PRECEDENTIAL

     UNITED STATES COURT OF APPEALS
          FOR THE THIRD CIRCUIT
               _____________

     No. 16-3470, 16-3552, 16-3867 & 16-3868
                 _____________

    IN RE: PETITION OF FRESCATI SHIPPING
COMPANY, LTD., AS OWNER OF THE M/T ATHOS
 I and TSAKOS SHIPPING & TRADING, S.A., AS
MANAGER OF THE ATHOS I FOR EXONERATION
      FROM OR LIMITATION OF LIABILITY
              (E.D. Pa. No. 2-05-cv-00305)

         UNITED STATES OF AMERICA

                        v.

 CITGO ASPHALT REFINING COMPANY; CITGO
   PETROLEUM CORPORATION; CITGO EAST
          COAST CORPORATION
         (E.D. Pa. No. 2-08-cv-02898)



CITGO Asphalt Refining Company; CITGO Petroleum
  Corporation; CITGO East Coast Oil Corporation,
                Appellants in Nos. 16-3470; 16-3552
    Frescati Shipping Company, Ltd.; Tsakos Shipping and
                        Trading, S.A.,
                      Appellants in No. 16-3867

                  United States of America
                    Appellant in No. 16-3868
                      _____________

       On Appeal from the United States District Court
            for the Eastern District of Pennsylvania
      District Court Nos. 2-05-cv-00305; 2-08-cv-02898
       District Judge: The Honorable Joel H. Slomsky

                 Argued November 8, 2017

         Before: SMITH, Chief Judge, HARDIMAN,
         Circuit Judge, and BRANN, District Judge*

                  (Filed: March 29, 2018)



Timothy J. Bergère        [ARGUED]
Alfred J. Kuffler
John J. Levy              [ARGUED]
Montgomery McCracken Walker & Rhoads

*
 The Honorable Matthew W. Brann, United States District
Judge for the Middle District of Pennsylvania, sitting by
designation.
                             2
123 South Broad Street
24th Floor
Philadelphia, PA 19109

Eugene J. O’Connor
Montgomery McCracken Walker & Rhoads
437 Madison Avenue
29th Floor
New York, NY 20022

Jack A. Greenbaum
22782 Buendia
Mission Viejo, CA 92691
      Counsel for Frescati Shipping Co. Ltd. and
        Tsakos Shipping and Trading S.A.

Matthew M. Collette
United States Department of Justice
Civil Division
Room 7212
950 Pennsylvania Avenue, N.W.
Washington, DC 20530

Stephen G. Flynn
United States Department of Justice
Torts Branch, Civil Division
P.O. Box 14271
Washington, DC 20044

Anne Murphy                    [ARGUED]
United States Department of Justice
                           3
Appellate Section
Room 7644
950 Pennsylvania Avenue, N.W.
Washington, DC 20530
      Counsel for United States of America

Benjamin Beaton
Jacqueline G. Cooper
Carter G. Phillips             [ARGUED]
Richard E. Young
1501 K Street, N.W.
Washington, DC 20005
      Counsel for Citgo Asphalt Refining Co.
         Citgo Petroleum Corp.
         Citgo East Coast Oil Corp.

George R. Zacharkow
Deasey Mahoney & Valentini
1601 Market Street
Suite 3400
Philadelphia, PA 19103
       Counsel for Intervenor Respondent
         International Liquid Terminal Ass’n
         American Fuels and Petrochemicals
          Manufacturers Ass’n




                           4
                           ________________

                               OPINION
                           ________________

SMITH, Chief Judge.

                            Table of Contents

I.    Introduction .......................................................... 6
II.   Background........................................................... 7
   a.   Facts ................................................................... 7
   b.   Procedural History ......................................... 13
III.  Jurisdiction and Standard of Review ............... 19
IV.   The Safe Berth Warranty .................................. 19
   a.   The Draft of the Athos I ................................. 21
   b.   Frescati’s Seamanship ................................... 27
V.    Wharfinger Negligence ...................................... 36
VI.   Subrogation and Equitable Recoupment......... 42
   a.   Subrogation and Subrogee-Specific
        Defenses ........................................................... 45
  b.    Equitable Recoupment .................................. 49
VII. Limitation of Liability under the Oil Pollution
      Act ........................................................................ 55
VIII. Prejudgment Interest Rate ................................ 58
IX. Conclusion........................................................... 61




                                         5
 I.   Introduction

       After a 1,900-mile journey from Venezuela to
Paulsboro, New Jersey, the M/T Athos I, a single-hulled
oil tanker, had come within 900 feet of its intended berth
when it struck an abandoned anchor on the bottom of the
Delaware River. The anchor pierced the Athos I’s hull,
causing approximately 264,000 gallons of crude oil to spill
into the river.

        The cost of cleaning up the spill was $143 million.
We are presented with the question of how to apportion
responsibility for that cost between three parties. The first
party comprises not only the shipowner, Frescati Shipping
Company, Ltd., but also the ship’s manager, Tsakos
Shipping & Trading, S.A. (collectively, “Frescati”).
Frescati, through an intermediary, contracted to deliver
crude oil to the second party, which is made up of several
affiliated    companies—CITGO          Asphalt      Refining
Company, CITGO Petroleum Corporation, and CITGO
East Coast Oil Corporation (collectively, “CARCO”). The
oil shipment was to be delivered to CARCO at its marine
terminal in Paulsboro. After the oil spill, Frescati paid for
the cleanup effort, and was eventually reimbursed $88
million by the third party to this litigation, the United
States, pursuant to the Oil Pollution Act (OPA) of 1990,
33 U.S.C. § 2701 et seq. Frescati and the United States
now seek to recover their cleanup costs from CARCO.

                             6
II.     Background
          a. Facts1

      The M/T Athos I was a single-hulled tanker ship,
measuring approximately 748 feet long and 105 feet
wide.2 As owner of the ship, Frescati chartered it to an
intermediary which assigned it to a tanker pool. CARCO
sub-chartered the Athos I from the tanker pool to deliver a
shipment of crude oil from Puerto Miranda, Venezuela, to
CARCO’s berth in Paulsboro, New Jersey. CARCO was
the shipping customer as well as the wharfinger who
operated the berth.

       The Athos I, carrying CARCO’s shipment, left
Venezuela in mid-November 2004 under the command of
the ship’s master, Captain Iosif Markoutsis. CARCO had




1
    The facts are undisputed unless otherwise noted.
2
   Single-hulled tanker ships drew the attention of
regulators and the public in the wake of the 1989 Exxon
Valdez oil spill off the Alaskan coast; the Exxon Valdez,
like the Athos I, was a single-hulled tanker. Single-hulled
ships were initially subjected to extra regulation, see, e.g.,
33 C.F.R. § 157.455, but have since been phased out of
operation in the United States in favor of double-hulled
ships. See 46 U.S.C. § 3703a.
                              7
instructed the Athos I to load to a draft3 of 37 feet or less
in Venezuela, and provided a warranty that the ship would
be able to safely reach the berth in Paulsboro as long as it
arrived with a draft of 37 feet or less. When the Athos I left
Venezuela, it had a draft of 36′ 6″. Over the course of the
Athos I’s journey, the ship burned fuel and the crew
consumed fresh water. As the ship grew lighter, it rode
higher on the water. By the time it reached the entrance to
the Delaware Bay, the Athos I was drawing 36′ 4″.
Because the fuel and fresh water were consumed from
tanks located in the stern, or rear, of the ship, the Athos I
was no longer sailing at an even keel; it was “trimmed by
the bow,” meaning that the bow, or front of the ship, was
deeper in the water than the ship’s stern. To return the ship
to an even keel, the Athos I took on approximately 510
metric tons of ballast to tanks in the rear of the ship.
Although the parties dispute how much the Athos I was
drawing as it approached CARCO’s berth, the District
Court found that the added ballast brought the ship’s draft
to 36′ 7″.

     The Athos I reached the entrance to the Delaware
Bay without incident on November 26th. All vessels

3
  A ship’s draft is the measurement from the water line to
the bottom of the ship’s hull, known as the keel. As a ship
loads cargo, it becomes heavier and sits lower in the water.
Its draft thereby increases.
                              8
traveling north from the Delaware Bay to the Delaware
River are required to use a Delaware River Pilot to
navigate the waters. At the appropriate time, a local river
pilot, Captain Howard Teal, Jr. boarded the ship and
guided it up the Delaware River until it reached a section
of the river near CARCO’s berth. At that point, a local
docking pilot, Captain Joseph Bethel, replaced Captain
Teal and began to navigate the ship to its berth at
Paulsboro. Captains Teal and Bethel both engaged Captain
Markoutsis in conversations about the Athos I, its passage
from the Delaware Bay to the Paulsboro berth, water
depth, underkeel clearance, and other local conditions.
The substance and sufficiency of those conversations are
disputed by the parties.

      CARCO’s berth is on the New Jersey side of the
Delaware River, directly across from Philadelphia
International Airport. To reach the berth from the main
river channel, ships must pass through an anchorage
immediately adjacent to the berth. The anchorage, known
as Federal Anchorage Number 9 or the Mantua Creek
Anchorage, is a federally-designated section of the river in
which ships may anchor; it is periodically surveyed for
depth and dredged by the Army Corps of Engineers, as
Corps resources allow. No government agency is
responsible for preemptively searching for unknown
obstructions to navigation in the anchorage, although the
Coast Guard, the National Oceanic and Atmospheric

                             9
Administration (NOAA), and the Corps of Engineers work
together to remove or mark obstructions when they are
discovered. Anyone who wishes to search for obstructions
in the anchorage may do so, but anyone wishing to dredge
in the anchorage requires a permit from the Corps of
Engineers.

       It was in this anchorage on November 26, 2004, at
9:02 p.m., that the allision occurred.4 The Athos I was only
900 feet—not much more than the ship’s length—from
CARCO’s berth. The ship was “just about dead in the
water” as Captain Bethel slowly positioned it to dock.
Suddenly, the ship began to list and oil appeared in the
river. At the time of the allision, the ship was in the middle
of a 180° rotation, guided by tugboats, and moving astern
and to port (backwards and to the ship’s left). The path
taken by the Athos I through the anchorage passed, at its
shallowest point, over a 38-foot shoal. Most of the
anchorage was deeper, and the depth of the river at the site
of the allision was at least 41.65 feet at the time.

       Captain Bethel immediately called the Coast Guard
to alert them to the spill, while Captain Markoutsis rushed
to the engine room and transferred oil from the breached


4
 An allision is “[t]he contact of a vessel with a stationary
object such as an anchored vessel or a pier.” Allision,
BLACK’S LAW DICTIONARY (10th ed. 2014).
                             10
cargo tank into another tank. The crew of the Athos I was
eventually able to stop the leak, but not before 264,321
gallons of crude oil had spilled into the Delaware River.

       The cleanup effort began almost immediately.
Although it was ultimately successful, it took months to
complete and the efforts of thousands of workers at a cost
of $143 million. The cause of the allision was not
discovered until more than a month later, when an
abandoned anchor was discovered on the riverbed. The
search for the obstruction that caused the allision proved
difficult. An experienced sonar operator using side-scan
sonar conducted the first search shortly after the allision,
but did not recognize the anchor.5 A second search by the

5
  Side-scan sonar is used to locate objects on the sea floor
and works like a camera, but using sound instead of light
to form an image. Single-beam sonar, by contrast, uses
sound to measure the depth along a single line traced by a
sounding mechanism known as a towpath. If an
obstruction is not located along the towpath, it would not
be detected, and even if the towpath crossed an
obstruction, the data would simply show a depth change
rather than the obstruction itself. Before the allision,
CARCO used single-beam sonar to survey its berthing
area and a small portion of the anchorage. The government
typically used single-beam sonar when it surveyed the
anchorage for depth and dredging purposes.
                            11
same operator, conducted several weeks later, eventually
discovered the anchor with the use of side-scan sonar in
combination with divers and magnetometers. The anchor
weighed approximately nine tons and was 6′ 8″ long, 7′ 3″
wide, and 4′ 6″ high. It has since been removed from the
river.

       The parties dispute the positioning of the anchor at
the time of the allision. An anchor like the one that
punctured the Athos I has two stable positions. It can sit at
rest in the “flukes-up” or “flukes-down” position. A
flukes-up anchor stands almost upright on its crown, with
the flukes pointed upward at a 65° angle, while a flukes-
down anchor has essentially tipped over, with both the
crown and flukes of the anchor lying horizontally on the
riverbed. In the flukes-up position, the anchor sticks up
approximately seven feet above the riverbed, but in the
flukes-down position, it rises only about 3′ 5″ above the
riverbed. The District Court found that the anchor was
flukes-up at the time of the allision, but CARCO asserts
that the anchor was flukes-down, pointing to side-scan
sonar data gathered as part of a geophysical study of the
Delaware River that showed the anchor was flukes-down
in 2001, three years before the allision.6 The anchor was


6
 The anchor was identified in the geophysical study data
only after the allision occurred. The parties agree that in
2001, the anchor was flukes-down, and that no one was
                             12
also flukes-down when it was discovered after the allision.
Between 2001 and the allision in 2004, 241 vessels went
to CARCO’s Paulsboro berth, and many others have
anchored in the anchorage over the years. The District
Court theorized that one of those anchored ships could
have dragged its own anchor chain along the riverbed,
catching on the abandoned anchor and shifting its position.
The court ultimately concluded that although the actual
cause of the anchor’s movement would never be known,
at some point between the geophysical study in 2001 and
the allision in 2004, the anchor shifted from flukes-down
to flukes-up. A flukes-down anchor would not have allided
with the Athos I if the Athos I’s draft was less than 37 feet;
a flukes-up anchor would have.

        Now, more than thirteen years after the allision, the
Athos I has been scrapped, the anchor removed from the
river, and the oil spill cleaned up. What remains is this case
for apportionment of cleanup costs.

          b. Procedural History

       This case, like the Athos I, has been on a long
journey. Over the past thirteen years, the matter has been
to trial before two different judges and heard on appeal

aware of the anchor’s existence before the allision—
except, perhaps, the still-unidentified owner who
abandoned it.
                             13
before two separate panels of this Court. We briefly
summarize that history.

       Litigation began shortly after the allision in January,
2005, when Frescati filed a “Petition for Exoneration from
or Limitation of Liability.” CARCO and others filed
claims for damages associated with the spill. Frescati then
filed a counterclaim against CARCO for its damages. The
United States eventually reimbursed Frescati for some of
its cleanup expenses pursuant to the OPA, and filed suit
against CARCO as a partial subrogee to some of Frescati’s
claims. The claims of Frescati and the United States
against CARCO were consolidated with CARCO’s
counterclaims and defenses, forming the litigation as it
exists today.

       The case was first tried in a forty-one-day bench
trial before the Honorable John P. Fullam. Judge Fullam
found that CARCO was not liable for the casualty in
contract, tort, or otherwise; Frescati and the United States
appealed. On appeal, we affirmed in part, vacated in part,
and remanded the case because the District Court had
failed to make appropriate findings of fact and conclusions
of law as required by Fed. R. Civ. P. 52(a)(1). In re
Frescati, 

, 189, 196–97 (3d Cir. 2013).

      We determined, among other things, that Frescati
was a third-party beneficiary of CARCO’s safe berth
warranty, and that the allision occurred in the approach to
                             14
CARCO’s terminal, meaning that CARCO had an
unspecified duty of care to Frescati in tort. We remanded
for the District Court to determine whether Frescati met
the conditions for the safe berth warranty to apply. We also
asked the District Court, if necessary, to determine the
appropriate duty of care CARCO owed Frescati and
whether CARCO breached that 

–15.

       Judge Fullam retired before the case was remanded.
Upon its return to the District Court, the case was assigned
to the Honorable Joel H. Slomsky as a successor judge
pursuant to Fed. R. Civ. P. 63. Under the terms of that rule,
Judge Slomsky certified his familiarity with the record and
recalled more than twenty witnesses over the course of a
thirty-one-day proceeding.

       The District Court held that CARCO was liable to
Frescati, and the United States as Frescati’s subrogee, for
breach of contract. CARCO’s contract included a
provision known as a safe berth warranty, which, for
purposes of this appeal, warrantied that CARCO’s berth
would be safe for the Athos I as long as the ship had a draft
of 37 feet or less and Frescati did not cause the allision
through bad navigation or negligent seamanship. The
District Court concluded that CARCO breached the
warranty because the Athos I had a draft of 36′ 7″ at the
time of the allision, exercised good navigation and
seamanship, and yet still hit an anchor within the
geographic area covered by the warranty. On appeal,
                             15
CARCO argues that the Athos I had a draft much deeper
than the warrantied depth of 37 feet, and that Frescati
demonstrated negligent seamanship by violating several
federal maritime regulations relating to underkeel
clearance and safe navigation.

      The District Court also found CARCO liable in tort
to Frescati,7 concluding that CARCO had a duty, as
operator of the berth, to search for obstructions in the
approach to its berth. Specifically, the District Court
concluded that CARCO had a duty to use side-scan sonar
to search for unknown obstructions to navigation in the
approach to its berth, and to remove any such obstructions
or warn invited ships—like the Athos I—of their presence.
Because CARCO had not taken any action to search for
obstructions, the District Court held CARCO liable in
tort—for the same amount for which it was liable in
contract. The District Court’s contract and tort holdings
independently support the judgment for Frescati.

      CARCO, in a motion for partial summary judgment
before the District Court, asked that its liability, like
Frescati’s, be limited under the OPA. Because CARCO
did not raise the defense until after the first trial and


7
 The United States is not a party to the tort claim, pursuant
to a partial settlement agreement it reached with CARCO
in 2009.
                             16
appeal, almost a decade into this litigation, the District
Court held that the defense was waived, and in the
alternative, that it failed on the merits.

        The District Court did, however, partially credit
CARCO’s equitable recoupment defense against the
United States. CARCO argued that the conduct of three
federal agencies—the Coast Guard, NOAA, and the Army
Corps of Engineers—misled CARCO into believing that
the United States was maintaining the anchorage free of
obstructions. In addition, CARCO argued that equity
requires the United States to bear the cost of the cleanup
rather than CARCO. The District Court ultimately reduced
the United States’ recovery against CARCO by 50%,
rather than acceding to CARCO’s request to eliminate its
liability entirely.

       Finally, the District Court held that Frescati was
entitled to prejudgment interest at the federal
postjudgment rate rather than the higher U.S. prime rate
requested by Frescati.

      The District Court ultimately awarded Frescati
$55,497,375.958 on the claims of breach of contract and

8
 Frescati’s liability under the OPA for the cost of cleaning
up the spill was limited to approximately $45 million. The
United States reimbursed it for the remaining $88 million
of its qualifying cleanup expenses. In addition to the $45
                            17
negligence, plus prejudgment interest of $16,010,773.75,
for a total judgment of $71,508,149.70. The United States,
after the court’s 50% reduction, was awarded
$43,994,578.66 on its subrogated breach of contract claim,
with prejudgment interest of $4,620,159.98, for a total
judgment of $48,614,738.64.

        All three parties now appeal. We will affirm the
District Court’s judgment in favor of Frescati on the
breach of contract claim and the prejudgment interest
award, as well as the District Court’s denial of CARCO’s
motion for partial summary judgment on its limitation of
liability defense. We will vacate the District Court’s
judgment in favor of Frescati on the negligence claim. We
will affirm in part the District Court’s judgment in favor
of the United States with respect to CARCO’s liability on
the subrogated breach of contract claim, but because


million in OPA damages, Frescati also incurred roughly
$10 million in damages that fell outside the scope of the
OPA’s liability cap—third-party claims; cleanup expenses
for recreational boats; the cost of removing the anchor and
the pump casing from the riverbed; a settlement with a
nearby nuclear power plant that had to shut down;
unrepaired hull damage to the Athos I, and other
miscellaneous expenses. Frescati’s contract recovery of
$55 million was based on both its OPA and non-OPA
damages.
                            18
CARCO’s equitable recoupment defense fails, we will
reverse and remand for further proceedings to recalculate
damages and prejudgment interest.

III.   Jurisdiction and Standard of Review

       The District Court had admiralty jurisdiction
pursuant to 28 U.S.C. § 1333(1). We have jurisdiction
over this appeal under 28 U.S.C. § 1291.

       “On appeal from a bench trial, we review a district
court’s findings of facts for clear error and exercise
plenary review over conclusions of law.” Norfolk S. Ry.
Co. v. Pittsburgh & W. Va. R.R., 

, 253 (3d
Cir. 2017). “A finding of fact is clearly erroneous when it
is completely devoid of minimum evidentiary support
displaying some hue of credibility or bears no rational
relationship to the supportive evidentiary data.” VICI
Racing, LLC v. T-Mobile USA, Inc., 

, 283 (3d
Cir. 2014); In re 

.

IV.    The Safe Berth Warranty

       CARCO promised that the Athos I would be
directed to a location “she may safely get (always afloat),”
a promise known as a safe port or safe berth warranty. JA
at 1211. Such a promise provides, among other things,
“protection against damages to a ship incurred in an unsafe
port to which the warranty applies.” In re 

.
                            19
      A port is deemed safe where the particular
      chartered vessel can proceed to it, use it, and
      depart from it without, in the absence of
      abnormal weather or other occurrences,
      being exposed to dangers which cannot be
      avoided by good navigation and seamanship.
      Whether a port is safe refers to the particular
      ship at issue, and goes beyond the immediate
      area of the port itself to the adjacent areas the
      vessel must traverse to either enter or leave.
      In other words, a port is unsafe—and in
      violation of the safe berth warranty—where
      the named ship cannot reach it without harm
      (absent abnormal conditions or those not
      avoidable by adequate navigation and
      seamanship).

(quotations and citations omitted). “[T]he safe
berth warranty is an express assurance made without
regard to the amount of diligence taken by the charterer.”

For our purposes, a safe berth warranty
promises that a ship with a draft less than the warrantied
depth is covered by the warranty in the absence of bad
navigation or negligent seamanship.

       Our prior opinion called for the District Court to
resolve three issues on remand: the draft limit beyond
which the safe berth warranty would not apply; the actual
draft of the Athos I at the time of the allision; and whether
                             20
the warranty was negated by bad navigation or negligent
seamanship. 

204 n.20.

       As an initial matter, the District Court found that the
safe berth warranty applied to ships drawing less than 37
feet, a finding neither party challenges on appeal. The
remaining issues, then, are whether the Athos I had a draft
of less than 37 feet, and if it did, whether bad navigation
or negligent seamanship by Frescati negated the warranty.

          a. The Draft of the Athos I

        The District Court found that the Athos I had a draft
of 36′ 7″ at the time of the allision. The court based this
finding on the undisputed draft of the Athos I at the time
of its departure from Puerto Miranda—36′ 6″—as well as
expert testimony regarding the condition of the ship and
its estimated draft at Paulsboro.9




9
  Frescati’s expert, Anthony Bowman, developed the
Seamaster software program, which allows him to enter
the measurements of a ship—including the weight,
dimensions, and strength of all its constituent parts, such
as the hull, cargo, and supplies—and calculate, among
other things, a ship’s draft. Having considered the ship’s
records, information about the ballast tanks, and his own
software, Bowman testified that at the time of the allision,
                             21
       CARCO challenges the District Court’s
determination of the Athos I’s draft, arguing that the
District Court improperly based its finding on a
speculative assumption about the orientation of the
abandoned anchor. Specifically, CARCO disputes the
District Court’s finding that the anchor shifted from a
flukes-down position to a flukes-up position sometime
between 2001 and the allision in 2004, a shift that caused
the anchor to intrude within the 37-foot safe depth
promised by CARCO. CARCO argues that the District
Court failed to make a finding as to the precise mechanism
by which the anchor shifted from flukes-down to flukes-
up. The anchor’s orientation matters; if the accident
occurred while the anchor was flukes-down, the Athos I
necessarily would have had a draft that exceeded the scope
of CARCO’s warranty.10




the Athos I had a draft of 36′ 7″. The District Court
credited his testimony.
10
   The District Court made undisputed findings of fact as
to the height of the anchor in a flukes-down position (41
inches or 3.42 feet) and the depth of the river at the time
and location of the allision (41.65 feet). Assuming for the
moment that the anchor was flukes-down, as CARCO
argues, the allision would not have occurred unless the
Athos I had a draft of at least 38.23 feet, or just under
                            22
       Broadly speaking, the District Court made three
findings of fact related to the anchor’s orientation. First,
the court and parties agree that, three years before the
allision, the anchor was in the flukes-down position.11
Second, the District Court found that at some point before
the allision, the anchor shifted into the flukes-up position.
Finally, after the allision, the anchor was eventually
discovered back in the flukes-down position—perhaps
unsurprising, given the force of its encounter with the
Athos I.

       CARCO attacks the second finding, arguing that
there was insufficient evidence in the record to support the
District Court’s suggestion that a “sweeping anchor chain”
could have caught the anchor and shifted it into the flukes-
up position.12


38′ 3″, significantly in excess of the warrantied draft of 37
feet.
11
  Experts for both sides were able to identify the flukes-
down anchor in a sonar scan performed in 2001 as part of
an independent geophysical study.
12
   Ships at anchor move with the tide, back and forth as the
tide comes in and goes out. The anchor chain drags or
“sweeps” across the riverbed as the ship floats, potentially
shifting the position of objects on the riverbed, and leaving
scour marks on the riverbed. Anchor chains also move
                             23
       We find CARCO’s arguments unconvincing,
primarily because the “sweeping anchor chain” theory,
however plausible or implausible, is not necessary to
sustain the District Court’s finding. Let us imagine a piece
of furniture (a sofa, perhaps, or an armchair) that has fallen
off the back of a pickup truck onto a roadway. One driver
reports seeing the furniture in the right lane. A while later,
a second driver hits the furniture. The second driver asserts
that the furniture was in the left lane when he struck it, and
provides evidence to that effect. A highway patrolman

along the river bottom when the anchor is pulled back onto
the ship. CARCO, for its part, characterizes the idea that
an anchor chain might have moved the abandoned anchor
as “fantastical,” “inexplicabl[e],” an “astonishing
assertion,” “facially implausible,” “pure and wild
speculation,”      “pure     speculation,”     “conjecture,”
“speculative and unsupported,” and, once again,
“implausible.” CARCO Opening Br. 4, 53–55; CARCO
Reply Br. 32. The District Court pointed out that scour
marks were found on the river bottom near the site of the
allision, but ultimately decided only that the anchor was in
the flukes-up position at the time of the allision. JA at 78
(“Although the actual cause of the anchor’s movement to
a ‘flukes-up’ position will never be known, the Court finds
that at some point after December 2001, this movement
occurred and the anchor was positioned in a ‘flukes-up’
orientation when it allided with the Athos I.”).
                             24
shows up later and finds the furniture once again in the
right lane. A court may find, without committing error,
that the furniture was in the right lane and moved to the
left without making a specific finding as to the precise
method by which the furniture moved from one lane to the
other. Perhaps another driver hit it; perhaps a pedestrian
tried to move it out of the road but did not finish the job.
When credible evidence shows that the second driver was
driving in the left lane, a finding to that effect does not
become error because the furniture was in the right lane
when the first driver passed, or changed position after—or
because of—the encounter with the second driver.

        Here, the record contains sufficient evidence to
support the finding that the anchor was, in fact, flukes-up
at the time of the allision. How exactly the anchor changed
position does not impact our sufficiency determination. As
an initial matter, the movement of the Athos I at the time
of the allision and the damage to its hull are sufficient to
show that the anchor was flukes-up. And substantial
evidence unrelated to the anchor showed that the Athos I
was drawing 36′ 7″ at the time of the allision—a draft at
which the allision would not have occurred had the anchor
been flukes-down. That is enough to support the District
Court’s finding that the anchor moved from flukes-down
to flukes-up.

     The movement of the ship and damage to its hull
shows that the anchor must have been flukes-up. The
                            25
District Court found that the Athos I was moving astern
and to port at the time of the allision, a finding CARCO
does not challenge. Based on that movement, the scoring
left by the anchor on the hull, the size and shape of the two
holes the anchor created, and the damage to the anchor
itself also supported the District Court’s finding that the
anchor must have been flukes-up at the time of the allision.
CARCO’s own expert witness, on cross-examination,
testified that if the Athos I were moving astern and to port,
the damage to the Athos I’s hull would necessarily require
a flukes-up anchor.13 JA at 1021–22.

       Nor did the District Court base its finding of a 36′ 7″
draft on the flukes-up anchor alone. While CARCO argues
that the anchor was flukes-down, and that therefore the
Athos I must have had a deep draft, the reverse is also true.
If the Athos I had a draft of 36′ 7″, then the anchor must
have been flukes-up. The District Court credited expert
testimony that the ship had a 36′ 7″ draft. The ballast tanks
contained no extra liquid that would have affected the
ship’s draft, a finding that CARCO does not challenge on
appeal. The ship left Puerto Miranda with a draft of 36′ 6″.
Visual observation of the ship by experts and
crewmembers immediately after the allision suggested the
Athos I had a 36′ 7″ draft before the allision. And, on

13
  CARCO’s theory at trial, abandoned on appeal, was that
the ship was not moving astern and to port.
                             26
appeal, CARCO fails to offer any suggestion as to how the
draft might have increased by more than a foot without the
crew’s knowledge or any evidence that the ballast tanks
were faulty.14

       We conclude there was no clear error in the District
Court’s determination that the Athos I had a draft of 36′ 7″
at the time of the allision. The ship was, therefore, within
the scope of CARCO’s safe berth warranty.

          b. Frescati’s Seamanship

       A safe berth warranty applies only in the absence of
bad navigation or negligent seamanship. CARCO argues
on appeal that Frescati violated several maritime
regulations related to the operation of single-hulled
tankers, and that those regulatory violations serve as
sufficient proof of negligent seamanship. The District
Court concluded that Frescati did not violate any relevant
regulations, and enforced the safe berth warranty. We

14
   The Athos I passed safely over a 38-foot shoal less than
fifteen minutes before the allision. JA at 203. It seems that
if the Athos I had a draft deep enough to hit the flukes-
down anchor (a minimum of 38.23 feet, see supra note 10),
it would have encountered the 38-foot shoal before it ever
encountered the anchor. A flukes-down anchor would
have been deeper than the 38-foot shoal even at the
anchor’s shallowest point. JA at 77, 78, 85.
                             27
agree with the District Court that Frescati did not violate
any relevant regulations.

       On appeal, CARCO argues specifically that Frescati
violated two federal regulations: 33 C.F.R. § 157.455 and
33 C.F.R. § 164.11. Section 157.455 applied to certain
single-hulled tankers during the period they were being
phased out of operation, while § 164.11 applies to certain
ships above 1,600 gross tons. 33 C.F.R. §§ 157.400,
164.01. Both sections applied to the Athos I at the time of
the allision.

       Section 157.455 requires the owner or operator of a
single-hulled tanker to provide certain written guidance to
the ship’s master for purposes of estimating the tanker’s
underkeel clearance.15 33 C.F.R. § 157.455(a). It also


15
     33 C.F.R. § 157.455(a)–(b) reads:
         (a) The owner or operator of a tankship, that
             is not fitted with a double bottom that
             covers the entire cargo tank length, shall
             provide the tankship master with written
             under-keel clearance guidance that
             includes—
               (1) Factors to consider when
                   calculating the ship’s deepest
                   navigational draft;
                              28
      (2) Factors to consider when
          calculating the anticipated
          controlling depth;
      (3) Consideration of weather or
          environmental conditions; and
      (4) Conditions which mandate
          when the tankship owner or
          operator shall be contacted
          prior to port entry or getting
          underway; if no such
          conditions exist, the guidance
          must contain a statement to
          that effect.
(b) Prior to entering the port or place of
    destination and prior to getting underway,
    the master of a tankship that is not fitted
    with the double bottom that covers the
    entire cargo tank length shall plan the
    ship’s passage using guidance issued
    under paragraph (a) of this section and
    estimate the anticipated under-keel
    clearance. The tankship master and the
    pilot shall discuss the ship’s planned
    transit including the anticipated under-
    keel clearance. An entry must be made in
    the tankship’s official log or in other
                      29
requires the master to use that guidance to plan the ship’s
passage, estimate the underkeel clearance, consult with the
relevant pilots who will guide the ship to its berth, and
make a log entry reflecting discussion of the ship’s
underkeel clearance with the pilot. 33 C.F.R.
§ 157.455(b). Section 164.11 mandates that the master
ensure the pilot is informed of certain information,
including the ship’s draft and tidal conditions.16 33 C.F.R.
§ 164.11.



            onboard    documentation  reflecting
            discussion of the ship’s anticipated
            passage.
33 C.F.R. § 157.455(a)–(b).
16
     33 C.F.R. § 164.11 reads:
         The owner, master, or person in charge of each
         vessel underway shall ensure that:
            ....
         (k) If a pilot other than a member of the
             vessel’s crew is employed, the pilot is
             informed of the draft, maneuvering
             characteristics, and peculiarities of the
             vessel and of any abnormal circumstances
             on the vessel that may affect its safe
             navigation.
                              30
       CARCO argues that Frescati was responsible for
three specific violations, each of which allegedly caused
the allision. First, CARCO claims that Frescati failed to
adequately plan the ship’s passage. Second, CARCO
claims that Frescati failed to estimate the Athos I’s
underkeel clearance. Finally, CARCO claims that Frescati
failed to ensure that an adequate master-pilot exchange
occurred, and made no log entry that would reflect such an
exchange.

       With respect to planning the passage, CARCO
argues that 33 C.F.R. § 157.455 requires a written voyage
plan. Frescati allegedly violated that requirement by
failing to finalize an official voyage plan document using
the Tsakos Voyage Plan form contained in the Tsakos
Vessel Operation Procedures Manual. See JA at 1178–85.

      The text of § 157.455 undermines CARCO’s
argument. The regulation does not itself require a written
voyage plan. Paragraph (a) of the regulation requires that
Frescati create “written under-keel clearance guidance,”
which must contain “factors to consider” when evaluating

         ...
      (n) Tidal state for the area to be transited is
          known by the person directing movement
          of the vessel . . . .
33 C.F.R. § 164.11.
                            31
draft, water depth, and weather conditions. Paragraph (b)
requires that the master plan the ship’s passage using those
“factors to consider” in the guidance required by
paragraph (a). Nowhere does this regulation require that
the master’s passage plan be in writing; the only reference
to a writing in paragraph (b) comes in the requirement that
some official log of the master-pilot conference be
recorded. CARCO conflates the passage plan requirement
of paragraph (b)—to consider certain relevant factors
when planning—with the “Voyage Plan” form contained
in Frescati’s Vessel Operation Procedures Manual. See JA
at 1180. The Voyage Plan form focuses on plotting the
course of the vessel from berth to berth; paragraphs (a) and
(b) of the regulation, on the other hand, serve to create a
reference list for the ship’s master of relevant factors to
consider when estimating underkeel clearance.

      Frescati satisfied the requirements of paragraph (a)
by providing written underkeel clearance guidance in
Section 3.417 of its Vessel Operation Procedures Manual.
JA at 1191. The Manual appropriately lists factors to
consider, including “sea state and swell,” “tidal


17
   The Vessel Operation Procedures Manual appears to
contain a typographical error listing the appropriate
section as 2.4 rather than 3.4, as it appears in the Table of
Contents. See JA at 1189, 1191.
                             32
conditions,” and “the effect of squat,”18 and suggests to the
master that 10% or 5% underkeel clearance margins would
typically be appropriate. 

satisfied the planning
requirement of paragraph (b) because the Athos I’s master,
Captain Markoutsis, considered factors like the sea state,
tidal condition, and the effect of squat. Even though
CARCO provided a safe berth warranty for a draft up to
37 feet, Captain Markoutsis loaded the ship to only 36′ 6″
because he was “afraid” of a 37-foot draft, and eventually
entered the Delaware River with a draft of 36′ 7″. In re

. The charts in the Athos I were
marked with the 38-foot controlling draft in the anchorage.
JA at 992. Captain Teal, the river pilot, testified that he
and Captain Markoutsis discussed the draft, wind,
visibility, and tides. We agree with the District Court that
Frescati fully complied with the planning requirement of

18
   “Squat is a hydrodynamic phenomenon, which occurs
when a ship is moving through the waters. As a ship moves
forward, it displaces a volume of water. The displaced
water rushes under the keel of the ship and creates a low
pressure area causing the ship to sink down toward the
riverbed. The faster a ship is moving, the more the ship
will sink down towards the riverbed. This process causes
a ship to be closer to the riverbed by increasing a vessel’s
draft.” JA at 70 (citations omitted).
                             33
§ 157.455(b)—that is, to use the factors listed in the
Vessel Operating Procedures Manual when planning the
passage.

      CARCO’s second argument is that Frescati violated
§ 157.455(b) because Captain Markoutsis failed to
estimate the Athos I’s underkeel clearance. The District
Court did not err in finding that Captain Markoutsis had
estimated underkeel clearance. Captain Markoutsis
discussed the draft, tidal conditions, and anticipated
underkeel clearance with Captain Teal. JA at 801–802.
They estimated that the ship would have at least 1.5
meters’ clearance—nearly five feet. 

and Markoutsis also discussed the draft and believed they
would have sufficient clearance. JA at 833, 837. CARCO
highlights that there is no evidence of written underkeel
clearance estimates, but § 157.455 does not require written
estimates.

       Finally, CARCO argues that the master-pilot
exchange required by § 157.455 and § 164.11 was
inadequate. In general, master-pilot exchanges are
intended to allow the master to share the navigational
characteristics of his ship with the pilot who will be
guiding it, and for the pilot to share local conditions such
as weather, depth, and the tide with the master. Section
157.455(b) requires that “[t]he tankship master and the
pilot shall discuss the ship’s planned transit including the
anticipated under-keel clearance. An entry must be made
                            34
in the tankship’s official log or in other onboard
documentation reflecting discussion of the ship’s
anticipated passage.” 33 C.F.R. § 157.455(b). Section
164.11 requires that the master ensure that

      [i]f a pilot other than a member of the vessel’s
      crew is employed, the pilot is informed of the
      draft, maneuvering characteristics, and
      peculiarities of the vessel and of any
      abnormal circumstances on the vessel that
      may affect its safe navigation. . . . [and that
      the] [t]idal state for the area to be transited is
      known by the person directing movement of
      the vessel.

33 C.F.R. § 164.11(k), (n).

       Captain Markoutsis was responsible for discussing
the     draft,    underkeel     clearance,   maneuvering
characteristics, and tidal state with the two pilots who
guided the Athos I. The testimony shows that Captain
Markoutsis did so, discussing all the relevant information
with both pilots, and that he recorded the conversation on
the signed Pilot Card, which served as sufficient
documentation of the master-pilot conference. The
District Court additionally credited Frescati’s expert
witness, Captain Betz, who observed both Captain Teal
and Captain Bethel testify. Captain Betz opined that the

                              35
master-pilot exchanges were adequate and customary in
all respects.

       Frescati operated the Athos I with neither bad
navigation nor negligent seamanship. Nevertheless, the
allision occurred. The District Court did not err in
concluding that the allision resulted from a breach of
CARCO’s safe berth warranty.

V.    Wharfinger Negligence

       CARCO wore two hats in its dealings with Frescati,
as a shipping customer and as a wharfinger. These dual
roles exposed CARCO to liability under two independent
legal theories. CARCO’s first role, as a shipping customer
that contracted with Frescati for delivery of a shipment of
crude oil, resulted in CARCO’s liability under the
contractual safe berth warranty, discussed above. The
second, as the wharfinger for the Paulsboro berth that was
the Athos I’s intended destination, resulted in the District
Court’s finding of negligence and CARCO’s
corresponding liability in tort.

       Both theories of liability independently support the
District Court’s judgment against CARCO. As a result,
our decision to affirm the judgment based on CARCO’s
contractual liability means that we are not required to
delve into the District Court’s tort analysis. However,
having reviewed that analysis, we harbor serious doubts

                            36
about the appropriateness of the court’s proposed duty of
care. For that reason, we are compelled to make clear that
we will affirm the District Court’s judgment based solely
on CARCO’s breach of contract.

      A wharfinger’s duty is more limited than that of a
shipping customer who has provided a safe berth warranty.
As we previously wrote:

      In the tort context, . . . a wharfinger is not a
      guarantor of a visiting ship’s safety, but is
      bound to use reasonable diligence in
      ascertaining whether the berths themselves
      and the approaches to them are in an ordinary
      condition of safety for vessels coming to and
      lying at the wharf. This is not an
      unconstrained mandate to ensure safe
      surroundings or warn of hazards merely in
      the vicinity. Instead, a visiting ship may only
      expect that the owner of a wharf has afforded
      it a safe approach. In being invited to dock at
      a particular port, a vessel should be able to
      enter, use and exit a wharfinger’s dock
      facilities without being exposed to dangers
      that cannot be avoided by reasonably prudent
      navigation and seamanship.

In re 

(quotations and citations
omitted). In short, and as a general matter, a wharfinger’s
                            37
duty is to use reasonable diligence to ascertain whether the
approach to its berth is safe for an invited vessel.19

       We remanded for the District Court to determine in
the first instance what reasonable diligence required of
CARCO under the circumstances of this case, and whether
CARCO breached that standard. 

the
District Court concluded that

      a reasonably prudent terminal operator
      should periodically scan the approach to its
      dock for hazards to navigation as long as
      ships are being invited there. In this case, the
      standard would require that side-scan sonar
      be used to search the approach for
      obstructions that are potential hazards to
      navigation. If an obstruction is located, a
      terminal operator is then required to remove
      it, and if the terminal operator cannot remove
      it, notice of the hazard must be given to



19
   We previously determined that the allision occurred in
the approach to CARCO’s berth—the geographic area
within which a wharfinger’s duty exists—and as a result,
CARCO had a duty to use reasonable diligence to provide
the Athos I with a safe approach. In re 

.
                            38
      incoming ships by marking it as a hazard
      and/or warning ships of its presence.

JA at 132. Because CARCO did nothing to look for
obstructions, the District Court held that it had breached
its duty.

       The District Court chose its standard by
determining what the “demands of reasonableness and
prudence” required. JA at 129. Citing Judge Learned
Hand’s famous formula from United States v. Carroll
Towing, 

(2d Cir. 1947), the court concluded
that the precaution of a preemptive side-scan sonar search
would be less burdensome than the probability of an
allision multiplied by the serious harm caused by a spill of
toxic substances like crude oil.

       We have doubts about the District Court’s
balancing of the cost of preventative measures on one hand
and the cost of potential accidents on the other. The court
found that a general scan of the approach to CARCO’s
berth and the berth itself would have cost between $7,500
and $11,000, and would have prevented the allision. Yet
in this very case, the targeted scan of the area where the
allision occurred, conducted only eight days after the
allision, did not identify the anchor. The first set of 93
side-scan sonar passes conducted by Frescati’s expert,
John Fish—at a cost of $38,577—identified a pump casing
on the river bottom. The anchor, however, went
                            39
unrecognized.20 We do not share the District Court’s
confidence that a general $11,000 scan of the approach
and berth would have “recognized” the anchor with
sufficient clarity to prevent the allision, given that a
targeted $38,000 scan for obstructions failed to do so.

       Beyond the questionable utility of side-scan sonar
as applied to this case, we doubt whether imposing a
specific duty to require side-scan sonar would be useful
for wharfingers in the ordinary course of their business.
Single-hulled vessels like the Athos I present unique risks,
and have been treated with special care by regulators. See,
e.g., 33 C.F.R. § 157.455. Today, as a result of those
unique risks, such vessels are no longer permitted to
operate in the waters of the United States. See 46 U.S.C.
§ 3703a (banning single-hulled oil tankers in the waters of
the United States after January 1, 2015). Furthermore,
side-scan sonar is not the only method available to detect
and recognize obstructions, as the District Court pointed
out.21 Even if we were to accept the court’s balancing of

20
   Fish testified that the side-scan sonar equipment
“detected” the anchor, but neither he nor anyone else
“recognized” it until after the second set of scans were
taken. JA at 927.
21
   The court determined that CARCO should have used
side-scan sonar to search for obstructions, but seemed
willing to accept that other methods of searching for
                            40
cost, risk, and the magnitude of the potential harm, the
high standard set forth in this case—involving a risky
single-hulled vessel—would not necessarily apply to
future cases, which will necessarily involve only double-
hulled vessels.22

       We are not unsympathetic to the position in which
we placed the District Court by asking it to specify the
duty of care at play in this case. The District Court has
conscientiously complied. And we stand by our previous
holding that CARCO had some duty to use reasonable
diligence to provide the Athos I with a safe approach to its

obstructions might accomplish the same purpose. It noted
that “side-scan sonar . . . is not the only method available
in the industry to search for hazardous debris. . . . Since
the standard of care involves factual issues, the methods
may vary when the conditions in the approach to each
terminal are examined.” JA at 132 n.109.
22
     Indeed, five years after the Athos I allision, the
Norwegian tanker SKS Satilla, carrying nearly 42 million
gallons of crude oil, allided with a sunken oil rig in the
Gulf of Mexico, sustaining “substantial damage to the port
side of her hull.” Findings of Fact and Conclusions of
Law, In re Ensco Offshore Co., No. 4:09-CV-2838, ECF
No. 185 at 3, ¶¶ 6–7 (S.D. Tex. Sept. 30, 2014). But
“[b]ecause the SATILLA [was] a double hulled vessel[,]
. . . there was no discharge of crude oil.” 

¶ 9.
                            41
berth—a duty it may or may not have breached. In re

. Nevertheless, given CARCO’s
independent liability in contract and our decision to affirm
on that basis, we will once again decline to outline
precisely what CARCO’s duty of reasonable diligence
entailed.

VI.   Subrogation and Equitable Recoupment

        This litigation does not implicate the interests of
only Frescati and CARCO. The United States reimbursed
Frescati for $88 million in cleanup expenses above the
liability limit established by the OPA. Consequently, the
United States became subrogated to Frescati’s claims, and
joined the fray by filing suit against CARCO in 2008.23


23
    The United States and CARCO reached a partial
settlement agreement before the first trial. Both the United
States and CARCO agreed to forgo any negligence claims
they might have had against one another. The parties
agreed that the United States would pursue only its
contract claim against CARCO. As a result, the United
States’ judgment against CARCO was based solely on
CARCO’s contractual liability under the safe berth
warranty. CARCO, for its part, reserved in the settlement
agreement
      each and every substantive and procedural
      right available to a defendant . . . including
                            42
       Frescati initially paid for the oil spill cleanup costs
as a “responsible party” under the OPA. See 33 U.S.C.
§ 2702(a). The OPA allows a responsible party like
Frescati to limit its liability to a specified sum; any cleanup
costs above that amount are reimbursed out of the Oil Spill
Liability Trust Fund.24 See 33 U.S.C. § 2704. Under this


      but not limited to the right to raise affirmative
      defenses under any theory or doctrine of law
      or equity, the right to assert setoff or
      recoupment and the right to assert
      compulsory           or         non-compulsory
      counterclaims other than a Claim for
      Contribution or Indemnity . . . .
JA at 391.
24
   The Oil Spill Liability Trust Fund, administered by the
Coast Guard, serves much like insurance for the oil
transportation industry. Companies that import oil into the
United States pay a per-barrel fee into the Trust Fund.
When a tanker vessel spills oil, the OPA assigns liability
for the cleanup to a “responsible party”—typically the
owner of the vessel from which the oil spilled. The
responsible party is liable for all cleanup costs associated
with the spill. If the costs exceed a liability cap established
by the OPA, the Trust Fund reimburses the responsible
party for all expenses above the statutory cap. Liability
under the OPA does not preclude a responsible party from
                              43
scheme, Frescati’s liability for the cost of the oil spill
cleanup was limited to approximately $45 million. The
Trust Fund reimbursed Frescati for its remaining cleanup
costs, which totaled approximately $88 million. The
United States then became statutorily “subrogated to all
rights, claims, and causes of action that the claimant
[Frescati] has under any other law.” 33 U.S.C. § 2715(a).
The United States pursued these claims against CARCO
as a “person who is liable, pursuant to any law, to the
compensated claimant [Frescati] or to the Fund, for the
cost or damages for which the compensation was paid.” 33
U.S.C. § 2715(c).

      Pursuant to the partial settlement agreement, the
United States limited itself to the same contractual claims
Frescati asserted. Because CARCO was liable to Frescati
in contract, it was also liable to the United States for the
amount the Trust Fund had reimbursed Frescati: nearly
$88 million. But CARCO asserted a defense against the
United States it did not assert against Frescati—equitable


bringing any claims it has against a third party under any
other law. The United States, to the extent the Trust Fund
has reimbursed the responsible party’s costs, steps into the
shoes of the responsible party as subrogee and may pursue
claims against a third party as if it were the responsible
party. Any recovery won by the United States is returned
to the Trust Fund to cover future oil spill reimbursements.
                            44
recoupment—and in response, the District Court reduced
the United States’ judgment by 50%. Both CARCO and
the United States appealed. CARCO argues that the
District Court erred by not eliminating the United States’
recovery, while the United States argues that the District
Court should have left the contract judgment untouched
and denied CARCO any equitable remedy. We conclude
that the District Court erred by reducing the United States’
judgment by 50%. The United States is entitled to a full
recovery.

          a. Subrogation and Subrogee-Specific
             Defenses

       As an initial matter, we note that the dispute
between CARCO and the United States presents an
unusual question about the nature of subrogation.
Subrogation itself is not unusual; in general terms, it
“simply means substitution of one person for another; that
is, one person is allowed to stand in the shoes of another
and assert that person’s rights against a third party.” US
Airways v. McCutchen, 

, 97 n.5 (2013). Most
often, it arises in the insurance context as a procedural
mechanism to allow an insurer (the subrogee) to step into
the shoes of its insured (the subrogor) after it has
compensated the insured for harm caused by a third party.
The subrogee, having stepped into the shoes of the
subrogor, is entitled to assert all of the subrogor’s rights
and claims against the responsible third party. Likewise,
                            45
the third party—now defending an action brought by the
subrogee—is entitled to assert every defense it otherwise
could have raised against the subrogor. In that vein, the
third party’s liability to a subrogee cannot be greater than
it would have been to the subrogor. Restatement (Third)
of Restitution & Unjust Enrichment § 24.

        All that is unexceptional. The unusual question
presented here is whether a third party may assert a
defense against a subrogee that it could not assert against
the subrogor. As we discussed above, CARCO is liable to
Frescati, the subrogor, in contract. Consequently, CARCO
is liable to the United States, the subrogee, under that very
same contract. But CARCO wishes to assert a defense
against the United States—namely, that equitable
recoupment requires the United States to bear the loss
rather than CARCO because of the allegedly misleading
conduct of three federal agencies—that it could not assert
against Frescati.

       The United States makes a related argument. Its
position is that the equitable recoupment defense,
predicated as it is on the conduct of federal agencies rather
than the contractual relationship between Frescati and the
United States, violates the statutory subrogation provision
of the OPA. Specifically, the United States argues that it
is entitled to “all [of Frescati’s] rights, claims, and causes
of action” under the OPA. 33 U.S.C. § 2715(a). Frescati’s
contractual right is not limited by CARCO’s claims
                             46
against the Coast Guard, NOAA, and the Army Corps of
Engineers; the United States asserting Frescati’s
contractual right should also not be so limited, and to do
otherwise would infringe on the United States’ statutory
entitlement. When Frescati has the right to a full recovery
under its contract, the argument goes, so does the United
States.

       We agree. CARCO may only assert defenses
against the United States’ subrogated claims which it
could have asserted against Frescati—including any
equitable recoupment defense it could have asserted
against Frescati. In its capacity as a subrogee, the United
States should be subject to the same treatment as Frescati.
Just as the United States, as subrogee, may only assert
Frescati’s claims, CARCO, as defendant, is not entitled to
extra defenses because the United States asserts Frescati’s
claims rather than Frescati itself. Of course, no party is
exempt from the Federal Rules of Civil Procedure. The
United States is subject to the ordinary procedural rules
governing counterclaims and third-party complaints, and
the OPA does not bar CARCO from asserting whatever
claims it has against the United States using those
recognized procedural mechanisms where appropriate.25


25
  This issue is complicated by the fact that the specific
defense asserted by CARCO, equitable recoupment, is
sometimes pleaded as a defense, and sometimes as a
                            47
       In this case, the only claim asserted by the United
States is Frescati’s contract claim. In re 

; JA at 390. It follows that CARCO’s equitable
recoupment defense must be directed toward the United
States’ contract claim. 

(declining to
preclude CARCO from raising “equitable defense[s] to the
Government’s subrogation claims”). If CARCO had other
cognizable claims against the three federal agencies
involved in regulating the Delaware River and the
anchorage, sounding in tort or otherwise, it was free to
assert them in a third-party complaint or counterclaim, just
as the United States was free to pursue other claims against
CARCO.26 In that light, we proceed to analyze CARCO’s

counterclaim. We do not mean to imply that CARCO
should have pleaded equitable recoupment as a
counterclaim rather than a defense. However it is pleaded,
“recoupment is in the nature of a defense arising out of
some feature of the transaction upon which the plaintiff’s
action is grounded,” and here, the plaintiff’s action is
grounded in Frescati’s contractual right. Bull v. United
States, 

, 262 (1935). To the extent CARCO
had cognizable claims against the Coast Guard, NOAA,
and the Army Corps of Engineers, it should have asserted
those claims directly, rather than as a defense to Frescati’s
now-subrogated contract claim.
26
  CARCO was also free to waive its claims against the
United States, and vice versa. Indeed, both CARCO and
                             48
equitable recoupment defense as it applies to the United
States’ contractual rights.

          b. Equitable Recoupment

       Equitable recoupment is a “principle that
diminishes a party’s right to recover a debt to the extent
that the party holds money or property of the debtor to
which the party has no right.”27 In re Frescati, 718 F.3d at


the United States waived certain rights in the 2009 partial
settlement agreement, including CARCO’s waivers of the
rights to bring a “Claim for Contribution or Indemnity . . .
whether based on principles of common law, contract,
quasi-contract or tort,” and “demand that the court reduce
or offset the damages awarded to the United States . . .
based on evidence that the negligence or fault of the
United States in failing to detect, mark and/or remove
underwater obstructions to navigation . . . caused or
contributed to the ATHOS I Incident.” JA at 389. At an
earlier stage in the litigation, the United States argued that
CARCO’s equitable recoupment defense amounted to a
violation of the settlement agreement. The United States
eventually waived that argument by failing to raise it at the
first trial, and so we need not consider it today. In re

.
27
  A classic example of recoupment is a situation in which
the statute of limitations is different for two related claims
                             49
214 n.35. For an equitable recoupment defense to succeed,
the defendant must possess a claim against the plaintiff
arising from the same transaction or occurrence as the
plaintiff’s suit, seeking relief of the same kind as that
sought by the plaintiff, in an amount no greater than that

arising out of the same transaction—when, for example,
the statute of limitations period during which the United
States may file a claim against a taxpayer for
underpayment of the income tax is longer than the period
during which a taxpayer may file a claim for a refund of
overpayment of the estate tax. The taxpayer (in this case,
the estate of a decedent) pays the estate tax and final year’s
income tax. Sometime later, after the statute of limitations
has run on the estate tax overpayment but not the income
tax underpayment, the government claims the taxpayer
owes additional income tax for the taxpayer’s final year.
Due to the increased income tax liability for the year, the
taxpayer now owes less in estate tax—but the statute of
limitations has already run, and the taxpayer cannot amend
the estate tax return. In an action brought by the
government to recover the extra income tax owed, the
taxpayer may assert an equitable recoupment defense for
the amount of the overpayment of the estate tax, even
though the statute of limitations has run and the taxpayer
would not otherwise have been able to recover the
overpayment. See generally Bull v. United States, 

(1935).
                             50
sought by the plaintiff. See Livera v. First Nat’l State Bank
of New Jersey, 

, 1195 (3d Cir. 1989).

       CARCO’s equitable recoupment defense faces at
least two serious obstacles. As an initial matter, the United
States questions whether CARCO possesses a “claim”
against it, rather than a generalized request for the court to
balance the equities. Second, the United States questions
whether CARCO seeks relief of the same kind as the
United States. On both points, CARCO fails to meet its
burden.

       CARCO’s claim, such as it is, appears to be that the
equities favor CARCO, and require the United States to
bear the cost of the spill. CARCO argues that the United
States, through the Coast Guard, NOAA, and the Army
Corps of Engineers, had responsibility for maintaining the
anchorage where the allision occurred free of obstructions.
In the alternative, if the agencies were not responsible to
preemptively search for obstructions, CARCO argues they
should have more explicitly made clear that they were not
conducting such searches. CARCO asserts that it
reasonably believed, based on the agencies’ conduct, that
the agencies were maintaining the anchorage free of
obstructions. Additionally, CARCO argues that equity




                             51
requires the Oil Spill Liability Trust Fund to bear the cost
of the cleanup rather than CARCO.28


28
   Though it is not necessary to our holding, we note that
these equities do not appear to favor CARCO. As to
agency regulation and maintenance of the anchorage
where the allision occurred, the District Court held that the
agencies did not have a duty to maintain the anchorage
free of obstructions. The United States does not
preemptively search for obstructions in the anchorage, it is
not responsible for doing so, and it did not tell CARCO
that it would do so. To the extent CARCO believed
otherwise, CARCO simply misunderstood the regulatory
structure and the responsibilities (and indeed, the
capabilities) of the agencies.
Additionally, to the extent—if at all—that the Coast
Guard, NOAA, and the Army Corps of Engineers were
responsible for the Athos I oil spill, reducing the recovery
of the United States in this case would not be equitable.
Beyond our concerns relating to subrogation (equity
would certainly not favor reducing Frescati’s recovery
under these circumstances), such a decision would impose
liability on the Oil Spill Liability Trust Fund, not the
responsible agencies. Any recovery based on the United
States’ subrogated claim flows back to the Trust Fund, out
of which the United States originally reimbursed Frescati.
26 U.S.C. § 9509(b)(3). The Trust Fund is not intended (or
                             52
       Equitable recoupment requires more than just a
request to balance the equities. CARCO points out that
although equitable recoupment most often arises in the
context of offsetting monetary claims, as in tax or
bankruptcy cases, it is not necessarily limited to those
situations. See, e.g., Oneida Indian Nation of New York v.


allowed by statute) to be used as a slush fund to cover the
liabilities of federal agencies. See 33 U.S.C. § 2712 (“Uses
of the Fund”).
As a final point, the purpose of the Trust Fund is not to
absorb the cost of cleaning up oil spills; indeed, almost the
opposite is true. The OPA creates a strict liability regime
for responsible parties, while capping that liability at a set
amount. But the Trust Fund was not designed to bear those
costs indefinitely; the subrogation provision of 33 U.S.C.
§ 2715 allows the United States, on behalf of the Trust
Fund, to pursue any claim a responsible party could have
brought against a third party under any law, in order to
recover the money paid out by the Trust Fund and preserve
the Trust Fund’s ability to respond quickly to spills in the
future. The OPA is intended to quickly compensate
victims of spills, minimize environmental damage, and
internalize the costs of oil spills within the oil industry.
The subrogation provision serves those purposes by letting
cleanup costs fall upon the liable party, rather than with
the Trust Fund.
                             53
New York, 

, 136–37 (N.D.N.Y. 2002)
(allowing an equitable recoupment defense in the context
of offsetting requests for declaratory judgments in a land
rights case). But CARCO still must assert some cognizable
claim, rather than simply a request for the Court to reduce
the United States’ damages in the interest of equity. Here,
CARCO has failed to do so.

        Neither does CARCO seek the same kind of relief
as the United States. The United States seeks contractual
relief, to which it is entitled by operation of statute. See 33
U.S.C. § 2715. CARCO, by contrast, seeks equitable
relief, or (on another reading) essentially tort-based relief
grounded in misrepresentation by the agencies. The
mismatched relief sought by CARCO and the United
States does not support CARCO’s equitable recoupment
defense.

      The requirement that a defendant seek the same
kind of relief as has been sought in the plaintiff’s claim is
a fundamental requisite for recoupment. The defense is not
intended to be a catch-all to allow any claims otherwise
barred by time, settlement, or statute to be heard as equity
seems to require. Equitable recoupment is intended to
allow only truly similar claims arising from the same
transaction to offset one another in the interest of equity
between the parties. As noted, equitable recoupment is
well-suited for disputes in which two claims arise out of
the same taxable event or the same contractual obligation,
                              54
 as often seen in tax or bankruptcy cases. When, as here,
 the plaintiff seeks relief on a contract, the defendant may
 not resort to equitable recoupment as a means to assert a
 non-contractual claim, whether sounding in an equitable-
 balancing analysis, in tort, or otherwise.

        CARCO has failed to meet its burden of
 establishing an equitable recoupment defense. It is liable
 to the United States in full.

VII.   Limitation of Liability under the Oil Pollution
       Act

         CARCO argues that a provision of the OPA, 33
 U.S.C. § 2702(d)(2)(B), limits its liability in this case to
 the same extent to which Frescati’s liability was limited—
 approximately $45 million. Because CARCO did not raise
 this defense with the requisite clarity until nearly ten years
 after this litigation began, the District Court concluded that
 CARCO waived it. We agree that the defense was waived.

        A District Court’s holding that an affirmative
 defense has been waived is reviewed for abuse of
 discretion. Cetel v. Kirwan Financial Group, Inc., 

, 506 (3d Cir. 2006). Waiver is appropriate if the
 party raising the defense did not do so at a “pragmatically
 sufficient time” and if the opposing party would be
 prejudiced if the defense were allowed. Charpentier v.
 Godsil, 

, 864 (3d Cir. 1991).

                              55
       Whether CARCO raised its defense at a
pragmatically sufficient time requires us to determine
when CARCO first raised the § 2702(d)(2)(B) defense.
CARCO argues that it first raised the limitation defense in
its 2005 answer to Frescati’s Amended Counterclaim by
referring to the OPA. The District Court concluded that
CARCO’s answer contained nothing that would have put
Frescati or the United States on notice that CARCO
planned to rely on a limitation of liability defense. In
general, “[a]n affirmative defense . . . ‘need not be
articulated with any rigorous degree of specificity, and is
sufficiently raised for purposes of [Fed. R. Civ. P. 8] by its
bare assertion.’ ” Moody v. Atl. City Bd. of Educ., 

, 218 (3d Cir. 2017) (quoting Zotos v. Lindbergh Sch.
Dist., 

, 361 (8th Cir. 1997)). Nevertheless, the
party asserting the defense must actually do so, and in a
way that gives fair notice of that defense.

       CARCO relies on the averment listed as its
“Seventh Separate Defense,” which reads simply: “The
claims and causes of action set forth in the plaintiffs’
Amended Counterclaim are barred in whole or in part by
the provisions of the Oil Pollution Act of 1990, 33 U.S.C.
§ 2701, et seq.” JA at 355. Noticeably absent from this
general averment is any specific citation to the limitation
of liability defense or even a description of the nature of
the defense. This is significant, because the OPA includes
a number of potential affirmative defenses. See, e.g., 33

                             56
U.S.C. § 2702(b) (limiting scope of damages for which the
OPA imposes liability); § 2702(c) (excluding certain oil
spills from OPA liability); § 2702(d)(1)(A) (shifting
liability under the OPA to a solely responsible third party);
§ 2702(d)(2) (limiting the liability of certain parties under
the OPA); § 2703 (“Defenses to liability”). CARCO’s
general reference to the entirety of the OPA did not
provide adequate information from which Frescati could
determine that CARCO was seeking to limit its liability
under § 2702(d)(2)(B). Nor did CARCO develop this
defense at any point before the first trial. For that reason,
CARCO’s unspecified reference to the OPA did not
provide the requisite fair notice to Frescati.

       Furthermore, Frescati would be prejudiced if the
defense were allowed. As the District Court found, if
CARCO had asserted its defense in a timely fashion,
fifteen days of depositions and trial testimony from seven
witnesses could have been avoided, along with the OPA
damages phase of the first trial.29


29
   Allowing CARCO to assert the defense after failing to
raise it at a practicable time wastes the District Court’s
resources as well.
      Affirmative defenses must be raised as early
      as practicable, not only to avoid prejudice,
      but also to promote judicial economy. If a
                             57
        CARCO did not clearly assert the limitation defense
 until nearly a decade after this action commenced, and
 over a year after the first trial and appeal had concluded.
 The District Court appropriately concluded that CARCO
 had not raised the defense at a pragmatically sufficient
 time, and that Frescati would be prejudiced if the defense
 were allowed. The District Court did not abuse its
 discretion in finding the defense waived.30

VIII.   Prejudgment Interest Rate

        The District Court awarded Frescati prejudgment
 interest of just over $16 million. Frescati, in its cross-
 appeal from the District Court’s judgment, argues that the
 District Court erred by using the federal postjudgment

        party has a successful affirmative defense,
        raising that defense as early as possible, and
        permitting a court to rule on it, may terminate
        the proceedings at that point without wasting
        precious legal and judicial resources.
 Robinson v. Johnson, 

, 137 (3d Cir. 2002).
 30
    It is worth noting that the United States similarly waived
 a defense by its failure to raise an argument in the first
 trial. We previously held that the United States waived its
 right to object to CARCO’s equitable recoupment defense
 on the basis that it violated the terms of the partial
 settlement agreement. In re 

.
                              58
interest rate set by 28 U.S.C. § 1961(a) to determine the
amount of the prejudgment interest award. Specifically,
Frescati argues that the District Court improperly believed
itself bound to use the federal postjudgment rate rather
than the higher U.S. prime rate because Frescati did not
present evidence of its borrowing costs.

      An award of prejudgment interest is reviewed for
abuse of discretion. Ambromovage v. United Mine
Workers of Am., 

, 981–82 (3d Cir. 1984); see
also Sun Ship, Inc. v. Matson Nav. Co., 

, 63
(3d Cir. 1986). When selecting an interest rate, the District
Court must keep in mind that the rate and corresponding
award “must be compensatory rather than punitive.” Del.
River & Bay Auth. v. Kopacz, 

, 634 (3d Cir.
2009).

      Here, the District Court awarded Frescati
prejudgment interest at the one-year Treasury rate—the
same rate used as the federal postjudgment interest rate.
See 28 U.S.C. § 1961(a). Importantly, the District Court
found that the postjudgment rate would “fairly and
adequately compensate Frescati for its losses.” JA at 183.

      Frescati argues that, in the absence of evidence of
borrowing costs, we should require the use of the U.S.
prime rate. We grant that, had the District Court chosen to
use the prime rate, it would not have abused its discretion
even without extensive proof of borrowing costs. Taxman
                             59
v. Bd. of Educ., 

, 1566 (3d Cir. 1996) (en
banc). Indeed, the prime rate is commonly used to
approximate the cost the defendant would have paid to
borrow in the market, and at least one court appears to
require it. See, e.g., Gorenstein Enters., Inc. v. Quality
Care-USA, Inc., 

(7th Cir. 1989) (requiring
use of the prime rate in certain circumstances); see also
Forman v. Korean Air Lines Co., 

, 450–51
(D.C. Cir. 1996) (“[T]he prime rate is not merely as
appropriate as the Treasury Bill rate, but more appropriate
. . . .”). In this Circuit, however, a district court is not
constrained to the use of only the prime rate: “[i]n
exercising [its] discretion, . . . the court may be guided by
the rate set out in 28 U.S.C. § 1961.” Sun 

; 

(“[A] court ‘may’ use the
post-judgment standards of 28 U.S.C. § 1961(a) [to
calculate prejudgment interest, though] it is not compelled
to do so.”).31

      The District Court determined that the federal
postjudgment rate “fairly and adequately compensate[s]




31
   Nor was it an abuse of discretion for the District Court
to adopt a variable interest rate. Interest accumulated for
more than a decade, and during that time prevailing
interest rates changed substantially.
                             60
Frescati for its losses.” JA at 183. Under our Court’s
precedent, the District Court acted within its discretion.

IX.   Conclusion

       The District Court’s order dated August 17, 2016
will be affirmed in part, vacated in part, and reversed in
part. The District Court’s judgment in favor of Frescati on
the breach of contract claim and the prejudgment interest
award will be affirmed. The District Court’s judgment in
favor of Frescati on the negligence claim will be vacated.
The District Court’s judgment in favor of the United States
will be affirmed in part with respect to CARCO’s liability
on the subrogated breach of contract claim, but the
judgment will be reversed and remanded for further
proceedings in light of our equitable recoupment ruling for
the purpose of recalculating damages and prejudgment
interest. The District Court’s order dated April 9, 2015,
denying CARCO’s motion for partial summary judgment
on its limitation of liability defense, will be affirmed.




                            61