The parties have agreed that this case may be submitted for decision on the briefs. See Fed.R.App.P. 34(f); 10th Cir.R. 34.1.2. The case is therefore ordered submitted without oral argument.

The trial court dismissed plaintiff Susan Williams' suit against the United States for lack of subject matter jurisdiction, holding that a federal flood control statute, 33 U.S.C. § 702c, immunizes the government from suit. We reverse and remand for further proceedings consistent with this opinion.

The facts of the case are largely undisputed. In June 1989, Ms. Williams' husband and two sons went fishing in the Verdigris River downstream of the Newt Graham Lock and Dam 18 of the McClellen-Kerr Arkansas River Navigation System. The Navigation System, including this lock and dam, is operated by the United States Corps of Engineers (COE). At the time in question, the dam's tainter gates were releasing approximately 12,000 cubic feet of water per second. This water posed no threat to the fishermen. In the course of moving a tug or tow boat with barges through the lock headed downstream, the lock operator released an additional 6,000 cubic feet per second of water across the original water flow. Ms. Williams' husband and two sons were swept away and drowned.

Ms. Williams sued the United States, alleging failure to warn the decedents adequately of the impending release of the 6,000 cubic feet of water per second to move boats through the lock. The government moved for dismissal based on lack of subject matter jurisdiction and failure to state a claim upon which relief can be granted or, in the alternative, for summary judgment. To support this motion, the government submitted the affidavit of a COE hydraulic engineer regarding the flow of water through the dam and lock on the date at issue. On this evidence, the trial court found that the government was immune from liability under the Flood Control Act ("the Act"), 33 U.S.C. § 702c, and it granted the government's motion to dismiss.

We review a dismissal for lack of subject matter jurisdiction de novo. Redmon ex rel. Redmon v. United States, 934 F.2d 1151, 1155 (10th Cir.1991). The issue before us is therefore whether or not the record establishes governmental immunity under 33 U.S.C. § 702c.1

The Supreme Court construed the immunity provision in United States v. James, 478 U.S. 597, 106 S. Ct. 3116, 92 L. Ed. 2d 483 (1986), holding that the government has broad immunity when it acts to control flood waters. The court stated that "[t]he Act concerns flood control projects designed to carry floodwaters. It is thus clear from § 702c's plain language that the terms 'flood' and 'flood waters' apply to all waters contained in or carried through a federal flood control project for purposes of or related to flood control...." Id. at 605, 106 S.Ct. at 3121. This court applied the James analysis in Boyd v. United States ex rel. United States Army Corps of Engineers, 881 F.2d 895 (10th Cir.1989), holding that the government is immune from prosecution if there is a "requisite nexus" or "necessary link between flood control activities and injuries sustained...." Id. at 900. For the purpose of deciding Boyd, we explicitly assumed that the lake in question "was created by the COE for flood control purposes." Id. at 899. Neither James nor Boyd considers what we find to be the threshold question in the instant case: Is there "a flood control project" that triggers the Act? Only if this question is answered affirmatively can we proceed to the issue of a nexus between flood control activities and the injury.

Other circuit courts have not explicitly defined "flood control project," but their analyses in similar cases begin by considering the character and purpose of the dam or levee involved. In finding that a lake was a flood control project, the Ninth Circuit cited the statute pursuant to which a dam was built and the fact that the COE monitored the lake's water level daily. McCarthy v. United States, 850 F.2d 558, 559 (9th Cir.1988), cert. denied, 489 U.S. 1052, 109 S. Ct. 1312, 103 L. Ed. 2d 581 (1989). (We note an unpublished decision following McCarthy, in which plaintiffs have petitioned for certiorari to the Supreme Court. Hiersche v. United States, 933 F.2d 1014 (9th Cir.1991), petition for cert. filed, 60 U.S.L.W. 3442 (U.S. Dec. 17, 1991) (No. 91-774).) The Third Circuit used the same indicia to find a flood control project in Dawson v. United States, 894 F.2d 70, 71 (3d Cir.1990). The Eighth Circuit also relied on the statutory basis for building a dam in finding that "[f]lood control was an essential component of the multiple purpose project" that expanded the lake in which an injury had occurred. DeWitt Bank & Trust Co. v. United States, 878 F.2d 246, 247 (8th Cir.1989), cert. denied, 494 U.S. 1016, 110 S. Ct. 1318, 108 L. Ed. 2d 493 (1990). The Fifth Circuit examined the purpose of dredging a flotation channel in finding that the channels, and therefore their dredging, "were inescapably part of a flood control project." Mocklin v. Orleans Levee Dist., 877 F.2d 427, 430 (5th Cir.1989). See also Fryman v. United States, 901 F.2d 79, 82 (7th Cir.), cert. denied, --- U.S. ----, 111 S. Ct. 295, 112 L. Ed. 2d 249 (1990) (endorsing broad governmental immunity for injuries such as drowning where flood-control project and activities increase probability of their occurrence).

We begin our analysis as these other circuit courts have, with the dam and lock themselves, and with the understanding that "Congress clearly sought to ensure beyond doubt that sovereign immunity would protect the Government from 'any' liability associated with flood control." James 478 U.S. at 608, 106 S.Ct. at 3122. In adopting this analysis, we explicitly reject the approach that the government and trial court took when they asked first and foremost whether the water involved in the injury was floodwater, either at the moment of injury or at any prior time. Immunity under the Act as interpreted in James depends not on the character or origin of the water, but on the purpose of the project and the nature of the activity creating the nexus with the injury. In the words of the James opinion, "The Act concerns flood control projects designed to carry floodwaters." James 478 U.S. at 605, 106 S. Ct. at 3121 (emphasis added).

An approach that elevated the character of the water over the purpose of the dam or levee would lead to absurd results. For example, a drop of water at the head of the Colorado River that was impounded in or that passed through a flood control project would thereby become floodwater, and would immunize the government from all liability for injury caused by that drop of water or any water with which it commingled, all the way from the headwaters to the Gulf of Mexico. By this approach the government would also be immune from liability for injuries caused by any activities taking place downstream of a flood control project, no matter what the activities and how negligently performed. This logic would provide immunity without analysis beyond finding the drop of water passing through a flood control project. There is no evidence in this record or in the cases cited that Congress intended the Flood Control Act to sweep so broadly or be applied so indiscriminately. Although James has been criticized for overreaching with its focus on the facility, see Fryman, 901 F.2d at 81 (positing absurd results at the fringes of the purpose-of-the-facility analysis), we apply its logic as clearly superior to and more rational than the proposed alternative.

Examining the record before us, we look first for evidence in the record that the dam and lock are a flood control project triggering the Act. The government relied almost exclusively on the affidavit of a COE hydraulic engineer on this point. This reliance is misplaced, however, as the affidavit contains a flat statement that "the lock and dam ... has [sic] no flood control capabilities." (Copley Aff. p 6 (emphasis added).)

The statutes pursuant to which the facility was built do not support finding this to be a flood control project, because they refer in the disjunctive to projects built for navigation or flood control. River and Harbor Act, 33 U.S.C. § 603a (1988) (amended 1945) (emphasis added); Flood Control Act of 1950, 33 U.S.C. § 701f-2 (1988) (referring to "river, harbor, or flood-control works") (emphasis added). The rules and regulations governing public use of the waters at issue here actually support a conclusion contrary to the government's argument, because they refer not to flood waters or flood control, but to "navigable reservoir areas" (36 C.F.R. § 311 (1971) (replaced 1973 by 36 C.F.R. § 327)), "water resource development projects" (id.), and a "navigation system" (33 C.F.R. § 207.275 (1991)). Finally, the statutory list of projects subject to federal flood control regulations does not include this facility. 33 C.F.R. § 208.11 (1991). The federal government has thus apparently decided to regulate the facility as a navigation system rather than as a flood control project.

On this record, we conclude that the dam and lock, which by the government's admission have "no flood control capabilities," were built for navigation, not flood control. This facility therefore does not trigger the Flood Control Act.

The COE affidavit describes facilities upstream of this lock and dam as "flood control projects," and characterizes the water that caused the injury in this case as originating partly from "flood control releases from upstream projects." (Copley Aff. p 6.) Projects upstream of the facility in question might qualify to trigger the Act if the proper record were made, but the mere presence of a qualifying project upstream of the injury would not immunize all governmental activities downstream. There must not only be a qualifying project to trigger the Act, but there must also be a nexus between the operation of that project and the downstream injury to invoke immunity.

While the record in this case (based on the bald but unrefuted affidavit of the COE) establishes that the upstream facilities are qualifying projects, there is no evidence whatever that the activity causing the injury--the increased release of water moving the boat and barges through the lock--was in any way connected with the upstream projects.2 The government's mere assertion that the water in which plaintiff's decedents drowned had passed through upstream flood control facilities is insufficient to invoke immunity; in fact, under the analysis described here, it misses the boat entirely.

We hold that the record before us fails to trigger the Flood Control Act, and as a result, the government does not have the benefit of immunity through the Act. We therefore hold that subject matter jurisdiction lies with the trial court, which may proceed to examine the merits of this case.

We REVERSE and REMAND for further proceedings consistent with this opinion.