Atlas Air, Inc. v. International Brotherhood of

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued September 7, 2018 Decided July 5, 2019

No. 17-7172

ATLAS AIR, INC. AND POLAR AIR CARGO WORLDWIDE, INC.,
APPELLEES

v.

INTERNATIONAL BROTHERHOOD OF TEAMSTERS, ET AL.,
APPELLANTS

Appeal from the United States District Court
for the District of Columbia
(No. 1:17-cv-01953)

Edward M. Gleason, Jr. argued the cause for appellants.
With him on the briefs were James Petroff and Trent R. Taylor.
Joshua D. McInerney entered an appearance.

Robert A. Siegel argued the cause for appellees. With him
on the brief were Rachel Janger, Michael G. McGuinness, and
Sloane Ackerman.

Before: GRIFFITH, Circuit Judge, and EDWARDS and
RANDOLPH, Senior Circuit Judges.

Opinion for the Court filed by Circuit Judge GRIFFITH.
 2
GRIFFITH, Circuit Judge: The district court issued a
preliminary injunction enjoining a union’s efforts to gain
leverage over two commercial air carriers during negotiations
over an amended collective bargaining agreement. Congress
permits courts to issue such injunctions in rare circumstances.
Because this is one of them, we affirm.

I

Atlas Air, Inc. and Polar Air Cargo Worldwide, Inc.
(collectively, “Atlas”) are global commercial air carriers that
operate domestic and intercontinental flights for the U.S.
military, DHL, and Amazon, among others. Atlas’s pilots are
represented by the International Brotherhood of Teamsters; the
International Brotherhood of Teamsters, Airline Division; and
the Airline Professionals Association of the International
Brotherhood of Teamsters, Local Union No. 1224. We refer to
them collectively as the “Union.”

In 2011, after a protracted negotiation process, the Union
and Atlas entered into a collective bargaining agreement
(CBA). The CBA prohibits the Union from engaging in a work
stoppage or slowdown and permits Atlas to seek an injunction
if the Union does so. The CBA also creates a process to resolve
any “grievance[s]” that Atlas has over the “interpretation or
application” of its provisions. Defs.’ Ex. 1 at 126-27, No. 17-
cv-1953 (D.D.C. Nov. 30, 2017), Dkt. No. 31-1.

Since entering the CBA, Atlas’s business model and
staffing demands have changed significantly because of the
rapid expansion of e-commerce. In the past, most of Atlas’s
business was international. Of late, the company’s focus has
shifted to a growing domestic market. The Union tried to work
“collaboratively with Atlas” to alleviate the growing pains
caused by this change rather than holding the company
 3
“accountable” to the precise terms of the CBA. J.A. 172-73.
But as domestic operations expanded, pilots’ frustrations
increased. In 2014, the pilots elected a new Chairman of the
Atlas Pilots’ Executive Committee, the body through which the
Union manages day-to-day representation of the pilots. Captain
Robert Kirchner ran on a platform calling for “strict contract
compliance.” J.A. 173. As he explained, if Atlas is “allowed to
bend and violate the terms of the [CBA] when it suits them,
[the company] will have no reason to negotiate changes to the
[CBA]” when it becomes amendable. Id. When he assumed the
role of Chairman in January 2015, Captain Kirchner launched
several communication tools to help educate pilots about their
rights and responsibilities under the CBA: “Atlas Teamsters
Action Message” podcasts (ATAM), “Atlas Pilots Crew Call”
question and answer sessions, “Chairman’s Update” emails,
and “CBA Chat” videos.

On February 16, 2016, about one year after Captain
Kirchner took office, the Union notified Atlas that it would
seek to amend the existing CBA. Around that same time, the
Executive Committee’s communications efforts picked up
steam. The day before the Union issued that notice, the
Communications Chairman, Captain Michael Griffith, asked
rhetorically on an ATAM, “Are you going to continue[] to sell
your talents for a quick buck, or are you going to stop doing the
Company favors and follow the CBA to the letter and give your
[Executive Committee] and Negotiation Committee the
leverage and power they need today?” J.A. 638. Over the
coming months, the Union repeatedly called on pilots to be “all
in,” “fly the CBA,” and “fly the contract.” E.g., J.A. 631, 650;
Pls.’ Ex. 51 at 3, No. 17-cv-1953 (D.D.C. Nov. 30, 2017), Dkt.
No. 5-54; Pls.’ Ex. 100 at 54, No. 17-cv-1953 (D.D.C. Nov. 30,
2017), Dkt. No. 27-4. It encouraged pilots to “SHOP,” or “stop
helping out Purchase,” named for the location of Atlas’s
headquarters in Purchase, New York. E.g., J.A. 640. According
 4
to Captain Kirchner, “SHOP” or “shopping” refers to the idea
that pilots should not “help out” Atlas “by permitting [it] to get
away with contract violations,” but should instead insist on
“strict contract compliance.” J.A. 178. The Union asked pilots
to “BOOT,” which stands for “block out on time.” E.g., J.A.
652. By contrast, Atlas encourages pilots to “block out”—i.e.,
push back from the gate—up to fifteen minutes early as a
matter of course, or even earlier with Atlas’s approval. The
Executive Committee also encouraged pilots to think more
carefully about when to call in sick or accept overtime work.

Atlas was unhappy with these efforts and the changes it
began to see in pilots’ behavior. Atlas viewed SHOP and
BOOT as part of a Union attempt to orchestrate a work
slowdown in an attempt to ratchet up pressure on Atlas during
their negotiations over an amended CBA. When Atlas could
not convince the Union to stop this behavior, the company
asked the district court for an injunction. The Union disputed
Atlas’s allegations and moved to dismiss for lack of
jurisdiction. After a three-day evidentiary hearing, the district
court determined that it had jurisdiction and entered a
preliminary injunction to prevent the Union from encouraging
pilots to “block out on time,” call in sick on short notice, and
refuse to volunteer for overtime shifts. Atlas Air, Inc. v. Int’l
Bhd. of Teamsters, 280 F. Supp. 3d 59 (D.D.C. 2017). The
Union appealed. We have jurisdiction pursuant to 28 U.S.C.
§ 1292(a)(1) and 29 U.S.C. § 110.

II

We begin by asking whether the district court had
jurisdiction to issue this type of preliminary injunction. Our
review is de novo. Foretich v. Am. Broad. Cos., 198 F.3d 270,
273 (D.C. Cir. 1999). To answer that question, we look to the
Railway Labor Act and the Norris-LaGuardia Act.
 5

A

In the Railway Labor Act (RLA), 45 U.S.C. § 151 et seq.,
Congress established different procedures to resolve two types
of labor disputes in the transportation industry, which we refer
to as major and minor disputes. Consol. Rail Corp. v. Ry. Labor
Execs.’ Ass’n (Conrail), 491 U.S. 299, 302 (1989). A major
dispute concerns the formation or amendment of a collective
bargaining agreement. Id. The process for resolving a major
dispute is complex and typically takes a long time. Only once
that process is complete may the company or the union alter the
status quo by engaging in a work slowdown or stoppage. Bhd.
of R.R. Trainmen v. Jacksonville Terminal Co., 394 U.S. 369,
378 (1969). Delaying the time at which labor or management
may use economic self-help encourages compromise and
prevents interruptions to commerce or carriers’ operations. Id.;
Detroit & Toledo Shore Line R.R. Co. v. United Transp. Union,
396 U.S. 142, 149-50 (1969); see 45 U.S.C. § 152, First. The
status quo requirement is thus at the “heart” of the RLA, and
may be enforced by injunction. Jacksonville Terminal, 378
U.S. at 377-78; see 45 U.S.C. § 152, First; Conrail, 491 U.S. at
302-03.

By contrast, a minor dispute involves a question about how
to interpret an existing collective bargaining agreement, like
the meaning of a term or whether the agreement permits a
certain action. Elgin, J. & E. Ry. Co. v. Burley, 325 U.S. 711,
723 (1945). As long as the contested “action is arguably
justified by the terms of the parties’ collective-bargaining
agreement,” we treat the dispute as minor. Conrail, 491 U.S. at
307; Air Line Pilots Ass’n, Int’l v. E. Air Lines, Inc. (Eastern),
869 F.2d 1518, 1521 (D.C. Cir. 1989). The resolution process
for a minor dispute is less involved, and there is no “general
statutory obligation . . . to maintain the status quo” while that
 6
process is ongoing. Conrail, 491 U.S. at 304. So although
“[c]ourts may enjoin strikes arising out of minor disputes” in
limited circumstances, they generally may not enjoin other
violations of the status quo. Id. If in doubt, the dispute is minor.
Eastern, 869 F.2d at 1521.

Labor disputes are also subject to the Norris-LaGuardia
Act (NLGA), 29 U.S.C. § 101 et seq., which Congress enacted
in response to concerns that federal courts were using their
injunctive power too often to the detriment of workers. Bhd. of
R.R. Trainmen v. Chi. River & Ind. R.R. Co., 353 U.S. 30, 40
(1957). To curtail such judicial interference, Congress stripped
federal courts of “jurisdiction to issue any . . . temporary or
permanent injunction in a case involving or growing out of a
labor dispute, except in . . . strict conformity” with various
procedural requirements. 29 U.S.C. § 101. The NLGA also
categorically eliminates jurisdiction to enjoin certain types of
conduct in “any labor dispute,” including work stoppages and
slowdowns. Id. § 104.

We cannot have jurisdiction to enjoin slowdowns or work
stoppages in major labor disputes in the transportation industry,
see 45 U.S.C. § 152, yet at the same time lack jurisdiction to
enjoin such conduct in “any labor dispute,” 29 U.S.C. § 104.
More than 50 years ago, the Supreme Court resolved this
conflict by holding that “the specific provisions of the [RLA]
take precedence.” Chi. River, 353 U.S. at 41-42, 41 n.23.
Courts therefore have jurisdiction to issue injunctions to
preserve the status quo in major disputes in the transportation
industry, Conrail, 491 U.S. at 302-03, but in keeping with the
goals of the NLGA, they should only do so if “that remedy
alone can effectively guard the plaintiff’s right[s],” Int’l Ass’n
of Machinists v. Street, 367 U.S. 740, 773 (1961). In order to
verify that such an RLA injunction is indeed essential, courts
must generally comply with the procedures set forth in the
 7
NLGA before issuing RLA injunctions. See 29 U.S.C. §§ 101,
105-109; United Air Lines, Inc. v. Int’l Ass’n of Machinist &
Aerospace Workers (United), 243 F.3d 349, 362 n.9 (7th Cir.
2001); Delta Air Lines, Inc. v. Air Line Pilots Ass’n, Int’l, 238
F.3d 1300, 1307, 1310 (11th Cir. 2001).

B

The Union sees this as a minor dispute. In its view, the
question of whether the Union’s conduct is permissible
concerns how best to interpret or apply the existing CBA, and
such grievances about the CBA’s interpretation or application
must be resolved using the specific procedures outlined in that
agreement. So because the Union has demonstrated that the
existing CBA “arguably resolves the dispute one way or
another,” the dispute is minor. See Union Br. 23. Atlas, by
contrast, claims this is a major dispute because the Union
encouraged pilots to push back from the gate exactly on time,
call in sick on short notice, and decline overtime in order to
pressure Atlas in negotiations for an amended CBA.

The district court agreed with Atlas, and so do we. A
dispute over the terms of a new or amended collective
bargaining agreement is unequivocally major. E.g., Nat’l R.R.
Passenger Corp. v. Transp. Workers Union of Am. (Amtrak),
373 F.3d 121, 123 (D.C. Cir. 2004). So too is a dispute over
conduct that “grows out of” the effort to negotiate that
agreement. Id. at 125-26 (where a proposed strike over
Amtrak’s funding “gr[ew] out of” a dispute over the terms of a
new collective bargaining agreement, and both the strike and
agreement would alter pay and working conditions, the strike
“gr[ew] out of” the major dispute). This makes perfect sense:
The purpose of the RLA’s status quo requirement is to maintain
the parties’ respective positions while they negotiate future
rights. See id. at 123. When one party alters the status quo in
 8
order to put “economic pressure[]” on the other to enhance its
own bargaining position, that conduct is part of a major dispute
and may be enjoined. United, 243 F.3d at 365 (finding that a
general slowdown campaign, which was designed to pressure
United during negotiations for a new collective bargaining
agreement, constituted a major dispute); accord Delta, 238
F.3d at 1307-08 (same, with respect to concerted effort to
decline overtime); see also Long Isl. R.R. Co. v. Sys. Fed’n No.
156, 368 F.2d 50 (2d Cir. 1966) (affirming order enjoining
union exhortations to “follow the rules strictly”). That is
precisely what Atlas alleges happened here.

This is not a case about whether the existing CBA arguably
permits the Union or its members to act in this manner, or
whether the CBA establishes a process to resolve disputes
about its interpretation. In fact, Atlas has consistently
maintained that this dispute cannot be resolved by the existing
agreement, and has disclaimed any reliance on the provision in
the CBA that prohibits slowdowns for purposes of this case. Cf.
Conrail, 491 U.S. at 311-12 (finding the dispute to be minor
where both parties’ arguments relied on the meaning of the
existing agreement, and whether a certain practice was
impliedly justified by that agreement); Eastern, 869 F.2d at
1524 (same, where both parties invoked a specific provision in
the existing agreement to justify their positions). Rather, as the
district court explained, this is a dispute “about whether the
Union may engage in a concerted campaign to alter the status
quo (by SHOP-ing, BOOT-ing and the like) in the midst of
what is unmistakably a major dispute (the negotiation of [an
amended] CBA) in order to apply economic pressure on the
company in those negotiations.” Atlas Air, 280 F. Supp. 3d at
80-81. Because the existing CBA does not even arguably speak
to whether this conduct is permissible when done in
furtherance of that particular goal, this is a major dispute.
 9
Atlas has presented compelling evidence in support of its
assertion that this is a major dispute. The Union’s own
statements demonstrate that it frequently encouraged pilots to
take the very actions Atlas challenges as a means to gain
leverage in the negotiations over amending the CBA. See
Amtrak, 373 F.3d at 124-25 (looking to unions’ words and
deeds to confirm Amtrak’s assertion that the dispute was
major). In ATAMs, Captain Griffith encouraged pilots to
“follow the CBA to the letter and give [the Executive
Committee] and Negotiation Committee the leverage and
power they need,” J.A. 638, and explained that “[SHOP] goes
hand-in-hand with following the CBA,” J.A. 640. The same
month the Union and Atlas finalized a negotiation protocol for
the amended CBA, Captain Kirchner told pilots to continue
their efforts to “SHOP, BOOT and push back on [Atlas’s]
tactics harder than ever as we are starting to get the movement
we desire.” J.A. 728. Don’t “start violating the CBA,” “resign
yourselves to the status quo,” or “abandon our quest for an
industry-leading CBA,” he implored. Id. Combined with the
additional examples in the record and district court opinion, see
infra, Part III.A; Atlas Air, 280 F. Supp. 3d at 78-79, 90-92, 97-
98, 101, Atlas has presented sufficient evidence to
demonstrate, for purposes of jurisdiction, that the challenged
actions grew out of the major dispute, see Amtrak, 373 F.3d at
124-25; United, 243 F.3d at 363; Delta, 238 F.3d at 1307-08.

The Union claims, however, that sections 8 and 7(e) of the
NLGA stripped the district court of its jurisdiction to enter this
injunction. Section 8 provides that a court may not enter an
injunction in a case involving a “labor dispute” if the plaintiff
“has failed to make every reasonable effort to settle [that]
dispute.” 29 U.S.C. § 108. Section 7(e) strips a court of
“jurisdiction to issue a temporary or permanent injunction in
any case involving or growing out of a labor dispute” unless
the court specifically finds “[t]hat the public officers charged
 10
with the duty to protect complainant’s property are unable or
unwilling to furnish adequate protection.” Id. § 107(e). The
Union is correct that these provisions can strip a court of
jurisdiction in a major dispute. See supra, Part II.A; Bhd. of
R.R. Trainmen v. Akron & Barberton Belt R.R. Co., 385 F.2d
581, 613-14 (D.C. Cir. 1967) (NLGA section 8); Green v.
Obergfell, 121 F.2d 46 (D.C. Cir. 1941) (NLGA section 7(e)).
But not here.

Atlas has shown that, as required by section 8 of the
NLGA, it made “every reasonable effort to settle [this]
dispute.” 29 U.S.C. § 108. Again, Atlas and the Union disagree
about how to characterize this labor dispute: Atlas says that it
needed to exert every reasonable effort to resolve the dispute
over the terms of the amended CBA, and has done so. The
Union disagrees and asserts that “[t]he NLGA ‘labor dispute’
between the parties . . . is indisputably over whether an
improper slowdown occurred under the parties’ existing
CBA.” Union Reply 14. To resolve that dispute, Atlas had to
use the “mandatory grievance and arbitration procedures” in
the CBA and RLA, among other things. Union Br. 34-36. It did
not, and the Union argues that the district court accordingly
lacked jurisdiction because “[t]he RLA ‘clearly states’ that this
arbitration requirement is jurisdictional” and that “[f]ederal
courts have no jurisdiction over RLA ‘minor’ disputes such as
this one.” Id. at 22 (first quoting Oakey v. U.S. Airways Pilots
Disability Income Plan, 723 F.3d 227, 237 (D.C. Cir. 2013),
and citing Conrail, 491 U.S. at 304). The problem with the
Union’s argument is that it is based on the assumption that this
case involves a “minor dispute” requiring interpretation of the
parties’ CBA which, as we have already explained, is incorrect.

The Union also argues that, per the CBA, Atlas must file a
grievance with the Union and submit to arbitration its claim
that the Union is directing an unlawful slowdown before the
 11
company can seek an RLA injunction in federal court. See 45
U.S.C. § 152, First. Because Atlas has not done so, it cannot
satisfy NLGA section 8. We find no merit in the Union’s claim,
which is based on a “frivolous or obviously insubstantial”
reading of the CBA. Conrail, 491 U.S. at 307, 310. The
relevant section of that agreement provides:

[T]he Union . . . agrees that during the term of the [CBA]
there will not be any complete or partial strikes, picketing,
[or] slowdowns . . . unless and until the parties’ rights to
self-help mature under the Railway Labor Act . . . . This
paragraph shall not alter or limit [Atlas’s] right, if any, to
obtain a court order enjoining such conduct by the Union
and or the [pilots] both collectively and individually.

J.A. 624 (emphasis added). As Atlas correctly notes, this
provision “reflects an agreement that the contractual no-strike
clause [would] be in addition to, and not in lieu of,” Atlas’s
right to ask a federal court for a status quo injunction under the
RLA. Atlas Br. 39 n.7. The Union offers no convincing
response to these arguments. Thus, Atlas was not required to
pursue the grievance/arbitration process set forth in the CBA
before asking a court to enjoin the Union’s concerted actions.

We agree with the district court that Atlas has made every
reasonable effort to resolve its disputes with the Union. Atlas
Air, 280 F. Supp. 3d at 85-86. With respect to the terms of the
amended CBA, the parties began negotiating in 2016;
mediated, arbitrated, and litigated a dispute related to those
negotiations; worked for eight months to create a framework to
govern the negotiation process; then spent about two and a half
months in negotiations before Atlas filed this complaint. See
id. at 72-73 (describing these efforts). Moreover, the district
court found that Atlas “made efforts to resolve the slowdown
short of litigation,” as demonstrated by communications with
 12
the Union and its counsel. Id. at 86. As the district court found,
on the present record, that is sufficient. Id. at 85-86; see United,
243 F.3d at 364-65; see also Grand Trunk W. R.R. Inc. v. Bhd.
of Maint. of Way Emps. Div., 497 F.3d 568, 572-73 (6th Cir.
2007) (finding section 8 satisfied where parties engaged in
unsuccessful negotiations more than once); San Antonio Cmty.
Hosp. v. S. Cal. Dist. Council of Carpenters, 125 F.3d 1230,
1238-39 (9th Cir. 1997) (same); cf. Aircraft Serv. Int’l, Inc. v.
Int’l Bhd. of Teamsters, 779 F.3d 1069, 1078 (9th Cir. 2015)
(en banc) (section 8 not satisfied where one party sought an
injunction before attempting to settle or negotiate with the
other).

Furthermore, even though section 7 of the NLGA applies
to all cases “involving or growing out of a labor dispute,”
courts have consistently interpreted subsection (e) to apply
only where one party has threatened violence against the
person or physical property of another. Donnelly Garment Co.
v. Dubinsky, 154 F.2d 38, 43 (8th Cir. 1946); Carter v. Herrin
Motor Freight Lines, 131 F.2d 557, 561 (5th Cir. 1942); Wilson
& Co. v. Birl, 105 F.2d 948, 950 (3d Cir. 1939). As the Third
Circuit explained, “it would be unreasonable to construe [this]
subsection to include losses which . . . the powers of the police
are hardly calculated to prevent,” like the loss of one’s
customers because of strikes or picketing. Wilson, 105 F.2d at
949-50. The one case from our court that the Union cites in
support of this argument fits with this rule. See Green, 121 F.2d
at 53-54 (finding that the district court lacked jurisdiction
because the record “show[ed] violence and destruction of
property” but there was no evidence that the public officers
were “unable or unwilling to furnish adequate protection”). As
the present record does not contain any evidence of threatened
or actual violence to Atlas’s persons or property, section 7(e)
is no bar to the district court’s authority to enjoin the Union’s
conduct. Accord Atlas Air, 280 F. Supp. 3d at 87.
 13

III

In general, a plaintiff seeking a preliminary injunction
must show (1) a likelihood of success on the merits; (2) a
likelihood of irreparable harm absent such relief; (3) that the
equities favor the plaintiff’s position; and (4) that the injunction
is in the public’s interest. Winter v. Nat. Res. Def. Council, Inc.,
555 U.S. 7, 20 (2008). But see Conrail, 491 U.S. at 303 (courts
may enjoin status quo violations in major RLA disputes
without a showing of irreparable injury). We review the
issuance of a preliminary injunction for abuse of discretion,
although we review the court’s underlying legal conclusions de
novo and factual findings for clear error. Davenport v. Int’l
Bhd. of Teamsters, 166 F.3d 356, 361 (D.C. Cir. 1999). “A
factual finding is clearly erroneous if it is ‘without substantial
evidentiary support or if it was induced by an erroneous
application of the law.’” Air Line Pilots Ass’n Int’l v. E. Air
Lines, Inc., 863 F.2d 891, 894 (D.C. Cir. 1988) (quoting Cuddy
v. Carmen, 762 F.2d 119, 124 (D.C. Cir. 1985)).

Because the Union’s argument on appeal is confined to the
question of whether Atlas has demonstrated a likelihood of
success on the merits, our analysis is as well. See Abdullah v.
Obama, 753 F.3d 193, 199 (D.C. Cir. 2014) (issues not raised
in opening briefs are forfeited).

A

To demonstrate a likelihood of success in this kind of case,
Atlas must show that the status quo changed during a major
dispute in violation of the RLA, see 45 U.S.C. § 152, First, and
provide “clear proof” that the Union participated in, authorized,
or encouraged that change, 29 U.S.C. § 106; see United, 243
F.3d at 363-66; Delta, 238 F.3d at 1309 & n.21.
 14

To determine whether the Union encouraged a change in
the status quo during a major dispute, we must know when that
dispute began. The Union says it started in January 2015 when
Captain Kirchner was elected head of the Executive Committee
and began “to educate the pilots about their contractual rights
and regulatory responsibilities.” Union Br. 6; see id. at 43-45.
It was in that context that he started calling for strict CBA
compliance. In Atlas’s view, the relevant date is February
2016, when the Union notified Atlas of its intent to negotiate
an amended CBA, intensified its messaging campaign, and
announced SHOP and BOOT. It is clear from the district
court’s analysis that the court agreed with Atlas. See Atlas Air,
280 F. Supp. 3d at 78-103. We do as well. As we have
explained, this case is about a major dispute over the terms of
an amended CBA. That dispute began in February 2016.

Of course, a union is allowed to educate its members about
their contractual rights and safety obligations, and in that
context, it may not be a problem to call for strict compliance
with the contract. A union may not, however, encourage strict
compliance with the terms of an existing agreement in an effort
to gain leverage in negotiations for a new or amended contract.
When a union changes the status quo in aid of such an effort,
the district court may enjoin the union’s conduct. That is just
what happened here with respect to blocking out, short-notice
sick calls, and overtime.

1. Blocking Out On Time

Atlas asserts that pilots were substantially more likely to
block out (i.e., push back from the gate) precisely on time after
the Union notified Atlas of its intent to negotiate an amended
CBA. Departing exactly on time, as opposed to when the
aircraft is “loaded and ready,” removes any potential “buffer,”
 15
meaning the flight is more likely to arrive late if it encounters
other issues, like headwinds or congestion. Id. at 100-01. This
threatens to cause cascading delays across flights by forcing
Atlas to shuffle runways, schedules, and gates, particularly at
Atlas’s biggest hubs where flights depart every two-to-three
minutes.

Atlas’s expert, Dr. Lee, presented statistically significant
evidence showing that from January 1, 2012 to February 15,
2016, nearly 80% of flights blocked out early and 13% blocked
out precisely on time. J.A. 464-67 (Lee Decl.). This
consistently low percentage of precisely on-time departures
reasonably suggests that pilots were not waiting to depart until
scheduled. But from February 16, 2016 to August 2017, about
34% of flights blocked out early and 53% blocked out precisely
on time. Id. In addition, since February 2016, “the percentage
of flights blocking out twenty-five or more minutes before the
[estimated departure time] dropped from 12.1% to 3.3%,” and
“flights unaffected by other delays have blocked out an average
of six minutes less early.” Atlas Air, 280 F. Supp. 3d at 100
(citing Dr. Lee’s analysis). Notably, these statistics exclude
flights delayed for reasons outside Atlas’s control, like weather
or customer requests, and flights experiencing cascading
delays.

The Union’s responses to this compelling evidence lack
merit. 1 First, the Union disputes the reliability of Dr. Lee’s

1
The Union objects that the district court did not make separate,
explicit findings about what constituted the pre-dispute status quo
with respect to any of the challenged conduct. While true, we detect
no error. Courts resolving these type of cases do not typically make
such express findings; instead, they compare statistics about
challenged behaviors from before and after the relevant dividing
time. See generally United, 243 F.3d at 354-55, 363; Delta, 238 F.3d
at 1310-11. This approach makes perfect sense to us. As the Union
 16
analytical models. He used departure delays, but in the Union’s
view, arrival delays are a better metric because that is what
customers care about. The Union’s expert, Daniel Akins, found
that Atlas’s “poor on-time [arrival] performance predates” the
Union’s February 16, 2016 notice, pursuant to section 6 of the
RLA, that it wished to negotiate an amended CBA (“section 6
notice”). Union Br. 12 (citing Mr. Akins’s analysis); see 45
U.S.C. § 156. The Union therefore reasons that Atlas cannot
blame the Union for these changes. The Union also claims that
Dr. Lee made up the “loaded and ready” concept; that is not
something airlines or pilots track, and Atlas does not have a
rule requiring or allowing pilots to depart when the aircraft is
“loaded and ready.” Union Br. 11-12.

As the district court found, there were good reasons for Dr.
Lee to use departure time. Customers may be primarily
concerned with arrival times, but the BOOT campaign
encouraged pilots to depart on time, and departure delays
contribute to arrival delays. That is why blocking out as soon
as the aircraft is able to depart (i.e., “loaded and ready”) is
beneficial—it decreases the risk that an aircraft experiencing
other delays will arrive late. A pilot who is ready and able to
depart early, but instead sits at the gate until the scheduled
departure time (i.e., waits to block out on time) cuts down the
buffer. Because we agree that departure time is an appropriate
metric, whether Atlas’s flights arrived chronically late before
the section 6 notice is immaterial to whether the BOOT
campaign had a statistically significant effect on departure
times. As for the Union’s objections to the “loaded and ready”

conceded at argument, the status quo includes practices that have
developed over time, as reflected by the manner in which the
employer or employee actually acts. So if the district court finds—as
it did here—that statistical evidence demonstrates that the employer
or employee consistently acted in one manner for a given period of
time, that can establish a pre-dispute “status quo.”
 17
metric, it matters not that Atlas’s expert coined this term to
explain his analysis to the district court, and that Atlas did not
have a formal policy requiring pilots to depart once the plane
was “loaded and ready.” Since voluntary conduct can be part
of the “status quo,” the pertinent question in this case is what
the pilots normally did before the Union informed Atlas of its
intent to negotiate an amended CBA. See, e.g., Delta Air Lines,
238 F.3d at 1309-10 (encouraging pilots to decline optional
overtime assignments violated the status quo). In other words,
the question is what pilots normally did before the Union
assertedly began trying to change the status quo. Atlas’s
statistical evidence shows that, prior to February 2016, pilots
frequently departed before the scheduled time and only rarely
departed precisely on time, and after February 2016, the
inverse was true.

Second, the Union argues that blocking out on time is not
illegal, but instead ensures that pilots are “in compliance with
[Federal Aviation Administration] rest, flight and duty time
regulation[s].” Union Br. 51. That may be so, but the issue is
not whether pilots should block out on time; it is whether pilots
altered their prior practice of blocking out before the scheduled
time, when possible. And the answer to that question, the
district court fairly concluded, was yes.

Finally, the Union offers various alternative explanations
for these changes, including “the increasing complexity and
size of [Atlas’s] operations,” the shift towards employing more
junior-level pilots, an “improved emphasis on fatigue training
and awareness, and the increasing age of the Atlas fleet.” Id. at
14. But these ideas would at best shed light on why fewer
flights blocked out before the estimated departure time. The
Union’s theories do not explain the significant increase in
flights blocking out exactly on time—from 13% to 53%. On the
present record, the most plausible explanation for this conduct
 18
is the one Atlas offered: even if they were ready earlier, pilots
waited to depart until the scheduled time.

Atlas provided clear proof that the Union encouraged
pilots to act in this manner. The record is replete with
communications from the Union exhorting pilots to stop
departing early. For example, on a June 2016 ATAM, Captain
Griffith reminded listeners that “the amendable date of our
current [CBA]” is approaching, and commended pilots on their
efforts to “show[] our particular giant how we can [SHOP] and
[BOOT] and even buckle down and say, first you pay me.” J.A.
653. In his June 2017 Chairman’s Update email, Captain
Kirchner told pilots that Atlas believes if it “delay[s] a new
CBA long enough, you will . . . abandon our quest for an
industry-leading CBA. This cannot be allowed to be the case!
YOU must SHOP, BOOT and push back on their tactics harder
than ever as we are starting to get the movement we desire.”
J.A. 728. Another Executive Committee member made the
point even more explicit in an August 2017 Crew Call:
“Blocking out on time is . . . advantageous to us by giving
leverage to our Negotiating Committee and our [Executive
Committee] and our stewards.” J.A. 769. Given this evidence,
the district court did not improperly conclude that Atlas is
likely to succeed in demonstrating that by encouraging pilots
to “block out on time,” the Union altered the status quo.

2. Short-Notice Sick Calls

A “short notice” sick call refers to a pilot’s decision not to
fly, typically due to illness, made on the same day as the pilot’s
next scheduled flight. Atlas argues that pilots began calling in
sick on short notice at a higher rate after the Union served its
section 6 notice. This made it more difficult to find substitute
crew members, increasing the risk that flights would be
cancelled or delayed.
 19

In support, Atlas provided statistical evidence that, from
January 2013 to February 2016, only 14.4% of sick calls were
made the same day the pilot was scheduled to fly. J.A. 451 (Lee
Decl.). This suggests that pilots were not typically waiting until
the day of the scheduled flight to notify Atlas of their
unavailability. But from February 2016 to September 2017,
that number was 20.4%. Id. Because the reason for this
percentage change was a spike in short-notice sick calls
beginning in October 2016, Atlas’s expert also compared
statistics from October 1, 2015 to September 20, 2016, and
from October 1, 2016 to September 20, 2017. Using those
dates, short-notice sick calls increased from 13.8% to 23.8%,
while sick calls made at least two days in advance decreased
from 44.5% to 31.6%. Id. Moreover, during the October 2015
to September 2016 period, the highest percentage of short-
notice sick calls in a given month was 18%. Id. But from
October 2016 to September 2017, the monthly percentage
never dropped below 18%, and it exceeded 20% in every
month but one. Id.

The Union does not challenge the assertion that short-
notice sick calls increased in frequency. Instead, the Union
argues that “overall sick leave usage at Atlas did not increase.”
Union Br. 54 (emphasis added). But that misses the point. The
increased percentage of sick calls made at the last minute
supports Atlas’s theory that pilots were waiting to call in sick
in order to disrupt scheduled flights. The Union also questions
the timing of these changes. Of course, a change in behavior in
October 2016 does not provide perfect support for Atlas’s
theory that pilots began acting in this manner in February 2016
in response to this major dispute. But it is unrealistic to expect
massive changes in behavior to occur right away, or all at once.
See United Air Lines, Inc. v. Air Line Pilots Ass’n Int’l (ALPA),
563 F.3d 257, 260, 264 (7th Cir. 2009) (affirming injunction
 20
for a “job action that began in 2006 [and] escalated in 2008”).
The question, then, is whether Atlas has offered sufficient
evidence connecting this statistically significant change to the
Union’s conduct, or whether the record suggests this occurred
for another reason.

That brings us to whether the Union participated in or
called for this change in behavior. Of course, there is nothing
inherently suspect about a union encouraging its members not
to work sick. What is suspect is a union reminding its members
not to work sick while acknowledging the legal consequences
of calling for a sick out and continually emphasizing that strict
contract compliance will provide negotiation leverage.
Combined, this kind of evidence can fairly be interpreted as
encouraging slowdown activities. See id. at 272-73 (affirming
district court’s conclusion that statistical evidence, combined
with coded directives “to engage in job actions,” like “fly the
contract,” warranted a status quo injunction).

Atlas offered several examples to support its claim that the
Union’s conduct went beyond merely reminding pilots not to
work sick. The day the Union provided its section 6 notice, the
host of a CBA Chat told viewers he had called in sick because
of a high fever, then pulled a hot water bottle from under his
shirt. The Chat closed with this reminder: “It’s your CBA. They
signed it. You use it.” J.A. 637. The Union was aware that
expressly calling for a “sick out” would be “unethical” and
might expose the Union to a large fine. See J.A. 636 (describing
$40 million fine levied against another airline union for
encouraging a sick out). But it did encourage pilots to be “all
in,” to SHOP, and to “fly the CBA.” E.g., J.A. 631, 650; Pls.’
Ex. 51 at 3, No. 17-cv-1953 (D.D.C. Nov. 30, 2017), Dkt. No.
5-54. The Union directly connected the CBA, these coded
terms, and the sick calls when it “commend[ed]” pilots for their
“vigilance in holding up the provisions of our CBA” and
 21
reminded them that, “in the midst of [this] period of increased
friction and hostility, . . . make absolutely certain that you are
fit for duty each and every time you report to work. Please, do
not fly sick or fatigued. All In, Every Day.” J.A. 774.

To be sure, Atlas’s argument is not ironclad. It could not
identify any specific example of an individual pilot who had
abused the sick leave policy. Atlas Air, 280 F. Supp. 3d at 89.
The fit between the Union’s language and the pilots’ conduct
is not exact, as these communications relate to calling in sick,
not calling in sick at the last minute. And, as the district court
acknowledged, “[s]tanding alone,” the claim that the Union
encouraged pilots not to fly sick “might not carry the day.” Id.
at 90. But the district court found that specific examples and
“more explicit language” were not necessary because, when
viewed in context, “it is evident that at least some of the
Union’s exhortations were tied to the ongoing labor dispute.”
Id. at 90-91. That was not reversible error.

3. Open Time

Finally, we turn to overtime, known in this industry as
“open time.” Airlines use open time when both the primary and
reserve crew are unable to fly as scheduled. Filling an open
time slot involves calling individual pilots in order of seniority
until Atlas finds someone willing to work that flight. Unfilled
open time creates cascading delays because the airline must
move pilots from other flights to cover these shifts. At its worst,
it may take days to get back to the normal schedule. As such,
although each pilot may choose whether to accept an available
open time shift, Atlas relies on the assumption that most open
time positions will be filled.

The district court found that Atlas presented sufficient
statistical evidence to support its claim that, since the Union
 22
served the section 6 notice, it has become more difficult to fill
open time. In 2015, Atlas made an average of one-and-a-half
calls to fill each open time assignment. J.A. 456 (Lee Decl.). In
2016, that number increased to over two calls per assignment,
and in 2017, to over three. Id. at 457. Still, more assignments
remained unfilled. From January 2015 to February 2016, about
2.5% of all open time assignments were not filled, and unfilled
assignments never exceeded 5% in a given month. See id. at
459. Between March 2016 and August 2017, around 10% of
total assignments were unfilled. See id. In eleven of those
eighteen months, the unfilled rate exceeded 5%; in three
months, it exceeded 20%. Id. As with sick leave, however, a
statistically significant shift in open time did not occur
immediately after the Union provided its section 6 notice in
February 2016. See id. Furthermore, although the overall trend
demonstrates an increase in unfilled open time, the statistics are
not consistent from one month to the next. Between November
2016 and March 2017, for instance, the percentage of unfilled
open time fluctuated from 11% to 42%, 5%, 0%, and 23%. Id.

The Union suggests that these changes resulted from a
decrease in total open time and a decline in the desirability of
those assignments. In support, the Union points to a purported
2015 policy change that required more pilots to “deadhead” on
company aircraft to get to the location of their next assignment.
Deadhead flights tend to be less direct, less comfortable, and
less reliable than commercial flights. The Union also suggests
that this change might have occurred because individual pilots
chose to decline open time for personal reasons, such as a
desire to spend more time with their families, or because they
did not need the extra money.

The district court rejected these alternative explanations as
“more supposition than evidence.” Atlas Air, 280 F. Supp. 3d
at 97. That was not clearly erroneous. The Union’s expert, Mr.
 23
Akins, was unusually equivocal about these ideas, emphasizing
that they were “possible” or “may” explain these changes in
behavior. J.A. 252-53 (Akins Decl.). He acknowledged that it
is “counterintuitive” to expect that “the share of open trips
uncovered would increase as open trips decreased.” Id. at 253.
We agree. We would naturally expect that, if there are fewer
opportunities to earn extra income, more pilots would be
interested in them. Furthermore, the Union did not offer any
evidence to support its claim that Atlas’s deadheading policy
had changed. See id. at 252 (Akins Decl. stating he was
“informed” of the deadheading policy shift but not providing
supporting evidence or affidavits). Given the questionable
logic of the desirability theory and the dearth of evidence
supporting the deadheading theory, the district court did not err
in accepting Dr. Lee’s reasoned analysis over Mr. Akins’s
unsupported speculation. 2 Although the district court did not
reject each theory of why an individual pilot might decline
open time, the court adequately made clear that it credited Dr.
Lee’s conclusion that this change did not merely stem from a
series of individual decisions.

Finally, the Union points to the temporal disconnect
between the section 6 notice and the alleged change in
behavior. As we said in our discussion of sick calls, large-scale

2
On appeal, the Union gestures at two other arguments, both of
which it forfeited. The idea that it is counterintuitive for pilots to
decline premium pay to gain bargaining leverage was not adequately
raised before the district court. Haselwander v. McHugh, 774 F.3d
990, 997 (D.C. Cir. 2014). And the Union mentioned only in passing
that Atlas has the authority to decide whether to add a trip to a pilot’s
schedule or designate it as open time and did not explain why that
might be important. Schneider v. Kissinger, 412 F.3d 190, 200 n.1
(D.C. Cir. 2005) (“It is not enough merely to mention a possible
argument in the most skeletal way, leaving the court to do counsel’s
work . . . .”).
 24
behavioral changes do not always happen overnight. Supra,
Part III.A.2. That is especially true when it means forgoing an
extra benefit, like overtime pay, so it was not erroneous for the
district court to conclude that the “precise temporal
correspondence that the Union demands is neither required nor
what one would reasonably expect to occur.” Atlas Air, 280 F.
Supp. 3d at 96. Moreover, although the amount of unfilled open
time was not consistent from month-to-month, statisticians
typically compare figures year-to-year or quarter-to-quarter.
And the fact remains that, from 2015 to 2017, it took more
phone calls to fill each open time slot, and more open time slots
still went unfilled. Based on this, the district court did not
improperly conclude that Atlas had carried its burden of proof.

Atlas presented sufficient evidence demonstrating that the
Union encouraged pilots to act in this manner. Although some
of these communications concerned the practice of accepting
open time out of seniority order in violation of the existing
CBA, see Union Br. 18-19, the vast majority did not. For
example, when a pilot asked during a Crew Call whether
“picking up open time flying harm[s] . . . [the Union’s] ability
to negotiate,” Captain Kirchner explained that “[e]verything
you do when you cut a corner on the CBA, when you go the
extra mile for [Atlas], that just solidifies and helps them in what
they are trying to do to us . . . and it prolongs the negotiation
process.” J.A. 721-22. He reminded pilots, “You are all
watching a huge amount of open time. . . . [U]nfortunately
some of our pilots are falling over backwards to help [Atlas]
out of a jam . . . . Think what you do before you do it because
you might be prolonging the agony” of “service failures,
scheduling failures, etc.” J.A. 735. The Executive Committee
did explain that whether to fly open time is “a personal
decision,” J.A. 736-38, and clarified that the Union could not
“reprimand” pilots who accepted open time flights, see J.A.
713, or “openly advocate [for] not picking up open time
 25
flying,” Pls.’ Ex. 113 at 10, No. 17-cv-1953 (D.D.C. Nov. 30,
2017), Dkt. No. 28-15. But the Union’s preferences were clear:
“[I]f you don’t fly open time, and that’s a personal decision
because you want to be with your family, and you don’t want
to help [Atlas] out, that’s a great personal decision. . . . [I]t’s
just time for people to do what’s right for the group, but do it
on a personal basis.” J.A. 737. In light of this anecdotal and
statistical evidence, the district court did not err in enjoining
the Union from encouraging pilots to decline open time.

B

Two final points warrant discussion. First, Atlas’s
September 2017 lawsuit was timely. RLA claims are subject to
a six-month statute of limitations. Atlas Air, Inc. v. Air Line
Pilots Ass’n, 232 F.3d 218, 226 (D.C. Cir. 2000). If events in
the six months leading up to the lawsuit are part of an ongoing
slowdown campaign or would themselves constitute violations
of the RLA’s status quo requirement, however, the action is not
time-barred. ALPA, 563 F.3d at 269-70 (ongoing slowdown);
see Local Lodge No. 1424 v. NLRB, 362 U.S. 411, 416 (1960)
(standalone violation). The statistical and anecdotal evidence
that we have described sufficiently demonstrates that the Union
was continuously violating the RLA in the six months before
Atlas filed suit.

Finally, the Union claims that the injunction is overbroad
in violation of the First Amendment and NLGA section 109.
The district court enjoined the Union, its “officers, agents,
employees, members employed by [Atlas], and all persons
acting in concert with any of them . . . from authorizing,
encouraging, permitting, calling [for], engaging in, or
continuing any strike, work stoppage, sick-out, concerted
refusal to volunteer for or to accept work assignments
(including, without limitation, open time flights), slowdown
 26
(including, without limitation, the existing SHOP and BOOT
campaigns), or other self-help against [Atlas] relating to the
dispute or disputes arising from the Section 6 notice.” J.A. 156.

Such language is common in these types of injunctions,
and presents no problem here. See, e.g., ALPA, 563 F.3d at 259-
60; Alton & S. Ry. Co. v. Bhd. of Maint. of Way Emps., 883 F.
Supp. 755, 765-66 (D.D.C.), amended on reconsideration, 899
F. Supp. 646 (D.D.C.), and aff’d, 72 F.3d 919 (D.C. Cir. 1995)
(per curiam) (unpublished table decision). The Union protests
that the district court erroneously enjoined lawful, protected
individual and group activity. But the district court was only
permitted to enjoin those “specific act[s] . . . expressly
complained of in the [complaint]” and about which it made
explicit factual findings. 29 U.S.C. § 109. As the complaint,
language of the injunction, and the court’s underlying findings
reflect, this case is about whether the Union violated the RLA
by encouraging pilots to jointly engage in actions designed to
obtain leverage in the negotiations for an amended CBA. See
Atlas Air, 280 F. Supp. 3d at 67. The injunction does not restrict
the rights of individual pilots whose actions are independent of
the Union, nor does it impermissibly prohibit lawful self-help
activities that would not run afoul of the RLA. For example, it
does not prohibit a pilot’s personal decision to refuse an open
time assignment. Rather, the injunction applies only to
“persons acting in concert with” the Union and its officers in
their efforts to obtain leverage in negotiations for an amended
CBA. J.A. 156; id. (enjoining only the “concerted refusal to
volunteer for or accept work assignments”); see Concerted,
NEW OXFORD AMERICAN DICTIONARY 352 (2d ed. 2002)
(“jointly arranged, planned, or carried out; coordinated”); see
also Concerted Action, BLACK’S LAW DICTIONARY (11th ed.
2019) (“An action that has been planned, arranged, and agreed
on by parties acting together to further some scheme or cause,
 27
so that all involved are liable for the actions of one another.”).
That complies with the First Amendment and NLGA.

IV

In sum, the district court had jurisdiction to enter a status
quo injunction in this major dispute, and did not abuse its
discretion in enjoining this conduct. The judgment of the
district court is affirmed.

So ordered.