parts kits that PECC received from P-E, PECC plainly did not need or use P-E technical manufacturing documentation to the same extent that Hitachi did. Viewed from this narrow perspective, the PECC royalty and the Hitachi royalty arguably would be comparable only after adjusting for this difference through something similar to the Horst technology adjustment. However, although the record is unclear as to whether PECC actually received complete technical documentation for all of its licensed finished products, PECC clearly had a contractual right to it, just as Hitachi did. PECC would have needed that documentation to procure manufactured and purchased parts in the manner implicitly hypothesized in petitioner's parts pricing theory. See supra pp. 145-151. Because we have accepted the parts pricing result urged by petitioner, with the*627 exception of a purported excess over arm's-length parts prices, a technology adjustment indeed appears unwarranted. Respondent supplements Dr. Horst's adjustments to the Hitachi royalty rate with other arguments to at least counter any further downward pressure on the adjusted royalty. Limitations on the geographic area covered can be a factor in arriving at the amount of an arm's-length royalty, sec. 1.482-2(d)(2)(iii)(c), Income Tax Regs., and although P-E limited Hitachi to sales in Japan and a select list of other countries, P-E granted PECC the right to sell worldwide. PECC did not sell its finished products worldwide, however, and in fact sold only to P-E. As shown by Gaynor Kelley's presentation to the P-E board of directors and PECC's initial application for an industrial tax exemption, this had been the plan from the start. See supra p. 26. Granted that P-E exploited established worldwide markets which PECC automatically tapped into, we have already accounted for this valuable marketing intangible in determining the P-E arm's-length resale margins for the PECC finished products. As to respondent's other points, it is not at all clear from the evidence how much *628 more technical assistance P-E provided to PECC than to Hitachi. Also, given the several countries other than Japan in which Hitachi could sell P-E products, the Hitachi royalty rate was not necessarily lower than otherwise because of P-E's failure effectively to penetrate the Japanese market. Petitioner, relying on Dr. Baumol, advocates different means to reduce the 7.5-percent Hitachi royalty and thereby render it comparable to the 3-percent PECC royalty that petitioner contends was arm's length. Dr. Baumol noted in his initial report that the Hitachi license was much broader in scope than the PECC license, in the sense that Hitachi could choose to manufacture and sell any or all of the low-priced PECC products plus many other higher priced P-E products. Petitioner accordingly characterizes the Hitachi royalty as including additional consideration for a valuable option on a broader range of intangibles. Dr. Baumol also observed that because Hitachi declined to manufacture the PECC products, this meant that Hitachi considered the opportunity worth something less than 7.5 percent. We are not persuaded. The notion that Hitachi perceived value in an "option", independent of the*629 intangibles actually used, is speculative. Although we have available some information concerning the pre-1971 dealings between P-E and Hitachi, we do not know what P-E products Hitachi planned to manufacture in 1971, nor what products P-E thought or hoped Hitachi would manufacture. Regarding the Hitachi decision not to manufacture PECC products, it may well be true that Hitachi in some sense valued the opportunity at less than 7.5 percent, but this point standing alone is not meaningful. For any number of legitimate business reasons, Hitachi could have decided that it would not manufacture and sell a particular P-E product in exchange for even a nominal royalty. Without knowing those reasons, we cannot evaluate the possible adjustment effect on the 7.5-percent royalty rate. Petitioner's final suggested rationale for adjusting the Hitachi royalty downward, and the only one that Dr. Baumol attempted to quantify, relates to the markup over P-E standard cost that PECC paid for parts. See supra p. 41. In his initial report, Dr. Baumol divided the $ 4.6-million parts markup by the approximately $ 75 million in PECC finished product sales, concluding that PECC was paying P-E *630 an implicit royalty over six percentage points higher than the recorded 3 percent. Petitioner, on brief, lowers the $ 4.6 million to $ 2 million because petitioner's parts pricing theory assumes that PECC paid some of the $ 4.6 million as arm's-length compensation for P-E parts-related services. See supra pp. 149-150. In essence, then, petitioner asks us to apply as an adjustment to the Hitachi royalty the portion of the $ 4.6-million parts markup that is not necessary to justify arm's-length parts prices. We have held, however, that the P-E parts prices to PECC were arm's length in the aggregate. There is no excess available for a royalty adjustment. See supra p. 152. In conclusion, we hold that the arm's-length royalty rates applicable to PECC sales of finished products to P-E are 7.5 percent for instruments and accessories and 3 percent for HC lamps. VII. Petitioner's Cost Plus Method In the notice of deficiency, respondent treated PECC as a consignment contract assembler and applied a cost plus method to determine the appropriate PECC profits. See supra p. 73. As petitioner views it, the notice of deficiency approach restricted PECC to a return for*631 its manufacturing function alone, without any return for its licensed intangibles. This approach, petitioner argues, establishes a ceiling for the income allocations derived from respondent's trial positions, which do ostensibly allow for a return on the licensed intangibles. Petitioner contends that respondent's notice of deficiency erred in four respects in applying the cost plus method, and that correcting these errors results in an income allocation to P-E roughly consistent with petitioner's trial position. According to petitioner, respondent in the notice: (1) Improperly excluded from the PECC cost base the cost of the materials that PECC purchased from P-E; (2) ignored a good portion of the location savings attributable to PECC; (3) incorrectly used operating profit markups rather than gross profit markups of 29 supposedly comparable electronics companies; and (4) erroneously assumed that the functions of the 29 companies were comparable to PECC's production of HC lamps. For purposes of their ceiling analysis, petitioner accepts the 29 electronics companies as comparable for the PECC instrument and accessory production. Petitioner's analysis raises a host of issues, most*632 of which we find unnecessary to address because of our holdings above that result in income allocations to P-E considerably less than called for under respondent's trial positions. However, because petitioner's "corrected" cost plus method undercuts even our income allocations to P-E, we will not entirely refrain from comment. Specifically, we address petitioner's asserted incomparability for the 29 electronics companies as relating to the PECCHC lamp production, and we also briefly address the applicable PECC cost base. Petitioner suggests that a much more appropriate comparable for valuing HC lamp manufacturing skills is ARC, which manufactured precision mercury capillary lamps for the Micralign system sold by the P-E Optical Group. See supra pp. 71-73. For these Micralign lamps, Arthur Andersen determined that ARC earned a gross profit markup on its cost of sales of approximately 200 percent for its 1976 through 1982 fiscal years. The effect of using ARC rather than the 29 electronics companies as an HC lamp comparable, according to petitioner, is an approximately $ 10-million reduction in the income allocation to P-E under the cost plus method in the deficiency notice. *633 On this record, however, we do not believe that the Micralign lamps were comparable to the PECCHC lamps within the meaning of section 1.482-2(e)(4), Income Tax Regs., which details the cost plus method. Dr. Baumol thought that the ARC manufacture and sale of Micralign lamps was comparable to the PECC manufacture and sale of HC lamps in several respects. As examples of similar sophistication, he mentioned the products, the work forces, and the manufacturing technology. He also noted the similar sizes of the companies and their fast responses to orders. Petitioner cites as additional support for comparability the testimony of people involved in either the manufacture of HC lamps or the manufacture of Micralign lamps. From a technical perspective, however, Dr. Baumol by his own admission lacked industry expertise, and petitioner's cited testimony fails to establish the comparability link between the manufacturing processes for the two lamps. See sec. 1.482-2(e)(4)(iii)(b), Income Tax Regs. Indeed, although P-E was an established HC lamp manufacturer by 1972, it did not set up its own Micralign lamp manufacturing facility until it learned the manufacturing process from ARC. *634 Moreover, the lamps were not even used in the same industries; the Micralign lamp was an integral component in very expensive equipment used to produce semiconductors. See sec. 1.482-2(e)(4)(iii)(a), Income Tax Regs. The technology flows between P-E and ARC, of indeterminate net effect, also preclude us from finding the two lamps comparable. P-E had a patent on the lamp mount that arguably enabled it to extract more profit from ARC, but ARC, for its part, had originally developed manufacturing know-how for the lamp itself and had transferred that know-how to P-E. Thereafter, ARC occasionally provided technical assistance. Finally, ARC's annual sales of Micralign lamps increased at a much higher rate than the PECC sales of HC lamps, indicating a large demand discrepancy with the potential for affecting relative profitability. As another of respondent's supposed errors in applying the cost plus method in the deficiency notice, petitioner argues that the PECC cost base should include the cost of the materials that PECC acquired from P-E. Whether petitioner is correct or not on this point, which we do not here decide, we need only note for present purposes that this argument appears*635 to detract from why petitioner chose to focus on the deficiency notice in the first place, which was to set a contract assembler ceiling on respondent's income allocations. Arguably, assigning the entire material cost to PECC weakens the contract assembler analog. VIII. Profit-Split Method One final word. Both parties submitted broad profit-split analyses, but only respondent sought sustenance for her position on this ground in her brief and then only as a check on the allocations she derived from the methods detailed in the section 482 regulations. Whatever benefit might have been supplied by a profit-split analysis, the calculations of the parties are not sufficiently grounded on basic facts to permit their use as a check on the conclusions we reach herein; there are significant differences between the parties, unresolvable on the record, in respect of the amounts of operating expenses, variable manufacturing profit, and location savings. Under these circumstances, we think it inappropriate to seek to develop such a check on our conclusions, particularly since, despite the difficulties we have encountered in reaching them, we are satisfied with the factual foundations*636 upon which they rest. To reflect the foregoing, An appropriate order will be issued. Footnotes 1. Statutory references are to the Internal Revenue Code as in effect for the years in issue. Unless otherwise indicated, Rule references are to the Tax Court Rules of Practice and Procedure.↩ 2. Unless otherwise clear from the context, our findings relate to the 7-year period in issue, and a reference to a year from 1975 to 1981 is to the P-E taxable year ending on July 31 of the referenced calendar year. For example, a reference to "1979" without more specificity is to the taxable year beginning August 1, 1978, and ending July 31, 1979; "October 1979", however, as an example of a more specific reference, means October of calendar year 1979.↩ 3. Consumables in this context are small items used directly with instruments, such as chart paper, lamps, columns, and bulbs. Although consumables could be thought of as a type of accessory, the term "accessory", as we use it, does not include consumables.↩ 4. An additional $ 26.1 million represented sales of component products, namely, printed circuit boards and mechanical and electromechanical assemblies. As earlier noted, the parties have settled the sec. 482 implications of the component products. See supra↩ p. 4. 5. Although autobalances and furnaces are not instruments, they are similar to instruments in terms of price and technical complexity. The original PECC financial statements and the parties' adjusted, stipulated product line income statements group these items with instruments rather than accessories.↩ 6. Gross profit is equal to net sales revenue less cost of sales. The cost-of-sales components are material, labor, and overhead; operating expenses are not included. A gross profit margin, in percentage form, is the gross profit amount divided by net sales revenue and then multiplied by 100. For convenience, we sometimes refer simply to a "gross margin" or "margin", meaning a percentage gross profit margin.↩ 7. Petitioner has conceded that a 30-percent discount should have applied throughout the years in issue. The stipulated effect is $ 1,609,040 additional gross income to P-E because of the reduced transfer price and $ 45,160 less gross income to P-E because of a smaller royalty base.↩ 8. PECC never assembled the Model F-11 gas chromatograph, to which the 1970 licensing agreements assigned a 3.5-percent royalty.↩ 9. Unrelated party sales by P-E, in this and all other Arthur Andersen calculations, were those made indirectly through USSD and P-E International and those made directly to foreign third parties; sales to foreign subsidiaries were not included. All references herein to P-E unrelated party sales are to this definition.↩ 10. The dispersive and nondispersive technologies differ, in part, in the method used to produce wavelength spectra. See supra↩ p. 7. 11. Both parties' experts assumed that the time between P-E's shipment to a customer and subsequent payment by that customer was the same for the Hitachi and PECC instruments. Accordingly, the relevant periods actually compared were from the time P-E paid the manufacturer to the time P-E shipped to the customer.↩ 12. The 3.5 percent resulted from 121/365 (the portion of the year) x 10.58 (the short-term borrowing rate).↩ 13. The 1.9 percent resulted from 27/365 (the portion of the year) x 37 (the Grossman cost of capital) x 0.7 (the 30-percent discount from list price).↩ 14. The 1.1 percent was the P-E net service cost for Spectra-Physics products (1.5 percent) less the P-E net service cost for PECC products (0.4 percent). See supra↩ p. 45. 15. Respondent has objected to this exhibit as inadmissible hearsay. Because we use this exhibit solely against petitioner's interest, we need not decide the hearsay question.↩ 16. Petitioner also invokes the cost plus method as applied to the ARC Micralign lamps, but because petitioner raises this method in another context that we discuss in due course, we defer comment at this point. See infra↩ p. 164. 17. When Dr. Horst described captive products and their pricing in his original report, he did so without reference to the possible influences of the competitive environment. Nonetheless, the parties consider competition part of the captive product analysis, as affecting the degree or the very existence of captivity, and we will do likewise.↩ 18. Respondent's reference to the purported market dominance of another major PECC product, the HGA furnace, is inappropriate here. Dr. Horst treated HGA furnaces and autobalances as instruments rather than accessories in his analysis, so any purported captive characteristics are irrelevant in the present accessory context. See also supra↩ note 5. 19. Respondent uses "manufacturing intangibles" in a broad sense to mean elements of value, other than marketing intangibles, in finished products. As examples applicable to P-E finished products, respondent on brief lists, among other things, technology, know-how, an experienced engineering staff, a vertically integrated structure that allowed a cost-efficient operation, an ability to innovate and respond to market demands for new techniques, and an extensive purchasing, production planning, and inventory control system.↩ 20. With a total gross margin of 52.0 percent of sales, cost-of-sales elements made up the remaining 48.0 percent of sales. Dividing the 19.3-percent manufacturing margin on sales by the 48.0-percent cost of sales gives a manufacturing markup on costs of 40.2 percent.↩ 21. For the PECC products, the cost-of-sales components in the gross margin computation were the combined costs of P-E (for the parts kits) and PECC (for labor and overhead).↩ 22. Respondent actually interjects this position between the preferred Comparable Norwalk Products method and the subordinated Proportionate Profits method, presumably because its cost plus orientation takes precedence over the Proportionate Profits profit split under the applicable regulations. However, Dr. Horst did not incorporate the Coleman/Sao Paulo cost plus implications into his detailed schedules and computations. For convenience, we have accordingly referred to the Proportionate Profits method as the foundation of respondent's secondary, rather than tertiary, trial position. ↩ 23. We do not agree with respondent that the terms of the General Licensing Agreement necessarily precluded PECC from having parts manufactured outside of Puerto Rico.↩ 24. Sec. 1.482-2(b)(3), Income Tax Regs., relied upon by neither party, provides that an arm's-length charge for services between related parties is generally deemed equal to the costs or deductions incurred by the service provider. However, when a seller of tangible property renders ancillary services in connection with the sale, the effect of such services is accounted for under sec. 1.482-2(e), Income Tax Regs., which describes the comparable uncontrolled price, resale price, and cost plus methods. Sec. 1.482-2(b)(8), Income Tax Regs.↩ 25. Dr. Horst did not make this adjustment for HC lamps because the Juniper/Instrument Lab agreement applied the royalty percentage to the invoice price of the first sale following manufacture, whether or not the purchaser was related to Instrument Lab.↩ 26. We have used "customer" and "user" interchangeably throughout this opinion only for convenience; we do not mean to suggest that this was an industry-wide practice.↩