In this patent interference action, two parties claim priority with respect to a patentable refinement on a process for synthesizing urea, an industrial chemical compound. Appellant Chemical Construction Company (Chemico) is the assignee of a United States patent on a process for synthesizing urea issued in 1965 to Ivo Mavrovic (the Mavrovic patent). This patent was based on a United States application filed by Mavrovic in June of 1961. Appellee Stamicarbon N.V. is the assignee of a United States patent application on a urea synthesization process which was filed by appellee Mathieu Bongard in March of 1966. In these proceedings, Stamicarbon has claimed that Bongard's application relates back to a Netherlands application filed by him in 1960.1

To provoke this interference action, Bongard copied two claims of the Mavrovic patent in 1966. In April of 1969, the Patent and Trademark Office declared the interference.2 In May of 1972, the Board of Patent Interferences awarded Bongard priority.3 This action was then initiated in the district court pursuant to 35 U.S.C. § 146.4

In the district court proceeding, neither party sought to prove actual invention; each relied on his application filing date for priority purposes. Chemico presented essentially three issues to the district court. First, it contended that the Bongard application did not teach the same process as the Mavrovic patent. Second, Chemico attacked Bongard's right to the Netherlands filing date, claiming that foreign applications filed prior to September, 1960 disclosed the same process; thus, the Netherlands application was not the earliest foreign application as required by 35 U.S.C. § 119, supra note 1. Finally, Chemico contended that the oath Bongard filed with his parent application was false. The district court determined (1) that the Bongard application and the Mavrovic patent disclosed the same process, (2) that Bongard's earlier foreign applications did not disclose the process set out in his Netherlands application and hence he was entitled to the filing date of the latter, and (3) that Bongard's parent application oath was not false.5 Accordingly, Bongard was awarded priority on the two interference counts.

On appeal, Chemico's principal contention, simply stated, is that the district court erred in concluding that the Bongard application discloses the refinement in urea synthesization technique taught by the Mavrovic patent. Because we agree with this contention, we reverse the judgment of the district court and do not reach other issues tendered by Chemico.

Although a district court proceeding pursuant to 35 U.S.C. § 146 is a trial de novo, it is governed by the strict rule of Morgan v. Daniels, 153 U.S. 120, 14 S. Ct. 772, 38 L. Ed. 657 (1894), that a Patent Office decision on priority of invention is not to be disturbed "unless the contrary is established by testimony which in character and amount carries thorough conviction." Id. at 125, 14 S.Ct. at 773. "The force of Morgan v. Daniels has been repeatedly recognized by this Court." Radio Corp. of America v. International Standard Electric Corporation, 232 F.2d 726, 729 (3rd Cir. 1956). We have said that the "Patent Office decision presents itself as a particularly heavy unit of weight which the evidence of the plaintiff in the district court must counterbalance." S & S Corrugated Paper Machinery Co. v. George W. Swift, Jr., Inc., 176 F.2d 358, 360 (3rd Cir. 1949).

Our review of the district court decision is governed, as to questions of fact, by the clearly erroneous rule. F.R.C.P. 52(a). The question for this court is "whether the District Court erred in confirming the Patent Office's award of priority of invention . . . ." Smith v. Carter Carburetor Corporation, 130 F.2d 555, 560 (3rd Cir. 1942). And where a factual matter is solely dependent on documentary evidence, we have more leeway, "(s)ince we are in just as good a position as the district court to evaluate (such) evidence . . . ." In re Multidistrict Litigation Involving Frost Patent, General Tire & Rubber Company, Appellant, 540 F.2d 601, 603 (3rd Cir. 1976). On questions of law, of course, we exercise plenary review.

With these precepts in mind, we turn to an examination of the processes at issue.

The prior art urea synthesis process which both Bongard and Mavrovic sought to improve involves taking a process stream (an aqueous solution combining the various chemical compounds necessary to synthesize ammonia) from a pressurized reaction vessel, expanding the volume to reduce the pressure, and then heating the mixture at the reduced pressure. The heating produces a gas, termed an off-gas, which is removed from the process stream, leaving behind a liquid mixture. In subsequent steps, the off-gas is partially condensed to an aqueous solution, and both the condensed and uncondensed portions of the off-gas are recycled to the reaction vessel for further use in the ongoing process.6 The process stream goes through further steps that are not here material.

Both Bongard and Mavrovic recognized that the efficiency of the process could be improved by minimizing the water content of the recycled off-gas condensate. They both seek to accomplish this end by reducing the water vapor content of the off-gas.

The Mavrovic patent calls for taking the process stream from the reaction vessel and reducing the pressure "under substantially adiabatic conditions." (Adiabatic means without addition or subtraction of heat.) This pressure reduction produces an off-gas having a low water vapor content. This off-gas is separated from the process stream. It is critical that this separation be accomplished without any heating of the process stream, either before or during separation.7 After the initial off-gas is separated, the process stream is heated and a second off-gas is produced in the prior art manner. The two off-gases are combined to form a single gas, which is then recycled.

In the district court, the parties disputed whether several points between the Bongard and Mavrovic processes were similar. See 401 F. Supp. at 391-94. We focus our inquiry on one of these points whether Bongard discloses the Mavrovic limitation that the initial separation must be accomplished prior to any heating.

The Bongard application summarizes its method for reducing the water content of the recycled aqueous solution as follows:

It has been found, in accordance with the invention, that (reduction of water content) can be realized in a simple way by omitting the direct heating to a higher temperature of the mixture to be expanded and the subsequent separation of the resulting liquid and gas phases, and instead leading this mixture after expansion and without preheating, through a preliminary separator and heating the resulting liquid phase, after this leading the gas-liquid mixture then formed through another liquid separator. It is found that in this case the water content of the gas phases is lower than with the old method, in which no preliminary separator is used.

The application refers to four schematic diagrams, known as "embodiments," of the Bongard invention. Figures 2, 3 and 4 are reproduced below:

NOTE: OPINION CONTAINS TABLE OR OTHER DATA THAT IS NOT VIEWABLEOur examination of these embodiments concerns only that part of the process occurring between elements one and four. Element one is the reaction vessel; element two is an expansion valve which accomplishes the process stream pressure reduction; element three is a heater. Elements four and forty are separators to allow for the separation of the off-gas from the process stream. The figure 3 element forty is described in Bongard's application as "(a) simple separatory (vessel) provided with a feed line for the liquid-gas mixture, arranged between a liquid discharge line at the base and a gas discharge line at the top of the vessel." The figure 4 element forty is described as "a liquid gas separator of a more complicated design" and is referred to as a "scrubber." It is not disputed and the district court found that separation in the scrubber is accomplished with heating. 401 F. Supp. at 393, n. 12.

That part of the figure 2 embodiment between elements one and four is a duplication of prior art.8 If separator forty in figure 3 is unheated, figure 3 duplicates the Mavrovic process. Since figure 4 involves a heated separation, it is materially distinguishable from the Mavrovic process, a fact recognized by the district court. 401 F. Supp. at 393.

Several aspects of the Bongard application do not expressly disclose elements of the Mavrovic process. Stamicarbon has thus relied on the law of inherent disclosure. This patent law doctrine holds that for process A to be held to duplicate process B, it is not necessary for process A to expressly disclose all the limitations of process B; rather, it is sufficient if the " necessary and only reasonable interpretation" of process A is that it duplicates process B. Dyer v. Field, 386 F.2d 466, 471, 55 CCPA 771 (1967). In a recent similar case where one party had copied counts from his opponent and relied on the doctrine of inherent disclosure, the court stated that the former had a "twofold burden":

First, one copying a claim from a patent for the purpose of instituting interference proceedings must show that his application clearly supports the count. * * * There must be no doubt that an application discloses each and every material limitation of the claims and all doubts must be resolved against the copier. * * * Second, where support must be based on an inherent disclosure, it is not sufficient that a person following the disclosure might obtain the result set forth in the counts; it must inevitably happen.

Tummers v. Kleimack, 455 F.2d 566, 569, 59 CCPA 846 (1972), quoting Dreyfus v. Sternau, 357 F.2d 411, 415, 53 CCPA 1050 (1966).

Chemico contends that the district court incorrectly applied the doctrine of inherent disclosure in deciding in favor of Stamicarbon. Chemico first argues that Bongard does not expressly disclose the critical limitation of the Mavrovic process that the initial separation be accomplished without heating. Chemico then notes that Bongard has disclosed three different inventive embodiments, only one of which is arguably consistent with Mavrovic. But, Chemico contends, figure 3 does not necessarily require a non-heated separation. Since figure 4 and language in the Bongard application indicate that the initial separation may be accomplished with heating, Chemico concludes that one skilled in urea synthesis would not inevitably duplicate the Mavrovic process by relying on the Bongard application.

Appellee Stamicarbon has responded that the only embodiment of the Bongard application at issue in this litigation is figure 3, which it claims duplicates the Mavrovic process. Stamicarbon concedes that figure 4 is inconsistent with the Mavrovic process but argues the inconsistency is irrelevant.

Although Chemico's contention was presented to the Patent Office, it was rejected without discussion.9 The district court, however, explicitly rejected this argument in its opinion. The court first concluded that the Mavrovic limitation that the separation be completed before heating the process stream is expressly disclosed by the Bongard application. The court relied for this conclusion on language in the Bongard application which requires leading the process stream "after expansion and without preheating, through a preliminary separator and heating the resulting liquid phase . . . ." 401 F. Supp. at 392-93 (emphasis added). Having concluded that the critical limitation was expressly disclosed, the district court nonetheless addressed the inherency issue. It found that figure 3 necessarily disclosed an unheated separator and that figure 4 introduced no ambiguity. 401 F. Supp. at 393. It is these conclusions that we must review under the strictures of Morgan v. Daniels, supra, and the clearly erroneous rule set out above.

Our starting point must be the district court's conclusion that Bongard expressly requires that the initial separation be accomplished without heating. If this conclusion is correct, the inherency argument becomes irrelevant.

The district court based its conclusion on Bongard's actual disclosure, which states:

. . . leading this mixture after expansion and without preheating, through a preliminary separator . . . .

It is apparent that the district court, in effect, read the word "preheating" to mean "heating." Such a reading is contrary to the natural meaning of the words. "Pre- heating" implies heating done before the next step, which next step is separation. A natural reading of the above quoted passage would take the language to preclude heating before separation. The passage simply does not speak to the presence or absence of heating during separation.

Reading the above passage to preclude heating before separation makes sense in light of prior art. The prior art processes on which both Bongard and Mavrovic sought to improve generally had a heater before the separator. Thus, in order to distinguish his invention from prior art, Bongard had to preclude heating before separation.10

Furthermore, to read the above-quoted passage as the district court did introduces an internal inconsistency into the Bongard application. The passage refers to the entire Bongard process, of which figure 4 is an embodiment. Bongard application p. 649. However, figure 4 shows heating during separation. If the passage is read to preclude heating during separation, it forbids what figure 4 discloses and commands. Inventors may not be linguistic artists in describing their creations, but it strains credulity to suggest that an obvious internal contradiction could pass unnoticed.

Evaluation of these factors leaves us "with the definite and firm conviction that a mistake has been committed." United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S. Ct. 525, 542, 92 L. Ed. 746 (1948). Accordingly, we hold that the district court's conclusion that Bongard expressly requires a non-heated initial separation is clearly erroneous. Stamicarbon thus must rely on the doctrine of inherent disclosure.

The law of inherency has been previously rehearsed. As those cases indicate, inherency "may not be established by probabilities or possibilities." Hansgirg v. Kemmer, 102 F.2d 212, 214 (Cust. & Pat.App.1939). It "is not sufficient that a person following the disclosure might obtain the result set forth in the count; it must invariably happen." Gubelmann v. Gang, 408 F.2d 758, 766 (Cust. & Pat.App.1969). Furthermore, in determining whether Bongard has disclosed the Mavrovic process, we may not, as Stamicarbon urges, focus only on figure 3. Rather, we must examine Bongard's disclosure "in its entirety," including the inconsistent figure 4. Dyer v. Field, 368 F.2d 466, 472, 55 CCPA 771 (1967). See McAninch v. O'Brien, 443 F.2d 1403, 1407, 58 CCPA 1333 (1971).

Bongard describes figures 3 and 4 as follows:

In Figures 2 and 3 the preliminary gas-liquid separators forty and seventy are indicated as simple separatory vessels provided with a feed line for the liquid-gas mixture, arranged between a liquid discharge line at the base and a gas discharge line at the top of the vessel. It is also possible, however, to apply a liquid gas separator of a more complicated design. A very satisfactory preliminary liquid-gas separator is a separator designed as a scrubber. The arrangement will then be as shown in Figure 4 where the reference numbers denote the same parts as in Figures 1, 2 and 3. (Emphasis supplied.)

The district court concluded that the "only reasonable" construction of this language is that the separator in figure 3 operates "without any heating . . . in contradistinction to the separator in Figure 4 wherein heat is applied to help effect separation." 401 F. Supp. at 393. The court apparently reasoned that since figure 4 shows heated separation, figure 3 must show non-heated separation or it would be merely duplicative.

We cannot accept this conclusion. The quoted language from the Bongard patent does not differentiate between heated and non-heated separators. Rather, it distinguishes between "simple separatory vessels" and those "of more complicated design." A "very satisfactory" example of the latter is one in which there is heating, as in figure 4, but there is no indication that the separators of a more complicated design necessarily include heating. The latter indication would be necessary to support the district court's conclusion. Furthermore, ample evidence in the record, including the cross-examination testimony of Stamicarbon's expert, indicates that a simple separator shown in prior art diagrams as in figure 3, was known in some prior art processes to operate with heating. Someone familiar with prior art could not determine from the above at best ambiguous passage of the Bongard application that the figure 3 separator must be operated without heating. Again, we must determine the district court's finding to be clearly erroneous.

Indeed, rather than mandating that figure 3 operates without heating, figure 4 and the language of the application indicate that the process can operate with heating. Except for the difference in separators, figures 3 and 4 are identical. By describing figure 3 and then saying "it is also possible . . . to apply a . . . separator of a more complicated design," one example of which is a heated separator shown in figure 4, Bongard implies that his process works equally well with heated or unheated separation.11 Heating thus appears to be of no consequence to the Bongard process and it is entirely possible that someone skilled in the art following the Bongard disclosure would use a heated separation.

Decisions of the Court of Customs and Patent Appeals reveal a similar analysis. In Dyer v. Field, 386 F.2d 466, 55 CCPA 771 (1967), the single interference count defined a bulky yarn, the critical requirement of which was that excess yarn was interwoven with and extended through a core yarn to form loops outside the core yarn. Field relied on a 1954 parent application to establish priority, and the sole issue on appeal was whether the parent application disclosed the counts. The parent application contained two embodiments diagramming the structure of the yarn; neither of these supported the counts. 386 F.2d at 469. For support of the count, Field relied on a third embodiment which diagrammed the yarn-making process; the description of the embodiment process required only that the yarn be distributed "throughout the process." The court disagreed with the Board's decision that this third embodiment description inherently supported the count:

We do not think that, merely because the core filaments are broadly disclosed by Field's 1954 specification to be distributed "throughout" the product, the excess yarn is necessarily interwoven with the core yarn or extends through it at a number of random locations as required by the count. While the quoted language of the 1954 specification could be construed as synonymous with the count limitations, it seems to us that, as Dyer points out, it is also consistent with a product in which the loops of the excess yarn are wrapped around the core yarn, rather than extending through it, at random points along its length. To be sure, the examiner relied upon the drawings of Field to support his construction of the count and specification language. But those drawings provide no support whatsoever for the examiner's interpretation of the Field disclosure, for they do not purport to illustrate a product in which excess yarn is interwoven with a core yarn, extending through it at random locations. Indeed, as observed earlier, Fig. 5 illustrates a product in which the excess yarn is wrapped around the core yarn, with no apparent opportunity for the core yarn to open to allow excess yarn filaments to pass through it, while Fig. 4 illustrates a product containing no separate core yarn at all.

Id. at 471. (emphasis in original.) Thus, even though Field's third embodiment could be read to support the counts, the court's examination of additional embodiments which were inconsistent with the counts led it to conclude that the Field parent application did not necessarily and inevitably support the counts.

Similarly in Noyce v. Kilby, 416 F.2d 1391 (Cust. & Pat.App.1969), the court refused to read an ambiguous embodiment as inherently disclosing a critical element of a count where there was an embodiment inconsistent with the count. In a comparable vein, the court in In re Dubbs, 77 F.2d 520 (Cust. & Pat.App.1935), affirmed a Patent Office decision rejecting an application for a patent, reasoning:

. . . in view of the fact that there is nothing in his specification to indicate that the claimed process would be carried out in the normal and usual operation of his apparatus, and in the absence of a disclosure of the steps called for in the appealed claims, the mere fact that the apparatus disclosed by him might permit the carrying out of such process does not warrant a holding that he is entitled to make such claims.

In light of the relevant case law and after careful examination of the entire Bongard application, we are thoroughly convinced, Morgan v. Daniels, supra, that both the district court and the Patent Office erred in awarding priority on the interference counts to Stamicarbon and Bongard.

The judgment of the district court will be reversed. Judgment will be entered awarding Chemico priority on both counts of the interference.

GIBBONS, Circuit Judge (dissenting).

This interference proceeding involves an alleged improvement in the process of synthesizing urea from carbon dioxide and ammonia. The prior art taught that when carbon dioxide and excess ammonia are subjected to high pressure in a reaction vessel a reversible reaction occurs in which is formed an intermediate product called ammonium carbamate. That intermediate product in turn decomposes into urea and water. The process stream comes from the high pressure reaction vessel as a mixture of excess ammonia, unconverted ammonium carbamate, urea and water. The end product of the process is a liquid mixture of urea and water which yields the desired urea. Before that end product is achieved the excess of ammonia and unconverted ammonium carbamate must be removed in the form of an off-gas which is recycled to the reaction vessel.

It is common to the Mavrovic and Bongard processes, and to the prior art as well, that the desideratum is to maximize the amount of water left in the urea stream and minimize the amount of water in the off-gas. The prior art taught that the decomposition of ammonium carbamate was accomplished by a simultaneous pressure reduction and heating. Both reduction in pressure and increase in temperature cause an increase in volume, and result in the decomposition of ammonium carbamate.1 But since the process stream leaving the reaction vessel also contains water, the addition of heat would tend to vaporize more water, thus increasing the water content of the off-gas to be recycled.2

The claimed invention is that of taking the off-gas in two stages; first when the ammonium carbamate decomposes solely as a result of a reduction in pressure; then at a later stage in the process when the remaining liquid mixture is again expanded by heating. As a result, when the two off-gas stages are recombined the recycled stream has less water than under the prior art process. This is apparently due to the fact that when the initial decomposition of ammonium carbamate by pressure reduction takes place, the process vaporizes less water than if the temperature of the process stream were increased at that stage.

The sole issue addressed by the majority is whether Bongard's Netherlands application filed in September, 1960, explicitly or inherently disclosed such an improvement over the prior art. The Board of Patent Interferences held that it did.3 The district court agreed.4 Both awarded priority to Bongard. Clearly both were right when they held that the Bongard application inherently disclosed that taking the off-gas from the stream in two stages, one of which was at the point of decomposition of ammonium carbamate resulting solely from pressure reduction rather than from heating, resulted in a recycle stream containing less water. This is the sole alleged invention in issue.

Mavrovic relies on two counts. In Count I the critical language reads:

". . . the improvement which comprises reducing the pressure of the urea synthesis effluent process stream under substantially adiabatic conditions thereby decomposing ammonium carbamate, separating a first off-gas having a low water content from the residual process stream."

As Judge Rosenn concedes in footnote 7, this count requires no more than separation of the first off-gas without heating. Decomposition of ammonium carbamate is accomplished in other words, solely as a result of pressure reduction, and the resulting gas is removed. Mavrovic's Count II also discloses separation of the off-gas described in Count I, without heating. Thus for present purposes both of Mavrovic's Counts can be treated as requiring separation of the first off-gas prior to heating the process stream after the stream has left the reaction vessel and passed through a pressure reduction valve.

It has been found, in accordance with the invention, that this (reduction of water content) can be realized in a simple way by omitting the direct heating to a higher temperature of the expanded mixture and the subsequent separation of the resulting liquid and gas phases, and instead leading this mixture after expansion and without preheating, through a preliminary separator and heating the resulting liquid phase, after this leading the gas-liquid mixture then formed through another liquid separator. It is found that in this case the water content of the gas phases is lower than with the old method, in which no preliminary separator is used.

The district court found this to be a disclosure of separation of the first off-gas prior to heating. Judge Rosenn, seizing on the single word "preheating" in the Bongard application, constructs a linguistic edifice totally divorced from the text of the Bongard disclosure, and totally divorced from the physical phenomena it describes. The heart of his argument is in a paragraph on page 651:

The district court based its conclusion on Bongard's actual disclosure, which states:

. . . leading this mixture after expansion and without preheating, through a preliminary separator . . . .

It is apparent that the district court, in effect, read the word "preheating" to mean "heating." Such a reading is contrary to the natural meaning of the words. "Pre heating" implies heating done before the next step, which next step is separation. A natural reading of the above quoted passage would take the language to preclude heating before separation. The passage simply does not speak to the presence or absence of heating during separation.

But the short quote is not Bongard's disclosure. Bongard explicitly disclosed:

(1) "omitting the direct heating to a higher temperature of the expanded mixture and the subsequent separation of the resulting liquid and gas phases";

(2) "instead (of expansion by heating as in the prior art) leading this (gas phase) mixture after expansion and without preheating to a preliminary separator";

(3) "heating the resulting liquid phase, after this leading the gas mixture then formed through another liquid separator";

(4) "the water content of the gas phases is lower than with the old method, in which no preliminary separator is used." (Emphasis supplied).

Judge Rosenn's artificial linguistic edifice depends upon his utter disregard of the plain language "omitting the direct heating to a higher temperature of the expanded mixture." It is understandable that the majority has difficulty comprehending the language of the Bongard disclosure, which is a translation into English from the original Dutch. Nonetheless, the awkwardness of the translation is no excuse for the majority's failure to pay attention to all the words of the disclosure. I think the source of the majority's error is in plainest relief at p. 652 n. 10, supra. Judge Rosenn asserts that " 'omitting the direct heating to a higher temperature,' does not expressly preclude heating during separation." But it does: the language of the disclosure is "omitting the direct heating to a higher temperature of the expanded mixture." (Emphasis added). The "expanded mixture" is what is in the preliminary separator, the "simple separatory vessel" of figure 3. It has passed from the reaction vessel, through an expansion valve, into the preliminary separator. At that point, it is the "expanded mixture," the direct heating of which is expressly omitted by the Bongard process. It was so found by the district court. That conclusion, far from being "clearly erroneous," is clearly correct. Moreover the majority opinion totally disregards the context of the disclosures in both the Mavrovic and Bongard applications. That context is the elimination of water vapor in the off-gas stream.

The Bongard application plainly discloses the elimination of heating in the expansion stage of the prior art. It plainly discloses use of a preliminary separator. It plainly discloses subsequent heating of the liquid stream to accomplish further degeneration of ammonium carbamate. It plainly discloses that the recombined gas streams result in a recycle stream lower in water content (the object of the invention). Despite these plain disclosures Judge Rosenn reads into the term "without preheating" a negative implication that heating does take place during preliminary separation.

The district court found that when Bongard referred to a "preliminary separator" in which the first off-gas was separated he referred to a vessel that did not use heat. In the context of the application a claimed invention for the reduction of water vapor in the off-gas no one who appreciates that heat boils water could read the disclosure otherwise.

If there were any room for doubt, and there is none, it would be removed by Bongard's explanation that the preliminary vessels shown in figure 3 of his application are:

". . . simple separatory vessels provided with a feed line for the liquid-gas mixture, arranged between a liquid discharge line at the base and a gas discharge line at the top of the vessel." 401 F. Supp. at 393, Trial Exhibit MX-3, p. 11.

The district court found that the necessary and only reasonable construction of the language of Bongard's application is that the separator in his figure 3 operates simply to allow whatever gas there is in the mixture to flow out the top of the separator and liquid to flow out the bottom, without any heating or other significant alteration. That is precisely the message, and the only message conveyed to me.

The majority refers to figure 4 of the Bongard application in support of its conclusion. Figure 4 of the Bongard process reveals the use of a "scrubber," or separator, which uses heat. However, that figure is not relevant to the count at issue in this appeal. The relevant disclosure is figure 3, using "simple separatory vessels" and it is irrelevant that Bongard concedes that alternatively "(i)t is also possible . . . to apply a liquid gas separator of a more complicated design," i.e., a heat-activated scrubber. The alternative possibility of using heat does not negate the fact that the basic Bongard disclosure omits heat. At the risk of being repetitious I emphasize that the Bongard application inherently disclosed 1) a urea synthesis process which effected a reduction of the water content of the recycle stream; and 2) accomplished this by means of an adiabatic process involving expansion and subsequent separation of the reaction vessel product without the addition of heat.

The district court did not err. One who read Bongard's 1960 application would know precisely as much about the method of elimination of water from the recycled gas stream of the urea process as would one who read Mavrovic's 1966 claims. The process of eliminating water from that stream is the sole invention. The majority has simply read out of the Bongard disclosure its express elimination of heat in the initial expansion and separation. The majority reads into figure 3 a heater which cannot, by any reasonable construction of that figure or of the text of the application, be found there. I do not understand what public purpose is served by such word games.

Appellee makes several other contentions to which Judge Rosenn refers. The majority does not reach them. I reject them for the reasons set forth in the opinion in the district court. I would affirm.