The Bike With the Mike

A Critique of Myers’ “Epipolar Geometric Analysis of Amateur Films Related to the Acoustical Evidence”

by Don Thomas , 3 Mar 2008

In 2004 an ABC News documentary entitled “Beyond Conspiracy” claimed that it had “concrete evidence” refuting the acoustical study presented to Congress in 1979. The latter study determined that the assassination gunfire was recorded over a microphone on a police motorcycle with the president’s motorcade. To be valid the acoustical findings require that a motorcycle with an open microphone was near the intersection of Houston Street and Elm Street at the time of the shooting (Fig. 1). Thus, a study of the films and newsreels provides an independent test of the acoustical hypothesis. The ABC documentary claimed that a computerized reconstruction based on the “Leon Zapruder film,” which it described as “the only film of the president’s murder,” proved that there was no motorcycle at the specified locations during the shooting.

Figure 1. The five positions of test microphones that recorded test shots that matched to impulse patterns on the DPD recording
Fig. 1. The five positions of test microphones that recorded test shots that matched to impulse patterns on the DPD recording. The photographic record was searched to see if a motorcycle was at these successive positions during the shooting.

The analysis was not actually presented in the documentary, however. It was not until 2007 that the evidence behind the documentary’s claim was made available. The graphic essay by Dale Myers, entitled “Epipolar Geometric Analysis of Amateur Films Related to Acoustics Evidence in the John F. Kennedy Assassination” (www.jfkfiles.com/jfk/html/acoustics.htm), presents the results of a computer-assisted effort to reconstruct the movement of the presidential motorcade, including the motorcycle escort, during the JFK assassination. In 1978 the HSCA had identified a motorcycle ridden by police officer H.B. McLain as the unit with the open microphone. McLain is seen in a film by Robert Hughes turning from Main onto Houston Street in Dealey Plaza a few seconds before the shooting. His last filmed position is 174 ft away from positon “1” in Fig. 1. From his reconstruction, Myers concludes that the last frame showing Hughes is only one-half second before the first shot, and thus McLain could not have been at the specified location at the time of the first shot without traveling at implausible speeds, in excess of 400 mph.

In this critique I will show that the study by Myers contains serious errors that invalidate his conclusions. In brief, Myers uses the Zapruder film as a chronometer for the assassination, but mis-synchronizes the acoustical evidence by placing the shooting one second earlier than it was. Also, Myers’ reconstruction of the motorcade procession is based on the positions of the vehicles that are seen first in the Hughes film and then later in the Zapruder film. But Myers misplaces the vehicles as seen in the Hughes film because he misplaced the position of Mr. Hughes. Myers then compounded the error in placement by over-estimating the speed of the vehicles. Thus, in his reconstruction the vehicles are further north and traveling faster than they really were. This combination of errors results in a time line which is about 3-1/2 seconds shorter than it would be without the errors. Removing those errors leaves McLain with approximately 4 seconds to cover the 174 ft (requiring a speed of about 25 mph) to reach the specified location. However, Myers’ reconstruction also includes an untestable assumption concerning the movement of the procession. Myers assumes a steady even pace, contrary to the memories of the police officers in the escort who recall a stop and go movement. If the latter rather than the former is true, then McLain had even more time to reach the acoustically required location for the first shot.

Caveat lector: the reader is advised that the essay by Myers contains a personal attack on the author of this critique for his support of the acoustical evidence and likewise, that Myers is the author of a previous book on the Kennedy assassination which argued against conspiracy theories. I have delayed responding to Myers in the hopes that someone more disinterested would point out the problems with his analysis, but so far only one person has done so, and on procedural rather than evidentiary grounds, and this anonymously to a discussion group which included myself and Myers. I find that Myers’ characterizations of the evidence in general and of my views in particular are actually caricatures; i.e., strawman arguments. Suffice it to say that anyone relying on Myers’ version of the evidence, and my analysis, does so at their own risk. In this essay I will confine my remarks to the analysis of the motorcade films relevant to the acoustical evidence.

The illustrations in Myers' study are professional quality. Unfortunately, they illustrate his opinions and conclusions better than the evidence itself. As pointed out by the previous critic, a major problem with the computer study is that Myers fails to provide the reader with the information that was put into the computer. This makes it quite difficult for a critical reviewer to judge the validity of the conclusions of the study; meaning that if it was garbage in, then it will be garbage out.

One measure of the reliability of Myers' analysis is his use of the term “Epipolar Geometric” in the title. Fig. 2 below illustrates the parameters of epipolar geometry.

Figure 2. Epipolar geometry: a method used in stereo imaging
Fig. 2. Epipolar geometry: a method used in stereo imaging.

In Fig. 2, P is the point of interest. OL and OR are two (left and right) optical device positions (actually the cameras' focal points). The points pL and pR are the images of P as projected on to an image plane, e.g., a photograph in its position at the instant it was exposed. EL and ER are the “epipoles.” The epipoles are the images of the opposing cameras projected on to the respective image planes. The epipolar line is the line defined by the two epipoles. The epipolar line is the baseline necessary for the triangulation process used in epipolar geometry to fix the position of object P. The problem is that the key analysis used by Myers involves images taken by Abraham Zapruder and by Robert Hughes. At no time does Zapruder appear in any of the Hughes frames nor at any time does Hughes appear in Zapruder’s film. Hence, there are no epipoles or epipolar lines or epipolar geometry involved in Myers' key analysis, or any other analysis as far as I can tell. Rather, Myers uses line of sight to estimate the positions of the cars in the motorcade, but without the exactitude implied by the use of “epipolar geometry.”

Myers, while doubtlessly skilled in the art of illustration, does not provide an accurate presentation of the acoustics, and certainly does not correctly relate my analysis of the acoustical evidence. This is critical because in order for the filmed evidence to constitute a fair test of the acoustical evidence the latter has to be posited correctly and Myers fails to do so. Myers attempts to further undercut my analysis by implying that it differs from that of the HSCA’s acoustical experts.

“While Thomas supported the HSCA conclusion that the assassination of President Kennedy was the result of a conspiracy, his own acoustic findings were at odds with virtually evey acoustic conclusion reached by the HSCA about the shooting in Dealey Plaza.”

Table II taken from HSCA Report, Volume 8, page 101Table II taken from HSCA Report, Volume 8, page 101.

Setting aside the hyperbole, Myers specifically contends that the HSCA’s acoustical experts found four suspect sounds on the police tapes to match with their test shots (I report five) and that the first synchronizes to Zapruder frame 160 (as opposed to my synchronization of the first shot with Z-175). But Myers’ contention is wrong on both counts. My numbers come directly from the BBN report.

The correct numbers are found in the BBN report in their Table II on page 101 of HSCA vol 8, reproduced herein. In column 1 of this table one finds the suspect patterns listed by their time history, that is, by their onset relative to the beginning of the open microphone segment. In the fifth column are the correlation coefficients which are the matching scores. The table legend indicates that a score of 0.5 or higher is statistically significant and thus, an objective match. Referring to this table one finds that five of the patterns have a score of 0.6 or higher, these being the ones designated at: 137.70, 139.27, 140.32, 145.15 and 145.61 sec. That is five, not four, matching sound patterns were found and reported by the HSCA acoustical experts in Table II and as shown in my Fig. 1 above which plots the microphone positions shown in the second column of the table.

Importantly, the time-history in Table II is the playback time from a tape recording made by the Dallas police. As mentioned in the footnote in Table II, the recording process was about 5% too slow, requiring a simple correction to adjust to real time. A second correction is also cited in the text of the BBN report necessitated by the subsequent detailed analysis by a second laboratory (Weiss & Aschkenasy) which demonstrated that the pattern identified by BBN at 145.15 sec included impulses that preceded slightly those recognized by BBN, placing the onset of this pattern 270 msec (=0.27 seconds) earlier. When these corrections are applied to the times in Table II the four time intervals among the five putative gunshots are 1.65, 1.1, 4.8 and 0.7 sec. Consequently, the time of the first acoustically identified shot synchronizes to Z-frame 175, not 160 as claimed by Myers. The math is simple, (4.8+1.1+1.65 = 7.55 sec), (7.55 sec x 18.3 fps = 138 Z-frames), and because the fourth putative shot came from the grassy knoll and the fatal shot is at Z-313, thus, 313-138 = frame 175. These intervals come directly from the BBN report. Alternatively, one can arrive at the same answer using the numbers in Table II. The grassy knoll shot is the pattern at 145.15 sec while the first shot is at 137.7 sec, thus, 145.15 – 137.7 = 7.45 sec. Expanding by 5% to correct from the playback time to real time, 7.45 x 1.05 = 7.82 sec. The WA analysis found that the initial pattern in the grassy knoll shot places the muzzle blast 270 msec earlier = 0.27 sec. Thus, 7.82 – 0.27 = 7.55 sec, and 7.55 x 18.3 fps = 138 Z-frames, and 313-138 = frame 175.

The values reported in my essays and lectures come directly from the data in the BBN report and thus are entirely in accord with the HSCA acoustical evidence. The discrepancy is not between my analysis and the acoustical evidence, but between the acoustical evidence and the way it was subsequently manipulated using non-acoustical evidence. The acoustical experts were told falsely that the murder weapon could be cycled in 1.6 sec, an error repeated by Myers. This erroneous, non-acoustical evidence, along with some tortured logic, led the HSCA to discard one of the acoustical matches as a false positive. This changed the number of shots from five to four, but this does not affect the timing of the first shot. In my analysis I reinstated the matching evidence because the grounds for declaring the match at 140.3 sec as a false positive were invalid. More importantly for this discussion, when the acoustical experts were asked to collaborate with an expert on the Zapruder film (Robert Groden) to attempt a synchronization of the acoustical evidence with the video evidence, they were told that any gunshot from the grassy knoll must have missed. But because the acoustical experts were not persuaded by that contention, perhaps aware that the forensic pathology panel had actually split on this issue, they produced two synchronizations (8 HSCA 107-108): one in which the fatal shot was set to a book depository shot, and one where the fatal shot was set to the grassy knoll shot. By using the erroneous of the two synchronizations, the one which assumes that the grassy knoll shot missed, Myers lops nearly a second off the time history, and of course, this helps make it appear that the motorcycle could not have been in the specified position.

In spite of some claims to the contrary (i.e. Gregg Jaynes) no films depict the positions where the acoustics places the open microphone at the time of the shots. However, one motorcycle, that ridden by H.B. McLain, was in a position in the motorcade both before and after the shooting, such that he might have been in the acoustically required locations. Those required locations, the test microphone positions where the suspect patterns matched to test shots, were distributed in the vicinity of the intersection of Elm and Houston Streets (Fig. 1). By referring to the Zapruder film, specifically to the sequence surrounding Z-175, the time of the first shot, one finds that the acoustically predicted location for the motorcycle is out of the camera’s view, blocked by structures and the crowd at the intersection. The motorcade vehicle in the immediate vicinity of the predicted location was the mayor’s car (sixth in the motorcade). Frustratingly, of the first nine cars in the motorcade the only one which is never visible in the Zapruder film happens to be the mayor’s car. But because of Myers’ faulty analysis this significant fact never surfaces. Between Z-frames 200 and 250, Zapruder panned his camera to the right providing a glimpse of the motorcade on Houston Street including views of the 7th, 8th, and 9th cars. Many researchers have searched these frames for any indication of the motorcycle and none has been found. This negative evidence means that at the time of the shooting, McLain’s motorcycle was either in exactly the right place predicted by the acoustical evidence (next to car-6) or much further back (next to car-10), the latter being Myer’s conclusion. This dichotomy, and its implications, was not made clear by Myers.

The importance of the mayor’s car to the issue can be seen in the analysis by Greg Jaynes. Jaynes’ flawed analysis appears at the http://mcadams.posc.mu.edu/Jaynes/. Jaynes claimed that McLain should have appeared in the uncropped version of the Altgens photo (synchronous with Z-255), and cites the absence as proof that McLain was not in the acoustically required position. But, the mayor’s car is also not visible in the Altgens photo. Jaynes should have provided a map showing, a) Altgens’ cameras scope of view and, b) the acoustically required location of the microphone. Such a map is here provided (Fig. 4) which shows that the required microphone location does not fall within the field of view in the uncropped Altgens photo.

Figure 3. A photo by James Altgens does not show McLain’s motorcycle
Fig. 3. A photo by James Altgens does not show McLain’s motorcycle. Courtesy of the Sixth Floor Museum at Dealey Plaza.

Figure 4. Altgens was at position A
Fig. 4. Altgens was at position A, the right side of his camera view intercepts the Dal—Tex building between the entrance and the corner. The acoustically required location for the motorcycle is at “m” a position between the 5th and 6th cars of the motorcade, 3.2 sec before the fatal shot.

Jaynes also claimed that the Hughes frame 20 showing McLain at the intersection of Main and Houston and the SS car at the intersection of Houston and Elm was equivalent to Z-frame 160. To refute that estimate Gary Mack argued that the LBJ car was already farther into the intersection at Z-133 than it is at H-20 and that therefore H-20 had to have been sometime earlier than Z-133, estimating that it was the equivalent of around Z-100. Myers’ new analysis reemphasizes that Jaynes was wrong, inasmuch as in Myers analysis H-20 is the equivalent of Z-138. Yet, Myers cites Jaynes as among the “credible analysts” whose work contradicts those of Mack and myself.

In order to constitute a serious test of the acoustical hypothesis, such critics as Jaynes and Myers were obligated to present that hypothesis in an accurate and cogent manner and they never do. The acoustical hypothesis is here presented (Fig. 5) so that it can be tested against the films.

Figure 5. The large circles represent the successive locations where the acoustics predicts the motorcycle would have been at the time of each successive shot
Fig. 5. The large circles represent the successive locations where the acoustics predicts the motorcycle would have been at the time of each successive shot. Location (b) is where the sonar analysis pinpointed the motorcycle at the time of the grassy knoll shot. Location (a) is McLain’s actual position as seen at Hughes frame 20. Location (c) is where McLain is seen in the Bond photo as he passed officer Hargis after the shooting. The open dots are the hypothesized location of the motorcycle at one second increments between points (a) and (c). Built into the hypothesis is the fact that the time between the first and last shot on the police recording was 8.3 sec and the distance from the first to the last circle, about 135 ft requires an average speed of 11-12 mph and thus there are nine open dots (8 spaces) covering that sequence. Also, just two seconds before the first putative shot the motor noise decreased sharply (75%), indicating that the officer had released the accelerator, and idled for the next 40 sec before the motor is heard to rev-up again. In accord with this evidence the motorcycle is depicted as slowing through the sequence beginning as it approached the first circle. The speed of the motorcycle after point (a) was fit to Hughes account in which he wrote that the first shot occurred about 5 sec after he stopped filming (H-20 occurs about 3 sec before the end of the film). This hypothesis was used to plot the predicted position of the motorcycle at Z-255, 3.2 seconds before the fatal shot and equivalent in time to the Altgens photo.

In the last photographic view of the subject motorcycle, in a film by Robert Hughes, officer McLain was just turning on to Houston Street at the intersection with Main Street alongside the tenth car in the motorcade. Myers calculates that the motorcycle is at a distance of 174.3 ft from the first acoustically required position. Myers concludes from his computer reconstruction of the motorcade that the last Hughes frame showing McLain synchronizes with Zapruder frame 150. If so, then McLain has only about a half second (by Myers' math) to go nearly a city block to be in the required location, or about a second and a half if one uses Z-175 as the time of the first shot.

However, the synchronization between the Zapruder film and the Hughes film depends on two parameters: the position of the motorcade vehicles on Houston street as they appear in the respective films, and the speed of those vehicles over the intervening distance. In both cases Myers’ estimates are off. The estimated position of the cars is based on alignments of the cars relative to fixed objects in the camera’s view and the position of the camera itself, in this case, Hughes camera.

Figure 6. Hughes’ actual location compared to placement by Myers
Fig. 6. Hughes’ actual location compared to placement by Myers.

But Myers has badly misplaced Hughes. Myers places Hughes 15.5 ft west of the center line of Houston Street and 8.8 ft south of the center line of Main Street. The first value is accurate but the second is not. It places Hughes in line with the traffic stripe separating the inner and middle east bound lanes of Main street (and Myers’ illustrations e.g. exhibit 83, show this). Actually Hughes was in line with the traffic stripe separating the middle from the outer lanes, which can be seen in the segment of the Hughes film of the oncoming motorcade on Main Street. This error results in a displacement of approx. 11 ft. from where Myers places him (see Fig. 6 for orientation).

Assuming that Myers’ line of sight estimates of the cars relative to inanimate objects is done correctly then the 11 ft northward displacement of Hughes must cause the cars to be similarly displaced from actual by about the same distance. If so it lops another 1-1/2 sec off the timeline available to McLain to arrive at the required position, because that is how long it would take the motorcade to move up 11 ft.

But Myers’ estimates of the vehicle speeds are also suspect. He calculates that the motorcade vehicles were moving at relatively steady speeds of around 8-10 mph on Houston Street. Most importantly (on p. 49) he calculates the speed of the Secret Service car (the fifth in the motorcade) at 8.7 mph. But the speed of the car is only 6.5 mph. In the Zapruder film the front bumper of the SS Car comes in to line with a lady dressed in red standing on the corner underneath the signal light at Z-144. The corresponding rear bumper reaches the lady in red 34 frames later at Z-178. That is, at 18.3 frames per sec it takes the SS car 1.86 sec to go one car length. With a length of 215 inches, about 17.9 ft, it has a speed of 9.6 ft. per sec, which equals 6.5 mph.

Through this combination of errors: setting the first shot earlier than it was, displacing the vehicles farther northward than they were, and having them travel faster than they were, Myers lops about 3-1/2 sec off the timeline of events. A precise estimate of the exact amount of time that McLain has between his last filmed position and the time of the first shot may not be attainable because it depends very much on how reliably one can extrapolate the speed of the motorcade. Because at these speeds it can take about two seconds to move one car length, it is clear that over distances of several car lengths any extrapolation is bound to have an ambiguity of at least a second or two. In contrast to Myers’ reconstruction which has the cars moving at a steady 9 mph on Houston Street, I have argued that the motorcade is moving in accordion fashion, slower through the turns than in the straight-aways, and moreover are traveling in a slow and surge mode. Jim Bowles (1993) interviewed the officers who were on Houston Street and gives this account on p. 325.

“McLain and others estimated the motorcade speed on Houston as no more than 2 or 3 mph, when they were moving.” And, “…vehicles which had just made their way slowly off Main onto Houston were now slowing and stopping again.” And, “Other vehicles, were barely moving if not stopped.”

Thus any instantaneous measurement of speed of the cars in motion is only a snapshot which is not applicable to the entire interval between the end of the Hughes film and the later view of the cars in the Zapruder film. Bowles further argued that the cars were not “coaches in a train.” But the “coaches in a train” model is exactly what Myer’s uses for his reconstruction and then rebukes me for suggesting that there could be ambiguity in these extrapolations.

To put the analysis in proper perspective consider these time estimates. Myers calculates McLain’s speed as he makes the turn from Main on to Houston at 14.7 mph. McLain is last seen at 174 ft from the acoustically required position. At a speed of 10 mph, the equivalent of 15 ft per sec, it would take him about 11 seconds to cover the distance. At 20 mph (=30 ft per sec) he could cover the distance in 5.8 sec. Thus at speeds of 21-29 mph, the distance could be covered in just 3-5 sec. Just from the obvious errors in Myers’ analysis one can see that McLain had the necessary time.

Although Myers claims that his measurements are consistent among the other films, the other films do not show the same vehicles in the same positions. And his conclusions based on the other films are also questionable. An important example is the Dorman film. The Dorman film shows a police motorcycle approaching the intersection of Elm and Houston at a time approximately 3-6 sec after the fatal shot. The officer could be either H.B. McLain or Jimmy Courson. A previous critic has already addressed some of the problems in Myers' identification of the officer as McLain and I would reiterate that Myers does not provide any evidence that would distinguish McLain from Courson. For

Figure 7. Frame from Skaggs film
Fig. 7. Frame from Skaggs film, courtesy Sixth Floor Museum at Dealey Plaza, Skaggs collection.

example, the cop in Dorman has a ticket book visible through the windshield and Myers cited this feature as an identifying character for officer McLain. But Figure 7 shows another motorcade officer (Chaney) with his ticket book in the same position; thus having a ticketbook in this position was certainly not unique to McLain, and there is no information on where Courson kept his ticketbook.

Myers attempts to refute my argument that the timing of the appearance of the officer in Dorman is consistent with Courson’s account that he saw Mrs. Kennedy and SS Agent Hill on the trunk of the limousine when he rounded the corner of Elm and Houston. Myers claims that this officer does not reach the intersection in time to see that event. But this claim is not supported by his synchronization. Mrs. Kennedy was on the trunk in the Zapruder film between frames 360 to 405. Hill stepped on to the rear bumper of the limo

Figure 8. A frame from the Darnell film
Fig. 8. A frame from the Darnell film showing all 4 mid-motorcade motorcycles.

during that sequence inducing her to return to her seat. By Myers' own account (on P. 103) the motorcycle cop appears in the last 40 frames of the Dorman film which he calculates to be equivalent to Z-frames 367-411. During those last frames the cop is at the intersection well beyond the wall which was the only inanimate obstruction to his view of the limousine. Thus, it is not at all clear how Myers can assert that the cop was not in a position to have seen Mrs. Kennedy and Agent Hill on the trunk as Courson said he did.

Regardless of their memories, if the officer in Dorman is Courson, then McLain has to be in exactly the position required by the acoustical evidence. As mentioned previously, the acoustical evidence requires that the microphone was in the vicinity of the mayor’s car (6th in the motorcade) at the time of the gunfire. Actually, because the acoustical evidence requires the bike with the mike to have an average speed of 11-12 mph during the shooting, the motorcycle is expected to pass the mayor’s car as it rounds the intersection. In the Zapruder film, the only car not seen at any time between the president’s limousine and the tenth car in the motorcade, happens to be the mayor's car. Similarly, in the Dorman film, one can see in one sequence, the cars ahead of the SS car (5th in line) and those behind the mayor’s car (6th in line) on Elm street, but not the space between the mayor’s car and the SS car, which happens to be the position indicated by the acoustical evidence for the location of the open microphone. Hence if the officer in Dorman is Courson, on the motorcycle next behind McLain, the only place McLain could have been is exactly where the acoustical evidence requires (Fig. 9).

Figure 9. The yellow block represents the mayor’s car in its position in the field of view at the start of the sequence of the Dorman film where the cop appears
Fig. 9. The yellow block represents the mayor’s car in its position in the field of view at the start of the sequence of the Dorman film where the cop appears. The green circle depicts the acoustically required position of the bike with the mike at the same moment, about 1.5 seconds before the fatal shot.

The negative evidence thus provides circumstantial support for the acoustical evidence in that while the acoustically required positions at the exact moment of the gunfire are not shown in any films, most every other position is, and McLain is not in the wrong position. The dichotomy is that McLain was either in exactly the right place, near the mayor’s car, or he was way back, no closer than the tenth car, which is where Myers believes he was. The reality is that until some new films or photographs surface, the presently available materials are consistent with either contention.

© Mary Ferrell Foundation. All Rights Reserved. |Press Room |MFF Policies |Contact Us |Site Map