The NASI Report
Toward a Resolution of the Bigfoot Phenomenon
By J. Glickman, Hood River, Oregon
Diplomate, American College of Forensic Examiners
|Due to server constraints this is an abridged version of the NASI Report.|
Analysis of 1967 Patterson Film
The 1967 Patterson Film, photographed on October 20, 1967 in Bluff Creek, California by Roger Patterson and Bob Gimlin is one of few pieces of photographic imagery that has not yet been proven to be a forgery. This film was evaluated by the author to see if it may be excluded from the set of Bigfoot phenomenon observations by demonstrating it is a forgery.
While it is not possible to prove the imagery in the Patterson film represents an uncataloged animal, it is possible to extract physical and biomechanical parameters which may exclude the possibility that it is a human in a costume. Analysis of the film was performed by converting it to digital data via a film scanner so that the imagery could be processed by computer. The original of the Patterson film is believed to have been photographed on Kodak Kodachrome film. The camera used was a Kodak K-100 16mm movie camera. A first generation copy of the film was made on Eastman 78 Safety film on November 8, 1968, a little more than one year after the original was shot. This first generation copy was first examined by the author in October, 1993 for shrinkage and brittleness to ensure it could withstand digitization. The film had been in storage for twenty-five years in the controlled environment of a bank safety-deposit box and was in excellent condition. The data in Table 6 was extracted from the film (as of October, 1993).
In May, 1994 the first generation copy was hand-carried via a commercial airline flight to Oxberry ATI, a com-pany located in Carlstadt, New Jersey. Oxberry is a manufacturer of high-resolution film scanners that are used by the motion picture industry. One scanner Oxberry manufactures is a modular high-resolution scanner which, when the proper parts are installed, is capable of digitizing 16mm movie film. As of May, 1994 Oxberry had never sold the 16mm components, so Oxberry was contracted to assemble a 16mm scanner from inventory and digitize the film.
Each of the 953 frames of the Patterson film were digitized three times, once each through red, green and blue filters. Each of these three colors was digitized using 12 bits (4,096 levels) at the rate of 2,656 pixels horizontally and 1,912 pixels vertically. This provided 36 bits per pixel at a digitization rate of slightly higher than 250 dots per millimeter (approximately 6,350 dots per inch) which was high enough to image the film grain. Each
frame of the Patterson Film yielded a 30,470,144 byte Silicon Graphics (SGI) file composed of a 512 byte header followed by the pixels organized in 3 planes, one for each color. These frame data files were buffered locally on a Silicon Graphics workstation until twenty frames were accumulated which were then written to an 8mm Exabyte tape cartridge in Unix tar format. The twenty frame files were written twice to each tape to minimize the chance of a tape defect rendering a frame inaccessible.
Because the total size of the dataset is nearly 30 billion bytes (GB) of data, it was impractical in 1994 to keep all of the frames on rotating magnetic storage (hard disk). Instead, each group of twenty frames was written to a recordable CD-ROM. The full dataset is comprised of 48 CD-ROMís which provide near-line, random access to the dataset. All data transfers were performed with 100% read-after-write verification. The software used to perform the image restoration, image enhancement, and other image processing is Clarity, a proprietary image processing software package developed by Photek. Clarity was run on a specially built ALR Evolution V, IBM-compatible computer. Other software used includes Adobe Photoshop and Kodak Shoebox.
Most of the images presented were processed with either image restoration or image enhancement software. Some of the enhancement algorithms operate only on gray-scale, or result in gray-scale images after processing. The original scanned images are RGB, and gray-scale images are the Y component of the YIQ transform. The image contained on the first generation copy that was scanned resulted from the optical superposition of a Kodachrome original onto Eastman Safety stock. This is important because the dyes used in the film for the three colors are not the same size. For this reason, in some instances, the green layer of the film is processed alone because it has the smallest grain and hence captured the highest spatially-resolute image. The image restorations involved motion and focal blur removal which was performed using FIR and IIR filters. Image enhancement included Wallis enhancements, homomorphic equalization, histogram equalization and curve adjustments.
Analyses can be performed on anthropometric and bio-mechanical properties of the subject observed in the 1967 Patterson film. Three types of analyses follow:
An accurate spatial reference has been established from research photographs taken by Byrne and Hodgson in 1972 at the Bluff Creek site [Byrne 1972]. While Patterson believed he set the Kodak K-100 16mm motion camera to 16 frames per second (fps) prior to his departure for Bluff Creek, he reported that in the circumstances surrounding the filming, notably the behavior of his horse, that he may have bumped the film speed
selection dial. Subsequent to the filming, Patterson did not check or record the position of the film speed selection dial. Even if Patterson did know the setting, the Kodak Archives reports that the speed tolerance of the K-100ís film carrier subassembly was ±10% resulting in a range of film speeds from 14.4 fps to 17.6 fps when the film speed selection dial is set to 16 fps [Kodak 1995]. An accurate spatial reference has been established from research photography taken by Byrne and Hodgson in 1972 at the Bluff Creek site. During the course of filming, Patterson filmed several dead trees some of which also appear in Byrne's 1972 research photography of the site, which is not normally useful by itself. However, also contained in Byrne's research photography is an assistant, Michael Hodgson, who is holding a vertical scale (apparently marked every six inches).
Hodgson reported his 1972 height as 6' 0" without shoes and weight as 150 lbs [Hodgson 1995]. Figures 4 and 5 were used to validate Hodgson's height. While Byrne took several photographs of Hodgson that day, the one shown is Figure 4 was used because Hodgson's feet were visible and Hodgson was in the same z-axis plane as the vertical scale. Even though the base of the vertical scale is not visible, the bottom of the scale was artificially extended to verify that it was in the same plane as Hodgson's feet. Because Hodgson and the vertical scale are in the same z-axis plane, Hodgson's height may be read from the scale. The photograph was digitally enhanced to make the washed-out scale markings visible, and shows Hodgson's height as 6' 1¾". Louise Robbins reports mean subject footwear height as 2", so Hodgson's height with footwear is within the expected range [Robbins 1985]. Therefore, Hodgson's total height including footwear is established as 6í 1¾".
Mensuration & Estimation
The mensuration method, which is lengthy, is not presented here for sake of brevity. Contact the author for complete information.
In a separate reference photograph taken by Byrne, Hodgson is seen moving through a z-axis image plane that the subject moves through in frame 326. Since the height of Hodgson has been established and because he is in the same z-axis plane as the subject, Hodgson's height may be used to measure the subject's height provided
that the scale of the two photographs can be matched. Both images contain a dead tree, the size of which did not change during the intervening 5 years which is used to match the scales of the two photographs. During scaling, three points were used to validate that the tree had not changed size. Figure 6 is a digital composite of the film subject and Hodgson in the same image plane with a common image scale established by the dead tree.
Note the successful planar alignment
of the feet in this matched-scale composite. Hodgson's height in pixels
is 276, and the height of the film subject is 327. A simple ratio is used
to compute the height of the film subject:
Thus, the computed height of the subject in the Patterson-Gimlin film is 7í 3½". An error analysis has not yet been undertaken, but in other similar forensic situations is typically ±1".
The waist of the subject in the film is modeled as an ellipse, with the height of the film subject used as the scale reference.
The major axis of the waist ellipse
was extracted from frame 61 and is 30.8". The minor axis of the waist
ellipse was extracted from frame 339 and is 20.4" (Refer to [Glickman
1997] for the extraction method]). Oblique and scale corrections were
required. The perimeter of an ellipse is approximated by [Hudson 1917]
(Note: a is ½ of the major axis, and b is ½ of the minor
Thus, the computed waist perimeter of the subject in the Patterson film is 81.3". An error analysis has not yet been undertaken.
The determination of chest size plays an important role in estimating mass and caloric requirements. The chest of the subject in the film is modeled as an ellipse (see Waist Perimeter above), with the height of the film subject used as the scale reference.
The major axis of the chest ellipse was extracted from frame 61 and is 31.4". The minor axis of the chest ellipse was extracted from frame 339 and is 20.9" [Glickman 1997]. Oblique and scale corrections were required. The perimeter of an ellipse is defined in Equation 5. Thus, using Equation 5 the computed chest perimeter of the subject in the Patterson film is 83". An error analysis has not yet been undertaken.
The methodology used for the extraction of the chest perimeter is discussed at length in [Glickman 1997].
The mass of all primates has been shown to be allometrically related to chest size [McMahon 1983]. Whether this equation is applicable to the subject of the Patterson film is open to debate. It is nonetheless interesting to note the mass estimated by this equation.
The allometric relationship that
relates chest circumference in centimeters to mass in kilograms is:
Substituting the chest circumference in centimeters, the estimated mass of the subject in the Patterson-Gimlin film is 887 kg or 1,957 lbs. An error analysis has not yet been undertaken.
The average stride length reported by Patterson, Gimlin and Titmus measured from the footprint impressions was 81.5" [Krantz 1992]. Byrne reports that this measurement may be in error, as it may have been measured from heel to toe, or toe to heel, rather than heel to heel, or toe to toe. Frame 308 shows a half-stride of the subject.
Using the previously computed subject height as a scale reference, the half-stride length is computed as 46.7" after oblique and scale corrections. Multiplied by 2 to make it equivalent to the average stride measurement reported is 93.4" which is substantially different than that reported. Two possible explanations for this are that the footprints measured by Patterson, Gimlin and Titmus were not made by the film subject, or that an error was made by Patterson, Gimlin and Titmus in measuring and reporting the stride length. With regard to the latter, the difference between the field and film measurements is 11.9" which is close to the 14.5" length of the plaster cast footprint impressions, the difference being 2.6" or
Given the error progagation of oblique and scale corrections, this is in within the expected range. These results suggest that Patterson, Gimlin and Titmus inadvertently measured from toe to heel. An error analysis has not yet been performed.
The length of the subjectís arm is computed from frame 326. Using the previously computed subject height as a scale reference, the subjectís arm length of 161 pixels is computed as 43". An error analysis has not yet been performed.
The arm length expected for this height in a human is 38.5" [Winters 1990]. The standard arm to height ratio is .44H [Winters 1990] (See Appendix D). The ratio extracted from the film is .49H. The arm length of the subject is 5.5 standard deviations from the human mean which is the 99.9999981 percentile or is present in one out of 52.5 million people [Weimer 1993].
This suggests that if the subject is a human in a costume that some form of arm prosthesis is in use. Finger and hand flexion is observed in the film which implies that the prosthesis must support flexion. The use of such a sophisticated prosthesis appears to be at odds with the year the film was made, the technology available at that time, and the financial resources of those involved with the filming.
The length of the subject's leg is computed from frame 326. Using the previously computed subject height as a scale reference, the subject's leg length of 150 pixels is computed as 40" accounting for the bent knee. An error analysis has not yet been performed.
The leg length expected for this height in a human is 46.4" [Winter 1990]. The standard leg to height ratio is .53H [Winter 1990]. The observed ratio extracted from the film is .46H. The leg length of the subject is 3.0 standard deviations from the human mean which is the 99.9 percentile and is present in one out of 1,000 people. While the length of the leg could be the result of a prosthesis, it is the opinion of the author that the probability of this is low because foot flexion is observed in the film.
Edge detection algorithms were run on key frames, including frame 352, with the hope that if the subject was a person wearing a costume, that a seam or interface in the costume would be detected [Gonzalez 1987]. No seams or interfaces were detected.
The morphology, or appearance of the film subject is compared to the great apes. Special attention is given to comparisons with the mountain gorilla because of the visual similarity. Hand, foot, face and body morphology are analyzed in the following sections.
The hand of the subject is seen clearly in only a few frames. Figure 7 shows the fingers extended whereas
Figures 8 and 9 show the hand clenched, thus hand flexion is evident in the film, demonstrating that the hand is not a solid, inflexible prosthesis. In Figure 10 a photograph of a clenched gorilla hand is shown for visual reference [Godwin 1994].
A series of hands from the great apes is shown in Figure 11 which illustrates the evolutionary changes from the gibbon to the human [Kogod 1993]. The notable adaptations include the shortening of the hand and fingers and the movement of the thumb upward.
It is the author's opinion that the subjectís hand appears to be between that of the gorilla and the human.
Several frames of the Patterson film clearly image the foot. Figure 12 shows the foot above the ground, whereas Figure 13 shows the foot undergoing flexion, which demonstrates that the foot in the film is not a solid, inflexible prosthesis.
Note that separate toes are visible in the subject. Note the wide heel, minimal instep, and other key features of the foot including the position of the instep and ball below the big toe which closely resembles the plaster cast of the foot imprint taken by Titmus (Figure 14) [Krantz 1992]. This result links the plaster casts to the film subject for the first time, suggesting that the plaster cast could have been made contemporaneously by the film subject.
A series of feet from the great apes is shown in Figure 15 which illustrates the evolutionary changes from the gibbon to the human [Kogod 1993]. The notable adaptations include the shortening of the foot and toes and movement of the big toe upward as its grasping role decreases.
It is the authorís opinion that the subject's foot appears to be between that of the gorilla and the human.
Subjectively, the face and head appear to be gorilla-like in profile, but human-like when viewed from the front. Figure 16 shows a frontal brow, sagittal crest and an ear location that are similar in appearance to that of the mountain gorilla (Figure 17). The bulbous region around the mouth (mouth plate), is less pronounced in the mountain gorilla, and absent in the human. The visible texture on the face of the subject suggests less facial hair than the mountain gorilla, but more than a human. The jaw of the subject is below the shoulder line as in a gorilla, whereas the human jaw is above the shoulder line. When the subject is seen in motion, the shoulders are rotated when the subject turns its the head toward the camera. This is mechanically required as a result of
the jaw being below the shoulder line, as in the mountain gorilla.
Figures 18 and 19 show a front-view of the subject's face which bears some subjective resemblance to a human.
It is the author's opinion that the relationship between the frontal brow, nose and mouth of the subject is between that of the mountain gorilla and the human. It is also the author's opinion that the relationship of the nose to the frontal brow-mouth plate plane places the subject between the mountain gorilla and the human.
There are several other morphological similarities between the film subject and the mountain gorilla. In Figures 20 and 21 the configuration and appearance of the lower neck and lower back musculature of the subject is shown to be similar to the mountain gorilla. In Figures 22 and 23 the configuration and appearance of the side-torso musculature is also shown to be similar to the mountain gorilla.
There are several ways the subject may be differentiated from traditional forgeries: non-uniform hair texture, non-uniform coloration and non-uniformity of hair length. Examples of sophisticated costumes include those fabricated for the motion pictures 2001 A Space Odyssey, Gorillas in the Mist, Congo, the Planet of the Apes series as well as Eugene OíNeilís stage drama "The Great Apes". Peter Elliott, its lead, is shown in costume in Figure 24.
Most costumes have uniform hair length and little variation in color. Less sophisticated costumes have baggy, shapeless bodies with little or no detectable musculature, non-visible buttocks, and inflexible hands and feet. The appearance and sophistication of musculature as seen in the Patterson film has not yet been reproduced in costumes in the entertainment industry.
Figure 25 shows the breasts of a gorilla and the subject.
The kinematics section investigates the motion of the subject's knee as compared to that of a human and illustrates one of the muscle groups seen in motion in the Patterson-Gimlin film.
Knee Kinematics Ė Knee Delta
People who have viewed the Patterson film have commented on whether the subject's walk looks like a human walk or not. The opinions have run the gamut from statements to the effect that "it walks exactly like a human", to "it walks nothing like a human".
These opinions have been rendered from a mixed set of film replay speeds, which is methodologically faulty because the recording speed is not known. Thus, these opinions have been purely subjective. Joint range-of-motion between the film subject and a human can be compared. While the elbow and wrist are two potential candidates, the knee was chosen because it may be the joint most differentiated from human motion parameters.
Preliminary objective gait data has been extracted from the film by digitizing the presumed positions of the hip joint, knee joint and ankle. First, the camera motion introduced by the cameraman was removed by extracting a 640 x 400 pixel subset of the 2,656 x 1,912 original relative to the hip of the subject.
The joints were digitized on-line on the computer using "play" an interactive forensic frame editing program (Figure 26) [Photek 1996].
These positions of the joints were estimated by observing the relationship of the surface deformation. For example, the hip joint was estimated by looking at the relationship of the torso to the upper leg and selecting the position that closely approximately the intersection of the medial axes. This type of estimation can be inaccurate and result in noisy data. At least four types of error can contribute to the noise including the:
The data extracted was not corrected for this oblique angle so as not to introduce additional noise into the data. If the digitized joint positions were corrected for this, the recorded angles would increase. The interior angle of the knee was computed as the difference between the angles described by the hip-knee and the knee-ankle line segments (Figure 27):
The knee theta for a human subject is shown in Figure 29 [Winters 1990]. The graphed data for the subject knee theta is a shown in Figure 30.
A cycle for the human knee contains two distinct phases, a swing phase and a weight transfer phase. The knee theta of the film subject shows a more gradual transfer of weight rather than a separate phase. Assuming the subject knee delta is not overwhelmed with noise, the film subject is not employing typical human locomotion.
What is important is whether a human can replicate the knee delta observed in the film. Could a 7í 3½:" tall human maintain the stable head position (relative to the ground) with the continuous forward motion observed in the film while emulating a stride length of 93". Preliminary experiments by the author suggest that while the stride can be duplicated, the continuous forward motion seen in the film can not. The knee delta data extraction must be repeated to verify the extracted graph, and the proposed experiment should be performed.
Moving Muscle Groups
In several places in the Patterson film, groups of muscles in motion can be seen, in the arms, back and legs. One example shown in Figure 28 is the equivalent of the quadricep muscle in the human which is seen expanding while it absorbs the weight of the subject. Also seen in this sequence is a structure similar to a knee cap, the shape of which changes like a human knee. This is particularly difficult to forge because of the need for surface conforming material. Surface plasticity in the side torso is seen near frame 352. This requires not only a conforming material, but a material with independent x and y plasticity to avoid detectable material folds.
To date the Patterson film has defied explanation, and it continues to do so after three years of rigorous forensic examination. If the film is a forgery it is the opinion of the author that it required substantial capital, which probably was not available to Patterson and sophisticated knowledge of primate anatomy and materials science.
If the Patterson film is of an uncataloged animal, its subject bears a strong resemblance to the mountain gorilla. Most of the comparative morphological analysis place the film subject either similar to the mountain gorilla or between a gorilla and a human.
The knee kinematics of the film subject do not appear to be normal for a human, but might possibly be simulated by a 7í 3½" human. It may be possible to conclusively demonstrate that a human cannot reproduce this motion with proper experimental design. It is the author's opinion that a scientist whose discipline is biomechanics might be able to prove that the subject in the film is or is not human from the kinematics data, and in particular the knee delta.
Motions visible in the film that are difficult to replicate in a costume include the knee cap, the upper leg muscle motion, and the surface plasticity. Also present is non-uniform hair length, non-uniform hair color, and well-defined body shape.
If only a single dimension of similarity was seen in the Patterson film it could easily be dismissed as a forgery. In the opinion of the author, the Patterson film is remarkable in the simultaneous presence of all of the dimensions listed above.
Despite three years of rigorous examination by the author, the Patterson film can not be demonstrated to be a forgery at this time.
This abridged version of the NASI Report, & Copyright 1998, North American Science Institute
Portions of this website are reprinted and sometimes edited to fit the standards of this website under the Fair Use Doctrine of International Copyright Law