In defense of Rosen references

By Sarah Burstein, Professor of Law at Suffolk University Law School

LKQ Corporation v. GM Global Technology Operations LLC, No. 2021-2348 (submitted but not decided) (oral argument recording available here)

On December 5, 2022, the Federal Circuit heard oral arguments in two cases between LKQ and GM. In one of them, LKQ asked the court to rule that the current primary reference requirement for design patent obviousness, as stated in In re Rosen and Durling v. Spectrum Furniture is inconsistent with the Supreme Court’s decision in KSR v. Teleflex.

The panel did not seem convinced that it had the authority to overrule those prior panel decisions and at least one judge expressed doubt that KSR even applies to designs. But the issue is an important one and well worth discussing, even if the panel declines to address the issue in this case.

As I’ve explained previously:

Under Federal Circuit case law, “the ultimate inquiry under section 103 is whether the claimed design would have been obvious to a designer of ordinary skill who designs articles of the type involved.” To this end, the Federal Circuit has created a two-part test. First, the court must identify a primary reference—“a something in existence, the design characteristics of which are basically the same as the claimed design.” If there is a proper primary reference, then “other ‘secondary’ references may be used to modify it to create a design that has the same overall visual appearance as the claimed design.” To qualify as a proper secondary reference, a design must be “so related [to the primary reference] that the appearance of certain ornamental features in one would suggest the application of those features to the other.”

The primary reference requirement comes from the CCPA case In re Rosen. These references are, therefore, sometimes referred to as “Rosen references.” Durling is perhaps the clearest statement of the Federal Circuit’s current § 103 test for designs, which incorporates the Rosen requirement.

There is certainly much to criticize about the Federal Circuit’s approach to § 103 and designs. For example, the “so related” language has a long pedigree but its meaning has never been particularly clear. And these days, it seems like the Federal Circuit hardly ever reaches step 2 of the Durling test because it has applied such a high standard of visual similarity for the first step that it sometimes seems hard to see any blue sky between a design that looks “basically the same” (and is, therefore, a proper Rosen reference) and one that is “the same” (and thus anticipates the claimed design).

As currently applied, the Durling test—and in particular the Rosen reference requirement—seems to be just as rigid, if not more so, as the TSM approach the Supreme Court rejected in KSR. But does that mean the Rosen requirement is the problem? Or is the problem the way it is being applied?

As I wrote in my very first design patent article, “[t]he primary reference requirement performs the valuable function of focusing the nonobviousness inquiry on the design ‘as a whole.’ If a new design is so different from other products of its type that no primary reference can be found, that would generally be strong evidence of nonobviousness.” I continue to believe that, properly and flexibly applied, this general approach makes sense for designs and is consistent with KSR. (If you’d like to read more, you can find those arguments here.)

Unmooring the § 103 inquiry from this primary reference requirement would allow challengers to pick and choose visual elements from the prior art to combine into “Frankenart.” The Frankenart approach would devolve the § 103 inquiry into whether a designer technically could have created the claimed design based on the prior art, as opposed to (the more proper question of) whether it would have been visually obvious to do so.

Yes, it’s true that challengers can effectively use a Frankenart approach in most utility patent cases. But a design patent covers a different type of invention that is claimed in a different way. It makes perfect sense to apply the general command of § 103 in a different way when it comes to designs.

It’s also worth noting that while the primary reference requirement looks different from how we do § 103 in most utility patent cases, it’s not totally without cognates. It is at least reasonably analogous to the “lead compound” approach used in some chemical cases.

To be clear, this is not to say that the Federal Circuit’s current § 103 case law is great. It’s not. The cases apply the proper primary reference requirement too strictly. And it’s not at all clear how Durling step two does or should work. But that doesn’t mean we should abandon the primary reference framework entirely.



11 thoughts on “In defense of Rosen references

      1. 2.1.1

        Wouldn’t its holding be applicable to narrowing down what could be considered to be the primary reference?


          The holding in Surgisil is that for anticipation the 102 reference needs to be the same article of manufacture as the claimed design. To date, this has not bled over to whether a primary reference in a 103 situation needs to be the same article of manufacture as the claimed design. The answer seems to depend on what a designer of ordinary skill would look to when designing articles of the type claimed. Under 103, that can be analogous art, i.e., not necessarily the same article of manufacture as the claimed design. Keep in mind that under 102, it’s an ordinary observer making the judgment of substantial similarity. Even more of an interesting question is whether the “comparison” prior art in an infringement analysis (under Egyptian Goddess) needs to be of the same article of manufacture as the patented and accused design, keeping in mind that in design patent law the test for infringement is the same as the test for anticipation (however misguided that “rule” is). This issue is undecided, but is before the Federal Circuit now in Columbia v. Serius.


            The USPTO has never come up with an analysis for determining the level of ordinary skill in the art.
            Are there any USPTO or Judicially-created tests for determining the level of ordinary skill of a design or determining the level of ordinary observation of an observer? I imagine that at least for the latter, there might be a test that is similar to the “likelihood of confusion” from the trademark side of things.


              I am not aware of any such tests. Although the issue of POSITA is frequently litigated, it’s a question of fact. And honestly, it usually has little bearing on the outcome of a case, i.e., it’s given lip service and the trier of fact simply goes on to determine substantial similarity. On the other hand, determining who is “an ordinary observer, giving such attention as a purchaser usually gives” is sometimes outcome determinative, e.g., in the old Arminak case involving the design of a sprayer on a bottle used extensively at retail, the court said since the ordinary observer is a corporate buyer rather than a retail purchaser, there was no infringement. link to

  1. 1

    Looking forward to comparing Paul’s statements and the usual well respected Design patent law’s (usual) rebuttals.

    (Yes, I do mean Mr. Saidman)


          My comments have been that this unique non-statutory 1o3 test for design patents ought to be legally challenged by an affected party, and as this blog indicates, it now has been [if belatedly?].

Comments are closed.