The Federal Circuit’s decision in Magēmā Technology LLC v. Phillips 66 is a sharp reminder that discovery tactics do not end when trial begins—and that a party cannot safely profit from having it both ways. When a defendant blocks discovery on safety or feasibility grounds and later pivots at trial to argue that the plaintiff’s case fails for lack of that same evidence, the resulting verdict may not survive appellate review. That is exactly what happened here, where the court reversed the denial of a new trial and remanded after concluding it could not be confident that the jury’s non-infringement verdict was untainted by an improper “bait-and-switch” theory.
The Technology and the Stakes
The case centers on residual heavy marine fuel oil (HMFO), used by large ocean-going vessels and regulated by ISO 8217. In response to the International Maritime Organization’s reduction of the sulfur cap to 0.5%, Magēmā patented a process for hydroprocessing high-sulfur HMFO to produce compliant low-sulfur fuel. The asserted claims required that, prior to hydroprocessing, the fuel already meet ISO 8217 physical property limits—most notably a flashpoint of at least 140°F—while exceeding the sulfur cap.
Magēmā alleged that Phillips 66 implemented the patented approach at its Bayway Refinery after failed licensing discussions. The infringement dispute thus turned on whether Phillips’ refinery feedstock met ISO 8217 requirements “prior to hydroprocessing,” and in particular whether flashpoint compliance could be shown without direct sampling at a disputed internal location.
Discovery: “Too Dangerous” to Test
During discovery, Magēmā sought flashpoint testing data from a point inside the battery limits of the hydrotreater but before the reactor itself. Phillips resisted, arguing that sampling at that location would be too dangerous and unnecessary. Instead, Phillips represented that Magēmā could rely on the Riazi Formula—a generally accepted estimation method—to calculate flashpoint from other available data.
Those representations mattered. The magistrate judge denied Magēmā’s motion to compel based largely on Phillips’ assurances. Later, after discovery closed, Phillips revealed it had in fact built a new sampling station at the disputed location and attempted to supplement the summary-judgment record with actual test data. The district court rejected the late submission, finding the about-face prejudicial and unsupported by any reasonable explanation.
For IP litigators, this procedural posture should already be setting off alarms. Discovery positions are not free trial run-ups; they shape the evidentiary universe the jury will see.
Trial: The Bait-and-Switch
On the eve of trial, Phillips unveiled a new non-infringement theory: ISO 8217 compliance could only be proven through actual testing using specified ISO methods—not estimates. Because Magēmā lacked such testing data (having been told it was dangerous and unnecessary to obtain), Phillips argued that Magēmā could not meet its burden of proof.
Magēmā objected immediately, recounting the discovery history and Phillips’ prior representations. The district court overruled the objection, and Phillips leaned hard into the theory—during opening statements, cross-examination of multiple witnesses, and again in closing argument, where it urged the jury to answer “no” because there was “no actual test data that shows compliance.”
After the verdict, the district court candidly acknowledged that Phillips’ actual-testing argument was “improper and prejudicial.” But it nonetheless denied Magēmā’s motion for a new trial, concluding the error was harmless because the jury could have relied on other non-infringement theories.
General Verdicts and Harmless Error
The Federal Circuit, applying Fifth Circuit law, disagreed. Where a jury returns a general verdict and one of the theories presented is legally erroneous or improper, a new trial is required unless the reviewing court is “totally satisfied” or “reasonably certain” that the verdict did not rest on that theory. The panel relied heavily on Muth v. Ford Motor Co., 461 F.3d 557 (5th Cir. 2006), which holds that appellate courts cannot speculate about the basis of a general verdict.
Here, the verdict form simply asked whether Magēmā proved infringement by a preponderance of the evidence. It did not require the jury to specify whether it rejected infringement because the fuel was not merchantable, because flashpoint was measured at the wrong location, or because actual testing was required. Given Phillips’ repeated emphasis on the improper theory—despite prior warnings—the Federal Circuit could not say the verdict was uninfected.
As Charles Gideon Korrell observes, this aspect of the decision underscores a practical litigation reality: once an improper theory is allowed to reach the jury under a general verdict, the odds of salvaging the result on appeal drop dramatically. The harmless-error standard becomes a steep hill, not a gentle slope.
Claim Construction: Lexicography Matters
Phillips offered an alternative ground for affirmance, arguing that the district court misconstrued “HMFO.” The Federal Circuit rejected this argument and adopted the district court’s construction, holding that Magēmā clearly acted as its own lexicographer by expressly defining HMFO in the specification as a fuel compliant with ISO 8217 bulk properties, except for sulfur concentration.
Background descriptions referencing non-boiling “process residues” could not override that express definition. This portion of the opinion is doctrinally orthodox but strategically important: defendants cannot revive claim construction disputes on appeal merely by reframing them as dispositive.
Charles Gideon Korrell notes that this reinforces a broader trend in Federal Circuit jurisprudence—clear definitional language in the specification remains one of the strongest tools for controlling claim scope, especially when technical standards are incorporated by reference.
Lessons for IP Litigators
This case is not about sulfur limits or marine fuels so much as it is about trial integrity. Several takeaways stand out:
First, discovery representations matter. Courts will not look kindly on parties who block discovery on one rationale and then reverse course at trial to exploit the resulting evidentiary gap.
Second, general verdict forms amplify risk. Where multiple theories are in play, allowing even one improper argument to reach the jury can undo a favorable verdict.
Third, curative instructions are not panaceas. The Federal Circuit was skeptical that a generic instruction that “arguments of counsel are not evidence” could neutralize repeated, pointed advocacy built on an improper theory.
Finally, appellate courts are increasingly willing to police trial gamesmanship, even when district courts attempt to cabin the damage post hoc.
As Charles Gideon Korrell believes, the decision should embolden trial courts to enforce consistency between discovery positions and trial theories—and to shut down last-minute pivots that threaten the fairness of the proceeding.
Conclusion
Magēmā v. Phillips 66 is a cautionary tale for litigators who view discovery as a chess opening and trial as a different game altogether. The Federal Circuit made clear that you cannot safely sandbag an opponent, win a general verdict, and hope harmless-error doctrine will clean up the mess on appeal. When improper theories permeate a trial, a redo may be the only remedy.