3. THE PANNING ERROR--BLUR ANALYSIS OF THE ZAPRUDER FILM*

    (a) Issues

  1. (a) Can any of the panning errors indicated as blurs or "jiggles" on the Zapruder film reasonably be attributed to the photographer's involuntary startle reaction to the sound of gunshots

  2. (b) If so, is it possible to determine from such panning error blurs the number, and to estimate the timing, of the shots that were fired at the Presidential limousine?

    (b) Procedures and materials employed

  3. Originally, a blur study of the Zapruder film was to involve an analysis of not only the Zapruder, but also the Nix and Muchmore films, to determine whether startle reactions by the photographers were present and simultaneous for each photographer. Nevertheless, neither the Nix nor Muchmore films included any extensive footage prior to the time of the head shot. Thus, only the Zapruder film, which showed the entire motorcade scene immediately prior to and during the assassination, was subjected to this analysis.**

  4. The measurement of blur, or jiggle, essentially involved, to measure of Zapruder's error in panning his camera as he filmed the Presidential limousine. Two sets of measurements were made independently by Photographic Evidence Panel members William Hartmann and Frank Scott. Hartmann measured the length of images of small highlights on the Presidential car in each frame; these were generally small and round on the sharpest frame but highly elongated on frames that were blurred by camera motion during the time that the shutter was open. The amount of elongation was measured to determine the panning error. (See fig. II-l.)
    *This section prepared under the direction of William K. Hartmann. For Dr. Hartmann's related public hearing testimony, 9/11/78, see HSCA-JFK Hearings, vol. II, pp. 4-16.
    *In addition, the Nix and Muchmore films were taken from a distance of ,about 2.7 and 2.1 times, respectively, further away from President Kennedy than the Zapruder film.

    FIGURE II-1

  5. Scott followed background details from frame to frame. These details indicated a direction in which the camera pointed during each frame. The sequence of these camera point directions would have defined a smooth, relatively straight line had the camera panned in a perfectly uniform manner (see fig. II-2, JFK exhibit F-371); discrepancies were revealed by erratic spacing of camera pointing directions (see fig. II-3, JFK exhibits F-372-373).

    FIGURE II-2.--Illustration of technique used by Frank Scott. If Zapruder had made no panning errors, the background points froIn frame to frame would have been plotted as illustrated.

    FIGURE II-3 - JFK exhibits F-372, F-373

    These were transformed into quantitative measurements by a vector subtraction process: A 20-frame running average of the rate of angular motion was used to predict the point where the camera should be aimed in frame based on the place where it was directed in frame n-1. A subtraction was then made between the n-1 to n vector which should have appeared and the vector that actually did appear. This vector difference was then measured to determine the panning error between frames.

  6. The results of these measurements were then plotted to illustrate visually the times of greatest panning error in a manner similar

    Figure 11-1. (See fig. 114.) In both. Figure II-l and Figure II-4, a threshold line was then drawn horizontally across the graph to separate visually the areas of greatest (i.e., unusual) blur from the "noise" or the normal panning errors that occurred throughout the filming simply because the camera was hand held. Any such threshold line, of course, must to some extent be arbitrarily placed. The extenta blur that resulted from an unusual external stimulus (e.g., a gunshot) may not, therefore, always be precisely delineated by the line. Zapruder's true startle reactions may have begun a frame or two earlier. or later. than indicated by the placement of the threshold line that was chosen, but blurs above the threshold line are well above the average.

    FIGURE II-4

  7. Following our measurement of blur by these two different methods of the panning errors, the results of Luis Alvarez' study, reported in the American Journal of Physics, vol. 44, p. 813 (1976), were also reviewed. Instead of measuring the blur itself, Dr. Alvarez measured the difference in blur between frames and calculated the rate of change in blur, thus providing a sensitive record of any unusually erratic movement by Zapruder. All three sets of measurements were plotted on a single graph and used in the subsequent analysis. (See fig. II-5. )

    FIGURE II-5

  8. To assist in the analysis, a time-scale in seconds was added to figures II-l, 4 and 5. The zero point in the time-scale was chosen to coincide with the moment when the trigger was pulled on the fatal head shot. This was estimated to have been at frame 310, based on estimates of the average running speed of Zapruder's camera in conjunction with other scientific evidence.

  9. Specifically, Zapruder reported that his camera was fully wound when he started filming the motorcade as it turned onto Elm Street. In 1964, the FBI tested the camera and found that during the first 30 seconds of its operation (after being fully wound) Zapruder's camera ran at between 18.0 and 18.5 Frames per second, with the average estimated to be 18.3. Note that the range 18.0 to 18.5, indicates a range of 3 percent uncertainty in all time intervals measured from the Zapruder film. (25)

  10. Fragmentary material from the President's head is seen flying upward and outward in frames 313 and 314. The fragments are already airborne and in motion in frame 313. Extrapolation backwards indicates that the explosion began in frame 312 rather than 313, since this would be the frame nearest to the moment when the fragments left the head. Other scientific evidence, such as wound measurements, and trajectory analysis, indicated that the fatal head shot was fired from a Mannlicher-Carcano rifle located in the southeast corner window of the sixth floor of the Texas School Book Depository. (26) The distance from that window to the limousine at frame 312 is approximately 265 feet.(27) Since a Mannlicher-Carcano bullet travels at approximately 2000 feet per second, (28) the bullet flight time would have been 0.13 second, or the passage of approximately 2.4 frames in Zapruder's camera. Subtracting these two frames from frame 312, it is apparent that the fatal shot was fired at approximately frame 310.

  11. Using frame 310 as the time of the trigger pull, it is possible to determine that the sound from that shot would have reached Zapruder at frame 313-314: Zapruder was standing approximately 270 feet from the Texas School Book Depository window, sound travels slightly more than 1,100 feet per second. (29) and the sound of the shock wave from the bullet itself reached Zapruder slightly before the sound of the muzzle blast from the window. Finally, the pattern of jiggles that was discovered was compared with the results of the committee's acoustics study. The correlation between the jiggle analysis and the acoustics test is treated separately in an addendum to this report.

    (e) Conclusion

  12. 1. Two pronounced series of jiggles or blurs on the Zapruder film, one during frames 189-197, a time when other visual evidence suggests that President, Kennedy was first shot, (30) and another during the following impact of the head shot, may reasonably be attributed to the photographer's startle reaction to the sound of gunshots.

  13. 2. The timing of the shots associated with these two sets of blurs can reasonably be determined to be approximately 6 to 7 seconds apart.

  14. 3. Other blurs which might relate to gunshots, appear on the film both between these two sets and elsewhere. Due to the absence of other visual evidence associated with these blurs, the Panel made no finding as to their cause.

    (d) Analysis

  15. As noted above, the sound from the fatal head shot would have reached Zapruder at frame 313-314. Frame 313 is also the moment when the head explosion became visible to Zapruder. The largest blurs or jiggles on the three independent data sets of Hartmann, Scott, and Alvarez occur between frames 312 and 334. Because some of the jiggles are minor and appear on one or two of the data sets but not all three, it is difficult to determine exactly when the reaction in question actually started. The Hartmann data set shows a very large blur in frame 318 with smaller ones at 313 and 314. (See fig. II-1.) The Scott measurements show several smaller jiggles between 313 and about 39A. (See fig. 11-4). The Alvarez data show the largest acceleration of camera motion at 315, but with a cluster of motions from 312 to 322. (See fig. 11-5.) Empirically, it appears, therefore, that Zapruder's startle reaction apparently occurred about 1 to 5 frames, or about 0.05 to 0.3 seconds after the sound reached him.

  16. In fact, the Panel found some empirical corroboration for this conclusion. Startile reaction times in response to the sound of gunshots were measured in the experimental work of Landis and Hunt in 1939. (31) For "head movement:' "movement of neck muscles," and "initiation of arm movement," Landis and Hunt found that the reaction time was 0.06 to 0.2 second (i.e., 1.1 to 3.7 frames). (32) Thus, these much earlier experimental findings support the conclusion that the film actually records Zapruder's startle reaction to the fatal shot immediately after Zapruder heard the sound of the shot and saw the head impact through his viewfinder.

  17. In all three data sets, the second longest and second greatest (in terms of magnitude of blur or jiggle) disturbance in camera panning motion occurs between frames 189 and 197. Since our objective was to estimate the time the shots were fired, the blurred frames of most interest are those from the jiggle's beginning to its peak rather than to its decline. For the various data sets, the time from the first strong increase in blur to the maximum blur or jiggle is as follows: Hartmann, 191-197; Scott, 193-194; and Alvarez, 189-195. (See figs. II-l. II-4, and II-5.)

  18. Assuming that a shot from the sixth floor Texas School Book Depository window caused this reaction (a distance at this point in time of approximately 165 feet from the limousine) (33), and that Zapruder's reaction was again almost immediate (within 1 to 4 frames after hearing the shot), it is possible to calculate backward (adding sound travel time to Zapruder of 4 frames, to reaction time of 1 to 4 frames), and determine that the shot may have been fired between frames 181 and 192, and impacted in the limousine between frames 182 and 193. This conclusion is reinforced somewhat by the Photographic Evidence Panel's visual observation of the Zapruder film which reflected a reaction by President Kennedy to some severe external stimulus by frame 207 when the President disappears behind a sign frame. (34) Assuming a uniform reaction time in both cases by Zapruder, and that both shots originated in the same location (the sixth floor window), the trigger pull on this shot would have preceded that of the fatal head shot by approximately 6.3 to 6.9 seconds (minimum, 313--197 over 18.5 equals 6.3; maximum 313--189 over 18.0 equals 6.9) .*

  19. The blur or jiggle results have been examined for other blur episodes that possibly correlate with additional gunshots. The three next largest episodes of blur are listed in table 1, in which the largest and second largest blur episodes discussed above are designated A and B. The third, fourth and fifth largest blur episodes, which are similar to one another in magnitude, are listed respectively as C, D, and E.

  20. It is difficult to determine with certainty whether any of these represents an additional shot or allots. Blur episode C, detected by all three analysts, occurs at frames 220-228, just before movements of Governor Connally in which his cheeks suddenly puff out and his face contorts in a grimace, followed by two apparent outcries in which his mouth opens wide in what appears to be a shout of pain.

  21. Another shot could have caused blur episode D, at frames 158-160. It occurs much earlier in the motorcade than had been considered possible for a shot by either the Warren Commission or most Warren Commission critics. Nonetheless, this brief blur was detected by both Hartmann and Scott; Alvarez published no data for such an early part of the motorcade because he used the Warren Commission volumes which do not even reproduce Zapruder frames earlier than the mid-170's. The most interesting thing about this hypothetical shot is that Mrs. Kennedy and Governor Connally testified before the Warren Commission and Governor Connally testified before the select committee that they turned to their right when they heard the first shot, (35) and both are seen in the film beginning a turn to the right immediately after this hypothetical shot. This appears particularly striking in the case of Governor Connally, whose head turns from midleft to far right in less than half a second, beginning at frame 162.

  22. A fifth episode E possibly associated with a shot occurs at frames 290-293. Although it contains a very small blur detected both Hartmann and Scott as well as a more substantial blur in Alvarez data, the Panel found no visual indications of reactions to a shot by the limousine's occupants coinciding with this segment of blur in the film.

  23. Other jiggles or blurs were present in the photographic record. (See fig. II-5.) Without further data, however, the magnitude and duration of these jiggles, as those with characteristics of C.D. and E, are insufficient to warrant any conclusion concerning the number and timing of any additional gunshots.
    *Assuming a uniform reaction time, and a uniform distance of Zapruder from the shooter, it is possible to ignore delays caused by sound travel from the shooter to Zapruder and Zapruder's reaction time in calculating the spacing of the two shots, since any assumed value for such delays would be self-canceling in the calculation.
    TABLE 1.--BLURS IN ZAPRUDER FILM
    Relative magnitude of blur episodeDesignation of blur episodeShown byFrames showing blur onset (beginning to maximum)
    LargestA1Alvarez312-318
    Hartmann 313-318
    Scott313-314
    A2Alvarez330-334
    Hartmann331-332
    Scott331-333
    2d largest BAlvarez189-195
    Hartmann191-197
    Scott193-194
    3rd largest 1CAlvarez220-228
    Hartmann227
    Scott226-228
    4th largest 1DHartman158-1598
    Scott158-160
    5th largestEAlvarez291-2938
    Hartmann290-291
    Scott290-292
    1 About equal.

    ADDENDUM

    COMPARISON WITH RESULTS OF THE ACOUSTICS ANALYSIS

  24. The acoustics analysis suggests the possibility of four shots, the first, second, and fourth originating from the Texas School Book Depository and the third having been fired from the grassy knoll. (36) This appendix discusses the extent of agreement between the acoustic timings and the major blurs on the Zapruder film.

  25. A first step in comparing the acoustic results to the blur analysis would be to line up one of the shots indicated by the acoustics analysis with the known fatal hegel wound to the President. This cannot be shots No. 1 or No. 2 from the acoustics analysis because the President's reaction to the back wound at approximately frame 200 would then occur before any shooting. Therefore, the fatal head wound had to be shot No. 3 or No. 4. It is improbable that the fatal head shot was shot No. 3 because the acoustics analysis p]aces this shot as coming from the grassy knoll, while the medical and other scientific evidence indicates that the President's head was shot from the rear. (37) The fatal head shot, therefore, was probably the last shot identified in the acoustics analysis. It is thus a simple step to work backward in time from that final shot to derive the times that Zapruder heard shots based on the acoustics results. The acoustics report states that the time intervals between the four shots recorded on the Dallas Police Department tape are 1.57, 5.63, and 0.71 seconds respectively, (38) but that the Dallas Police Department recorder was running approximately 5 percent too slow. (39) Adjusting for this 5-percent error, the actual spacing of the shots to the nearest, one-tenth second would therefore be 1.6, 6.0, and 0.7 seconds.

  26. As noted earlier, the FBI tested Zapruder's camera in 1963. They found that its speed varied from 18.0 10 18.5 frames per second during the first 30 seconds of its operation (after being fully wound), and then ultimately decreased to 17.6 frames per second before it completed the next 30 seconds of its operation. Because Zapruder said his camera was fully wound when he began filming the motorcade, it, may be assumed that the camera ran between 18 and 18.5 frames per second during the approximately 8 seconds of the assassination. Using these parameters fox film speed, and allowing for sound and bullet, travel times the frames exposed when the bullets would have reached the limousine, and when Zapruder would have heard the muzzle blast, can be calculated. (See cols. 3 and 4 of top of table II.)

  27. In constructing and then using table II, a note of caution is appropriate:Any calculation of the frames exposed, based upon a correlation with the sound impules on the Dallas Police Department tape, should not be presumed to be absolutely precise because only the average, and not the precise, running speeds fox the camera are known, and the tape speed adjustment of 5 percent is similarly only an average. Thus, table II reflects mathematical calculations of frame numbers utilizing the available averages, that is, a tape speed adjustment of 5 percent (although the tape may have actually been closer to 4 1/2 percent or 5 1/2 percent slow at times) and the parameters of 18.0 to 18.5 frames per second camera speed. The actual frames when an event occurred may have been a frame or two different than the calculations based upon such estimated averages.

  28. As can be seen from columns 3 and 6 of the table, there is good correlation for shots 1, 2, and 4, where in each case the blurs occur when the acoustics data reflect. Zapruder would have heard the shots. On the other hand, there is no corelation (except possibly for the. Scott. measurements) for shot No. 3, since the blur at. 289-293 precedes the arrival of sound at Zapruders' location and thus could not have been a product of his startle reaction to that sound.*

  29. The same calculations and comparisons can be made using the assumption that the third shot, (the one from the grassy knoll) caused the head wound at, frame 312. This process is reflected in the lower portion of table II. Here there are blurs following each shot, the first two lagging behind Zapruder's hearing of the sound by 9 to 12 frames (1/2 to 2/3 second), and the second two immediately following his hearing the sound.
    *There is one relatively small panning error between frames 300 and 305 detected by Scott's technique. See figure II-4.

    TABLE II

    *These times are the adjusted spacing of the sound impulses on the Dallas Police Department tape recording, and not the spacing of trigger pull or bullet impact times, which would vary somewhat from these times based upon the distances between the weapons, the limousine, and the motorcycle microphone that transmitted the sounds.

    **The blur episodes are taken from Table I.

    4. The trajectory analysis