OPINION

RICHARD A. POSNER, Circuit Judge, sitting by designation.

In 1997, Abbott Laboratories filed this suit for patent infringement against TorPharm, Inc., a manufacturer of generic pharmaceutical drugs, and affiliates of TorPharm, notably Apotex, Inc., its parent. (TorPharm has since been absorbed into its parent, so I have dropped it from the caption and throughout this opinion refer to TorPharm as Apotex.) The district judge initially assigned to the case granted summary judgment in favor of Abbott on both validity and infringement. 156 F. Supp. 2d 738 (N.D. Ill. 2001). The Federal Circuit affirmed the ruling on validity but remanded for a trial on infringement. 300 F.3d 1367 (Fed. Cir. 2002). I was designated to conduct the trial ( 28 U.S.C. § 291(b)) — a bench trial, because Abbott was seeking only equitable relief. Indeed, it could not seek damages, because Apotex had not yet begun to market its generic substitute for Abbott's product. At the conclusion of the trial, I held that Apotex had infringed, Abbott Laboratories v. TorPharm, Inc., 309 F. Supp. 2d 1043 (N.D. Ill. 2004), and after considering further submissions from the parties, dealing with relief, I issued an injunction, No. 97 C 7515 (N.D. Ill. Mar. 31, 2004), which the Federal Circuit affirmed. 122 Fed. Appx. 511 (Fed. Cir. 2005). The case is back before me on a motion by Abbott to rule that Apotex has violated the injunction. "When an equity case ends in a permanent injunction, the trial court, with or without an explicit reservation of jurisdiction, retains jurisdiction to enforce the injunction, as by contempt proceedings." McCall-Bey v. Franzen, 777 F.2d 1178, 1183 (7th Cir. 1985); see also Riverdale Cotton Mills v. Alabama Georgia Mfg. Co., 198 U.S. 188, 195 (1905). Necessarily, therefore, the judge who issues an injunction has the authority to interpret it if a dispute arises over its scope, In re China Peak Resort, 847 F.2d 570, 571-72 (9th Cir. 1988), vacated on other grounds under the name California State Bd. of Equalization v. Sierra Summit, Inc., 490 U.S. 844 (1989); Wilson v. Alexander, 276 F. 875, 881 (5th Cir. 1921) — and indeed is in a better position than anyone else to do so. Schering Corp. v. Illinois Antibiotics Co., 62 F.3d 903, 908 (7th Cir. 1995); In re Chicago, Rock Island Pacific R.R., 860 F.2d 267, 272 (7th Cir. 1988). But the issue here is not the meaning of the injunction; the meaning is clear; the issue is whether the injunction has been violated. If it has been, even if in good faith, that is a civil contempt, McComb v. Jacksonville Paper Co., 336 U.S. 187, 191 (1949); Additive Controls Measurement Systems, Inc. v. Flowdata, Inc., 154 F.3d 1345, 1353 (Fed. Cir. 1998), and I can so rule without imposing monetary sanctions or imprisonment. Abbott is not requesting such sanctions at this time; it is content with a determination that Apotex has violated the injunction and with an "extension" (modification) of the injunction to eliminate any possible ambiguity about its reaching the "new" Apotex product.

A finding of contempt requires proof by clear and convincing evidence, e.g., Additive Controls Measurement Systems, Inc. v. Flowdata, Inc., supra, 154 F.3d at 1355-56, Preemption Devices, Inc. v. Minnesota Mining Mfg. Co., 803 F.2d 1170, 1172-73 (Fed. Cir. 1986), a rule that is appropriate because of the summary character of contempt proceedings. See, e.g., KSM Fastening Systems, Inc. v. H.A. Jones Co., 776 F.2d 1522, 1524 (Fed. Cir. 1985). That is the standard that will guide my inquiry.

The patent (actually two patents, but I will pretend they're one because they don't differ in any respect material to this opinion) is on a chemical called divalproex sodium, which Abbott sells under the trade name Depakote for the treatment of epilepsy and other ailments. The chemical is a complex of two molecules, one of sodium valproate and one of valproic acid. Although either sodium valproate or valproic acid would have all the desired therapeutic properties of the drug, combining them produces a crystalline solid that is easier to manufacture into pills than either of its constituents.

For reasons never adequately explained, Abbott's patent is limited to divalproex sodium in the form of an "oligomer" consisting of about 4 to 6 units of divalproex sodium. Just as the sodium valproate and valproic acid molecules combine to form the divalproex sodium complex, so two or more of these complexes might join together to form a larger structure, such as a crystal, which might consist of millions or billions of identical units, whether atoms, molecules, or looser complexes. A "polymer" is an assemblage of a large number of smaller molecules. At the opposite extreme is a "monomer," implying no discernible linkage with neighboring units of the same chemical. An oligomer, as the name implies (cf. "oligopoly"), is an assemblage of several rather than, as in the case of a polymer, many identical units.

In the infringement trial, Abbott presented compelling evidence that Apotex's product is indeed an oligomer, and therefore infringes; hence the injunction. Of course there would be no infringement if, as Apotex argued, Abbott's product was not an oligomer; for then Abbott's patent would be invalid. In the first round of this litigation, however, the Federal Circuit held unequivocally that Abbott's divalproex sodium is indeed an oligomer. 300 F.3d at 1372, 1378. That was the law of the case when I tried the infringement claim; and it might seem that now, by virtue of the doctrine of collateral estoppel, it precludes Apotex from relitigating the issue, since the finding undergirds the final judgment enjoining it. But collateral estoppel precludes relitigation of an issue in a subsequent case, and the present proceeding, initiated by a motion filed in an existing case (Abbott's infringement suit against TorPharm and Apotex), is not a subsequent case, and so it is the law of the case doctrine, rather than collateral estoppel, that limits relitigation. KSM Fastening Systems, Inc. v. H.A. Jones Co., supra, 776 F.2d at 1524, 1529.

The Federal Circuit in the first round of appeals made findings concerning Apotex's product that strongly suggested that it was identical to Abbott's. Nevertheless, it was at least conceivable that differences in the production process (Abbott uses a solvent in manufacturing divalproex sodium while Apotex just mixed the sodium valproate and the valproic acid and heated the mixture to produce divalproex sodium) might somehow prevent the oligomeric structure from forming in Apotex's product. However, evidence at the trial, at which Abbott's principal witness — an eminent supramolecular chemist named Jerry Atwood — reported the results of five different types of test that he had conducted on a batch of Apotex's product left no doubt in my mind that the product was indeed an oligomer; that despite the difference in process, it was identical to the patented product. (Apotex had conducted no tests.) There is no need to summarize the evidence or my analysis of it; the reader can consult my previous opinion. 309 F. Supp. 2d at 1047-54.

One further point about the trial is worth nothing, however. Apotex wanted to prove, mainly through testimony by Dr. Bernard Sherman, Apotex's principal, that Atwood's tests, and the earlier tests that had convinced the Patent Office that Abbott's product was an oligomer, were fatally flawed, that it was not an oligomer and therefore the patent (limited, as I said, to a divalproex sodium oligomer) was invalid. Sherman, I ruled, was no more qualified to testify about Abbott's tests than about Apotex's. He is a chemical engineer, not an analytical chemist. I noted also that the issue he wanted to reargue, concerning the chemical structure of Abbott's product (was it or was it not an oligomer?) was not only beyond his professional competence, but was made irrelevant by the Federal Circuit's decision upholding the validity of Abbott's patent.

In March 2005, a year after I issued the injunction, another Canadian company, Nu-Pharm, Inc., filed with the Food and Drug Administration an Abbreviated New Drug Application (ANDA) for a divalproex sodium product. Nu-Pharm is a tiny (six employees) company formerly owned by Apotex. It markets drugs for Apotex in Canada. It has never marketed a drug in the United States, and it does not develop or test drugs. The product for which Nu-Pharm filed the ANDA had been developed by Apotex, is owned by it, and, as Sherman has testified, "we [Apotex] will get our profit out of it, anyway."

It is obvious, and I find, that Apotex's choice of Nu-Pharm to file the ANDA was a subterfuge intended to give Apotex a crack at another district judge, who might, in an infringement suit by Abbott, conclude that it was a different, and noninfringing, product from the one I had enjoined. Abbott did sue Nu-Pharm for infringement, because it was unaware that Nu-Pharm was merely a stalking horse for Apotex. Had Abbott known, it would not have bothered with the other suit ( Abbott Laboratories v. Nu-Pharm, Inc., No. 05 C 3714 (N.D. Ill. filed June 24, 2005), which under the local rules was quite likely to end up before another judge; see N.D. Ill. Local Rule 40.3(b); it would have moved, as it now has, for relief in the original proceeding. (The judge in the Nu-Pharm suit has stayed proceedings pending my ruling on Abbott's motion.) I infer that Sherman hoped to get a favorable decision or ruling in that case before Abbott woke up, penetrated the veil, and sought to enforce its injunction against the new ANDA.

Under Federal Circuit case law, I may determine infringement in this contempt proceeding only if a summary proceeding is adequate for deciding whether the redesigned product (the subject of the Nu-Pharm ANDA) infringes the patent, and thus violates the injunction; otherwise I must remit the issue of infringement to a separate suit charging infringement by the new product (in this case, it would be Abbott's suit against Nu-Pharm and Apotex, pending before the other judge.) Specifically, I must do that if I find that "there is more than a `colorable difference' in the accused and adjudged devices," which is to say "if there are substantial open issues with respect to infringement to be tried," since "the presence of such disputed issues creates a fair ground for doubt that the decree has been violated." KSM Fastening Systems, Inc. v. H.A. Jones Co., supra, 776 F.2d at 1530, 1532; see also California Artificial Stone Paving Co. v. Molitor, 113 U.S. 609, 614 (1885); Conoco, Inc. v. Energy Environmental Int'l, L.C., No. 05-1363, 2006 U.S. App. LEXIS 21036, at *39 (Fed. Cir. Aug. 17, 2006); Stryker Corp. v. Davol, Inc., 234 F.3d 1252, 1260 (Fed. Cir. 2000).

The injunction extends to any "generic divalproex sodium" manufactured by Apotex that had been "found to be infringing." At the evidentiary hearing on Abbott's motion to determine that Apotex through its tool Nu-Pharm was violating the injunction, which I conducted on August 30, Atwood testified that he had performed on the new product three of the five tests that he'd performed on the old one (freezing-point depression, vapor phase osmometry, and FAB mass spectrometry — the others are melting point and solubility) on the new one and that a consulting firm, SSCI, Inc., had conducted a fourth, a melting-point analysis, plus four tests that had not been done in the original litigation — powder X-ray diffraction, solid state 13C nuclear magnetic resonance spectroscopy, infrared spectroscopy, and elemental analysis. He testified that all the tests (besides his own tests, he had reviewed the test data and results of the tests conducted by SSCI.) had revealed, within the margins of error accepted by the chemistry profession, that Nu-Pharm's product was chemically identical to both Abbott's product and the original Apotex product, including in structure (i.e., oligomeric). True, he testified only that the tests that he had performed were within the accepted margins of error, but his expert report stated that "all of the experiments conducted or commissioned by SSCI . . . were scientifically sound, based on accepted scientific procedures, and produced reliable results," implying that their results, too, were within the accepted margins of error, for otherwise the results would not have been reliable.

Apotex responded that it had altered the process of making its product by raising the heat level, but it presented no evidence that the alteration (or a different process, called "spray dry," that it also used to make the new divalproex sodium product) necessarily or even probably would alter the oli-gomeric structure. From admissions in the deposition testimony of Dr. Sherman, it is plain, and I find, that he altered the manufacturing process not on the basis of any scientific evidence that it would produce a different product (for different processes can produce the identical product) but because he was unshaken in his belief (now rejected by two courts — this court and the Federal Circuit) that divalproex sodium is never an oligomer, and he wanted additional shots at persuading a court. Thus, in changing the process, Sherman was hoping to come up with a product that would be found by a court to be noninfringing, not necessarily a product that was noninfringing. He admitted not only that testing "might or might not" show that the Nu-Pharm product was different from Abbott's product and from Apotex's original product, but also that the Nu-Pharm product might be even more likely to infringe Abbott's product than his original product was. This is stubbornness carried to the point of contumacy.

Apotex did, however, present the testimony of a reputable chemist, Michael Hursthouse, who conducted a test that had not been conducted in the original proceeding — single crystal X-ray diffraction analysis — which showed, he testified, that the new product was probably a polymer rather than an oligomer. Another chemist who testified for Apotex, Peter Stephens, modified Hursthouse's model of the new product to make it consistent with data that Stephens had obtained in a second test — powder X-ray diffraction analysis — a test also performed by SSCI, on behalf of Abbott, on the new Apotex, old Apotex, and Abbott samples. Stephens testified that his analysis also indicated that the new Apotex product was probably a polymer. But the analyses by Apotex's expert witnesses did not indicate that the new product was different from the old one, because they did not conduct their tests on that product. For all that appears, the test results would have been identical — with one exception: Stephens did run his powder X-ray diffraction test on the Abbott sample, and, consistent with SSCI's results from this test, his preliminary results indicated that it was identical to the new Apotex sample. That of course was a finding helpful to Abbott.

A third expert witness for Apotex (Michael Cima) testified that changing the process for making divalproex sodium could change the chemical structure of the resulting product, but that to determine whether it had changed it would require testing that he had not done.

At the conclusion of the evidentiary hearing, Apotex's counsel asked for more time before I ruled, in order to enable Apotex's experts to conduct their tests on Apotex's original product. I asked the parties to submit posthearing briefs that would include discussion of whether Apotex should be given additional time to present evidence. Although I asked for a brief, not for evidentiary submissions, Apotex has included with its brief affidavits reporting new test results and new analyses of the tests conducted previously. I did not authorize such submissions — only briefs on whether to authorize them — and they are hereby stricken. I have, however, considered three affidavits submitted after the August 30 hearing that presented testimony that otherwise could have been presented live at the hearing — an affidavit from Atwood, concerning his critique of Hursthouse's experiments, that I specifically requested from Abbott at the hearing, plus Hursthouse's response to the new affidavit, plus a further affidavit from Atwood replying to the response.

Upon consideration of the posthearing briefs, I have decided that to permit Apotex to extend this proceeding with further evidentiary hearings would simply reward its contumacy. The hearing that I held on August 30 revealed, consistent with Dr. Sherman's deposition, that Apotex had conducted no tests on the Nu-Pharm product before Nu-Pharm at Apotex's behest filed the ANDA in which it certified that the product was noninfringing.

A person or firm that has been enjoined has a duty to make a good-faith effort to comply with the injunction. For the principle, see Stotler Co. v. Able, 870 F.2d 1158, 1163 (7th Cir. 1989); American Fletcher Mortgage Co. v. Bass, 688 F.2d 513, 517 (7th Cir. 1982), and Peppers v. Barry, 873 F.2d 967, 969 (6th Cir. 1989), and for its application to patent injunctions Stryker Corp. v. Davol, Inc., 75 F. Supp. 2d 741, 743 (W.D. Mich. 1999), affirmed, 234 F.3d 1252 (Fed. Cir. 2000). So it behooved Apotex to test its new divalproex sodium product to make sure it wasn't an oligomer. It did not do this. It changed the process, but without obtaining a scientific opinion that this would change the structure, let alone testing the new product and Abbott's product to see whether the new product infringed. Even after being challenged by Abbott's motion it ran no tests to determine whether the new product was in fact different from the old one. One of the chemists who testified for Apotex at the hearing (Stephens) performed his test on the Abbott product, and "eyeballing" the result (but going no further) thought it looked the same. The fact that he did not complete his comparison of the Abbott product with the new Apotex product raises an inference that he feared what the result would be.

Because there is no credible evidence that the new Apotex product is different from the old, the enjoined, one and much evidence to the contrary, the threshold test for contempt of a patent injunction has been met: there is no more than a colorable difference between the new product and the infringing old product. (By "colorable" is meant "slight." Here, there is no difference at all.) But the Federal Circuit, wishing to encourage pharmaceutical companies to design around patented products, has ruled that a finding of "no more than a colorable difference" merely establishes that it is proper to determine infringement in the contempt proceeding rather than having to confide the issue to determination in a new infringement suit. Additive Controls Measurement Systems, Inc. v. Flowdata, Inc., supra, 154 F.3d at 1349-50; KSM Fastening Systems, Inc. v. H.A. Jones Co., supra, 776 F.2d at 1532.

It might seem that if the old product infringed and the new product is identical to the old product, it must infringe as well. But that is not correct. A product is infringing only if it infringes the claims in the patent, and the patented product might not have embodied the claims. Suppose Abbott's product hadn't been an oligomer; then even if Apotex's new product were identical to Abbott's, it would not infringe the patent, because the patent is limited to divalproex sodium in oligomer form. That is why the Federal Circuit considers it "error for a court to compare in its infringement analysis the accused product or process with the patentee's commercial embodiment or other version of the product or process" and holds that "the only proper comparison is with the claims of the patent." Zenith Laboratories, Inc. v. Bristol-Myers Squibb Co., 19 F.3d 1418, 1423 (Fed. Cir. 1994); see also Spectrum Int'l, Inc. v. Sterilite Corp., 164 F.3d 1372, 1381 (Fed. Cir. 1998); KSM Fastening Systems, Inc. v. H.A. Jones Co., supra, 776 F.2d at 1530. But as I noted earlier, Abbott's product has been found to be an oligomer, and thus to be within the patent claims. So if Apotex's new product is identical to Abbott's — and the clear and convincing evidence demonstrates that it is — it must be an oligomer too.

In any event there is fresh, independent, and compelling evidence that Apotex's new product is an oligomer — the evidence that Professor Atwood presented at the August 30 hearing. From the results of the freezing-point depression, vapor phase osmometry, and FAB mass spectrometry tests that Dr. Atwood performed in this phase of the case, he concluded that the Nu-Pharm product was an oligomer. The experiments were part of the battery of experiments that convinced me in the first case that Apotex's (old) product was an oligomer. Atwood obtained the same results when he ran these tests on the Nu-Pharm samples as he had obtained when he had run the tests on the Abbott and (old) Apotex samples, and the results were further supported by his review of the battery of tests performed by the consulting firm. The result of the melting-point test, performed by the firm, is especially instructive because, as I noted in the original case, melting point is strongly probative of oligomeric structure. The melting point of the Nu-Pharm samples was within the 98-100°C range set forth in the patent.

Atwood was, moreover, more credible than Apotex's experts. Stephens, for example, omitted from his expert report his preliminary conclusion that the Abbott and Nu-Pharm products were identical (the conclusion was brought out on cross-examination), and Cima made no attempt to explore or explain the chemical structure of any divalproex sodium product. There are further problems with the evidence given by Stephens, and problems with Hursthouse's evidence as well. I mentioned their failure (except for one test by Stephens — the result of which supported Abbott's case!) to run their tests on the Abbott and (old) Apotex products. In addition, Hursthouse's model did not agree well with the data, as judged by "R value," a measure of the discrepancy between the hypothesis and the data used to test it. An R value much in excess of 5 to 10 percent "would be ground for suspicion." Jack D. Dunitz, X-Ray Analysis and the Structure of Organic Molecules 184 (1995). As calculated by Hursthouse himself, the R value for his test results was 18.74% and an astronomical 40.43% when he employed an alternative method of computation. And when Stephens adjusted the Hursthouse model to be consistent with his own data, the R value jumped to 41.51%. Stephens admitted that an R value over 40% is "crap."

The ratio of observations to atomic parameters (spatial and thermal characteristics) is also used to assess the reliability of the type of analysis of chemical structure conducted by Hursthouse and Stephens. Hursthouse and Atwood agreed that a ratio of 10 to 1 is preferred and that any study in which the ratio of observations to parameters falls below about 5 to 1 is unreliable. From examination of Hursthouse's raw data and his expert report, Atwood concluded that Hursthouse had used only 2,772 of the 12,301 observations he had collected, resulting in a ratio of observations to parameters of only 2.15 to 1. Although Hursthouse balked at Atwood's contention that he had used only 2,772 observations, he neither provided any support for his rejection of Atwood's contention nor quarreled with Atwood's estimate of a 2.15 to 1 observations-to-parameters ratio. Nor did he confute Atwood's testimony that the crystal on which Hursthouse performed his test was of an "exceedingly small size" and in fact "below the minimum size recommended for study of organic compounds by [single crystal X-ray diffraction analysis, Hursthouse's method]." Atwood also testified that the crystal was of "low quality." The significant problems with Hursthouse's experiment deprive his results of any probative value. (Atwood also complained about manipulations that Hursthouse had made in "refining" the data so that it would fit a polymeric model.)

Atwood testified that Stephens had had to alter Hursthouse's model of divalproex sodium in several ways in order to get it to match his powder X-ray diffraction data. One of these ways was by stretching the bonds between atoms. By stretching the bonds, Stephens weakened them, with the result that there was a "break" every seven units, suggesting an oligomer — which is to say that Stephens himself was offering evidence that supported Abbott! Even Hursthouse suggested that it was possible that the Nu-Pharm sample could be a series of oligomers. Atwood also questioned whether the samples on which Apotex's experts conducted their tests are representative of the bulk material from which they were drawn.

One mustn't miss the forest for the trees. One must consider the evidence in its entirety, including Sherman's seeming indifference to patent law and injunctions and the pregnant failure of his experts to run their tests on the Abbott and original Apotex products. The evidence taken as a whole overwhelmingly favors Abbott.

All that Apotex is left to argue is that a contempt proceeding is an improper vehicle for determining infringement if expert evidence is presented. The Federal Circuit has never so held, but on the contrary in Additive Controls Measurement Systems, Inc. v. Flowdata, Inc., supra, 154 F.3d at 1349, approved the district court's decision to conduct a contempt proceeding in which expert evidence was presented. And it has explained that the need for expert evidence counsels against the contempt route only when the defendant has made a good-faith effort to comply with the injunction, Arbek Mfg., Inc. v. Moazzam, 55 F.3d 1567, 1570 (Fed. Cir. 1995), which Apotex did not do. As Sherman made only minor modifications, unsupported by any scientific rationale, to a process that had generated a product previously adjudged infringing and enjoined, it behooved him to conduct scientific testing to determine whether his product was indeed not an oligomer and therefore not infringing. His failure to conduct any scientific testing and his cavalier disregard for judicial findings that the patent was valid and infringed by his original product, as well as his use of Nu-Pharm as a stalking horse, were the antithesis of good faith.

Were I to hold that contempt proceedings were impermissible in this case, it would be difficult to imagine any pharmaceutical patent case in which a contempt proceeding was permissible. No judge is qualified to determine by looking at a heap of powder (as Apotex's attorneys invited me to do) whether a change in the process by which a chemical is made has altered the chemical structure of the product without scientific testing conducted by experts. If Apotex can elude a finding of contempt, enjoined infringers need make only a trivial change in the process by which their infringing product had been made and present shoddy scientific evidence that the new product is different. Such infringers would not even need to have a good faith belief that the changes in the process had resulted in a noninfringing product. They would scour the scientific community for new experts and new tests that gave the desired result and go from judge to judge arguing that what was in reality the same substance held to have infringed was actually noninfringing. Injunctions in pharmaceutical patent cases would be worthless and the Federal Circuit's characterization of contempt proceedings as "a shield protecting the patentee against an infringer's flagrant disregard for court orders" would be a sour joke. Id. at 1570.

Apotex insists quite rightly that attempting to invent around a valid patent is legitimate conduct. Indeed, one of the purposes of requiring that a patent explain in detail how to make the patented product is to facilitate inventing around (and thus trim the economic consequences of the patent monopoly, as well as to facilitate greater innovation) by giving other inventors information that may enable them to figure out how they can create a product that does not infringe but that confers similar benefits on consumers. But the privilege of inventing around is not a privilege to violate an injunction against infringement by producing the identical product enjoined as a patent violation and defending such behavior by saying that the defendant was simply trying to invent around. Unreconciled to a binding judicial determination that the divalproex sodium product patented by Abbott is an oligomer, Apotex is determined to keep making divalproex sodium until new tests or new testers convince some judge somewhere of the soundness of an opinion that he is not even professionally competent to form.

The injunction has been violated. Should the violation continue, Apotex — a large, successful, and sophisticated enterprise — will be risking heavy sanctions for its willful disobedience of the injunction. For the present, however, it will suffice to extend the injunction to embrace the Nu-Pharm ANDA, Conoco, Inc. v. Energy Environmental Int'l, L.C., supra, 2006 U.S. App. LEXIS 21036, at *39; the extended injunction is attached to this opinion.