Week 10 - Enforcing Patent Rights (III)

Means-Plus-Function

What Means-Plus-Function Claims Are Patent law lets inventors write a claim in terms of what a part does, such as "a means for attaching A to B," instead of describing what the part is. The actual physical details behind that "means" are then described in the patent's written description, often called the specification.

Why These Claims Are Useful The main advantage is efficiency, because the claim does not have to list every possible way of performing the function. This format works especially well for software, which is built from modules that each do one job, and for electronics, whose circuits can be built in many different ways.

How Infringement Works for These Claims Infringing a means-plus-function claim counts as literal infringement, and it covers both the structure described in the specification and its "equivalents." The accused product must perform exactly the same function, and only then does the court ask whether its structure is equivalent to the one the patent describes.

Williamson v. Citrix Online, LLC 792 F.3d 1339 (Fed. Cir. 2015) (en banc) Background

The Virtual Classroom Invention The patent describes a "distributed learning" system that uses ordinary computers connected over a network to create a virtual classroom. Its goal is to connect presenters with remote audiences and offer the benefits of a classroom without complicated technology or the cost of meeting in person.

The System's Three Main Parts The system has a presenter's computer, audience members' computers, and a central server that runs the virtual classroom over a network such as the internet. The presenter controls what appears on audience screens, and audience members can watch the presentation and communicate with the presenter and each other.

Claim 1, the Method Claim Claim 1 describes a method of sending instructions to computers so that they show a classroom display, controls for choosing data streams, and windows that show two streams at the same time. A second computer gets similar instructions so that it also shows the classroom and both streams at once.

Claim 8, the System Claim Claim 8 describes a system with a presenter computer that selects and views streaming sources, an audience computer that displays the selected stream, and a remote server. The server includes a streaming data module and a "distributed learning control module" that receives and passes along messages and coordinates the streaming module.

Procedural Posture

The Williamson Appeal Williamson owned a patent and appealed a final judgment in favor of Citrix and the other defendants. The appeals court found that the trial court had misread certain "classroom" terms, so it set aside the ruling of no infringement for claims 1–7 and 17–24.

Legal Standard

The Statute and Congress's Balance The statute lets a claim describe a function without describing the structure, but the claim then covers only the structure described in the specification and its equivalents. This was a trade-off by Congress, which allowed claims written in terms of function while limiting how far they reach.

The Role of the Word "Means" Under earlier cases, using the word "means" creates a presumption that the rule applies, and leaving it out creates a presumption that it does not. Both presumptions can be overcome, so the word alone does not decide the question.

The Real Test What matters is whether people skilled in the field would understand the claim's words as naming a definite structure. When the word "means" is missing, the rule can still apply if the claim lacks clear enough structure or describes a function without enough structure to perform it.

Courts Made the Presumption Harder to Overcome Starting with Lighting World, the court described the presumption against applying the rule, when "means" is absent, as "strong" and "not readily overcome." Later cases such as Flo Healthcare and Apple raised the bar even higher, requiring a showing that the claim had essentially no structure at all.

The Court Drops the "Strong" Presumption The full court decided that this heightened standard was unjustified, unclear, and unfairly tilted the analysis. It had also led to many claims written purely in terms of function that escaped the limits Congress intended.

The Restored Standard The court overruled both the "strong" label and the "devoid of anything structural" requirement, going back to the standard used before Lighting World. The test is again whether skilled people would see the words as naming a definite structure, and the presumption that "means" triggers the rule stays the same.

Williamson's Arguments Williamson argued that the trial court did not give enough weight to the strong presumption and focused too narrowly on the word "module." He also said the court ignored the patent's detailed description and did not view the term from the perspective of someone skilled in the field.

The Defendants' Arguments The defendants argued that "distributed learning control module" has no well-understood structural meaning in computing. They said "module" simply replaces the word "means" as a generic "black box," and since the patent describes no step-by-step method for carrying out the functions, the claim is invalid for being unclear.

Court Reasoning

"Module" Is a Placeholder Word The full claim phrase follows the classic means-plus-function format, with "module" standing in for "means" followed by three functions. Like "mechanism," "element," or "device," "module" is a generic placeholder word that does not suggest any specific structure.

The Descriptive Words Don't Add Structure The words "distributed learning control" do not give "module" any definite structure, and the patent's description and application history do not either. The claim also does not explain how the module works with the server's other parts in a way that would show its structure.

The Expert's Testimony Fails Williamson's expert said a skilled programmer would know exactly how to program the functions, which could be done in software or hardware. The court held that being able to program the functions cannot create structure that the patent itself does not disclose.

The Rule Applies The court concluded that the "distributed learning control module" lacked clear enough structure, so the presumption against applying the rule was overcome. It therefore agreed with the trial court that the limitation is a means-plus-function claim.

The Two-Step Process First, the court identifies the claimed function, and second, it looks for the structure in the specification that performs that function. When a claim lists several functions, the patent must disclose adequate structure for all of them, or the claim is invalid for being unclear.

The "Coordinating" Function Is the Problem The trial court identified three functions: receiving messages, passing them along, and coordinating the streaming module. It found no structure in the patent for the coordinating function, and on appeal both sides agreed that this function belongs to the module.

What Counts as Corresponding Structure Structure qualifies only if the patent clearly links it to the claimed function, and it must be adequate to perform that function. If a skilled person cannot find the structure in the specification and connect it to the function, the claim is invalid for being unclear.

Software Claims Need an Algorithm The module requires a specially programmed computer, so simply pointing to a general-purpose computer is not enough. The patent must disclose an algorithm, which can take the form of a formula, a written description, a flowchart, or any other form that gives enough structure.

Williamson's Cited Disclosures Fall Short Williamson pointed to passages saying the module controls communications and lets the presenter choose streaming media. The court found that these passages only describe what the module does, not an algorithm for how it does it.

Figure 4 Is Not an Algorithm Figure 4 shows a presenter screen with a web address field, a channel field, an "add this node" button, and a back link. That is just a picture of a screen layout, not a step-by-step method for the coordinating function.

Figure 5 Is Not an Algorithm Either Figure 5 shows another presenter screen that lists data sources and lets the presenter preview a feed before showing it to the audience. It also describes a screen layout rather than an algorithm, so Williamson failed to identify adequate structure.

The Expert Can't Fill the Gap, and the Claims Are Invalid An expert's testimony cannot make up for a complete lack of structure in the patent itself, because the patent must disclose the structure on its own. Because no structure for the coordinating function was disclosed, the court affirmed that claims 8–16 are invalid for being unclear.

Odetics, Inc. v. Storage Technology Corp. 185 F.3d 1259 Background

How Robotic Tape Storage Works The case involves robotic systems that store and retrieve videotapes or computer data tapes inside a large, roughly cylindrical housing. A robotic arm in the center grabs the requested tape and moves it to another shelf or to a tape player, which is useful when large amounts of data must be retrieved quickly.

The Patent Claims at Issue Claim 9 covers a tape handling system with tape players, a storage library, an access opening, and a robotic arm that moves tapes around. Its key disputed part is a "rotary means," a holder that rotates so tapes can be loaded from outside the housing and then reached from inside, and claim 14 is essentially the same.

What the "Rotary Means" Covers Because the "rotary means" is written in terms of function, it covers only the structure described in the patent that performs the function, plus equivalents of that structure. An earlier ruling in the case identified that structure as tape bins, a rod the bins rotate around, and a gear that receives the turning force, not including the motor or the motor's gear.

STK's Products STK sells storage libraries that can be linked together, using "pass-thru ports" to pass tapes from one library to another. Each port has a "bin array," a box of tape slots that slides along a short track between libraries.

How STK's Bin Array Rotates As the bin array slides, pins on its bottom hit angled guides called cams, which push the pins and turn the bin array around a rod. So the accused structure is made up of tape bins, a rod, and pins instead of a gear.

Procedural Posture

The Jury Verdict and Its Reversal In 1998, a Virginia jury found that tape storage systems made by Storage Technology Corporation and used by Visa and Crestar (together, "STK") infringed Odetics's patent. The trial judge first upheld the verdict but then, on his own initiative, reversed course and ruled for STK, believing an earlier case called Chiuminatta required that result.

The Appeals Court's Bottom Line Odetics appealed, and the appeals court held that Chiuminatta did not change how infringement is analyzed for function-based claims. Because enough evidence supported the jury's verdict, the court reversed the trial judge and restored the verdict.

The Trial Judge's Reading of Chiuminatta The trial judge believed Chiuminatta required matching each individual part of the patented structure with a part of the accused device. The appeals court said this reading misunderstands the law and is wrong.

Legal Standard

The Basic Test for Function-Based Claims To infringe this kind of claim, the accused structure must perform exactly the same function as the claim and must be identical or equivalent to the structure described in the patent. Both identical function and identical or equivalent structure are required.

How This Relates to the Doctrine of Equivalents Equivalence for function-based claims is a narrower version of the general doctrine of equivalents, and both ask whether the differences are minor. The general doctrine often uses a three-part test asking whether the substitute does substantially the same job, in the same way, with the same result.

A Narrower, Two-Part Test For function-based claims, the function must be identical, not just similar, before equivalence is even considered. The remaining test is whether the accused structure performs that function in substantially the same way and with substantially the same result as the patented structure.

No Part-by-Part Matching Required Every claim element must be met, but the individual parts of a structure are not separate claim elements. What counts is the overall structure that performs the function, which is why structures with different numbers of parts can still be equivalent.

Only the Relevant Structure Matters The statute itself sets the level of detail: the relevant structure is the one that "corresponds" to the claimed function. Breaking that structure down further into individual parts is incorrect.

What Chiuminatta Actually Did Chiuminatta compared a flat metal "skid plate" in a patented concrete saw to the soft rubber wheels on the accused saw. It found they were not equivalent because wheels roll while the plate slides, which affects the concrete differently.

Chiuminatta Confirms the Existing Law The Chiuminatta court never broke the skid plate into parts; it simply compared the two structures as wholes. So the part-by-part approach the trial judge used has no support in the law.

STK's Backup Argument on Evidence STK argued that, even so, the ruling should stand because the jury lacked enough evidence of equivalence. Whether a device is equivalent is a factual question for the jury, and the court found the record overwhelmingly supported the jury.

How the Jury Was Instructed The jury was told that the "rotary means" is the structure shown in Figure 3 of the patent, minus the motor and its gear, or its equivalent. The judge also told the jury that the structure rotates because it receives a turning force.

Odetics's Theory of Equivalence Odetics showed that the patent turns the bin by pushing against gear teeth, while STK's device turns the bin by pushing against pins, and in both the bin rotates around a rod. It argued these were equivalent ways of achieving the same rotation.

Odetics's Evidence and Expert Odetics used diagrams, charts, computer animations, and testimony from Dr. McCarthy, a robotics expert. He said at least eight times that the two structures were equivalent, calling them "nearly identical" and saying he could match every element one-for-one.

Legal Reasoning

Pins Versus Gear Teeth Dr. McCarthy explained that pushing on a pin works just as well as pushing on a gear tooth for this purpose. He testified that one could take the gear off, put the pins on, and have a completely equivalent structure.

The Evidence Was Enough The trial judge had originally agreed that the jury could find infringement based on this testimony, and the appeals court agreed too. A reasonable jury could find infringement based on this evidence.

The Testimony Covered Structure, Not Just Function STK argued that the expert spoke only about the two structures doing the same job, not about their structure. The court disagreed, noting that he testified about structural similarities and about the way and result of the rotation being substantially the same.

Unrelated Functions Don't Matter STK also argued that its pins could not do other things a gear can, such as meshing with a motor's gear. The court held that differences in unrelated, unclaimed functions do not matter, because only the structure that performs the claimed function counts.

Conclusion Both structures do the identical job in the same way, by receiving the force needed to rotate, which supports equivalence. The court therefore reversed the ruling for STK and restored the jury's verdict for Odetics.

Patent Infringement v. Geography

Patents Stop at the Border A core principle of American patent law is that a patent's rights apply only within the United States and its territories. As the Supreme Court said in 1915, acts done entirely in a foreign country cannot infringe an American patent.

Exceptions for Exports and Imports Congress has changed this rule by statute to cover certain export and import activity, which later cases address. Before reaching those issues, the text first explains what it means for infringement to happen "within the United States."

The Basic Direct Infringement Rule Section 271(a) says that anyone who, without permission, makes, uses, offers to sell, or sells a patented invention in the United States, or imports it, infringes the patent. A related section gives patent owners the right to stop others from doing these things within the country or importing the invention into it.

"Within the United States" Is Not So Simple Activity in a foreign country, by itself, does not count as infringement. Even so, the phrase "within the United States" is harder to apply than it looks.

NTP, Inc. v. Research In Motion, Ltd. 418 F.3d 1282 (Fed. Cir. 2005) Background

The Technology Involved The case involves systems that connect regular email systems to wireless radio networks. This lets a mobile user receive email over a wireless network.

The Campana Invention The inventors' key idea was to send email both over regular wires and over radio waves to a portable wireless device. The user could read messages on the device and later transfer them to a desktop computer.

Why the Invention Was Useful Sending messages to the portable device meant the desktop computer did not have to be turned on or carried around to receive email. The patents did not, however, describe a way to write and send messages from the portable device.

RIM and the BlackBerry System RIM is a Canadian company based in Waterloo, Ontario, that sells the BlackBerry system for sending and receiving email on a small wireless device. The system includes the handheld device, email-forwarding software, and access to a nationwide wireless network.

Push" Email and Its Versions The BlackBerry uses "push" technology, which sends email to the device automatically without the user having to check for it. There are different versions: a Desktop version installed on a personal computer, a Corporate version installed on a company's mail server, and an Internet version.

Seamless Fit With Existing Email In every version, the BlackBerry software blends smoothly into the user's existing email system. The email system does not notice the BlackBerry software, and no changes to it are needed.

How Messages Reach the Relay When new mail arrives, the software copies it, encrypts it, and sends it to the BlackBerry "Relay," which is located in Canada. In the Corporate version, the software grabs the email before it reaches the user's computer, so the computer does not need to be on, and users can set their forwarding preferences.

Delivery to the Handheld From the Relay, the message goes to a partner wireless network, which delivers it to the user's BlackBerry almost instantly. This happens without the user doing anything, which is what "push" email means.

Benefits of Push Email The main benefit is that users no longer have to connect to the mail server to check for new email. This saves time and lets users respond to important messages right away.

Sending Email From a BlackBerry BlackBerry users can also send email, and sending follows the same path as receiving, just in reverse. Messages sent from the device look exactly like messages sent from the user's desktop, coming from the same address and appearing in the same "sent" folder.

Procedural Posture

The BlackBerry Appeal RIM, the maker of the BlackBerry, appealed a jury verdict that its system infringed NTP's patents, along with an award of about $53.7 million in damages. The trial court also permanently barred further infringement but put that order on hold during the appeal.

The Trial Court's Ruling and Jury Verdict The trial court ruled that the Relay's location in Canada did not prevent infringement and told the jury so. The jury found RIM liable for direct infringement and for encouraging and contributing to infringement, which meant its customers were also direct infringers.

Legal Standard

RIM's Main Defense RIM argued that because the Relay is located in Canada, it could not be liable for infringement as a matter of law. The court noted that section 271(a) applies only to infringement happening within the United States, since American patent laws do not reach beyond its borders.

Why This Case Is Harder Usually it is easy to tell whether infringement happened in the United States. Here it is harder because the invention is a system or method with many parts or steps, and those parts can work together even when they are in different places.

NTP's Claim and RIM's Motion NTP claimed that RIM infringed by making, using, selling, offering to sell, and importing BlackBerry products and services. RIM asked the court to rule that the Canadian Relay, which matches the patents' "interface" requirement, put the system outside United States law, but that argument did not affect one patent whose claims lacked that requirement.

RIM's Argument on Appeal RIM did not dispute that its customers used the system, but it argued that this use did not happen "within the United States." Relying on the Deepsouth case, RIM claimed that the entire system had to be inside the United States, and that the Canadian Relay, which it called the system's "control point," ruled out infringement.

The Question and Unclear Statute The court asked whether using or selling a patented invention counts as infringement when one part or step happens abroad. The statute's wording treats "within the United States" as a separate requirement, so it does not clearly answer what happens when part of the invention is located elsewhere.

Deepsouth Doesn't Answer the Question In Deepsouth, the Supreme Court held that making all the parts of an invention in the United States and shipping them abroad for assembly was not infringement. That case involved an invention assembled entirely outside the country, unlike this case, where the system is partly inside and partly outside.

The Decca Navigation Case In Decca, the government was sued over a radio navigation system that used three transmitting stations, one of them in Norway. Since one claim required all three stations, the court had to decide how patent law applies when one part of a system is located abroad.

Decca on "Making" the System The Decca court found it hard to say the system was "made" in the United States, because one part was in Norway. It noted that as much of the system as possible was in the United States, but it did not clearly resolve whether the system was "made" there.

Decca on "Using" the System The Decca court found it clearer that the system was "used" in the United States, because navigators used the Norwegian station's signals wherever they received them. It considered it important that the United States owned and controlled the equipment and benefited from its use in the country.

The Lessons From Decca Decca shows that having one part of a system abroad does not automatically prevent infringement. It also shows that the answer may differ depending on the type of infringing act and the type of claim, so the court analyzed system claims and method claims separately.

The Meaning of "Use" Infringement happens where the infringing act takes place, and courts read the word "use" broadly to mean putting something into service. A patented system is "used" where the system as a whole is put into service, meaning where it is controlled and where its benefits are received.

Method Claims Are Different Using a method is different from using a system, because a method is a series of steps that must actually be carried out. A method is not infringed unless every one of its steps is performed.

Every Step Must Happen in the United States Unlike a system, whose parts are used together, a method is used by performing each step. So the court held that a method is used "within" the United States only if every step is performed in the country

Could RIM Infringe by Selling the Methods? Since the "use" theory failed for the method claims, the court considered whether RIM infringed them by selling or offering to sell them. No earlier case had decided whether a method can be "sold" in the United States when it would be performed partly abroad, so this was a new question.

The Statute and Method Claims The statute does not define "sells" or "offers to sell," and it does not say which infringing acts apply to which types of claims. Section 271(a) simply wrote existing law into the 1952 Patent Act, and the exact boundaries of infringing a method claim have never been clearly set.

The Ordinary Meaning of "Sale" Following an earlier case, the court looked to the ordinary dictionary meaning of "sale," which involves transferring property or ownership for a price. It is hard to apply this to a method, because performing steps for payment does not necessarily transfer any property.

Congress's View From 1987 Committee reports from Congress in 1987 said that a process patent gives the right to stop others from using the process, while the rights to make or sell do not directly apply. An earlier case from the court likewise said a method is directly infringed only by someone actually performing it.

The 1994 International Trade Agreement In 1994, Congress added "offering to sell" and "importing" to the statute to carry out an international trade agreement. That agreement protects processes directly only from being used, while protecting sales and imports of products made by those processes.

Court Reasoning

The System Claims Were Used in the United States RIM's American customers controlled the sending of messages and benefited from them within the United States. So the Canadian Relay did not, as a matter of law, prevent infringement of the system claims.

Rejecting RIM's Attempt to Distinguish Decca RIM argued its case was different because the Relay controls the system and is located outside the United States. The court found this beside the point, because customers using their handhelds in the United States were using the system as a whole there, so the jury properly considered the system claims.


The Method Claims Fail Each of the method claims at issue included a step involving the "interface," which only the Canadian Relay performed. So RIM's customers could not directly infringe those methods, which also meant RIM could not be liable for encouraging or contributing to their infringement.

Legislative History Points to Use Only A Senate report on the 1994 changes reflected the same view, extending sale and import protection to products, not to the processes themselves. Overall, the legislative history shows that Congress understood method claims could be directly infringed only by use.

The "On Sale" Bar Doesn't Change the Answer Under a different rule, offering to perform a method too early can make a patent invalid, but the court has refused to carry that rule over to infringement. The Supreme Court has also warned that courts should not expand patent rights without a clear signal from Congress, and Congress's signals here point to a narrow reading.

RIM Did Not Sell the Methods The court did not hold that method claims can never be infringed by selling, only that RIM's performing some steps as a service for its customers did not count as selling the methods. Selling the handheld devices was not enough either, so RIM did not sell or offer to sell the patented methods in the United States.

Microsoft Corp. v. AT&T Corp. 550 U.S. 437 (2007) Background

AT&T's Patent and Windows AT&T has a patent on a device that digitally encodes and compresses recorded speech, and Windows contains code that lets a computer process speech in the patented way. However, Windows does not infringe until it is installed on a computer, just as a computer without Windows does not infringe.

The Court's Answer The question is whether Microsoft is liable for foreign-made computers loaded with copies of Windows made abroad from a master disk or electronic file sent from the United States. The Court said no, because the copies actually installed were made abroad, so Microsoft did not supply those parts "from the United States."

How Microsoft Distributes Windows Microsoft creates Windows at its headquarters in Redmond, Washington, and sells it to users and computer makers around the world. It sends foreign manufacturers a master copy on a disk or by encrypted transmission, and they make their own copies to install on computers sold abroad.

AT&T's Lawsuit AT&T's patent covers a computer that can encode and compress speech, and everyone agreed Windows enables this. In 2001, AT&T sued Microsoft in New York over Windows installations both in the United States and abroad

What Microsoft Admitted Microsoft admitted that it directly infringed by installing Windows on its own computers during development. It also admitted that it encouraged infringement by licensing Windows to makers of computers sold in the United States.

The Dispute Over Foreign Computers Microsoft denied liability for the master copies it sent abroad, and AT&T argued this was supplying parts from the United States under section 271(f). Microsoft responded that software is intangible information and not a "component," and that the foreign-made copies were not supplied from the United States.

Procedural Posture

The Lower Courts Ruled for AT&T The trial court held Microsoft liable under section 271(f), and a divided Federal Circuit panel agreed. The Supreme Court agreed to hear the case and reversed.

Legal Standard

Reading the Exception Narrowly There are reasonable arguments on both sides about whether section 271(f) should cover Microsoft's conduct. Because the provision is an exception to the rule that patent law stops at the border, the Court read it narrowly and left any changes to Congress.

The Deepsouth Shrimp Machine Case Congress passed section 271(f) in response to Deepsouth, a 1972 case about a patented shrimp-deveining machine. Deepsouth admitted it could not make and sell the machine in the United States, but it argued it could make the parts here and sell them to foreign buyers to assemble abroad.

Congress Responds With Section 271(f) Congress then passed section 271(f) to treat supplying a patented invention's parts from the United States as infringement. The first part covers supplying all or a large portion of the parts in a way that actively encourages their combination abroad.

The Second Part of the Statute The second part covers supplying even a single part that is specially made for the invention and has no substantial use other than infringing. The supplier must know the part is specially made and intend for it to be combined abroad in a way that would infringe if done in the United States.

The Two Questions The first question is when, or in what form, software counts as a "component." The second is whether Microsoft supplied the foreign computers' components "from the United States."

Two Ways to Think About Software Everyone agreed software can sometimes be a component, but they disagreed about when. Software can be thought of in the abstract, like the notes of Beethoven's Ninth Symphony, or as a physical copy on a disk, like the sheet music.

The Parties' Positions AT&T argued that software in the abstract counts as a component. Microsoft and the federal government argued that only a physical copy of software can be a component.

Legal Reasoning

The Court Agreed With Deepsouth Under the law at the time, making or using a patented product outside the United States was not infringement, so the foreign buyers did not infringe. Because no one infringed, Deepsouth could not be liable for helping, and it was not a direct infringer either, since it never made, sold, or used the fully assembled machine in the United States.

Unpatented Parts and Congress's Role The machine's individual parts were not patented, so exporting them unassembled did not infringe. The Court rejected the patent owner's argument that this reading was too narrow, explaining that a patent on a combination protects only the fully assembled whole and that any expansion had to come from Congress.

No Reach Beyond the Border The Court stressed that American patent law was never meant to operate outside the United States. Without a clear signal from Congress, courts had no authority to stop the making and sale of parts for assembly and use abroad.

Why the First Question Decides the Second If the components are the copies actually installed abroad, AT&T cannot argue they were supplied from the United States, since they were made overseas. If abstract software is the component, it would not matter that the master copies themselves were never installed.

Components Must Be Combinable The Court read the statute's words in their ordinary sense, and the statute covers only parts that can be combined to form the invention. Abstract software cannot be inserted, downloaded, or installed until it exists as a copy, so it is just an idea and not a combinable component.

The Blueprint Comparison Abstract software is like a blueprint, which may give precise instructions for building a device but is not itself a part of that device. AT&T did not argue otherwise, and an earlier case held that sending chip-making instructions abroad was not covered by section 271(f).

AT&T Says Software Is Different AT&T argued that software, unlike a blueprint, is a standalone product that works with many computers and can be updated or deleted after installation. It also argued that software keeps working inside the computer, while a blueprint's job is done once the device is built.

The Court Is Not Persuaded Blueprints can also be developed, bought, and sold separately, and stores sell copies of software, not software in the abstract. Before software can run, be updated, or be deleted on a computer, a physical copy must be delivered to it.

The Copying Step Matters AT&T argued that copying software is so easy that it should not matter legally. The Court held that copying is what makes software a usable part, just as a machine that makes thousands of sprockets an hour is not itself a part of every device using those sprockets.

Only a Copy Is a Component Congress could have written the law to cover information, instructions, or tools used to make components, but it did not. So a copy of Windows, not Windows in the abstract, is a "component."

The Federal Circuit's View on Supplying Under a normal reading, the copies installed abroad were supplied from abroad, not from the United States. The Federal Circuit had disagreed, reasoning that copying software is part of supplying it, so sending one master abroad to be copied covers all the foreign copies.

The Court Agrees: Only the Supplied Parts Count The statute applies to the very parts supplied from the United States, not to copies of them. The copies installed abroad did not even exist until third parties made them outside the United States.

Ease of Copying Is Not the Test Many things besides software are easy to copy, such as keys, machine parts, chemicals, biological substances, and paper documents. The statute says nothing about copying, which weighs against treating copies made abroad as supplied from the United States.

The Presumption Against Reaching Abroad Any remaining doubt is settled by the rule that United States law is presumed to apply only at home, a rule that is especially strong in patent law. The Patent Act itself grants rights only within the United States and for imports into it.

Respecting Other Countries Courts assume Congress respects other nations' legitimate interests when it writes laws. Foreign activity is generally governed by foreign law, which may strike a different balance between inventors, competitors, and the public.

The Presumption Weighs Against AT&T This presumption pushes strongly against treating intangible code as a component and treating copies made abroad as supplied from the United States. It still applies even though section 271(f) specifically addresses activity abroad, because it helps define how far that exception reaches.

AT&T's Domestic-Conduct Argument Fails AT&T argued the presumption does not apply because section 271(f) covers only conduct in the United States. The Court disagreed, because AT&T's reading would turn one shipment from the United States into liability for every copy later made and sold abroad, so AT&T's remedy is to get and enforce foreign patents.

The "Loophole" Argument AT&T argued that this ruling creates a loophole, because software makers can avoid liability simply by having copies made abroad from a master sent from the United States. The Federal Circuit had agreed, saying the law should be read to fit modern technology and industry practices.

The Loophole Is for Congress to Close The Court understood the concern but declined to stretch the statute to fit new technology. It noted that section 271(f) does not cover exporting blueprints, templates, or prototypes, and nothing suggests Congress meant to treat Microsoft's master copies differently.

Congress Fixed Only the Deepsouth Gap Section 271(f) responded to Deepsouth, which involved exporting physical kits of machine parts that were themselves assembled abroad. Congress did not address other possible gaps, such as sending instructions or materials abroad so that copies can be made there.

Congress Knows How to Address Copying Congress is aware of how easily software can be copied and addressed digital copying in copyright law through the Digital Millennium Copyright Act in 1998. Any change to patent law to reflect how software is distributed should come from Congress, so the Court reversed the Federal Circuit.

Why Dissent believed Congress intended software to be included

Software Is a Component Justice Stevens reasoned that if exporting a stock of specially made knives for a patented machine counts, a master disk is like a warehouse of components that Microsoft expected to be installed abroad. He added that software, unlike a blueprint, actually causes the infringement, like a player-piano roll rather than sheet music, and its only intended use is infringing, so he would have ruled for AT&T.

Life Technologies Corp. v. Promega Corp. 137 S. Ct. 734 (2017) Background

The Question in the Case This case is about how patent law applies to international supply chains. Section 271(f)(1) bars supplying "all or a substantial portion" of a patented invention's parts from the United States to be combined abroad, and the Court asked whether supplying just one part can count.

The Court's Answer The Court held that a single part is not a "substantial portion" of an invention's parts under section 271(f)(1). Because only one part of the invention here was shipped from the United States, the Court reversed and sent the case back.

The Genetic Testing Kit Patent The Tautz patent covers a kit that takes small samples of DNA and makes many copies of a specific section, a process called amplification. The resulting DNA profiles are used by police for forensic identification and by medical and research institutions around the world.

The Kit's Five Parts Both sides agreed that the patented kit has five parts. These are primers that mark the section of DNA to copy, nucleotides that serve as building blocks, an enzyme called Taq polymerase, a buffer solution, and control DNA.

The License Arrangement Promega held the exclusive license to the patent and gave Life Technologies a sublicense to make and sell the kits for certain law enforcement uses worldwide. Life Technologies made four of the five parts in the United Kingdom and made only the Taq polymerase in the United States.

Promega's Lawsuit Life Technologies shipped the Taq polymerase to its United Kingdom plant, where it was combined with the other four parts. Four years into the deal, Promega sued, claiming Life Technologies sold kits outside the licensed uses and that shipping the enzyme from the United States made it liable under section 271(f)(1).

The Statute's Wording At trial, the parties disputed what "all or a substantial portion" of the parts means. The statute makes liable anyone who supplies all or a substantial portion of a patented invention's parts from the United States in a way that actively encourages combining them abroad in a way that would infringe if done in the United States.

Procedural Posture

The Jury Verdict and Trial Court Ruling The jury found that Life Technologies willfully infringed the patent. The trial judge then overturned the verdict, ruling that supplying only one part, the Taq polymerase, did not count as a "substantial portion."

The Federal Circuit Reverses The Federal Circuit restored the verdict, holding that supplying a single part can sometimes create liability. Relying on a dictionary meaning of "substantial" as "important" and on expert testimony that the enzyme was a "main" and "major" part, it found that the enzyme counted.

The Supreme Court's Holding The Supreme Court framed the question as whether supplying one part of an invention with several parts infringes under section 271(f)(1). It held that it does not.

Legal Standard

Amount or Importance? The first question was whether "substantial portion" refers to quantity, meaning how many parts, or quality, meaning how important they are. Life Technologies and the federal government argued it means a number greater than one, while Promega argued a single important part could be enough.

The Word Alone Is Unclear The Patent Act does not define "substantial," so the Court looked to its ordinary meaning. Dictionaries define it as either "important" or "large in amount," so the word alone does not settle the question.

Can One Part Ever Be Enough? Next, the Court asked whether a single part could ever count as a "substantial portion," and it answered no. The statute consistently uses the plural word "components," which shows that more than one part is required.

Legal Reasoning

The Surrounding Words Point to Quantity A word's meaning is shaped by the words around it, and both "all" and "portion" refer to amounts. "All" means the whole amount, "portion" means part of a whole, and nothing in the nearby text supports a meaning based on importance.

Every Word Should Count The statute refers to a substantial portion "of the components," and if importance were what mattered, Congress could have just said "a substantial portion of a patented invention." Reading "substantial" as importance would make "of the components" unnecessary, so only the quantity reading gives every word meaning.

Rejecting a Case-by-Case Approach Promega asked the Court to let juries decide case by case, using either a quantity test or an importance test. The Court refused, because its job was to resolve the ambiguity, not leave juries across the country to interpret the statute on their own.

Rejecting a Combined Test The Court also declined to adopt a test that looks at both the number of parts and their importance. Although judging importance might seem helpful in close cases, the statute's structure gives little support for it.

An Importance Test Wouldn't Work Nearly every part of an invention is important, since most inventions would not work without any one of their parts. Promega could not name a single part of the kit that would fail its own "importance" test.

Quantity Is Clearer and Workable Neither Promega nor the Federal Circuit offered an easy way for courts or businesses to rank the importance of parts. The Court concluded that the quantity reading fits the statute best and is practical to apply.

The Companion Provision Covers Single Parts Section 271(f)(2), the companion provision, covers supplying "any component," singular, that is specially made for the invention and is not a common commodity. Reading the first provision to require multiple parts lets the two provisions work together.

Two Provisions, Two Different Jobs The first provision asks whether the supplied parts amount to a substantial portion of all the parts, while the second asks whether a single specially made part was supplied. The Court agreed the provisions cover different situations, as it had noted in Microsoft v. AT&T, and said they differ in how many parts must be supplied.

Protecting the Second Provision's Role Reading the first provision to cover any single part would leave little room for the second provision. It would also undermine the second provision's specific focus on single parts that are specially made for the invention.

Plural Means Plural A general rule says plural words can include the singular unless the context says otherwise. Here, the statute's text, context, and structure show that Congress meant plural when it said "components" and singular when it said "component."

A Narrow Holding The Court did not decide how close to "all" the parts a "substantial portion" must be. It held only that one part is not enough, which was all it needed to decide the case.

History Supports the Ruling Congress passed section 271(f) in response to the Deepsouth decision, which held that making or using a patented product outside the United States was not infringement. The new law filled that gap by reaching parts made in the United States but assembled overseas.

The Ruling Fits Congress's Intent A supplier can be liable for shipping a substantial portion of the parts, plural, even when they are combined abroad, or for shipping a single part that is specially made and not a common commodity. When a product is made abroad and only one common part comes from the United States, the activity falls outside the statute.

Conclusion The Court held that "substantial portion" refers to quantity, not importance, and that the provision does not cover supplying just one part of an invention with several parts. It therefore reversed the Federal Circuit and sent the case back for further proceedings.

Eli Lilly & Co. v. American Cyanamid Co. 82 F.3d 1568 Background

Brand-Name Versus Generic Drug Makers This case is another round in the ongoing fight between "pioneer" drug companies that develop new drugs and generic drug distributors. Eli Lilly sued several companies distributing a generic drug, claiming that importing and selling it in the United States infringed Lilly's patent on a process for making a related compound.

Brand-Name Versus Generic Drug Makers This case is another round in the ongoing fight between "pioneer" drug companies that develop new drugs and generic drug distributors. Eli Lilly sued several companies distributing a generic drug, claiming that importing and selling it in the United States infringed Lilly's patent on a process for making a related compound.

The Drug Cefaclor The drug at issue is cefaclor, an antibiotic that works against many kinds of bacteria and belongs to a family built on a shared core chemical structure called the cephem nucleus. Every known commercial method of making cefaclor first produces an in-between compound called an enol, which then goes through several more steps to become cefaclor.

Lilly's Patents Lilly developed and patented cefaclor in 1975 and was its only maker and seller in the United States until recently. Lilly also obtained patents on ways of making cefaclor, including processes for making the in-between enol compounds, but many of those patents have expired.

The Patent in This Case In 1995, Lilly bought the '085 patent, whose claim 5 covers a method for making enol compounds, including one called "compound 6." Compound 6 is similar to the in-between compound Lilly uses to make cefaclor, and the patent was set to expire on July 3, 1996.

How Compound 6 Differs From Cefaclor Compound 6 and cefaclor share the same core structure but have different chemical groups attached at three positions. Each difference helps make cefaclor an effective antibiotic that can be taken by mouth, because it improves the drug's bacteria-fighting activity and strength and allows it to work orally.

Four Steps From Compound 6 to Cefaclor Turning compound 6 into cefaclor takes four separate chemical steps. Each step swaps or removes a chemical group, producing a series of in-between compounds and finally cefaclor.

The Generic Distributors In April 1995, Zenith and American Cyanamid received approval from the Food and Drug Administration to sell cefaclor in the United States, and Biocraft had applied for approval but had not yet received it. All three obtained large amounts of cefaclor made in Italy by a company called Opos.

Procedural Posture

Lilly Files Suit On the same day Zenith and Cyanamid won approval, Lilly obtained the '085 patent and sued all four companies. It asked the court to declare that importing Opos's cefaclor infringed its patents and to order the defendants to stop importing it.

The Trial Court's Main Findings After a three-day hearing, the trial court focused mainly on whether Lilly was likely to win its infringement claim. It found that Lilly would probably show its patent was valid but would probably not show that the patent was infringed.

Legal Standard

The Law on Imported Products The key law is the Process Patent Amendments Act of 1988, which makes it infringement to import, sell, or use in the United States a product made abroad using a process patented in the United States. However, the law does not apply if the product is "materially changed" by later processing before it is imported.

Why Congress Passed the 1988 Law Before 1988, a process patent owner could sue only if someone used the process in the United States, not if they used it abroad and then imported the products. The owner's only option was to ask the International Trade Commission to block the imports, so the 1988 law closed this gap by making such imports, sales, and uses infringement.

Limits for Changed Products Congress worried about products made abroad by a patented process but then changed or built into other products before import. It addressed this by saying a product no longer counts as "made by" the process once it is "materially changed" or becomes a trivial, nonessential part of another products

Legal Arguments

Lilly's Purpose-Based Argument Lilly argued that the "materially changed" phrase should be read in light of the law's purpose, which is to protect the economic value of process patents. Before the law, that value could be undermined by using the process abroad and importing the product, so Lilly argued the law should cover any scheme that undercuts the patent's value.

Lilly's Proposed Test Under Lilly's view, a product is not "materially changed" if its main commercial use is to be converted into the imported product. Because cefaclor is the only product made from compound 6 that is sold in the United States, Lilly argued the conversion is not a material change, however large the chemical differences.

Court Reasoning

Compound 6 Was Materially Changed The trial court found that compound 6 and cefaclor differ significantly in structure and properties, including their effects on living organisms. Relying on a Senate report, it held that the steps turning compound 6 into cefaclor change its basic usefulness, so Lilly was unlikely to win.

No Irreparable Harm The trial court also found that Lilly had not shown it would suffer harm that money could not fix if the court refused to stop sales. Lilly did not get the usual assumption of such harm because it had not made a strong showing that it would win, and money damages would be enough if Lilly eventually won.

Other Factors Not Needed Since Lilly failed to show either a likely win or harm that money could not fix, the court did not need to consider the other factors. Those factors are the balance of hardships between the parties and the public interest.

The Argument Has Appeal The court acknowledged that Lilly's argument has real appeal, because a stingy reading could weaken the law's ability to address the problem Congress targeted. This is especially true for chemicals, where simple, routine reactions can dramatically change a product's structure and properties.

But Purpose Can't Override the Words A law's purpose helps interpret its words but cannot replace them, and the court could not stretch "materially changed" as far as Lilly wanted. The statute focuses on changes to the product itself, which does not fit Lilly's test based on whether the patent's economic value is harmed.

Lilly's Test Has No Limits Lilly pointed out that only four routine steps separate compound 6 from cefaclor, but it offered no limiting principle based on the compound's structure or the steps involved. Under Lilly's test, even ten complex reactions would not count as a material change, as long as compound 6's main use in the United States was to make cefaclor.

A Strange Result Under Lilly's Test Lilly's test treats compound 6 and cefaclor as basically the same because compound 6 has no other commercial use in the United States. That would mean compound 6 becomes "materially different" from cefaclor if it later gains other uses, even though the two compounds themselves have not changed at all.

The Court Rejects Lilly's Test The court refused to make the question of material change depend on whether the first compound has other products with economic value. It declined to adopt Lilly's reading of the statute.

The Correct Test: Size of the Change Instead, the court looks at how significant the change is between the product of the patented process and the imported product. For chemicals, a material change most naturally means a significant change in the compound's structure and properties, and the change from compound 6 to cefaclor cannot easily be called immaterial.

The Shared Core Doesn't Make Them the Same Compound 6 and cefaclor share the core structure behind all antibiotics in this family, but thousands of compounds share it, and many are very different from one another. Beyond that core, the two compounds differ in four important ways, matching the four chemical steps between them.

More Than a Minor Tweak Simply adding or removing a protective chemical group might not, by itself, be a material change. The conversion here involves much more than that, so the statute's text does not support Lilly's claim that it is likely to win.

The Legislative History Is Mixed Both sides cited Congress's records, and each found some support, including statements favoring strong protection that is "not easily evaded." Other passages about in-between chemical compounds suggest a narrower reading, so the history does not clearly favor Lilly or cast doubt on the trial court's analysis.

Why Rader Dissent believed in danger of loophole

Judge Rader's Separate Opinion Judge Rader wrote separately, disagreeing with the majority's reasoning about the "material change" standard. He criticized the majority for relying on legislative history that it admitted had something for each side, and he described that history as a bitter battle between brand-name and generic drug makers.

A New Loophole Judge Rader warned that the decision creates a large loophole that effectively denies protection to process patents on in-between products rather than final products. It would leave a patented process unprotected whenever it is not the only way to make an in-between product, even if it is the most efficient way.

Compound 6 and Cefaclor Are Essentially the Same In light of the law's purpose, Judge Rader viewed compound 6 and cefaclor as essentially the same product, because compound 6's only commercial use in the United States is making cefaclor, which is just four simple, well-known steps away. He argued the court should read "material change" to serve the law's main goal of protecting process patent owners, rather than sorting through confusing legislative history.