Back to Annotated Bibliography



Gregory F. Zehner
Human Effectiveness Directorate
Crew System Interface Division
Wright-Patterson AFB, Ohio 45433-7022

June 2001

Interim Report for the Period January 1978 to October 2000

Pages i - xii and 1 - 127

NOTE: The following are selected extracts from or annotations regarding the subject publication. 


     "This project was presented as a dissertation to Ohio State University. However, it was a group effort. A number of people helped in various stages of its' completion. Ken Kennedy, who has been my friend and mentor at AFRL, helped develop the aircraft measurement methods, and assisted me in gathering the T-38 data. Jeff Hudson also helped gather and organize data, and was one of the team that developed the Multivariate Models program discussed in Chapter 5. Richard Meindl also helped develop that technique. Joyce Robinson has helped me assemble databases for many years, and continued her kind and patient support on this project. Finally, Patrick Files did the initial editing of the manuscript and Tina Brill helped with formatting.


     Designing aircraft cockpits to accommodate to wide range of body sizes existing in the US population has always been a difficult problem for Crewstation Engineers. The approach taken in the design of military aircraft has been to restrict the range of body sizes allowed into flight training, and then to develop standards and specifications to ensure that the majority of the pilots are accommodated. Accommodation in this instance is defined as the ability to: 

     This dissertation describes a methodology for correcting this problem and demonstrates it by predicting pilot fit and performance in the USAF T-38A aircraft based on anthropometric data. The methods described can be applied to a variety of design applications where fitting the human operator into a system is a major concern. A systematic approach is described which includes: defining the user population, setting functional requirements that operators must be able to perform, testing the ability of the user population to perform the functional requirements, and developing predictive equations for selecting the future users of the system.

     To each of these people I offer my thanks and a cold one the day this dissertation is accepted.



     Two recent policy decisions by the U.S. Government have created an immediate need for anthropometric data and accommodation performance data for people of extreme body size in USAF cockpits. In addition, the ability to predict accommodation levels based on an individual's anthropometric data has become very important.

     The first of these policy changes occurred when the Secretary of Defense (Aspin Memorandum, Apr 93) and Congress expressed the need for the services to expand opportunities for military women by opening career paths that had previously been restricted to males. This has resulted in a small number of women being trained in and assigned to Fighter Aircraft.

     This policy change has created a problem. All existing USAF aircraft were designed to accommodate a male pilot population with a minimum Stature of 64 inches and a minimum Sitting Height of 34 inches. Traditional cockpit design practice was to perform anthropometric surveys on the existing pilot population and to use summary statistics from those surveys as design requirements for aircraft. On the small end of the design range, 5th percentile male pilot values for critical body dimensions were used as minimum design points. Those members of the population smaller than the minimum design values sometimes had to stretch in order to be accommodated. Unfortunately, of those females meeting the minimum [pilot training] entry requirements (~45% of military women) a very large percentage fall below 5th percentile male values. On the large end, 95th percentile male values were used as design limits. Larger pilots may have clearance and escape problems.

     Previous experience has shown that assignment of individuals to aircraft in which: they are too small to adequately reach switches and controls, see over the nose to land, achieve full rudder throw with brakes, move the control stick to the full range of it's capability, or have escape clearance problems, are at increased risk for mishap.

     The second policy change occurred when Congress and the Department of Defense directed the Joint Primary Air Training System (JPATS) to accommodate a much wider range of body sizes than are currently allowed to enter flight training. The JPATS aircraft will be the primary trainer for both the USAF and Navy for the next 30 or so years. This change in design philosophy was necessary because body size restrictions for becoming a pilot prevent the majority of women from entering flight training. While smaller males will also benefit from a change in design philosophy, the largest impact will be felt in the female military population. Unfortunately, this policy change has the potential to dramatically increase body size fit problems.

     The JPATS aircraft was designed to accommodate 97% of the "general female military population." While this group must meet all of the other criteria for entry into flight training, it is not subjected to the 34 inch Sitting Height and 64 inch Stature limitation. It appears that individuals of 31 inches in Sitting Height and 58 inches in Stature will be able to fly the JPATS aircraft. For that reason, the US Air Force is now considering expansion of the body size entrance requirements (AFI 48-123) for Undergraduate Pilot Training (UPT). This change is intended to provide essentially equal opportunity for both genders for entry into flight training.

     At the same time, larger pilots are also being allowed to enter flight training. The current maximum size for pilots is 40 inches in Sitting Height, and 77 inches in Stature. While the large body size restriction has been in place for several years, some individuals have had the size requirements waived, and been permitted to become USAF pilots.

     While it will be possible for pilots of extreme body size to operate the JPATS aircraft when it is completed, these pilots must continue training in either the T-1 (Tanker/Transport trainer) or the T-38 (Fighter/Bomber trainer). After that training they will be assigned to one of the other 40 or so types of aircraft in the USAF fleet. Our previous experiences in evaluating accommodation in some of these aircraft indicated pilots smaller than the 5th percentile or larger than the 95th percentile design requirements could have difficulty operating them. Therefore, a much larger percentage of the population will be at even greater risk if entrance requirements are relaxed.

     While currently only a few accident investigations have reported body size as a cause of the mishap, we appear to be very near the limits of current aircraft accommodation. A change to pilot entrance requirements could create a very dangerous situation.

     This research project focuses on the T-38 aircraft. This aircraft was selected since it is the next step (after JPATS) in flight training for pilots headed to the Fighter/Bomber track of Specialized Undergraduate Pilot Training. Five questions related to accommodation are addressed in this research.

Anthropometric Samples

     1) What are the anthropometric profiles of the current male and female pilot populations, and, the potential pilot populations if size restrictions are removed?

     Chapter 2 addresses sample construction. That is, the creation of several anthropometric datasets. These datasets must be representative of current male and female pilots as well as those individuals who could be pilots if anthropometric restrictions for entry into flight training were not in place. The USAF has not performed an anthropometric survey on female members since 1968 or male pilots since 1967. Because those surveys are now outdated, a sample representative of the current population is needed.

     To create current datasets [representative of USAF flying populations], the 1988 U.S. Army Anthropometric Survey (Gordon et al, 1989) 'datapool' was used. In the Army survey, researchers used a stratified sampling strategy for age categories and over-represented specific ethnic/racial groups. This was done so that in the future if there are demographic shifts in the Army population, restructured subsets could be constructed which keep the "working database" current. The datapool includes over 200 measurements on more than 5,000 subjects. Using a similar philosophy, . . . [the] Army datapool is restructured to match USAF demographic profiles.

     This was accomplished by selecting subjects from the Army datapool representative of the age, race, and height/weight profiles of the USAF population. In doing so, the significance of each of these parameters on anthropometric dimensions was studied. Age was examined since growth is not always complete in the military population, and because pilots must be college graduates. This cuts the lower end of the pilot age distribution off at 21 years. Younger subjects may need to be excluded from the dataset due to incomplete growth. Age categories of 5 years were compared to check for secular and growth differences within the datapool. Similar statistical approaches were then applied to examine ethnic differences in anthropometric distributions.

     The results of these tests indicate that it may be improper to combine African-American and European-American samples in the same dataset in the proportions existing in the current USAF pilot population (~85% European-American) because significant differences in body type may  be hidden in the summary statistics. It may be necessary to separate these groups for statistical analysis because the accommodation problems each group encounters may be quite different.

     Next, since Height and Weight restrictions for the Air Force are different from those of the Army, comparisons of their effect on the resulting samples are necessary. Weight differences obviously effect many well correlated anthropometric dimensions (such as Waist Circumference or Hip Depth). A key examination was to assure that all of these restrictions have not resulted in a violation of the multivariate normality assumption used in other analyses. Bimodal distributions may result from combining two very different samples.

Operational Requirements

     2) What tasks must be performed in an aircraft to safely and effectively operate it?
Chapter 3 addresses the establishment of the "operational requirements" for the T-38. These requirements establish the pass/fail criteria which pilots must perform to safely operate that particular aircraft. While it is obvious that all controls must be reachable in an aircraft, which ones must be reached in an emergency condition? In an emergency, the inertial reel restraint system may lock, or, due to adverse G forces, the pilot may be pushed into a difficult position from which to reach a particular control. For these reasons, critical reaches as well as minimum visual fields (to see the landing zone, or other aircraft in a formation) were defined. This research was done at the Instructor Pilot Training School at Randolph Air Force Base, Texas. This school is a unique resource since it is where instructor pilots  are trained. The entire syllabus of training maneuvers as well as student errors and emergency procedures for recovery from them are the focus of this training. A panel of Instructor Pilots and Safety Officers was assembled to discuss and define the operational requirements for the aircraft. 

     The areas defined are: minimum external visual field, the "critical controls list" (which controls need to be accessible during emergency situations where the pilot may have a locked inertial restraint system or be unable to reach a long distance), adequacy of rudder pedal and brake reach, the necessary range of stick/yoke mobility, and adequate clearance space for control operation and ejection.

Cockpit Mapping

     3) By using "cockpit mapping" techniques, can the performance of an individual in a particular cockpit be accurately predicted from anthropometric measurements, and, can these data be used to predict accommodation percentages for the population?

     Chapter 4 describes the anthropometric evaluation used to determine which body sizes are able to meet the minimum accommodation criteria once the operational requirements set has been defined. Cockpit Mapping is the technique used to make measurements on a sample of subjects performing the operational requirements in a crewstation. Regression equations based on sample data are then used to predict performance levels for the population. The methods which will be used in this research require at least 20 test subjects representing as well as possible the extremes of body size within the potential user population. Samples of roughly this size were decided upon based on previous experience with these types of data. Typically, some data editing is required. If fewer than 20 subjects are used it becomes difficult to determine which subject data should be considered outliers. 

     When combined with the critical tasks list discussed earlier, these data can be used to assess the impact of accommodation limits on the entire population in terms of the percentage which can or cannot operate a particular aircraft safely. By applying the results of the performance evaluation in the cockpit to the datasets constructed to represent the pilot population, the severity of the non-accommodation problem that exists for the current pilot population as well as the severity of the problem if anthropometric entrance requirements are changed can be determined.

Future Design Criteria

    4) What anthropometric statistical methods should be used to design future cockpits so that accommodation levels can be increased?

     Chapter 5 presents the creation of new statistical techniques for the design of future aircraft. The traditional method of design uses lists of 5th and 95th percentile values for a large number of dimensions. Primarily body segment lengths. Nearly all current USAF aircraft were designed in this way. Unfortunately, this method leads to many errors and misconceptions since percentiles are not additive, and do not describe variability in body proportions. A multivariate technique for describing body size variability should be used to specify new aircraft design and existing aircraft modifications.

     Using a Principal Components technique developed by Meindl, Hudson, and Zehner (1993), several small subsets of body types which exhibit the range of size and proportional variability existing in the larger population will be constructed. If the body size variability exhibited by these subsets is accommodated into a new aircraft design, then the target percentage of the total population will. This system is now in place for the design of new USAF aircraft.

Crewstation Design Methodology

     5) Using the data information described above, what methodology should be used to incorporate anthropometric information into the design of an aircraft?

     Chapter 6 describes a step-by-step methodology for using these data in the design of a cockpit. This methodology should be used in place of outdated Military Design Standards such as 1333 C (Aircrew Station Accommodation Criteria For Military Aircraft). This Standard uses the traditional "percentile man" philosophy as well as a number of seemingly arbitrary design rules in crewstation designs.

     While this dissertation addresses a very specific design problem, the methodologies described can be applied to a variety of design applications where fitting the human operator into a system is a major concern. A systematic approach which includes: defining the user population, setting functional requirements that operators must be able to perform, testing the ability of the user population to perform the functional requirements, and where necessary, developing new design criteria and methods that assure accommodation, is the key to a successful human engineering design.



Dataset Construction
Age Structure
Combined Samples
Body Fat



T-38 Operational Requirements
External Vision Requirements



     ... Each area of accommodation ... involve different numbers of subjects, depending on the amount of variability we expect. For example, overhead clearance is a straightforward measure in which clearance above the head is added to the subject's Sitting Height. When the seat is positioned full down, the subject's Sitting Height plus the clearance space sum to the largest Sitting Height that could be seated with no head clearance. Because there is little variability in results, just four large subjects are averaged to arrive at the final value. For reach to controls however, subject results vary a great deal because of harness fit, strength, motivation, and a number of anthropometric variables. We use a larger number of subjects and perform multiple regression analysis to produce the final results for this area of accommodation. ... 

     For the T-38, we examined seven aspects of anthropometric accommodation:

     1. Overhead clearance.
     2. Rudder pedal operation.
     3. Internal and external visual field.
     4. Static ejection clearances of the knee, leg, and torso with cockpit structures (i.e. canopy bow).
     5. Operational leg clearances with the main instrument panel.
     6. Operational leg  clearance with the control stick motion envelope and pilot's ability to attain the full range of stick travel.
     7. Hand reach to controls.

     In aspects of accommodation (overhead clearance and vision, for example), anthropometric relationships are obvious and fairly simple. Overhead clearances are directly related to Sitting Height. Vision out of the aircraft, primarily ONV [Over the Nose Vision], is directly related to Sitting Eye Height. For these measures, multiple anthropometric dimensions are unnecessary to explain accommodation levels. 

     Other measures of accommodation are more complex. For example, operational clearance of the body with the control stick motion envelope can be restricted as the stick is pulled aft. There often is not room between the thighs to roll the aircraft ...  . Limitation of stick motion is influenced by Sitting Eye Height, Thigh Circumference, and Buttock-Knee Length. The relationship between the upper seat positions (used by pilots with small Sitting Eye Height) and Thigh size seems to be the most critical. ... .

     As the seat is raised to improve external vision, the range of stick travel side-to-side increases ... . ... large pilots will typically use the full-down seat position, and the control stick is usually so far above the thighs that interference does not occur. However, small pilots are typically adjusted as high in the seat as possible to gain adequate over-the-nose vision. In this seat position, the stick often contacts their thighs. Also, pilots with long legs are typically able to spread their knees apart, making a greater space available between the thighs for control stick movement. Small pilots may not be able to spread their legs while keeping their feet on the rudder pedals" 

[There appears to be an oversight at this point in the outline of this report. Whereas, further on, there is a section entitled "LARGE PILOT ACCOMMODATION," there is no corresponding and necessary section entitled "SMALL PILOT ACCOMMODATION." The latter title should logically appear at this point.]

"Test Sample

     The T-38 study of small pilot accommodation included 22 small test subjects, each equipped in the full complement of flight gear used by the Air Education and Training Command. Prior to measurement of their capabilities in the cockpit, each subject was measured on 18 traditional anthropometric dimensions ... . ... subjects were selected to represent the small size extremes of the population while retaining a reasonably normal distribution for each measure. [The Figure below] compares this sample ... to the USAF baseline population ... . 

Forward Vision Over the Nose

     ... Vision ... was measured in two body postures in the front cockpit and one in the rear. In the front crewstation, ONV was measured with the subjects looking straight ahead over the nose of th4e aircraft. Subjects were instructed to keep their heads level (i.e. in the Frankfort Plane). An Abney Level ... [see below] was used to measure the depressed elevation angle to the ground over the nose of the aircraft." [Unfortunately, in this photograph, the Abney Level is not adjusted properly for final reading.]

Reach to Rudders

     Like ONV, the ability to reach and actuate rudder pedals and brakes is affected by seat position. A pilot with very short legs may lower the seat to reach the rudder pedals. However, minimum vision levels (and, therefore, seat position) must be maintained throughout a mission. Under normal circumstances pilots should not be allowed to excessively sacrifice external vision. The pilot who is small in Sitting Height will have to adjust the seat [upward] to achieve adequate vision. This moves the pilot farther away from the rudder pedals. If the seat can be lowered and acceptable vision out of the aircraft maintained ..., the pilot can improve access to the rudder pedals. 

     [Many aircraft require] ... very little rudder input when in the air except during slow flight "gun jinks". These radical maneuvers are used when trying to avoid enemy fire. The pilot slams full rudder and quickly pushes the stick in various directions causing extreme movements of the aircraft. In addition to jinks, maneuvering on the ground and maintaining control in case of a blown tire on landing or takeoff require the ability to apply full rudder and brake simultaneously. Measurements were made in a number of seat positions so that the effect of seat movement could be calculated.

     In this analysis subjects placed their feet on the rudders with their toes [forward part of the shoe sole] on the brakes. Full ruder throw was defined as full rudder input, and full brake, with the knee fully extended. The subject was tightly restrained and not allowed to slide forward in the seat. This method of positioning the foot is an intentionally conservative estimate: under certain flight conditions, a great deal of strength is required to hold the pedal in. 

     Measurement was made to the rudder adjust position where the subject could just actuate the rudder and brake. A regression equation was developed using rudder position and leg length, and the leg length equated to a full aft rudder adjustment was calculated. 

     The measurement which best identifies the minimum leg length required to reach full rudder throw is a combined leg length. Buttock-Knee Length and Knee Height Sitting are summed to arrive at a new [artificial] measure called Comboleg. For example, if a [minimal] 42-inch combined leg length [Comboleg] is required to obtain full rudder throw, it does not matter if an individual has a 23-inch Buttock-Knee Length and 19-ing Knee Height Sitting or a 22-inch Buttock-Knee Length and 20-inch Knee Height Sitting. Their reach to rudders ... will be the same. The correlation between Comboleg and rudder adjust position is .96.

     The graph below shows miss distance (negative numbers) [and excess reach] to full rudder and brake for a variety of leg lengths. With the seat in the full-up position, a combination leg length [Comboleg] of 43 inches is required to attain full rudder and full brake simultaneously.

Arm Reach to Controls

     Pilots must be able to reach and operate hand controls to safely fly an aircraft. In normal flight conditions, with the inertial reels unlocked, this is not a difficult task. Under adverse-G conditions, however, when there is an inadvertent reel lockup, small pilots will have difficulty reaching many controls. ... 

     ... Several factors other than body size affect reach capability in an aircraft cockpit. The design, fit, and adjustment of harnesses, personal protective equipment, survival gear, body strength, and motivation, all influence the act of reaching. Due to these factors, reach is the most difficult area of accommodation to accurately quantify. For that reason, we liberally edited outlier subjects. Subjects more than 2 standard errors away from the predicted values for a given reach ... were examined for possible deletion. 

     Reach to controls was based upon two harness configurations ...: first, with the reels locked and shoulders against the seat back. This is referred to as a Zone one restraint condition (MIL STD 1333C). Next, we evaluated reach in Zone two, where the reels are locked but shoulders are allowed to reach out toward the control with a maximum stretch. [Zone 2 is illustrated below.] ... In Zone 3 ... the harness is not locked and the subject is allowed to lean forward to gain access to controls. All subjects were able to reach all controls of interest in a Zone 3 harness configuration. 

     Reach was initially measured in the full-up seat position, and then repeated in a lower seat position to determine the change in reach ability for an increment of seat adjustment. Measurements were taken from the interface point on the body [the hand] to its respective contact point on the control. ... Miss or excess distances were measured and regressed against body dimensions to determine the body sizes and proportions just able to ... [reach].

      Reach to a particular control is a function of arm length [as indicated in the measurement] (Span) and torso height. Torso height plays a large role in seat adjustment, since the pilot must seek at least minimally adequate vision. Moving the seat up, however, [typically] moves the pilot further from some controls [i.e., those below shoulder level]. 

     Arm reach may also be affected by the width of the shoulders, primarily because of the restraint system. ... Wide-shouldered subjects are relatively free to move around the shoulder straps while stretching. ...

     Seat position effects were calculated by averaging differences in reaches for each subject between the full-up seat position and the down-one inch seat position. The results indicate that for each inch the subject lowers the seat, miss distance to the throttle is reduced by 0.9 inches (range = .25 to 1.75 inches). A 2.5-inch smaller Span measurement would reduce miss distance by 0.9 inches. [The author gives no evidence here to support this latter conclusion.]  ... Again, in the analysis, subjects were ... positioned so that they would see at least the minimum -11 degrees visual angle [over the nose of the aircraft]. 

Prediction of Reach to Throttle -95% Confidence

     [As can be seen in the above chart, negative values are used to report GREATER reach capability. A confusion arises when it is reported, as above, that a SMALLER Span REDUCES miss distance. One must interpret "reduce" to mean a move from more to less negative, and possibly into positive values - or, as is used here, reflecting lesser reach capability. As we saw in the Reach to Rudders section this interpretation is reversed - positive values indicate greater reach with the leg, negative values indicate less.] 

     ... two steps ... [are] necessary to determine the percentage of the various populations accommodated on reach ... . Two steps were required because, if a pilot's arms are too short to reach the controls, he or she may be able to lower the seat to get closer to the controls. Lowering the seat is acceptable if the subject still has adequate (-11 degrees) ONV in the lowered seat position. During data analysis, therefore, we mathematically adjust the seat so that each person ... sees -11 degrees over the nose. From that seat position, we determined if the subject could reach [the control]... . 

Total Accommodation Rates for Small Pilots 


     [The author includes a section titled "Test Sample" within his discussion of small pilot accommodation. Contrary to expectation, however, he did not include a similar discussion here, his discussion of large pilot accommodation.]

Overhead Clearance

     Inadequate overhead clearance in an aircraft ... can interfere with pilot performance and can be an ejection hazard. If the pilot is unable to sit erect with his head firmly in contact with the seat headbox, poor spinal positioning could result in an injury during ejection. Also, pilot mobility and his or her ability to check the sky for other aircraft directly behind (the "six o'clock" position) is reduced. Both of the3se problems are exaggerated when the aircraft is under negative G-forces or is inverted. The pilot's head is then forced into the canopy.

     During these measurements, the pilot sat erect with the head held in the Frankfort Plane (horizontal line of sight).[*] The space between the top of the head and the underside of the canopy was measured. In addition, clearance space had to be verified in a manner to ensure that the pilot could place his head fully into the head box before ejection and have sufficient side space for checking the sky ... directly behind him ... . ... We began with the subject in the full down seat position and adjusted the seat upwards until the head contacted the canopy. His or her mobility to turn and "check six" were then tested, and the seat was adjusted down until head mobility was acceptable. Seat position was recorded, and the distance from the seat full-down position was added to the subject's Sitting Height. ... [This value, then represented the absolute maximum accommodated Sitting Height.] 

     [* The definition of Frankfort Plane is unrelated to "the horizontal line of sight. The classical definition is as follows: "The standard horizontal plane of orientation of the head, realized when the lowest point in the margin of the left eye socket (orbit) and the left tragion ([an approximation of the] superior margin of the external auditory meatus) are in a common horizontal plane" - from A Collation of United States Air Force Anthropometry (U), K. W. Kennedy, AAMRL-TR-85-062, Harry G. Armstrong Aerospace Medical Research Laboratory, Wright-Patterson Air Force Base, OH, January 1986 - and others. A horizontal line of sight may approximate but is independent of the Frankfort Plane and can be horizontal in a great variety of head positions.]

     Since helmet designs in the military are subject to change, these measurements were taken two ways: bareheaded for overall clearance, and with the lHGU-55/P (the current flight helmet) to test mobility. When a new helmet comes into the inventory, the HGU-55/P data may become obsolete and will [or may] need to be replaced. ...

     Sitting Height is the only anthropometric variable of interest for overhead clearance. The correlation between Sitting Height and Overhead Clearance is -.92. ...

Leg Clearance

Leg Clearance to the Canopy Bow

     ... Clearances for escape were measured to the Canopy Bow ... to ensure the pilot would not strike this structure during ejection. ...

Operational Shin Clearance

     ... [O]perational clearance was measured forward from the shin to the bottom edge of the main instrument panel to ensure ejection clearance the pilot has space to operate the rudders. ... 

Large Pilot Final Accommodation Percentages


Stick Interference with the Thigh

     One final anthropometric accommodation problem [,stick Interference with the thigh,] exists that we were unable to quantify. When the seat is full-up, there is very little space between the thighs for stick roll authority (pulling the stick full aft and moving it left and right all the way to its limits). This problem is made worse if the pilot has short legs. For small subjects, reach to rudders is so difficult that the knee is fully extended and the pilot is unable to spread the thighs apart to make room for ... stick [travel]. However, the relationship between body size measures and stick/thigh interference is unclear. The correlation between body size measures and stick interference problems was near zero. However, 13 of 19 subjects tested with the seat full-up had stick movement restricted by one to two inches. ... 



Percentile Limitations
Regression Modeling
The USAF Multivariate Accommodation Method
Bivariate Distribution for Accommodation
Principal Component Analysis
Cockpit Accommodation Example
Racial/Ethnic Variability
Male Comparisons



External Vision
Internal Vision
Overhead Clearance
Reach to Rudders
Shin Clearance
Escape Clearances
Arm Reach to Controls
Control Stick Range of motion



     There are 32 references. 




Back to Annotated Bibliography