On July 31, 1973, a Delta Air Line DC-9 aircraft crashed during an approach in fog to Logan International Airport in Boston, Massachusetts. Eighty-three passengers, the crew of five and a Delta cockpit observer died in the accident.

The various lawsuits arising out of the accident were transferred for pretrial proceedings in the District of Massachusetts. See, e. g., In re Delta Airlines Crash at Boston, Massachusetts, on July 31, 1973, 395 F. Supp. 1405 (Jud.Pan.Mult.Lit.1975); In re Delta Airlines Crash at Boston, Massachusetts, on July 31, 1973, 373 F. Supp. 1406 (Jud.Pan.Mult.Lit.1974). All claims were consolidated for a non-jury trial on the issue of liability.

Delta did not contest its own liability to the estates of deceased passengers but sought contribution and indemnification from the United States under the Federal Tort Claims Act, 28 U.S.C. § 1346(b), based on the alleged negligence of an air traffic controller who handled the flight. The airline also sought recovery for the loss of the aircraft and other consequential damages. Individual suits against the United States were brought as well by the families of crewmembers barred under workmen's compensation statutes from bringing suit against Delta.

Delta conceded at the eleven-day trial that crew negligence contributed to the accident, but argued that negligence on the part of air traffic control was also a contributing factor. The individual plaintiffs sought to prove that the controller was negligent and that this negligence was a proximate cause of the crash. The Government took the position that the crew's negligence was the sole cause of the accident.

On February 19, 1976, the court filed a lengthy opinion which concluded "that the sole and exclusive cause of the accident was the negligence of the pilot and co-pilot of D 723". In re Aircrash Disaster, 412 F. Supp. 959, 998 (D.Mass.1976). Judgment for the United States was entered and notice of appeal was subsequently filed by Delta and by the individual plaintiffs. We affirm.

* On July 31, 1973, at approximately eight minutes after eleven in the morning (1508:05 Greenwich Mean Time, hereinafter GMT), Delta flight 723 (D 723), a regularly scheduled passenger flight from Burlington, Vermont to Boston, Massachusetts, crashed into a seawall at the perimeter of Logan International Airport while making an instrument landing system (ILS) approach in fog to runway four right (4R).

D 723 had made an unscheduled stop at Manchester, New Hampshire to pick up some passengers whose flight had been cancelled due to the weather. At 10:14 a. m. (1414 GMT), D 723 was informed that its departure from Manchester would be delayed because of traffic delays at Logan caused by adverse weather conditions. The flight eventually left Manchester at approximately ten minutes before eleven (1450 GMT) for the 18 minute flight to Logan.

Shortly after takeoff, D 723 contacted Boston Approach Control Arrival Radar # 1 (Approach) which was being manned by Mr. Taylor. Approach told the flight to plan on receiving radar vectors for an instrument landing approach to runway 4R. A radar vector is "a (magnetic compass) heading issued to an aircraft to provide navigational guidance by radar." See 412 F. Supp. at 965 n. 6. Approach also gave D 723 its only en route weather advisory: "Weather is partial obscuration, estimated four hundred overcast, mile an' a half and fog." The flight climbed to 4,000 feet and at 10:52 a. m. (1452 GMT), D 723 was identified by Approach on the radar scope. Four minutes after the plane left Manchester, Approach instructed it to "fly heading now one eight zero (180o ), radar vectors ILS four right." This meant that the plane was to fly south and would be given further headings which would feed it into the instrument landing system for runway 4R at Logan.

Because of the low ceiling and limited visibility at Logan Airport, D 723, and the other aircraft flying into Logan on July 31, were required to make an instrument landing system approach. The two major components of the instrument landing system are the localizer and the glide slope, see 412 F. Supp. at 965 nn. 7 & 9. The localizer is an electronic beam which, in combination with radio equipment in the plane, allows a pilot to align his aircraft so that it is headed straight for the centerline of the intended runway for landing. The localizer course for runway 4R is 035o . If the plane deviates to the right or the left of this course, the aircraft's instrumentation tells the pilot which way to turn to reintercept the localizer, i. e., to get back on course. The glide slope, also an electronic beam which operates in combination with a receiver in the plane, aids the pilot in making the descent. A plane which stays on the glide slope will eventually touch down at the intended point on the runway, the ILS touchdown zone. FAA regulations and Delta procedures require that when, during the descent, an altitude known as the decision height (DH) is reached 216 feet above sea level for runway 4R at Logan the Captain must determine if the landing environment is in sight and whether a safe landing can be made.1 If not, he must execute a missed approach by applying power and climbing back up to altitude. Otherwise, the landing approach continues, although the Captain is free to discontinue the approach should he deem it necessary. There was testimony at trial that a missed approach may be executed in a DC-9 at any time up to and including touchdown.

All flights approaching Boston on the morning of July 31, 1973, were under the control of air traffic controllers who handle the air traffic from a darkened room where each airplane is tracked on a radar scope. See 412 F. Supp. at 972 n. 18. The function of approach control is to promote the "safe, orderly and expeditious" flow of traffic arriving and departing from a terminal area. See id. at 978 n. 22 (definition of approach control service). Approach controllers operating in these terminal areas, including Boston, do so according to procedures set out in the Federal Aviation Administration's handbook, Terminal Air Traffic Control (the Manual). The Manual provides that action verbs in the imperative mean that a procedure is mandatory. The Manual also prescribes duty priorities, the first being to separate aircraft, the second to tend to "second priority services" that do not involve air traffic separation and the third to give "additional services to the extent possible". See 412 F. Supp. at 964 n. 6 quoting Manual P 28. Included in the last category is the dissemination of weather information. See Manual P 361.

Paragraph 1360 of the Manual in use on July 31, 1973, required an approach control to issue certain information and instructions to an aircraft before it reached the so-called approach gate. This would constitute a second priority duty. The approach gate for runway 4R at Logan is a point 6.3 miles from the runway landing threshold and 1 mile outside the outer marker, the final approach fix. Id at 965 n. 8. The following information should have been given: 1) D 723's position relative to the outer marker (the final approach fix); 2) a vector to intercept the final approach course; 3) a clearance for the aircraft to make the instrument approach; and 4) an instruction to go ahead and monitor Boston Local Control (the Tower) then contact it upon reaching the final approach fix or, alternatively, an instruction to switch over to the Tower immediately. The handbook gives this example: "Three miles from outer marker, turn left heading zero one zero, cleared for I-L-S runway three six approach. Monitor one one eight point three; report to tower when over the outer marker." Furthermore, P 1351 of the Manual states as a guideline that the maximum localizer interception angle to be given when a plane is 2 miles or more outside the approach gate is 30o . Paragraph 1352 requires, by its use of verbs in the imperative, that the controller "vector aircraft to intercept the localizer course at least 2 miles from the approach gate and at an altitude not above the glide slope." (Emphasis original) As will be seen, D 723 was never given its position relative to the approach fix and was not turned over to the Tower until well after passing the approach gate. Further, it was given an interception angle larger than 30o and was not instructed to descend to an altitude "not above the glide slope."

With this background, we return to D 723. About 6 minutes after it took off from Manchester, the flight, which had leveled off at 4,000 feet, was told to descend to 3,000 feet. Approach never assigned a lower altitude. Between 10:57 a. m. and 11:04 a. m., Approach issued various compass heading changes to D 723 which brought it west of Boston and around to the south of Logan Airport in preparation for the instrument approach to runway 4R. Just past 11:04 a. m. (1504:27 GMT), less than 4 minutes before the crash, Approach said: ". . . and Delta seven two three fly heading of zero eight zero (080o ) and intercept the localizer course and fly it inbound over." D 723 replied, "Okay zero eight zero for interception." The transmission from Approach constituted "a vector to intercept the final approach course" and was given more than 2 miles and approximately 1 1/2 minutes prior to D 723's reaching the approach gate. Therefore, the second of the required four pre-approach instructions was given before the flight reached the approach gate. The angle of intercept, however, was 15o greater than that suggested by the Manual. The intercept heading of 080o necessitated a heading change, or intercept angle, of 45o as the airplane made its left turn on to the localizer course of 035o . Of more significance, however, no clearance to descend to an altitude "not above glide slope" was given; this was contrary to mandatory Manual procedures. Thus, the flight remained at 3,000 feet at a point when planes making the approach would, according to testimony, customarily have been descended to 2,000 feet.

Some thirty seconds prior to arriving at the approach gate, D 723 asked if it were "cleared for ILS". Approach responded, "Yes, seven two three is cleared for ILS, yes." The flight profile indicates that this transmission from Approach occurred before D 723 reached the approach gate. Thus, the third of the four mandatory instructions was timely given to D 723, albeit at the instigation of the aircraft. However, at no time was the flight given the first mandatory instruction, its position relative to the final approach fix; furthermore, at the point that D 723 had passed the approach gate and reached the final approach fix, it had not received the fourth mandatory instruction, to monitor the Tower frequency or, alternatively, to report to the Tower.

The plaintiffs claim that the fact D 723 was not turned over to the Tower was critical because a significant weather change had taken place at Logan subsequent to the weather report relayed to the aircraft after its take off from Manchester. At almost 6 minutes past 11 (1505:42 GMT), the Tower informed Eastern 572, the plane immediately preceding D 723 on the approach to 4R, ". . . the fog's movin' back in from the south across the airport now it's just approaching runway four right". D 723 was less than a mile from the approach gate, had not been told to monitor the Tower and did not hear this transmission. Had D 723 contacted the Tower at the outer marker, testimony suggests that it would have received the fog warning. This would have occurred shortly after 11:06:14 a. m., almost 2 minutes prior to the accident. At 7 minutes past 11 (1507:03 GMT), about 1 minute before the crash when D 723 was well inside the outer marker, but before it had been switched to Tower, there was an internal communication from the Tower to the radar room that the fog was "coming right back across again so you can get ready for some (missed approaches), it's real thick again." Mr. Taylor, the approach controller handling D 723, testified that he did not hear anything about the fog bank.

Meanwhile, D 723's approach continued with the co-pilot flying the aircraft and the Captain handling the radio communications and other chores. D 723 crossed the outer marker 1 minute and 51 seconds before impact at a speed of 207 knots which was, according to the testimony, over 40 knots too fast. The flight profile shows, and various witnesses testified, that D 723 did not execute a good approach; it was off course to the left of the localizer until the final moments and was never established on the glide slope. See id. at 999. Nevertheless, there was testimony by Delta's expert that the approach would have been acceptable had the ceiling been at 400 feet as initially reported to D 723. Plaintiffs argued that the excessive intercept angle and the fact that D 723 was forced to begin its approach 1,000 feet too high contributed to the deviations. While there was testimony tending to confirm this view, there was other testimony, and the district court found, that the deviations were caused by the crew's negligent reliance on the flight director, a sophisticated instrument that had been set in the wrong mode.

Fifty-one seconds before the accident and only at the request of the local control coordinator whose job it was to coordinate approaches, Approach cleared D 723 to land and told it, finally, to go to the Tower frequency. This clearance was received 2 minutes and some 4 miles after the plane had crossed the approach gate and at a point in the approach where the cockpit voice recorder reveals the crew was concerned with problems in flying the plane.2 The aircraft was 3 1/2 to 4 miles from the touchdown zone and was at 700 feet having just descended below the glide slope on its approach.

D 723 did not contact the Tower immediately. In the seconds after the transmission from Approach, the Captain made a remark which indicated that the plane was going too fast, see 412 F. Supp. at 982; 25 seconds before impact, apparently perceiving a malfunction in the flight director, the Captain ordered a switch to "raw data" or primary instrumentation. Only then, 22 seconds before the crash, did D 723 contact the Tower which responded, 2 seconds later as the plane was descending through 325 feet, "Cleared to land four right, traffic's clearin' at the end, the RVR shows more than six thousand, a fog bank is movin' in, it's pretty heavy across the approach end." D 723 acknowledged the Tower's clearance and weather advisory 13 seconds before the accident by saying simply, "seven two three". This was the last transmission from the aircraft and it was made fractions of a second before the plane descended through decision height. Insofar as the cockpit recorder shows, the crew did not make mandatory altitude call-outs nor did anyone refer to altitude throughout the approach.3 Discussion in the cockpit following receipt of the fog warning indicates no vocalized reaction whatever to the reported fog bank; instead, both the co-pilot who was flying the plane and the pilot voiced concern only with getting the airplane on course. Despite the focussed efforts, the aircraft was never successfully established on the localizer. Moreover, inexplicably, it descended in the last 11 seconds at a rate of descent of 1200 feet per minute, almost twice the normal rate of 650-700 feet per minute which D 723 had, during the latter phases of the approach, maintained. D 723 crashed 4153 feet short of the touchdown zone on runway 4R and about 165 feet to the right of the runway centerline. See 412 F. Supp. at 963.

Three planes which immediately followed D 723 on the ILS approach were successively cleared to land. (The Tower was unable to see the crash site because of fog and for some reason did not realize that D 723 was not accounted for.) Two of them executed missed approaches because they were unable to see the runway environment due to fog. The third was told by the Tower to make a missed approach, apparently as a result of first reports of a plane down on runway 4R. Upon confirmation of the crash, the airport was closed.

In order to complete the factual picture as it pertains to the issues in this case, it is necessary to backtrack in time a bit and focus on Approach's difficulties with another aircraft during D 723's approach.

Just past 11:02 a. m., while D 723 was in the process of being vectored by Approach and was still almost 3 minutes away from the approach gate, Allegheny 666 (AL 666) reported in the Millis holding pattern 20 miles southwest of Boston at 8,000 feet. AL 666, however, had been assigned 9,000 feet and another aircraft, Eastern 1020 (EA 1020), was already at 8,000 feet in the holding pattern. Approach reacted to this potential conflict immediately by directing EA 1020 to make a left turn to the northeast that would take it toward Logan and out of the holding pattern and by instructing AL 666 to maintain its southwesterly heading which would take it out of the holding pattern in a direction away from Logan. Approach then gave D 723 a vector of 150 o and instructed another aircraft at Millis to descend out of 7,000 feet to 6,000 feet. Next Approach verified EA 1020's left turn and, satisfied that it was now safe to do so directed AL 666 to reverse its course, return to Millis and hold.

At 11:04 a. m. (1504:27), D 723 was given its final vector for intercepting the localizer, but no position report, no clearance for the approach and no instruction to monitor the Tower. Fifty-two seconds later, Approach attempted to verify AL 666's altitude. AL 666 did not reply and Approach repeated its radio call to the flight four more times. It was after the fourth of these five attempts to raise AL 666 that D 723 asked if it were cleared for the approach. Approach replied simply "yes, seven two three cleared ILS, yes" then made a fifth call to AL 666, this time receiving a response culminating in a confirmation at 11:06 a. m., that AL 666 was still at 8,000 feet. Eleven seconds later, D 723 crossed the outer marker inbound. Approach continued to handle the numerous other traffic in its jurisdiction including setting up EA 1020 and another aircraft for their approaches to runway 4R. There were, however, no communications to D 723 between the time it was cleared for the ILS approach at 11:05:41 a. m., when the plane was roughly 1 mile outside the approach gate and almost 2 miles from the outer marker, and 11:07:14 a. m., when the aircraft, then 51 seconds and slightly less than 2 miles from the site of impact, was, at the request of the local control coordinator, cleared by Approach to land and to switch to the Tower.

Delta premised its theory of Government liability on the contention that had Manual procedures been followed, D 723 would have been aware that the fog bank was in the vicinity of runway 4R prior to beginning its approach and would not have encountered the heavy fog unexpectedly in the last critical stages of flight, become disoriented, lost control of the aircraft and crashed. An essential part of this theory is that at some point the crew had ground contact then lost it in the fog. Delta further contended that the cumulative effects of the nonstandard service led to an unstabilized condition which made it impossible for the crew to react quickly enough when it flew into the fog bank. The individual plaintiffs argued that failure of Approach to turn D 723 over to the Tower prevented D 723 from receiving timely warning of the thickening fog and also produced a number of last minute controller-cockpit communications which seriously distracted the crew from attention to essential landing procedures. The Government countered that the controller was, during the entire approach of D 723, properly concentrating his attention on maintaining separation of another aircraft, AL 666, and that, in any event, the safe completion of D 723's flight was entirely in the hands of the crew.

The district court found that the controller's attention was

"appropriately focussed on concentrated communication efforts with (Allegheny 666) and other aircraft, coordination efforts with other controllers concerning intended routes for other approaching aircraft, the sequence of aircraft being vectored to their approach courses, and the avoidance of potential conflicts with departing aircraft."

412 F.Supp. at 991. In particular the court found that concern over Allegheny 666 caused the controller to delay ordering Delta to switch to the Tower frequency. Id. at 900.

Although Delta had conceded its negligence from the onset of the trial, the court also made specific findings in support of its conclusion that the crew had been grossly negligent. There was negligence, the court found, in the crew's reliance on the flight director's faulty indications, its failure to control the aircraft's speed, or to stabilize on the localizer and glide slope, its permitting the aircraft to dive in the last moments of the flight and its failure to make a decision at decision height regarding whether to continue the approach. Id. at 966. The court ruled ultimately that the plaintiffs had not proven "by a preponderance (of the evidence) either that ATC (air traffic control) personnel were negligent or that the conduct of ATC, however, characterized, was a proximate cause of the crash either in whole or in part." Id. at 988.

Once the Government undertakes to provide services otherwise not required of it, those services must be performed in the exercise of due care and the Government will be liable for injuries shown to have been proximately caused by the lack of such care. Ingham v. Eastern Air Lines, Inc., 373 F.2d 227, 236 (2d Cir.), cert. denied, 389 U.S. 931, 88 S. Ct. 295, 19 L. Ed. 2d 292 (1967). See Indian Towing Co. v. United States, 350 U.S. 61, 76 S. Ct. 122, 100 L. Ed. 48 (1955); Yates v. United States, 497 F.2d 878, 884 (10th Cir. 1974).4 In determining whether or not the Government should be held concurrently liable for the crash of D 723, the appropriate inquiry is whether in the circumstances the approach controller was under a duty to provide the flight with those services which were omitted, and, if so, whether his failure to give the information and instructions had a "reasonably close causal connection" with the accident. See W. Prosser, Law of Torts § 30 at 143 (4th ed. 1971). For the reasons set forth below we believe that the approach controller, Mr. Taylor, did have a duty to both the crew and the passengers to comply more fully than he did with the mandatory approach procedures outlined in the Manual; and we do not accept the district court's finding that concern about AL 666 was a satisfactory excuse for what was tantamount, on occasion, to forgetting D 723. The district court's finding, however, that the controller's omissions did not proximately cause the accident is not clearly erroneous and we therefore affirm.

Other courts have said that, as a basic premise, the Government's duty of care in the maintenance and promotion of a safe and efficient air traffic control system is defined in part by the provisions in the procedure manuals. See Spaulding v. United States, 455 F.2d 222, 226 (9th Cir. 1972); Gill v. United States, 429 F.2d 1072, 1075 (5th Cir. 1970). See generally Ingham v. Eastern Air Lines, Inc., supra, 373 F.2d at 233-36.5 This duty extends to both flight crews and their passengers. See, e. g., In Re Air Crash at New Orleans (Moisant Field), 544 F.2d 270, 273 (6th Cir. 1976). While failure to conform to every mandatory Manual procedure, however trivial the deviation, would not necessarily constitute negligence, and while it might not be negligent to deviate from established procedures in the face of a higher priority concern, nonetheless a substantial and unjustified failure to follow procedures made mandatory by the Manual is persuasive as an indication of a lack of due care.

Appellants complain that Mr. Taylor failed in four respects to give standard service to D 723: he gave a final vector to the localizer which required too great an intercept angle; he omitted the position report; he kept the aircraft at 3,000 feet when it should have been at 2,000 feet to intercept the glide slope from below; and he failed to turn the flight over to local tower control at or before the aircraft reached the outer marker. The first of these was contrary to recommended but not mandatory procedures. The other three lapses represented nonstandard service in violation of specific mandatory Manual provisions. We discuss each of the four factors in turn.

The Manual prescribes, as a guideline, that the maximum localizer intercept angle for a plane two miles or more outside the approach gate be 30o . Manual at P 1351. Within two miles, the angle is reduced to 20o . Id. D 723 intercepted the localizer approximately two miles from the approach gate. Delta's expert testified that the greater the intercept angle the more difficult it is to get stabilized quickly on the localizer course. The farther away from the outer marker the turn is made, however, the less significant the amount of turning the aircraft must do becomes, simply because there is more time to stabilize the heading before the final approach begins at the outer marker. Thus, the reason for the suggested maximum heading change is not related to the pilot's ability successfully to execute the turn required by a larger intercept angle, but rather to the need to minimize the maneuvering required to get set up for the final approach. A very large intercept angle quite close to the outer marker will increase the crew workload and diminish, to some degree, the chances that the aircraft will be stabilized on the localizer heading as early in the final approach phase as desirable.

The district court found that Approach was not negligent in giving to D 723 a 45o intercept, 15o greater than that recommended. The finding rested in part on the fact that the 45o intercept, as well as greater ones, had been used successfully with other aircraft that day. 412 F. Supp. at 988.6 There was also evidence that intercept angles in excess of 30o were not unusual. This, combined with the fact that the angles suggested in the Manual are guides and not absolute values, and that pilots are free to refuse approaches which they find difficult to handle, leads us to accept the district court's finding that the vector to the final approach course which required a turn of more than 30o was not, by itself, indicative of a lack of due care.

The three deviations from mandatory procedures seem to us more serious; they cannot be explained as acceptable customary procedures. While experts testified that on occasion individual services are not provided in compliance with the Manual, this is the exception and not the rule. Mr. Taylor himself testified that he normally gave arriving aircraft the services required by the Manual and none of the witnesses could recall an instance where all of the services not given to D 723 were similarly withheld from another flight.

We start with Approach's failure to give D 723 its position relative to the outer marker, information which must be given to the crew before it is 3 miles from that point. Delta's expert testified that the aircraft's distance from the outer marker cannot be determined from aircraft instrumentation, and that the knowledge assists the crew in planning speed reduction in preparation for the approach. While it is no doubt true that the crew knew or should have known by a mental time-distance-speed calculation its general position south of Boston, the radar controller was better able to know more exactly the aircraft's position. D 723 had not been responsible for its own navigating but had been following headings given to it by Approach. In light of the mandatory Manual procedure and the fact that the information is customarily given to all incoming flights, the crew and, more especially, the passengers, were entitled to rely on Approach to give the position report.

A further troublesome factor in the handling of D 723 is that the flight was kept at 3,000 feet, and, as a result, was required to intercept the glide slope from above. While all agreed that occasionally pilots do this in making instrument approaches, there was also expert testimony that pilots are trained to intercept the glide slope from below.7 Further, the sophisticated instrumentation in airliners is designed with such an intercept in mind. Having to intercept from above greatly increases crew responsibilities and workload. It is, for example, difficult for a jet aircraft to descend rapidly and still keep the speed down to that desired during the approach. It is the purpose of the procedure to achieve the minimum crew workload by making this very exacting phase of flight as uncomplicated and routine as possible. That pilots should be capable of handling the additional problems caused by the intercept does not relieve the controller of his duty to the crew and passengers to adhere to procedures established to make the pilots' tasks simpler and to enhance safety.

Finally, Mr. Taylor did not turn D 723 over to the Tower at or before the outer marker as he was required to do by P 1360d of the Manual.

"The principal purpose of this paragraph is to ensure that frequency changes are made prior to passing the final approach fix. However, at times it will be desirable to retain an aircraft on the approach control frequency to provide a single-frequency approach or other radar services. When this occurs, it will be necessary to relay tower clearances or instructions to preclude changing frequencies prior to landing or approach termination."

Manual P 1360 d. Note. Mr. Taylor had no reason to keep D 723 on the Approach frequency. Once he discovered the flight was still with him, he turned it over to the Tower instead of undertaking to relay all communications himself. As a result, the crew was required to switch frequencies and to engage in last minute communications with the Tower at a time when concentration should have been focussed on the instrument approach and, more particularly, on the mandatory altitude callouts as the aircraft reached decision height.8 This is the type of distraction that the procedure seeks to prevent.

Another consequence of the failure to turn D 723 over to the Tower at or before the outer marker was that D 723 did not learn of the fog bank in the vicinity of runway 4R until twenty seconds before the crash. While appellants do not argue that radar controllers must warn pilots about airport weather conditions,9 this is the kind of information crews can reasonably expect to receive from the local Tower controller who is in a position to observe and relay last minute weather changes that cannot be observed from the cockpit or the radar room. The duty to turn the aircraft over to the Tower at the appropriate time and place arises in part from the desirability of landing aircraft receiving information of a distinctly local character.

The established Manual procedures for setting up aircraft for the instrument approach and for turning them over to local control are important to the safe and efficient termination of flights conducted in instrument flight conditions. The fact that the procedures here in question are made mandatory by the Government indicates not only that they are not to be treated casually but that pilots may rely on their being followed. Moreover, it is clear from all the testimony that the three omissions by Mr. Taylor contributed to an increased cockpit workload. We therefore cannot treat Mr. Taylor's lapses as trivial. Pilots and passengers alike are entitled to expect better service from the air traffic control system and, although 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. Accord, Moisant Field, supra, 544 F.2d at 273. Unless there were circumstances which would reasonably have caused a controller to act as Mr. Taylor did, the handling of D 723 was negligent. We turn, then, to the problems caused by Allegheny 666 to determine if they provided sufficient excuse for the lapse in service to D 723.

After reviewing the record, we are left with a strong conviction that the separation problem created by Allegheny 666 did not justify the controller's failure to give D 723 standard service. Neither the transcripts of radio communications between Approach and the numerous aircraft it was handling, nor the testimony of Mr. Taylor supports the district court's finding that Mr. Taylor was involved in an emergency situation over a seven minute period.

At approximately two and one half minutes after eleven (1502:25 GMT), AL 666 reported in the Millis holding pattern at 8,000 feet, the same altitude already occupied by EA 1020. (D 723 was at the time on a radar vector flying a heading of 220o .) Approach responded quickly to the potentially serious conflict by directing both aircraft to leave the holding pattern, EA 1020 to the northeast toward Logan Airport and AL 666 to the southwest. It was soon determined that the two aircraft had followed instructions and the immediate problem was resolved. Approach continued to control other aircraft including D 723 to which new vectors were issued. At 11:03 a. m. (1503:28 GMT), Approach told AL 666 to reverse its course and to return to Millis and hold. Mr. Taylor testified that at the time he gave this instruction he was satisfied that, from a separation point of view, it was safe to send AL 666 back to Millis. Almost a full minute later, at 11:04:27 a. m., Approach gave D 723 a final vector to intercept the localizer, but no instruction to descend and to go to Tower frequency, no clearance for the approach and no position report. Not only does it appear from the transcript that there was no reason for Mr. Taylor to withhold a complete standard clearance at that time, but Mr. Taylor himself testified that he had intended to give the full clearance and he didn't know why he had not. Nevertheless, Approach continued to handle the various aircraft in its jurisdiction, vectoring some in readiness for approaches and descending others within the holding pattern. During the early part of minute 11:05 a. m., five attempts to raise AL 666 and to reconfirm its altitude as 8,000 feet were unsuccessful. At 11:05 a. m. (1505:40 GMT), D 723 asked if it were cleared for the ILS approach and the response was "Yes, seven two three is cleared for ILS yes." Shortly thereafter, AL 666 reported that it was still at 8,000 feet10 and Approach took the time to discuss with the flight AL 666's expected approach clearance time. D 723 crossed the outer marker at 11:06:14, after discussion between Approach and AL 666 had, for the moment, ceased, yet Approach still did not turn D 723 over to the Tower and only did so when reminded by the local control coordinator. Mr. Taylor, in his testimony, as much as admitted that he had simply forgotten about D 723.

A controller's job is a busy one requiring the juggling of many aircraft at once and the skillful handling of a constantly changing situation. Too much attention to any one aircraft or problem, however, may mean that others in the system do not get the service to which they are entitled. Mr. Taylor may have been preoccupied with AL 666; he had not in twenty years of experience had two airplanes report in a holding pattern at the same altitude. Certainly immediate, concentrated attention to the problem was necessary. Once the potential crisis was defused, however, there was no longer a legal excuse for failing to deal properly with all of the aircraft under his control. That Mr. Taylor may have been reacting to feelings of worry and concern, while understandable, will not justify the Government for failing to afford appropriate services when it could have done so. We must hold, therefore, that Mr. Taylor's handling of D 723 was negligent.

Our finding that Mr. Taylor's omissions constituted negligence does not end the inquiry, however. Though falling below the requisite standard of care, an act or omission is not actionable unless also shown to be a substantial cause of the injury complained of. We consider next the district court's finding that there was no causal connection between the way in which D 723 was handled by air traffic control and the accident.

Our finding of controller negligence in failing to follow established mandatory procedures is grounded on the premise that the unexcused violation by a federal employee of procedures established by the Government which have as their purpose the protection of those who were in fact harmed constitutes negligence. This is analogous to the familiar tort principle that violation of a statute or administrative regulation is either negligence per se, see Restatement of Torts (Second) § 288B(1), or evidence of negligence, see Guinan v. Famous Players-Lasky Corp., 267 Mass. 501, 516, 167 N.E. 235, 242 (1929). Cf. Gill v. United States, supra, 429 F.2d at 1075 (federal regulations may impose duties and standards of conduct on actors in suit under Federal Tort Claims Act). While such a finding exposes the Government to the possibility of liability it must also be shown that the negligent conduct is a legal cause of harm; here there must be a proven causal relationship between Mr. Taylor's omissions and the crash of D 723. Accord, Spaulding v. United States, supra, 455 F.2d at 225-26. See generally Moisant Field, supra.

Section 431 of the Restatement (Second) of Torts states in part,

"The actor's negligent conduct is a legal cause of harm to another if (a) his conduct is a substantial factor in bringing about the harm."

See also Teasdale v. Beacon Oil Co., Inc., 266 Mass. 25, 27-28, 164 N.E. 612, 613 (1929). Thus, "(i)t is not enough that the harm would not have occurred had the actor not been negligent", Restatement (Second) of Torts § 431, comment a. The act must be a "substantial" cause in bringing about the harm.

"The word 'substantial' is used to denote the fact that the defendant's conduct has such an effect in producing the harm as to lead reasonable men to regard it as a cause, using that word in the popular sense, in which there always lurks the idea of responsibility, rather than in the so-called 'philosophic sense', which includes every one of the great number of events without which any happening would not have occurred. Each of these events is a cause in the so-called 'philosophic sense', yet the effect of many of them is so insignificant that no ordinary mind would think of them as causes."

The question of legal cause is for the factfinder; the issue involves essentially no legal judgments. Therefore, we must uphold the district court's finding unless it is clearly erroneous.

In attempting to establish that Mr. Taylor's conduct was a substantial cause of the crash of D 723, Delta and the individual plaintiffs had the burden of persuasion.11 See Restatement (Second) of Torts § 433B; Prosser, supra at 241. They were thus required to convince the district court as factfinder that it was more likely than not that the omissions by Mr. Taylor were a "substantial" cause of the accident. This they failed to do. The district court found that the sole cause of the crash was the negligence of the Delta crew and in so doing made the factual determination that there was not sufficient proof of a causal connection between air traffic control service, however characterized, and the accident.

While causation in fact is often a determination that can be made based on inferences from common knowledge, the present case is one where laymen, unaided, could not be expected to draw such inferences. Expert testimony was of utmost importance. The district court, having heard all the evidence, was entitled to weigh the conflicting testimony of expert witnesses and to draw all reasonable inferences from the testimony it credited.12

Viewing all of the evidence, even in the light most favorable to appellants, we think that the district court had ample support for its finding that there was no causal relationship between air traffic control service and the accident. Delta's expert conceded that no one of the irregularities in service would, by itself, cause a problem for a well-trained airline crew. He explained that, standing alone, the vector requiring a 45o intercept would be no problem, that failing to receive a position report prior to reaching the outer marker would not preclude a satisfactory approach, that having to intercept the glide slope from above would not be an insurmountable task, that not being told to go to the Tower at or before the outer marker and being switched well inside the outer marker would not necessarily prevent successful completion of the approach. Although stating that a combination of any two or more would adversely affect the cockpit workload, the expert testified that even so, the lack of services and the attendant problems "would (not) necessarily cause any great alarm", and that, in fact, the approach of D 723 was not unmanageable and was not, as a practical matter, totally unacceptable for everyday purposes until the very last stages of flight. There is evidence that the crew was not concerned by the lack of air traffic control service. As the aircraft intercepted the localizer some two miles from the outer marker, D 723 called Approach, asked if it were cleared for the ILS approach, then acknowledged the clearance by saying "alrighty". There was testimony that this implied that the crew felt ready to make the approach. It can thus be reasonably inferred that in the early stages of the ILS approach, the crew of D 723 was not disturbed by the omissions of Mr. Taylor.

In contrast to the somewhat tentative testimony on behalf of Delta, the Government's expert, an FAA flight examiner in the DC-9, testified unequivocally that the nonstandard air traffic control service received by D 723 could not have caused the accident. He testified that the Captain had, at all times, not only the option but the obligation to make a missed approach if too many difficulties developed. A missed approach can be executed in a DC-9 from very low levels and in a matter of seconds. A reasonable inference from the fact that D 723 did not go around and apparently made no attempt to do so would be that even in the last stages of the flight, none of the additional workload caused by the lack of service to D 723 appeared to the Captain to create a problem of major proportions even with the added factor that the flight director was not functioning properly and the aircraft was, for much of the approach, well left of the localizer.13

As to the specific effects of the various omissions, there was convincing evidence that the excursions to the left of the localizer course were related to the faulty signals from the flight director and not to the angle of intercept provided by Approach. The flight profile shows that the 45o turn onto the localizer was accomplished without excessive deviation to the right of course as might be expected if the required turn was too great. The cockpit voice recorder reveals that during the approach the First Officer expressed concern about the flight director and, ultimately, that the Captain ordered that he stop using the sophisticated device altogether. It was therefore reasonable for the district court to find it more likely than not that the cause of the deviations from the localizer had nothing whatever to do with the service or the lack thereof by the controller.

We find more troublesome the question of whether the lack of a position report and the intercept above glide slope were causally connected to the aircraft's speed control difficulties and the fact that D 723 was never stabilized on the glide slope and eventually flew into the ground short of the runway. The answer is that D 723 had apparently coped adequately with any speed and rate of descent problems by the time it had reached decision height. There was expert testimony to the effect that from the outer marker inbound, a gradual reduction in speed took place so that shortly before the aircraft reached decision height, it had also reached the target speed of 124 knots. Moreover, although the aircraft was never established on the glide slope, the rate of descent during most of the period was within normal limits and, according to expert testimony, the plane's glide path flattened out just after it dropped below the glide slope. That the speed and the rate of descent were basically under control toward the end of the approach is corroborated to some extent by the testimony of Delta's expert that the approach was, until the end, acceptable, if not ideal. A reasonable inference, then, is that whatever difficulties might have been caused by poor controller service had been handled successfully by the time D 723 reached decision height. While unquestionably something occurred at about the time the plane reached decision height the rate of descent went from approximately 700 feet per minute to 1200 feet per minute and the speed increased to 130 knots where it remained until impact on the evidence we think that the district court was entitled to find that the controller services were unrelated to the last minute loss of control.

It was Delta's theory, in essence, that had the weather been as reported, the sloppy approach would not have mattered. However, because D 723 was in an unstabilized position at decision height, due in part to the service it had received from the controller, the crew was unable to deal successfully with the fog bank which was purportedly entered unexpectedly after establishing some ground contact. Thus, the really important factual question in this case is whether the crash would not have happened if D 723 had been switched to the Tower frequency at the outer marker and so had learned of the fog bank at the beginning of the final approach instead of a minute and a half later, less than 20 seconds before impact. On this point, airline pilots testified that the information about the fog bank would be useful and welcome at any phase of the approach. Delta's expert explained that the new information would require rapid evaluation and decision-making of the type regularly engaged in by pilots. The witness further stated that the transmission D 723 received from the Tower would not indicate to him that he would be unable to complete the approach to a landing but only that he might have to fly through some fog before reaching the runway. Moreover, while there was some testimony that the weather report D 723 received early in its flight estimated 400 feet overcast, 1 1/2 miles in fog would not have caused a pilot to be particularly concerned about the possibility of a missed approach, there was other testimony by witnesses for both Delta and the Government that, given the forecast for the area, the weather report received, and the unpredictability of fog, especially near the water, any pilot would be anticipating the possibility of a missed approach that morning. Finally, the cockpit voice recorder reveals that the crew did not react at all to the Tower transmission reporting the fog bank when it was finally received. We cannot say that the court could not reasonably have inferred that, as Delta's expert indicated, the pilots simply did not consider the report to have any major bearing on the probability that they would be able to go ahead and land. If, as one might conclude from both the testimony and the evidence of a lack of reaction to the additional weather information, the crew continued for some reason to believe that the weather at the airport was basically as it had been reported to them earlier, the argument that receiving news of the fog a minute and a half sooner would have made the crew act differently is not conclusive. The district court was entitled to find that Delta and the individual plaintiffs did not show by a preponderance of the evidence a causal connection between D 723's having learned of the fog bank only late in the approach and the accident.

Without an established causal connection, the fact that the controller did not provide proper services in several respects is not a sufficient basis for holding the Government liable.

Affirmed.