38a. The nonissue of blurring

Introduction
   In the last year or two, the issue of blurring of the Zapruder images surrounding the head shot has been raised in certain quarters. Its alleged effects range from presenting the false impression of a forward snap to its preventing researchers from concluding anything about the president's motions after the head shot. It is now being claimed that the head-shot part of the film is essentially useless unless it is "deblurred" by complex mathematical procedures. The potential derogatory role of blurring has emerged as the most serious criticism of this monograph. This section shows that blurring is actually a nonissue.
    The most recent attention to blurring began with an article by David Wimp entitled "A Motion Blur Analysis of the Zapruder Film." This undated manuscript was posted on Ronald Hepler's web site (at http://server3002.freeyellow.com/rhepler/Motion%20Blur.htm ) during fall 2002 or earlier. It consists of three parts, "The Effect of Motion Blurring on Contrast Edges," "A Method for Measuring Distances in Motion Blurred Images," and "Measurements of JFKs [sic] Front to Back Head Movements in the Zapruder Film." The author used the tools presented in the first two parts to derive new positions of JFK's head in frames 304 through 319, which were presented in the third part. David Wimp's measurements differ radically from Josiah Thompson's measurements in Six Seconds in Dallas, which this monograph is based on. Wimp's results have the back of the head beginning to move forward by 311, then rapidly forward in 312, and only slightly more rapidly in 313. The head reaches its forward peak in 314 and then moves backward, first slowly and then more rapidly.
    The scenario through at least 314 cannot be. First of all, a cursory viewing of the film shows no perceptible movement in the few frames before 313, where the frames are sharp enough to reveal any movement that might have been present. Just view the "medium" images on "Images of an Assassination" and you will see. Alternatively, view a subset of the images here. Relative to Mrs. Kennedy's head, JFK's head did not move. In particular, there is no perceptible movement from 311 to 312, where Wimp claims to find major movement. Furthermore, the head can be seen to begin moving backward immediately after 313, not forward as Wimp calculates. I do not know where he went wrong, having not studied his method in detail. I just know that his method is wrong because it gives results counter to obvious features of the film.
    The second problem with Wimp's analysis is lack of context—it makes no effort to see whether its results mesh with the other physical evidence. In effect, he presents only half an argument, which is logically unacceptable. It is also wrong, for the other physical evidence shows that his conclusions cannot be right. (More on this below.)
    Josiah Thompson found Wimp's article and commented favorably on it at the JFK Lancer meeting in Dallas in late November 2002. He followed it up with a note to Lancer's discussion group on 25 November 2002, in which he stated that he had known for some time that his previous calculations were wrong because they were based on blurred frames. He noted that he specifically believed that his earlier conclusion about a "small forward movement of JFK's head followed by a huge back and leftward snap of his head and body" was mistaken. He went on to state that if his interpretation of Wimp's graphs was correct, "it seems to me that the movement of the head shows no forward movement that could be ascribed to a shot from the rear." This is another statement that suffers from lack of context.
     The Wimp/Thompson argument leads to dead ends at every turn. The first dead end is the idea that the deblurred motions reveal no quick forward movement from a rear hit. Aside from seeing the forward motion in 312–313, we know that it existed because there was a rear hit, as evidenced by the ballistics and the NAA—the lead fragments from JFK's brain match the lead from the large fragment found in the front seat (and dismatch the fragments associated with the body shot), and the two large fragments from the front seat were traced ballistically to Oswald's rifle in the TSBD. Thus JFK's head did indeed snap forward. The calculations that form the bulk of this monograph show that the head would have moved forward by an amount consistent with the measured 2.2 inches. The second dead end is the notion that the rearward motion of JFK's head is hardly faster or more abrupt than its forward motion (Wimp's plots). That means that the rearward motion can't be ascribed to a bullet, either, because the forward motion can't be. That leaves the totally illogical situation that none of the motions after the head shot can be ascribed to bullets. Something had to make his head snap forward and backward—it didn't just snap forward and backward because it felt like it.
    Thus David Wimp's revised measurements and interpretations of JFK's movement are wrong on several counts: (1) They place major movement at 311 to 312, where there is none; (2) They halve the movement between 312 and 313 (1 inch vs. the observed 2 inches); (3) They are inconsistent with the known rear hit; and (4) They nonsensically require that no bullets caused any of the motions observed.

The noneffect of blurring
    Clearly, Wimp's revised motions are wrong in both fact and inference. But what is the right version? The original measurements by Thompson, with only slight revisions, are correct and up to the task. One need only reread Thompson's Appendix B to see how carefully these measurements were made. But what about the blurring that is now claimed to threaten the entire set of numbers? It is a huge and unjustified overreaction, akin to throwing the baby out with the bathwater. Trashing the whole data set is too pat and easy. There is a better way.
    To appreciate the strength of direct measurements on the existing frames, one must go back and review the frames carefully. Although more than half of the frames are blurred, nearly all of them can still be used. Consider first the infamous frame 313. It is true that this frame contains significant horizontal blurring. But it is also true that the back of the head is still defined fairly sharply, and can be positioned accurately. The nature of the blurring can be seen from the bright spots on the rollbar, which are brightest on the right and strung out toward the left. If you look carefully at the back of the head, you will see the same pattern but in black: a dark head with a slightly dark extension to the left (the blurring). That blurring is easy to compensate for. If you still need convincing that the head has moved seriously to the right from 312, just compare its angle and distance relative to Mrs. Kennedy in 313 vs. 312. The head clearly rotates toward the right (forward). We will say more about the position in 313 under "Smoothing."
    What about the blurring in other frames? The two worst cases are 318 and 319, where it is severe (as Zapruder was reacting to seeing JFK's head blown open). The blurring in 318 is so bad that Thompson didn't report a position of the head, and I consider 319 to be nearly as bad. But all the other frames seem usable to me, and several are very sharp. Interestingly, 312 is one of the sharpest. This sharpness can be used to determine very easily that the head did not move forward from 311 to 312.
    The above evaluations can be criticized because they are subjective. They are supported by objective graphical analysis, however. The central question is whether the data as measured by Thompson are robust, that is, whether the correct movements can be discerned over the inevitable noise of blurring and measurement error. The following figures show clearly that the data are robust and that the true movements can be discerned quite easily. The procedure has three steps: (1) identify candidate large-scale features; (2) identify smaller-scale features within the larger ones; and (3) determine whether the smaller features are genuine signals or just noise (from blurring or measurement error).
    Step 1. We begin with Figure 1, which shows the raw position, where "raw" refers to Thompson's averaged measurements as presented in Appendix B of Six Seconds in Dallas. The values are normalized to frame 312 (frame 312 set equal to zero). Note that frame 318 is missing. This figure appears to have six or seven large-scale features, depending on whether certain features are considered separately or together: (1) a long, slow rearward movement from 301 to 312; (2) an abrupt, rapid forward movement from 312 to 313; (3) an equally abrupt reversal of motion at 314; (4) a period of rearward motion and increasing velocity from 314 into the blank frame at 318; (5) a period of decreasing rearward speed from 319 into 321; (6) a reversal to forward motion at 321; and (7) a prolonged period of forward motion and increasing speed through 330.
    Step 2. Smaller-scale features, possibly perturbations, appear in at least the first and last of these major features. The first is the irregularities between 301 and 312; the last is the two irregularities at 324 and 326, both being high relative to the surrounding frames. A third case may be 319 and 320, which do not form a smooth pattern with 321 and 322. In contrast, the frames in the major rearward lurch (314 through 317) are remarkably regular.

Figure 1. Raw position (with respect to 312) from Thompson's Appendix B.

    Step 3. In evaluating the smaller features, we first recognize that blurring and random measurement error will both appear as "noise" in plots like Figure 1. Functionally, their effects are the same. They can be distinguished from each other, however, at least in principle, by displaying blurred and unblurred points differently and seeing whether the blurred points are noisier. This is done in Figure 2 below, with the unblurred points being indicated by white circles. The results are clear: the blurred points are no noisier than the unblurred points (with more of the blurred points being smooth than noisy); the unblurred points are no smoother than the blurred points; and the two sets of points work equivalently to form the large features. The best example of blurred and unblurred points working together is the long period of forward movement between 321 and 330. Except for 324, which is too high, the ten blurred and unblurred points jointly create a remarkable picture of steady forward movement and acceleration. Blurring or measurement error has affected only one of ten points. Perhaps the most important example of this regularity, though, is frames 313 through 321, which constitute the rearward lurch. Here the pattern is clear and smooth despite the blurring in four of the five frames. Note especially how the unblurred 316 fits perfectly into the smooth pattern established by the blurred 313 to 315 and 317. A  similar situation holds for points 301 through 312, where JFK's head is slowly drifting backward. Although the blurred points tend to be lower than the unblurred ones, a common pattern of slow, steady drifting is clear, which is confirmed by examining the frames visually.
    Thus steps 1–3 applied to Figure 2 demonstrate that the effects of blurring are small, random, and unimportant. Blurring is a nonissue for understanding the head shot and its motions.

Figure 2. Raw position with minimal blurring indicated by circles.

Back to Objections and Responses