The holder of a patent for a device used to cut grooves in duct board so that it can most practically be used to form ducts for central air conditioning and heating systems appeals the determination by the trial court that the patent was not infringed by a competing machine. The issues require only the application of long settled principles. Differing with the trial judge, we conclude that the plaintiff's device was equivalent in function and operation to the patented machine and that it, therefore, infringed the defendant's patent.

The principles that guide us have been frequently stated and restated. We sum them up without lengthy citation. The validity of a patent and its interpretation are both questions of law. Each may depend upon factual inquiries. After the proper construction of a patent is determined, whether the patent, thus interpreted, has been infringed is a question of fact.

A patent is obviously infringed if the accused device incorporates its teaching literally read. In reading a patent to determine whether it has thus been infringed, its claims must be read in connection with its specifications and its file history; the patent's claims cannot be construed to reach beyond the teachings expressed in the patent. Minor modifications are, therefore, sufficient to avoid literal infringement. Such alterations may, however, result in a device that appropriates the invention and incorporates its innovative concept into a device that performs substantially the same function in substantially the same way to obtain substantially the same result. The "doctrine of equivalents" protects the patent and its inventor from such abuse.

The ultimate question, as the title of the rubric implies, is whether the accused device is functionally equivalent to the patented one. What constitutes equivalency is determined in the context of the scope of the prior art, the nature of the patent, the real invention disclosed by its specifications and examples and the extent of innovation accomplished. A pioneer patent disclosing a function never before accomplished is entitled to broader protection than one that makes minor improvements on known technology. The range of equivalents is also limited by the patentee's surrender or amendment of claims in response to demands of the patent examiner, a process that creates "file wrapper estoppel."

In applying these rules, if the question involved is a legal one, we are free on appeal to reexamine the answer given by the trial court. If it is, however, factual, then we defer to the trial court's findings and may reverse them only if they are clearly erroneous. Fed.R.Civ.P. 52(a). See Kaspar Wire Works, Inc. v. Leco Engineering & Machine, Inc., 575 F.2d 530, 533 n.1 (5th Cir. 1978). These precepts are taught by a host of cases, a number of which are cited in our own recent opinion in Studiengesellschaft Kohle mbH v. Eastman Kodak Co., 616 F.2d 1315 (5th Cir. 1980).

Central air-conditioning and heating systems require ducts to carry the heated or cooled air from the central unit to various areas. Ducts should be insulated for energy efficiency. One commonly used insulation is factory-made in a form called "duct board;" this is a sheet of matted fiberglass held together by a binder to form a board with a vapor barrier such as aluminum foil on one side. Duct board is usually shipped by the manufacturer in large flat sheets. On the job, the sheets are cut and formed into the sizes and the three-dimensional shapes required.

The thick board cannot readily be bent into shape; the method used to fabricate the desired ducts is to cut grooves into the board, removing the insulating material but leaving the vapor barrier intact, then to fold the board into the desired shape. The edges brought into contact are then resealed. The duct sections thus formed are joined to make up the air carrying system. The typical duct is rectangular, but the same method can be used to produce other shapes.

Two decades ago grooves were cut in duct board by the hand use of cutting tools. As the use of duct board increased, machines were developed to reduce the hand labor. By 1965 two machines were available; in each the duct board was held in a fixed place and cutting tools, mounted on a tool bar, were drawn across the board to cut the grooves. Another tool, made by the Howard C. Forrest Co. and known as the Forrest machine, used cutters fixed in place. The duct board was fed horizontally across the cutters by a pair of feed rollers a set of conveyor rollers that looked much like rubber tires mounted on a rotary shaft. The cutters were mounted on a support. A roller located on the opposite side of the board from the cutters held the board against the cutting blades as the board was moved across the cutters. The knife blades were located so that they made the deepest cut in the board just behind the top of that roller. The cutting blade used was triangular, but there was testimony that the machine could be adapted to use knives that would cut shiplap grooves.

It is easier to form duct board into the desired shape when the grooves are cut shiplap rather than triangular. Shiplap grooves also permit square corners to be formed, which is difficult to accomplish with triangular grooves. The patent (U.S. Patent 3,605,534, referred to as the '534 patent or the Glass Master patent) is for a machine to cut several shiplap grooves in duct board while it is passing through the machine so that it can be fabricated into duct sections. The invention covered by the '534 patent provided for the first time a machine that in a single pass would cut a modified shiplap groove without piercing the foil, so that only a single stapling and taping of a flap was necessary when folding a grooved board into rectangular shape to form a duct. To make this duct by hand required 12 minutes. It could be made in a fraction of that time on the machine.

Application for the '534 patent was filed in May 1967. The '534 patent is owned by defendant Willson, and Glass Master Products is Willson's exclusive licensee. Another duct board cutting machine, called the Fiber Dragon, is made by the plaintiff, Weidman Metal Masters Company.

The '534 patent teaches a machine having cutters mounted on a tool bar support, with feed rollers mounted on opposite sides of the support for feeding a sheet of duct board across the cutters. Each of the cutters includes at least one knife blade formed to produce the desired deep cut in the board close to the foil. Each knife blade includes a flat bottom section for cutting directly over a blade support roller. The knife blade is shown as having a second section situated above this cutter and below the knife blade carrier, designed to make the blade resilient. This prevents the blade support roller from exerting excessive force on the lowermost flat section of the knife blade.

The heart of the '534 invention lies in the relationship of three elements: the tool bar or support (which is rigid on Glass Master), the knife blade and the support roller. The '534 patent teaches that the position of the knife blade against the duct board is controlled by the support roller, and that two of these elements should be rigid and the third relatively resilient so that the knife blade, guided by the support roller, will rise and fall as the roller rotates in order to maintain its proper cutting position. When these three elements are used in the functional relationship conceived by the inventor, the machine removes all of the insulation material without piercing the vapor barrier and leaves a shiplap groove that facilitates easy assembly of a duct.

The patent teaches that the preferred form of the device employs a rigid tool bar and a rigid support roller with a resilient knife blade. While passing through the machine the duct board is slightly compressed between the blade and the support roller so that the foil vapor barrier first touches the support roller in front of the leading edge of the knife blade and remains in contact with the roller as the foil slides under the blade. When the knife blade is positioned behind the top dead center of the roller, as in the Weidman machine, the same effect is achieved. The duct board expands as it passes over the top of the roller and presses against the downward arc of the roller. The foil remains in contact with the roller and the expansion pressure continues to hold the foil in position against the roller so that the blade strips the insulation material without piercing the foil. If the support roller rotates imperfectly in the Glass Master machine, the resiliency in the knife blade prevents the imperfection from causing the roller and blade to contact with one another and cause the blade to tear the foil vapor barrier. The '534 patent suggests, as even the Fiber Dragon proponents admit, that the resilient means necessary for operation of the contemplated device might be provided by providing resilience in another of the three elements, the support roller, but it makes no express claim for resilience in either the support roller or the tool bar.1

Weidman began production of its Fiber Dragon in 1973. In the Fiber Dragon machine the knife blades do not cut directly on a roller. The knives are mounted so that they make their deepest cuts behind a guide roller and in a plane below the plane defined by the tops of the lower feed and guide rollers. The Fiber Dragon machines do not have resilient knife blades.

Before building their first Weidman machine, Weidman's employees had examined a Glass Master machine, and the Fiber Dragon machine is in many aspects a copy of Glass Master. It uses the same three basic elements, the tool bar, the cutting knife blade and the support roller (a guide roller in the Fiber Dragon). Two of these elements are designed to be rigid and one is in fact resilient, the tool bar. Whether the resiliency is purposefully provided is not clear; it is evident that, given relative rigidity in the other members, the Fiber Dragon would tear the vapor barrier unless resiliency was supplied by the tool bar.

In operation of the Fiber Dragon, the duct board does not merely roll tangentially to the guide roller but, as a result of compression, remains in contact with the top two inches of the roller. The compression not only forces the duct board downward against the roller but upward against the knife blade. If nothing yielded, then the knife blade could cut the foil. Instead the upward force lifts the blade slightly more than 3/32 of an inch. This is accomplished by bending the tool bar, which is two inches square but ten feet long. Thus the indispensable resilience is provided by the flexibility of the tool bar.2

A second basic difference between the Fiber Dragon and the '534 patent is in the position of the knife blade in relationship to the guide roller. On Glass Master the knife is immediately over the top of the roller, a position taught by the patent which causes the roller to support the knife blades; the knife on the Fiber Dragon is positioned behind the apogee of the roller. The expert who testified on behalf of Weidman found in this the only substantial difference between the two machines. If, however, the knife on the Fiber Dragon were positioned over the roller top, it would cut better. Moving the knife back slightly makes the cut less consistent and results in leaving more fiber in the cut.

The trial judge's oral remarks, which indicate his understanding even more clearly than the written findings he signed upon submission, limited the patent to a device with a resilient blade positioned directly over the roller. He concluded that the claims of the patent, which call for a roller acting as a support for knife blades, could not be read to include a machine in which the roller was not acting as a blade support. This denied the patent any range of equivalents. He then found the Fiber Dragon different from the invention claimed by Patent '534 in the two ways discussed: the position of the blade in relationship to the roller and the reliance on either the knife blade or the rollers to provide resiliency during cutting.

Claim 1 is set forth in the margin.3 At least one of its descriptions is not literally infringed. The last paragraph of claim 1 teaches that one of the knife blades and one of the blade supports includes resilient means to avoid excessive force on the knife blade (which would result in piercing the vapor barrier). The Fiber Dragon includes no resilient means in either member. We find it unnecessary similarly to analyze each of the remaining ten claims. In his written findings of fact, prepared by counsel for Weidman, the trial judge did so and concluded there was no literal infringement. Although we do not necessarily subscribe to his entire interpretation of each claim, and we are mindful that the interpretation of the patent is a question of law, we find his reading of salient portions of each claim legally correct, and we do not find erroneous the factual conclusion that the claim, thus read, is not literally infringed.

Inventions, however, are not limited to the preferred embodiment disclosed in a patent. Patent construction is not a matter of pure literalism and slavish adherence to the words used. Merry Manufacturing Co. v. Burns Tool Co., 335 F.2d 239, 244 (5th Cir. 1964). A patent is to be construed as a contract with the intent of the parties as the lodestar. It is the real invention claimed and granted protection which we seek to determine. Del Francia v. Stanthony Corp., 278 F.2d 745 (9th Cir. 1960). See Laitram Corp. v. Deepsouth Packing Co., 443 F.2d 928 (5th Cir. 1971).

In Fiber Dragon the knife is rigid but the cutter is carried on a resilient tool bar. The machine in operation relies upon a long slender tool bar to provide the necessary resiliency. Claim 1 reads in terms directly upon the structure of the Fiber Dragon except that the '534 patent calls for the knife blade means or the blade support to be resilient. The Fiber Dragon merely transposes parts that perform the identical function. Such transposition does not avoid infringement.

The alteration of the knife's relationship to the roller is but a slight repositioning of an element of a combination and the consequent acceptance of a slightly less efficient job. This does not avoid infringement. The positioning of the knife behind, rather than above, the guide roller does not alter the fact that, in both machines, the roller exerts force through the duct board against the knife blade, thus positioning the blade for the optimum cut. Without resiliency, this would cause the knife to tear the vapor barrier, particularly when the blades dulled and had to be moved closer to the roller to achieve the desired cut. The patent teaches both the effect of the roller on the knife and the need for resiliency to utilize that effect. Moreover, in both machines the compression of the duct board between a roller and the knife blade causes the foil vapor barrier to remain in contact with the roller as it passes by the forwardmost cutting edge of the knife. The operation and the results achieved by the Fiber Dragon are essentially the same as those of the device disclosed by '534 save for inferior performance.

It is not necessary for us here to determine how near to the forefront of pioneers patent '534 comes. Although it did achieve a result never before accomplished, even the minimum equivalency due to any patent normally forbids the mere reversal of a function of two parts and the small movement of one part to avoid literal infringement by accepting a less efficient job.

Decision of an infringement claim requires a court to look at the heart of the invention. If it is appropriated, the patent is infringed. Harrington Manufacturing Co. v. White, 475 F.2d 788 (5th Cir.), cert. denied, 414 U.S. 1040, 94 S. Ct. 542, 38 L. Ed. 2d 331 (1973). Normally infringement is a question of fact. Ziegler v. Phillips Petroleum Co., 483 F.2d 858, 867 (5th Cir.), cert. denied, 414 U.S. 1079, 94 S. Ct. 597, 38 L. Ed. 2d 485 (1973). However, in this case, the critical analysis provided by the district judge involved the construction of the patent. When this is so, legal questions are intertwined with the factual issues and we are not strictly limited by the clearly erroneous standard in our review. Id. As we construe the patent, its crucial teaching was the relationship between the support roller and the knife blade achieved when the roller transfers force through the duct board to a knife resilient enough to react to that force. Neither the position of the knife in relation to the roller nor the source of the resiliency is crucial, although both are important, as evidenced by the inferior performance of the Weidman machine. However, in regarding these factors as limitations on the patent and finding no infringement for that reason, the district judge erred.

The trial judge found the patent valid. Weidman argues alternatively that its device can be found to infringe '534 only as the result of reading that patent so broadly as to make it invalid when compared with prior art, embodied in a machine called the Forrest machine. The Forrest machine as it was used before the date of filing the '534 application, was not capable of making a modified shiplap cut. It utilized V-blades and left a space between the duct board and the cutters. The sole function of the roller in the Forrest machine was to hold down and guide the board across the blade. When, during the course of the trial, an experiment was conducted outside the court room by putting shiplap tools on Forrest, the foil was torn and there were problems in getting the knives to work. There was apparently no resiliency in the Forrest blade apparatus, and the guide roller did not operate to position the blade relative to the duct board vapor barrier. The '534 patent was, therefore, neither anticipated nor made obvious by Forrest.

For these reasons, the judgment of the district court that the Fiber Dragon does not infringe the '534 patent is REVERSED, the determination that the patent is valid is AFFIRMED, and the case is REMANDED for further proceedings, consistent with this opinion.APPENDIX

NOTE: OPINION CONTAINS TABLE OR OTHER DATA THAT IS NOT VIEWABLE