Some Thoughts About Unexpected Results Jurisprudence

Mar 2004 – Article
86 J. Pat. & Trademark Off. Soc'y 169 (2004).

By Harris A. Pitlick[1]

I. An Overview

Evidence of unexpected results is no doubt the most prevalent form of evidence of nonobviousness relied on by patent applicants during patent examination.Indeed, there are hundreds of published cases wherein the sufficiency of such evidence to rebut a case of prima facie obviousness was in issue.

Whether such rebuttal evidence is sufficient is a finding of fact.[2]Therefore, it is difficult to fashion an objective standard by which a patent applicant can determine beforehand (short of an agreement with the patent examiner) whether its evidence will be sufficient.

Some elements of a standard are well-settled. The evidence asserted as unexpected must actually have been obtained.[3]The evidence must include a comparison with the closest prior art,[4] which prior art must actually exist,[5] although in some cases, an "indirect" comparison may suffice.[6]The evidence (as well as evidence of non-obviousness generally) must be commensurate in scope with the claims to which it pertains,[7] although as I have previously argued, such evidence must be commensurate in scope with only so much of the claimed subject matter (in terms of scope) that is prima facie obvious.[8] Attorney argument, unsupported by factual evidence, is insufficient.[9]Superiority of, or difference in, results, if not shown to be unexpected, is insufficient.[10]The difference in results should indicate some practical advantage.[11]Evidence of unexpected results need not be described in the specification,[12] nor must an advantage shown to be unexpected be described in the specification if it would inherently flow from a use that was described.[13]

Evidence of unexpected results has been dismissed as not statistically significant by a number of decisions of the Board of Patent Appeals and Interferences (board),[14] although the Court of Customs and Patent Appeals (CCPA) has indicated that where the patent applicant has not asserted statistical validity, the board may not dismiss the evidence simply because statistical validity has not been shown.[15]

II. A Tale of Two Cases

Two fairly recent cases decided by the United States Court of Appeals for the Federal Circuit (Federal Circuit) --In re Soni[16] and In re Geisler[17] -- demonstrate that unexpected results jurisprudence is still in a state of flux.In Soni, the evidence of unexpected results was found to be sufficient; in Geisler, it was not.

In Soni, the claimed invention was drawn to particular compositions comprising a particular type of polymer having a molecular weight greater than 150,000. The court succinctly set forth Soni's position as follows:

Soni's patent specification states that the claimed compositions have significantly improved physical and electrical properties compared to compositions using polymers having a molecular weight below 150,000. To illustrate this point, the specification describes a number of tests comparing the properties of a composition of the invention composed of polyethylene having a molecular weight of 203,000 with a comparative composition composed of polyethylene having a molecular weight of 148,000. The data show at least a fifty-fold increase in tensile strength for the higher molecular weight composition compared to the lower molecular weight composition. The data also show at least a five-fold increase in peel strength as well as improved resistivity and recovery behavior properties. From these data, the specification concludes that "[t]he tensile, peel, resistivity behavior and recovery tests show significantly improved properties for a polymer having a molecular [weight] of 203,000 compared to one having a molecular weight of 148,000, which are much greater than would have been predicted given the difference in their molecular weights."[18]

The issue before the court was whether Soni had shown unexpected results sufficient to rebut a prima facie case of obviousness, to which Soni admitted.[19]

The board had found that the record was devoid of any factual data to support the above- emphasized statement. The board concluded that while the data was some evidence of nonobviousness, it was insufficient to outweigh the evidence of obviousness.[20]

The court, in a 2-1 decision, reversed. Citing various precedent that conclusory statements in the specification regarding unexpected results, unsupported by objective evidence, are not sufficient, and that mere improvement in properties does not always suffice to show unexpected results, the court nevertheless found that Soni's specification was not limited to such statements; it contained specific data indicating improved properties, and that the improvement was substantial.[21]The court held:

[W]hen an applicant demonstrates substantially improved results, as Soni did here, and states that the results were unexpected, this should suffice to establish unexpected results in the absence of evidence to the contrary.[22]

The court found that the PTO had not provided any persuasive basis to question Soni's comparative data and assertion that the demonstrated results were unexpected.[23]

The dissent found the majority's holding contrary to the court's precedent[24] and suggested that it "may be inherently unworkable."[25]

Particularly, the dissent argued that "the majority overturns a well-settled facet of the law of rejections for obviousness by eliminating altogether one of the two requirements of a successful rebuttal case of unexpectedly improved results-namely, objective proof that the observed improvement was indeed unexpected."[26]The dissent found that

[n]either the specification nor any post-rejection submission contains objective evidence tending to establish either (1) a baseline of expected improvements against which to measure the observed improvements, or (2) the lack of any such baseline expectation in the relevant prior art, as a result of which all degrees of improvement would be unexpected. Without establishing a baseline or the unavailability of one, however, unexpectedness cannot be proved.[27]

The dissent agreed with the board's finding that there was insufficient factual evidence to support Soni's assertion of unexpected results, and that this finding was consistent with the court's precedent on what must be shown to demonstrate that improved results are actually unexpected.[28]The dissent accused the majority of excusing Soni from the requirement that an improvement's unexpectedness be demonstrated objectively.[29]

The dissent further criticized the majority for providing no guidance on how to decide in future cases whether unexpected results are substantial enough to rebut a prima facie case of obviousness.[30]The "substantiality" requirement, according to the dissent, will breed more litigation to the Federal Circuit when unexpected results are in issue.[31]

Finally, the dissent found the "in the absence of evidence to the contrary" element of the court's "substantially improved results" formulation perplexing, since it "eliminates the applicant's obligation to come forward with evidence, the obligation that was the heart of the established rule [and] it assumes that an unfair burden has been placed on the applicant."[32]

The dissent's conclusion is set forth in full below:

I question the desirability of the majority's new rule for assessing an applicant's rebuttal evidence of unexpectedness, according to which there need not be any objective evidence of unexpectedness other than the inventor's unsupported assertion that an artisan would not have expected so great an improvement in light of the changes made. Even if desirable, such a rule is certainly not the one our cases have established. Nor does the majority merely create an exception to the settled rule; it altogether abolishes the rule in favor of a "substantially improved results" standard that will often be met. Many cases will be decided differently in the future as a consequence.

Because of dramatic improvements here, the majority sets off on a dramatic departure from the law as it stood before this case, a departure I do not think it either explains or justifies. Nor will the new rule be easy to apply or predictable in its application. If the old rule is too harsh here, then at most a narrowly defined exception could perhaps be crafted. Instead, the majority upends settled law for all cases in all arts. Like the Board, I would follow the rule of cases such as Merck[33]and Lindner[34] and would thus affirm the Board's rejection of Soni's application.[35]

It is difficult to quibble with the dissent's analysis.Nothing in the precedent cited by the dissent or any other precedent on the issue that I could find appears to support the new standard annunciated by the majority.[36]To put it in other words, the majority's standard seems to work a balancing test--the greater the magnitude of the difference in results, the smaller is the burden to show that the difference is unexpected.

In Geisler, an issue was whether a recited thickness of 50-100 Angstroms in a protective layer of a claimed multi-layer reflective article rendered the article non-obvious over an analogous prior art article. The prior art protective layer thickness was disclosed as sufficient to provide the protection desired and still be colorless, for example, 100 to 600 Angstroms, and preferably 200 to 300 Angstroms, and in general, should not be less than about 100 Angstroms for suitable protection.

The specification included results of two tests for testing coated surfaces for wear resistance. Geisler tested protective layers of different thicknesses ranging from 900 Angstroms to 50 Angstroms. The court summed up the data as follows:

The first test was conducted shortly after the coating process was completed, and the second test was conducted "a few days later." The first test showed that, in general, the thinner protective layers exhibited better wear resistance. The second test showed the same trend, but in that test the thinner protective layers exhibited a somewhat greater and more consistent increase in wear resistance. For example, in the first test the 50 Angstrom layer exhibited about 20 percent better wear resistance than the 300 Angstrom layer, while in the second test the 50 Angstrom layer exhibited about twice as much wear resistance as the 300 Angstrom layer. After setting forth the two sets of test results, the application concluded that a protective layer of 50 Angstroms "provides a very good protection and, moreover, is very inexpensive in its production."[37]

The examiner held the claimed invention to have been obvious, based on the overlap of the respective ranges of the invention and prior art at 100 Angstroms.The examiner relied on precedent holding that overlapping ranges are prima facie obvious. The court's opinion does not indicate what the examiner's specific findings, if any, were on the test results.

The board affirmed. The court characterized the board's decision as follows:

The Board's decision that Geisler failed to rebut the prima facie case of obviousness is based on two findings: (1) that Geisler did not satisfy his burden of proving that the claimed protective layer produced unexpected results in the claimed thickness range of 50 to 100 Angstroms; and (2) that Geisler failed to show that the prior art of record . . . taught away from the claimed invention.[38]

The court affirmed.

According to the court, Geisler conceded that the examiner was correct in finding that the claims were prima facie obvious, because the claimed thickness range of 50 to 100 Angstroms for the protective layer overlapped at its end point with the thickness range of 100 to 600 Angstroms disclosed by the prior art.[39]

The test results in Geisler's specification were not accompanied with any description that they were unexpected, nor was an affidavit or declaration subsequently submitted during prosecution on the issue.Geisler instead asserted common sense, i.e., the thicker the protective coating, the greater the expected protection.The court dismissed this assertion as "just attorney argument."[40]

In response to Geisler's argument that their test results were contrary to assumptions in the prior art that a thickness less than 100 Angstroms would not be sufficient, the court found that "[s]uch an assumption is contrary to Geisler's evidence that a coating of 50 Angstoms thick offers better wear resistance than a coating 100 Angstoms thick[.]"[41]

Noting the age of the applied prior art, the court continued:

But the fact that one group of inventors 15 years earlier may have made an assumption that was contrary to Geisler's test results does not prove that Geisler's results would be regarded as unexpected by one of ordinary skill in the art at the time of Geisler's application.[42]

The court then distinguished Soni.While the court acknowledged Geisler's 26% greater wear resistance from 100 Angstrom thickness to 50 Angstrom thickness, it found that this did not compare with the "substantially improved results" of a 50-fold improvement in tensile strength shown in Soni.[43]The court further distinguished Soni by noting the lack of any statement by Geisler about unexpected results, either in the specification or through other evidentiary submissions, other than the above-discussed attorney argument about common sense.[44]The court found that Geisler did not satisfy either of the Soni prerequisites, which are:

[W]hen an applicant demonstrates substantially improved results, as Soni did here, and states that the results were unexpected, this should suffice to establish unexpected results in the absence of evidence to the contrary.[45]

In rejecting Geisler's argument that the prior art taught against their 50-100 Angstrom range, the court focused on the prior art disclosure of "about" 100 Angstroms as providing "suitable protection," and found that the prior art both "suggests" benefits to be derived from keeping the layer as thin as possible, and provides motivation to "explore thickness levels below that range" [i.e., "about 100 Angstroms."][46]

Geisler is troubling on a number of levels, as now discussed.

Since the finding of prima facie obviousness was based on the overlap of a single point, i.e., 100 Angstroms, it is only at that thickness (and perhaps some thickness a little less than 100 Angstroms, given the court's finding of "about" 100 Angstroms[47]) that Geisler should have had the burden of showing unexpected results.[48]But since the court found that Geisler did not show unexpected results for the entire claimed thickness range of 50 to 100 Angstroms, it must be assumed that the court thought Geisler had the burden for this entire range.

The court's dismissal of Geisler's "common sense" argument as "just attorney argument" seems rather harsh. Geisler's argument seems to make good sense, and would seem to merit at least some substantive rejoinder as to why Geisler's results were not necessarily unexpected. Would the Federal Circuit dismiss as "just attorney argument" one that asserts that an invention that make the sun rise in the west and set in the east is an unexpected result? Indeed, Geisler's "common sense" argument would appear to have greater probative value than Soni's unsupported assertion that their improvement was much greater than would have been predicted. Instead of dogmatically in every case dismissing statements of counsel as not probative, one would hope that examiners and the PTO reviewing tribunals would be able to distinguish mere puffery which, I believe, is the type of attorney argument meant to be ignored, from observations that anyone would understand are truly unexpected.

Even more troubling is the court's finding, in response to Geisler's argument that their test results were contrary to assumptions in the prior art that a thickness less than 100 Angstroms would not be sufficient, that "[s]uch an assumption is contrary to Geisler's evidence that a coating of 50 Angstoms thick offers better wear resistance than a coating 100 Angstoms thick[.]"[49]The court thus improperly used Geisler's evidence against them, treating the evidence as if it were in the prior art.[50]

By commenting on the 15-year age of the applied prior art, the court raised a potential problem for all future proponents of unexpected results evidence.The law is clear that the pertinent time frame for determining whether unexpected results has been shown is the time the claimed invention was made. But the practical result of the court's comment is that the older the prior art, the more likely a showing of at least superior results will be rejected.How does one show what was expected at the time the invention was made when evidence relied on as the closest prior art is not close in time to the claimed invention?[51]While the court noted what Geisler lacked in terms of evidence of unexpected results, a supporting opinion affidavit or declaration by one of ordinary skill in the art would appear to be a minimum requirement.

The court's comparison of the amount of improvement in Geisler compared to that in Soni is further troubling, and substantiates the prophetic nature of theSoni dissent. Soni raised the bar so high on what would be regarded as substantial improvement that the court felt compelled to find Geisler's 26% improvement, although not in so many words, insubstantial.How many practitioners out there have seen a 50-fold improvement over the closest prior art?

Finally, the court's comment about motivation to "explore" thicknesses below 100 Angstroms, even if there were no disclosure that the thickness should not be less than about 100 Angstroms, sounds a lot like "obvious to try," which, as the court has stated on numerous occasions, is insufficient to make out a case of legal obviousness.[52]

What is one to do? Surely, a legitimate explanation by a person of ordinary skill in the art, such as the inventor, of why results are unexpected, either as part of the specification as filed or during patent prosecution, should suffice to overcome the negative dicta, if not the holding, in Geisler, even if the substantiality of the difference in results is questionable.

III.Some Suggestions

The remainder of this article focuses on one of the above-discussed well-settled elements

--that relating to the closest prior art as the necessary object of comparison.A sports analogy may be apt.

Closest prior art may be viewed as the admission ticket that must be presented before one can even get in the game. If comparison is not made to the closest prior art, any unexpected results evidence will be ineffective.

To continue the analogy, it is also generally assumed in many sports that the home team has an advantage over the visiting team. Thus, a patent applicant would want the game played on its turf, viz., the closest prior art is what the patent applicant decides, not what the examiner decides.

In many cases, the closest prior art, in terms of real world experience of persons of ordinary skill in the relevant art, is the prior art whose problems are being addressed by the invention. That prior art may be embodied in a patent or printed publication, or it may be otherwise publicly known. The application on such an invention may even have comparative data in the specification between the claimed invention and this prior art, showing results for the former to be better, if not unexpected, compared to results for the latter. If the inventor, or his patent attorney, has followed the case precedent, the specification would also explain why the results are unexpected, if so. Alternatively, if not part of the original disclosure, the evidence may subsequently be submitted in an affidavit or declaration.But this may be an exercise in futility if the examiner decides that other prior art is closer.

How is the closest prior art determined? As stated in In re Merchant:

Given the enormous variety of technologies and claimed subject matter, no all-encompassing principle or test can be delineated for determining the closest prior art. However, an almost self-evident guideline would appear effective in most cases. A comparison of the claimed invention with the disclosure of each cited reference to determine the number of claim limitations in common with each reference, bearing in mind the relative importance of particular limitations, will usually yield the closest single prior art reference.[53]

I think the more significant aspect of the Merchant test is its qualification about the inability to delineate an all-encompassing principle or test, as opposed to its "almost self-evident guideline" in "most" cases. Indeed, a guideline is simply that.An imaginative approach is thus not foreclosed.

Suggestion 1: Consider challenging the examiner's finding of closest prior art.

It is already well-settled that the prior art from which the closest prior art is chosen is not necessarily limited to prior art relied on by the examiner.As stated in In re Holladay:

While in most cases the closest prior art will be the art relied on by the examiner, this is not necessarily so. It is conceivable that two or more pieces of prior art could be equally close to the invention, and yet only one of them applied against the claims by the examiner.

Where two pieces of prior art are in fact equally close to the claimed invention, there is no logical reason for requiring an applicant to make a comparison with one instead of the other. On the other hand, practical considerations favor allowing the applicant to choose between them. Prior art devices are sometimes unavailable for testing. They may be disclosed in "paper patents" on inventions which have never been reduced to practice. Where the applicant uncovers a piece of prior art actually used in the real world and establishes that its teachings are equal to the relevant disclosure in a "paper patent" relied upon in the examiner's rejection, it would be unfair not to permit the applicant to make his comparison with the commercially used piece of prior art.[54]

The force of Holladay is somewhat limited, because the prior art in question relied on by the applicant was found to be equal in closeness to that relied on by the examiner. Indeed, highlighting the above-emphasized term from Holladay, the CCPA in In re Johnson[55] subsequently explained that Holladay did not give an applicant license to choose between equally close pieces of prior art where the teachings were not equal. "Thus Holladay is premised on the notion that the teachings of the closest prior art references are sufficiently similar so that the testing of one compound showing unexpected results would provide the same information as to the relevant teachings of the other equally close references."[56]

Despite its limiting facts, Holladay nevertheless stands for the proposition that the examiner is not the sole arbiter of what is the closest prior art.

Suggestion 2: Consider presenting evidence on the issue of closest prior art.

Whether the issue of what is the closest prior art is a question of fact or a question of law is one that has not yet been decided by the Federal Circuit, although the Federal Circuit has stated the proposition in a number of non-precedential decisions that a finding by the board that an applicant did not compare to the closest prior art is a question of fact.[57]There does not appear to be any reason why the issue would not be a question of fact. If a question of fact, then an affidavit or a declaration asserting that particular prior art is the closest prior art, with suitable explanatory support, should be given some weight.[58]Thus, faced with a disagreement with an examiner of what the closest prior art is, such a declaration or affidavit may sway the examiner in your favor.

Such a declaration can be a vehicle by which facts can be introduced to avoid a finding of closest prior art that relies on a simple tally of most limitations or elements in common. These facts can include a comparison of the problems addressed by the invention and problems addressed by the prior art. Prior art addressing the same problems as the invention can in many, if not most, cases be considered to be closer prior art than prior art that may coincidentally have more elements in common with the invention but which is addressing a different problem.[59]

The statement "bearing in mind the relative importance of particular limitations" in the Merchant test would have greater meaning if the problem being addressed by the invention can be considered an implicit, and relatively important, limitation for this limited purpose. Indeed, as part of the consideration of the invention as a whole, required by 35 USC 103, the problem addressed must be considered.Simply counting and comparing explicit claim limitations, without more, would appear to fall short of the statutory mandate.

Suggestion 3: Consider the question of nonanalogous art.

In furtherance of the notion that the problem addressed by the invention must be considered in determining the closest prior art, the possibility that prior art relied on by the examiner as closest, say by virtue of number of claim limitations in common only, is nonanalogous prior art, should not be dismissed.The test for analogous art has been stated in In re Wood, as follows:

In resolving the question of obviousness under 35 USC 103, we presume full knowledge by the inventor of all the prior art in the field of his endeavor. However, with regard to prior art outside the field of his endeavor, we only presume knowledge from those arts reasonably pertinent to the particular problem with which the inventor was involved. . . . The rationale behind this rule precluding rejections based on combination of teachings of references from nonanalogous arts is the realization that an inventor could not possibly be aware of every teaching in every art. Thus, we attempt to more closely approximate the reality of the circumstances surrounding the making of an invention by only presuming knowledge by the inventor of prior art in the field of his endeavor and in analogous arts.

The determination that a reference is from a nonanalogous art is therefore twofold. First, we decide if the reference is within the field of the inventor's endeavor. If it is not, we proceed to determine whether the reference is reasonably pertinent to the particular problem with which the inventor was involved.[60]

Where the prior art asserted as closest is not reasonably pertinent to the particular problem with which the inventor was involved, focus should be on whether it is from the field of the inventor's endeavor. Whether prior art is nonanalogous is a question of fact.[61]Thus, an affidavit or a declaration explaining why one of ordinary skill in the art should not be presumed to have any knowledge of the prior art asserted as closest should be given weight.To that end, it is obviously in the applicant's interest to give the "field of endeavor" element a relatively narrow breadth or at least narrow enough to be outside the field of endeavor of the asserted closest prior art.[62]The examiner, having (presumably) found this p