dissenting.
I conclude that the district court did not commit clear error in dismissing the complaint, and I respectfully dissent.
To my mind, this is a straightforward case. When plaintiff Capitol Indemnity Corporation failed even to allege a basis for federal jurisdiction in its first complaint, the district court allowed Capitol to amend its complaint, but directed it to submit sufficient evidence to demonstrate diversity of citizenship within ten days, at which point the court would resolve the matter. This is a perfectly “rational mode of inquiry” in accord with our precedents, see Osborn v. United States, 918 F.2d 724, 730 (8th Cir.1990), and the court does not contend otherwise.
In response to the district court’s directive, Capitol submitted a two-page affidavit, which contained only a conclusory allegation concerning Capitol’s principal place of business (on which the court properly declines to rely, ante at 835-836), and three relevant statements of fact: (1) “Capitol is incorporated in the State of Wisconsin,” (2) “Capitol does not maintain a place of business in Arkansas,” and (3) “Capitol’s business of writing insurance policies and surety bonds in the State of *838Arkansas is conducted entirely through the use of independent agents who are not employees of Capitol.” The district court concluded that the evidence-presented by Capitol was insufficient to establish diversity of citizenship. I agree with that conclusion, and the court does not contend otherwise. That should end the analysis.
The court, however, proceeds to find that the district court committed clear error because it failed to “look beyond the evidence” and draw on its “general knowledge of commonly known information,” ante at 836, even though Capitol never cited or argued any such “commonly known information” to the district court or to this court. The court concludes that this general knowledge is sufficient, under the newly adopted “total activities” test for determining a principal place of business, to prove that Capitol’s principal place of business cannot be in Arkansas. I disagree with this analysis.
I accept, of course, the general proposition that a court may take judicial notice of certain adjudicative facts, including those “generally known within the territorial jurisdiction of the trial court.” Fed.R.Evid. 201(b). But a district court is required to take judicial notice of an adjudicative fact only when “requested by a party and supplied with the necessary information.” Fed.R.Evid. 201(d). Given that Capitol never asked the district court to consider “general knowledge of commonly known information,” and never supplied any necessary information in that regard, it was within the district court’s discretion whether to consider such information on its own initiative, and it was not required to do so. Fed.R.Evid. 201(c); FDIC v. Houde, 90 F.3d 600, 607-08 (1st Cir.1996). The court should not reach out for facts necessary to establish diversity jurisdiction, because “[jjudicial notice is an inappropriate device for remedying a failure of proof.” Glover v. Cole, 762 F.2d 1197, 1200 n. 6 (4th Cir.1985); see also United States v. Hawkins, 76 F.3d 545, 551 (4th Cir.1996).
Assuming for the sake of argument that the district court was required sua sponte to consider “general knowledge” about insurance companies, the scope of what may be judicially noticed about an insurance company is limited. The court is forced by the paltry record to “presum[e]” that Capitol’s senior claims examiner (who served as an affiant in the district court) is an employee and not an independent agent, ante at 836, but it is entirely possible that an insurance company would use independent contractors as claims examiners. I daresay it is “commonly known” that the insurance industry in the 21st century increasingly is outsourcing such functions as claims processing and administration, underwriting, accounting, and information technology services. See, e.g., Doug McPhie, Thumbs up on Outsourcing, CROSS CURRENTS, Summer 2002, at 12-14, 21 (recognizing growing trend in life insurance industry toward outsourcing of information technology and business processes, and citing examples of outsourced policy administration functions, such as underwriting and claims adjudication), http://www.ey.com/global/download.nsfi' Belgium_E/Cross Currents_Summer_2002 /$file/CrossCurrents_Summer_2002.pdf.1 These industry developments show that *839one cannot reliably make broad assumptions about the number and function of employees at an insurance company. Moreover, the presumed existence of certain core employees, managers, and directors does not tell us anything about where they are located, including whether they are in a single state or dispersed among different venues. While it may be possible to infer from the barebones affidavit and common knowledge that Capitol has some sort of “home office” outside Arkansas, much beyond that is left to speculation.
More problematic is the legal conclusion that “general knowledge” about insurance companies necessarily demonstrates diversity of citizenship in this case. The mere existence of a judicially noticed home office outside Arkansas, especially when the nature of such an office is speculative, does not establish that Capitol’s principal place of business is outside Arkansas. The very cases cited by the court to illustrate the “total activities” test, ante at 836, demonstrate the point.
As the court observes, Capitol failed to prove that it sold insurance policies anywhere other than Arkansas. The court’s authorities agree that “[w]hen virtually all of the corporate business is conducted in one state, but the headquarters and policy-making functions are conducted in another, the situs of the corporate business assumes greater importance.” Associated Petroleum Producers, Inc. v. Treco 3 Rivers Energy Corp., 692 F.Supp. 1070, 1074 (E.D.Mo.1988). If, for example, Capitol were incorporated in Wisconsin for the purpose of selling insurance exclusively in Arkansas (a scenario that the record and the court’s judicially noticed facts do not exclude), decisions applying the total activity test indicate that the company’s principal place of business would be Arkansas. See id. at 1075 (where sole officers and employees, company headquarters, policy-making functions, and place of incorporation were in Missouri, but 92% of sales were in Kentucky, principal place of business was Kentucky); North Star Hotels Corp. v. Mid-City Hotel Assocs., 696 F.Supp. 1265, 1270 (D.Minn.1988) (where Texas was plaintiffs state of incorporation, residence of officers and directors, situs of bookkeeping and accounting tasks, and location of corporate offices where policy decisions were made and records were maintained, but sole source of income was hotel in Minnesota managed and operated through management agreement, principal place of business was Minnesota); White v. Halstead Indus., Inc., 750 F.Supp. 395, 399 (E.D.Ark.1990) (where company’s executive offices were in North Carolina, but majority of goods, sales, and employees were associated with two plants in Arkansas, principal place of business was Arkansas); Hanna Mining Co. v. Minnesota Power & Light, 573 F.Supp. 1395, 1400 (D.Minn.1983) (where company was created to hold and operate parent’s interest in Minnesota mining project, principal place *840of business was Minnesota rather than state where executive and administrative offices were located), aff'd, 739 F.2d 1368 (8th Cir.1984). Given the paucity of evidence in the record about Capitol, even when it is augmented by the judicially noticed existence of a home office outside Arkansas, I do not find clear error in the conclusion that Capitol failed to prove diversity of citizenship.
As the district court observed, it may well be that Capitol could demonstrate easily that its principal place of business is outside Arkansas. It should remain free to do so in other litigation. But Capitol failed to make the requisite showing in this case, and I see no good reason to strain both the doctrine of judicial notice and the “total activities” test to create diversity jurisdiction over this lawsuit. I would affirm the judgment of the district court, and I respectfully dissent.
. See also, e.g., Chris Pryer, Outsourcing to Play Larger Role Among Insurance Companies, OUTSOURCING JOURNAL, January 2003 (describing outsourcing of insurance processes, including field-based services such as auditing; policy administration and billing; financial recoveries, which comprises subrogation and premium collection; and data management and regulatory reporting), at http://www.outsourcing-journal.com/ issues/jan2003/insurance.html; William R. Pape, The Fewer the Merrier, INC. MAGA*839ZINE, Dec. 1998 (reporting that in 1998, outsourcing allowed GeneraLife Insurance Company of America to employ a staff of only 13, some of them college students, to manage more than 3,000 independent insurance agents), http://www.inc.com/maga-zine/19981201/5420.html; Florida Dep’t Ins., 2001 Property and Casualty Target Market Conduct Examination of DeSoto Insurance Company, at 4 (March 8, 2002) (insurance company relied on third-party managing general agent to provide underwriting, production, marketing, policy issuance, premium billing and collection, premium accounting, claims adjusting, auditing, payroll issuance and various other services, while managing general agent subcontracts the policy administration and claims administration functions to third-party servicing providers), http://www.fldfs.com/companies/pc/Exams/ Desoto_030802_Rpt.Pdf.