McCONNELL, Circuit Judge.
In June of 2004, Willis Ray McPhail, a Kiowa County, Oklahoma farmer, was
killed when he started his tractor by bypassing its malfunctioning circuitry. When it started, the tractor suddenly lurched backward and ran him over, inflicting fatal injuries. His widow, Barbara Jean McPhail, brought suit in state court on her own behalf and on behalf of his estate against Deere & Co. (“Deere”), the manufacturer of the tractor. After the case was removed to federal court, the district court granted summary judgment in favor of Deere. Although it concluded that there was sufficient evidence to go to a jury regarding whether the tractor was unreasonably dangerous, as that term is used in Oklahoma, it found that a sticker affixed to the tractor provided sufficient warning to cure the defects in the tractor’s design. We disagree, and reverse the grant of summary judgment. Mrs. McPhail also appeals the district court’s denial of her motion to remand the case to state court. She argues that Deere has failed to establish the amount in controversy and that the parties are no longer diverse because of the identification of “John Doe” defendants. We examine whether diversity jurisdiction exists, and conclude that it does.
I. BACKGROUND
Willis McPhail purchased a used 1981 Deere 4440 tractor at an auction approximately eight years ago. On the morning of his death, in June 2004, the tractor would not start, due to a malfunction in its starting circuitry. Because of the urgency of getting the hay crop in, Mr. McPhail decided to bypass start the tractor as a stopgap measure; the alternative was to lose the crop. Mr. McPhail was not the first farmer to do this. Plaintiff submitted evidence that significant numbers of farmers have resorted to bypass starting when the starting circuitry on their tractors failed.
See generally McMurray v. Deere & Co.,
858 F.2d 1436 (10th Cir.1988). When a tractor’s starting circuitry becomes inoperative, the tractor can still be started if the user creates a new electrical circuit that bypasses any failures in the normal circuitry. To effectuate a bypass start, the user stands on the ground near the engine and lays a screwdriver across the electrical terminals on the starter and the solenoid to close the circuit necessary for engine ignition. Mr. McPhail did this twice that morning. On his first attempt, the tractor started successfully, in neutral, but the front-end loader would not work. Mr. McPhail turned off the tractor and attempted to fix the front-end loader. Then, standing between the front and rear wheels on one side of the tractor, he attempted a second bypass start. This time, according to his employee Darrel Coffelt, the transmission engaged and the tractor suddenly lurched backward at high speed, running over Mr. McPhail. He died from the injuries later that day.
Mr. McPhail’s tractor was designed with a neutral-start switch, which prevents engine ignition if the transmission is in gear. But because this switch is part of the tractor’s normal starting circuitry, bypass starting the tractor also bypasses this safety mechanism. Thus, nothing in the design of the tractor protects users from the unexpected movement of the tractor if it is bypass started while in gear. In the 1970s and 1980s, Deere became aware that farmers were bypass starting their tractors, that tractors started in gear were also prone to sudden movement, and that this was leading to accidents. In response, for models manufactured after 1984, Deere began using an “Engagement Override Valve,” which prevents the transmission from engaging, even if the tractor is in gear, unless the tractor’s operator cycles the clutch or manually returns the tractor to neutral. For older models, Deere undertook a safety campaign, starting in
1989, to warn users of the dangers of bypass starting.
Deere first identified all of the pre-1984 models whose transmission could engage when the tractor is bypass started. For these tractors, Deere attempted to contact all known owners to educate them on the dangers of bypass starting and to install a safety kit on their tractors. The safety kits included a plastic cover for the tractor’s solenoid to prevent the operator from accessing the terminals used in bypass starting, and a warning sticker to affix near the solenoid cover. The sticker is a vibrant red and contains the following words: “DANGER” and “Start only from seat in park or neutral. Starting in gear kills.” The warning also contains two cautionary depictions. One illustration, contained within a prohibitory red circle with a slash through it, shows a screwdriver being placed over a solenoid terminal. The second image depicts a person standing on the ground near the tractor and being run over because the tractor is moving forward.
It is undisputed that this warning was present on Mr. McPhail’s tractor on the morning of the accident. However, there is no evidence that a Deere representative contacted Mr. McPhail, rather than a previous owner, as part of the education component of the campaign. There were remnants of the solenoid cover in place, suggesting it had been installed at some point. But it had been broken off prior to the accident and nothing in the record indicates whether the cover was present when Mr. McPhail bought the tractor at auction.
Two other features of the Deere 4440 are important. First, it employs a Quad-Range Transmission (QRT), which functions via a hydraulic (or “wet”) clutch. When a tractor with a QRT is started, hydraulic pressure must build up within the clutch before the tractor will move, whether or not the tractor is in gear. Because of this, the tractor’s engine will act as if it is in neutral for a short interval after it is started, even if it is in gear. Once the hydraulic pressure is adequate to engage the transmission, the tractor will suddenly move under full power. This is in contrast to the behavior of a manual (or “dry”) clutch. If a tractor with a “dry” clutch is started while the tractor is in gear, the transmission is engaged during engine ignition. The tractor will begin to move immediately but, because the engine has not fully started, the movement is less powerful.
A second significant feature, as explained by one of Mr. McPhail’s expert witnesses, is that “[t]he ‘Quad-Range’ transmission controls provide ‘false’ or ‘apparent’ neutral positions which can cause the operator to believe the tractor is in neutral, when it is not. The control system does not provide a positive visual or tactile indication of a true neutral position.” ApltApp., Vol. Ill, at 175. In other words, the gear shift on this model of tractor may appear to be in neutral when in fact the tractor is in gear.
II. DIVERSITY JURISDICTION
Mrs. McPhail originally filed this suit in Oklahoma state court. Invoking diversity jurisdiction, Deere removed the case to federal court. Mrs. McPhail challenged removal through a motion to remand. On appeal, she continues to assert that federal subject matter jurisdiction is lacking because the jurisdictional requirements of 28 U.S.C. § 1332(a) have not been met, for two independent reasons. First, she contends that the citizenship of unidentified (“John Doe”) defendants has destroyed diversity. Second, she contends that Deere never carried its burden to prove the requisite amount in controversy. Our review
of a denial to remand is de novo.
See Huffman v. Saul Holdings Ltd. P’ship,
194 F.3d 1072, 1076 (10th Cir.1999).
A. Diversity of “John Doe” Defendants
Mrs. McPhail first argues that the citizenship of unidentified defendants is a bar to removal. In her original complaint, Mrs. McPhail named three “John Doe” defendants who have since been identified as in-state distributors. Under 28 U.S.C. § 1332(a) the citizenship of all defendants must be different from the citizenship of all plaintiffs.
Strawbridge v. Curtiss,
7 U.S. (3 Cranch) 267, 2 L.Ed. 435 (1806);
Salt Lake Tribune Publ’g Co. v. AT & T Corp.,
320 F.3d 1081, 1095-96 (10th Cir.2003). Mrs. McPhail urges, therefore, that the identification of these defendants has destroyed diversity and that the case must be remanded. She also asserts that she has been unable to amend her complaint to substitute these named defendants because necessary information is contained within records controlled by Deere.
Under the federal removal statutes the presence of “John Doe” defendants at the commencement of an action creates no impediment to removal.
See
28 U.S.C. § 1441(a) (“[T]he citizenship of defendants sued under fictitious names shall be disregarded.”);
Australian Gold, Inc. v. Hatfield,
436 F.3d 1228, 1234-35 (10th Cir.2006). Thus, the diversity of citizenship between Deere and McPhail was sufficient to support the statutory diversity requirement at the time of removal. The question, then, is whether subsequent identification of potential defendants destroys complete diversity and requires remand to state court.
To be sure, if a non-diverse party is added to the complaint at any time prior to final judgment, the case must be remanded to state court. 28 U.S.C. § 1447(c) provides that “[i]f at any time before final judgment it appears that the district court lacks subject matter jurisdiction, the case shall be remanded.” Further, § 1447(e) states: “[i]f after removal the plaintiff seeks to join additional defendants whose joinder would destroy subject matter jurisdiction, the court may deny joinder, or permit joinder and remand the action to the State Court.” Although § 1447(e) speaks of joinder, it has been held to apply when the complaint is amended to replace “John Doe” defendants with defendants identified by name.
See Casas Office Mach., Inc. v. Mita Copystar Am., Inc.,
42 F.3d 668, 673-75 (1st Cir.1994) (noting legislative history indicating Congressional intent that this section apply to identification of “John Doe” defendants).
As § 1447(e) indicates, however, the plaintiff does not have an absolute right to join such parties. Federal Rule of Civil Procedure 15(a)(2) allows amendments only with leave of the opposing party or the court.
Further, under Rule 19 the district court must determine whether the party sought to be joined is indispensable. If so, Rule 19 requires the court either to join the party, in which case remand is necessary under § 1447(e), or to deny join-der, in which case Rule 19(b) also requires that the action be dismissed.
If the de
fendant is not indispensable, Rule 20(a)(2) permits joinder at the discretion of the district court.
See State Distrib., Inc. v. Glenmore Distill. Co.,
738 F.2d 405, 416-17 (10th Cir.1984). In exercising this discretion, the district court “typically considers several factors [including] whether the amendment will result in undue prejudice, whether the request was unduly and inexplicably delayed, [and whether it] was offered in good faith....”
Id.
at 416. If the district court determines that joinder is appropriate, § 1447(e) requires remand to state court. If the district court decides otherwise, it “may deny joinder.” 28 U.S.C. § 1447(e).
We have no occasion here to apply these principles to the district court’s decision, because Mrs. McPhail never attempted to amend her complaint. She informs us in her brief that she did not learn the identities of these potential defendants until shortly before the district court granted summary judgment, and states that her ability to add these individuals to her complaint was dependent on information controlled by Deere. Accordingly, as matters now stand, the complaint names parties who are wholly diverse, and there is no basis for remanding to state court. If Mrs. McPhail attempts to amend her complaint on remand, the district court will determine whether to grant leave to amend, whether any additional named parties are indispensable, and, if not, whether permissive joinder is appropriate. For now, Deere is the only named defendant. Having never been presented with a motion to amend the complaint to substitute non-diverse defendants in place of the unidentified defendants, the district court did not err in determining that it has diversity jurisdiction.
B. Amount in Controversy
Subject matter jurisdiction under 28 U.S.C. § 1332(a) requires, in addition to diversity of citizenship, an amount in controversy in excess of “$75,000, exclusive of interest and costs.” Mrs. McPhail asserts that Deere never established this amount in controversy and that the district court therefore erred by denying her motion to remand.
Determinations of sufficiency of the amount in controversy are governed by an odd set of rules. The evident purpose of diversity jurisdiction is to protect out-of-state defendants from possible home-town prejudice of jurors from the plaintiffs state. As James Madison explained in the Virginia ratifying convention:
It may happen that a strong prejudice may arise, in some states, against the citizens of others, who may have claims against them. We know what tardy, and even defective, administration of justice, has happened in some states. A citizen of another state might not chance to get justice in a state court, and at all events he might think himself injured.
3 Jonathan Elliot,
The Debates in the Several State Conventions on the Adoption of the Federal Constitution
533 (2d ed. 1836);
see also Bank of the United States v. Deveaux,
5 Cranch (9 U.S.) 61, 87-88, 3 L.Ed. 38 (1809) (Marshall, C.J.).
With this purpose in mind, one might expect the rules for determining sufficiency to be at least as favorable to the out-of-state as the in-state party. Alternatively, bearing in mind that federal jurisdiction is exceptional and that parties seeking to invoke federal jurisdiction must demonstrate that their case falls within its reach,
see, e.g., United States v. Hays,
515 U.S. 737, 743, 115 S.Ct. 2431, 132 L.Ed.2d 635 (1995), one might expect the burden of
proof to fall on the proponent of federal jurisdiction. Neither expectation proves entirely true. If an in-state plaintiff suing an out-of-state defendant wishes to be in federal court, all the plaintiff needs to do is allege an amount in excess of $75,000 and he will get his way, unless the defendant is able to prove “to a legal certainty” that the plaintiffs claim cannot recover the alleged amount.
See St. Paul Mercury Indem. Co. v. Red Cab Co.,
303 U.S. 283, 289, 58 S.Ct. 586, 82 L.Ed. 845 (1938);
Woodmen of World Life Ins. Soc’y v. Manganaro,
342 F.3d 1213, 1216-17 (10th Cir.2003). On the other hand, if the in-state plaintiff wishes to remain in state court, all it needs to do is to refrain from alleging any particular sum in its prayer for relief (assuming that is permitted, as it often is, under state rules of civil procedure), and, according to this and most other courts, the defendant is required to prove jurisdictional facts by a “preponderance of the evidence” such that the amount in controversy may exceed $75,000.
Martin v. Franklin Capital Corp.,
251 F.3d 1284, 1290 (10th Cir.2001) (citing cases). Thus, when the proponent of federal jurisdiction is the party that does not need it, mere allegations suffice; but when the proponent of federal jurisdiction is the party in whose interest diversity jurisdiction was created, actual proof of jurisdictional facts is required, at a stage in the litigation when little actual evidence is yet available. This set of rules bears no evident logical relation either to the purpose of diversity jurisdiction, or to the principle that those who seek to invoke federal jurisdiction must establish its prerequisites.
See generally
Alice M. Noble-Allgire,
Removal of Diversity Actions When the Amount in Controversy Cannot be Determined from the Face of Plaintiffs Complaint: The Need for Judicial and Statutory Reform to Preserve Defendant’s Equal Access to Federal Courts,
62 Mo. L.Rev. 681 (1997).
The requirement that a defendant must prove facts in support of the amount in controversy by a “preponderance of the evidence,” which derives from
McNutt v. General Motors Acceptance Corp.,
298 U.S. 178, 189, 56 S.Ct. 780, 80 L.Ed. 1135 (1936),
raises a further puzzle: in most removal cases, there is little “evidence” one way or another. In most cases, the defendant must file a notice of removal within thirty days after receiving the complaint.
See
28 U.S.C. § 1446(b). Pre-re-moval discovery in state court is unlikely to have produced helpful information by then. “Even if a defendant immediately were to serve interrogatories or requests to admit upon the plaintiff, many states give the opposing party thirty days or more in which to answer interrogatories. The defendant, therefore, either must hope that the plaintiff will respond well before the ordinary deadline or, in the alternative, must seek a court order to shorten the time period.” Noble-Allgire, 62 Mo. L.Rev. at 731 (footnote omitted). And if the plaintiff moves quickly to challenge removal in federal court, there may not be time to produce more evidence in federal discovery before the court decides to rule.
See
Fed.R.Civ.P. 16(b)(2) (the district court’s scheduling order, which sets forth the timing for disclosures, may be issued as late as ninety days after the defendant makes an appearance); Fed R. Civ. P. 26(f)(1) (the parties must meet to plan for discovery, at the latest, twenty-one days before the district court’s order under Rule 16(b)(2) is due). If the judge holds a
hearing under Rule 12(b)(1), the defense can present evidence, but may have little to work with.
Misunderstood, this system is liable to serious abuse. However, in a recent decision the Seventh Circuit has clarified this jurisprudential mess, explaining the proper role of the “preponderance of the evidence” standard, and how defendants may have a chance at satisfying it.
Meridian Security Ins. Co. v. Sadowski,
441 F.3d 536, 540-43 (7th Cir.2006) (Easterbrook, J.). The “preponderance of the evidence” standard applies to jurisdictional facts, not jurisdiction itself. As the court put it, “[w]hat the proponent of jurisdiction must ‘prove’ is contested factual assertions ... ^jurisdiction itself is a legal conclusion, a
consequence
of facts rather than a provable ‘fact’.”
Id.
at 540-541. Applying this rule to the same situation we confront here — a removing defendant claiming diversity jurisdiction in the face of a silent complaint — the court offered the following solution:
the removing defendant, as proponent of federal jurisdiction, must establish what the plaintiff stands to recover. We have suggested several ways in which this may be done — by contentions, interrogatories or admissions in state court; by calculation from the complaint’s allegations!;] by reference to the plaintiffs informal estimates or settlement demands!;] or by introducing evidence, in the form of affidavits from the defendant’s employees or experts, about how much it would cost to satisfy the plaintiffs demands. The list is not exclusive; any given proponent of federal jurisdiction may find a better way to establish what the controversy between the parties amounts to, and this demonstration may be made from either side’s viewpoint (what a judgment would be worth to the plaintiff, or what compliance with an injunction would cost the defendant). Once the estimate has been made — and contested factual allegations that support the estimate have been established in a hearing under Rule 12(b)(1) by admissible evidence ... — then the
St. Paul Mercury
standard comes to the fore, and the case stays in federal court unless it is legally certain that the controversy is worth less than the jurisdictional minimum.
Id.
at 541^42 (citations omitted). We also note that to the extent that a defendant must rely on the federal discovery process to produce this evidence (perhaps because there was no time to do so in state court) he may ask the court to wait to rule on the remand motion until limited discovery has been completed,
see, e.g., McCraw v. Lyons,
863 F.Supp. 430, 435 (W.D.Ky.1994) (“If Plaintiff seeks remand, Defendant may request that the Court delay a decision pending completion of limited discovery.”), although we do not decide under what circumstances a district court would abuse its discretion by refusing such a request.
Meridian
eliminates the double standard that would come from misunderstanding what “preponderance of the evidence” requires. The proponent of federal jurisdiction must prove contested
facts;
and because a defendant has no control over the complaint, he cannot put a large sum of money in controversy simply by demanding it, as a plaintiff often can. But once those underlying facts are proven, a defendant (like a plaintiff) is entitled to stay in federal court unless it is “legally certain” that less than $75,000 is at stake. If the amount is uncertain then there is potential controversy, which is to say that at least $75,000 is in controversy in the case.
Although this Court’s opinions have not been entirely clear on this issue, the ap
proach we adopt today is consistent with their holdings and analysis. In
Laughlin v. Kmart Carp.,
50 F.3d 871, 873 (10th Cir.1995), we held that where a state court complaint does not identify a specific amount that the plaintiff seeks to recover, the burden is on a defendant seeking removal to demonstrate that this jurisdictional prerequisite is satisfied by “affirmatively establishing]” in the petition that the amount in controversy exceeds the statutory requirement. The plaintiff in
Laughlin
alleged wrongful constructive discharge and sought damages in excess of $10,000.
Id.
at 872. The defendant attached the petition to its notice of removal, but failed to suggest any potential value of the claims in the petition.
Id.
at 873. Thus, we were presented with a petition and a notice of removal that both only referred to damages in excess of $10,000.
Id.
We found this insufficient to satisfy the amount in controversy requirement.
Id.
at 874.
In
Martin v. Franklin Capital Corp.,
251 F.3d 1284 (10th Cir.2001), the defendant’s notice of removal totaled up all of the dollar figures contained in the plaintiffs complaint, but some of these dollar amounts were essentially background information that were not linked to the plaintiffs attempts to recover damages. Because defendant’s notice of removal depended on an erroneous “construction of the [plaintiffs] pleading,” we found jurisdiction lacking.
Id.
at 1291. In the course of doing so, we stated that the defendant must “establish the jurisdictional amount by a preponderance of the evidence,” 251 F.3d at 1290. It would have been more precise to say that the defendant must affirmatively establish jurisdiction by proving jurisdictional
facts
that made it
possible
that $75,000 was in play, which the defendants in
Martin
failed to do. It is only the jurisdictional facts that must be proven by a preponderance — not the legal conclusion that the statutory threshold amount is in controversy. As explained in
Meridian:
To recap: a proponent of federal jurisdiction must, if material factual allegations are contested, prove those jurisdictional facts by a preponderance of the evidence. Once the facts have been established, uncertainty about whether the plaintiff can prove its substantive claim, and whether damages (if the plaintiff prevails on the merits) will exceed the threshold, does not justify dismissal. Only if it is “legally certain” that the recovery (from plaintiffs perspective) or cost of complying with the judgment (from defendant’s) will be less than the jurisdictional floor may the case be dismissed.
441 F.3d at 543.
In light of this clarification of the rule, we now consider whether Deere has proven the facts necessary to supports its assertion that this case may involve more than $75,000. Mrs. McPhail contends that because her complaint does not specify the total amount of damages sought, Deere did not carry its burden. However, a plaintiff cannot avoid removal merely by declining to allege the jurisdictional amount. This would allow frustration of the purpose of diversity jurisdiction, which is, after all, to protect the out-of-state defendant. Still, in the absence of an explicit demand for more than $75,000, the defendants must show how much is in controversy through other means.
First, the defendant may rely on an estimate of the potential damages from the allegations in the complaint.
See Meridian,
441 F.3d at 541. A complaint that presents a combination of facts and theories of recovery that may support a claim in excess of $75,000 can support removal. For example, the Fifth Circuit has upheld
removal based on a complaint seeking recovery in tort for “alleged damages for property, travel expenses, an emergency ambulance trip, a six day stay in the hospital, pain and suffering, humiliation, and her temporary inability to do housework after the hospitalization.”
Luckett v. Delta Airlines, Inc.,
171 F.3d 295, 298 (5th Cir.1999). The court made this determination by “[rjeading the face of the complaint,” even though it did “not specify the numerical value of the damage claim.”
Id.
Second, beyond the complaint itself, other documentation can provide the basis for determining the amount in controversy — either interrogatories obtained in state court before removal was filed, or affidavits or other evidence submitted in federal court afterward.
Meridian,
441 F.3d at 541-42;
Manguno v. Prudential Prop. & Cas. Ins. Co.,
276 F.3d 720, 723 (5th Cir.2002) (allowing “summary-judgment-type evidence,” such as affidavits, to support jurisdiction). This roughly parallels a plaintiffs right, under Fed.R.Civ.P. 10(c), to make “[a] copy of a written instrument that is an exhibit to a pleading ... a part of the pleading for all purposes.” In some situations, this sort of evidence alone will allow a defendant to support its claims regarding the amount in controversy. For example, where a defendant has allegedly breached a contract and the plaintiff seeks damages in an indeterminate amount, a defendant might support jurisdiction by attaching a copy of the contract, valued at more than $75,000, to the notice of removal. Or it might “introduc[e] evidence, in the form of affidavits from the defendant’s employees or experts, about how much it would cost to satisfy the plaintiffs demands.”
Meridian,
441 F.3d at 541-42 (citing
Rubel v. Pfizer Inc.,
361 F.3d 1016 (7th Cir.2004)).
Furthermore, a plaintiffs proposed settlement amount “is relevant evidence of the amount in controversy if it appears to reflect a reasonable estimate of the plaintiffs claim.”
Cohn v. Petsmart, Inc.,
281 F.3d 837, 840 (9th Cir.2002). Acknowledging that the use of a settlement offer would not be permissible at trial as evidence to establish “ ‘liability for or invalidity of the claim or its amount,’ ” the Ninth Circuit has held that it is permissible for a district court to consider settlement offers when deciding the jurisdictional question.
Id.
at n. 3 (quoting Fed.R.Evid. 408). We agree. The amount in controversy is not proof of the amount the plaintiff will recover. Rather, it is an estimate of the amount that will be put at issue in the course of the litigation. To this end, documents that demonstrate plaintiffs own estimation of its claim are a proper means of supporting the allegations in the notice of removal, even though they cannot be used to support the ultimate amount of liability.
In light of these principles, we think Deere has met its burden. As required under Oklahoma pleading rules, Mrs. McPhail did not plead any specific amount in damages.
See
Okla. Stat. tit. 12, § 2008(A)(2). Rather, she sought relief “in an amount in excess of Ten Thousand Dollars.” ApltApp., Vol. I, at 18. She also sought “actual damages, punitive damages as determined by the trier of fact, plus pre and post judgment interest, costs, attorney fees and such other and further relief which the Court deems appropriate,” again without citing any specific sum.
Id.
In its notice of removal, Deere cited the allegations in the complaint and represented that the amount in controversy in the case exceeded $75,000. Deere also incorporated several emails and letters that shows that counsel for Deere had a conversation with one of Mrs. McPhail’s attorneys that included a discussion of the value of the claim. ApltApp., Vol. II, at 28. Deere’s counsel interpreted comments
made during this discussion as meaning that Mrs. McPhail was stipulating that she would be seeking more than $75,000 in damages.
Id.
Mrs. McPhail’s attorneys consistently denied this concession, stating that any figures discussed between counsel were spoken informally “in general settlement discussions” and did not necessarily reflect the actual amount in controversy.
Id.
at 29. Counsel for Mrs. McPhail also wrote a separate letter refusing to concede an amount in controversy in excess of $75,000, inviting Deere’s attorneys to undertake their own “economic analysis” to ascertain the amount in controversy.
Id.
at 31. While Mrs. McPhail’s counsel stated in this correspondence that it “may very well be” true that the amount in controversy would exceed $75,000, he refused to stipulate to any amount other than the $10,000 required by Oklahoma law.
Id.
This information, which was presented to the district court with the notice of removal, is sufficient to support diversity jurisdiction. The complaint itself sets forth all of the underlying facts: Mrs. McPhail alleged that Deere’s defective design led to Mr. McPhail’s “severe bodily injuries” which in turn led to “permanent and progressive injury” and his “wrongful death.” Aplt.App., Vol. I, at 16. In addition, she also cited the Oklahoma wrongful death statute, and requested “all relief enumerated therein.”
Id.
at 17. That statute allows recovery for medical and burial expenses, the loss of consortium and the grief of the surviving spouse, the mental pain and anguish suffered by the decedent, pecuniary loss based on earning capacity, and grief and loss of companionship; it also permits an award of punitive damages.
See
Okla. Stat. tit. 12, § 1053. For actions found by a jury to exhibit “reckless disregard,” the jury has the discretion to award any amount up to $100,000 or the amount awarded in actual damages, whichever is greater.
See
Okla. Stat. tit. 23, § 9.1(B)(1).
Given these allegations and the nature of the damages sought, the complaint on its face may be sufficient by itself to support removal. We need not decide the question based on the complaint alone, however, because the correspondence between counsel that was incorporated in the notice of removal demonstrates that Mrs. McPhail’s attorneys also believed the amount in controversy “very well may be” in excess of $75,000. Although Mrs. McPhail’s counsel attempted to downplay this as a general figure discussed in “the context of settlement,” we agree with the Ninth Circuit that this is relevant evidence of the value of the claim. Even assuming the amount in controversy was not apparent from the face of the complaint, this background information provides enough supplementary information for the district court to conclude that it is not legally certain that Mrs. McPhail will recover an amount less than $75,000. The district court was therefore correct in concluding that Deere proved the facts necessary to establish that more than $75,000 was in controversy.
III. THE ADEQUACY OF THE WARNING
We turn now to the merits. Because jurisdiction is based on diversity, we apply the law of the forum state, in this case the Oklahoma law of products liability. A plaintiff may recover under a theory of strict products liability in Oklahoma if she can prove that: “1) the product was the cause of the injury; 2) a defect existed in the product at the time it left the manufacturer’s possession and control; and 3) the defect made the product unreasonably dangerous to the plaintiff or his property.”
McMurray v. Deere & Co., Inc.,
858 F.2d
1436, 1439 (10th Cir.1988) (applying Oklahoma law). Under Oklahoma law, the test for whether a product is unreasonably dangerous is whether the product is “ ‘dangerous to an extent beyond that which would be contemplated by the ordinary consumer who purchases it, with the ordinary knowledge common to the community as to its characteristics.’ ”
Id.
(quoting
Kirkland v. Gen. Motors Corp.,
521 P.2d 1353, 1362-63 (Okla.1974) (further internal quotation marks omitted).
However, even where a product’s design defect makes the product unreasonably dangerous, Oklahoma law does not impose liability if the product contains a warning that adequately addresses the known risks of use.
See Smith v. U.S. Gypsum Co.,
612 P.2d 251, 253-54 (Okla.1980). The Oklahoma Supreme Court has enunciated the relevant standard:
If a product is potentially dangerous to consumers, a manufacturer is required to give directions or warnings ... as to its use. If these warnings cover all foreseeable use and if the product is not unreasonably dangerous if the warnings and directions are followed, the product is not defective in this respect. If the warnings are unclear or inadequate to apprise the consumer of the inherent or latent danger, the product may be defective; particularly where a manufacturer has reason to anticipate danger may result from the use of his product and the product fails to contain adequate warning of such danger, the product is sold in a defective condition.
Id.
Further, Oklahoma law does not require a manufacturer “to foresee that [professionals who use its product will] fail to read its warnings, and then use [the product] in a manner that manufacturer’s instructions expressly warned against.”
Hutchins v. Silicone Specialties, Inc.,
881 P.2d 64, 67 (Okla.1993). Finally, “there is no duty to warn a knowledgeable user of the product of the dangers associated therewith.”
Duane v. Oklahoma Gas & Elec. Co.,
833 P.2d 284, 286 (Okla.1992).
The district court concluded that, had there been no warning on the tractor, a triable issue would exist regarding whether the tractor was unreasonably dangerous. Citing common failures in the starting circuitry, the need that this created for farmers to bypass start their tractors, the fact that the tractor was designed in such a way that bypass starting was easy to accomplish, and the potential for unexpected movement associated with bypass starting, the district court found that there was sufficient evidence in the record for reasonable jurors to find that the tractor was unreasonably dangerous. Deere does not challenge this conclusion.
We confronted a similar question in
McMurray v. Deere & Co., 858
F.2d 1436 (10th Cir.1988). In
McMurray,
the plaintiffs decedent had died when he attempted to bypass start his tractor while it was in gear.
Id.
at 1437-38. In that case, the district court had issued a jury instruction that required the jury to find in favor of the defendant if it found that bypass starting the tractor was an abnormal use of the product.
Id.
at 1438-39. Oklahoma law only allows such a defense “where the method of using a product is not that which the manufacturer intended or is a use that could not reasonably be anticipated by a manufacturer.”
Id.
at 1442 (quoting
Fields v. Volkswagen of Am., Inc.,
555 P.2d 48, 56 (Okla.1976)). We found that “Deere intended ... that the tractor would be used for agricultural purposes” and that Mr. McMurray, when he bypass started it, was “attempting to use [his tractor] for a proper purpose.”
Id.
at 1442-43. Accordingly, we held that Mr. McMurray did not use his “in an abnormal manner or in an unforeseeable way.”
Id.
at 1443. Since
McMurray,
Deere implemented its bypass start campaign, which resulted in the warning being placed on Mr. McPhail’s tractor.
Because this sticker was present, Oklahoma law also requires Mrs. McPhail to prove that the warning is “unclear or inadequate to apprise the consumer of the inherent or latent danger.”
Smith,
612 P.2d at 254. The district court held that Deere’s warning was adequate or, in the alternative, that Mr. McPhail’s awareness of the dangers of bypass starting rendered the warning unnecessary. Mrs. McPhail appeals the district court’s grant of summary judgment regarding the adequacy of the warning.
We agree with Mrs. McPhail that there are genuine issues of material fact that should have precluded the grant of summary judgment on this issue.
The sticker on Mr. McPhail’s tractor contains two sentences. First, it tells the operator of the tractor to “[s]tart only from seat in park or neutral.” Aplt.App., Vol. II, at 163. Second, it warns that “[smarting in gear kills.”
Id.
The warning also contains the word “DANGER” and an illustration of a person standing on the ground being run over by the tractor as it moves forward. •
Id.
We find that these cautionary phrases, taken in the light most favorable to the party opposing summary judgment, give rise to an issue of fact regarding the clarity and adequacy of this warning. Specifically, the sticker does not warn a user that the tractor’s transmission can engage even though the tractor appears to be in neutral. Instead, the warning expressly connects the danger of death to “starting in gear.” Although the first sentence instructs the operator to “[sjtart only from seat,” it also highlights the need to start in “park or neutral.” This implies that the dangers of bypass starting might be averted by making sure that the tractor is in park or neutral, not in gear, when it is started. Yet, Mrs. McPhail’s evidence makes it apparent that this is not the case.
Deere argues that if farmers simply complied with the first four words of the warning — “start only from seat” — they would not be able to bypass start and would not be at risk. True. But they might also not be able to use the tractor. The record shows that Deere is aware that normal use of the tractor sometimes makes it necessary to bypass start the tractor, which cannot be accomplished from the seat. Mr. Kirk Ney, a Division Engineer for Deere, acknowledged the need that farmers have to bring in their crops during harvest time, because it is tied to their livelihoods. ApltApp., Vol. Ill, at 64. He testified that some failures in the starting system are “easily fixable” and may be fixed by the tractor’s operator, instead of a technician, “depending on ... what’s wrong.”
Id.
at 63. He went on to state that “of course” it would be necessary to diagnose the problem.
Id.
Finally, he acknowledged that Deere’s service technicians are trained to diagnose starting circuitry problems by bypass starting tractors, and that even Deere service technicians had been injured while bypass starting tractors they were diagnosing.
Id.
at 78-81. This testimony, taken in the light most favorable to Mrs. McPhail, could lead a reasonable juror to conclude that Deere has anticipated that farmers will need to bypass start their tractors, and the warning fails to account for this need.
The district court decision does not take into account this key shortcoming in the warning: the inference that the danger arises from starting the tractor when it is not in neutral or park. Mrs. McPhail provided evidence that the behavior of the Quad-Range Transmission masks whether the tractor is in gear. The affidavit from Mrs. McPhail’s expert witness states that the gear selector for this model tractor could appear to be in neutral when the tractor was actually in gear. These characteristics of the tractor give rise to the inference that even an operator who takes care to ensure that the tractor is in neutral cannot be sure that unexpected movement will not occur. Mr. Coffelt testified that, after the accident, the tractor was traveling in reverse at “a pretty good rate of speed” and that he “couldn’t believe it, it was going backwards instead of forwards.” ApltApp., Vol II., at 102. These facts support an inference that the tractor was in a gear that Mr. McPhail had not deliberately selected.
The sticker affixed to Mr. McPhail’s tractor warned: “Danger. Start only from seat in park or neutral. Starting in gear kills.” This informs a tractor’s operator of the potentially lethal consequences of bypass starting a tractor in gear. However, reading the warning in the light most favorable to the plaintiff, it may also suggest that starting the tractor in neutral or park eliminates that danger. While the warning contains cautionary depictions regarding bypass starting generally and tractor motion caused by bypass starting, the warning does not caution against the possibility that even a tractor that appears to be in neutral or park may move under engine power. Indeed, by stating that “starting in gear kills,” it tends to suggest that if the user is sufficiently careful in ensuring that the tractor is in park or neutral, all will be well. Absent some warning about “false” or “apparent” neutral settings on this transmission, we are unable to find, as a matter of law, that this warning overcomes the dangerous conditions of the tractor.
The district court found, alternatively, that Mr. McPhail’s own knowledge of the dangerous condition of the tractor would make it impossible for Mrs. McPhail to prove the causation element necessary for recovery on a theory of products liability. However, the evidence in the record regarding Mr. McPhail’s knowledge relates to the dangers of bypass starting generally, and not to the specific dangerousness at issue here. That is, Mr. McPhail may have been aware that a tractor will move if it is bypass started while it is in gear. Nothing in the record shows that Mr. McPhail also knew that a tractor that appeared to be in neutral could, in fact, be capable of unexpected movement. Similarly, the evidence Deere presented regarding the knowledge of farmers in the community also addressed only the dangers of starting in gear, not the danger that a tractor could appear to be in neutral when in fact it was not. Mrs. McPhail has presented evidence that this model of tractor was capable of potentially lethal behavior that is not addressed by the tractor’s warning. This evidence creates a genuine factual dispute with regard to the adequacy of the warning, and makes summary judgment in favor of Deere inappropriate.
IV. CONCLUSION
The district court did not err in exercising jurisdiction. However, we find that there are genuine issues of material fact regarding whether the warning clearly and adequately addresses the tractor’s ability to be in gear and move under power even when it appears to be in neutral. Accordingly, we REVERSE the district court’s grant of summary judgment in favor of
Deere, and REMAND the case for further proceedings. Appellant’s motion to seal the appendix, Vol. Ill, is GRANTED.