airplanes converging
Getting your Trinity Audio player ready...
7 min read

There I was in a left echelon trailing position in a close formation of two KR-2’s with my hangar partner, Dick Osterhof.  It was a typical Sunday morning flight from DeQuincy, Louisiana (5R8), with visual meteorological conditions and Gulf Coast haze.  We were not going anywhere in particular, just staying close to our home base at about 3,000 feet.  Dick was leading our flight path and navigating, while I spent 99 percent of my time looking at him, keeping about 20 feet of clear space between us.

While keeping an eagle eye on Dick’s KR-2, I noticed a tiny black speck show up about an inch or two above his half-bubble canopy.  The speck eventually sprouted a fuselage, twin-engine nacelles and a T-tail.  By the time the wing panels outboard of the engines became big enough to see, along with the turbine exhaust pipe exiting the near side nacelle, I was measuring four G’s on my panel accelerometer and depressing my control stick microphone switch, saying with urgency, “DICK, PULL UP”!

collision diagram

Figure 1. Blue outlines show actual aircraft midair threat geometry. Red Beechcraft shows hypothetical virtually invisible alternative location.

I estimated that only about five seconds passed from the first moment of the dot being visible over Dick’s canopy until flight path intersection.  The dot blossomed into what could only be a Beechcraft Super King Air.  Our 140 mph airspeed, plus the King Air’s capability of 350 mph added up to a potential 500 mph closing velocity.  (A King Air pilot obeying Federal Aviation Regulation Part 91, Section 91.117a, would have been flying no more than 288 mph below 10,000 feet MSL—a trivial distinction.)

By pulling up so abruptly, Dick’s KR-2 disappeared below my instrument panel.  His invisibility inspired my choice to go straight up, knowing that I had begun my evasion maneuver at least a couple of seconds before he could begin anything like my own pull up.

Having asked Dick to “pull up,” I anticipated that he was likely coming up where the floor of my KR-2 under my rudder pedals was blocking my view of his flight path.  Therefore, pushing my stick forward to recover towards his assumed vertical trajectory seemed like a losing idea.  Having just missed a King Air, I did not want to run into a KR-2.

Pulling the stick back to reverse course before my airspeed sagged below 50 mph was my guaranteed safe path away from Dick and back the way I’d come.  By the time I was upside down heading away from him, I was sure I did not want to complete a loop headed back in Dick’s general direction.  Rather than take a chance on a loop completing a path back towards running into him, I rolled halfway around to right-side-up, let the nose drop, and leveled out after my first ever Immelmann.

Once I was settled in level flight headed safety away from Dick, I radioed him to see where he was.  There would be no sense in turning back together until we both knew exactly where to find the other.  After being separated by almost a mile, we finally found each other for the return flight back to our hangar at DeQuincy.

Dick eventually explained later that at my call to “pull up,” he delayed taking action long enough to visually acquire the King Air for himself.  Not knowing where I’d gone, he decided against following me up in the world’s loosest unplanned vertical formation.  He chose to dive away from the King Air, thereby automatically accelerating away from me in a way I could never overtake even if I tried to hit him.  Both being engineers, we’d each manipulated potential and kinetic energy in a way that was guaranteed to keep us apart, no matter what the other pilot might do, for at least several minutes after missing a three-aircraft midair collision.

Dick estimated later that the King Air passed about 15 feet directly over his canopy.  The King Air was close enough to see its rivets, along with its pilot’s aviator sunglasses.  The clearly visible pilot’s facial expression was one of stoically blissful ignorance, lacking any clue that he’d detected the grave danger he’d shared with the two of us.  His lack of any facial recognition of his mortal peril, along with the total absence of any observable deviation from his straight-and-level flight path, convinced Dick that the King Air pilot never saw us.

Reflecting back on the flight geometry in Figure 1, we realized that if the King Air had been coming from the left, as shown with the red position in Figure 1, instead of from the right, as it actually was in the blue position, then my left echelon trailing position, with eyeballs 99% looking ahead to the right directly at Dick, would have made it nearly impossible for me to detect the King Air during the five seconds it might have been visible on the left side until too late.  I would have been looking in the wrong direction, to the right.  The “high definition” fovea of the average eyeball retina can only perceive fine detail in about a 10 degree angle.  Alternatively, if I had been flying in a right echelon trailing position, my 99% visual duty cycle looking ahead to the left at Dick would have made it nearly impossible to detect the King Air coming from the front right until too late, as with the blue outlines in Figure 1.

collision diagram

Meanwhile, Dick, being the leader of our flight of two, was operationally expected to aviate, navigate, communicate, and look for traffic coming from the forward arc of our flight trajectory.  After all, my Job One was to not run into him, requiring that I watch him almost all the time.

Imagine flight leader Dick looking outside for traffic scanning from one side to the other in 10-degree foveal arcs for about one to two seconds in each high definition arc as required for the eye-brain electro-chemical system to actually detect and perceive a collision threat.  A 180-degree forward scan for collision threat traffic would then require 18 views of at least one second each.  A hypothetical two-second scan at the flight instruments between every outside scan of 18 seconds would then add up to about 20 seconds for the scan cycle.

So what happens to the probability of a pilot detecting a collision threat if a threatening aircraft is only visible for five seconds of a 20-second visual scan cycle, as it was in our Figure 1 Louisiana scenario?  To a first approximation, five seconds divided by 20 seconds gives a one-in-four probability of even looking in the correct direction to detect the threat before two aircraft colide.  Then 15 seconds of the 20-second visual scan cycle has the threatened pilot looking in the wrong direction 75 percent of the time, with no chance of detecting and evading the collision threat.  Never mind that detecting the threat in the tail end of five-seconds of perceivable, detectable visibility may still leave too little time to late to even begin an effective collision threat evasion maneuver.

I learned two unforgettable lessons about aviation safety probability physics from this encounter.

(1)  Flying at cardinal altitudes (1000, 2000, 3000, 4000, …), where others are liable to also be flying at the same cardinal altitudes, maximizes the danger of a midair collision.  Following the crowd into accurate flight at cardinal altitudes is a losing bet that wastes the far safer random altitudes in between cardinal altitudes.

(2)  It is impossible for any pilot to visually scan outside the cockpit with enough diligence to guarantee a safe landing any time where closing path traffic is visible even one second shorter than the length of time between the beginnings of a first and second visual scanning pattern looking for collision threat aircraft.  The limitations of the fovea and the eyeball-brain blindness outside the foveal high definition arc guarantee that the National Transportation Safety Board will forever be using accident analyses with copy-and-paste boilerplate about the “the inherent limitations of the see-and-avoid concept” for detecting and avoiding collision threats.

This near midair collision inspired my 1997 peer-reviewed, published analysis supporting the above two lessons learned at:

My 2018 Experimental Aircraft Association (EAA) webinar convincing a supermajority of 89 percent of voting pilots of the above two lessons is available for viewing by current EAA members at:

My 2023 Colorado Pilots Association webinar updating my 2018 EAA webinar is available at


Robert Patlovany
Latest posts by Robert Patlovany (see all)
10 replies
    • Robert Patlovany
      Robert Patlovany says:

      Do you believe that ADS-B guarantees you won’t have a midair collision? Have you noticed how many fully equipped ADS-B aircraft are featured in NTSB accident reports?

      While I am very happy to fly with ADS-B now, the FAA System Safety Handbook, December 30, 2000, Section 3.6, Safety Order of Precedence, Priority 1 is “design to eliminate risks.” Priority 3 is “provide warning devices.” Priority 4 is “develop procedures and training.” Both ADS-B and TCAS are Priority 3 systems that, combined with Priority 4 pilot and controller procedures and training, can never be as effective as a Priority 1 system, so says the FAA System Safety Handbook, and the MIL-STD-882 standard used as its template.

      My Priority 1 recommendation is random altitude cruising to eliminate or minimize collision risk manufactured by cruising at common altitudes. This recommendation fixes the “navigation paradox” described by FAA Chief Scientist, Robert Machol, as early as 1975, whereby aircraft violating altitude assignments to integer-thousand-foot altitudes are more likely to avoid midair collisions. The paradox is that the most accurate pilots and autopilots “earn” the greatest collision risk.

      My 2023 CPA webinar promotes the 1928 Altimeter-Compass Cruising Altitude Rule (ACCAR) as a physics-based Priority 1 alternative to cruising at common altitudes, in general, and the hemispherical cruising altitude rule (HCAR), in particular. The hemispherical cruising altitude rule in FARs 91.159 and 91.179 violates the FAA’s own very poorly (or not at all) implemented Priority 1 by actually designing midair conflicts into the airspace control system. Then, extremely expensive Priority 3 systems of TCAS, ADS-B, and air traffic controllers following Priority 4 training can “save the day” in the last seconds before impact and look like heroes. Many make big money selling Priority 3 and 4 solutions. These systems based on humans and hardware naturally fail as expected in 2002 over Germany and 2006 over Brazil. What happens to the firefighter who gets caught setting a fire to look like a hero when putting it out?

      Imagine a hypothetical theory that a disease can be manufactured that can be cured with expensive medicines, thereby making a profit that makes the risk of the disease worthwhile to those making the profit. The HCAR FARs would be the manufactured disease cured by TCAS/ADS-B/ATC. In contrast, random altitude cruising flight is the no-cost-to-pilots-and-airplane owners, physics-based, Priority 1 innovation that harnesses the science of what the FAA should have learned about the navigation paradox from its own chief scientist as early as 1975. The physics-based Priority 1 ACCAR cheap fix then becomes like some safe and effective aspirin that is anathema to those selling expensive Priority 3 and 4 mitigations to regulation-manufactured risk.

      We were not really cruising anywhere in particular below 3000 feet AGL However, we did make the point proven in my 1997 paper and my 2018 and 2023 webinars that IFR and VFR hemispherical cruising altitudes are unsafe for everybody, no matter what they are doing. VFR traffic below 3000 feet AGL may legally cruise at ANY altitude, even IFR altitudes. Just because something is legal does not make it safe, just as obeying an unsafe regulation does not make it safe. Thanks for adding your skepticism about the safety of legal IFR cruising altitudes below 3000 feet AGL that provide no systematic safety separation whatsoever from VFR pilots legally cruising below 3000 feet AGL at any altitude at all.

      • David Dickins
        David Dickins says:

        Thank you for the fascinating and somewhat sobering article about the reality of collision avoidance or not. I have never understood why the FAA is still so insistent that “see and avoid” with visual scans is the first line of defense. The situational awareness offered by on-screen traffic targets (ADS-B or TCAS) is superior to any visual scan even if the pilot has 20 year old eyeballs. I’m always shocked how hard it is to detect traffic even when you have a target on your iPad to orient your scan angle. I’m not saying stop looking out the cockpit but to recognize the power of modern technology to avoid traffic conflicts well in advance of when the threat would be apparent visually.

  1. Barry Arnold
    Barry Arnold says:

    I had a close (very close) encounter with a King Air who was descending on a reciprocal heading but on a different frequency the first I saw of him was when he appeared out of the clouds heading straight at me, I adopted the dive and steep turn escape route and survived. Took a considerable time to get my heart out of my throat and the pulse rate back to nearly normal. My father was a 727 pilot with TAA and he once told me that if two modern jet airliners are on a head on collision course, by the time the pilots see each other (if they are even looking out of the cockpit) it is too late.

  2. Bob Wieneke
    Bob Wieneke says:

    “Imagine a hypothetical theory that a disease can be manufactured that can be cured with expensive medicines,” … sound a bit like COVID?? (LBNR).
    Seriously, GREAT article, Robert! I’ve used this (off-altitude flight) myself, but not with specific forethought. Having read this article, I now plan to do it with purpose! I wonder if the same dynamics might have applied to airways, and if the GPS-direct routing we commonly use now has helped us avoid conflicts that might have occurred when we were confined to those narrow corridors?


Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *