The railroad bridge across the upper Mississippi River near Sabula, Iowa, is more than a century old. (The bridge is just south of Savanna, Illinois, where Route 52 crosses the river.) Authorized by the Secretary of War in 1880, built in 1881, and substantially renovated between 1905 and 1908, the bridge has a deck too low for river traffic to pass. Transit is through a pin-connected swingspan, which, when river traffic approaches, rotates 90 on the pivot to produce two channels, each 154 feet wide. The river's shipping channel is 300 feet, so the bridge effects a substantial constriction. To pass the bridge a towboat and its fleet of barges must line up perfectly, even when wind and currents combine to push the assembly off the centerline.

On May 5, 1997, the M/V Megan Beesecker was moving downriver with a tow of 12 barges, arranged 3 across and 4 deep. The tow was 105 feet wide, and the assembly (including the towboat) was 780 feet long and weighed about 21,000 tons. This is a common size; 15-barge fleets (lashed 3 5) also are frequent. Pilot James Jarvis planned to use the western (or Iowa) channel, but a wind from the northwest pushed the tow off its centerline. Jarvis then tried to maneuver the tow through the eastern (or Illinois) channel but entered at an angle, and the barges hit the bridge. Admiralty calls this an allision (distinguished from a collision, in which moving vessels meet). I&M Rail Link, the bridge's owner, filed this admiralty action demanding compensation for the damage to the bridge; Northstar Navigation, which operates the Megan Beesecker, filed a counterclaim seeking compensation for the damage to the barges and towboat. The district court granted summary judgment against Northstar, applying the principle of The Oregon, 158 U.S. 186, 192-93 (1895), and The Louisiana, 70 U.S. (3 Wall.) 164, 173 (1865), that a vessel is presumptively at fault when it hits a stationary object. Northstar sought to rebut this presumption by arguing that the Sabula Bridge is an unreasonable obstruction to navigation. See The Pennsylvania, 86 U.S. (19 Wall.) 125 (1873). The permit, issued in 1880 under what is now 33 U.S.C. sec.401, was amended as recently as April 1996, so the Sabula Bridge cannot be condemned out of hand. But 33 U.S.C. sec.512, which provides that "[n]o bridge shall at any time unreasonably obstruct the free navigation of any navigable waters of the United States", applies to all bridges, not just those built without permits. The Coast Guard, which today administers both sec.401 and sec.512, see 49 C.F.R. sec.1.46(c)(3), (5), has found that the Sabula Bridge unreasonably obstructs navigation because its channels are too narrow. A report issued late in 1995 sums up:The horizontal opening of 154 feet does not provide adequate clearance for contemporary commercial tows. This leaves about 24 feet of clearance on each side of the tow when transiting the bridge. This requires commercial tows to be perfectly aligned with the opening when approaching and transiting the bridge to avoid contacting the bridge. This is not what the navigational characteristics of the waterway dictate. If the bridge provided adequate horizontal clearance, tows transiting this reach of the river would assume the angular characteristics required of a turn. A standard tow would occupy a width of at least 172 feet in order to steer the turn through the railroad bridge.

The permit requires the owner to renovate or even remove the Sabula Bridge if necessary to permit unhindered movement on the river. On June 17, 1996, the Commandant of the Coast Guard took action based on the 1995 report. The Commandant "determined that the [Sabula Bridge] is an unreasonable obstruction to navigation" and directed its owner to reconstruct the bridge so that

The movable span shall provide a horizontal clearance of no less than 300 feet measured normal to the channel and a vertical clearance of no less than 52 feet above the two percent flowline or 60 feet above normal pool, whichever is greater, in the open position and 6 feet above the two percent flowline or 18 feet above normal pool in the closed position. These clearances are necessary for the reasonable needs of navigation.

This order was issued more than ten months before the Megan Beesecker and her tow struck the Sabula Bridge. Yet to this day I&M Rail Link has neither sought judicial review of the Commandant's order nor taken the first step to comply with it (which, according to 33 U.S.C. sec.514, is the preparation and submission of a plan to alter the bridge).

In the district court, I&M Rail Link argued that the Coast Guard's finding that the Sabula Bridge is an unreasonable obstruction to navigation should be ignored, because sec.512 is part of the Truman-Hobbs Bridge Act of 1940, 33 U.S.C. sec.sec. 511-23. An order issued under this Act leads to a cost-sharing proceeding, and the United States has promised to bear part of the reconstruction expense "attributable to the necessities of navigation." 33 U.S.C. sec.516. The district court agreed with I&M Rail Link's position that the only significance of the Commandant's order (and the findings on which it was based) is to facilitate a draw on the Treasury. On this interlocutory appeal under 28 U.S.C. sec.1292(a)(3) from the grant of summary judgment to I&M Rail Link, the principal question is whether the district court's understanding of the Coast Guard's report and decision is correct.

All bridges with piers in the water or decks lower than the tallest ship obstruct navigation. Usually the degree of obstruction is reasonable in relation to the commercial benefits of the bridge. Sometimes, however, changes in the nature of water-borne commerce or the conditions of the waterway render unreasonable an obstruction that was justified when built. Congress promised that, when such a change occurs, the United States will bear part of the renovation expense. But that is not all the Act does. Section 512 forbids maintenance of a bridge that has become an unreasonable obstruction, and sec.519 imposes civil and criminal penalties on bridge owners that fail to comply with a renovation order. What is more--and what is more important--the Act supplies funding only if a bridge has become an unreasonable obstruction to navigation. I&M Rail Link treats the unreasonable-obstruction finding as a matter of form, a kind of administrative wink and nudge needed to liberate public funds. But we have no reason to think that Congress contemplated fraudulent findings; both the structure of the Act and the evident seriousness with which the Coast Guard takes its task show that public funds areavailable only if the bridge really is an unreasonable obstruction to navigation.

Perhaps the Truman-Hobbs Act as a whole was a concession to bridge owners, a victory for an interest group, but like most statutory packages it is a complex deal, with some features making life easier for the proponents and other features attaching prices. See Jerry L. Mashaw, Greed, Chaos, & Governance: Using Public Choice to Improve Public Law 81-105 (1997); Daniel A. Farber & Philip P. Frickey, Law and Public Choice: A Critical Introduction 88-115 (1991). Here a subsidy comes hand in hand with a finding that the bridge is an unreasonable obstruction to navigation. Such a finding has many consequences; one may be an effect on negligence litigation under the admiralty jurisdiction.

If the Coast Guard's finding were an unelaborated ukase, it would do Northstar little good in rebutting The Oregon's presumption. But the Coast Guard gave several reasons, and these reasons have independent significance for the suit. One is that, when the bridge was built, vessels moving on the river were smaller. A 154-foot channel was ample. Today, with wider, longer, and heavier vessels, the channel is too narrow. With a smaller vessel, the 154-foot span is adequate even if the pilot makes an error; today only precise navigation allows a transit without incident. The Coast Guard's report observes that towboats must maneuver near the bridge until wind and current, coupled with the vessel's momentum, permit a transit. That takes time, which increases the expense of navigation. Sometimes the towboat just can't do the job alone; power must be applied at both ends of the tow to keep the assembly straight. The Coast Guard added that "approximately 90% of the river pilots use helperboats at this bridge on some occasions." Helperboats, which are small tugboats, are costly and add to the delay in passing the bridge.

The Sabula Bridge is struck repeatedly. For the 11-year period 1980 through 1990, the bridge was hit at least 63 times. (Some allisions are not reported, so the Coast Guard views this as a conservative estimate.) There were 27,663 recorded commercial tows during this period, so at least 1 in every 439 transits produced an allision. I&M Rail Link says that this shows that the bridge is safe; with careful navigation a pilot can transit without incident. But the Coast Guard drew a different inference: it concluded that a bridge should have a much lower accident rate. What would one think of a highway bridge struck by one out of every 500 passing cars? Or an airport designed so that one of every 500 landings produced an accident? What if one of every 439 trains passing over the Sabula Bridge derailed? An accident rate of 1 in 439 is too high for river transportation, the Coast Guard concluded. Doubtless some of these allisions could have been avoided by prudent navigation, but others must be chalked up to the fact that the bridge was designed for navigation in a different era.

Perhaps there would be less reason for concern if the Sabula Bridge were the only tight squeeze, and all other bridges had 300-foot spans to match the main navigation channel. But they don't. The degree of obstruction to navigation depends on the risks and costs of navigating the river from origin to destination. A series of bridges with a design similar to the Sabula Bridge may well pose costs substantially greater than the expense of renovation, and thus represent unreasonable obstructions to navigation in the sense that the value of precautions exceeds their discounted costs. See United States v. Carroll Towing Co., 159 F.2d 169, 173 (2d Cir. 1947); cf. Reardon v. Peoria & Pekin Union Ry., 26 F.3d 52 (7th Cir. 1994). The Sabula Bridge is the third-most-frequently struck on the Mississippi: it was hit at least 204 times from 1972 through 1996; the two "leaders" were hit 481 and 350 times during this period, and three other bridges were struck at least 100 times during these years. Other bridges are hit rarely if ever. It is unlikely that all of these allisions can be attributed solely to pilots' negligence. Calling the most-frequently-struck bridges unreasonable hazards to navigation makes a good deal of sense.

If the Coast Guard may find the Sabula Bridge an unreasonable obstruction based on the cost and accident data, then so may the trier of fact in admiralty, where Learned Hand's famous Carroll Towing formula for negligence originated. Findings in the Coast Guard's report are more than adequate to overcome The Oregon's presumption. See Folkstone Maritime, Ltd. v. CSX Corp., 64 F.3d 1037, 1050-51 (7th Cir. 1995). Perhaps Jarvis should have reversed the Megan Beesecker's engines rather than attempting to use the Illinois channel; perhaps he blundered earlier when using the current to line up for the transit; but perhaps this was one of the allisions that are to be expected given the clash between an 1880 design and the nature of modern traffic on the Mississippi. See Charles Parrow, Normal Accidents: Living with High-Risk Technologies (rev. ed. 1999). The trier of fact must give an answer without resort to presumptions. Although the Coast Guard's findings may well be conclusive for some purposes under Monongahela Bridge Co. v. United States, 216 U.S. 177 (1910), the question remains whether the shortcomings of the bridge caused this accident. Because admiralty uses a comparative-fault regime, the possibility that Jarvis and the bridge's design each bear some responsibility cannot be overlooked. The judgment is reversed, and the case is remanded for trial.