dissenting.
There is no question that the pilot bears primary responsibility for the tragic accident at issue here. Nevertheless, that is only part of the story. This case raises important questions about the scope of an air traffic controller’s duty of care to a plane flown under visual flight rules *72(“VFR”) after radar contact initiated by the pilot is then lost. The majority accepts the district court’s finding that any possible breach of duty by the controller did not contribute to the crash. I do not agree. If a controller’s duty to separate aircraft from terrain continues after radar contact is lost, we could not conclude that the controller’s conduct had no effect on the accident. I believe that Santiago did have such a duty and that, on this record, he was required to attempt radio communication with the airplane in time to separate it from El Yunque. Because the record shows that, if Santiago had done so, the plane crash might have been prevented, I cannot conclude that Santiago’s breach of duty did not contribute to the accident. I therefore respectfully dissent.
I.
Paragraph 5-5-9 of the Air Traffic Control Manual (“ATOM”) requires air traffic controllers to “separate aircraft from prominent obstructions depicted on the radar scope,” and specifies a minimum separation distance of three miles for aircraft like the one flown by Wojciechowicz.18 The district court held that this provision applies only to flights operating under instrument flight rules (“IFR”) and further concluded that, even if VFR flights were within the scope of ¶ 5-5-9, Santiago did not violate the provision because Wojeiechowicz’s plane was off radar by the time the three-mile limit was reached. The majority assumes for the sake of efficient analysis that ¶ 5-5-9 applies to VFR flights, and it holds that the district court did not commit clear error in finding no breach of duty because, even though radar contact with Wojeiechowicz’s plane was lost, the controller could reasonably rely on the pilot to see and avoid terrain and obstacles.
I do not understand how the majority can reach that conclusion. If ¶ 5-5-9 applies to VFR flights, it can only be construed to impose a legal obligation on controllers to take all reasonable steps to separate aircraft from obstructions without regard to the VFR pilot’s separate duty to prevent accidents. As plaintiffs argue, the very point of applying ¶ 5-5-9 to VFR aircraft is to provide a second layer of protection against accidents like the one that occurred in this case. It is precisely in eases of unforeseeable pilot negligence that ¶ 5-5-9 would play its most critical role for VFR flights. It thus makes no sense to say that the provision imposes a duty whose outer limit is reached when the controller reasonably could rely on the pilot’s own duty of care. Such a conclusion defines the concurrent duty of the controller out of existence.
To resolve this case, therefore, we cannot avoid directly addressing two legal questions: (1) whether ¶ 5-5-9 applies to VFR aircraft, and, if so, (2) the scope of the controller’s duty under that provision. I consider each of those in turn.
II.
Notwithstanding its assumption that ¶ 5-5-9 applies to VFR flights, the majority discusses the nature of that provision at some length and appears to conclude that the classic rules of statutory interpretation do not apply to it. It points out that the *73ATOM is neither a statute nor a regulation, but only an internal FAA guideline that does “not have the force of law.” It effectively treats the provision’s scope as a question of fact by endorsing the district court’s reliance on the government’s expert, Edward Henderson, who testified that ¶ 5-5-9 could not apply once Wojciechowicz’s plane disappeared from the radar screen. Henderson also testified that he had never seen ¶ 5-5-9 used for VFR aircraft, which led the district court to conclude that the provision does not apply to VFR flights.
Paragraph 5-5-9, however, constitutes more than background guidance or advisory criteria whose applicability is left to the controller’s discretion. Although this court has acknowledged the difficulty of distinguishing between law and practicepartieularly where informal documents such as guidelines and handbooks are involved, see Nieves-Villanueva v. Soto-Rivera, 133 F.3d 92, 100-01 (1st Cir.1997) — no such difficulty exists here. A Federal Aviation Regulation (“FAR”), 14 C.F.R. § 65.45(a), states that “[a]n air traffic control tower operator shall perform his duties in accordance with ... the procedures and practices prescribed in air traffic control manuals of the FAA, to provide for the safe, orderly, and expeditious flow of air traffic.”19 Because part 65.45(a) mandates compliance with air traffic control manuals, the manuals themselves are incorporated into the regulation and acquire the force and effect of law. See Ross v. United States, 640 F.2d 511, 517 (5th Cir.1981) (per curiam, affirming on the basis of the district court’s opinion) (stating that provisions of a similar air traffic control manual have the effect of FAA regulations); see also Rodriquez v. United States, 823 F.2d 735, 739 (3d Cir.1987) (stating that FAR’s “have the force and effect of law”); In re N-500L Cases, 691 F.2d 15, 28 (1st Cir.1982) (same).20
Thus, ¶ 5-5-9 has the status of law and, as such, its meaning must be determined in accordance with ordinary principles of statutory construction rather than by means of expert testimony. See United States v. Lachman, 387 F.3d 42, 50-52 (1st Cir.2004) (applying canons of statutory interpretation to federal regulations); Nieves-Villanueva, 133 F.3d at 100 (“[I]t is for the judge, not the lawyers or the witnesses, to inform the jury of the law *74applicable in the case and to decide any purely legal issue.”). I thus begin by examining the language of the provision. See Lachman, 387 F.3d at 50 (stating that, “ ‘if the language of a statute or regulation has a plain and ordinary meaning, courts need look no further and should apply the regulation as it is written’ ” (quoting Textron, Inc. v. Comm’r, 336 F.3d 26, 31 (1st Cir.2003))).
The paragraph as written does not include an exception for VFR aircraft and, in the absence of such an exclusion, there is no basis for concluding that the controller lacks separation responsibility for VFR flights that are receiving radar services. See Wojciechowicz v. United States, 576 F.Supp.2d 241, 252 (D.P.R.2008) (stating that Wojciechowicz was receiving “basic radar service”). Although it is conceivable that VFR flights by definition would fall outside provisions requiring controllers to separate aircraft from obstructions displayed on their radar scopes, the next ¶ of the ATOM belies any such notion. Paragraph 5-5-10 addresses separation distances from “adjacent airspace” in which radar separation is also being used, and it excludes certain VFR aircraft from its dictates. See ATOM, FAA Order 7110.65M § 5-5-10(c) (“The provisions of sub-para[graph]s a and b do not apply to VFR aircraft being provided Class B, Class C or TRSA services.”).21 This attention to VFR flights reinforces the interpretation that such flights are covered by the radar separation provisions unless explicitly excluded. See § 5-5-10(d). “ ‘[Wjhere Congress includes particular language in one section of a statute but omits it in another section of the same Act, it is generally presumed that Congress acts intentionally and purposely in the disparate inclusion or exclusion.’ ” Russello v. United States, 464 U.S. 16, 23, 104 S.Ct. 296, 78 L.Ed.2d 17 (1983) (quoting United States v. Wong Kim Bo, 472 F.2d 720, 722 (5th Cir.1972)) (alteration in original); Trenkler v. United States, 268 F.3d 16, 23 (1st Cir.2001).22 That principle has no less force in this context.
Construing ¶ 5-5-9 to apply to VFR flights does not conflict with the VFR pilot’s responsibility to see and avoid obstructions. Controllers and pilots have a concurrent duty to maintain safety, see, e.g., Delta Air Lines, 561 F.2d at 392 (“[Ajlthough pilots are undoubtedly responsible for the safety of their aircraft, controllers are concurrently responsible for adhering to procedures which minimize the difficulties for the crew.”), and the FAA has acknowledged that rules relating to pilots and controllers may overlap. Section 5-5-l(e) of the AIM states:
The responsibilities of the pilot and controller intentionally overlap in many areas providing a degree of redundancy. Should one or the other fail in any manner, this overlapping responsibility is expected to compensate, in many cases, for failures that may affect safety.
*75Imposing the duty to separate under ¶ 5-5-9 on controllers for all flights receiving basic radar services is one obvious method to protect against unexpected “failures that may affect safety.” Hence, I can only conclude that the plain, unlimited language of ¶ 5-5-9 governs and that the provision applies to VFR as well as IFR flights. I therefore must consider the scope of the duty imposed by the provision and whether it was breached in the circumstances of this case.
III.
The majority explains why the district court could properly find that Santiago did not act unreasonably in failing to contact Wojeiechowicz before the plane disappeared from radar, when it was still approximately 4.7 miles from El Yunque. Paragraph 5-5-9 prescribes a mandatory minimum of three miles, and the record shows that Wojeiechowicz still had plenty of time at the 4.7-mile point to change course and avoid crashing into the mountain. The majority errs, however, in deferring to the district court’s conclusion that Santiago’s duty ended when radar contact was lost. Nothing in ¶ 5-5-9 suggests that the duty to separate aircraft from obstructions automatically dissolves when a plane receiving radar services suddenly disappears from the screen. Indeed, no one would suggest that the loss of radar contact with a flight operating under instrument rules terminates the controller’s duty to provide separation services.
The majority holds that, in the absence of “mandatory language stating when a controller must take action to attempt to maintain a separation,” the district court permissibly treated the question of duty “as one of what a reasonable controller would have done on these particular facts.” I have no problem with the majority’s statement of the general principle.23 The majority goes on to conclude, however, that the district court committed no clear error in finding that it was reasonable for the controller to assume that the VFR pilot knew where he was and could see the mountain, and, on that basis, acted reasonably in failing to immediately initiate contact with the pilot when the plane disappeared from radar.
As I have explained, ¶ 5-5-9 has no purpose in the VFR context unless it is to serve as backup protection against the possibility of pilot negligence. In imposing a concurrent duty on controllers to separate VFR flights from obstructions, ¶ 5-5-9 effectively incorporates an assumption that pilot negligence is foreseeable. The fact that Wojeiechowicz, as a VFR pilot, should have been aware of his surroundings is thus besides the point, and the controller’s duty under ¶ 5-5-9 is no different for VFR flights than for IFR flights. If a plane is no longer visible on the radar screen, the only way for the controller to be sure there is proper separation between that aircraft and prominent obstructions is to contact the pilot directly. From that perspective, it is apparent that Santiago’s duty under ¶ 5-5-9 could not automatically end with the loss of radar contact. Rather, as with an IFR flight, the duty to maintain separation required him to make *76reasonable efforts to contact Wojciechowicz to ensure that the aircraft remained three miles away from El Yunque. It was foreseeable that, in the event of pilot negligence, Santiago’s failure to take that step could be disastrous.
Santiago’s own actions and testimony indicate that he was aware of his responsibility to attempt to contact Wojciechowicz. Although Santiago testified that short-term loss of radar contact with an aircraft is not unusual, he acknowledged that if an aircraft disappears from his scope, he would, “most frankly,” inform that aircraft that radar contact was lost. He explained that he does not necessarily inform a pilot that radar contact is lost when the plane’s data is in “coast” mode- — the short period immediately after the loss of radar contact when the computer predicts the flight’s likely path — because planes sometimes will “go into coast ... and then they’ll come back.” The flight transcript shows that Santiago attempted to reach Wojcieehowicz’s plane by radio more than a dozen times, starting at 43 seconds after the plane’s coast data disappeared from the radar scope (roughly the time of the crash).24
The district court’s view that Santiago’s duty under ¶ 5-5-9 ended when the plane was no longer visible on radar is thus unsupportable. That legal error infected the district court’s fact-finding and undermines the majority’s reliance on it. Examining the record in light of the proper scope of ¶ 5-5-9, the remaining question is whether Santiago breached his duty to maintain separation by failing to contact the pilot for 43 seconds after all radar information about the flight ended. I now turn to that question.
IV.
In one sense, it is undisputed that Santiago failed to perform his duty to ensure three miles of separation between the plane and prominent obstructions. The aircraft was permitted to penetrate the three-mile “ring” around El Yunque and crashed 1.43 miles from the mountain’s peak. Wojciechowicz, 576 F.Supp.2d at 246. The majority is correct, however, that simple breach of the duty does not amount to negligence, and the issue before us turns on “what a reasonable controller would have done on these particular facts” to fulfill his duty under ¶ 5-5-9. Santiago did attempt to contact the pilot. Hence, the specific question to be addressed is whether his delay in doing so was reasonable.
The record shows that a reasonable fact-finder would have to conclude that it was not. Santiago knew or should have known, based on the information displayed on his scope, that at the moment actual radar contact was lost, the airplane was only 1.7 miles away from the three-mile ring — the minimum distance for separation specified in ¶ 5-5-9. See Wojciechowicz, 576 F.Supp.2d at 259. He also knew that the plane was headed toward the mountain. Even if a three-mile separation was no longer feasible, Santiago at least needed to take all reasonable steps to contact Wojciechowicz early enough to guide the plane away from the mountain in case the pilot inexplicably failed to change direction on his own.
The government’s expert testified that Wojciechowicz could have maneuvered the plane away from the accident site within 17 or 18 seconds. The last coast data appeared on radar at about 45 seconds before the crash. Id. Accepting the end of *77all data as a reasonable trigger for the controller’s concern, and even allowing for a brief delay as Santiago came to the realization that the plane had not reappeared on the radar scope, Santiago still had more than thirty seconds to attempt to turn the aircraft before the impact occurred — nearly double the time needed. He waited, however, until virtually the time of impact to initiate radio contact. On this record, that delay can only be found unreasonable.25 I realize that thirty seconds is not much time. In the business of air traffic control, however, split seconds matter, and the stakes are enormous. If Santiago had fulfilled his duty under ¶ 5-5-9 of the ATOM- — to “separate [the] aircraft from [the] prominent obstruction[ ] depicted on the radar scope” — this accident may have been avoided.26
The majority observes in a footnote that there was no evidence presented that Santiago could have reached the plane by radio if he had tried earlier. To the contrary, Santiago and Wojciechowicz had been communicating by radio, showing that the radio was functioning properly. There is no basis for concluding that the radio would have suddenly failed before the plane crashed.
Nor is there any reason to think that Santiago could not have guided the change of course within the 17 or 18 seconds the expert estimated would ordinarily be necessary. The record shows that, despite the lack of radar contact, Santiago had some awareness of the airplane’s location. After the crash, he directed numerous planes and police helicopters directly to the crash site, at one point stating that “once you get close to the area I’ll give you a point out.” Therefore, regardless of the pilot’s own disorientation, Santiago had sufficient knowledge of the plane’s whereabouts to accomplish the required separation.
Y.
The tragic result in this case might have been avoided if the government had per*78formed its duty under ¶ 5-5-9 to maintain a three-mile separation between Wojciechowicz’s plane and El Yunque. The plain language of the provision extends the duty to separate aircraft from obstructions to IFR and YFR flights alike, and that duty does not terminate when radar contact is lost unexpectedly. Indeed, even under VFR conditions, the unexplained loss of all radar contact when a plane is heading directly toward an obstruction and nearing the three-mile safety barrier should trigger the controller’s immediate concern. To fulfill his duty under ¶ 5-5-9, the controller had to act quickly to contact Wojciechowicz so that, if necessary, he could guide the pilot in separating the plane from El Yunque. In the circumstances here, Santiago’s 43-second delay before attempting such contact was unreasonable.
In concluding that the district court committed no clear error in evaluating the facts, the majority misconstrues the law and the principle of concurrent duty. The government has undertaken to act as a safeguard to potential pilot negligence, and when it performs that function negligently, its actions cannot be insulated from review because of the pilot’s errors. Sadly, it takes only a few seconds of inattention on the part of an air traffic controller for tragedy to strike. I would reverse and remand with instructions that the district court apportion responsibility for the tragedy between the pilot and the controller under Puerto Rico’s tort law.
. Paragraph 5-5-9, which is titled "Separation from Obstructions,” states in relevant part:
[SJeparate aircraft from prominent obstructions depicted on the radar scope ... by the following minima: 1. When less than 40 miles from the antenna — 3 miles.
ATCM ¶ 5-5-9.
. Although the regulation refers to "control tower operators], ” courts have included air traffic controllers within its scope. See, e.g., Campos Viuda de Courtois v. United States, 778 F.Supp. 585, 591 (D.P.R.1991); In re N-500L Cases, 517 F.Supp. 825, 834 (D.P.R.1981), aff'd, 691 F.2d 15 (1st Cir.1982).
. Contrary to the majority's assertion, evaluating the ATCM as a provision of law is not inconsistent with First Circuit precedent. The case cited by the majority, Federal Express Corp. v. Rhode Island, 664 F.2d 830, 835 (1st Cir.1981), holds that violations of the ATCM "do not necessarily constitute negligence" — a point with which I agree — but it does not hold that manual provisions do not have the force of law. Indeed, Federal Express Corp. equates mandatory ATCM provisions with statutes and ordinances. See id. ("Although slight deviations from manual procedures do not necessarily constitute negligence, we have previously indicated that 'a substantial and unjustified failure to follow procedures made mandatory by the Manual is persuasive as an indication of a lack of due care.’ The Rhode Island courts have embraced a similar position, holding that in an action based on violation of a statute or ordinance, 'evidence of injury proximately caused by the violation is prima facie evidence of defendant's liability ....’") (citations omitted).
This is not to say that FAA documents that contain only advisory criteria, and not binding rules, must also be given the force and effect of law. Ross, 640 F.2d at 518 (noting that another FAA manual consisted of "advisory criteria rather than binding FAA regulations").
. Like ¶ 5-5-9, ¶ 5-5-10 specifies the separation distance that controllers must maintain depending on the distance between the aircraft and the radar antenna, or sensor, that is being used by the controller. After explicitly excluding certain categories of VFR flights from its coverage, the provision continues by directing a "radar handoff” or termination of radar service for VFR flights approaching the excluded types of airspace under the control of another air traffic control facility.
. The majority notes that the FAA-published Aeronautical Information Manual (“AIM”) expressly states that separation services are not provided to VFR aircraft in Class E airspace. The district court found that Wojciechowicz's plane entered Class G airspace one minute before the collision and flew in and out of Class G from that time until the crash, including "at least the 10 seconds prior to colliding with the terrain.” Wojciechowicz, 576 F.Supp.2d at 251.
. The majority asserts that I have improperly treated the question of the reasonableness of the controller's actions under ¶ 5-5-9 as one of law rather than fact. This misrepresents my analysis, which is consistent with our decision in Federal Express Corp., 664 F.2d at 835. We held there that “[t]he existence and extent of a duty of care are questions of law; whether any such duty has been breached and whether proximate cause exists are questions for the factfinder, whose determination is binding on appeal unless clearly erroneous.” Id. My analysis considers ”[t]he existence and extent” of the controller’s duty under ¶ 5-5-9 as a question of law and the issue of breach as a factual question.
. The district court found that the plane crashed about two seconds later, "approximately 45 seconds” after the last coast data appeared. Wojciechowicz, 576 F.Supp.2d at 259.
. The majority states that reliance on a duty to separate within thirty seconds after the coast data disappeared introduces a new theory to the case. That is not so. Plaintiffs make the following argument in their brief:
[T]he district court’s finding that "[t]he last 'coast' data block on the Aircraft was seen at 2:21:42 p.m., approximately 45 seconds prior to the crash,” also undercuts the court’s conclusion that it would have been impossible for Santiago to have complied with § 5-5-9.... Based upon any reasonable view of the evidence, Controller Santiago had sufficient information and time to issue a separation instruction to the Aircraft, as required by § 5-5-9. This analysis is also consistent with the testimony of the government’s piloting expert who testified that Mr. Wojciechowicz, the pilot, could have climbed the Aircraft out of its predicament at a point as late as 17 or 18 seconds prior to impact.
Although the plaintiffs' timing discussion focused primarily on the contention that Santiago needed to contact the plane before it reached the three-mile minimum distance from obstructions, the underlying theory is the same whether the focus is the three-mile minimum or separation from the mountain itself — i.e., that Santiago had a duty to act before it was too late.
. I do not suggest that the government bore full, or even substantial, responsibility for the crash. Under Puerto Rico law, where the relative fault of the plaintiff and defendant is taken into account, a factfinder properly could have concluded that the government was responsible for only a small percentage of the damages. See, e.g., Campos Viuda de Courtois, 778 F.Supp. at 590 C"[I]f both the plaintiff and defendant are at fault, the plaintiff can still recover, but his recovery is limited to the proportion of damages sustained by the plaintiff that were proximately caused by the defendant's negligence.”); P.R. Laws Ann. tit. 31, § 5141 ("Concurrent imprudence of the party aggrieved does not exempt from liability, but entails a reduction of the indemnity.”).